Identification of Myxomaviral Serpin Reactive Site Loop Sequences That Regulate Innate Immune Responses*

The thrombolytic serine protease cascade is intricately involved in activation of innate immune responses. The urokinase-type plasminogen activator and receptor form complexes that aid inflammatory cell invasion at sites of arterial injury. Plasminogen activator inhibitor-1 is a mammalian serpin that binds and regulates the urokinase receptor complex. Serp-1, a myxomaviral serpin, also targets the urokinase receptor, displaying profound anti-inflammatory and anti-atherogenic activity in a wide range of animal models. Serp-1 reactive center site mutations, mimicking known mammalian and viral serpins, were constructed in order to define sequences responsible for regulation of inflammation. Thrombosis, inflammation, and plaque growth were assessed after treatment with Serp-1, Serp-1 chimeras, plasminogen activator inhibitor-1, or unrelated viral serpins in plasminogen activator inhibitor or urokinase receptor-deficient mouse aortic transplants. Altering the P1-P1′ Arg-Asn sequence compromised Serp-1 protease-inhibitory activity and anti-inflammatory activity in animal models; P1-P1′ Ala-Ala mutants were inactive, P1 Met increased remodeling, and P1′ Thr increased thrombosis. Substitution of Serp-1 P2–P7 with Ala6 allowed for inhibition of urokinase but lost plasmin inhibition, unexpectedly inducing a diametrically opposed, proinflammatory response with mononuclear cell activation, thrombosis, and aneurysm formation (p < 0.03). Other serpins did not reproduce Serp-1 activity; plasminogen activator inhibitor-1 increased thrombosis (p < 0.0001), and unrelated viral serpin, CrmA, increased inflammation. Deficiency of urokinase receptor in mouse transplants blocked Serp-1 and chimera activity, in some cases increasing inflammation. In summary, 1) Serp-1 anti-inflammatory activity is highly dependent upon the reactive center loop sequence, and 2) plasmin inhibition is central to anti-inflammatory activity.

Serine proteinase inhibitors, designated serpins, represent the majority of proteinase inhibitors in the circulating blood, functioning to balance thrombotic and thrombolytic cascades (1)(2)(3). The reactive center loop (RCL) 2 of the prototypical serpin molecule consists of 20 -27 amino acids near the C terminus, approximately P17-P10Ј, dependent upon the individual serpin, and defines serine proteinase target(s). The specificity of serpin recognition of protease targets is thought to principally reside in the P1-P1Ј scissile bond, which acts as a pseudosubstrate for target proteases, forming a suicidal inhibitory complex (1)(2)(3). The RCL exists in a strained configuration that is required for full inhibitory function, during which the target protease cleaves the P1-P1Ј and is then bound and dragged across the serpin face (10 -14), rendering the enzyme inoperative. This RCL loop structure thus also has a significant role in serpin to target protease specificity (4,5).
Activation of serine protease cascades, both in the thrombotic and thrombolytic cascades, is associated with activation of the innate immune response. Excessive or prolonged up-regulation of innate immune responses, also termed inflammation, accelerates plaque growth and even rupture, resulting in increased local thrombosis and arterial occlusion (heart attacks and strokes) or, conversely, outward remodeling and aneurysm formation. Innate immunity is an integral part of vascular healing (6 -10). The urokinase-type plasminogen activator (uPA) and the uPA receptor (uPAR) form complexes that initiate thrombolysis, reducing clot formation, and act to up-regulate inflammatory cell responses (6 -8). Increased uPA and uPAR are detectable after acute arterial injury in a wide range of cells (endothelial cells, monocytes, smooth muscle cells, and T lymphocytes) (11)(12)(13), where the uPA⅐uPAR complex activates matrix metalloproteinases and growth factors, initiating collagen and elastin breakdown and allowing cell invasion and proliferation. The uPA⅐uPAR complex also alters cellular activation and intracellular signaling (11)(12)(13). In mouse models, uPA deficiency reduces intimal hyperplasia, and in rabbits excess uPA increased plaque growth after vascular injury, whereas uPAR expression increased macrophage invasion in ApoE null mice (14), and uPAR deficiency did not alter plaque growth (15).
Whereas serpins can represent 2-10% of serum proteins, the exact mechanism of serpin-mediated anti-inflammatory activity in vivo in a complex living circulatory system is only partially defined. To determine the reactive center site sequences responsible for Serp-1-mediated anti-inflammatory activity, we have assessed the effects of a series of Serp-1 RCL chimeras, designed to mimic known mammalian and viral serpins, in mouse aortic transplant models. To ascertain the specificity of Serp-1 inhibitory activities, the effects of the mammalian serpin, PAI-1, and two unrelated cross-class viral serpins, CrmA and Serp-2 (31,32), were also assessed in these models. Serp-1-mediated regulation of innate immune responses was then correlated with specific RCL sequences and protease inhibition.
A single infusion of purified Serp-1, individual Serp-1 RCL chimeras, or control serpin proteins (0 -15 g; 0 -500 ng/g of body weight in 0.2 ml) were given by intravenous tail vein injection immediately after transplant, when aortic pulsation was visible (Table 1). Twelve PAI-1 Ϫ/Ϫ mice had 50 ng of MCP-1 (Cedar Lane, Hornby, Canada) given by intraperitoneal injection for analysis of cellular migration 18 h after treatment with saline or 1.5 g of PAI-1 or Serp-1 (33). Twelve Balb/c mice (3 mice/treatment group) were tested for bleeding time with no treatment or after intravenous injection of saline and 1. Histology-A 0.5-0.6-cm length of transplanted donor aorta and adjacent recipient aorta was cut into two 0.25-0.3-cm pieces, fixed, and paraffin-embedded, and two 5-m sections were cut and stained with hematoxylin and eosin for morphometric analysis, providing four sections per aorta for analysis (24,(27)(28)(29)(30). Areas of plaque, thrombus, lumen, and internal elastic lamina and the number of invading mononuclear cells per high power field were measured using the Empix Northern Eclipse program (Empix Imaging Inc., Mississauga, Canada) and a Sony Power HAD3CCD video camera attached to the microscope and calibrated to the microscope objective, as described (27)(28)(29)(30). The mean total cross-sectional area of the intima, using sections with the largest detectable plaque, was calculated for each aortic specimen. A score for aneurismal dilation (0 -4ϩ) was derived based upon the IEL measurement.
Fluorescence-activated Cell Sorting Analysis-Peritoneal exudates from mice injected with MCP-1 were washed with phosphate-buffered saline containing 2% fetal bovine serum and treated with RBC lysis buffer (0.15 M ammonium chloride, 10 mM potassium bicarbonate, 0.1 mM EDTA, pH 7.4). RBC-free cells were isolated by layering with 0.5 ml of fetal bovine serum and centrifuging at 500 ϫ g for 5 min. Cells were identified using fluorescein isothiocyanate-labeled CD4 and phycoerythrin-labeled CD11b (FACSCalibur; BD Biosciences) (38).
Expression and Purification of Serp-1 and Serp-1 Chimeras-Serp-1 CHO and Serp-1 VV were harvested and purified from the supernatant of a recombinant Chinese hamster ovary (CHO) cell line (Biogen, Inc. (Boston, MA) and Viron Therapeutics, Inc. (London, Canada)) and vaccinia vector (VV)-infected Buffalo green monkey kidney cell supernatants (24,(27)(28)(29)(30), respectively, as previously described. Serp-2 and CrmA (cytokine response modifier A) as well as two Serp-2 RCL mutants D294A and D294E were purified as previously described (29,31,32). For Serp-1 RCL mutant construction, the complete native Serp-1 gene, including the signal sequence, was inserted into the thymidine kinase locus of VV strain WR under the control of a synthetic late promoter by homologous recombination (25,26,29,39), followed by multistage mutagenesis of the P7 to the P2Ј sites using the Altered Sites II mutagenesis system (Promega) cloning approach ( Table 2). In brief, a silent AlwNI site was inserted at the P8-P7 codon boundary (nucleotide position 934 of Serp-1 open reading frame) without altering the Serp-1 protein sequence to allow for the use of AlwNI and BglI for P7-P2Ј region excision. The complete Serp-1 sequence, including the signal sequence, was inserted into the thymidine kinase locus of VV strain WR as described (26,39). A BamHI/HindIII fragment containing the complete Serp-1 open reading frame was cloned into the pAlter-1 vector (Promega). Purified pAlter-Serp-1 was denatured and annealed to three mutagenic oligonucleotides simultaneously: the ampicillin resistance repair oligonucleotide (Promega) and two oligonucleotides were used
Analysis of Serp-1 and Chimera Protease Inhibition-Serp-1 or individual Serp-1 RCL chimeras (180 nM) were incubated with a slight molar excess of target proteinases, uPA, tPA, plasmin, thrombin, trypsin, cathepsin G, bovine chymotrypsin, or human neutrophil elastase, obtained from Sigma, in 100 mM NaCl, 2 mM CaCl 2 , 0.005% Triton X-100, 100 mM Tris-HCl, pH 7.5. Samples were separated on a 4 -20% linear gradient or 10% Tris-glycine SDS-polyacrylamide gels using the Laemmli buffer system (23,27,28,39) and transferred to Hybond C-extra (Amersham Biosciences) nitrocellulose by electroblotting, blocked in TBS (150 mM NaCl, 2.5 mM KCl, 25 mM Tris-HCl, pH 7.4) containing 5% (w/v) skimmed milk powder and 0.1% (v/v) Tween 20. Bands containing Serp-1 were identified by 0.05% (v/v) monoclonal anti-Serp-1 antibodies AQ.H9 and AG.F11 (kindly provided by Leona Ling, Biogen, Inc.) and secondary goat anti-rabbit antibody conjugated to horseradish peroxidase (Bio-Rad). Labeled bands were visualized using ECL detection (Amersham Biosciences) on Eastman Kodak Co. x-ray detection type film. Reactions with complete protease cleavage of Serp-1 were repeated at shorter incubation times to test for transient, high molecular weight complexes. Reactions in which the serpin appeared unchanged at the end of the incubation period were repeated at higher enzyme concentrations for longer incubation periods. Serp-1 and the Serp-1 (SAA) and (Ala 6 ) constructs were also assessed for thermal stability. All three serpins precipitated with loss of activity at temperatures greater than 60°C, indicating that altered activity was not indicative of a latent state in which proteins are stable at higher temperature. Standard assays for inhibition were also performed as previously described (25,39,41) to assess inhibitory activity of Serp-1, Serp-1 (SAA), and Serp-1 (Ala 6 ) for uPA and plasmin (chromogenic pefachrome substrate) (Sigma). 0.19 g/ml plasmin was added to reactions containing 0.75 mM S-2251 substrate and varying concentrations of Serp-1 (at 0 -6 g/ml) in the reaction buffer (100 mM Tris-HCl, 100 mM NaCl, 2 mM CaCl 2 , 0.005% Triton X-100, pH 7.5). The pseudo-first-order rate constant (k obs ) at each inhibitor concentration was determined from a nonlinear regression fit, and data were collected every 5 min (300 s).
Statistics-Mean cell count from three high power fields and mean plaque area from individual animals were assessed for significant change by unpaired Student's t test and analysis of variance with Fisher's protected least significance difference post-test analysis (41). Regression analysis was also used to assess correlations for histologic parameters and for serpin protease inhibition kinetics.
Despite the preservation of an intact Serp-1 RN sequence at the P1-P1Ј site, the Serp-1 (Ala 6 ) chimera did not retain anti-inflammatory or anti-atherogenic activity, producing instead an accelerated inflammatory cell response (p Ͻ 0.03) (Fig. 1, C and E). Mononuclear cell invasion into the intimal and adventitial layers (p Ͻ 0.0001) (Fig. 1, C and E) and mean plaque area (p Ͻ 0.002) (Fig. 1, C and F) increased 2-3-fold after treatment with the Serp-1 (Ala 6 ) chimera when compared with all other treatments (Fig. 1, E and F). Areas of chronic, organized thrombosis (Fig. 1, C and G) were also profoundly increased after Serp-1 (Ala 6 ) treatment (p Ͻ 0.028). Associated with increased plaque area, inflammatory cell invasion, and thrombosis in the Serp-1 (Ala 6 )treated mice was marked positive (outward) remodeling with focal aneurysm formation (p Ͻ 0.01) (Fig. 1, C-H).
In summary, 1) Serp-1, but not the Serp-1 RCL mutants, is capable of reducing inflammatory responses and plaque growth, and 2) altering the Serp-1 P1-P1Ј scissile bond results in loss of anti-inflammatory activity, and mutation of P2-P7 exacerbates inflammation, thrombosis, and plaque growth with attendant aneurismal dilation.
Infusion of the Serp-1 (CrmA) chimera with CrmA P1-P1Ј amino acid sequence DC neither decreased nor increased mortality (Table 1) (p ϭ 0.698), inflammatory cell response, or plaque growth at 4 weeks follow-up (Fig. 1, E and F), as noted above. We have concluded that the infusion of unrelated, foreign viral serpins does not reproduce the antiinflammatory and anti-atherogenic effects of Serp-1 infusion but can for some (CrmA and D294A) initiate a proinflammatory response. The substitution of the Serp-1 P1-P1Ј sequence with CrmA P1-P1Ј DC sequence, however, did not initiate a proinflammatory response, indicating that the combination of a negatively charged Asp residue at P1 with a polar uncharged Cys at P1Ј in the Serp-1 RCL did not increase inflammation, such as was seen with Serp-1 (␣1-PI) and (Ala 6 ) chimeras.
Conversely, the Serp-1 (Ala 6 ) and the Serp-1 (PAI-2) mutants increased thrombosis (Fig. 2G, p Ͻ 0.037) and outward arterial remodeling (Fig. 2H, p Ͻ 0.043) in uPAR Ϫ/Ϫ allografts. The prothrombotic effect of the Serp-1 (PAI-2) chimera and the increased remodeling produced by Serp-1 (Ala 6 ) were greater in the uPAR Ϫ/Ϫ aortic allograft model (p Ͻ 0.037 and p Ͻ 0.004, respectively, upon comparison with saline) (Fig. 2, E-H)) and when compared with Serp-1 (PAI-2) and Serp-1 (Ala 6 ) mutant treatment after PAI-1 Ϫ/Ϫ aortic allograft transplant (Fig. 1). A significant increase in plaque area was also detected in the uPAR Ϫ/Ϫ model after Serp-1 (PAI-2) and Serp-1 (Ala 6 ) treatment (Fig. 2F). Despite the increased thrombosis, remodeling and plaque growth, none of these chimeras increased mortality during the term of this study (Table 1) (p ϭ 0.355). These findings suggest that uPAR expression in the donor aorta is not required for the increased thrombosis and aneurismal remodeling observed after treatment with some of the Serp-1 mutants. However, uPAR is present in the recipient aorta and in circulating cells in the recipient mouse and thus may still have a role in thrombosis and aneurysm formation.
Effects of Viral and Mammalian Serpins on Thrombosis and Mortality-Both PAI-1 and Serp-1 target tPA and uPA but diverge in their ability to bind and inhibit plasmin, thrombin, and factor Xa. PAI-1, similar to Serp-1, has also been demonstrated to have anti-inflammatory and anti-atherogenic activity in selected animal models, whereas Serp-1 can compensate for PAI-1 deficiency in mouse PAI-1 Ϫ/Ϫ aortic allografts (29). In order to assess the relative efficacy of Serp-1 and PAI-1 for inhibiting inflammatory cell responses and plaque growth, we compared PAI-1 and Serp-1 treatment of PAI-1 Ϫ/Ϫ aortic allograft transplants. The mortality for Serp-1-treated mice after aortic transplant ranged from 0 up to 14.29%, less than or equal to that seen in salinetreated controls, where mortality was 0 -31.25% (Table 1) (p ϭ 0.079 for Serp-1). In stark contrast, rat PAI-1 treatment resulted in acute thrombosis in the transplanted aorta with early (less than 24-h) mortality in all mice treated (Table 1) (p Ͻ 0.0001). Mouse PAI-1 treatment was also tested to rule out a possible reaction to rat PAI-1 in the mouse models, but infusion of mouse PAI-1 also produced acute thrombosis with 100% mortality (p Ͻ 0.0001) ( Table 1). PAI-1 infusion also caused a marked early thrombotic response with 100% mortality in uPAR Ϫ/Ϫ aortic allograft (Table 1) and uPAR Ϫ/Ϫ isograft transplants (Table 1). Serp-1 treatment did not alter mortality in the uPAR Ϫ/Ϫ isograft or the uPAR Ϫ/Ϫ (0 -28.57%) allograft transplant models (Table 1).
PAI-1 infusion was also tested in PAI-1-deficient isografts where PAI-1 expression should be absent or minimal. In PAI-1 Ϫ/Ϫ to PAI-1 Ϫ/Ϫ isograft transplants, infusion of PAI-1 did not increase mortality ( Table 1) (mortality 33.3%). We have concluded that the presence of the Arg-Met P1-P1Ј RCL sequence in PAI-1 can produce anti-inflammatory and anti-atherogenic activity in some animal models, but in the presence of circulating PAI-1 in recipient mice, infusion of low dose PAI-1 increased thrombosis.
Analysis of Clotting Time and Platelet Activation after Serp-1 and PAI-1 Treatment-Histology analysis represents findings at 4 weeks follow-up in the aortic transplant segments but does not allow analysis of early changes in the arterial wall immediately after transplant. Thus, early effects of Serp-1, PAI-1, and the Serp-1 RCL chimeras on bleeding time, platelet activation, and mononuclear cell invasion were also assessed.
The bleeding time for wild type Balb/c mice was significantly shorter for PAI-1 than for Serp-1 treatment (p Ͻ 0.014) (Fig. 3A). Although both serpins reduced the bleeding time, only PAI-1 produced a significant and reproducible reduction in the bleeding time (p Ͻ 0.016). Adherence of platelets, isolated from normal human volunteers, to fibronectin was also assessed in the presence and absence of treatment with Serp-1 and PAI-1, in vitro. Serp-1 reduced human platelet adhesion to fibronectin (p Ͻ 0.05), whereas PAI-1 increased platelet adhesion (p Ͻ 0.05) (Fig. 3B).
Effects of Serp-1, PAI-1, and Serp-1 RCL Mutant on Peritoneal Monocyte Invasion-Serp-1 reduced migration of mononuclear cells into mouse ascites in response to intraperitoneal injection of MCP-1 in PAI-1 Ϫ/Ϫ mice. PAI-1 infusion, in contrast, had no effect on monocyte migration into ascites fluid after injection of MCP-1, indicating that Serp-1 has a greater inhibitory activity for monocyte invasion, even in mice lacking PAI-1 (p Ͻ 0.01) (Fig. 3C). Serp-1 (SAA) had no effect. Serp-1 also reduced the total number of migrating mononuclear cells in normal Balb/c mice after stimulation with MCP-1 (p Ͻ 0.02) (Fig. 3D), with associated equal reductions in CD4-positive T cells (p Ͻ 0.011) (Fig. 3E) and CD11b-positive monocytes (data not shown) as assessed by fluorescence-activated cell sorting analysis. The Serp-1 (Ala 6 ) RCL chimera conversely increased total cell invasion (p Ͻ 0.02) (Fig. 3D) with significant and equivalent percentage increases in CD4 ϩ T cells (p Ͻ 0.011) (Fig. 3E) and CD11b-positive mononuclear cells (not shown) into mouse peritoneal fluid when compared with Serp-1. We have concluded that Serp-1 inhibits early mononuclear cell invasion in response to chemokine activation, whereas PAI-1 does not. In contrast, the Serp-1 (Ala 6 ) chimera produced a marked increase in mononuclear cell invasion.
Kinetics of Protease Inhibition-Specific P1-P1Ј sequences and protease-inhibitory activity for Serp-1 and individual chimeras were also compared with the capacity for each serpin and chimera to alter cell invasion, thrombosis, plaque growth, and aneurysm formation after aortic transplant. Western blot analysis of serpin binding to selected proteases and protease cleavage as well as analysis of kinetics of uPA and plasmin inhibition are provided ( Table 2). Serp-1, as previously reported (26,39), bound to and inhibited tPA, uPA, plasmin, and factor Xa but was cleaved by thrombin. The Serp-1 (PAI-2) RCL chimera failed to interact with uPA, tPA, or plasmin but formed inhibitory complexes with thrombin and was cleaved by trypsin, human neutrophil elastase, and cathepsin G, exhibiting a product of ϳ50 kDa, consistent with cleavage at or near the RCL. Serp-1 (PAI-2) is also cleaved by chymotrypsin, without the formation of an inhibited complex. The Serp-1 (␣1-PI) chimera binds tPA and Thr, forming SDS-stable complexes. However, no inhibited complex was observed with either uPA or plasmin, both of which cleaved Serp-1 (␣1-PI) without forming a visible inhibitory complex. Serp-1 (␣1-PI) was cleaved by trypsin, chymotrypsin, human neutrophil elastase, and cathepsin G. Serp-1 (CrmA) had properties similar to those observed for Serp-1 (␣1-PI), forming a visible inhibitory complex with tPA, but similar to the Serp-1 (SAA) mutant, it failed to bind to uPA, plasmin, or thrombin and was cleaved by trypsin, human neutrophil elastase, cathepsin G, and chymotrypsin. Serp-1 (Ala 6 ) failed to form SDS-stable complexes with any of the proteinases tested in gel shift assays, but a 50-kDa cleaved form was observed upon incubation with uPA, tPA, plasmin, thrombin, trypsin, and chymotrypsin. Incubation of Serp-1 (Ala 6 ) chimera with human neutrophil elastase and cathepsin G failed to exhibit either substrate or inhibition activity.
Kinetic analysis of inhibition of uPA and plasmin was also performed for Serp-1 and the Serp-1 (SAA) and Serp-1 (Ala 6 ) mutants to further define specific changes in protease inhibition. These studies detected specific interactions for Serp-1 (Ala 6 ) with uPA and plasmin, interactions not detectable by simple gel shift assays (Fig. 4). Altering the P1-P1Ј Arg-Asn sequence resulted in loss of uPA inhibition, as seen with the Serp-1 (SAA) mutant. The rate of reaction (K a ) for wild type Serp-1 with uPA was 2.59 ϫ 10 5 M Ϫ1 s Ϫ1 ; K a for plasmin was 4.8 ϫ 10 4 M Ϫ1 s Ϫ1 ; K a for Serp-1 (SAA) was inactive or equal to 0 for uPA and plasmin; and K a for Serp-1 (Ala 6 ) was 2.05 ϫ 10 4 M Ϫ1 s Ϫ1 for uPA and with relative  6 ) not only has lost the capacity to inhibit plasmin but displays modestly increased plasmin activity. Serp-1 (Ala 6 ) concentrations were 0 (a), 6 g/ml (1), 3 g/ml (2), and 1.5 g/ml (3). These curves have higher slopes than the control curve (a), indicating an increase of the reaction velocity.  (p < 0.016). Serp-1 reduced the bleeding time but not significantly (A). Adhesion of platelets from normal volunteers to fibronectin was significantly reduced by Serp-1 treatment (p Ͻ 0.05) and increased with PAI-1 treatment (p Ͻ 0.05) (B). Serp-1 infusion, but not PAI-1 infusion, reduced mononuclear cell invasion into mouse ascites in PAI-1 Ϫ/Ϫ mice after intraperitoneal injection of MCP-1 (C). Total numbers of invading mononuclear cells were also reduced after intravenous injection of Serp-1 into normal mice with MCP-1 intraperitoneal injection (D) but markedly increased after injection of Serp-1 (Ala 6 ). Fluorescence-activated cell sorting analysis similarly demonstrated a decrease in CD4-positive cells migrating into mouse ascites with MCP-1 injection after treatment with Serp-1 but an increase after treatment with Serp-1 (Ala 6 ) (E). ANOVA, analysis of variance. enhancement of activity for plasmin (Fig. 4). Retention of the Serp-1 P1-P1Ј RN sequence in Serp-1 (Ala 6 ) reduced inhibitory activity for uPA (Table 2). However, mutation of the P2-P7 arm to a series of 6 Ala residues not only removed the capability to inhibit plasmin but in fact produced a modest increase, up to 1.3-fold, in plasmin activity, as detected by the increased slope of curves in the plasmin assays (Fig. 4B). Upon comparison of curves with control, there is an increase of reaction velocity along with increased Serp-1 (Ala 6 ) concentrations (Fig. 4B).
The stoichiometry of inhibition (SI) for Serp-1 was measured for uPA and plasmin at 1.0 and 1.9, respectively (40). Both SI values are in the range deemed consistent with the inhibitory pathway rather than the substrate pathway; SI values between 1 and 2 are indicative of an inhibitory pathway. Using the SI value multiplied by the K a provides an adjusted K a , as previously reported by Schechter and Plotnick (40). The calculated adjusted K a values for Serp-1 do not vary significantly from prior K a analyses (2.59 ϫ 10 5 for uPA and 9.12 ϫ 10 4 for plasmin). If one makes the assumption that the SI values for the wild type Serp-1 can be used to calculate the adjusted K a for Serp-1 (Ala 6 ) with uPA, the adjusted K a remains 2.05 ϫ 10 4 . No SDS stable complexes were, however, detectable for Serp-1 (SAA) and Serp-1 (Ala 6 ) mutants when incubated together with uPA or plasmin on gel shift analysis, and thus calculation of SI for these mutants is inaccurate. Further detailed analyses will be needed to provide a final accurate assessment of the kinetics for each serpin analyzed in these studies.

DISCUSSION
The arterial wall has a highly modifiable architecture with responses to injury that range from cellular proliferation with negative or inward remodeling to cellular necrosis with outward or positive remodeling and all possible variations between. These highly plastic responses in the arterial wall are initiated by a host of cells that include activated platelets, macrophages, inflammatory T cells, and smooth muscle cells (8 -10, 42, 43). The balance between synthesis and degradation of connective tissue and proliferation and necrosis of cells is regulated in part by serine proteases in the thrombotic and thrombolytic cascades and by their attendant regulatory serpins (2, 6, 7, 9, 11-13, 15, 44). This balance can determine the final outcome of vascular responses to injury, specifically vascular healing, lumen obliteration, or aneurysm formation.
We report here a viral serpin, Serp-1, that binds to and inhibits plasminogen activators yet demonstrates a remarkable selectivity for inhibition of inflammatory responses without associated thrombosis. In this series of studies, Serp-1 significantly reduced early platelet adhesion and monocyte migration following aortic allograft transplant. Selective changes in the reactive center loop of Serp-1 markedly altered the balance between inflammation, thrombosis, and healing in the vascular wall. Our studies have demonstrated that 1) the Serp-1 P1-P1Ј sequence with Arg (a bulky, positively charged amino acid) and Asn (a polar uncharged amino acid), both containing NH 2 groups, provides a unique anti-inflammatory and anti-atherogenic activity, 2) substitution of either the P1 or the P1Ј amino acid results in a loss of anti-inflammatory activity, and 3) altering the P2-P7 RCL arm amino acid sequence increases plasmin activity and induces a widely divergent proinflammatory response. We have also demonstrated that, despite similarities in protease targets, the mammalian serpin, PAI-1, induces a markedly prothrombotic response, whereas Serp-1 has a neutral effect on thrombosis and can reduce platelet activation in vitro.
The mammalian serpin, PAI-1, which is similar to Serp-1 in terms of substrate specificity, binds plasminogen activators but displays a very different profile of activity in this allograft model, producing local thrombosis, vascular occlusion, and inevitable early fatality. PAI-1 is known to regulate and block thrombolysis with the potential for increased thrombosis (16,45,46), but PAI-1 has also been reported to have both anti-inflammatory and proinflammatory actions in differing animal models (15, 16, 19 -21, 29). Despite the shared capacity of Serp-1 and PAI-1 to each bind and inhibit plasminogen activators, these two serpins differ in their reaction rates for plasminogen activators (Table 2) (29,39). PAI-1, but not Serp-1, displayed prothrombotic activity in the mouse tail bleeding time measurements and increased platelet adhesion, whereas Serp-1 reduced platelet adhesion to fibronectin, consistent with prior reports of increased thrombosis with PAI-1 (16,45). Serp-1 thus either has additional functions or potentially lacks domains that cause platelet activation and thrombosis. These additional antithrombotic actions of Serp-1 may be the result of the much lower reaction rates for Serp-1 than PAI-1 with respect to plasminogen activators or may be due to an additional inhibition of clotting factor proteases, such as factor Xa (26,39).
Substituting Met for P1 in the Serp-1 (␣1-PI) chimera increased aneurismal dilation and resulted in a loss of anti-inflammatory and antiatherogenic activity. Substituting Thr for the P1Ј in the Serp-1 (PAI-2) chimera also removed anti-inflammatory and anti-atherogenic activity and caused a slight increase in thrombosis in PAI-1-deficient allografts. Serp-1, Serp-1 (PAI-2), and Serp-1 (Ala 6 ) all share an Arg at the P1 scissile bond site, but the presence of Arg in the P1 position was not enough to provide anti-inflammatory or anti-atherogenic activity. Substitution of the wild type Serp-1 P2-P7 RCL arm amino acid sequence of Ala, Ile, Thr, Leu, Ile, Pro (AITLIP) with 6 Ala residues represents a substitution with nonpolar aliphatic Ala for the wild type Serp-1 P2-P7 sequence, which is predominantly composed of nonpolar, aliphatic amino acid R groups (5 of 6). This 6-Ala substitution in the Serp-1 (Ala 6 ) chimera produced an excessively proinflammatory protein that induced excess plaque growth, thrombosis, and aneurysm formation.
The properties of the Serp-1 (Ala 6 ) chimera are of particular interest in that while retaining the P1-P1Ј RCL site of Serp-1 (RN), alteration of the RCL P2-P7 arm with Ala 6 caused a profound acceleration of all of the functions that native Serp-1 inhibited. Unlike wild type Serp-1, the Serp-1 (Ala 6 ) chimera induced a marked increase in thrombotic and inflammatory responses along with accelerated plaque growth and aneurysm formation. The proinflammatory activities of the Serp-1 (Ala 6 ) chimera as demonstrated here differ from the other serpins and chimeras tested in that the Ala 6 substitution of the P2-P7 RCL residues altered binding to and inhibition of uPA, reducing the K a modestly ( Table 2), and also increased plasmin activation (Fig. 4) with resulting proinflammatory functions and extensive local aneurysm formation ( Fig. 1). A loss of inhibitory activity for uPA and plasmin and an increase in plasmin activity would be predicted to increase promatrix metalloproteinase activation with increased inflammatory cell invasion and aneurysm formation. Certainly, the P1-P1Ј and adjacent RCL amino acid residues have been previously documented to alter activity of other serpins, such as anti-thrombin (1,4,47,48), and additionally other domains of serpins are known to alter serpin-mediated inhibitory actions (1,49,50).
Serp-1-mediated inhibition of inflammatory cell invasion and plaque growth was dependent upon the expression of uPAR in the donor aortic transplant artery, whereas the prothrombotic activity of Serp-1 (PAI-2) and the aneurysm formation induced by Serp-1 (Ala 6 ) are not dependent upon uPAR expression. The increase in inflammatory cell invasion, thrombosis, plaque growth, and aneurysm formation in the uPAR Ϫ/Ϫ allografts for some of the Serp-1 mutants may be associated with either the absence of uPAR in the donated aorta or with the presence of uPAR expression in adjacent recipient aorta or in circulating inflammatory cells in the recipient blood stream. Prior work has indicated that uPAR is associated with but not always required for cell migration, plaque growth, and aneurysm formation (46). The aortic allograft model used here provides a model of extensive vascular inflammation and rapid plaque growth. Although inducing a highly inflamed vascular environment, it should be noted that this is an artificial model utilized to mimic the marked inflammatory vascular responses seen in transplanted organs, so-called chronic rejection. Thus, immune allograft responses in addition to innate inflammatory responses to surgical trauma are both induced, providing a robust model where the inflammatory and immune responses are up-regulated.
We conclude that the highly potent viral serpin, Serp-1, has unique anti-inflammatory and anti-atherogenic activities, distinct from the mammalian serpin PAI-1. Serp-1 is also unique in displaying both platelet and monocyte inhibitory actions. Whereas the anti-inflammatory and anti-atherogenic actions require Arg-Asn at the Serp-1 P1-P1Ј, the mobile RCL P2-P7 arm adjacent to the P1-P1Ј scissile bond also has a key role. When the P1-P1Ј is altered, Serp-1 activity transforms to a dramatically more proinflammatory state and promotes thrombosis, plaque growth, and aneurismal dilation. Mutation of the P2-P7 sequence in the RCL to Ala 6 in the Serp-1 (Ala 6 ) mutant removed tPA, uPA, and plasmin protease-inhibitory activity and enhanced plasmin activity. The altered protease-inhibitory activity and enhanced plasmin activity of Serp-1 (Ala 6 ) promoted an excessive proinflammatory response with aneurysm formation.