The Serpin Database Back

Tryptase is a 31 kDa, glycosylated, trypsin-like enzyme that is stored in and released from all human mast cells. Tryptase is not inhibited by blood plasma SERPINs, such as anti-thrombin III and a1-proteinase inhibitor. The active protease exists as a homo-tetramer with the four active sites facing a central cavity and SERPINs are thought to be too large for their inhibitory loops to reach the active sites in the central core (Pereira et al., Nature 392:306,1998). Tightly bound heparin stabilizes both the tetrameric structure and activity. In spite of its resistance to inhibition by SERPINs, tryptase cleaves several large proteins at specific places. We have determined five cleavage site sequences in four human proteins. Fibrinogen is cleaved at Arg572 in the alpha chain, destroying the Arg-Gly-Asp binding site recognized by cellular intergrins and at Lys21 in the beta chain, which appears to be responsible for preventing thrombin initiated clotting (Thomas et al., Biochemistry 37:2291, 1998). Cleavage kinetics for fibrinogen at a-Arg572 and at b-Lys21 are similar with Km values of 0.20 and 0.26 ÊM, respectively, and 30 fold lower than the Km values of thrombin, indicating that fibrinogen is a preferred substrate. Tryptase also specifically cleaves high molecular weight kininogen at Arg431 (in the positively charged region), human vitronectin at Lys88 (between domains0, and pro-urokinase at Lys158 (at the normal activation site). Examination of these data using multiple sequence alignment reveals a surprising lack of homology on either side of the cleavage sites, indicating that tryptase has specificity only for Lys or Arg and not for surrounding residues. Tryptase cleavage of several large proteins argues against size exclusion of SERPINs from the tetrameric core of tryptase as the sole explanation for resistance to inhibition and suggests that the resistance of tryptase to SERPINs may be related to the lack of specificity displayed by tryptase. Tryptase Cleavage Sites

Fibrinogen-a TSYNR572GDST
Fibrinogen-b RPLDK21 KREE
HMW Kininogen HKHER431DQGH
Vitronectin QAQSK88GNPE
Pro-Urokinase RPRFK158 IIGG

Supported by the East Tennessee State University Research and Development Committee and the American Heart Association, Tennessee Affiliate, davidj@ETSU.edu

CHARACTERIZATION OF A MYXOMA-VIRUS SERPIN-LIKE PROTEIN
Jean-Luc GuÇrin, Christelle Camus, Jacqueline Gelfi, Stephane Bertagnoli, Corine Boucraut-Baralon and FrÇdÇrique Messud-Petit. Labo INRA-ENVT Microbio Moleculaire - 23 chemin des Capelles 31076 Toulouse cedex FRANCE

Myxoma virus (MV), a member of the Poxvirus family, is the causative agent of myxomatosis, a fatal disease of the european rabbit. The MV genome is a linear double stranded DNA molecule, which encodes several factors important for the subversion of the immune system of the host. These virulence genes are mainly located within or near the Terminal Inverted Repeats. We cloned and sequenced a right-end region of MV genome and identified a 801 bp Open Reading Frame (ORF) coding for a polypeptide belonging to the serpin superfamily. There are two serpins identified so far on the MV: SERP-1 binds to several targets and is an anti-inflammatory molecule (1); Serp-2 is essential for virus virulence, and has both anti-inflammatory and anti-apoptotic effects (2), which correlates with its putative anti-caspase activity (3). The newly discovered MV-serpin-like protein, temporarily called Serp-3, needs to be characterized biochemically. Insertional inactivation of Serp-3 ORF leads to a significant attenuation of virulence in vivo, measured by an increased survival rate of infected rabbits, and a limited dissemination to secondary sites of infection. The main histopathological features are the presence of so-called secondary myxomas in the lymph nodes draining the inoculation site of rabbits infected with the wild-type MV, whereas there is no secondary myxomas in rabbits infected with a serp3-deletion mutant. Both the wild-type and the serp3-deletion mutant virus replicate at comparable levels in vivo. Serp3, a novel serpin-like protein, may represent a significant virulence factor of MV, likely acting in synergy with other viral proteins. Biochemical analysis are in progress to determine its target (s). References: 1 - Lomas D.A., D.L. Evans, C. Upton, G. McFadden, and R.W. Carrell, 1993. Inhibition of plasmin, urokinase, tissue plasminogen activator, and C1s by a myxoma virus serine proteinase inhibitor. J.Biol.Chem. 268:516-521. 2 - Petit F., S. Bertagnoli, J. Gelfi, F. Fassy, C. Boucraut-Baralon, and A. Milon, 1996. Characterization of a Myxoma virus-encoded serpin-like protein with activity against Interleukin-1ß converting enzyme.
J. Virol. 70: 5860-5866. 3 - Messud-Petit F., J. Gelfi, M. Delverdier, M.-F. Amardeilh, R. Py, G. Sutter, and S. Bertagnoli, 1998. Serp2, an inhibitor of the Interleukin-1ß converting enzyme, is critical in the pathobiology of Myxoma virus. J. Virol. 72: 7830-7839.

COMPARISON OF THE NATIVE AND CLEAVED FOLDING PATHWAYS OF a1-ANTITRYPSIN.
Stephen P. Bottomley and Ellie L. James Department of Biochemistry and Molecular Biology, Monash University, Victoria.

a1-Antitrypsin (a1-AT) the archetypal member of the SERPIN family, functions as an inhibitor of neutrophil elastase. a1-AT displays a highly conserved tertiary structure consisting of 9 a-helices and 3 b-sheets. X-ray crystallographic and biochemical data have shown that a1-AT can adopt the native, cleaved and latent structures. These data demonstrate that areas such as the reactive center loop and A b-sheet are extremely flexible. The cleaved-state, in which the cleaved reactive center loop is inserted into the A b-sheet, affords the protein very different physical characteristics. In particular the cleaved-state is thermodynamically far more stable than the native state. Cleavage of the serpin produces two fragments that remain non-covalently bound. Recently the small C-terminal fragment, termed SPAAT (short peptide from a1-AT) has been identified in human tissue where it is bound to the extracellular matrix. SPAAT also functions as an efficient competitive inhibitor of elastase and cathepsin G. However due to the high stability of the cleaved-state it is unclear how SPAAT could be produced. To begin to understand this problem we have initiated studies examining and comparing the folding and unfolding pathways of both native and cleaved a1-AT. Detailed analysis of the a1-AT unfolding pathway using a variety of spectroscopic approaches has suggested that unfolding is at least three-state, Native Æ Intermediate Æ Unfolded (Native a1-AT) From our biophysical data we were able to demonstrate that unfolding occurs via disruption of the A and C b-sheets followed by the B b-sheet. In contrast the unfolding of cleaved a1-AT occurred in a two-state process with only the native and unfolded states being populated. Native Æ Unfolded (Cleaved a1-AT) Cross-linking and gel-filtration studies showed that SPAAT was not released until extremely high denaturant conditions were reached. Interestingly, we managed to renature the cleaved-state. The structural changes we observed upon refolding were similar to that of the native molecule refolding. The biological implications of these data to serpin structure and function will be presented.

HUMAN PROTEINASE INHIBITOR 6 (PI-6) REQUIRES ARG IN THE REACTIVE CENTRE LOOP TO INHIBIT CATHEPSIN G
Jiuru Sun, Fiona L. Scott, and Phillip I. Bird Department of Medicine, Monash University, Box Hill Hospital, Box Hill, Vic. 3128, Australia

Human proteinase inhibitor 6 (PI6) is an intracellular serpin that uses Arg341 as the reactive site P1 residue to neutralize trypsin-like serine proteinases, or Met340 as the P1 residue to inhibit chymotrypsin. Recently we found that PI6 is a potent cathepsin G inhibitor but did not identify the P1 residue required for this interaction. Mouse SPI-3 is an intracellular serpin with 76% amino acid sequence similarity to human PI-6. It has a P1 Arg and efficiently inhibits trypsin and thrombin. However,it differs from PI-6 in having Val instead of Met in the adjacent upstream position, suggesting it may exhibit a different inhibitory profile against chymotrypsin or cathepsin G. To identify the likely P1 residue in PI-6 required for cathepsin G inhibition, the enzyme-inhibition pattern of the PI-6 and SPI-3 was examined in more detail. Site-directed mutagenesis was used to change the P1 and P2 residues, and the mutants were expressed in the methylotropic yeast Pichia Pastoris. As expected, both PI-6 and SPI-3 formed SDS-stable complexes with thrombin, trypsin, cathepsin G and chymotrypsin. Kinetic analysis revealed that PI-6 is a very efficient inhibitor of trypsin, cathepsin G and chymotrypsin with pseudo first order association rate constants (Kass) at 22 C of 1.7 x 107M-1S-1, 7.6 x 106M-1S-1and stoichiometries of inhibition (SI)1.0, 1.0,1.0 respectively. While SPI-3 inhibits trypsin strongly, it inhibits cathepsin G and chymotrypsin less efficiently than PI-6 with pseudo first order association rate constants of 6.8 x 106 M-1S-1, 2.0 x 105 M-1S -1, 2.3 x 105 M-1S - 1 , and SIs of 1.1, 2.5 and 11.8 respectively. P2 mutation of PI6 (P2 MetrAla or MetrVal) had little effect on the inhibitory activity of PI6 towards trypsin and cathepsin G, however these mutations reduced chymotrypsin-inhibition activity about 20-fold (Kass 3 x 105M-1S-1, 8.2 x 105M-1S-1, SI 8.1 and 6.9). By contrast, a P1ArgrAla mutation completely abolished trypsin and cathepsin G inhibitory activity. These results suggest that the P1 residue for cathepsin G is Arg341 and the P2 residue plays an important role in the inhibition of cathepsin G and chymotrypsin but not trypsin.

MURINE HOMOLOGS OF HUMAN MNEI (MONOCYTE/ NEUTROPHIL ELASTASE INHIBITOR)
Hong Gao, Jessica Cooley, Weilan Zeng and Eileen Remold-O'Donnell, The Center for Blood Research, Harvard Medical School, Boston MA 02115

MNEI (Elastase Inhibitor of Monocytes and Neutrophils) is a potent in vitro inhibitor of the prevalent neutrophil granule proteases, elastase, cathepsin G and proteinase 3, which are major mediators of human inflammatory pathology. MNEI, an ovalbumin-related serpin (Ov-serpin) is encoded in humans by a single copy gene, clustered with PI-6 and PI-9 on chromosome 6p24-ter, one of two Ov-serpin loci. We report here the identification and characterization of the apparent mouse ortholog of human MNEI and describe also additional mouse Ov-serpin gene(s) highly related to human MNEI. Initially, we sought related mouse genes by a Blast search of Genbank dbEST (Expressed Sequence Tags) division. Fifteen clones were identified as highly homologous to human MNEI; several were obtained and sequenced. Comparison revealed one full length and 14 partial clones, all encoding a single mRNA species (Mnei-A), which had been reverse-transcribed and cloned at least nine times independently from thymus, spleen, T-cells, mammary gland, fetal tissue and pooled organs. No other closely related murine cDNA clone was found. In contrast to the simple cDNA findings, Southern blots showed that mice, unlike humans, have more than one genomic band hybridizing with a probe from the Mnei-A exon 7 (reactive center) region. Southern blots with 3 untranslated and an intronic probe also detected 2-3 hybridizing bands, indicating that homology extends to non-coding regions, i.e., that mice have multiple genes that are very similar to human MNEI. A genomic lambda phage library from 129/Sv mice was screened by hybridization with Mnei-A exon 7 labeled with 32P. Ten phage clones were isolated and grouped based on partial sequence of amplified exon 7. Three phage clones were identical to Mnei-A exon 7 cDNA. Five phage clones were identical to each other in partial exon 7 sequence and ~85% identical to Mnei-A. The assignment of five phage clones to the same gene (Mnei-B) was verified by their overlapping BamH1 restriction fragmentation patterns. For the remaining two phage clones, exon 7 sequence could not be obtained with Mnei-A primers. Subcloning and sequencing has generated complete structure for Mnei-A and Mnei-B. Each gene shares the 7 exon 6 intron structure of human MNEI. A third murine gene, Mnei-C is presently represented by a partial clone. It is highly likely that Mnei-A is the ortholog of human MNEI based on highest overall homology (82% identity of cDNA nucleotide sequence; 81% identity of amino acid sequence). Also, the reactive center loop sequence is identical in Mnei-A and human MNEI and different in Mnei-B. Finally, Mnei-A, like human MNEI, appears to be prevalent in hematopoietic cells since it is represented by at least 9 independent clones in cDNA databases, whereas Mnei-B and Mnei-C are not represented, suggesting that their expression may be limited to restricted tissues. Generation of PAI-1 mutants and examination of the modulating function of PAI-1 on the adhesion of breast cancer cells to the endothelium Nuria Arroyo de Prada, Manfred Schmitt, Viktor Magdolen, Olaf G. Wilhelm Dept. of Gynecology and Obstetrics, TU MÅnchen, Germany Plasminogen activator inhibitor type 1 (PAI-1) is a multifunctional 50 kD protein that plays a key role in physiological processes such as fibrinolysis and pathophysiological events like migration and invasiveness of tumor cells. PAI-1 belongs to the serpin superfamily. The major known function of PAI-1 is the inhibition of the serine proteases plasminogen activator of the urokinase-type (uPA) and tissue-type (tPA). A feature shared with a further serpin, PAI-2. Moreover, PAI-1 not only possesses a binding site for uPA/tPA but it also can interact with the extracellular matrix and plasma protein vitronectin, with the glycosaminoglycan heparin and with fibrin, which constitutes the major proteinic component of blood thrombi. The additional binding properties and the spontaneous conversion of PAI-1 into a latent, inactive form confer PAI-1 a unique and special position among serpins. The significant influence of PAI-1 on malignant processes such as adhesion, migration and the degradation of the extracellular matrix by cancer cells is not merely restricted to the inhibition of uPA or tPA; it rather relies in the modulation of the interactions between the urokinase system and the extracellular matrix by PAI-1. Despite this recognition it is presently unknown, which function is played by the additional PAI-1 binding characteristics in carcinogenic processes. Thus, the aim of our work is to elucidate the importance of the binding properties of PAI-1 that are not shared by other plasminogen activator inhibitors in processes such as invasion and metastasis. The binding regions of PAI-1 to its known interacting partners have been roughly characterized. Our present work is focussing on the generation of PAI-1 mutants lacking these regions or including substitutions in one of the different binding domains. The mutations were designed after computer modelling (in cooperation with Dr. Andreas Bergner, Max-Plank-Institut for Biochemistry, Dept. Prof. Dr. W. Bode). The mutational strategy is a single-step method by reverse PCR. These mutants will be characterized in order to determine their binding properties as well as their inhibitory capacity towards uPA. Moreover, active mutants will be tested and compared to wild-type PAI-1 and PAI-2 in adhesion assays. These assays were established in our lab and represent a model in order to test the adhesion capacity of breast cancer cells to an endothelial monolayer composed of HUVEC cells. This process is of critical importance in intra- and extravasation events. The assay enables the assessment of the modulating action of PAI-1 on the adhesion of breast cancer cells.

IMMUNOLOGICAL DETECTION OF CONFORMATIONAL NEOEPITOPES ASSOCIATED WITH THE SERPIN ACTIVITY OF PLASMINOGEN ACTIVATOR INHIBITOR TYPE-2
Darren N. Saunders1, Kathy M.L. Buttigieg1, Alison Gould2, Virginia McPhun 3 and Mark S. Baker1 1Department of Biological Sciences, University of Wollongong, Wollongong 2522, Australia, 2Biotech Australia Pty Ltd, PO Box 20, Roseville 2069, Australia and 3John Curtin School of Medical Research, Division of Cell Biology, Australian National University, Canberra 2601, Australia

The physiological roles of plasminogen activator inhibitor-2 (PAI-2) are not yet well understood. Kinetic studies suggest a role in the regulation of plasminogen activator-driven proteolysis in many cell types. This study describes a monoclonal antibody (2H5), which uniquely recognises neoepitope determinants on PAI-2 appearing after thermodynamic relaxation of the molecule. Enzyme-linked immunosorbent assays and native polyacrylamide gel electrophoresis immunoblotting confirmed the specificity of 2H5 for urokinase type plasminogen activator-PAI-2 complexes. Examination of the affinity of 2H5 for complexes formed between PAI-2 and a synthetic 14 mer reactive site loop peptide, PAI-2 treated with tissue plasminogen activator, or thrombin suggests that the 2H5 epitope is determined exclusively by sequences found only on PAI-2 following proteolytic cleavage of the Arg 380-Thr 381 bond and insertion of the reactive site loop into b-sheet A. Peptides lacking both the P13 (Glu 368) and P14 (Thr 367) residues did not induce a conformational change or affect the inhibitory activity of PAI-2, indicating that one or both of these residues are critical for PAI-2 function. To our knowledge, this is the first description of a monoclonal antibody that can distinguish conformational changes in PAI-2 related specifically to its potential biological function(s).Monoclonal antibodies directed to PAI-1 and PAI-2: Epitope-mapping and analyses of their inhibitory capacity Bernd MÅhlenweg, Nuria Arroyo de Prada, Elke Guthaus*, Niko Schmiedeberg* Olaf G. Wilhem, Horst Kessler*, Manfred Schmitt, Viktor Magdolen Dept. of Gynecology and Obstetrics and Dept. of Org. Chemistry*, TU MÅnchen, Germany The serine protease urokinase-type plasminogen activator (uPA) bound to its cell-surface receptor contributes to the invasive and metastatic capacity of tumor cells. The proteolytic activity of uPA is controlled by the serpins plasminogen activator inhibitor type-1 and 2 (PAI-1, PAI-2). Evaluation of the prognostic significance of both inhibitors in several clinical studies have indicated that high levels of PAI-1 antigen predict poor patient prognosis while high levels of PAI-2 are related to good prognosis. Thus, these inhibitors play different roles in tumor biology. A series of monoclonal antibodies directed to PAI-1 and PAI-2 have been generated in various laboratories. However, most of these antibodies are not well defined with respect to the epitope present in the target protein. Thus, we aimed to identify the epitopes of these mAbs to be able to use the mAbs as highly specific tools for the analysis of the role of these serpins in malignant tumors. Selecting conserved regions between PAI-1 and PAI-2, we created a total of 10 fusion proteins encoding different portions of PAI-1 and PAI-2. All recombinant proteins - including wild type PAI-1 and PAI-2 - were expressed in E. coli with an N-terminally located histidine6-tag and, subsequently, purified under denaturating conditions by Ni2+-chelating chromatography. After renaturation the proteins were coated on ELISA-plates and probed with various mAbs directed to PAI-1 or PAI-2. Binding of the mAbs to the recombinant proteins were detected with peroxidase-conjugated rabbit-anti mouse IgGs. All antibodies reacted with high sensitivity and specificity. In many cases we were able to map the epitopes of the mAbs to regions containing about 80-130 aa using the PAI-1/PAI-2-chimera. However, some antibodies only reacted with the respective wild-type protein indicating that these mAbs are directed to conformational epitopes located in different regions of the protein. The experiments were paralleled by Western blot analyses. To evaluate the precise epitope of PAI-1 mAbs we synthesized 20mer peptides spanning the complete wt-PAI-1-sequence (aa 1 - 381). These peptides overlap with seven amino acids to each side with their neighbouring peptide. Using these peptides as antigen for the mAbs we were able to map a series of non-conformational antibodies to epitopes of about six aa length. In further experiments we investigated the ability of the mAbs to inhibit PAI-1 and PAI-2 activity. After preincubation with the respective mAb the remaining PAI-1/PAI-2 activity was determined. Most mAbs displayed no inhibitory effect, but some mAbs were able to inhibit PAI-1 activity significantly. In conclusion we have developed a rapid screening method for mAbs directed to PAI-1 and PAI-2. Using this method and synthetic peptides we have devised well-defined and epitope-mapped mAbs to PAI-1 and PAI-2. These mAbs will be used in functional assays to further explore the role of both serpins in tumor biology.

BIOCHEMICAL CHARACTERISATION OF SERPIN2A, A NOVEL INTRACELLULAR PROTEIN EXPRESSED IN HAEMOPOIETIC CELLS.
Emma C Morris1, Timothy Dafforn1, Lynne Hampson2, Ian N Hampson2 and Paul B Coughlin1 1Department of Structural Medicine, University of Cambridge UK. 2CRC Department of Experimental Haematology, Paterson Institute for Cancer Research, Manchester UK

In recent years it has become apparent that a class of serpins exists which are retained, at least in part, within the cell. These proteins lack a cleavable secretion signal sequence, have oxidisable residues close to the reactive, are truncated at the C-terminus and have conserved genomic structures. Several human members of this group have been described in two chromosomal clusters at 6p25 and 18q21.3. We have studied the murine haemopoietic serpin2A and found that it has some features in common with other intracellular serpins including in that it lacks a cleavable secretion signal, a predicted P1-P1' sequence of Cys-Cys but by contrast it has a C-terminal extension containing two further cysteine residues and is most closely related to human antichymotrypsin. Serpin2A was originally identified as a gene expressed in primitive haemopoietic cells and down regulated upon differentiation. It is also expressed in lymphoid cells upon T-cell activation. We have performed immunofluorescence and laser confocal scanning microscopy studies and shown that the serpin is present within the cytosol. Furthermore, a significant proportion of cells show prominent nuclear localisation. This pattern of expression and cellular localisation suggest that serpin2A plays a role in regulating important events in cell growth and/or apoptosis and it is also the first member of a new class of intracellular serpin. In order to understand the molecular events underlying the biological role of serpin2A we have produced a recombinant protein and studied its biochemical properties. A 6xHis-tagged protein using the baculovirus expression system and purified by chromatography on NTA-Agarose. The key question to be answered was whether serpin2A could function as a protease inhibitor in vivo. Examination of its deduced amino acid sequence showed that the proximal hinge region of the reactive site loop was typical of inhibitory serpins. This prediction was tested by cleaving the reactive site loop with trypsin and chymotrypsin. This caused S->R transition as shown by a typical change in mobility on transverse urea gradient gels and is consistent with potential for inhibitory activity. Corroborative circular dichroism studies of stability during thermal denaturation are in progress. Although serpin2A has features of a protease inhibitor its reactive site does not allow us to predict a target protease. We have therefore tested its ability to form an inhibitory complex with a panel of serine and cysteine proteases. So far we have not been able to detect inhibitory activity although the serpin appears to a substrate for several of these proteases. The unique Cys-Cys motif at the serpin2A reactive site together with its cysteine containing C-terminal region suggest that the biological activity of the protein will be sensitive to redox conditions. Molecular modelling suggests that the C-terminus could interact with and modulate the availability of the reactive site. Preliminary studies in which the protein has been cleaved by plasmin then analysed by reduced and non-reduced SDS-PAGE suggest that the C-terminus is linked to the body of the molecule by a disulphide bond. We propose that the C-terminal extension of the protein may be involved not only in modulating the availability of the reactive site but also in localisation within the cell and also that both of these activities may be modified by cellular redox conditions. Our ongoing studies are directed at identifying a physiological target protease for serpin2A and elucidating the way that its unusual biochemical properties may contribute to roles in regulating important molecular processes in cell growth or apoptosis. Identification of an unusual serpin-like gene From activated human t cells. M.A. Sharkey, D.M. Worrall. Department of Biochemistry, University College Dublin, Belfield, Dublin 4, Ireland. The serpin superfamily has expanded enormously in recent years and there are undoubtedly more as yet undiscovered members. Current and future genome and expressed sequence tag (EST) sequencing projects are likely to be the major source of information for the identification of new serpin genes. ESTs are short, single pass cDNA sequences generated from randomly selected library clones. Examination of EST databases yielded a number of fragments homologous to serpins, including a human gene expressed in activated T cells. This 369bp EST, which showed highest similarity to the ovalbumin-type serpin PI 6, was used to design PCR primers, and RT-PCR established that this transcript was being produced in HeLa cells. Using a PCR-based strategy (Israel, D.I., Nucleic Acids Res., 1993; 21(11): 2627-2631), we have screened a HeLa cell lambda cDNA library and have isolated a clone which may contain the open reading frame of a novel protein. The library clone insert has been fully sequenced and is approximately 1,700 bp in length. It includes an open reading frame of 291bp which would give rise to a protein with a predicted Mr of 10.5kDa. Although this clearly couldn't be a full length serpin, the amino terminal half of the predicted protein product is similar in sequence to the carboxy terminal region of serpins. It has an amino acid tract which could constitute a reactive site loop with a predicted P1-P1' of Arg-Thr, which is similar to PAI-2. Studies to express and evaluate the inhibitory potential of the recombinant protein are currently underway.

THE EFFECT OF REACTIVE CENTER LOOP INSERTION ON THE KINETICS OF BINDING OF PLASMINOGEN ACTIVATOR INHIBITOR-1 TO tPA.
Ingrid M. Verhamme, Jan-Olov Kvassman, Herbert R. Halvorson, Daniel A. Lawrence# and Joseph D. Shore Henry Ford Health Sciences Center, Division of Biochemical Research, Detroit, MI., 48202-2689. #American Red Cross Holland Laboratory, Department of Biochemistry, Rockville, MD., 20855.

It was recently demonstrated that when loop insertion is blocked in PAI-1, it acylates tPA more than 50 times slower ( ~0.06 s-1) than the unmodified inhibitor ( ~3 s-1), and reacts as a substrate (1). With tPA, the steps beyond Michaelis complex formation are limited by the removal of the distal peptide fragment from the enzyme substrate pocket. This step is slow if loop insertion is prevented but becomes much faster in the inhibitory mechanism in which loop insertion effectively separates the acylated enzyme from the newly formed N-terminus. Effects of viscosity on the PAI-1 binding kinetics should make it possible to distinguish whether the peptide inserted in sheet A to block loop insertion affects the bimolecular binding step, the subsequent bimolecular reaction steps or both, and to quantitate these effects. A characteristically different response to viscosity is expected for binding of inhibitory and substrate PAI-1 to tPA: based on the results cited above they should be distinguished as 'sticky' and 'non-sticky' ligand, respectively. To test these predictions we generated two forms of PAI-1 in which the loop was constrained in two different ways, a) by an octapeptide bound to position 4 in b-sheet A and b) by the replacement of the threonine in the RCL hinge region by an arginine. The two constrained forms were completely hydrolyzed by catalytic amounts of the tissue type plasminogen activator (tPA) with only small differences in kcat and Km. The respective Km values, 20 and 32 nM, are comparable to the Kd reported for binding of unmodified PAI-1 to the inactive Ser195Ala mutant tPA. RCL insertion in inhibitory and loop-constrained PAI-1 was measured directly, by using P9Cys PAI-1 labeled with the fluorescence probe NBD at position P9 (2). Consistent with their substrate behavior, the constrained forms exhibited rates of RCL insertion at least 3000 times slower than unmodified PAI-1. If enzyme acylation is irreversible, the overall rate of reaction of PAI-1 with proteinase should be independent of loop insertion. We found, however, that the RCL constrained PAI-1 forms exhibited a five fold smaller apparent second order rate constant for binding to tPA than inhibitory PAI-1. To resolve this ambiguity we studied the effects of viscosity on the two binding reactions. The rate of formation of the association complex was the same for inhibitory PAI-1 and both substrate forms. However, inhibitory PAI-1 binding exhibited a linear dependence on the reciprocal relative viscosity, characteristic of 'sticky binding', whereas binding of the substrate forms was distinctly hyperbolic, indicating 'non-sticky' behavior. The 'sticky' behavior of inhibitory PAI-1 is consistent with the linear dependence of the pseudo first order rate constant for inhibitor binding on PAI-1 concentration (0.03 - 12 M) observed in competitive p-aminobenzamidine displacement from the tPA active site. The results show that the Michaelis complex formed with inhibitory PAI-1 is stabilized much faster than it dissociates whereas this relation is reversed in the reaction with loop constrained PAI-1. The identical 'non-sticky' binding behavior of two substrate PAI-1 forms in which loop constraint was generated differently, argues against a nonspecific viscosity effect, and is clearly consistent with the difference in the acylation rate constant observed in inhibitory and substrate PAI-1.
(1) J.-O. Kvassman, I. Verhamme and J.D. Shore, Biochemistry 37, 15491-502, 1998.
(2) J.D. Shore, D.E. Day, A.M. Francis-Chmura, I. Verhamme, J. Kvassman, D.A. Lawrence and D. Ginsburg, J. Biol. Chem. 270, 5395-5398, 1995.

THE ROLE OF PLASMINOGEN ACTIVATOR INHIBITOR 2 (PAI-2) IN APOPTOSIS OF HUMAN PERIPHERAL BLOOD MONOCYTES
Alex Fragoyannis, Nuala A Booth, Helen Ritchie. Department of Molecular & Cell Biology, University of Aberdeen, Institute of Medical Sciences, Aberdeen AB25 2ZD

Plasminogen activator inhibitor 2 (PAI-2) is a member of the ov-serpin sub-family, and functions as an inhibitor of plasmin generation. PAI-2 is synthesised by activated monocytes, where PAI-2 is secreted and is also maintained as an intracellular protein. The intracellular pool of PAI-2 has been suggested to have an additional role to that of inhibition of plasminogen activation. Intracellular PAI-2 has been shown to inhibit apoptosis of transfected cells. Here, we have investigated the role of PAI-2 in apoptosis of human monocytes. Monocytes were isolated from human peripheral blood by centrifugation on Ficoll-Paque and subsequent adherence to plastic. Apoptosis was induced either by removal of serum from the culture medium, or by addition of 20 mM hydrogen peroxide (H2O2) to cells. Apoptosis was assessed using the terminal transferase dUTP nick end labelling (TUNEL) method, in conjunction with morphological changes and nuclear staining using DAPI. Apoptotic nuclei were expressed as a percentage of total nuclei. Monocytes were either untreated or treated with lipopolysaccharide (LPS), thrombin or phorbol myristate acetate (PMA) prior to induction of apoptosis. All these agents activate monocytes and upregulated synthesis of various proteins, including PAI-2. Both secreted and intracellular forms of PAI-2 were upregulated in response to LPS (100 ng/ml), thrombin (1 U/ml) or PMA (100 nM) following an overnight incubation. Intracellular PAI-2 antigen was increased significantly (p<0.01) to ~1400, 800, and 500 ng/mg total cellular protein in PMA, LPS and thrombin treated monocytes, respectively compared to untreated cells (~100 ng/mg). Apoptosis occurred rapidly in untreated monocytes, and a significant increase was detectable in apoptotic cells at 1 h (p<0.01), compared to control cells. The percentage of apoptotic nuclei ranged from 20% (1 h following induction of apoptosis) to 60-80% within 4 h of apoptosis. Caspase-3 activity was increased following induction of apoptosis. Monocytes that had been stimulated with LPS, thrombin or PMA were protected from apoptosis. The percentage of apoptotic nuclei in monocytes that had been treated with LPS, thrombin or PMA was 10%, 30% and 10% respectively at 4 h. This represented a significant decrease in apoptosis (p<0.01) compared to untreated monocytes (60-80% apoptotic nuclei). Antisense technology was employed to investigate a possible protective effect of PAI-2 on apoptosis. An antisense oligonucleotide that was directed to the start of PAI-2 mRNA, and a control scrambled oligonucleotide were successfully transfected into human peripheral blood monocytes. The antisense oligonucleotide blocked translation of PAI-2 mRNA, and PAI-2 antigen levels following LPS stimulation were decreased by 60-80 %, compared to monocytes transfected with a scrambled control oligonucleotide. Inhibition of PAI-2 using antisense technology had no effect on apoptosis of monocytes, which demonstrated that intracellular PAI-2 does not interact with the apoptotic pathways induced by serum removal and hydrogen peroxide. Inhibition of alternative apoptotic pathways in monocytes by PAI-2 is currently being investigated. The mechanisms for the protective effects of thrombin, LPS and PMA on apoptosis of human monocytes are also being studied. INtracellular Plasminogen Activator Inhibitor type-2: Function and gene regulation Toni M. Antalis, Andreas Suhrbier, Vicki Mann, Brett Stringer and Steven Ogbourne The Queensland Institute of Medical Research, Post Office Royal Brisbane Hospital, Brisbane 4029 Queensland, Australia Plasminogen activator inhibitor type-2 (PAI-2) was one of the first identified members of a unique and growing subclass of serpins, called ovalbumin-like serpins, which lack a typical amino terminal sequence and exists in both cytosolic and secreted forms. While PAI-2 is well characterised as an inhibitor of the extracellular serine proteinase, urokinase-type plasminogen activator, our data suggests an additional and distinct intracellular function for PAI-2 as a regulator of intracellular signalling pathways. PAI-2 expression is restricted to a small number of cell types but may be strongly induced as a consequence of cell activation, cell differentiation and under certain pathological conditions. We are investigating both the function of intracellular PAI-2 and mechanisms involved in the regulation of the PAI-2 gene. Cytosolic PAI-2 expression in HeLa cells is associated with a range of phenotypes, including resistance to TNFa mediated apoptosis, resistance to lytic infection by certain cytopathic viruses and altered sensitivity to certain chemotherapeutic drugs. Molecular changes associated with intracellular PAI-2 activity include mitochondrial changes and altered gene expression. These data suggest that intracellular PAI-2 expression influences cell signalling, leading to the hypothesis that the levels of certain transcriptional regulatory proteins might be regulated by specific cytoplasmic proteinases, which may in turn be controlled by proteinase inhibitors such as PAI-2. The PAI-2 gene is regulated by mechanisms which both suppress and activate PAI-2 gene transcription. Our investigations using DNase I hypersensitivity analysis, interspecies promoter comparisons, functional reporter gene assays and gel shift studies have identified several elements inportant for PAI-2 gene regulation. Comparisons of the human and mouse PAI-2 promoters shows that they are highly conserved, suggesting that the mouse and human genes are regulated similarly. We have identified an upstream silencer (PAUSE-1) functions that suppresses gene transcription in a position, orientation and promoter specific manner. UV cross-linking experiments demonstrate a single PAUSE-1 binding protein (PAUSE-1 BP) of approximately 67kD. Gel shift analyses show that unlike PAI-2 expression, the PAUSE-1 BP is ubiquitously expressed. Given the extensive range of expression and the fact that other silencers similar to PAUSE-1 have been identified in other genes, PAUSE-1 BP is likely to be a key regulator of gene transcription. A positive regulatory element upstream of the PAUSE-1 element which overcomes silencer-mediated repression in a cell specific manner has been localised to a 330 bp upstream region which is rich in potential transcription factor binding sites. Identification and characterisation of these elements will provide insights into the molecular basis for the regulation of PAI-2 gene transcription.

CO-EXPRESSION OF HEPARAN SULFATE PROTEOGLYCANS AND SERPINS IN THE OVARIAN FOLLICLE
M. Princivalle, S. Hasan, G. Hosseini and A. de Agostini Clinic of Infertility and Gynaecological Endocrinology, Department of Gynaecology and Obstetrics, Geneva University Hospital, 1211 Geneva 14, Switzerland

In mammals, ovulation involves the breakdown of the follicular wall and an inflammatory reaction with vascular permeabilization and subsequent fibrin deposition. Activated serine proteases from the plasminogen activator and coagulation cascades are controlled by serpins present in the ovary such as plasminogen activator inhibitor-1 (PAI-1), protease nexin-1 (PN-1) and antithrombin III (AT). Anticoagulant heparan sulfate proteoglycans that bind and activate AT (aHSPG) are synthesized by rat ovarian granulosa cells (Hosseini et al. J.Biol.Chem. 271:22090,1996) and we aim at the elucidation of their role in ovarian physiology. To identify the serpins potentially interacting with aHSPG have localized aHSPG, AT, PN-1 and PAI-1 in serial sections of rat ovaries taken at different stages of stimulated and natural cycles. Localization of aHSPG in rat ovary cryosections was determined by specific 125I-AT-binding followed by autoradiography. During follicular development, aHSPG are undetectable in follicles from the primary stage to the preantral stage. In antral follicles, a strong labeling is seen on granulosa cells while the oocyte, theca layers and the follicular basement membrane are not labeled. Maximum signal for aHSPG was observed at ovulation, followed by a decrease in postovulatory follicles. Thogether with previous in vitro data, these observations demonstrate that FSH determines the onset of aHSPG expression in granulosa cells. Serpins were localized in the ovary by immunoperoxidase labelling. AT was absent from granulosa cells of preantral and early antral follicles. In large antral follicles, a diffuse staining appeared in the peri-antral region and extended to the entire granulosa cell layers in the periovulatory period. Luteal cells are weakly labeled, and theca layers remain negative at all stages. Thus, AT seems to be synthesized locally in follicles, as confirmed by Northern blot analysis, where AT mRNA was found in the ovary in amounts comparable to the liver and the kidney. This is the first observation of local expression of AT in the ovary, suggesting its involvement in controlling proteolytic activities outside of the vascular bed. PAI-1 and PN-1 were localized in the ovary in agreement with published data (Liu et al. Eur.J.Biochem. 195:549,1991; Hagglund et al. Endocr. 137:5671,1996). aHSPG, AT, PN-1 and PAI-1 were co-localized in ovary serial cryosections. In antral follicles, aHSPG is present on granulosa cell layers of antral follicles, together with PN-1 and AT, while PAI-1 does not colocalize with aHSPG. During follicular development, PN-1 appears first in secondary follicles, aHSPG appears in early antral follicles and AT appears in preovulatory follicles. aHSPG, PN-1 and AT are co-localized in preovulatory and ovulatory follicles. After ovulation, aHSPG and AT co-localize in forming corpus luteum, while PN-1 disappears. In corpus luteum, aHSPG are present together with PAI-1. These data show that aHSPG are co-localized first with PN-1 in early antral follicles, and in the periovulatory period with both AT and PN-1; after ovulation, aHSPG are colocalized with AT and later, in mature corpus luteum, with PAI-1. Thus, aHSPG localization in the ovary is adequate to interact with PN-1 and AT at the time of ovulation. As the localization of HSPG depends from the nature of their core protein, we have analyzed the expression in rat ovary of the HSPG membrane-spanning core proteins syndecan and ryudocan and of the basement membrane core protein perlecan. We have detected expression of syndecan and perlecan in rat ovary by Northern blot analysis. These data indicate that aHSPG localized on granulosa cells could be attached to syndecan-1 core protein. In conclusion, we postulate that granulosa cell aHSPG are mainly present on syndecan-1 core protein and participate in the regulation of serine protease activity at ovulation by localizing the serpins AT and PN-1 in activated forms on the granulosa cell layer.

THE MYXOMA VIRUS SERP2 PROTEIN INHIBITS ICE AND GRANZYME B WEAKLY IN VITRO, AND IS NOT ABLE TO INHIBIT APOPTOSIS IN COWPOX VIRUS INFECTED CELLS, UNLIKE CRMA
Peter C. Turner1, M. Carmen Sancho1+, S.R. Thoennes1, A. Caputo2, R.C. Bleackley2, and Richard W. Moyer1* 1Dept. of Molecular Genetics and Microbiology, University of Florida, Gainesville, FL 32610-0266 2Biochemistry Department, University of Alberta, Edmonton, Alberta T6G 2H7, Canada +present address: Centro de Investigaciones Biol¢gicas (CSIC), 28006 Madrid, Spain

The serp2 protein of the leporipoxvirus myxoma virus, like the cowpox virus serpin crmA, has Asp at the P1 position in the reactive site loop (RSL) and has been reported to inhibit the cysteine proteinase ICE (caspase-1) in vitro, as does crmA. However, serp2 and crmA proteins share only 35% identity overall. crmA forms a stable complex with ICE that can be detected by native PAGE. Following crosslinking with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), the complex between crmA and the p20 subunit of ICE was visible by SDS-PAGE. Radiolabeled serp2 protein was synthesized by coupled transcription-translation in vitro. Serp2 protein was unable to form a detectable complex with ICE that could be visualized by native PAGE. Crosslinking likewise did not allow us to visualize a complex between ICE and serp2 by SDS-PAGE. Consistent with these observations, purified His10-tagged serp2 protein inhibited ICE much more weakly (Ki = 80 nM) than did crmA (Ki = 4 pM). Serp2 like crmA was able to form a stable complex with the serine proteinase granzyme B that was visible by denaturing PAGE, but inhibition of granzyme B by serp2 (Ki = 420 nM) was less effective than by crmA (Ki = 100 nM). Purified serp2 protein was infeffective as an inhibitor of human caspases- 2, -3, -4, -5, -6, -7, -8 or -9. Serp2 protein has been shown [F. Messud-Petit et al. (1998) J.Virol. 72:7830-7839] to be essential for full virulence of wild type myxoma virus in rabbits. We have found that the serp2 nucleotide sequence is identical for the T1 and Lausanne strains of myxoma virus, indicating conservation. The serp2 protein was tested for its ability to inhibit apoptosis by using a system involving cowpox virus (CPV) and LLC-PK1 pig kidney cells. Wild type CPV does not cause apoptosis in infected LLC-PK1 cells, but a CPV mutant deleted for the crmA gene does trigger apoptosis on infection as indicated by biochemical and morphological criteria [C.A. Ray and D.J. Pickup (1996) Virology 217:384-391; J. Macen et al., (1998) J. Virol. 72:3524-3533]. A CPV recombinant expressing serp2 and deleted for crmA was tested for the induction of apoptosis in LLC-PK1 cells. The recombinant gave full apoptosis as judged by DAPI staining of nuclei and by induction of cleavage activity against the caspase-3 substrate DEVD-AMC. Serp2 is therefore unable to substitute for crmA in terms of inhibiting apoptosis in CPV-infected cells. Serp2 may inhibit rabbit ICE and granzyme B more effectively than the human enzymes. However, the inhibition spectra of serp2 and crmA in vitro are clearly distinct, and serp2 is not functionally equivalent to crmA in cowpox virus infections. The natural proteinase target(s) for serp2 remains to be discovered, and appears to be different from the targets for crmA.

SERPIN EXPRESSION IN MALIGNANT HUMAN TISSUES AND IN GYNECOLOGIC CANCER CELL LINES.
Paula M. Boucher, Diane Palmieri and Frank C. Church. Department of Pathology and Laboratory Medicine, School of Medicine, University of North Carolina, Chapel Hill, NC, 27599-7035, U.S.A.

"Tumor-associated proteolysis" has a vital role in the basic mechanism of invasion and metastasis of most cancers. Not only the tumor cells produce one of these systems, the plasminogen activators and their inhibitors, but also the surrounding inflammatory cells, endothelial cells and fibroblasts. Plasminogen activator inhibitors (PAI-1, PAI-2 and PAI-3) are well known for their role in hemostasis and in the regulation of both tissue-type and urokinase-type plasminogen activators. However, increasing evidence supports a role for PAIs in cancer invasion and metastasis. Increased expression of PAI-1 was found to be a poor prognostic indicator for overall survival in breast cancer. We hypothesize that a "balance" of serpin expression exists in normal cells, and that the inappropriate expression of PAIs may contribute to the metastatic potential of cancer cells. To assess expression of PAI-1, PAI-2 and PAI-3 (also named Protein C Inhibitor) in malignant human tissue, a cDNA panel of human tumors propagated in nude mice was examined by PCR. PAI-1 mRNA expression was seen in prostate, pancreas, ovary and one each of two samples from lung and colon tumors. Expression of PAI-2 mRNA was seen only in the pancreatic tumor sample. mRNA levels for PAI-3 were found in the pancreatic tumor, and expressed to a lesser extent in lung and colon tumor samples. These results were compared to mRNA expression from Northern blots containing 23 normal human tissues. PAI-1 was not expressed in normal colon or pancreas as compared to tumor samples from these tissues and PAI-3 mRNA was not seen in normal colon or lung as compared to the tumor samples. We examined the expression of PAIs in gynecologic cancers. We examined four breast, six ovarian, four cervical and three endometrial cancer cell lines for PAI-1, PAI-2, and PAI-3 mRNA expression via rt-PCR. Of the four breast cancer lines examined, PAI-1 mRNA expression was seen only in the highly invasive MDA-MB-231 cells. PAI-2 expression was seen in the MDA-MB-231 cells and the highly invasive MDA-MB-435 cell line. However, PAI-3 expression was seen only in the mildly invasive MCF7 cells. In the ovarian cell lines examined, all six expressed some degree of PAI-1 mRNA. PAI-2 message was detectable in the 420, 429 and CAOV-3 cell lines, all of which are non-tumorigenic. PAI-3 was detectable in the three non-tumorigenic cell lines as well as the highly tumorigenic 194 and SKOV-3 cell lines. PAI-1 and PAI-2 were expressed in two of four cervical cancer cell lines respectively, while PAI-3 was not visible in any of the cervical cell lines examined. No mRNA expression of the PAIs was detectable in the three endometrial cancer cell lines examined. We also assessed the expression of the tumor suppressor serpin maspin in the above tissues and cell lines. Interestingly and in contrast to the PAIs, maspin mRNA was expressed in all the human tumor tissues from the tumor panel. In the four breast cancer lines, maspin mRNA expression was seen in MCF7 and normal human mammary epithelial cells, but not in the MDA-MB-231 and MDA-MB-435 cell lines. Maspin message was detectable in both non-tumorigenic and highly tumorigenic ovarian cell lines. Maspin mRNA expression was detected in some of the cervical and endometrial cell lines. Our results show varying expression of PAIs and maspin between normal and malignant tissues and in different cancer cell lines, which supports the hypothesis that the balance of serpin expression might influence the malignant potential of cancer cells. expression of barley serpin genes, cloning of wheat serpins, and a loop deletion in arabidopsis serpin sequences Soren K. Rasmussena, Thomas H. Robertsa,b, Henrik O. Jensena,b, Lars Hellgrena and Jorn Hejgaardb aPlant Biology and Biogeochemistry Department, PBK-301, Riso National Laboratory, DK-4000 Roskilde, Denmark. bDepartment of Biochemistry and Nutrition, Technical University of Denmark, DK-2800 Lyngby, Denmark. We are studying the properties of plant serpins, the expression of their genes and the identity of their target proteinases. Cereal grain serpins are potent inhibitors of proteinases of the chymotrypsin family, forming SDS-stable complexes1. Recombinant barley serpin Zx (rZx) was tested against various blood coagulation factors and found to have an inhibitory specificity similar to that of antithrombin III2. Poly(A)+ RNA from genes encoding barley serpins Z4, Z7 and Zx has been isolated from tissues throughout development. RT-PCR indicated that transcript corresponding to each serpin was found in endosperm and embryos, with expression of Z7 genes appearing to be biphasic with seed development. Z7 transcript was present in trace amounts in roots and was undetectable in leaves, while Z4 and Zx mRNA were present in both these tissues. Northern dot blots indicated that the levels of Z7 and Z4 mRNA in grains were much higher than those in the vegetative tissues and much higher than the level of Zx mRNA in any tissue. Putative serpins have been detected in roots and leaves, and immunomicroscopy is being conducted to localize plant serpins within tissues. Five Z-type serpin cDNAs have been cloned from wheat. These serpins share 80-94% amino acid sequence identity but vary at several positions within the RSL. We will present kinetic data describing the interaction between these serpins and a range of proteinases including the few purified plant serine- and cysteine proteinases available. Five of the six Arabidopsis thaliana putative serpin gene sequences in EMBL/GenBank encode proteins with an amino acid sequence deletion of up to 23 amino acids corresponding to the absence of the surface loop linking helix I1 to strand 5A. Each of the Arabidopsis putative serpins has one or two cysteines in the predicted P1-P2' positions. We plan to clone one or more Arabidopsis serpins and study their properties.
1Dahl SW, Rasmussen SK, Hejgaard J (1996). Heterologous expression of three plant serpins with distinct inhibitory specificities. J Biol Chem 271: 25083-25088
2Dahl SW, Rasmussen SK, Petersen LC, Hejgaard J (1996). Inhibition of coagulation factors by recombinant barley serpin BSZx. FEBS Lett 394: 165-168

BASIC FIBROBLAST GROWTH FACTOR, UROKINASE-TYPE PLASMINOGEN ACTIVATOR, UROKINASE-TYPE PLASMINOGEN ACTIVATOR RECEPTOR, AND PLASMINOGEN ACTIVATOR INHIBITOR-1 IN SMALL CELL CARCINOMA OF LUNG AND NON-SMALL CELL CARCINOMA OF THE LUNG.
Takeshi Wajima, and Daniel D. Von Hoff, MD, Texas A&M University Health Science Center and Central Texas Veterans Health Care System, Temple, and Institute for Drug Development, University of Texas Health Science Center at San Antonio, Texas, USA

Small cell carcinoma of the lung (SCCL) is biologically different from non-small cell carcinoma of the lung (NSCCL). SCCL grows rapidly, responds to chemotherapy, and is more dependet on the coagulation mechanism than NSCCL. Increased levels of basic fibroblast growth factor (b-FGF) have been detected in urine of patients with lung cancer. B-FGF plays an important role in angiogenesis and tumor growth. When b-FGF is released from the basement membrane, it may induce messenger RNA for urokinase-type plasminogen activator (UPA). The UPA binds to urokinase-type plasminogen activator receptor (UPAR) and converts plasminogen to plasmin. The UPA is inhibited by plasminogen activator inhibitor-I (PAI-1), possibly also by PAI-2. B-FGF, UPA, UPAR, and PAI-1 are present in the lung cancer tissues. UPA, UPAR and PAI-1 are involved in the degradation of the extracellular tumor matrix and have been implicated in cell proliferation, invasion, and metastasis. The aim of this study is to determine whether b-FGF, UPA, UPAR and PAI-1 levels differ between SCCL and NSCCL, and to determine whether increased levels of b-FGF, UPA, UPAR and PAI-1 in the lung cancer cells relate to tumor colony formation in vitro. Twenty-five lung cancer specimen (7 SCCL, and 18 NSCCL) were processed according to the stem cell assay techniques (Cancer Res, 1984;43:1926) and single cell suspensions were made. Single cells suspensions were washed in phosphate buffered saline, resuspended and incubated in a hypotonic medium. This was followed by disintegration with polytran, freezing and thawing. After ultracentrifugation supernatant was collected. Antigen levels of b-FGF, UPA, UPAR and PAI-1 in extracts of single cell suspensions were measured by EISA method. A mixture of single cell suspension and double-enriched CMRL were incubated on agar plates for 14 days. A "glow" specimen was defined > 20 colonies per plate. Three of 7 SCCL were "grow". Eleven of 18 NSCCL were "grow" Remained were "no grow". There was no significant difference between SCCL and NSCCL in antigen levels of b-FGF, UPA, UPAR, and PAI-1. However, UPA and PAI-1 levels tended to be higher in NSCCL. The "grow" specimen had significantly higher levels of b-FGF than "no grow" group. Antigen levels of UPA, and UPAR were higher in "grow" specimens than those in "no grow". The PAI-1 levels were similar in both the "grow" and "no grow" specimen. Higher levels of b-FGF are associated with higher levels of UPA and UPAR. Lung cancer cells with high intracellular levels of b-FGF, UPA and UPAR appeared to have more growth potential in vitro.

MODULATION OF PAI-1 ACTIVITY TROUGH INDUCED CONFORMATIONAL TRANSITIONS IS DEPENDENT ON GLYCOSYLATION
A. Gils, E. Brouwers, I. Knockaert and P.J. Declerck Laboratory for Pharmaceutical Biology and Phytopharmacology, Faculty of Pharmaceutical Sciences, Katholieke Universiteit Leuven, Belgium

Plasminogen activator inhibitor-1 (PAI-1) is a unique member of the serpin superfamily because of its conformational and functional flexibility. PAI-1 is synthezised as a labile active protein that converts spontaneously to a latent conformation. A non-inhibitory stable substrate conformation of PAI-1 has been identified. However, Triton X-100 (TX-100), a non-ionic detergent, can induce in non-glycosylated E.coli PAI-1 a non-stable substrate-like conformation. This non-stable substrate-like conformation is an intermediate in the two consecutive first-order transitions (i.e. active substrate latent) that occur when non-glycosylated active PAI-1 is incubated in the presence of TX-100 at 37¯C (1). Different systems have been used to express PAI-1 e.g. CHO-cells, E.coli cells, HT 1080 cells,¨ To date no differences have been observed between glycosylated and non-glycosylated PAI-1. In the present study, we have investigated the effect of glycosylation on the inactivation of PAI-1 by TX-100 and on the associated pathways of conformational transitions. Whereas in the absence of TX-100, the observed PAI-1 stability was independent on the source of PAI-1, the addition of TX-100 revealed major glycosylation-dependent differences in the inactivation process of PAI-1. Incubation at 0¯C in the presence of TX-100 resulted in a conversion of the active conformation to a substrate-like conformation for all PAI-1's examined. The rate (k) of this conversion was 0.00016 s-1, 0.00121 s-1 and 0.00024 s-1 for non-glycosylated PAI-1, CHO-PAI-1 and HT 1080-PAI-1, respectively. As mentioned above incubation of non-glycosylated PAI-1 with 0.2% TX-100 at 37¯C results in two consecutive conformational transitions (with k1 = 0.029 s-1 and k2 = 0.011 s-1) ultimately resulting in a complete conversion to the latent conformation. However, incubation of recombinant CHO-PAI-1 under these conditions yielded significantly different pathways of conformational transitions i.e. a rapid conversion (k>0.035 s-1) of part (39%) of the active conformation into a stable substrate conformation and a slower conversion (k=0.0004 s-1) of the remaining part (61%) of the active conformation into the latent conformation. In contrast, time-dependent analysis of the conformational transitions in HT 1080-PAI-1 at 37¯C in the presence of TX-100 revealed a simultaneous conversion (k=0.0002 s-1) of the active conformation into both a stable substrate and a stable latent conformation. In conclusion, this is the first report describing significant differences between glycosylated and non-glycosylated PAI-1. Both the rate and the pathways of induced conformational transitions depend strongly on the glycosylation. Based on the 3D-structure, we hypothesize that the glycosylation of residue 329 located on s5A may play a significant role in these conformational transitions.
(1) Gils, A. and Declerck, P.J. (1998) Modulation of plasminogen activator inhibitor 1 by Triton X-100 - Identification of two consecutive conformational transitions. Thromb.Haemost., 80, 286-291. c

DIFFERENTIAL INHIBITIONS OF THROMBIN AND FACTOR Xa BY ANTITHROMBIN SER380 MUTANTS.
Hiroko Shirotani, Yoshiko Maruki and Takehiko Koide Dept of Life Science, Faculty of Science, Himeji Institute of Technology, Hyogo 678-1297, Japan

Antithrombin is a plasma serpin which functions as a regulator of blood coagulation by mainly inhibiting thrombin and factor Xa. However, to be a fast reactive inhibitor, antithrombin requires heparin (or heparan sulfate) as a cofactor. This is considered to be due to a unique form of the reactive site loop which is partially inserted into the A-b-sheet of the molecule. To study a role of Ser380 located at the bottom of insertion (P14 position) of the reactive site loop in antithrombin, we expressed 12 mutants (Gly-, Ala-, Thr-, Pro-, Val-, Ile-, Leu-, Tyr-, Asp-, Asn-, His-, and Arg-mutants) of Ser380 in BHK cells and examined their secretions from BHK cells and their abilities to form complexes with thrombin (T-AT) and factor Xa (Xa-AT) in the presence and absence of heparin. All of the 12 mutants of Ser380 were secreted at a normal level from BHK cells. After 5-min incubation with thrombin or factor Xa in the absence of heparin, the order of T-AT formation among these mutants was Tyr,Gly>>Wt>Asn,Ile>Ala,Pro,Thr>Leu,Arg>>Asp,His,Val. The Tyrand Gly-mutants showed 170% T-AT formation, taking the value of wild-type antithrombin (Wt) as 100%, while the Asp-, Valand His-mutants formed little (<8%) T-AT. The last three mutants, however, showed about 50%, 40% and 25% Xa-AT formation, respectively, giving the order of Xa-AT formation as follows: Wt>Tyr>Pro,Gly>Asp,Ile,Val>His>Asn,Ala,Arg,Thr,Leu. After 60- or 120-min incubation, the order of T-AT formation became Tyr,Gly>Thr,Wt,Ile,Leu>Val,Arg>Ala>> Asn,Pro>>Asp,His, suggesting that the Gly- and Tyr-mutants are faster reactive and more potent thrombin inhibitors than Wt, the Val-mutant is a slow reactive but potent thrombin inhibitor, and the Asn-mutant is a fast reactive but weak thrombin inhibitor. The order of Xa-AT formation after 60- or 120-min incubation was Gly>Wt>Tyr>>Thr>Ile>Asp, Ala,Val>Pro>Arg, His, Leu,Asn, suggesting that the Thr-mutant is a slow reactive Xa inhibitor, and the Pro-mutant is a fast reactive but poor Xa inhibitor. After 5-min incubation with thrombin or factor Xa in the presence of heparin, the order of T-AT formation was Ala,Tyr>Wt,Gly,Thr>Ile>Leu>Val,Arg>>Asn,Pro>>Asp,His, and that of Xa-AT formation was Gly>Thr>Tyr,Wt>>Asp,Ala>>Val>Pro,His,Leu,Asn,Ile. The Ala- and Tyr-mutants showed 117% and 111% T-AT formations, respectively, while the Gly- and Thr-mutants showed 168% and 128% Xa- AT formations. Interestingly, the Asp-mutant showed no T-AT formation, but 63% Xa-AT formation. In contrast, the Ile- and Leu-mutants showed 75% and 62% T-AT formations, respectively, but essentially no Xa-AT formations. In conclusion, Ser380 located at the bottom of inserted proximal hinge region of the reactive loop of antithrombin could be replaced by at least 12 different amino acid residues without defective effects on secretion. Among these 12 mutants of Ser380, the Gly- and Tyr-mutants were more reactive with thrombin than Wt, and the Gly-mutant was a better factor Xa inhibitor than Wt in the absence of heparin. In the presence of heparin, the Ala- and Tyr-mutants were more reactive with thrombin than Wt, and the Gly- and Thr-mutants showed more potent Xa inhibitory activity than Wt. Differential inhibitions of thrombin and factor Xa in the presence of heparin were observed for the Asp-mutant and the Ile- and Leu-mutants in opposite directions.

COMPLEXATION OF UROKINASE-TYPE PLASMINOGEN ACTIVATOR WITH ITS TYPE-1 INHIBITOR INDUCES A MAJOR CONFORMATIONAL CHANGE IN THE PROTEASE CATALYTIC DOMAIN
R. Egelund, T.E. Petersen, and Peter A. Andreasen Department of Molecular and Structural Biology, Aarhus University, Denmark. E-mail: reo@mbio.aau.dk

Serpins form stable complexes of 1:1 stoichiometry with their target proteinases by reaction of their P1-P1' peptide bond with the proteinase active site, resulting in formation of an ester bond between the active site serine of the proteinase and the P1 residue of the serpin [1]. An important question concerning the mechanism for serpin inhibition is why the ester bond is not hydrolysed as in normal serine proteinase-catalysed peptide bond hydrolysis. There are two principal possibilities. One is that access of water to the active site of the proteinase is restricted. Another is that insertion of the reactive centre loop into b-sheet A of the serpin together with the concomitant translocation of the proteinase is accompanied by conformational alterations of the proteinase molecule and its active site. We have now employed the method of proteolytic susceptibility to address this question. Free PAI-1 contains a number of peptide bonds with hypersensitivity to non-target proteinases, while uPA-complexed PAI-1 is totally resistant to digestion [2]. In contrast, while the serine proteinase domain of free uPA is totally resistant to proteolysis, the serine proteinase domain of PAI-1-complexed uPA can be digested by a number of non-target proteinases. This is also true for uPA in complex with protease nexin-1, but not for uPA in complex with non-serpin inhibitors or for pro-uPA. We mapped the endoproteinase Asp-N and trypsin cleavage sites in the serine proteinase domain of uPA by reducing and non-reducing SDS-PAGE and reverse phase HPLC fractionation, followed by N-terminal sequence analysis and in some cases mass spectrometry. Localisation of the identified cleavage sites in the three-dimensional structure of the serine proteinase domain of uPA allowed us to conclude that they all localise to the C-terminal b-barrel of the serine proteinase domain, including cleavage exactly in the S1 specificity pocket. These data provide evidence that the serine proteinase domain of uPA undergoes a conformational change making it more flexible upon complex formation with a serpin. Furthermore, the induced flexibility of the serine proteinase domain is restricted to the C-terminal b-barrel in and around the specificity pocket. We therefore propose that the stable serine proteinase-serpin complex is trapped at the acyl-enzyme intermediate step due to a distortion of the specificity pocket, resulting in a disruption of the optimal positioning of the ester bond relative to the active site of the proteinase.
1. Egelund, R., Rodenburg, K.W., Andreasen, P.A., Rasmussen, M.S., Guldberg R.E., and Petersen, T.E. (1998) An ester bond linking a fragment of a serine proteinase to its serpin inhibitor. Biochemistry 37, 6375-6379.
2. Egelund, R., Scousboe, S.L., Sottrup-Jensen, L., Rodenburg, K.W., and Andreasen, P.A. (1997) Type-1 plasminogen activator inhibitor. Conformational differences between latent, active, reactive-centre cleaved and plasminogen-activator-complexed forms, as probed by proteolytic susceptibility. Eur. J. Biochem. 248, 775-785.

ROLE OF b-STRAND 5A RESIDUES IN MODULATION OF THE ACTIVITY OF PLASMINOGEN ACTIVATOR INHIBITOR-1 BY SOLVENT, MONOCLONAL ANTIBODIES, AND VITRONECTIN
S. Jensen*, S.L. Schousboe*, T. Kirkegaard*, R. Egelund*, K.T. Preissner# , K.W. Rodenburg*, and P.A. Andreasen* * Department of Molecular and Structural Biology, Aarhus University, Denmark, and #Department of Biochemistry, Justus-Liebig-University, Giessen, Germany. E-mail: sgj@mbio.aau.dk

The inhibitory activity and the rate of latency transition of PAI-1 can be modulated by vitronectin, monoclonal antibodies, and the composition of the solvent. We have used limited proteolysis and site-directed mutagenesis to analyse the role of b-strand 5A residues in activity and latency transition modulation. In detergent-containing buffers, PAI-1 assumed a conformation with an increased tendency to substrate behaviour at 0¯C. This conformation was characterised by an increased proteolytic susceptibility of specific bonds in b-strand 5A and in the b-strand 1A/b-strand 2A/a-helix E-region, which forms a flexible joint during loop insertion [1, 2]. Glycerol rescued PAI-1 from cold-induced substrate behaviour, concomitantly with reducing the proteolytic susceptibility. Substitution of the b-strand 5A residue K325 with A, disrupting a hydrogen bond between b-strand 5A and the loop connecting a-helix F and b-strand 3A, partially abolished cold-induced substrate behaviour. Vitronectin, binding to the b-strand 1A/a-helix E-region, delayed latency transition at pH 7.4, but not at pH 8.1. The K325A substitution drastically changed the response to vitronectin: The latency transition of PAI-1/K325A was insensitive to vitronectin at pH 7.4, but accelerated by vitronectin at pH 8.1. Vitronectin rescued PAI-1, but not PAI-1/K325A, from cold-induced substrate behaviour in detergent-containing buffers. The K325A substitution potentiated the activity-neutralising effect of PAI-1 by two monoclonal antibodies against the a-helix F region, activity-neutralisation being associated with increased substrate behaviour. We conclude that the conformation of b-strand 5A is affected by agents binding to or near the flexible joint-region, and that K325 plays a crucial role in relaying the changes in the flexible joint-region to changes in b-sheet A movements and loop insertion.
[1] Stein, P.E. and Chothia, C. (1991) J. Mol. Biol. 221, 615-621
[2] Egelund, R., Schousboe, S.L., Sottrup-Jensen, L., Rodenburg, K.W. and Andreasen, P.A. (1997) Eur. J. Biochem. 248, 775-785

PROTEIN C INHIBITOR (PCI): POSSIBLE ROLE AS A RETINOIC ACID BINDING AND DELIVERING SERPIN.
Margarethe Geiger, Ingrid Jerabek, Maria J. Prendes, Michael Krebs, Bernd R. Binder. Department of Vascular Biology and Thrombosis Research, University of Vienna, Austria.

The serpin family of glycoproteins includes inhibitors of serine proteases as well as non-inhibitory members with other biological functions (e.g. the hormone precursor angiotensinogen, the hormone carriers corticosteroid binding globulin and thyroxin binding globulin). It is unknown whether or not inhibitory serpins might have additional non-inhibitory functions. We studied binding of 3H-labeled hydrophobic hormones (estradiol, progesterone, testosterone, cortisol, aldosterone, and retinoic acid) to different inhibitory serpins (antithrombin III, heparin cofactor II, plasminogen activator inhibitor-1, and PCI). 3H-all-trans-retinoic acid (RA) bound in a dose and time dependent way to PCI, while none of the other 3H-hormones used bound to any of the serpins analyzed. Binding of 3H-all-trans-RA to PCI was specific and could be competed in a dose dependent way by unlabelled all-trans-RA, 9-cis-RA, and retinol. We also analyzed the effect of PCI on the signaling function of RA using HT-1080 fibrosarcoma cells, which - in response to RA - increase their synthesis of tissue plasminogen activator (tPA). When HT-1080 cells were incubated with 9-cis or all-trans-RA, PCI enhanced in a dose dependent manner RA-induced tPA synthesis. These data suggest that besides its role as an inhibitor of serine proteases PCI might also function as a hormone binding and delivering serpin. We therefore analyzed the presence of PCI in cells and tissues requiring retinoic acid for differentiation processes. By means of immuno-histochemistry, immun-electron microscopy, RT-PCR, and in situ hybridization we could localize PCI in the epidermis, in keratinocytes, in platelets and megakaryocytes, as well as in other bone marrow derived cells and peripheral blood leukocytes. Together with previously published data (high concentrations of PCI throughout the male reproductive tract) this tissue localization of PCI strongly supports the hypothesis that PCI might play a role in the local delivery of retinoic acid.

TOWARDS THE ROLE OF PROTEIN C INHIBITOR IN VIVO: DEVELOPMENTAL EXPRESSION, AND GENERATION AND CHARACTERIZATION OF GENETICALLY MODIFIED MICE.
Gerry T.M. Wagenaar, Hanneke J. van Vuuren, Merone Girma, Margriet J. Tiekstra, Maaike Verschuren and Joost C.M. Meijers. Department of Haematology, University Medical Center, Utrecht, The Netherlands.

Protein C inhibitor (PCI) is a heparin-dependent serine protease inhibitor (serpin) with a large target specificity. In plasma, PCI has procoagulant activity by inhibiting activated protein C (APC) or the thrombin-trombomodulin complex, and anticoagulant activity by inhibiting coagulation factors XIa, Xa, thrombin and kallikrein. PCI is also known as plasminogen activator inhibitor-3 after its discovery in urine, in which it acts as an inhibitor of urokinase. PCI is present in many body fluids and is expressed in many organs in adults in man and rodents. The highest PCI expression is observed in the male reproductive organ. To determine the role of PCI in vivo, we decided to generate genetically modified mice with a targeted gene disruption of the mouse PCI gene and transgenic mice in which human PCI is overexpressed. To anticipate on possible developmental effects in PCI knockout mice we first studied the embryonic expression of mouse PCI mRNA in wild type mice using in situ hybridization. Mouse PCI mRNA could be detected from 11 days of gestation onward and was expressed in epithelia of tubular structures of the gastrointestinal tract, the respiratory system, the urogenital system and the central nervous system. This suggests a role for PCI in the maintenance of fluidity in these tubular structures. In addition PCI mRNA was expressed in neural cells, plexus choroideus, bony structures, and smooth and skeletal muscle cells. For the knockout model a targeting vector containing 4.4 kb of homologous sequence of the mouse PCI gene, including exons I to V, and the 1.8 kb positive selection marker hygromycine, cloned into exon II, was prepared, linearized and electroporated into embryonic stem (ES) cells. 3 out of 600 targeted clones showed correct homologous recombination, which was confirmed by Southern blotting. These ES cell clones were injected into blastocysts, which gave rise to 4 male chimeric mice after implantation into pseudo pregnant females. The chimeric mice are currently bred to obtain heterozygous and homozygous PCI knockout mice. In the transgenic model a targeting construct was generated in which human PCI cDNA expression was driven by the liver-specific mouse albumin promoter/enhancer. We characterized mice of 3 transgenic lines, which did not show gross developmental abnormalities. In these mice liver-specific expression of human PCI mRNA was observed in hepatocytes only, showing a heterogeneous expression pattern. In two transgenic lines PCI mRNA was expressed pericentrally and in the other line periportally. A functional human PCI protein was secreted in to the circulation (plasma levels: 3-5 mg/ml in heterozygous and 10 mg/ml in homozygous transgenic mice). Functional activity of the transgene was demonstrated by the inhibition of human APC activity in the presence of heparin in plasma of transgenic mice. Interestingly, mice that showed a periportal expression pattern of human PCI mRNA had the highest specific activity of the transgene. The results strongly indicate that these genetically modified mice are suitable models for studying the in vivo role of PCI. Donor-Donor Energy Migration (DDEM) for Determining Intramolecular Distances in the Latent wild type PAI-1. HÑgglîf P, Bergstrîm F, Karolin J, Fa M, Wilczynska M, Johansson L B.-è, and Ny T Department of Medical Biochemistry and Biophysics and Physical Chemistry, UmeÜ University, S-901 87 UmeÜ, Sweden PAI-1 is a major inhibitor of plasminogen activators (PAs). Although PAI-1 is produced as an active molecule, it has a metastable conformation and can exist in three interconvertible forms (active, latent and substrate-like). To investigate structural-functional aspects and the molecular dynamics of PAI-1 we have taken advantage of the lack of cysteines in the PAI-1 molecule and replaced different amino acid residues with cysteine pairs, thereby creating attachment sites for extrinsic fluorescent probes for intramolecular distance measurement. After expression in E. coli and purification to homogeneity, the PAI-1 mutant proteins were converted to the latent form. The cysteines in the mutated PAI-1 were then labelled with the fluorescent probe BODIPY. To calculate intramolecular distances between the two fluorescent donors in latent PAI-1, a recently developed donor-donor energy migration (DDEM) method (1) was applied. The PAI-1 residues substituted by cysteines forms a triangle in which the length of each arm has been determined. The donor-donor energy migration measurements revealed distances: between the residues Cys106 and Cys185 in Val106Cys: His185Cys double mutant of PAI-1 to be 54 Ò 3è, between residues Cys106 and Cys266 in Val106Cys: Met266Cys double mutant of PAI-1 to be 55 Ò 3 è and between residues Cys185 and Cys266 to be 38 Ò 3è in His185Cys: Met266Cys double mutant. This is in reasonable agreement with values of 60.8è, 55.1è and 30.8è respectively, which were determined by X-ray diffraction of the latent wild type PAI-1 (2). Although the accuracy of this method for intramolecular distance measurements, is not at atomic level of resolution, it gives the opportunity to measure distances also in active PAI-1 and to determine conformational changes that take place during complex formation and following interaction with cofactors.
(1) Karolin, J.; Fa, M.; Wilczynska, M.; Ny, T.; Johansson, L. B-è. Biophysical Journal Vol74 January 1998 11-21
(2) Mottonen J., Strand A., Symersky J., Sweet R.M., Danley D.E., Geoghegan K.F., Gerard R.D., Goldsmith E.J. (1992) Nature, 355: 270-273.

MOLECULAR IDENTIFICATION OF NOVEL SECRETORY VESICLE SERPINS. ENDOPIN 1 AND 2: POTENTIAL ROLE IN PRONEUROPEPTIDE PROCESSING.
Vivian Hook, Shin-Rong Hwang, Brent Steineckert, Thomas Toneff, Catherine Sei, and Sukkid Yasothornsrikul, Dept. of Medicine, University of California, San Diego.

The biosynthesis of peptide neurotransmitters and hormones requires proteolytic processing of protein precursors into active neuropeptides. It is known that the extent of proneuropeptide processing can be regulated. For example, proenkephalin processing is limited in adrenal medulla, but is more extensive in brain; the product (Met)enkephalin acts as an endogenous analgesic. These observations, and others, suggest a possible role for endogenous protease inhibitor(s) that may regulate processing. Our initial studies indicated the presence of ACT immunoreactivity in secretory vesicles of adrenal medulla (chromaffin granules) which is abundant in enkephalins. These secretory vesicles also contain several proneuropeptide processing enzymes including the subtilisin-like prohonnone convertases (PCs), the cysteine protease' prohormone thiol. protease,' and a 70 kDa aspartyl protease. To identify potential endogenous protease inhibitor(s) in adrenal medulla, molecular cloning of ACT-like proteins in adrenal medulla utilized a combination of cDNA library screening, RT-PCR, and genomic DNA cloning. This work resulted in the isolation of two related serpin cDNAs that represent novel, neuroendocrine serpins with homology to a1-antichymotrypsin (ACT). These serpins are referred to as endopin 1 and 2 to indicate neuroendocrine serpins. Endopin 1 and 2 contain reactive site loop that are characteristic of serpin protease inhibitors. Moreover, alignment predicts that endopin 1 and 2 possess different target protease specificities. Endopin 1 was predicted to possess Arg as the P1 residue, suggesting that the endopin 1 inhibits target protease(s) cleaving at Arg or basic residues, consistent with the cleavage specificity of proneurtiopeptide processing enzymes. However, endopin 2 was predicted to possess Ser as the P1 residue, suggesting inhibition of proteases cleaving at hydrophobic residues. Recombinant NHis-tagged endopin 1 potently inhibited trypsin that cleaves at basic residues (10-20 nM), but did not inhibit proteases cleaving at hydrophobic residues including elastase and subtilisin A. In contrast, recombinant N-His-tagged endopin 2 potently inhibited elastase (nM), but did not inhibit trypsin. These results demonstrate the protease-specific nature of these serpins. Specific antibodies demonstrated the localization of endopin 1 and 2 to isolated secretory vesicles of adrenal medulla. In primary cultures of adrenomedullary chromaffin cells, functional secretion of endopin 1 and 2 with (Met)enkephalin, upon KCI depolarization, also demonstrated localization of these endopins to secretory vesicles. Importantly, endopin 1 and 2 inhibited the proenkephalin cleaving enzyme known as 'prohormone thiol. protease.' which is colocalized with endopin 1 and 2 in these secretory vesicles. In summary, endopin 1 and 2 represent unique secretory vesicle serpins, with different target protease specificities, that may participate in the regulation of proneuropeptide processing enzymes within secretory vesicles. In vivo modulation of the acrosome reaction in the female genital tract by a CBG-like SERPIN Miska W, Baltes P, ÿSÝnchez R & Henkel R Department of Dermatology & Andrology, Justus Liebig University, Giessen, Germany ÿFaculty of Medicine, Frontera University, Temuco, Chile Introduction The acrosome reaction (AR) is prerequisite for successful mammalian fertilisation. The AR is a modified exocytotic process, in which the outer acrosomal membrane fuses with the plasma membrane of the spermatozoon, resulting in its complete loss and release of the acrosomal content. Enzymes thereby released are thought to play a role in the penetration of spermatozoa through the outer oocyte investments. Human follicular fluid (hFF) contains a component that is able to induce the AR in spermatozoa. Results?&?Discussion Previous experiments have demonstrated the protein character of the acrosome reaction-inducing substance (ARIS), and according to the current literature, steroids also seem to play an important role in AR induction. Our recent findings have shown that hFF from which proteins and/or steroids (protease-, DCC treatment) had been removed could not induce the AR. However, after removal of steroids, the AR-inducing activity of hFF could be restored by exogenous progesterone, but only in the presence of intact protein. In gel filtration experiments with 3H-progesterone-labelled hFF elution of the radioactive signal in the high-molecular weight range, corresponding to bound progesterone, was found. Based on these results we suggest that the effect of ARIS is a synergistic action of progesterone and a progesterone-binding protein. The protein has been found to be immunologically identical with the corticosteroid-binding globulin (CBG), which has already been described and serves as a transport protein for progesterone and cortisole in the plasma. Our findings also show that the AR can be induced by CBG-progesterone complex at nanomolar concentrations, where both progesterone and the protein alone are unable to bring about an effect. In the culture medium of human cumulus oophorus cells, ARIS was detected by its biological activity as well in Western blot analysis. Moreover, secretion of a CBG-like protein was found in endothelial cells of the tubulus depending on the hormonal cycle. Both immunological and radiochemical investigations strongly indicate that human cumulus cells actively express and secrete a CBG-like progesterone-binding protein. Conclusions CBG and, therefore, ARIS is a member of the superfamily of serine protease inhibitors (SERPINs). The SERPINs have numerous regulatory functions by means of their inhibitory effects. In the case of transport proteins, to which CBG belongs, they have lost their inhibitory properties, but have retained the typical structure and functional characteristics of SERPINs. CBG is able to bind and transport small molecules and then to release them at specific targets after SERPIN-specific conformation changes in the CBG molecule through the action of a serine protease. Based on these findings, the following mechanism is suggested for the role of the CBG-progesterone complex in the induction of the AR: Progesterone is transported by CBG to the spermatozoon where CBG is cleaved by a plasma membrane-bound protease, leading to conformational changes within the protein. This causes local release of progesterone and, consequently, increases the local concentration of the hormone, possibly stimulating a Ca2+-channel, which leads to an influx of Ca2+ and finally induces the AR.

GENETIC VARIATION IN THE 3' FLANKING SEQUENCE OF THE a1-ANTITRYPSIN GENE IS ASSOCIATED WITH INCREASED PROGRESSION OF ATHEROSCLEROSIS IN THE LOPID CORONARY ANGIOGRAPHY TRIAL.
Philippa J Talmud*, Steve G Martin*, Stephen L Sturley+, Steve E Humphries*, Marja-Riitta Taskinen# and Mikko SyvÑnne#. * Royal Free and University College Medical School, London. + Columbia University, New York. # Helsinki University Hospital, Helsinki.

a1-antitrypsin (AAT) is an important plasma serine-protease inhibitor, its major function being to limit neutrophil elastase released during inflammation. AAT prevents damage to elastic tissue in the lung, as well as the gut, liver and vasculature. AAT deficiency leads to emphysema and liver disease. The finding by Lobritto et al1 that human apoB bound to AAT in a yeast two-hybrid system prompted us to examine whether there was an association between AAT gene variation and progression of coronary atherosclerosis. We studied the TaqI (Oct1G/A) polymorphism in the 3' flanking sequence of the AAT gene. The loss of the TaqI site (Oct1-A) destroys an Oct-1 binding site leading to reduced AAT gene expression in the acute phase in carriers. DNA samples were genotyped from 313 individuals who had participated in LOCAT, a study of atherosclerosis progression in Finnish men, treated with gemfibrozil or placebo, who had undergone coronary artery bypass surgery. Focal and diffuse atherosclerosis had been assessed by angiography at zero and 2´ years. The rare allele (Oct1-A) frequency was 0.05 (95%CI 0.032-0.66),in both treatment and placebo groups. Individuals who carried the (Oct1-A) allele had significantly greater progression in focal atherosclerosis than individuals homozygous for the common Oct1-G allele [change in minimal lumen diameter -0.169mm (+0.206) vs -0.057mm (+ 0.169), respectively, p=0.001], however this was independent of treatment. There was no effect of genotype on diffuse progression of disease or on levels of lipids or clotting factors. These results are the first report that genetic variation in the AAT gene determines atherosclerosis progression. The mechanism of this effect, we propose, is due to 'protective' AAT, bound to apoB-LDL, being recruited into the plaque where it inhibits the degradative effects of elastase and other proteases. Reduced levels of AAT in Oct1-A carriers could thus promote atherogenesis.
1 Lobritto, SJ, Lewin B, Behari A, Abrahams J, Deckelbaum, RJ and Sturley SL (1998) Circulation Supplement. I-108 abstract 552.

ANTICOAGULANT PROPERTIES OF ANTITHROMBIN III-a AND -b IN A TNF STIMULATED ENDOTHELIAL CELL CULTURE SYSTEM.
J Rîmisch. Centeon Pharma GmbH, Reserach, 35002 Marburg, Germany.
INTRODUCTION. Antithrombin III (ATIII) plays an important role in the regulation of haemostasis. The two isoforms of ATIII, named a- and b-, reveal different affinities to glucosaminoglycans like heparan sulfate as part of the vascular matrix. Different anticoagulant properties and functions of the isoforms might be expected, if the endothelium is stimulated as observed after TNF release in the course of sepsis. METHODS. Cultured HUVEC were stimulated with TNF-a and were subsequently incubated with recalcified human plasma or ATIII deficient plasma (ATIII-DP), both of which contained a clot inhibitor. Samples of resting and stimulated cells were drawn at baseline and up to 30 min after recalcification. The concentrations of the prothrombin fragment F1+2 were quantified by ELISA. Increasing concentrations of the purified ATIII-a and -b isoforms were added to the ATIII-DP and the impact of the F1+2 generation was studied. In addition, the TNF stimulated HUVEC were incubated with FVIIa in order to facilitate accelerated F1+2 generation in a recalcified FVII-DP, which was added after removal of unbound FVIIa. The impact of the ATIII isoforms on the inhibition of cell bound FVIIa was investigated by subsequent incubation with ATIII, washing and addition of recalcified FVII-DP. RESULTS. Compared to unstimulated cells, TNF stimulated endothelial cell monolayers induced an accelerated generation of F1+2, which was even more pronounced using ATIII-DP. Addition of ATIII 3 0.5 IU/ml to the deficient plasma resulted in a significant reduction of the F1+2 release. In this model we did not observe, however, a statistically significant difference between the isoforms. Both of them also had a significant but equivalent inhibitory effect on the FVIIa loaded HUVEC. CONCLUSION. Reduced plasma ATIII levels resulted in a significant increase of F1+2 generation mediated by the stimulated endothelium. Both ATIII isoforms significantly reduced the procoagulant potency. No significant difference between ATIII-a and ATIII-b was observed so far in the test-system employed.

ANTITHROMBIN III-a AND -b: INHIBITION CHARACTERISTICS OF PLASMATIC PROTEASES AND INACTIVATION BY ELASTASE.
J. Rîmisch. Centeon Pharma GmbH, Research, 35002 Marburg, Germany.
INTRODUCTION. The two isoforms of Antithrombin III, ATIII-a- and -b, differ in their affinities to glycosaminoglycans There is however limited information whether both isoforms reveal equivalent inhibitory potencies towards different proteases. In the course of septic events the consumption of ATIII and the degradation by proteases contribute to the reduction of the plasma inhibitory capacity. We investigated the resistance of ATIII-a- and ATIII-b against the inactivation by elastase. METHODS. The ATIII isoforms were isolated by affinity cromatography from an ATIII concentrate. Increasing concentrations of ATIII-a or -b were incubated with purified human thrombin, FXa, FIXa or plasmin in the absence or presence of heparin. The ATIII concentrations to inhibit 50% of each protease activity were determined (IC50) employing chromogenic assays. Global coagulation tests were performed using ATIII deficient plasma. The isoforms were incubated with elastase and their residual thrombin inhibitory activities were determined. RESULTS. The comparison of the ATIII isoforms revealed equivalent inhibitory potencies towards thrombin and plasmin in the absence of heparin, whereas IC50 values for ATIII-b were found to be lower for FXa and FIXa. No difference between the isoforms was observed upon addition to ATIII deficient plasma regarding the prolongation of APTT; only slightly more prolonged plasma recalcification times were found for ATIII-b. In the presence of heparin the IC50 values of both ATIII-a and ATIII-b were significantly lower showing a trend to a slightly better inhibition by ATIII-b. The inactivation rates by elastase were found to be comparable for both isoforms under the conditions chosen. CONCLUSION. In particular in the absence of heparin there was a difference in the inhibitory potency of ATIII-a and -b towards the amidolytic activities of distinct coagulation proteases. The observed distinctions might be of importance in local events. An equivalent inactivation by elastase was found for ATIII-a and ATIII-b.

ALPHA-1-ANTITRYPSIN (AAT) GENE REGULATION: MODULATION BY A 3' ENHANCER IN RESPONSE TO THE CYTOKINES INTERLEUKIN-6 AND ONCOSTATIN-M (OM)
Noor Kalsheker, Kevin Morgan and Peter Marsters. Division of Clinical Chemistry, School of Clinical Laboratory Sciences, Queen's Medical Centre, University of Nottingham NG7 2UH, UK

There are at least two tissue-specific promoters and two enhancers which regulate the 1-antitrypsin gene. Cytokine-induced expression is modulated by a 3' enhancer. We previously demonstrated that interleukin-6 modulates expression via this enhancer in Hep G2 cells (1). Oncostatin-M induces a 30-fold increase over basal AAT expression in human lung alveolar epithelial cells (2) and has a similar effect to IL-6 in Hep G2 cells, i.e. two to three-fold effect. IL-6 mediates its effects via the transcription factor NF-IL6, whereas OM mediates its effects via the STAT family of transcription factors. Our previous studies, using reporter gene constructs containing combinations of AAT enhancer regions in transfection of Hep G2 cells, demonstrated that basal regulation proceeds via 5' enhancers, whereas the IL-6 induced up-regulation requires both 5' and 3' enhancers. Incorporation of a STAT-like binding site, located in the AAT flanking sequence 20 bp downstream of the previously used 3' enhancer, into a reporter gene construct confers an OM (100 ng/ml) response in Hep G2 cells, resulting in a three-fold increase in activity over basal (p = 0.0004). Electrophoretic mobility shift assays show that this region of the AAT gene can bind STAT 3, therefore implicating this member of the STAT family. The IL-6 response (mean of 2.7-fold increase over basal) with the STAT construct, was no greater than previously seen, but the combined response to IL-6 and OM (4.8-fold, p<0.0001) was higher, presumably due to the STAT site (p = 0.028). This suggests the presence of two independent pathways for AAT regulation. Interestingly, constructs which contained a naturally occurring enhancer mutation, demonstrated a reduced OM response (1.9-fold over basal). In vivo studies confirm the findings of a diminished acute-phase response arising from the enhancer mutation. Our studies define a pathway that regulates local AAT production, which may be important in controlling local tissue injury, and these findings may be relevant to other serpins which respond to these cytokines.
This work was supported by EU grant BMH4-CT96-0152 as part of the Biomed 2 Eurolung Consortium.
1. Morgan K, Scobie G, Marsters P, Kalsheker N (1997). Biochim. Biophys. Acta, 1362: 67-76.
2. Sallenave J-M, Tremblay GM, Gauldie J, Richards CD (1997). J. Interferon & Cytokine Res., 17: 337-346.

SELF-ASSOCIATION OF AN INTERMEDIATE IN THE UNFOLDING OF MASPIN, A TUMOR-SUPPRESSING SERPIN
Ting Liu, Philip A. Pemberton*, and Andrew D. Robertson Department of Biochemistry, University of Iowa, Iowa City, IA 52240 USA & *LXR Biotechnology, 1401 Marina Way South, Richmond, CA 94804 USA

Maspin is a tumor suppressor protein (Mr 42,000) found in normal human mammary epithelium but that is missing in many breast cancers [Zou et al. (1994) Science 263, 526-529]. Recombinant maspin inhibits tumor cell motility, invasion and metastasis [Sheng et al. (1996) Proc. Natl. Acad. Sci. USA 93, 11669-11674]. Maspin thus has potential value as an anti-cancer therapeutic. Sequence analysis places maspin in the serpin family and a molecular model has been proposed on the basis of structural homology [Fitzpatrick et al. (1996) Prot. Eng. 9, 585-589]. The molecular mechanism by which maspin acts is unknown, although recent evidence suggests that tissue plasminogen activator may be a target [Sheng et al. (1998) Proc. Natl. Acad. Sci. USA 95, 499-504]. Many serpins adopt multiple conformations as an integral part of their function. In addition, many serpins are involved in "conformational" diseases, where destabilizing mutations lead to self-association and loss of normal function [Carrell & Gooptu (1998) Curr. Opin. Struct. Biol. 8, 799-809]. We are studying the folding and stability of maspin in an effort to understand the relationship between its structure and function, to investigate the possibility of alternative conformations in maspin and their role in function, and as background for the possible design of improved maspin molecules for therapeutic purposes. Urea denaturation of maspin at pH 7 and 25¯C and at protein concentrations ranging from 0.01 - 0.2 mg/ml has been monitored by circular dichroism (CD) and intrinsic tryptophan fluorescence. Two distinct transitions are observed. The midpoint of the first transition is about 2 M urea at all protein concentrations, but the midpoint of the second transition increases with increasing protein concentration. Refolding of denatured maspin leads to complete recovery of native-like spectroscopic signals at the lower protein concentrations, but only partial recovery is observed at higher concentrations of maspin. These results suggest that maspin undergoes a partial unfolding in the first transition, leading to an intermediate which tends to self-associate. We hypothesize that, in general, this intermediate is populated under solution conditions that destabilize native maspin. We further hypothesize that the resulting self-association reaction is at least partially responsible for the anomalous dose-response data for maspin in bioassays, where higher concentrations of maspin lead to a diminished biological response. The molecular properties of the unfolding intermediate and maspin multimers are under investigation.

2.6A STRUCTURE OF BETA ANTITHROMBIN AND COMPLETE STRUCTURE OF LATENT ANTITHROMBIN
X.Y. Pei, Structural Medicine Division, Department of Haematology, Cambridge Institute for Medical Research, University of Cambridge, Cambridge CB2 2XY, UK, R. Skinner, Unilever, UK, J.P. Abrahams, Biophysical Structural Chemistry, Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University; Einsteinweg 55 / P.O. Box 9502 2300RA, Leiden, The Netherlands, and R.W. Carrell, Structural Medicine Division, Department of Haematology, Cambridge Institute for Medical Research, University of Cambridge, Cambridge CB2 2XY, UK.

Antithrombin is the major inhibitor of serine proteases of the blood coagulation cascade but its most active form is the minor beta-antithrombin isoform that constitutes only 5-10% in proportion to the main, alpha-antithrombin form. Beta-antithrombin differs from alpha-antithrombin in lacking a carbohydrate sidechain at the Asn135 glycosylation site (Peterson & Blackburn, 1985). Beta-antithrombin binds heparin and heparan sulphates with a higher affinity than does alpha-antithrombin and this explains its physiological function as the first-line inhibitor of coagulation proteases (Turko et al., 1993). Here we present the crystal structure of a dimer of beta-antithrombin at 2.6A, one molecule being in the active inhibitory form (I-form) with linkage through its reactive center loop (RCL) to a second molecule in the latent conformation (L-form). Both the I- and L- forms of beta-antithrombin show clear but different changes at Asn135 when compared to alpha-antithrombin. In particular the internal orientation, with well-defined H-bonding, of Asn135 in the L-form provides an explanation for the constant fraction of unglycosylated antithrombin (beta-antithrombin) present at synthesis, though the likely main cause is the coding at this site of an Asn-X-Ser sequence which is known to inefficiently result in glycosylation (Picard, et al, 1995). The principal change indicating a direct reason for the higher heparin affinity of beta-antithrombin is a shift of one of the main binding residues, Arg129, to an optimal orientation. The likely reason for the overall increase in affinity is the greater flexibility of the hD-s2A loop (containing Asn135) that undergoes conformational rearrangement on the binding of heparin. The completion of the structure of beta-antithrombin at 2.6A has been accompanied by further rounds of refinement of the structure of alpha-antithrombin at 2.6A (from the data of Skinner et al, 1997). These now together provide a complete structure of both the active and the latent molecules with the only gaps being in the mobile portions of the amino terminus. The results along with the structure of dimeric alpha-antithrombin complexed with the heparin pentasaccharide (Jin et al, 1997) allow a comparison of the conformational changes involved in inhibitory activation and provide a basis for the design of novel anticoagulant drugs.

THE MECHANISM OF ANTICOAGULANT ACTIVATION OF ANTITHROMBIN BY HEPARIN
Whisstock J.C.1, Pike R.N. 1, Carrell R.W. 2, Jin L. 2 and Lesk A.M.2 Dept of Biochemistry and Molecular Biology, Monash Uni., Melbourne, VIC. 2Dept of Haematology, University of Cambridge.

Antithrombin is responsible for the inhibition of thrombin and factor Xa and thus plays a critical role in the regulation of the blood coagulation cascade. Uniquely, antithrombin requires heparin for full activity against its cognate proteases. The X-ray crystal structure of antithrombin bound to heparin pentasaccharide (HP) - a unit within full-length heparin responsible for specific and tight binding - has recently been described (Jin et al., 1997). This structure reveals a remarkable conformational change in the entire molecule upon binding to pentasaccharide, resulting in a greater exposure of the Reactive Centre Loop (RCL). Here we analyze the mechanism of conformational change in antithrombin by means of a structural comparison of free native antithrombin (F-AT) with pentasaccharide bound native antithrombin (P-AT). We divide the structures into: rigid fragments that move during the conformational change, hinge regions and areas of plastic deformation. These data allow us to relate the conformational change to the binding of pentasaccharide and we propose a novel mechanism for pentasaccharide activation. Five fragments were identified. The top half of the molecule (i.e. hA, hG, hH, the B-sheet, the C-sheet and part of s5A/s6A) represents one major rigid unit, and the bottom half (hB, hC, hF, hI and the bottom half of the A-sheet) another major fragment. The other three fragments are smaller, with hD, hE and the top of s2A/s3A all representing separate rigid units that shift upon HP binding. The binding of HP to F-AT results in movement of hD relative to hA and Arg 47. This shift in hD appears to cause a rotation in the bottom half of the molecule which is then transmitted to the top of the molecule (s2A/s3A) with consequent loop expulsion. We relate these data to the known high and low affinity natural variants of antithrombin and suggest a structural basis for altered heparin affinity. Some variants are relatively straightforward to explain, i.e. mutations of residues shown to be critical HP binding residues, or catastrophic mutations which would obviously perturb secondary structure. Others variants appear to require a more subtle explanation. One particularly fascinating mutation is the recently described antithrombin Wibble T85M (Beauchamp et al., 1998), in which the rate of pentasaccharide-induced conformational change is four times higher than in a-antithrombin. T85 is buried underneath the heparin binding site and packs against the D-helix to which the HP binds. The structural comparison indicates that T85 is located in a critical position and plays a mechanistic role during the conformational change.
References Beauchamp, N.J., Pike, R.N., Daly, M., Butler, L., Makris, M., Dafforn, T.R., Zhou, A., Fitton, H.L., Preston, F.E., Peake, I.R. and Carrell, R.W. Blood 92: 2696-2706 (1998).
Jin, L., Abrahams, J.P., Skinner, R., Petitou, M., Pike, R.N. and Carrell, R.W. Proc Natl Acad Sci U.S.A. 94: 14683-14688 (1997).<

Zn2+ AND PCI ARE IMPORTANT REGULATORS OF HUMAN GLANDULAR KALLIKREIN 2 ACTIVITY.
Janita Lîvgren, Dept. Laboratory Medicine, Malmî, Sweden, Kalervo Airas, Dept. Biochemistry, Turku, Finland, and Hans Lilja, Dept. Laboratory Medicine, Malmî, Sweden.

INTRODUCTION & OBJECTIVES: Human glandular kallikrein 2 (hK2) is a serine protease expressed by the prostate gland that has 80% sequence similarity with prostate specific antigen (PSA or hK3). HK2 has trypsin like activity and cleaves its substrates carboxy terminal of single or double arginines. HK2 has been shown to be able to autoactivate and activate the zymogen form of PSA (proPSA) and urokinase and effectively cleave fibronectin. All these functions may be of importance in the progression of prostate cancer. The prostate contains extremely high 5 - 10 mM concentrations of zinc which is excreted to prostatic fluid by the epithelial cells. The zinc concentration of the prostate is significantly decreased in prostate cancer. Our objective was to study the regulation of hK2 activity by zinc as it could be a delicate way to regulate enzyme activity. The association kinetics of hK2 and PCI were studied as hK2 in seminal plasma is found in complex with this inhibitor, which is also expressed in the prostate. The ability of SLPI, PSTI, ACT, and Aprotinin to inhibit hK2 was also tested. METHODS: The inhibition of hK2 by Zn2+ and PCI, SLPI, PSTI, ACT, and Aprotinin was studied by kinetic measurements. The Zn2+ inhibition results were fitted to different inhibition models until a model which explains the data satisfactorily was found and inhibition constants could be calculated. The association constant of hK2 and PCI without and with different heparin concentrations was calculated using an equation for slow binding inhibition.
RESULTS: HK2 is inhibited by Zn2+ in a manner which involves binding two zinc ions. The second Zn2+ binding site may be generated between two adjacent hK2 molecules. The binding of one Zn2+ has either an enhancing effect or no effect on the activity and the binding of the second Zn2+ leads to inhibition. The binding constants for the two zinc ions were KZn1= 4.6Ò3.9 ÊM Zn2+ and KZn2= 3.2Ò0.7 ÊM Zn2+. The association constant for hK2 and PCI was 2.0 x 105 M-1s-1. Heparin (10 ÊM) increased the association constant to 8.7 x 105 M-1s-1. None of the other inhibitors tested were able to efficiently inhibit hK2.
CONCLUSIONS: HK2 activity is likely to be regulated in the prostate and prostatic fluid by Zn2+. The inhibition mechanism allows enzymatic activity providing that hK2 has a high affinity for the substrate. The inability of prostate cancer cells to accumulate and excrete zinc may at some point affect this control mechanism and lead to increased hK2 activity. PCI is a very efficient inhibitor of hK2 and could be another means of controlling hK2 activity in the prostate. In seminal plasma hK2 is rapidly inhibited by PCI originating from the seminal vesicles. The half-life for the formation of the PCI-hK2 complex was calculated to be 1 s at the PCI and hK2 concentrations, 4 ÊM and 0.2 ÊM respecively, found in seminal plasma.

THE HEPARIN BINDING SITEIN PROTEASE NEXIN 1 IS REQUIRED FOR RETENTION AND DEGRADATION BUT NO THE BINDING OR INTERNALISATION OF  INHIBITORY COMPLEXES OF PROTEASE NEXIN WITH UROKINASE-TYPE PLASMINOGEN ACTIVATOR
Robert J. Crisp, Mary F. Knauer and Daniel J. Knauer, Department of Developmental and Cell Biology, University of California, Irvine, CA 92697.

Human protease nexin-I (PN1) is a 43 kDa member of the SERPIN family, restricted to extravascular compartments. The primary inhibitory targets of PN1 are thrombin (Th) and urokinase-type plasminogen activator (uPA). In the present studies we have investigated the mechanism of clearance of Th:PN1 and uPA:PN1 complexes in human fibroblasts, using two reagents: 1) an antibody directed against the low density lipoprotein receptor-related protein (LRP) binding site in PN1 (47Pro-Ile58)1, and 2) a variant form of PN1 deficient in heparin binding (PN1-K7E)2. In previous studies, we demonstrated that the heparin binding site in PN1 is an important component in the clearance mechanism of Th:PN1 complexes. Thrombin:PN1 complexes first bind to cell-surface heparins via the heparin binding site in PN1, and are subsequently internalized by the LRP1,2,3. However, other studies have shown that when PN1 forms complexes with uPA, the initial cell-surface binding and internalization of uPA:PN1 complexes requires the urinary plasminogen activator receptor (uPAR) and is mediated by the protease moiety of the complexes. Since this would imply two distinct pathways for PN1 internalization, we sought to investigate the roles, if any, of the LRP and heparin binding sites of PN1 in uPA:PN1 complex catabolism. Presently, we demonstrate that anti-47Pro-Ile58 IgG potently inhibits the LRP mediated inter-nalization of Th:PN1 complexes, while it has no effect on uPA:PN1 internalization. The antibody also had no effect on cell-surface binding of Th:PN1 or uPA:PN1 complexes. Complexes between the heparin binding deficient variant of PN1 and uPA (uPA:PN1-K7E) bound to the cell-surface to the same degree as native uPA:PN1 complexes, ruling out a role for the heparin binding site in PN1 in the cell-surface binding of uPA:PN1 complexes. These data strongly suggest that the clearance of uPA:PN1 complexes and Th:PN1 complexes utilize different penultimate receptors, prior to LRP-mediated internalization, and that the LRP binding site in PN1 is not required for uPA:PN1 clearance. The simplest interpretation of these data is that Th:PN1 complexes utilize heparin as an initial cell-surface binding site, while uPA:PN1 complexes utilize the uPAR. In contrast to the cell-surface binding studies, measurements of the clearance rates of uPA:PN1-K7E, and Th:PN1-K7E complexes, which are both heparin binding incompetent, revealed that neither type of complex was cleared efficiently by HF cells. This result was somewhat surprising in the case of uPA:PN1 complexes, since the cell-surface binding of uPA:PN1 complexes was shown to be heparin independent. A careful analysis of the LRP-mediated internalization of these complexes revealed that the complexes were internalized at the same initial rate as native uPA:PN1 complexes, but were rapidly exocytosed. Binding experiments conducted in the pH range of 3.5 to 7.0, revealed that cell-surface bound uPA:PN1-K7E complexes were released between pH 5.5 and 5.0, whereas native uPA:PN1 complexes remained bound, but were rapidly released by the addition of soluble heparin. These data are consistent with a mechanism in which uPA:PN1 complexes bound to the uPAR are released at acidic pH, and re-bind to heparin sulphate in the endosome, resulting in targeting to lysosomes4.
1Knauer, M.F., Hawley, S.B., and Knauer, D.J. (1997) J. Biol. Chem. 272:12261.
2Knauer, M.F., Kridel, S.J., Hawley, S.B., and Knauer, D.J. (1997) J. Biol. Chem. 272:29039.
3Knauer, M.F., Crisp, R.J., Kridel, S.J., and Knauer, D.J. (1999) J. Biol. Chem. 274:275.
4Crisp, R.J., Knauer, D.J., and Knauer, M.F. (1999) submitted to J. Biol. Chem.

ROLE OF ARG129 AND LYS125 OF ANTITHROMBIN IN HEPARIN BINDING AND ALLOSTERIC ACTIVATION
Umesh Desai, Susan C. Bock, Ingemar Björk, Peter G. W. Gettins and Steven T. Olson. Center for Molecular Biology of Oral Diseases and Dept. of Biochemistry and Molecular Biology, Univ. of Illinois-Chicago, Chicago, IL, USA, Depts. of Medicine and Bioengineering, Univ. of Utah, Salt Lake City, UT, USA, Dept. of Veterinary Medical Chemistry, Swedish Univ. of Agricultural Sciences, Uppsala, Sweden.

The importance of Arg129 and Lys125 of antithrombin in binding a sequence-specific heparin pentasaccharide and in the allosteric activation of the serpin by the pentasaccharide was quantified from the effects of mutating these residues in recombinant antithrombins expressed in BHK or baculovirus-infected insect cells. Equilibrium binding studies showed that mutation of Arg 129 to His or Gln, the latter to mimic a natural antithrombin variant, decreased heparin pentasaccharide affinity 700-1300-fold at physiologic pH and ionic strength. Similarly, mutation of Lys 125 to the isosteric Met reduced heparin affinity 500-fold. These reductions in affinity demonstrate that both residues are major contributors to heparin binding free energy. The salt dependence of the equilibrium binding interactions indicated that mutation of Arg129 and Lys125 resulted in the loss of two and three ionic interactions, respectively, as well as nonionic interactions, suggesting that either basic residue functions cooperatively to promote multiple interactions with heparin. Rapid kinetic studies showed that the Arg129(His mutation minimally affected the initial low affinity interaction with heparin pentasaccharide but reduced the forward rate and increased the reverse rate of the subsequent conformational change leading to a high-affinity interaction and inhibitor activation. By contrast, the Lys125( Met mutation resulted in a significant reduction in the initial low heparin affinity binding step (25% of the total binding free energy change) as well as in defects in the forward and reverse rates of the conformational activation step. Binding of heparin pentasaccharide to both Arg 129 and Lys 125 variant antithrombins produced normal accelerations of factor Xa inhibition (300-fold), indicating that the mutations solely affected heparin binding and not the ability of the serpin to be allosterically activated. These results indicate that Arg129 and Lys125 are key residues which act cooperatively to establish the high-affinity interaction of antithrombin with heparin. While both residues are important for inducing antithrombin into the high heparin affinity state and for stabilizing this activated state, Lys125 is additionally critical for mediating the initial recognition of heparin by antithrombin.
Supported by NIH grants HL39888, HL30712, HL45486 and HL49234, Swedish Medical Research Council grant 4212 and European Community Biomed grant

BMH4-CT96-0937. MEASUREMENT OF a1-ANTITRYPSIN - IgA COMPLEXES IN BRONCHOALVEOLAR LAVAGE FLUID AND SERUM OF MILD ASTHMATIC SUBJECTS.
Summers C, Foreman R.C, Howarth P.H and Warner J.A

Increased levels of a1-antitrypsin-IgA complexes have been shown to be prognostic of disease severity in several inflammatory conditions, including rheumatoid arthritis. This study used a novel ELISA to detect and quantify these complexes in the bronchoalveolar lavage fluid (BALF) and serum of 31 mildly asthmatic subjects. We also examined the effect of six weeks therapy with two widely used inhaled asthma therapies, budesonide and nedocromil. 31 mild asthmatics underwent BALF and serum sample collection before being allocated to six weeks inhaled therapy with either budesonide, nedocromil or a placebo. Repeat BALF and serum samples were collected on completion of therapy. 96 well microtiter plates were coated with a polyclonal goat antibody to human a1-antitrypsin and samples incubated for two hours. Complexes were detected using a rabbit anti-human IgA horseradish peroxidase (HRP) conjugated antibody. A substrate for HRP was added and the colour development halted by the addition of 1M sulphuric acid. Samples were standarised to a pooled mixed BALF. a1-antitrypsin-IgA complexes were detected in both BALF and serum. There was no apparent relationship between the level of complexes in the BAL and those in serum, although there was a strong relationship between the amount of a1-antitrypsin-IgA complexes and the level of immunoreactive a1-antitrypsin detected in BALF (r =0.811, p<0.05). The level of complexes in the placebo group increased from a median of 95 to 153 AU/ml, while remaining unchanged in the nedocromil group ,median of 109 to 104 AU/ml, and decreasing slightly in the budesonide group, from a median of 76 to 58 AU/ml, though this did not attain statistical significance. In summary, we can detect a1-antitrypsin-IgA complexes within the BALF and serum of mild asthmatics, which suggests the presence of auto-antibodies. These auto-antibodies may be important in the regulation of a1-antitrypsin activity. The lack of relationship between the level of complexes in BALF and serum suggests that the a1-antitrypsin-IgA complexes are not simply derived from the serum. This study also found that two widely used asthma therapies have no significant modifying effects on the levels of a1-antitrypsin-IgA complexes within the BALF and serum.

IMPORTANCE OF THE RANDOM COIL BETWEEN HELIX D AND STRAND 2A IN HEPARIN ACTIVATION OF ANTITHROMBIN.
Jennifer L. Meagher, Steven T. Olson, and Peter G.W. Gettins, Department of Biochemistry and Molecular Biology, and Center for Molecular Biology of Oral Disease, University of Illinois at Chicago

The binding of pentasaccharide heparin to antithrombin induces a conformation change that greatly increases the rate of inhibition of the target proteinase factor Xa. Although the structure of antithrombin bound to pentasaccharide heparin has been solved (Jin et al, Proc. Natl. Acad. Sci., 94, 14683-14688, 1997), questions still remain as to the mechanism of the activating conformational change induced by heparin. The extension of helix D in the heparin bound state, as seen in the crystal structure, led us to hypothesize that the key event in heparin activation is movement of the linker connecting helix D to b-sheet A as a result of heparin-induced extension of helix D. To test this hypothesis, we created variant antithrombins in which we sequentially removed up to four amino acids (134-137) from the random coil connecting helix D and b-sheet A to simulate the effects of helix D extension. The variants retained the ability to form covalent complex with thrombin as seen on a SDS gel, indicating that the mutations did not adversely affect the serpin inhibitory mechanism. The initial fluorescence intensity of the variants was increased in comparison to control antithrombin from 1.12 to 1.4-fold, suggesting that the mutations altered the native conformation of the protein. In the presence of full length heparin, the 40% fluorescence enhancement seen for wild type antithrombin was reduced to 15% for a variant with 1 amino acid removed and completely abolished when 3 amino acids were removed. The mutations also reduced heparin affinity of the variants, from approximately 20 to 80-fold for full length heparin. The basal rate of inhibition of factor Xa was slightly decreased for all the variants, approximately 0.5-fold compared to control antithrombin. The most significant effect of the mutations was a large (3 to 10-fold) reduction in the rate of inhibition of factor Xa in the presence of heparin. Although the combined perturbations are not as expected from simple simulation of helix D extension, they do implicate this region in transmission of heparin-induced conformational effects to the reactive center loop. Supported by grant HL49234 from the National Institutes of Health P14E (S380E)

ANTITHROMBIN IS FULLY ACTIVATED WITHOUT HEPARIN
Akiko Futamura and Peter G. W. Gettins, Department of Biochemistry and Molecular Biology, University of Illinois at Chicago

Regulation of the inhibitory activity of antithrombin, the principal inhibitor of the blood clotting proteinases factor Xa and thrombin, is accomplished by binding to heparin. It has been proposed that, in the unactivated state, the reactive center loop of antithrombin has residues P15 and P14 incorporated as part of b-sheet A and with the side chain of residue P14 consequently buried in the protein interior. It has been further proposed that, upon heparin binding, P15 and P14 are expelled from the sheet, permitting a more appropriate conformation of the reactive center loop for interaction with factor Xa. We therefore created an antithrombin variant in which serine at position 380, 14 residues N-terminal from the reactive bond and at a hinge point in the structure, was replaced by glutamate to test the hypothesis that there is an equilibrium between the two states, P14 inserted and P14 expelled, and that changes in the rate of inhibition of factor Xa result from a shift in equilibrium between these states. In the absence of heparin, P14E antithrombin fluorescence is enhanced 30-40% compared to wild-type antithrombin. Although heparin binds tightly to P14E and gives a 2-3nm blue shift, there is no further increase in antithrombin fluorescence. This suggests that the P14 residue of the reactive center loop is not in contact with W225 and the reactive center loop is already expelled from b-sheet A. The P14E variant is an inhibitor of proteinases and has a more than hundred fold increased basal rate of inhibition of factor Xa, after correction for an increased SI. The rate is comparable to that of antithrombin fully activated by high affinity heparin pentasaccharide. Binding of full length heparin (HAH) causes only a 12-fold additional increase in rate. In contrast, the basal rate of inhibition of thrombin is similar to that of wild-type antithrombin. Addition of HAH increases the rate of inhibition of thrombin by 3000 fold, as expected from a bridging contribution. Taken together, these findings suggest that the serine-to-glutamate mutation in P14E antithrombin has shifted the equilibrium almost completely toward the P14-expelled conformation to give a fully activated inhibitor of factor Xa in the absence of heparin. Further increase in rate of inhibition of proteinase by binding HAH results principally from bridging contributions.
Supported by grant HL49234 from the National Institutes of Health. PROTEINASE DEPENDENT

CHANGES IN THE HORMONE BINDING SITE OF THYROXINE BINDING GLOBULIN.
Scott A. Suda, Peter G.W. Gettins, and Philip A. Patston. Department of Biochemistry and Molecular Biology, and Department of Oral Medicine and Diagnostic Sciences, University of Illinois at Chicago, Chicago, IL 60612, USA.

Thyroxine binding globulin (TBG) is the major carrier of the thyroid hormones triiodothyronine (T3) and thyroxine (T4) in plasma. Although TBG is a non-inhibitory serpin it converts to a more stable form after cleavage at the P4'-P5' Thr-Phe bond with neutrophil elastase, consistent with loop insertion into b-sheet A taking place (1). The proposed location for the hormone binding site is in the b-barrel formed by b-sheets B and C, and is thus adjacent to the reactive center loop (2). Such a location for the hormone binding site might be expected to be sensitive to reactive center loop cleavage and subsequent conformational change, however it was reported that no change in affinity for T4 resulted from cleavage with elastase. This is in contrast to a related serpin, cortisol binding globulin, for which a ten-fold reduction in affinity for cortisol after reactive center loop cleavage was demonstrated (1). We have reinvestigated the effect of TBG cleavage on hormone affinity by taking advantage of the specific binding of a fluorophore, 1,8-anilinonaphthalene sulfonic acid (ANS), to the hormone binding site of TBG (3). When bound to TBG there is a four-fold increase in the fluorescence intensity of ANS at 475 nm (excitation at 375 nm). By titrating ANS into TBG the Kd for native TBG was determined to be 0.2:M and for cleaved TBG to be 0.38:M. This provides the first evidence that reactive center loop cleavage causes a change in the environment of the hormone binding site such that affinity for ligand is altered. Evidence that ANS binds to the hormone binding site is given by the stoichiometric displacement of ANS by T4 as shown by the linear reduction in fluorescence to a 1:1 molar ratio of T4:TBG. The affinity of T3 for native and cleaved TBG was measured by displacement of ANS by T3 from 0.1mM TBG, as followed by the reduction in fluorescence. By this method the Kd of T3 for native TBG was determined to be 0.39nM and that for cleaved TBG to be 0.73nM. The total T3 concentration in serum is ~1.5nM, therefore a two-fold reduction in the Kd could lead to a large percentage increase in free T3. Binding of T3 to native TBG followed by reaction with elastase and analysis on SDS-PAGE showed that the TBG-T3 complex was resistant to cleavage by elastase, compared to TBG alone. Elastase cleaves at the P4'-P5' bond which is at the beginning of s1C, and as such is part of the hormone binding site. Protection from cleavage by T3 binding not only provides support for this region being part of the binding site, but also suggests that the in vivo cleavage site is not the P4'-P5' bond, and that elastase is not the proteinase which carries out this cleavage. Nevertheless, the proteinase-dependent release of T3 from TBG, as a result of a typical serpin conformational change, provides a rationale for a serpin being utilized as a T3 carrier. In addition, it provides another example of the versatility of the native metastable serpin structure, with the consequence that there are unique opportunities for regulation of the biological function of serpins, as a result of the conformational states the serpin can adopt.
1. Pemberton, P.A., Stein, P.E., Pepys, M.B., Potter, J.M., and Carrell, R.W. (1988) Nature 336, 257-258. Hormone binding globulins undergo serpin conformational change in inflammation.
2. Jarvis, J.A., Munro, S.L.A., and Craik, D.J. (1992) Prot. Eng. 5, 61-67. Homology model of thyroxine binding globulin and elucidation of the thyroid hormone binding site.
3. Green, A.M., Marshall, J.S., Pensky, J., and Stanbury, J.B. (1972) Science 175, 1378-1380. Thyroxine binding globulin: Characterization of the binding site with a fluorescent dye as a probe.

AMINO ACID RESIDUES OF HEPARIN COFACTOR II REQUIRED FOR STIMULATION OF THROMBIN INHIBITION BY SULPHATED POLYANIONS
Niall S. Colwell 1, Michael J. Grupe 2 and Douglas M. Tollefsen 2 1 Department of Pharmacology and Therapeutics, University College, Cork, Ireland. 2 Division of Hematology, Department of Internal Medicine, Washington University School of Medicine, St. Louis, Missouri 63110, U.S.A.

A variety of sulphated polyanions in addition to heparin and dermatan sulphate stimulate the inhibition of thrombin by heparin cofactor II (HCII). Previous investigations indicated that the binding sites on HCII for heparin and dermatan sulphate overlap but are not identical. In this study we determined the concentrations (IC50) of various polyanions required to stimulate thrombin inhibition by native recombinant HCH in comparison with three recombinant HCII variants having decreased affinity for heparin (K173Q), dermatan sulphate (R189H), or both heparin and dermatan sulphate (K185N). Pentosan polysulphate, sulphated bis-lactobionic acid amide, and sulphated bis-maltobionic acid amide resembled dermatan sulphate, since their IC50 values were increased to a much greater degree (>8-fold) by the mutations R189H and K185N than by K173Q (<1.5-fold). By contrast, the IC50 values for fucosylated chondroitin sulphate, chondroitin sulphate E, dextran sulphate, and fucoidan were minimally affected. Only in the case of heparin was the 1C50 increased to a greater degree by both K173Q and K185N (>6-fold) than by R189H (<1.5-fold). Noneofthe polyanions significantly stimulated inhibition of thrombin by an N-terminal deletion mutant of HCII (del 1-74). These results suggest that, like dermatan sulphate and heparin, other polyanions stimulate HCII primarily by an allosteric mechanism requiring the N-terminal acidic domain. The Unfavorable Reversible Acylation Mechanism for Serpin Action Demonstrated for PAI-1 Inhibition of uPA Jan-Olov Kvassman, Ingrid Verhamme and Joseph D. Shore Henry Ford Health Sciences Center, Division of Biochemical Research, Detroit, Michigan We recently demonstrated that when loop insertion in PAI-1 is blocked by binding an octapeptide analog of the P14-P7 part of its reactive center loop (RCL) to position 4 in b-sheet A, it acylates tPA more than 50 times slower than unmodified PAI-1 (0.06 versus 3 s-1), whereas trypsin is acylated at the same rate (20 s-1) by the two forms (1). The peptide blocked PAI-1 acts as a substrate that is completely hydrolyzed at the reactive center by catalytic amounts of proteinase. We attributed the discrepancy between tPA and trypsin to residues in the tPA '37-loop' (VR1) which are absent or different in trypsin and which retain specifically the distal part of the PAI-1 RCL in the tPA substrate pocket, promoting a reversal of the acylation step in the 'docking' complex. We argued that if loop insertion is blocked, slow passive displacement of the 'exosite' bonds determines the rate of acylation and enzyme turnover. With unmodified PAI-1, tPA can be pulled from its docking site by the highly exergonic step of loop insertion which forces acylation at a much higher rate. We suggested that a previously implicated salt bridge between P4' Glu in PAI-1 and Arg-39 in tPA (2) has a decisive role in promoting the Michaelis complex over the acyl enzyme. We have measured the effect of blocking loop insertion in PAI-1 on its rate of acylation of uPAa, using rapid quenching combined with SDS-PAGE and photometric techniques. With unmodified PAI-1 the rate of uPA acylation was 17 s-1 whereas when loop insertion is blocked this rate was reduced to 0.7 s-1. This rate reduction is consistent with the fact that uPA, as opposed to trypsin, carries a 37-loop very similar to that of tPA, including a stretch of 4 positively charged residues. As for tPA, the uPA 37-loop confers much of the enzyme's specificity for PAI-1 (3). The transient accumulation of the acyl enzyme intermediate was clearly observed in the reaction of uPA with the peptide-blocked substrate PAI-1 form. The ratio between the intrinsic forward and reverse rate constants of the acylation step was evaluated to 0.2, based on the ratio of the acyl enzyme to the Michaelis complex at steady-state. The acyl enzyme intermediate was not observed in the reaction of PAI-1 with saturating tPA. Hence, our data indicates that although PAI-1 acylation of both enzymes is intrinsically unfavorable, this is much more pronounced when tPA is the target. The fact that uPA cannot form the salt bridge to the P4' Glu (it has a threonine at position 39) offers an immediate explanation for this difference. This would explain also the overall higher rates observed with uPA than with tPA and the 2.5-fold smaller effect of blocking loop insertion on the rate of acylation. Subjected to the same forces, the distal RCL fragment dissociates faster from the substrate pocket of uPA than that of tPA. aHigh molecular weight uPA was generously provided by Dr. Donald E. Eisenhauer at Abbott Laboratories, North Chicago, Illinois. 1. Jan-Olov Kvassman et al., (1998) Biochemistry 37, 15491-15502 2. Ed L. Madison et al. (1989) Nature 339, 721-724 3. John D. Sipley et al. (1997) Proc. Natl. Acad. Sci. USA 94, 2933-2938 PLASMINOGEN ACTIVATOR INHIBITOR TYPE 2 IN CHRONIC VENOUS LEG ULCERATION C.L. Bunn1, S.J. Wysocki2, N.J. Trengove2, Y.K. Vandongen2, and M.C. Stacey2 1Biotech Australia, PO Box 20, Roseville, NSW, 2069, Australia; 2University Department of Surgery, Fremantle Hospital, WA, 6160, Australia. Plamsinogen activator inhibitor type 2 (PAI-2) is a serpin most commonly found in monocytes, pregnancy, plasma, and skin. It is a specific inhibitor of urinary plasminogen activator (uPA), and to a lesser extent, tissue plasminogen activator (tPA). Several publications have suggested a role for uPA in the tissue proteolysis which accompanies chronic venous leg ulceration, and consequently PAI-2 has therapeutic potential for this disease. A pilot scale clinical trial was performed in a double blind, placebo controlled design, primarily to assess the safety of PAI-2. PAI-2 was administered topically once per day for five consecutive days, and patients followed for 28 days. All 20 patients with chronic venous leg ulcers received compression bandaging, and 11 patients received PAI-2, with 9 receiving placebo. Samples of wound fluid were taken over the first day of PAI-2 administration, and once on five other trial days. There were no adverse events related to PAI-2. The uPA activity levels in wound fluid decreased over 6 hours following PAI-2 administration, but returned to previous levels by 24 hours. The median ulcer area in the PAI-2 treated group decreased about 24% over 28 days, whereas the placebo group increased slightly (p = 0.059). These results suggest that PAI-2 may be beneficial in the treatment of chronic venous leg ulcers, and a larger clinical trial is in progress.
This work was supported by Biotech Australia Pty Ltd.

EVIDENCE FOR THE POLYMERIZATION OF Z & SIIYAMA VARIANT a1-ANTITRYPSIN WITHIN THE SECRETORY PATHWAY OF XENOPUS LAEVIS OOCYTES D. J. Oakley & R. C. Foreman. University of Southampton, School of Biological Sciences, Bassett Crescent East. SO16 7PX. UK.

The mechanism of loop-sheet polymerisation has been shown to have an important role in the formation of aggregates of a1-antitrypsin that occur in the liver and lungs of patients with a1-antitrypsin deficiency (1,2). Crystallographic analysis and predictive molecular modelling have interpreted the effects of point mutations known to cause a clinical deficiency in terms of their ability to promote loop sheet insertion (3,4). However, while loop-sheet polymers may undoubtedly form in the endoplasmic reticulum lumen of hepatocytes, is this the direct cause of the secretory blockade? In this study canine microsomal membranes (cmm) were used in an in vitro translation system and injected into Xenopus oocytes to reconstruct the secretory pathway (5). This surrogate system was used to examine the relative secretion of M a1-antitrypsin and the deficiency variants Z and Siiyama synthesised in vitro compared to that produced from in ovo injection of mRNA (6). Microsomes were added to a cell free reticulocyte lysate and a1-antitrypsin mRNA translated in the presence of [35S]-methionine to produce a 49kDa protein sequestered in the microsomal vesicles. Microsomes were retrieved by sucrose density gradient centrifugation and the resuspended material injected into Xenopus oocytes. Secretion of radio labelled protein was assessed by fluorography of SDS gels. In parallel experiments a membrane fraction was prepared from RNA injected oocytes and analysed by native PAGE and western blotting. The results show that the secretion of M a1-antitrypsin (37%) is significantly higher (p>0.01) than that of the Z (17%) and Siiyama (9%) variants from oocytes injected with canine microsomes. This demonstrates that exogenous microsomes can integrate successfully with components of the oocyte secretory machinery and that the secretory phenotype of a1-antitrypsin variants is maintained under these conditions. We could detect no high molecular weight forms of a1-antitrypsin within microsomes prior to injection into oocytes suggesting that if loop sheet aggregation is responsible for the secretory blockade it must occur subsequently. Further work has shown that high molecular weight forms of a1-antitrypsin can be detected in a membrane fraction from oocytes injected with variant Z and Siiyama but not M encoding mRNA (Fig. 1). This novel secretory system may be useful in examining the early post translational events of a1-antitrypsin biosynthesis and that of other proteins with aberrant handling in the ER. Fig. 1. Western blot analysis showing the presence of higher molecular weight forms of a1-antitrypsin protein in the ER of oocytes injected with Z and Siiyama, but not M mRNA. OO indicates retrieved oocyte ER fragment, S indicates secreted protein, Cntl lane oocytes received no injection.
1. Lomas et al., (1990). Nature 357, 605-607
2. Elliott et al., (1998). Am. J. Respir. Cell. Mol. Biol., 18, 419-425
3. Lomas & Carrell (1993). Am. J. Physiol., 265, 211-219
4. Stein & Carrell (1995). Struct. Biol., 2, 96-112
5. Paiement et al., (1990). Am. J. Anat., 187, 183-192
6. Sidhar et al., (1995). J. Biol. Chem. 270, 8393-8396

EXPRESSION, PURIFICATION AND CRYSTALLISATION OF RECOMBINANT HUMAN ANGIOTENSINOGEN.
Peter Stanley, Rosa Streatfeild-James and Penny Stein. Department of Structural Medicine, Cambridge Institute of Medical Research, Hills Road, Cambridge, UK.

The renin-angiotensin system is a major regulator of salt and water homeostasis and has a key role in the control of blood pressure. Angiotensinogen is the only known substrate for renin, a member of the aspartyl proteinase family of proteins, and the precursor to the angiotensin peptides which include one of the most potent vasoactive hormones, angiotensin-II. Angiotensinogen, a member of the serpin family, does not undergo the stressed to relaxed transition typical of inhibitory serpins and has not been shown to function as a serine proteinase inhibitor. This suggests that the positioning and configuration of the reactive site loop of angiotensinogen is likely to resemble that of ovalbumin, the best studied non-inhibitory serpin, but the precise three-dimensional structure of angiotensinogen is unknown. An aim of this work is to produce recombinant angiotensinogen suitable for crystallisation and diffraction. The human angiotensinogen gene was PCR amplified from a liver cDNA library and subcloned with a C- terminal (His)6 fusion into an expression vector under the control of the metallothionein promoter. Following stable transfection and expression in Baby Hamster Kidney (BHK)-21 cells, the fusion protein was purified by a combination of anion exchange and nickel affinity chromatography. As we found previously with angiotensinogen obtained from plasma, angiotensinogen expressed in BHK cells provided material acceptable for functional assays but it did not form crystals suitable for diffraction. Human angiotensinogen is a globular glycoprotein with four putative asparagine-linked glycosylation sites. The plasma protein has a molecular weight 55-65kDa and p14.3-4.9, depending on the degree of post-translational N-glycosylation, with carbohydrate accounting for 13-14% of its mass. Differences in the level of glycosylation may contribute to heterogeneity of the protein sample and the lack of suitable crystals. Non-glycosylated protein is generally more appropriate for this purpose since it is more homogeneous. A system for the expression of recombinant angiotensinogen in Escherichia coli was therefore established using the pET-22b T7 promoter vector containing a pelB leader sequence to direct angiotensinogen to the bacterial periplasm. However, angiotensinogen largely remained in the cytoplasm, retaining the pelB sequence. A construct was made without the pelB sequence but keeping the C-terminal (His)6-tag to enable purification of angiotensinogen from the bacterial cytoplasm. This system expressed a single 50kDa form of soluble angiotensinogen at the level of 1mg/litre culture. Kinetic studies showed that the lack of glycosylation did not affect either the efficient processing by renin or the thermal stability of the recombinant angiotensinogen. Further purification and crystallisation trials are being performed. The solution of a crystal structure for angiotensinogen will provide valuable insight into the function of this non-inhibitory prohormone serpin. It will allow modelling of the angiotensinogen-renin complex and prediction of those residues important in this key interaction of the renin-angiotensin system. A crystal structure of angiotensinogen will also allow predictions to be made about the functioning of angiotensinogen variants.

THE EFFECT OF A REDUCING END EXTENSION ON PENTASACCHARIDE BINDING BY ANTITHROMBIN
Klara J. Belzar, Timothy Dafforn, Maurice Petitou, Jean Marc Herbert, Robin W. Carrell and James A. Huntington. Department of Haematology/Division of Structural Medicine, University of Cambridge, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, Hills Road, Cambridge, CB2 2XY. Haemobiology Research Department, Sanofi Recherche, Toulouse, France.

Most of the anticoagulant activity of the widely prescribed drug heparin is exerted by a unique pentasaccharide sequence (denoted DEFGH) which binds tightly to plasma antithrombin (AT). Binding is a two-step process resulting in an activating conformational change and an increase in inhibitory activity towards its target protease Factor Xa. The crystal structure of a high-affinity pentasaccharide bound to AT (Jin, et al., 1997) mapped the binding domain lower down helix D than had been predicted (van Boeckel, et al., 1994), indicating that the C-terminal end of helix D was more likely involved in the binding of non-reducing end extensions present in full-length heparin. Recent binding studies with truncated pentasaccharides have indicated that non-reducing end residues (DEF) are crucial for the initial binding step and the reducing end residues, G and H, stabilise the activated, high-affinity conformation (Desai, et al., 1998). To further characterise the role of the reducing end in heparin binding we studied the effects of a reducing end extension on pentasaccharide binding to AT. For this purpose, a synthetic heptasaccharide (denoted DEFGHGH) was created by the addition of a GH disaccharide to the reducing end of the core pentasaccharide. The pentasaccharide and heptasaccharide produced identical enhancements of plasma AT fluorescence and rate of Factor Xa inhibition, indicating that the activating conformational change was not affected by the disaccharide extension. Pentasaccharide and heptasaccharide bound AT with identical affinities at low ionic strengths I<0.2, resulting in a slope of 3.3 and y-intercept of -4.6 from the linear fit of the log of the dissociation constants against the log of the ionic strengths. However, at higher ionic strengths 3I=0.2 heptasaccharide bound AT with a 2-fold greater affinity than physiological pentasaccharide (Kd = 251Ò 32.4 and 481Ò 7.3 nM at I=0.3, respectively). The slope was unaffected at I>0.2, but the y-intercept of -4.8 indicated that the increased affinity was attributable to an increase in the strength of the non-ionic interactions. Stopped flow studies gave similar rates at I=0.15, however, at I=0.3 the rate of dissociation was significantly decreased with heptasaccharide binding to AT. We propose a model to account for this second mode of heptasaccharide binding, where the heptasaccharide shifts two residues towards the C-terminal end of helix D thus occupying the extended heparin binding site. Molecular modelling studies support our frameshift model by predicting new non-ionic interactions between Lys133 and residues D and E, and by replacing the ionic interactions to the 3-0-sulphate on residue F with hydrogen bonds.
1. Jin, L., Abrahams, J.P., Skinner, R., Petitou, M., Pike, R.N., and Carrell, R.W. (1997) Proc. Natl. Acad. Sci. USA. 94, 14683-14688.
2. van Boeckel, C.A.A, Petitou, M. (1994) Structural biology. 1, 423-425.
3. Desai, U.R., Petitou, M., Bjîrk, I. and Olson, S.T. (1998) J. Biol. Chem. 273 (13), 7478-7487.

ALPHA1-ANTICHYMOTRYPSIN DEFICIENCY: CONFORMATIONAL TRANSITIONS IN CHRONIC AIRFLOW OBSTRUCTION.
B. Gooptu, W-S.W. Chang, D.A. Lomas. Department of Medicine, University of Cambridge, and Division of Structural Medicine, Cambridge Institute for Medical Research, Hills Road, Cambridge, CB2 2XY, UK

Alpha1-antichymotrypsin is a proteinase inhibitor of the serpin superfamily, similar to a1-antitrypsin both in terms of its overall structure and its function. It protects the lungs from proteolytic attack by inhibiting enzymes with its mobile reactive centre loop. The importance of this role is underscored by our recent observation that the protein is in an inactive, latent conformation in the lungs of patients with chronic bronchitis and emphysema (J. Biol. Chem. 1998; 273:3695-3701). Furthermore, two deficiency variants of a1-antichymotrypsin (229ProrAla and 55LeurPro) have been associated with chronic airflow obstruction. The effect of these mutations on the structure and function of a1-antichymotrypsin has been assessed. The variant a1-antichymotrypsins were prepared by site-directed mutagenesis, and then expressed in E. coli, purified to homogeneity, and characterised. 229ProrAla a1-antichymotrypsin displayed normal kinetic and biochemical characteristics with the exception of a greatly increased tendency to form reactive looprbeta pleated sheet linked polymers, comparable to that seen in severe deficiency variants of a1-antitrypsin. 55LeurPro a1-antichymotrypsin purified as 3 non-cleaved conformers: an active, native form and an inactive, more stable species shown to be the latent protein with its reactive-loop inserted intramolecularly; the third conformation, also inactive, was one previously uncharacterised within the serpin superfamily and showed properties intermediate between the native and latent conformations. Taken together these data show that a1-antichymotrypsin deficiency results from conformational transitions inactivating the protein as an inhibitor and exacerbating lung disease.

THE INTERACTION OF NATIVE, LATENT AND CLEAVED PLASMA ALPHA-1-ANTITRYPSIN WITH HUMAN NEUTROPHIL DEFENSIN.
P Sivasothy 1, TR Dafforn1, PS Hiemstra 2 and DA Lomas1. [1] Respiratory Medicine Unit, Dept. of Medicine, University of Cambridge, Hills Rd, Cambridge, UK; [2] Dept. of Pulmonology, Leiden University Medical Center, The Netherlands.

Upon activation neutrophils release the content of their azurophilic granules, which includes human neutrophil elastase (HNE) and human neutrophil defensin (human neutrophil peptide, HNP), into the local microenvironment. HNP is a member of a family of amphipathic cysteine and arginine rich peptides (29-34 amino acids) that are in a predominantly b sheet structure. Defensins have antibacterial, antifungal, antiviral, cytotoxic, chemotactic and opsonising actions. If unrestrained HNE and HNP will cause host tissue damage. It has been suggested that a1-antitrypsin has a role in regulating the inflammatory processes of these azurophilic granule components (Am J Respir Cell Mol Biol 1995 12:351-357). Alpha-1-antitrypsin is the archetypal member of the serpin superfamily and may adopt conformations other than its native state: reactive loop cleaved a1-antitrypsin has the cleaved loop inserted into the A b-sheet and latent a1-antitrypsin has the intact loop inserted into the A b-sheet. We report on the effect of the HNP on plasma M, S and Z a1-antitrypsin and the latent and cleaved conformations of M a1-antitrypsin. HNP reduces the inhibitory activity of M, S and Z a1-antitrypsin against bovine a1-chymotrypsin in a concentration dependent manner from a 1:1 molar ratio, with complete loss of activity at 30:1. Incubation of HNP-1 with M, S, Z and latent a1-antitrypsin in 30:1 molar ratio at 37¯C resulted in the loss of native and latent a1-antitrypsin bands on non-denaturing PAGE gels and the appearance of discrete ladder bands on acid PAGE gels. This interaction was monitored by intrinsic tryptophan fluorescence which showed an exponential decay over time with the rate dependent on the concentration of HNP. Native a1-antitrypsin underwent a more rapid conformational change than the latent form when incubated at 0.25 mg/ml with HNP at 1:1 to 30:1 molar ratio ( HNP: a1-antitrypsin) at 37¯C . Cleaved a1-antitrypsin was more resistant to HNP when compared to either M a1-antitrypsin or latent a1-antitrypsin, and incubation revealed a slower decline in intrinsic tryptohan fluorescence and a slower loss of cleaved band on native PAGE. HNP has with the ability to disrupt lipid membranes in a detergent like manner due to its amphipathic structure. HNP did not mediate its effect on of a1-antitrypsin by a detergent action as incubation of a1-antitrypsin with the non-ionic detergents NP-40 and Ipegal ( 1-5%v/v) at 37¯C resulted in discrete polymer bands on native PAGE that had a different appearance on acid PAGE from the multimer of a1-antitrypsin with HNP. The antimicrobial function of HNP has been shown to be impaired in the explanted airways of cystic fibrosis patients. It has been proposed that this results from raised intra-airways salt concentration. In vitro studies of the effect of NaCl on the antibacterial activity of HNP now support this mechanism. Incubation of a1-antitrypsin with HNP at 1:30 molar ratio revealed that there was a slower loss of the native a1 antitrypsin band and decline in intrinsic fluorescence as the salt concentration was increased from 0 to 300 mM NaCl. Taken together our data shows that the interaction of a1-antitrypsin with HNP is dependent on conformation of a1-antitrypsin, salt concentration and is distinct from the actions of non-ionic detergents.
This work was funded by the MRC (UK) and Wellcome Trust.

PROPAGATION OF INACTIVATION OF ANTITHROMBIN AND THE ONSET OF THROMBOSIS
Aiwu Zhou, James Huntington, Robin Carrell Department of Haematology, University of Cambridge, CIMR Wellcome Trust IMRC Building, Hills Road CB2 2XY

We previously showed how the onset of thrombosis could result from a change in conformation of antithrombin to its monomeric inactive latent form [Blood, 92:2696-2706, 1998]. Nevertheless there is still a discrepancy between the severity and nature of the thrombotic events associated with such conformational variants, as compared to thromboses resulting from a simple genetic deficiency of antithrombin. A genetic deficiency with a 50% deficit in plasma antithrombin, results in an increased risk of thrombosis but this is usually not life-threatening, particularly before full adult life. By comparison, the presence of a conformationally unstable antithrombin, with a much less deficiency of inhibitory activity, predisposes to atypically sudden and severe thromboses. Our new findings, presented here, explain this discrepancy by showing that the transition of a molecule of the abnormal antithrombin to the latent conformation is accompanied by its immediate linkage to a molecule of normal antithrombin to give a dimer with an eletrophoretic mobility close to that of normal a-antithrombin. The formation of the dimer results in inactivation of the molecule linked to the latent antithrombin, though since the heterodimer readily dissociates there is only a 13% additive loss of activity with a-antithrombin. With b-antithrombin, which constitutes 5- 10% of the total plasma antithrombin and is the primary inhibitor of thrombosis in circulation, the dissociation is much less marked with the additive loss of activity being 21% rising to 33% on stabilisation of the dimer with heparin. The potential physiological consequences of the combination of these effects is particularly significant with b-antithrombin, where the likely in vivo concentrations (latent to b-antithrombin is 10:1) result in a 65% loss of activity of the heparin-bound inhibitor. This selective and additive loss of activity of b- antithrombin provides an explanation for the unexpected severity of thrombotic episodes in heterozygotes with conformationally unstable antithrombins. The identification of this immediate formation of an electrophoretically observed dimer explains the previous difficulties in detecting the transition of antithrombin to the latent conformation in various experimental systems, including in crystallisation trials and in therapeutic concentrations during pasteurisation. A kinetic mechanism for the polymerisation of a1-antitrypsin T. R. Dafforn1, R. Mahadeva1,2, P. R. Elliott1,2, P. Sivasothy1,2, D. A. Lomas1,2 Department of 1Haematology & 2Respiratory Medicine Unit, Department of Medicine, Cambridge Institute for Medical Research, University of Cambridge, MRC Centre, Hills Road. Cambridge. CB2 2XY UK. The mutation in the Z deficiency variant of a1-antitrypsin perturbs the structure of the protein to allow a unique intermolecular linkage. These loop-sheet polymers are retained within the endoplasmic reticulum of hepatocytes to form inclusions which are associated with neonatal hepatitis, juvenile cirrhosis and hepatocellular carcinoma. The process of polymer formation has been investigated here by intrinsic tryptophan fluorescence, fluorescence polarisation, circular dichroic spectra and extrinsic fluorescence with 8-anilino-1-naphthalenesulphonic acid (ANSA) and tetramethylrhodamine-5-iodoacetamide (5-TMRIA). These biophysical techniques have demonstrated that a1-antitrypsin polymerisation is a two stage process and have allowed the calculation of rates for both of these steps. The initial fast phase is unimolecular and likely to represent temperature induced protein unfolding whilst the slow phase is bimolecular and associated with loop-sheet interaction and polymer formation. The naturally occurring Z,S and I variants and recombinant site-directed reactive loop and shutter domain mutants of a1-antitrypsin were used to demonstrate the close association between protein stability and rate of a1-antitrypsin polymerisation. Taken together these data allow us to propose a kinetic mechanism for a1-antitrypsin polymer formation which involves the generation of an unstable intermediate which can form polymers or generate latent protein.

HETEROPOLYMERISATION OF S, I AND Z a1-ANTITRYPSIN AND LIVER CIRRHOSIS
Ravi Mahadeva1, Wun-Shaing W. Chang1, Timothy R. Dafforn1, Diana J. Oakley3, Richard C. Foreman3, Jacqueline Calvin2, Derek G.D. Wight4, David A. Lomas1 1Respiratory Medicine Unit, Department of Medicine and Department of Haematology, Cambridge Institute for Medical Research, University of Cambridge, Departments of 2Clinical Biochemistry and 4Pathology, Addenbrooke's NHS Trust, Hills Road, Cambridge, CB2 2QH UK and 3Division of Cell Sciences, School of Biological Sciences, University of Southampton, Bassett Crescent East, Southampton, SO9 3TU, UK.

The association between Z a1-antitrypsin deficiency and juvenile cirrhosis is well recognised and there is now convincing evidence that the hepatic inclusions are the result of entangled polymers of mutant Z a1-antitrypsin. Four percent of the northern European Caucasian population are heterozygotes for the Z variant but even more common is S a1-antitrypsin which is found in up to 28% of southern Europeans. The S variant is known to have an increased susceptibility to polymerisation although this is marginal compared with the more conformationally unstable Z variant. There has been speculation that the two may interact to give cirrhosis but this has never been demonstrated experimentally. This hypothesis was raised again by the observation reported here of a mixed heterozygote for Z a1-antitrypsin and another conformationally unstable variant (I; 39ArgrCys) identified in a 34 year old man with cirrhosis related to a1-antitrypsin deficiency. The conformational stability of the I variant has been characterised and we have used fluorescence resonance energy transfer to demonstrate the formation of heteropolymers between S and Z a1-antitrypsin. Taken together these results indicate that not only may mixed variants form heteropolymers but that this can causally lead to the development of cirrhosis.
This work was supported by Anglia and Oxford Regional Health Authority and the Cystic fibrosis trust (UK)

EFFECTS OF BINDING OF PLASMINOGEN ACTIVATOR INHIBITOR-1 TO VITRONECTIN: FORMATION OF LARGE MULTIVALENT COMPLEXES.
Kenneth H. Minor*, Shannon Eaker*, Ingrid Verhamme#, Joseph D. Shore#, and Cynthia B. Peterson*, #Dept. of Biochemistry, Henry Ford Health Systems, Detroit, MI 48202, and *Dept. of Biochemistry and Cellular and Molecular Biology, Univ. of Tennessee, Knoxville, TN 37996

Plasminogen activator inhibitor type-1 (PAI-1) is a serine protease inhibitor that is the physiological inhibitor of urokinase and tissue-type plasminogen activators. As such, PAI-1 represents a key regulatory component of fibrinolysis, a biological process important to homeostasis, cellular migration, tissue repair, and tumor metastasis. Unlike other serine protease inhibitors, PAI-1 is unstable in its active conformation, and readily converts to a stable, but inactive conformation. However, PAI-1 is stabilized in its active conformation through an interaction with the abundant glycoprotein vitronectin. Furthermore, nearly all PAI-1 found in vivo is bound to vitronectin and therefore in the active form. Recent work from this laboratory using equilibrium analytical ultracentrifugation has demonstrated that PAI-1 binds to vitronectin with an apparent 2:1 stoichiometry and associates to higher order species (M ~320,000) comprised of 4 PAI-1 and 2 vitronectin molecules (Blackburn, M. N., Podor, T. J., Shaughnessy, S. G., and Peterson, C. B., submitted). Further characterization of the kinetics of formation and breakdown of these complexes and effects of the higher order complexes on function of vitronectin and PAI-1 have been pursued. These efforts demonstrate apparent cooperative binding interactions between PAI-1 and vitronectin, producing high-molecular weight species that exhibit altered cellular adhesive properties. Binding of vitronectin to PAI-1 was monitored from changes in fluorescence of an NBD probe attached covalently to a unique cysteine introduced by site-directed mutagenesis of PAI-1 at position 119. This interaction is associated with large increases in fluorescence which result from an increase in hydrophobicity in the vicinity of the fluorescent probe. Binding isotherms are sigmoidal in shape, indicating positive cooperativity between the two proposed PAI-1-binding sites on vitronectin. Size exclusion HPLC was used to monitor complex formation following the mixing of various ratios of PAI-1:vitronectin to evaluate the time course of complex formation and the sizes of species produced. Shortly after mixing, complexes of approximately 160 kDa and 320 kDa are present, along with a higher molecular weight component. The 160 kDa complex observed is consistent with the idea that PAI-1 and vitronectin initially associate into a 2:1 stoichometric complex, and the higher molecular weight species represent higher order complexes of vitronectin and PAI-1. Sedimentation velocity experiments in the analytical ultracentrifuge, analyzed using the time-derivative (g*S) method, were also conducted to evaluate sizes of complexes that form in mixtures of vitronectin and PAI-1. Consistently, sedimenting species that correspond to 1:2 and 2:4 complexes of vitronectin:PAI-1 are observed. Does the self-association of vitronectin into the higher order complexes alter the physiological properties of vitronectin? What is the effect of PAI-1 binding on the cellular adhesion function of vitronectin? Effects of PAI-1-induced formation of higher order complexes on the adhesive functions of vitronectin were investigated using assays that measure binding to the integrin, GP IIbIIIa. Microtiter plates were coated with the integrin GP IIbIIIa and then incubated with varying concentrations of vitronectin in the absence and presence of PAI-1. Following incubation and a series of washes, vitronectin bound to integrin was evaluated using a monoclonal antibody for vitronectin. In conjunction with association into multivalent complexes, the binding of vitronectin to the integrin was enhanced in the presence of PAI-1. These findings extend the current thinking regarding PAI-1 and its regulation of adhesion properties of vitronectin by demonstrating that the multivalent, higher order complexes have apparent higher affinity for cell surface receptors. From these studies, it appears that PAI-1 binds to vitronectin in a cooperative manner, leading to the formation of multivalent complexes that enhance the adherent properties of vitronectin.

EFFORTS TO INCREASE ADHESION OF TRANSDUCED ENDOTHELIAL CELLS USING A RETROVIRAL VECTOR CONTAINING PLASMINOGEN ACTIVATOR INHIBITOR-1 OR VITRONECTIN.
Christine R. Schar and Cynthia B. Peterson, Department of Biochemistry, Cellular and Molecular Biology, The University of Tennessee, Knoxville, TN 37996

Synthetic vascular grafts are routinely employed in reconstructing arteries. It has been observed that these synthetic vascular grafts have a limited life span in vivo. Attempts to increase the life span of these grafts have utilized endothelial cells to seed the surface of the synthetic graft. The use of a retrovirus for gene therapy has been investigated as a potentially effective way to transduce seeded cells with genes that could reduce the rate of graft failure by regulating cellular proliferation or controlling the inflammation associated with graft failure. It has been observed that synthetic grafts seeded with cells transduced by a retrovirus exhibit only limited endothelialization whereas non-transduced counterparts exhibit endothelialization as early as two weeks. The transduced endothelial cells fail to express intact b1 or b3 integrin subunits. Since integrins are important in mediating interactions between the cell and the extracellular matrix (ECM), this suggests that the transduced endothelial cells have impaired adhesive ability potentially due to the loss of integrin integrity. Can over-expression of proteins that regulate cell:matrix interactions be effective at improving the adhesive properties of transduced endothelial cells? Vitronectin and plasminogen activator inhibitor-1 (PAI-1) have both been implicated in regulation of cell interactions with the ECM of the vasculature. Therefore, both vitronectin and PA1-1 have a potential therapeutic value in restructuring the vasculature by mediating cellular adhesion and migration or pericellular proteolysis. The balance between PAI-1 and vitronectin in the vasculature can influence the adhesive state of the cell. Can this balance be exploited to confer an adhesive character to transduced cells? To this end, endothelial cells have been transduced with either PAI-1 or vitronectin. Transduced endothelial cells will secrete either PAI-1 or vitronectin into the culture medium. Secreted vitronectin from transduced endothelial cells may cause the cells to become more adherent by providing a surface to promote cell adhesion. Alternatively, the elevated levels of vitronectin may lead to an up-regulation of integrin expression. On the other hand, PAI-1 secreted from the transduced endothelial cells may decrease the level of active serine proteases and thereby decrease proteolysis of integrins or the ECM. However, PAI-1 may prevent interaction of the urokinase receptor on cells with vitronectin in the matrix by competing for overlapping binding sites in the somatomedin B domain of vitronectin. Thus, the effects of PAI-1 on integrins would be expected to increase cellular adhesion, while effects on the urokinase receptor would decrease the adhesive ability of the cells. Human umbilical vein endothelial cells (HUVEC) have been transduced with recombinant retroviral vectors containing either PAI-1 or vitronectin. The transduced HUVEC cells were screened to confirm the introduction of the target genes by Southern analysis as well as polymerase chain reaction. Expression of intact message for either of the proteins was detected by Northern blotting. Western blots using antibodies specific for a2, a5, av, b1 and b3 integrin subunits were performed to evaluate the integrity of the a and b subunits in the transduced cells.

DEVELOPMENT AND PROPERTIES OF NEUTROPHIL-RESISTANT HUMAN ANTITHROMBINS
Pedram Zendehrouh, Aiqin Lu, Yancheng Zuo, Veronique Picard, Mohamad Aman Jairajpuri, Theodore G. Liou and Susan C. Bock, University of Utah, Departments of Medicine and Bioengineering, Salt Lake City, UT, USA, 84132.

The reactive loop of human antithrombin III (ATIII) is sensitive to cleavage by neutrophil elastase. Elastase inactivation of infused plasma-derived ATIII concentrates may limit the efficacy of ATIII supplementation in inflammatory disorders, including sepsis and DIC. We have developed vessel wall and matrix-directed neutrophil-resistant antithrombins (NR-ATIIIs) by disabling the antithrombin asparagine-135 N-glycosylation consensus sequence to generate b-ATIII-like molecules with high affinity for heparin and heparan sulfate proteoglycans, and by modifying the reactive loop to introduce elastase resistance. Initial substitutions were based on natural thrombin recognition sequences predicted to be elastase resistant (e.g., fibrinogen, a1-proteinase inhibitor Pittsburgh, factor XIII and prothrombin), and additional modifications were introduced in subsequent rounds of mutagenesis to improve inhibition and stability properties further. Several recombinant antithrombins with preserved anticoagulant activity and increased resistance to inactivation by elastase and neutrophils were obtained. Compared to plasma antithrombin, one NR-ATIII mutant exhibits greater than 20-fold increased resistance to cleavage by human neutrophil elastase, similar progressive anti-thrombin activity, and twice the rate of fXa inhibition. Thrombin and fXa inhibition stoichiometries were respectively 2.6 and 2.3, and in contrast to plasma ATIII, cleaved inhibitor was observed in gels of complex formation reactions. Thus, the rates of encounter complex formation with thrombin and fXa, and partitioning into the substrate pathway are all increased for this mutant. Another NR-ATIII mutant has greater than 40-fold increased resistance to elastase, 30% of plasma ATIII progressive anti-thrombin activity and 3.5-fold increased anti-fXa activity. Thrombin and fXa inhibition stoichiometries were respectively 2.9 and 2.5, with cleaved inhibitor also observed in gels of complex formation reactions. Compared to plasma ATIII, this mutant has a similar encounter rate with thrombin, but increased partitioning into the substrate pathway. This is in contrast to its greatly increased encounter rate with fXa, and increased substrate partitioning as well. As predicted on the basis of their prolonged half-lives in the presence of purified HNE, the mutants exhibited similarly increased resistance to inactivation by IgG-stimulated human neutrophils. Therefore, these NR-ATIIIs may be useful for controlling in vivo activation of thrombin and thrombin coagulant and non-coagulant functions in inflammatory disorders where there are high neutrophil and neutrophil elastase burdens.

A1-ANTICHYMOTRYPSIN AND ALZHEIMER'S DISEASE: A-SHEET INSERTION IN FULL LENGTH AB PEPTIDES INDUCES SHEET-LIKE POLYMERISATION
B. Gooptu 1, L. Serpell 2, W.-S. W. Chang 1, D.A. Lomas 1. Departments of Medicine and Haematology, University of Cambridge, Division of Structural Medicine, Cambridge Institute for Medical Research 1, and Division of Neurobiology, Laboratory of Molecular Biology 2, MRC Centre, Hills Road, Cambridge, CB2 2QH

Alzheimer's disease is a term describing a heterogeneous condition characterised clinically by cognitive impairment and dementia, and histologically by a constellation of 'hallmark' lesions. Evidence has steadily accumulated to support a causative role for the formation of one such lesion - the deposition of amyloid in plaques primarily composed of the Alzheimer's peptide Ab1-42. In 1988 a1-antichymotrypsin, a member of the serpin superfamily and an acute phase protein, was co-localised to Alzheimer's amyloid plaques. Subsequently, studies of the interactions between Ab peptides and a1-antichymotrypsin have indicated both fibrillogenic and disaggregatory effects of the serpin on the peptide. We have studied these interactions in the general context of serpin-peptide interactions and binary complex formation. Recombinant wildtype a1-antichymotrypsin was expressed in E. coli, purified to homogeneity, and incubated with 100-fold molar excess of Ab1-42 at 37¯C over 48 hours. This resulted in a complete loss of native a1-antichymotrypsin to high molecular mass species, assessed by non-denaturing PAGE and Western blot analysis. The control incubation without peptide showed no appreciable loss of native protein. These high molecular mass species dissociated to intact, monomeric a1-antichymotrypsin and peptide following boiling in 1% w/v SDS. Electron microscopy of the co-incubated material showed very large sheet-like structures, previously undescribed, and very few structures with typical Ab fibrillar morphology when compared to control Ab peptide incubations. The formation of a1-antichymotrypsin: Ab1-42 multimers was conformation-specific and did not occur with latent or reactive loop cleaved a1-antichymotrypsin, or with a1-antichymotrypsin complexed with the P14-P3 reactive loop peptide of antithrombin. Native a1-antichymotrypsin was not taken up into multimers when Ab1-42 was present at less than a 25-fold molar excess relative to a1-antichymotrypsin (at 0.25mg/ml), nor on incubation of a1-antichymotrypsin with C- and N-terminal sections of the full length Ab peptides. The interaction was also serpin-specific as it did not occur with native a1-antitrypsin or antithrombin. However a1-antichymotrypsin when incubated with the Ab1-40 peptide interacted identically to a1-antichymotrypsin incubated under the same conditions with Ab1-42. These data are consistent with the findings of Janciauskiene et al (Nat. Struct. Biol. 1996, 3:668-671). They provide evidence of a specific interaction between Ab1-42 or Ab1-40 and a1-antichymotrypsin, requiring peptide insertion into the A-sheet of the serpin at the s4A position and a subsequent polymerisation process. This process is likely to involve non-inserted sections of Ab peptide molecules, and to occur in perpendicular directions, thus generating the observed sheet structures.

HIGH RESOLUTION (2.0 è) CRYSTAL STRUCTURE OF WILDTYPE a1-ANTITRYSIN
Peter R. Elliott, Xue Y. Pei and David A. Lomas Departments of Medicine and Haematology, University of Cambridge, Cambridge Institute for Medical Research, UK
a1-antitrypsin is the most abundant circulating proteinase inhibitor and historical the archetype of the serpin proteinase inhibitors. We present here the highest resolution structure to date of an intact, active, serpin. This new 2.0 è crystal structure of intact wildtype a1-antitrypsin confirms our previous structure of a1-antitrypsin (1, 2) and shows with much greater confidence the b-strand canonical conformation the reactive loop, which is ideal for docking with active site of the cognate proteinase. A stabilizing salt bridge between P5 glutamate of the reactive loop and arginies 193, 223 and 281 in the body of the molecule provides strong evidence that this conformation is not an artifact of crystallization but represents the circulating conformation of a1-antitrypsin in vivo. The b-strand conformation of the reactive loop explains the ability of the Z, Siiyama and Mmalton mutants of a1-antitrypsin to form inactive loop sheet of a second molecule to form a dimer that then extends to form chains of up to 20 molecules in length. These inactive polymers have been shown to form in the hepatic inclusions of a1-antitrypsin that are associated with liver cirrhosis and have also been demonstrated in plasma and in the lungs of affected individuals. Our high-resolution structure of a1-antitrypson provides much improved side chain clarity in all areas but most usefully around a hydrophobic pocket that is filled on reactive loop cleavage and on polymer formation. This pocket provides a potential target for rational drug design to prevent the formation of polymers and the associated liver cirrhosis, plasma deficiency and emphysema.
REFERENCES:
[1] Elliott, P.R., Abrahams, J-P and Lomas, D.A., (1998), Wild-type a1-antirypsin is in the canonical inhibitory conformation. J. Mol. Biol.. 275, pp419-425.
[2] Elliott, P.R., Lomas, D.A., Carrell, R.W. and Abrahams, J-P, (1996), Inhibitory conformation of the reactive loop of a1-antitrypsin. Nature Struct. Biol.. 3. pp676-681.

ALPHA1-ANTITRYPSIN VARIANTS IN RELATION TO ATOPY IN SOUTH AFRICAN PATIENTS WITH ULCER AND NON-ULCER DYSPEPSIA
M-Christine Gaillard*, Pierre Redelinghuys#, Obedy Mwantembe*, Ernest Song*. *Department of Medicine, University of the Witwatersrand, Johannesburg, South Africa # Department of Microbiology, University of the Witwatersrand, Johannesburg, South Africa

Alpha 1-antitrypsin (a1AT) is an anti-inflammatory mediator. Deficiency or dysfunction of this serpin is associated with several inflammatory diseases. We have previously shown significant associations of the M2 allele of a1AT in Caucasian patients with atopic asthma, and the V allele in Negroid patients with focal glomerulosclerosis and evidence of exposure to Mycobacterium tuberculosis. We investigated the prevalence of variance of a1AT in South African patients with dyspepsia with or without H. Pylori infection. Sixty-two Negroid patients (35 women), and 59 Caucasians (33 women) were investigated. The controls were race matched asymptomatic blood donors. The patients were compared to 52 Negroid controls (18 women), and 240 Caucasians (98 women). a1AT variants were identified by isoelectric focusing. Serum levels were quantified by laser nephelometry. H. Pylori was detected by the CLO-Urease test. Serum IgE levels were quantified by a flouro-immuno assay. Statistical analysis was done by means of the two-tailed Chi-squared test, and the Student's t-test. Both patient groups demonstrated an increased frequency of non-M1 variants: Negroids (13%), controls (3%),(p=0.032); Caucasians (21%), controls (12%), (p=0.04). The V allele was increased in Negroid dyspeptics (8%), controls (1%), (p=0.02). In Caucasian dyspeptics, the M2 phenotype (9%) and the F allele (2%) were increased compared to controls (1%) (p<0.00001), (0.2%)(p=0.01) respectively. Both groups showed raised serum a1AT levels: Negroid patients (1.62Ò0.34g/l), controls (1.34Ò0.13g/l) (p=0.0004); Caucasian patients (1.74Ò0.38g/l), controls (1.13Ò0.12g/l) (p=0.00001). The 2 groups demonstrated raised serum IgE levels: Negroid patients (92.4Ò87.8KU/L), controls (52Ò33KU/L), (p=0.031); Caucasian patients (150.2Ò168KU/L), controls (22.8Ò16.6KU/L), (p=0.02). H. pylori infection prevalence was 64% in the Negroid patients, and 52% in the Caucasian patients. There was no difference with the respective controls. The association of specific a1AT alleles with dyspepsia, raised a1AT and IgE levels, may suggest an atopic component in this disease which could result in an inflammatory response to H. pylori infection or food allergens.

PROTEIN C INHIBITOR REGULATES THE THROMBIN-THROMBOMODULIN COMPLEX IN THE ACTIVATION OF TAFI AND PROTEIN
C. Laurent O. Mosnier, Marc G.L.M. Elisen, Joost C.M. Meijers and Bonno N. Bouma. Dept. of Haematology, University Medical Center, Utrecht, the Netherlands.

Protein C inhibitor (PCI), originally identified as an inhibitor of activated protein C (APC), inhibits a broad variety of coagulation factors such as factor Xa and thrombin. Recently, PCI was identified as a potent inhibitor of thrombin bound to the endothelial cell receptor thrombomodulin. This is an interesting observation since thrombin bound to thrombomodulin has distinct substrate specificities compared to thrombin in the absence of thrombomodulin. Furthermore, a specific inhibitor of this complex has not been identified before. The two substrates in plasma of which thrombomodulin stimulates the activation by thrombin are protein C and Thrombin Activatable Fibrinolysis Inhibitor (TAFI). TAFI is a carboxypeptidase B-like enzyme that after activation down regulates fibrinolysis by removing C-terminal lysines from partially degraded fibrin that play a role in plasminogen binding and activation. Since TAFI is activated by thrombin, TAFI activation can be controlled by enzymes that regulate the formation of thrombin, such as activated protein C (APC). APC inactivates the factors Va and VIIIa, thereby down regulating the formation of thrombin, resulting in a reduced activation of TAFI. Activation of protein C, therefore, results in an up regulation of fibrinolysis while activation of TAFI results in a down regulation of fibrinolysis. Since activation of both TAFI and protein C is stimulated by thrombomodulin and are therefore potentially susceptible for inhibition by PCI, we determined how PCI regulates the balance between TAFI activation and inhibition of TAFI activation by APC. In a purified system, activation of both TAFI and protein C could be inhibited by PCI. PCI had no direct effect on activated TAFI, whereas the direct inhibition of APC activity by PCI accounted for less than 2% of the total inhibition. In a plasma system, TAFI activation was determined using tissue factor induced coagulation and 1 nM thrombomodulin. TAFI activation was increased in PCI depleted plasma (PCIdP) compared to normal plasma. Reconstitution of PCIdP with purified PCI restored the inhibition of TAFI activation corresponding to that in normal plasma. This indicates that PCI is capable of inhibiting the activation of TAFI both in a purified system and in plasma. Similar to TAFI, activation of protein C is stimulated by thrombomodulin and is therefore also susceptible to inhibition by PCI. Because APC in turn inhibits the activation of TAFI by down regulation of the thrombin formation, inhibition of APC generation by PCI should lead to increased TAFI activation. For this, TAFI activation was determined in protein C and PCI double depleted plasma (PC/PCIdP) in the presence of a relative high thrombomodulin concentration (10 nM TM), which results in sufficient stimulation of protein C activation to completely inhibit TAFI activation in normal plasma. In contrast, stimulation of TAFI activation by thrombomodulin reaches its maximum at a thrombomodulin concentration of about 3 nM. In the PC/PCIdP large amounts of TAFI were activated. Addition of PCI to PC/PCIdP had no effect since stimulation of TAFI activation was saturated at this thrombomodulin concentration. In contrast, reconstitution of PC/PCIdP with protein C almost abolished the activation of TAFI caused by the inhibitory effect of APC on the activation of TAFI. However, when PC/PCIdP was reconstituted with both PCI and protein C TAFI activation increased again compared to PC/PCIdP reconstituted with only protein C. This time, PCI inhibited the generation of APC resulting in a decreased inhibition of the activation of TAFI by APC while abundant thrombin-thrombomodulin complexes remained to stimulate the activation of TAFI. This suggests that PCI can up regulate TAFI activation by the inhibition of APC activation. PCI is therefore an important regulator in the balance between coagulation and fibrinolysis by inhibiting both the activation of TAFI and of protein C. However whether PCI up regulates or down regulates fibrinolysis is dependent upon environmental factors, of which the thrombomodulin concentration is an important determinant. Analysis of the Anti-atherogenic and Regulatory Activity by Serp-1, a Viral Anti-Inflammatory Serpin Alexandra Lucas, Piers Nash, Hai yan Guan, Stephen Little, Erbin Dai, Li Ying Liu, Denny Ivanova, Grant McFadden. Robarts Research Institute, London, Ontario, Canada. Rationale: Serp-1 is a secreted viral glycoprotein, a serpin that binds and inhibits tissue- (tPA) and urokinase-type plasminogen activators (uPA) as well as plasmin. Plasminogen activator inhibitor-1 (PAI-1) is a vascular serpin that inhibits tPA and uPA in a similar manner to Serp-1 and plays a pivotal regulatory role in atherosclerotic plaque growth after angioplasty injury. We have recently reported that Serp-1 markedly reduces plaque growth after angioplasty and aortic allograft transplant in rat and rabbit models. Methods: Serp-1, an active site mutant, rat PAI-1, and Serp-1/serpin P1-P1' active site chimeras were infused immediately after iliofemoral artery angioplasty injury under general anesthetic. The Serp-1 active site was replaced by the P1-P1' active site from PAI-1, PAI-2, al-chymotrypsin, and nexin. Ilio-femoral plaque growth was assessed at 4 weeks follow up by morphometric analysis of Hematoxylin and eosin histological sections. Plasminogen activator and PAI- 1 enzyme activity were also measured in the rat iliofemoral arterial sections at 0 to 24 hours after angioplasty and treatment with Serp-1 or serpin chimeras by quantitative RT-PCR. Results: Atheroselerotic plaque area was significantly reduced at 4 weeks follow up by Serp-1 and the PAI-2 and al- chymotrypsin chimeras (p<0.01), but not by the ala- ala active site mutant. Decreased tPA and increased PAI-1 and uPAR mRNA and protein levels were detected at early times after balloon injury and treatment with Serp-1. PAI-1 and the serpin chimeras did not alter tPA, PAI-1 or uPAR mRNA levels. Conclusion: Inhibition of atherogenesis by Serp-1 is associated with upregulation of PAI-1. Similar reduction of plaque development by Serp-1/serpin chimeras is not associated with increased PAI-1 mRNA. The serpin backbone as well as the active site is important to the anti-atherosclerotic activity of Serp-1. A 2.85è

STRUCTURE OF AN ACTIVATED ANTITHROMBIN REVEALS THE LINK BETWEEN THE HINGE REGION AND THE HEPARIN BINDING DOMAIN
Airlie McCoyY, Xue Pei, Robin W. Carrell, Peter G. W. Gettinsx and James A. Huntington. Department of Haematology, Division of Structural Medicine, CIMR, Cambridge, UK CB2 2XY, YMRC-LMB, Structural Studies Division, Cambridge, UK CB2 2QH, Department of Biochemistry and Molecular Biology, UIC, Chicago, IL USA.

Antithrombin is unique among the serpins in that it circulates in a conformation which is kinetically inactive towards its target proteinase, factor Xa. Conformational activation occurs upon binding of a pentasaccharide sequence unique to heparin which results in a rearrangement in secondary structure in the heparin binding region and expulsion of the hinge-region from b-sheet A. The recently solved structure of antithrombin bound to a high-affinity synthetic pentasaccharide revealed an elongation of helix D towards its C-terminus, and of helix A towards its N-terminus (1). This finding was in support of the hypothesis that helix D elongation played a role in affecting the expulsion of the hinge-region and thus was responsible for conformational activation (2). Although the structures of native and pentasaccharide bound antithrombin reveal the conformation of the two end states, the molecular mechanism between the end states remains open to speculation. In particular, two questions remain: 1) How is the native conformation stabilised in the partially loop-inserted form? 2) How are structural changes in the heparin binding region translated to the hinge region? Our approach was to determine the crystal structure of a conformationally activated antithrombin, which has a fluorescein labeled cysteine at position 380 (P14), in order to see how full activation is achieved in the absence of oligosaccharide and if, and how, the structural effects of the substitution in the hinge-region are translated to the heparin binding region. We solved the structure of S380C-fluorescein antithrombin to 2.85è resolution using the cassette crystallisation procedure where latent and active antithrombin is added at a one-to-one ratio. Data were collected at Daresbury station 7.2 on a single frozen crystal. The fluorescein moiety could clearly be seen and made contacts with the surface of the antithrombin molecule. Although in solution the antithrombin variant behaved as a loop expelled, activated antithrombin (3), in the crystal the hinge region was inserted to the extent seen in native antithrombin with very little perturbation of the peptide backbone. The heparin binding region, however, had structural changes consistent with the structure of the pentasaccharide bound structure. In particular, there was a 1.5 turn extension of helix D and a 1 turn extension of helix A. The apparent inconsistency between the native-like structure of the hinge region and the pentasaccharide-activated structure of the heparin binding region provided a unique opportunity to compare the detail of the surface interactions responsible for the linkage between the two regions. We hypothesise a novel mechanism for conformational activation of antithrombin whereby the native conformation is stabilised by ionic interactions between the P13 Glu and residues on the loop extending from helix F, and that rearrangement of the C-terminal loop of helix D upon pentasaccharide binding results in a significant reorientation of the P13 sidechain resulting in loop expulsion.
Reference List
1. Jin, L., Abrahams, J.P., Skinner, R., Petitou, M., Pike, R.N., and Carrell, R.W. (1997) Proc.Natl.Acad.Sci.U.S.A. 94, 14683-14688
2. van Boeckel, C.A., Grootenhuis, P.D., and Visser, A. (1994) Nat.Struct.Biol. 1, 423-425
3. Huntington, J.A. and Gettins, P.G. (1998) Biochemistry 37, 3272-3277