Immobilization of the Distal Hinge in the Labile Serpin Plasminogen Activator Inhibitor 1

Abstract

The serpin plasminogen activator inhibitor-1 (PAI-1) plays an important role in the regulation of the fibrinolytic activity in blood. In plasma, PAI-1 circulates mainly in the active conformation. However, PAI-1 spontaneously converts to a latent conformation. This conversion comprises drastic conformational changes in both the distal and the proximal hinge region of the reactive center loop. To study the functional and conformational rearrangements associated solely with the mobility of the proximal hinge, disulfide bonds were introduced to immobilize the distal hinge region. These mutants exhibited specific activities comparable with that of PAI-1-wt. However, the engineered disulfide bond had a major effect on the conformational and associated functional transitions. Strikingly, in contrast to PAI-1-wt, inactivation of these mutants yielded a virtually complete conversion to a substrate-like conformation. Comparison of the digestion pattern (with trypsin and elastase) of the mutants and PAI-1-wt revealed that the inactivated mutants have a conformation differing from that of latent and active PAI-1-wt. Unique trypsinsusceptible cleavage sites arose upon inactivation of these mutants. The localization of these exposed residues provides evidence that a displacement of αhF has occurred, indicating that the proximal hinge is partly inserted between s3A and s5A. In conclusion, immobilization of the distal hinge region in PAI-1 allowed the identification of an “intermediate” conformation characterized by a partial insertion of the proximal hinge region. We hypothesize that locking PAI-1 in this transition state between active and latent conformations is associated with a displacement of αhF, subsequently resulting in substrate behavior. The serpin plasminogen activator inhibitor-1 (PAI-1) plays an important role in the regulation of the fibrinolytic activity in blood. In plasma, PAI-1 circulates mainly in the active conformation. However, PAI-1 spontaneously converts to a latent conformation. This conversion comprises drastic conformational changes in both the distal and the proximal hinge region of the reactive center loop. To study the functional and conformational rearrangements associated solely with the mobility of the proximal hinge, disulfide bonds were introduced to immobilize the distal hinge region. These mutants exhibited specific activities comparable with that of PAI-1-wt. However, the engineered disulfide bond had a major effect on the conformational and associated functional transitions. Strikingly, in contrast to PAI-1-wt, inactivation of these mutants yielded a virtually complete conversion to a substrate-like conformation. Comparison of the digestion pattern (with trypsin and elastase) of the mutants and PAI-1-wt revealed that the inactivated mutants have a conformation differing from that of latent and active PAI-1-wt. Unique trypsinsusceptible cleavage sites arose upon inactivation of these mutants. The localization of these exposed residues provides evidence that a displacement of αhF has occurred, indicating that the proximal hinge is partly inserted between s3A and s5A. In conclusion, immobilization of the distal hinge region in PAI-1 allowed the identification of an “intermediate” conformation characterized by a partial insertion of the proximal hinge region. We hypothesize that locking PAI-1 in this transition state between active and latent conformations is associated with a displacement of αhF, subsequently resulting in substrate behavior. Plasminogen activator inhibitor-1 (PAI-1) 1The abbreviations used are: PAI-1, plasminogen activator inhibitor 1; t-PA, tissue-type plasminogen activator; u-PA, urokinase-type plasminogen activator; RCL, reactive center loop. is the most important physiological inhibitor of tissue-type plasminogen activator (t-PA) and urokinase-type plasminogen activator (u-PA) (1Kruithof E.K. Tran Thang C. Ransijn A. Bachmann F. Blood. 1984; 64: 907-913Google Scholar). It plays an important role in the regulation of the fibrinolytic activity in human blood (2Van Meijer M. Pannekoek H. Fibrinolysis. 1995; 9: 261-276Google Scholar). In addition to its role in fibrinolytic processes, PAI-1 has been demonstrated to be involved in a variety of other (patho)physiological processes. PAI-1 is a member of the superfamily of the serpins. The three-dimensional structure of these glycoproteins consists of three β-sheets, nine α-helices, and a reactive center loop (RCL, P16–P10′). Typically their active site comprises the P1–P1′ (bait) residues, located on the RCL, thereby serving as a pseudo-substrate for the proteinase (3Schechter I. Berger A. Biochem. Biophys. Res. Commun. 1967; 27: 157-162Google Scholar). N-terminally, the RCL is connected via the so-called proximal hinge (P14–P18) to β-strand 5 of the large central β-sheet A (s5A). C-terminally, the RCL is connected to s1C (i.e. the distal hinge region). PAI-1 is unique among the serpins because of its functional and conformational flexibility. The active conformation of PAI-1 inhibits its target proteinases by the formation of a stable, inactive complex. After the formation of an initial, reversible Michaelis-like complex, the proteinase cleaves the active site of PAI-1 and forms a stable, covalent complex resulting in the inactivation of the proteinase (4Lawrence D.A. Nat. Struct. Biol. 1997; 4: 339-341Google Scholar, 5Wright H.T. Scarsdale J.N. Proteins. 1995; 22: 210-225Google Scholar). PAI-1, which has a high second order rate constant for inhibition of t-PA of 2.0 × 107m–1 s–1 is distinguished from other serpins by the facile insertion of its RCL. This property is thought to be responsible for the spontaneous transformation of the unstable, active form of PAI-1 with a half-life (t½) of only 2.5 h at 37 °C to a stable, inactive “latent” conformation (6Hekman C.M. Loskutoff D.J. J. Biol. Chem. 1985; 260: 11581-11587Google Scholar). In this latent conformation residues P15 to P4 of the RCL insert into the β-sheet A to form s4A, and residues P3 to P10′ form an extended loop at the surface of the molecule (7Mottonen J. Strand A. Symersky J. Sweet R.M. Danley D.E. Geoghegan K.F. Gerard R.D. Goldsmith E.J. Nature. 1992; 355: 270-273Google Scholar). Latent PAI-1 can be reconverted to the active form by denaturation and refolding (8Hekman C.M. Loskutoff D.J. Biochemistry. 1988; 27: 2911-2918Google Scholar). Apart from their distinct functional properties, distinction between the active and latent conformations of PAI-1 can also be made based on differences in their proteolytic susceptibility (9Egelund R. Schousboe S.L. Sottrupjensen L. Rodenburg K.W. Andreasen P.A. Eur. J. Biochem. 1997; 248: 775-785Google Scholar, 10Gils A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar, 11Urano T. Ihara H. Takada Y. Nagai N. Takada A.R. Biochim. Biophys. Acta. 1996; 1298: 199-208Google Scholar). Depending on the mobility of the RCL, the presence of co-factors (e.g. heparin) and the proteinase used, PAI-1 can also be cleaved in a substrate-type reaction, resulting in an inactivated inhibitor (12Declerck P.J. De Mol M. Vaughan D.E. Collen D. J. Biol. Chem. 1992; 267: 11693-11696Google Scholar, 13Munch M. Heegaard C.W. Andreasen P.A. Biochim. Biophys. Acta. 1993; 1202: 29-37Google Scholar, 14Urano T. Strandberg L. Johansson L.B. Ny T. Eur. J. Biochem. 1992; 209: 985-992Google Scholar, 15Gils A. Declerck P.J. Thromb. Haemostsis. 1998; 80: 531-541Google Scholar). Some serpins require secondary interactions for efficient inhibition of the target proteinase. Additional interaction sites between PAI-1 and t-PA that stabilize the reversible, initially formed complex have been suggested (8Hekman C.M. Loskutoff D.J. Biochemistry. 1988; 27: 2911-2918Google Scholar, 16Chmielewska J. Ranby M. Wiman B. Biochemistry. 1988; 251: 327-332Google Scholar, 17Lawrence D.A. Strandberg L. Ericson J. Ny T. J. Biol. Chem. 1990; 265: 20293-20301Google Scholar, 18Madison E.L. Goldsmith E.J. Gerard R.D. Gething M.J. Sambrook J.F. Nature. 1989; 339: 721-724Google Scholar). The residues P4′, P5′, and P9′ of PAI-1 were assumed to be important for this interaction (18Madison E.L. Goldsmith E.J. Gerard R.D. Gething M.J. Sambrook J.F. Nature. 1989; 339: 721-724Google Scholar, 19Van Meijer M. Roelofs Y. Neels J. Horrevoets A.J. Van Zonneveld A.J. Pannekoek H. J. Biol. Chem. 1996; 271: 7423-7428Google Scholar). As previously shown (20Stoop A.A. Eldering E. Dafforn T.R. Read R.J. Pannekoek H. J. Mol. Biol. 2001; 305: 773-783Google Scholar), stabilizing mutations tend to be located in areas implicated in mobility of the serpin structure. The latency transition requires an additional flexibility, in which strand 1C is peeled away from sheet C and passed through the gate region (7Mottonen J. Strand A. Symersky J. Sweet R.M. Danley D.E. Geoghegan K.F. Gerard R.D. Goldsmith E.J. Nature. 1992; 355: 270-273Google Scholar). Overall, the stabilizing mutations can be grouped into five distinct subdomains of PAI-1, each of which has been associated with the latency transition: the vitronectin-binding domain, the helix F/loop subdomain, the helix C subdomain, the proximal hinge, and the sheet C/“back of the gate” region. Based on these five subdomains, the following model for the consecutive steps involved in the latency transition of PAI-1 was proposed. First, the proximal hinge region of the RCL would insert up to the point where s1C starts to peel away. Judging from the structure of the α1-antichymotrypsin δ-form (21Gooptu B. Hazes B. Chang W.S. Dafforn T.R. Carrell R.W. Read R.J. Lomas D.A. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 67-72Google Scholar), about four residues of the P region can insert without movement of s1C. To allow further insertion of the P region, the helix F/loop subdomain will have to move aside with concomitant conformational changes within the vitronectin-binding domain and the C-helix, whereas s1C (P′ region) passes through the gate region (22Kruger P. Verheyden S. Declerck P.J. Engelborghs Y. Protein Sci. 2001; 10: 798-808Google Scholar). Finally, insertion of the remainder of the RCL takes place, resulting in the latent conformation. Based upon this proposed pathway, we hypothesized that immobilization of the distal hinge region (s1C) (a) might prolong the functional half-life of PAI-1, (b) might prevent typical latency transition (i.e. full RCL insertion), but (c) should not hamper the initial steps (i.e. partial insertion of the proximal hinge and displacement of αhF). Characterization of mutants in which disulfide bonds were engineered to immobilize the distal hinge region revealed that the functional stability is indeed increased and that a conformational transition occurs in which PAI-1 is locked into a transition state between active and latent PAI-1. Importantly, this “intermediate” conformation is characterized by a partial insertion of the proximal hinge and a displacement of αhF, resulting in altered functional properties. Oligonucleotides for mutagenesis and sequencing were purchased from Amersham Biosciences. Mutagenesis was performed using the QuikChange™ XL site-directed mutagenesis kit from Stratagene (La Jolla, CA). Pfu Turbo™ DNA polymerase was obtained from Stratagene. The PCRs were done with the Mastercycler® gradient from Eppendorf (Hamburg, Germany). The expression vector pIGE20 was provided by Innogenetics (Ghent, Belgium), together with the bacterial strains Escherichia coli DH1λ for cloning and MC1061 for expression as well as the pAcI plasmid encoding the thermolabile repressor. Plasmid DNA purification was performed using a Nucleobond™ purification kit (Macherey-Nagel, Germany). Agar and Luria Broth growth medium were purchased from Invitrogen. Dithiothreitol and the proteinase inhibitors leupeptin, phenylmethanesulfonyl fluoride, pepstatin, benzamidine HCl, and antipain were from Sigma. SP-Sepharose™ Fast Flow, Activated Thiol-Sepharose® 4B and 4B were purchased from Amersham Biosciences. The and kit was obtained from The The substrate was obtained from was purchased from Sigma. t-PA was a of from human and of high and in a was a from of provided and were obtained from Sigma. human PAI-1 were as previously S. Declerck P.J. 1995; Scholar, S. Declerck P.J. Biochim. Biophys. Acta. 1997; Scholar). human PAI-1 was in E. coli as previously A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar). were obtained from for was performed with the using gradient and DNA was with the to the for the sequencing kit from Amersham The mutants were using the and for the of PAI-1 is as PAI-1-wt PAI-1 with mutations for PAI-1 in which has been by as of and of each were used in the The of an initial denaturation of at °C by of denaturation at at an and at The was by a of at of was to h at 37 the vector (i.e. the of was in E. coli DH1λ by on A was in of and the DNA was using the Nucleobond™ purification and were by The presence of the together with the of was by and purification was performed as previously A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar, E. Eur. J. Biochem. with the was in E. coli MC1061 together with the pAcI the for a thermolabile repressor. were at °C in medium with and on an at were and the PAI-1 expression was by the to After the were the was in and a of proteinase inhibitors 2.5 A benzamidine antipain phenylmethanesulfonyl and the were in a The was by at × for at The obtained was to was performed at °C using and SP-Sepharose™ Fast was with a The was with a and to a of and and the After a with were using a gradient in were and a with The was with and PAI-1 was using a gradient in for the purification of the mutants residues, was from resulting in a of disulfide bond were to for of their and in °C The of PAI-1 in was at of and forms was done on Activated The was in on a at a rate of After a with the A the initial (i.e. the was using with at a rate of The to the of the of the disulfide was done with a and kit This kit provides a for based upon a by R. A.R. Biochem. 1993; Scholar, R. A.R. 1995; 251: Scholar). PAI-1 were with This molecule the of on that have high R. A.R. Biochem. 1993; Scholar). a an with which a of the active The of is using the The were using the of forms was by because the form the PAI-1 were with to a between and After and for at were by performed both and using gradient The were by with and by with the of the of PAI-1 and PAI-1 PAI-1 were with to a between and and for at 37 °C as in the with the proteinases at a The was by and for at The were by performed using gradient with the The were by with and by with the as previously A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar). of the were to a between and using the to a with The were at 37 and were at The activity was by with a of t-PA for at 37 °C and of formed by and The of active PAI-1 at the was as a of the initial of active PAI-1, and the was used for of the of the between t-PA and of t-PA was used in the of this and the mutants. of the of t-PA and by PAI-1 and PAI-1 rate of inhibition of t-PA and by PAI-1 and PAI-1 was as A. Declerck P.J. J. Biol. Chem. 1997; Scholar), with In and t-PA were used at of and PAI-1 and its at of order were at and and the were by the addition of the A. Declerck P.J. J. Biol. Chem. 1997; at a PAI-1 activity and without on the t-PA the activity used to t-PA activity Chang C. Thromb. 1984; Scholar, A. J. Knockaert I. Declerck P.J. Thromb. 2001; Scholar). order rate were as A. Declerck P.J. J. Biol. Chem. 1997; Scholar). The rate of inhibition of the proteinases by mutants (i.e. and was by of complex formation using these PAI-1 were at 37 °C with a of the and the formation of the covalent complex were at and The was by and for at The rate of complex formation was using a of the rate of inhibition × of and were with to a between and were at a of and the were at 37 between and a was and the was by and for at The were by performed using gradient and The of the cleavage sites was revealed through of the denaturation were as A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar, A. Declerck P.J. J. Biol. Chem. 1997; Scholar). were to in and for h at a constant from to was by and the was on PAI-1 was by using and as and PAI-1 is as a of the initial The is the at which of the is these active PAI-1 has been shown to have a and the substrate form has been shown to have a In the latent and cleaved form a high to denaturation A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar). of differences was using of were not to be of for into (a) the of mutations on the activity and stability of PAI-1, (b) the of the in the three-dimensional and (c) a of between inserted the following disulfide mutants were and of the mutants were also and and revealed a complete of activity and were not further of the disulfide PAI-1 of and purification to of forms yielded the only that as an form to of the of the in with used However, from the demonstrated that of forms were of form in and in a of the of PAI-1 and PAI-1 of PAI-1-wt with t-PA revealed the presence of the three typical functional forms and at as the mutants revealed the presence of the three for at these where to a complete three disulfide mutants also revealed the presence of the three typical functional However, with PAI-1-wt the of was increased for PAI-1-wt, t-PA and inactivation at 37 as of PAI-1 with with with with were obtained a with as of PAI-1 with with were obtained a with in a As the inactivation of PAI-1-wt at 37 °C in the formation of latent PAI-1, whereas the of substrate A in functional was for the for and which a increased substrate behavior. In inactivation of the disulfide mutants at 37 °C was mainly associated with a in substrate for PAI-1-wt, It was also that cleavage of PAI-1-wt, and is with the of cleavage (i.e. for PAI-1-wt and and for about In the half-life for cleavage of is about not in with obtained of the for mutants exhibited a functional stability comparable with that of PAI-1-wt In contrast disulfide mutants had a up to increased PAI-1-wt. of the of t-PA and by PAI-1 and PAI-1 differences for and both with t-PA and to a (i.e. and a behavior. (i.e. to for the t-PA and u-PA, of the active a second order rate constant of inhibition comparable with that of PAI-1-wt The other with a second order rate constant of inhibition with that of PAI-1-wt was for the substrate order rate of inhibition of t-PA and by PAI-1-wt and 2.0 × × × × × × × × × 2.0 × × × × 2.0 × × × × × × × Fast in a and of the active conformation of the forms of the disulfide mutants were comparable with with active PAI-1-wt not Strikingly, digestion of the inactivated forms of the disulfide mutants revealed a digestion with that for inactivated PAI-1-wt. As inactivated PAI-1-wt is not cleaved by and whereas digestion of the inactivated mutants with revealed cleavage in the RCL and for an of the RCL. of inactivated PAI-1-wt and yielded with a of to a cleavage at (9Egelund R. Schousboe S.L. Sottrupjensen L. Rodenburg K.W. Andreasen P.A. Eur. J. Biochem. 1997; 248: 775-785Google Scholar, 10Gils A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar). In inactivated disulfide mutants yielded a digestion pattern and resulting in the of and with the and of the sequencing of these revealed that cleavage had at of inactivated PAI-1-wt and and and with and and trypsin and PAI-1 in the of proteinase. and PAI-1 with the proteinase for and denaturation of PAI-1-wt and PAI-1 inactivation revealed comparable °C as previously for PAI-1-wt A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar). of PAI-1-wt of most of the in a increased to denaturation with °C with the of these inactivated as a latent conformation (with a inserted as demonstrated previously for latent PAI-1-wt A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar). In the to denaturation of and the disulfide mutants virtually for the of forms and with their functional of denaturation of PAI-1 at 37 °C and at 37 in a Based on a model the consecutive steps involved in the latency transition of PAI-1, a movement of s1C away from sheet C (7Mottonen J. Strand A. Symersky J. Sweet R.M. Danley D.E. Geoghegan K.F. Gerard R.D. Goldsmith E.J. Nature. 1992; 355: 270-273Google Scholar, P. Verheyden S. Declerck P.J. Engelborghs Y. Protein Sci. 2001; 10: 798-808Google Scholar, De De Declerck P.J. Nat. Struct. Biol. 1995; Scholar), we hypothesized that immobilization of s1C in a of the active conformation but would not prevent in the proximal of s1C in a of the half-life but not prevent functional of these inactivated forms their target proteinases t-PA and revealed that not as latent forms but as substrate-like This that in contrast to a latent the RCL of these inactivated mutants is but in contrast to their active the RCL has a distinct conformation. proteolytic digestion with trypsin revealed the presence of previously cleavage sites in these inactivated mutants. of these cleavage sites in the of αhF (i.e. located in the s3A and is for a of αhF in these inactivated distinct from that in active and latent a digestion pattern was with an inactivated substrate A. N. Knockaert I. Declerck P.J. Biochim. Biophys. Acta. 1998; Scholar), a between this functional (i.e. and the conformational of denaturation that in contrast to latent forms by the full insertion of the RCL and a increased to the conformation of these inactivated substrate-like is comparable with that of forms C and We hypothesize that upon inactivation of these disulfide resulting in the of substrate-like the distal hinge region is not (i.e. the as in the active whereas the proximal hinge region partly inserted insertion by the immobilization of resulting in a of αhF of the proximal hinge of the RCL in with displacement of αhF provides an thereby the serpin of through the It is well that the interaction between a serpin and its target proteinases can in a cleaved serpin in the formation of a covalent complex of the the substrate is the of RCL insertion formation of the and to of the of the proteinase. to H.T. Scarsdale J.N. Proteins. 1995; 22: 210-225Google Scholar, Biochemistry. 2000; Scholar, Scholar, S. R. Goldsmith E.J. Nat. Struct. Biol. 2001; Scholar), is assumed that for activity (a) full of the RCL is (b) upon initial interaction with the target proteinase insertion has (c) partial insertion only starts formation of the and this insertion and movement of the proteinase in a displacement of αhF, which as a the for proteinase to the covalent complex. that not only the J. D.A. Biochemistry. 1997; but also the of is of the in the substrate has been hypothesized that displacement of αhF interaction an a Scholar). study that insertion of RCL and displacement of αhF is should not to interaction with the target proteinase but in a with proteinase These might that partial insertion De B. R. J. Nat. Struct. Biol. of the hinge region is a in the of the formation of a complex A. Knockaert I. Declerck P.J. Biochemistry. 1996; 35: 7474-7481Google Scholar, De De Declerck P.J. Nat. Struct. Biol. 1995; Scholar, M. M. J. Johansson Ny T.R. Nat. Struct. Biol. 1997; 4: Scholar, Read R.J. Carrell R.W. Nature. 2000; Scholar), cleavage and partial insertion should each other Strikingly, this is in with a H.T. Scarsdale J.N. Proteins. 1995; 22: 210-225Google that the substrate of be by a partial of These also in with the that in the cleavage of the P1–P1′ bond insertion of the hinge region M. M. Ny T. J. Biol. Chem. 1995; Scholar, D.A. D. D.E. J. Biol. Chem. 1995; Scholar). have suggested that the mobility of strand 1C be important to the of serpins F. Protein Sci. Scholar, I. I. J. Biol. Chem. 1992; 267: Scholar). that in the mobility of strand 1C not the activity of the molecule Chang W.S. J. Biol. Chem. 1997; Scholar). the conformational changes in PAI-1 by the immobilization of that of strand 1C from β-sheet C is not for the activity of PAI-1. However, the for the of t-PA with that the is efficient the of the RCL is interaction with is by these in flexibility. It should be that this effect on the is not by the at because the at had effect on an as a both with t-PA and This in the of both to form with the The is most in the distal hinge region because this region is assumed to be important for the rate of formation of the initial, reversible, complex. In we have shown that immobilization of s1C not the in PAI-1 but its of to the formation of an in which the RCL is and αhF is substrate behavior. of will the for substrate in PAI-1 and other serpins.