Portal de Periódicos da CAPES

Lysine 156promotes the anomalous proenzyme activity of tPA: X-ray crystal structure of single-chain human tPA

1997; Springer Nature; Volume: 16; Issue: 16 Linguagem: Inglês

10.1093/emboj/16.16.4797

1460-2075

Martin Renatus,

Click Chemistry and Applications

Article15 August 1997free access Lysine 156 promotes the anomalous proenzyme activity of tPA: X-ray crystal structure of single-chain human tPA Martin Renatus Corresponding Author Martin Renatus Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Richard A. Engh Richard A. Engh Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Biochemical Research Center, Boehringer Mannheim GmbH, Nonnenwald 2, D-82372 Penzberg, Germany Search for more papers by this author Milton T. Stubbs Milton T. Stubbs Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Robert Huber Robert Huber Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Stephan Fischer Stephan Fischer Biochemical Research Center, Boehringer Mannheim GmbH, Nonnenwald 2, D-82372 Penzberg, Germany Search for more papers by this author Ulrich Kohnert Ulrich Kohnert Biochemical Research Center, Boehringer Mannheim GmbH, Nonnenwald 2, D-82372 Penzberg, Germany Search for more papers by this author Wolfram Bode Corresponding Author Wolfram Bode Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Martin Renatus Corresponding Author Martin Renatus Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Richard A. Engh Richard A. Engh Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Biochemical Research Center, Boehringer Mannheim GmbH, Nonnenwald 2, D-82372 Penzberg, Germany Search for more papers by this author Milton T. Stubbs Milton T. Stubbs Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Robert Huber Robert Huber Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Stephan Fischer Stephan Fischer Biochemical Research Center, Boehringer Mannheim GmbH, Nonnenwald 2, D-82372 Penzberg, Germany Search for more papers by this author Ulrich Kohnert Ulrich Kohnert Biochemical Research Center, Boehringer Mannheim GmbH, Nonnenwald 2, D-82372 Penzberg, Germany Search for more papers by this author Wolfram Bode Corresponding Author Wolfram Bode Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany Search for more papers by this author Author Information Martin Renatus 1, Richard A. Engh1,2, Milton T. Stubbs1,1,3, Robert Huber1, Stephan Fischer2, Ulrich Kohnert2 and Wolfram Bode 1 1Max-Planck-Institute of Biochemistry, Department of Structural Research, D-82152 Martinsried, Germany 2Biochemical Research Center, Boehringer Mannheim GmbH, Nonnenwald 2, D-82372 Penzberg, Germany 3Institute for Pharmaceutical Chemistry, Philipps-University Marburg, Marbacher Weg 6, D-35032 Marburg, Germany *Corresponding authors. E-mail: [email protected] or E-mail: [email protected] The EMBO Journal (1997)16:4797-4805https://doi.org/10.1093/emboj/16.16.4797 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Tissue type plasminogen activator (tPA) is the physiological initiator of fibrinolysis, activating plasminogen via highly specific proteolysis; plasmin then degrades fibrin with relatively broad specificity. Unlike other chymotrypsin family serine proteinases, tPA is proteolytically active in a single-chain form. This form is also preferred for therapeutic administration of tPA in cases of acute myocardial infarction. The proteolytic cleavage which activates most other chymotrypsin family serine proteinases increases the catalytic efficiency of tPA only 5- to 10-fold. The X-ray crystal structure of the catalytic domain of recombinant human single-chain tPA shows that Lys156 forms a salt bridge with Asp194, promoting an active conformation in the single-chain form. Comparisons with the structures of other serine proteinases that also possess Lys156, such as trypsin, factor Xa and human urokinase plasminogen activator (uPA), identify a set of secondary interactions which are required for Lys156 to fulfil this activating role. These findings help explain the anomalous single-chain activity of tPA and may suggest strategies for design of new therapeutic plasminogen activators. Introduction The chymotrypsin family serine proteinases normally are secreted as single-chain zymogens with weak reactivity towards unspecific active site reagents (such as DFP or NPGB) and virtually no reactivity towards peptidic inhibitors such as peptidyl chloromethylketones (Robinson et al., 1973; Morgan et al., 1974; Kerr et al., 1975). They are activated by proteolytic liberation of a highly conserved N-terminus (typically with the sequence IVGG at residues 16-19 in chymotrypsinogen numbering, which is used for tPA throughout this publication; see Table I and Lamba et al., 1996). The free α-amino group forms a solvent-inaccessible salt bridge with the Asp194 carboxylate group in the 'activation pocket' or 'Ile16 cleft' specific for the Ile-Val sequence. The formation of the salt bridge creates a functional substrate recognition site by reorienting Asp194 compared with the proenzyme structure. This reorientation restructures the surrounding activation domain, which includes the oxyanion hole, the catalytic triad and the S1 specificity pocket (Freer et al., 1970; Bode et al., 1976; Fehlhammer et al., 1977; Huber and Bode, 1978). The consequent low and high activities of the zymogen and the cleaved forms, respectively, enabled the evolution of robust regulation of critical biological processes in higher organisms, especially via activation cascades such as found in blood coagulation. Despite the common activation mechanism, the chymotrypsin family proteinases are highly specific for substrates and inhibitors. Table 1. Important residues cited in this study: residues in chymotrypsinogen numbering and the tPA numbering system Chymotrypsinogen numbering tPA numbering Cys1 Cys264 Gln10 Gln273 Arg15 Arg275 Ile16 Ile276 Lys17 Lys277 Phe21 Phe281 His37 His297 His57 His322 Tyr99 Tyr368 Ser110 Ser379 Lys143 Lys416 His144 His417 Lys156 Lys429 Arg174 Arg449 Thr186 Thr461 Asp194 Asp477 Ser195 Ser478 Ser214 Ser497 Lys222 Lys505 Pro244 Pro527 Exogenous substances, such as the free dipeptide Ile-Val, can activate the zymogen without proteolytic cleavage. The Ile-Val dipeptide binds in the Ile16 pocket and mimics the activating N-terminus (Bode and Huber, 1976; Bode, 1979). Interactions with very tight binding inhibitors can also induce the formation of an active conformation, as seen for example in the complex of bovine pancreatic trypsin inhibitor (BPTI) with trypsinogen (Bode et al., 1978). Many properties of the single-chain proenzyme form are adequately described by a two-state model: an inactive conformation or set of conformations is in equilibrium with active forms that possess a structured activation domain (Huber and Bode, 1978; Bode, 1979; Coletta et al., 1990). The partition between active and inactive forms depends on environmental conditions, such as the presence of the Ile-Val dipeptide (Bode, 1979), the Ile16 N-terminus (Hedstrom et al., 1996) or other ligands at the substrate-binding site (Bode et al., 1978; Bolognesi et al., 1982). More detailed analyses of the activation of trypsinogen to β-trypsin indicated the presence of several distinguishable activation levels in the transition from zymogen to active enzyme (Coletta et al., 1990). Several crystal structures of serine proteinases and their cognate zymogen forms in different environments are known, most notably chymotrypsin/chymotrypsinogen (Freer et al., 1970; Birktoft and Blow, 1972; Wang et al., 1985) and trypsin/trypsinogen (Bode et al., 1976; Fehlhammer et al., 1977; Huber and Bode, 1978; Bode, 1979). Tissue type plasminogen activator (tPA) is a chymotrypsin family serine proteinase central to fibrinolysis (for review, see Madison, 1994). It selectively activates the zymogen plasminogen to plasmin by the activation cleavage described above. Plasmin then degrades fibrin with relatively broad specificity and also activates single-chain tPA (sc-tPA); the latter process closes a positive feedback cycle. Plasminogen activation by tPA is strongly stimulated in the presence of fibrin and is inhibited principally by the fast-acting serpin plasminogen activator inhibitor 1 (PAI-1). It is similar in many ways to uPA (urokinase plasminogen activator), with similar substrate and inhibitor specificities, sequence, structure and function. In contrast to tPA, uPA is found mainly in tissues where it seems to initiate the plasmin-mediated proteolysis of extracellular matrix proteins. Despite their functional and structural similarity, the activities of the single-chain forms of uPA and tPA differ significantly. In contrast to uPA and other serine proteinases, sc-tPA has significant activity toward small substrates compared with the two-chain form (tc-tPA): it has low 'zymogenicity'. The 'zymogenicity' of a serine proteinase designates the extent of activation upon cleavage, or in other words the ratio of the catalytic activities of the two- and single-chain forms. The low zymogenicity of tPA (5-10: Rånby et al., 1982; Tate et al., 1987; Madison et al., 1993) is in stark contrast to the high zymogenicity of trypsin/trypsinogen (107: Bode, 1979). Other serine proteinases have intermediate zymogenicities, such as factor XIIa and urokinase with values of 4000 and 250 (Silverberg and Kaplan, 1982; Lijnen et al., 1990), respectively. The structural basis for these differences, and in particular for the low zymogenicity of tPA, has been unclear, although several factors have been identified. For example, chymotrypsinogen and trypsinogen possess a 'zymogen triad' Asp194-His40-Ser32, which stabilizes Asp194 and the activation domain in a non-functional conformation (Bode, 1979; Madison et al., 1993). These stabilizing factors are absent in tPA and many other serine proteinases. To explain the anomalously low zymogenicity of tPA, salt bridge interactions between Asp194 and one of several lysine residues have been proposed; Lys17 (for tPA numbering see Table I) (Wallén et al., 1983), Lys143 (Petersen et al., 1990) and Lys156 (Nienaber et al., 1992; Lamba et al., 1996; Tachias and Madison, 1997) appeared most likely from model considerations. Understanding the unusual activity levels of tPA is complicated by the co-factor interactions of fibrin binding. Full-length human tPA, like its physiological substrate plasminogen, binds to fibrin via interactions with the finger domain. In the presence of fibrin, plasmin cleavage by both the sc-tPA and the activated two-chain form is stimulated such that the efficiencies of both forms are similar. In comparison, the catalytic efficiency of tPA's amidolytic activity against chromogenic substrates is stimulated only weakly in the presence of fibrin (Bringmann et al., 1995). This suggests that fibrin stimulation properties occur via a combination of several mechanisms: (i) a major effect by promoting tPA and plasminogen interaction through locally increased concentrations (Hoylaerts et al., 1982; Higgins et al., 1990); (ii) a major effect by promoting suitable orientations and accessibility of the plasminogen substrate cleavage site (Hoylaerts et al., 1982; Lamba et al., 1996); and (iii) an effect promoting the amidolytic efficiency of tPA itself. The latter effect is small [a factor of 4 for human tPA (Bringmann et al., 1995)] but presumably arises from suitable effects in the catalytic domain, slightly altering the catalytic state of the molecule. We have determined the crystal structure of the catalytic domain of recombinant human sc-tPA in a covalent complex with dansyl-Glu-Gly-Arg-chloromethylketone inhibitor. The structure reveals that the Lys156 side chain is bound in the Ile16 pocket and forms an asymmetric salt bridge with Asp194 (see Figures 2 and 3). Asp194, the active site and the substrate recognition sites are arranged as in tc-tPA (Lamba et al., 1996). This provides evidence that Lys156 plays a role in a non-proteolytic, self-activating process in sc-tPA, leading to its low zymogenicity. A comparison of the sc-tPA structure with other serine proteinase structures also provides a rationale for the higher zymogenicities of other serine proteinases which also have a lysine residue at position 156 but lack the necessary residues in the surrounding environment. Kinetic data confirming the role of Lys156 in the catalytic efficiency of full-length tPA have been published very recently by Tachias and Madison (1997). Figure 1.Stereo ribbon plot of the catalytic domain of sc-tPA in 'standard' orientation. The inhibitor (dansyl-Glu-Gly-Arg-chloromethylketone) shown by green sticks is covalently bonded to Ser195 and His57 of the catalytic triad. Key sc-tPA residues are shown as yellow sticks: Asp102, His57 and Ser195 of the catalytic triad; Asp189 at the base of the S1 specificity pocket; and Asp194 and Lys156 which form a salt bridge in the activation pocket. The red ribbon shows the conformation of the N-terminal activation loop which includes the plasmin cleavage site. Some loops arranged around the active site are labelled: the mainly disordered 37 loop and the 110 loop to the east; the partially disordered 186 loop to the south-west is in proximity to the activation loop. The figure was made with SETOR (Evans, 1993). Download figure Download PowerPoint Figure 2.Stereo plot of the activation domain environment of sc-tPA involving residues of the activation pocket, the activation loop and Lys156. Lys156 (blue) stabilizes the active conformation of sc-tPA via formation of a salt bridge with Asp194 (red). The strength of this interaction is presumably increased by the concerted solvent shielding effect of the (red) activation loop and of the hydrophobic residues Ile16, Phe21 and His144 (yellow). Lysines 17 and 143, which were considered potential activators of sc-tPA (Wallén et al., 1983; Petersen et al., 1990) are also shown (blue). Residues His188, Arg186A and Asn186F of the 186 loop (yellow) are presumed to be responsible for the conformational stabilization of this loop. Download figure Download PowerPoint Figure 3.Superposition of the activation domains of sc-tPA (yellow) and tc-tPA(bisb) (green) (Renatus,M., Bode,W., Huber,R., Stürzebecher,J., Parsa,D., Fischer,S., Kohnert,U. and Stubbs,M.T., submitted). The activation domains are identically formed although the mechanisms by which the active enzyme conformations are achieved (i.e. salt bridge formation with Asp194) differ. The 'active' conformation of Asp194 is stabilized by Lys156 of sc-tPA, but by the N-terminal primary ammonium group of Ile16 in tc-tPA. In the two-chain form, the aliphatic part of the Lys156 side chain is packed between side chains of Ile16, His144 and Phe21, while its distal ammonium group is anchored via hydrogen bonds to Ile16 O and Gly18 O; the first three N-terminal residues Ile16, Lys17 and Gly18 of the B chain are part of a three-strand β-sheet including the segments Lys143-His144 and Asp189-Ala190 (Lamba et al., 1996). In the single-chain form, the activation pocket is shielded from the solvent by the activation loop, in particular through the side chains of residues Ile16 and Phe21, and by the His144 side chain rendering the the salt bridge Lys156Nζ-Asp194Oδ more stable; the activation loop is further fixed in its position by the interaction between the side chains of Lys17 and Glu145. The 186 loop is shifted in sc-tPA relative to the position seen in tc-tPA(bisb), with residues Arg186A, Gln186F and His188 its stabilization. Download figure Download PowerPoint Results Overall structure The overall three-dimensional structures of the catalytic domain of human tc-tPA in complex with benzamidine (Lamba et al., 1996) and with a bis-benzamidine inhibitor (Renatus,M., Bode,W., Huber,R., Stürzebecher,J., Parsa,D., Fischer,S., Kohnert,U. and Stubbs,M.T., submitted), called here tc-tPA(b) and tc-tPA(bisb) respectively, correspond to that of sc-tPA. Optimal superposition of the sc-tPA and tc-tPA main chains excluding mobile or poorly defined regions (Ser7-Arg15, Lys36-Arg37D, Arg110D, Ser186B-Ala186G, Pro244) and using a distance cutoff of 1.5 Å identifies 238 equivalent Cα positions. The main differences are found in the activation domain (Bode et al., 1976; Huber and Bode, 1978) and in the adjacent surface loops. The catalytic domain of sc-tPA is roughly spherical, with a radius of 25 Å (see Figure 1). The 17 residues of the nominal A chain (residues N-terminal to the cleavage site Arg15-Ile16 of tPA) and the 252 residues of the nominal B chain are continuously linked and, with a few exceptions described below, ordered in the crystal. As in all chymotrypsin family serine proteinases, the molecule consists of two six-stranded barrel-like sub-domains; these pack against each other at an interface which forms the catalytic site. Although the two molecules in the asymmetric unit are not crystallographically identical, no notable differences between them were observed. Many of the features specific to tPA involve interactions with surface elements and loops of the catalytic domain (Lamba et al., 1996). Since tPA is active in its single-chain form, the comparison of these loops with the two-chain structure is particularly relevant to zymogenicity and the unique aspects of single-chain activity. The 37 insertion loop points away from the molecular surface into the solvent (to the 'east' of the active site cleft in Figure 1). As in both tc-tPA structures (Lamba et al., 1996), this insertion is only partially defined in the electron density map. No electron density is observed for the six residues Lys36, His37, Arg37A, Arg37B, Ser37C and Pro37D. Although this loop is in the vicinity of a symmetry-related molecule, apparently no specific interactions exist to stabilize a single conformation. This loop has been shown to be of fundamental importance for the interaction with its inhibitor PAI-1 (Madison et al., 1989, 1990a,b) and for fibrin specificity and stimulation (Bennett et al., 1991; Paoni et al., 1993b). The loop flexibility in tPA (and also in uPA, Spraggon et al., 1995) is probably associated with induced conformational changes upon interaction with PAI-1. The 186 loop, 'south-west' of the active site cleft, contains an insertion of eight residues compared with α-chymotrypsin. The open and extended loop protrudes away from the molecular surface into the solvent. It is bounded on its eastern side (orientation of Figure 2) by the 'activation loop' (see below). Excepting Arg186A and His188, which pair on the molecular surface and approach Arg15 (see Figure 3), and Asn186H, whose Nδ2 is linked with the carbonyl atoms of Ile16 and Gly18, the side chains of this loop do not make stabilizing interactions with other residues on the molecular surface. Part of the loop (Arg186A-Gly186C) is close to a symmetry-related molecule, but also for this near approach neither hydrogen bonds nor hydrophobic interactions were evident. In the conformation of sc-tPA, the 186 loop is close to the entrance frame of the S1 pocket (Ser214-Gln221A) and the turn formed by the succeeding residues Lys222, Asp223 and Val224. The importance of this loop for general function is not clear, since position 186G is deleted in mouse and rat tPA. This region seems to be involved in fibrin stimulation properties (Paoni et al., 1993a); this might, however, be species specific. The entire 186 loop was defined in tc-tPA(bisb), disordered in tc-tPA(b) (Lamba et al., 1996) and disordered at Gly186C, Gly186D and Pro186E in sc-tPA. Crystal contacts between the 186 loop and a symmetry-related neighbour might account for the order of the loop in tc-tPA(bisb). The density for the ordered residues of the 186 loop in sc-tPA shows that the loop is shifted relative to the position seen in tc-tPA(bisb), although the backbone conformations are similar where they are ordered (an approximate hinge position is evident in Figure 3). The crystal packing of the sc-tPA structure does not exclude the conformation of the 186 loop seen in tc-tPA(bisb). Also, the differing conformation of the adjacent 'activation loop' in sc-tPA (see below) would not obviously exclude the conformation of the 186 loop in tc-tPA(bisb) (see Figure 3). Different pH levels in the crystals may contribute to differences in the loop structure. The tc-tPA(b) crystals were grown at pH 7.3 (Lamba et al., 1996), the tc-tPA(bisb) crystals at pH 5.0 and the sc-tPA crystals at pH 7.2. For the tc-tPA(bisb) structure, the stabilization of the 186-loop in its particular conformation was attributed to a cluster of three residues: Arg186A, Gln186F and especially His188, whose protonation state presumably varies in the pH range. The 'autolysis' loop, Tyr141-Tyr151, located to the south of the active site cleft, is well defined by electron density. As in tc-tPA, the hydrophobic residues Leu147, Pro149 and Phe150 are exposed to the solvent, suggesting that this region might participate in specific interactions with substrates, inhibitors and co-factors (Lamba et al., 1996; Ke et al., 1997). The side chain of Tyr151 extends into the S2′ subsite. The side chain of Glu145 forms a salt bridge with Nζ Lys17 in tc-tPA(bisb) and in sc-tPA, although the Cα position of Lys17 is quite different in both cases. Insertion loops 186 and 110 (characteristic for plasminogen activators) lie approximately on the long axis of the catalytic domain and at opposite poles (Figure 1). Although the 110 insertion loop is completely defined in tc-tPA(b), the single residue Arg110D is undefined in sc-tPA. In tc-tPA(bisb), the three residues Asp110A, Ser110B and Ser110C are undefined; the equivalent insertion loop of uPA (Spraggon et al., 1995) shows high mobility. Thus this loop may be flexible under physiological conditions. No hypothesis about its function has been reported so far. The last residue at the C-terminus, Pro244, and the side chain of Arg243, which are close to the '110 loop', are also undefined in sc-tPA. N-terminal segment, activation loop and Ile16 cleft In the hemisphere opposite to that of the active site cleft, the N-terminal segment of the catalytic domain of human tPA is linked to the rest of the molecule by the disulfide Cys1-Cys122. In the structure reported here, it comprises the first three N-terminal residues Ser1E-Tyr1D-Gln1C of the full-length tPA, the two residues Ser1B and Thr1A of the linker between kringle 2 and the catalytic domain, and the initial part of the activation peptide (Cys1-Arg15). The 'activation loop' comprises residues Gln10-Phe21, including the activation cleavage site Arg15-Ile16 (see Figures 2 and 3). Residues Cys1-Lys17 follow the molecular surface in an almost extended conformation, forming a short antiparallel β-sheet with inter-main chain hydrogen bonds (Gln10N…Phe21O, Phe11O…Phe21N, Ile16N… Gly19O, Ile16O…Gly19N). Segment Ile16-Lys17-Gly18-Gly19 forms an almost regular 1,4-tight turn (Figures 2 and 3). As in the crystal structure of the catalytic domain of tc-tPA, the first four N-terminal residues (Ser1E-Tyr1D-Gln1C-Ser1B) of the nominal A chain are not defined by electron density. From Thr1A upstream, almost all main chain atoms of the A chain (with the exception of Gly2) and all side chain atoms (with the exceptions of Gln8 and Gln10) are defined by electron density. The geometry of the residues from Thr1A to Gln5 are very similar to the corresponding residues of tc-tPA, with Arg4 and Glu137 forming a salt bridge. Carboxy-terminal to Tyr6, the polypeptide chain in sc-tPA follows the molecular surface, while in tc-tPA the corresponding part of the A chain is not defined by electron density and might point into the solvent. Ile16, Phe21 and His144 form a hydrophobic cluster and partially shield the Ile16 cleft from solvent (see Figures 2 and 3). Several residues of the Ile16-Gly19 turn segment form additional stabilizing interactions with the adjacent autolysis and 186 surface loops: the carbonyl oxygen of Ile16 forms a hydrogen bond with the Nδ2 of Asn186H, Nζ of Lys17 with Oϵ1 of Glu145 and the carbonyl oxygen of Gly18 with N of Asp189. Glu144 is stabilized further by an interaction with the carbonyl oxygen of Cys220. The net effect of these interactions is to stabilize the 'activation loop' segment Phe11-Leu20 in a conformation resembling a flap or lid which shields the Ile16 pocket and the salt bridge Lys156-Asp194 from solvent. The Lys156-Asp194 salt bridge interaction confirms the hypothesis (Nienaber et al., 1992; Lamba et al., 1996) that Lys156 is particularly important for the low zymogenicity of tPA, which has now been supported by mutagenesis studies (Tachias and Madison, 1997). The side chains of Lys17 (Wallén et al., 1983) and Lys143 (Petersen et al., 1990), other candidates hypothesized to form the decisive salt bridge interaction with Asp194, extend away from the Ile16 pocket and are exposed to the bulk solvent; Lys17Nζ forms a salt bridge with Glu145Oϵ. The Ile16 pocket is shielded by the activation turn (Ile16-Gly19) and the hydrophobic cluster of Ile16, Phe21 and His144. Both shielding elements presumably increase the strength of the salt bridge interaction and may be an additional requirement for single-chain activity (see below for a discussion of single-chain activity relative to other Lys156-containing serine proteinases). The shielding position of the 'activation loop' is stabilized by: (i) hydrophobic interactions (between Ile16 and the aromatic groups of Phe11, Phe21 and Phe150); (ii) the formation of a salt bridge between Lys17 and Glu145 (Glu145 Oϵ2 makes a further interaction with Cys220 O); and (iii) the hydrogen bonds Asn186HNδ…Ile16O and Gly18O…Asp189N. By comparison, the tc-tPA structures are identical to sc-tPA for main chain atoms C-terminal to Leu20 and, as described above, N-terminal to Gln5. N-terminal to Leu20, the main chain of tc-tPA changes its direction abruptly at residues Gly18 and Gly19 (ψ-Gly18: −144° in tc-tPA and −21° in sc-tPA, φ-Gly19: 68° in tc-tPA and 177° in sc-tPA, see Figure 3). Consistent with their conservation among serine proteinases, Gly18 and Gly19 possess the flexibility required upon activation cleavage to allow Ile16 to insert into the Ile16 pocket and to form the internal salt bridge with Asp194. Lys156 adopts a g+-rotamer conformation in tc-tPA and a g−-rotamer conformation in sc-tPA. The side chain of Lys156 in tc-tPA is exposed to solvent and forms hydrogen bonds with carbonyl oxygen atoms of the activation peptide at Gly18 and Gly17 (see Figure 3). Inhibitor binding The inhibitor molecule as bound into the functional active site of sc-tPA is shown in Figure 4. The continuous density indicates the two covalent bonds between the P1 Arg group (residue ArmI3) and the enzyme, as is known from other serine proteinase structures complexed with chloromethylketone inhibitors (Bode et al., 1992; Brandstetter et al., 1995; Spraggon et al., 1995; Mather et al., 1996). The bonds exist between His57Nϵ2 and the methylene carbon and between Ser195Oγ and the carbonyl carbon, representing a tetrahedral hemiketal structure (Powers, 1977). Figure 4.Dansyl-Glu-Gly-Arg-chloromethylketone inhibitor (orange) and part of the active site of sc-tPA (yellow) superimposed on the final 2Fo−Fc electron density contoured at 0.9σ, oriented as in Figure 1. The figure was made with Main (Turk, 1992). Download figure Download PowerPoint The inhibitor backbone abuts the segment Ser214-Gly217, forming a short antiparallel slightly twisted β-sheet interaction. The side chain of ArmI3 binds in the S1 pocket, forming symmetric twin salt bridges between Nη1/Nη2 and Oδ1/Oδ2 of Asp189 at the bottom of the pocket. Nη1 forms an additional hydrogen bond to the carbonyl oxygen of Gly219 (d = 2.7 Å), as seen for substrate analogues containing an Arg at P1 as well as amidine-derived inhibitors in complex with serine proteinases. GlyI2 occupies the rather small S2 pocket of tPA and approaches the Oη group of Tyr99. This is consistent with the tPA preference for Gly as a substrate P2 residue (Ding et al., 1995; Madison et al., 1995). The side chain of GluI1 forms weak hydrogen bonds with the N of Gly219 and with the side chain of Gln192. The GluI1 carbonyl oxygen forms a favourable hydrogen bond with N Gly216. GluI1 and the dansyl group are also involved in crystal contacts. The N-terminus of the inhibitor is masked with a dansyl group. Although this group, with its combined aromatic moiety and distal amine, might be considered an ideal S4 binding group for tPA (Renatus,M., Bode,W., Huber,R., Stürzebecher,J., Parsa,D., Fischer,S., Kohnert,U. and Stubbs,M.T., submitted; see also the related discussion for factor Xa, Brandstetter et al., 1996), it does not bind into the hydrophobic S3/S4 cleft of tPA (Lamba et al., 1996), maybe due to steric hindrance caused by the dimethyl moieties. Rather, the group binds in an aromatic depression in a symmetry-related neighbour formed by residues Thr186, Tyr163, Trp133, Arg167 and Asp223. As with loop conformations, this binding geometry may determine crystal packing, or, c