Mutants of Plasminogen Activator Inhibitor-1 Designed to Inhibit Neutrophil Elastase and Cathepsin G Are More Effective in Vivo than Their Endogenous Inhibitors*

Neutrophil elastase and cathepsin G are abundant in-tracellular neutrophil proteinases that have an impor-tant role in destroying ingested particles. However, when neutrophils degranulate, these proteinases are re-leased and can cause irreparable damage by degrading host connective tissue proteins. Despite abundant endogenous inhibitors, these proteinases are protected from inhibition because of their ability to bind to anionic surfaces. Plasminogen activator inhibitor type-1 (PAI-1), which is not an inhibitor of these proteinases, possesses properties that could make it an effective inhibitor of neutrophil proteinases if its specificity could be redirected. PAI-1 efficiently inhibits surface-seques-tered proteinases, and it efficiently mediates rapid cellular clearance of PAI-1-proteinase complexes. Therefore, we examined whether PAI-1 could be engineered to inhibit and clear neutrophil elastase and cathepsin G. By introducing specific mutations in the reactive center loop of wild-type PAI-1, we generated PAI-1 mutants that are effective inhibitors of both proteinases. Kinetic analysis shows that the inhibition of neutrophil proteinases by these PAI-1 mutants is not affected by the se-questration of The as activity the versus the The inhibits in a molar Two for inhibitory

Neutrophils are the first defensive cells to extravasate from the circulation into infected areas where their primary role is to ingest foreign particles and eliminate them by using an arsenal of bactericidal, hydrolytic, and oxidative agents (1)(2)(3). The direct action of neutrophils is temporary because they degranulate soon after reaching the affected area, but their intracellu-lar proteinases can have a lasting effect.
In many chronic inflammatory disorders where there is a persistent influx and degranulation of neutrophils, their proteinases can overwhelm endogenous proteinase inhibitors, cause tissue degradation, and augment the inflammatory response. The broad substrate specificities of neutrophil elastase and cathepsin G are similar to the digestive proteinases, pancreatic elastase and chymotrypsin, respectively. Like their pancreatic counterparts, the neutrophil proteinases can degrade most proteins, including cross-linked extracellular matrix proteins such as collagens and elastin. In addition to degrading extracellular matrix proteins, these neutrophil proteinases can also intensify the host inflammatory response by both proteolytic and non-proteolytic mechanisms (4 -7).
The activities of neutrophil elastase and cathepsin G are primarily regulated by the serine proteinase inhibitors (serpins), 1 ␣ 1 -proteinase inhibitor (␣ 1 PI, also called ␣ 1 antitrypsin) and ␣ 1 -antichymotrypsin (␣ 1 ACT), respectively. These inhibitors are abundant plasma proteins, present in the circulation at high concentrations (8). The importance of these inhibitors in regulating neutrophil proteinase activity in extravascular tissues has been demonstrated in individuals with ␣ 1 PI deficiency who are more likely to develop early onset emphysema because of degradation of lung elastin (9,10).
Neutrophil elastase and cathepsin G are very basic proteins (pI Ն8.5) that bind negatively charged molecules like DNA and heparin with high affinity. Their binding to these anionic surfaces limits the accessibility of their active site to both small substrates and serpins. ␣ 1 PI and ␣ 1 ACT are very effective inhibitors of neutrophil elastase and cathepsin G in solution phase, but efficiency is greatly reduced in the presence of these surfaces. The reason for this is thought to be because of the phase separation of neutrophil elastase and cathepsin G from ␣ 1 PI and ␣ 1 ACT due to the inability of the inhibitors to bind these surfaces. Partitioning of these proteinases from their endogenous inhibitors could explain the tissue damage associated with many chronic inflammatory disorders because they are protected from inhibitors by anionic macromolecules present in cell debris (11)(12)(13).
Unlike ␣ 1 PI and ␣ 1 ACT, which are efficient inhibitors of proteinases in solution phase, other serpins are efficient inhib-itors of their target proteinases when they are bound to a surface. A classic example is antithrombin-III, whose rate of inhibition is enhanced by heparin (14). Another serpin that is able to inhibit proteinases bound to surfaces is plasminogen activator inhibitor type-1 (PAI-1), which is the principal inhibitor of the plasminogen activators (PAs), urokinase-type plasminogen activator and tissue-type plasminogen activator in vivo (15,16). Both PAs bind to a variety of surfaces, including cellular receptors, fibrin, and heparin. Surface binding, however, does not protect them from inhibition by PAI-1 (17)(18)(19)(20). Besides the ability of PAI-1 to inhibit PAs bound to receptors and surfaces, PAI-1 is able to mediate rapid cellular clearance of the target proteinase.
Upon complex formation with a proteinase, PAI-1 undergoes a rapid conformational change that increases its affinity for the clearance receptors of the low density lipoprotein (LDL) receptor family. These include the LDL receptor-related protein (LRP), the very low density lipoprotein receptor, and Megalin (21)(22)(23). We previously showed that the high affinity binding of PAI-1-proteinase complexes to LRP was independent of the proteinase but was mediated through a cryptic site on PAI-1 that is exposed when it is in a covalent complex with a proteinase. Furthermore, this enhanced clearance was significantly more efficient compared with clearance of proteinases, such as thrombin, in complex with other serpins, including antithrombin-III, ␣ 1 PI, and heparin cofactor-II (22). Thus, PAI-1 not only inhibits proteinases but also specifically promotes their clearance. In this study we have taken advantage of these innate properties of PAI-1, which normally is cleaved by these proteinases, and designed mutations in the reactive center loop (RCL) of PAI-1, which converts it to an efficient inhibitor of either pancreatic elastase or chymotrypsin or of the neutrophil proteinases, neutrophil elastase and cathepsin G. The efficiency of these PAI-1 mutants is not adversely affected by surfaces such as heparin or DNA, and they are markedly more efficient at promoting the cellular clearance and degradation of neutrophil proteinases compared with their natural serpin inhibitors. Together, these properties make the PAI-1 mutants more effective in vivo than their endogenous inhibitors at reducing the neutrophil proteinase concentrations in a model of lung inflammation.
Preparation of PAI-1 Mutant Inhibitors of Neutrophil Proteinases-A preferred method for producing PAI-1 mutants utilizes the commercially available Transformer site-directed mutagenesis kit (Clontech, Palo Alto, CA) (22). This kit uses a mismatched primer form of mutagenesis and is compatible with double stranded plasmid DNA. Thus the mutants can be constructed directly onto existing expression plasmids. The PAI-1A346V-V343A and PAI-1R346F mutants were made using the oligonucleotides (5Ј-CATAGCCTCAGCCGTCATGGCCCCC-GAG-3Ј and 5Ј-CATAGTCTCAGCCTTCATGGCCCCCGAGGAG-3Ј, respectively). These PAI-1 mutants were expressed and purified as described (24) and isolated and purified as described by Kvassman and Shore (25). Preparations with significant endotoxin levels as determined by coagulase kit (Bio-Whittacker, Walkersville, MD) following the phenyl-Sepharose step were further purified by either additional passes through phenyl-Sepharose or Detoxi-Gel (Pierce) until endotoxin levels were at or below 10 units/mg. Biacore Analysis of the Binding of the Proteinases, Inhibitors, and Their Complexes to LRP-Affinities of pancreatic elastase, neutrophil elastase, chymotrypsin, and cathepsin G for LRP either in complex with ␣ 1 PI, ␣ 1 ACT, PMSF, or PAI-1 mutants were measured by surface plasmon resonance using a BIA 3000 optical biosensor (Biacore AB, Uppsala, Sweden). Purified human LRP was immobilized at the level of 3000 response units. Remaining binding sites were blocked by 1 M ethanolamine, pH 8.5, and unbound proteins were washed out with 0.5% SDS. A flow cell with immobilized ovalbumin at the level of 500 response units was used as a control for nonspecific protein binding. All binding reactions were performed in standard HBS-P buffer, pH 7.4, containing 10 mM HEPES, 150 mM NaCl, and 0.005% Tween 20. Binding of proteases, inhibitors, and their complexes to LRP was measured at 25°C at a flow rate of 30 l/min for 4 min, followed by 4 min of dissociation. Chip surfaces were regenerated with subsequent 1-min pulses of 1 M NaCl, pH 4.0, and 1 M NaCl containing 10 mM NaOH, followed by 2 min of washing with HBS-P. Binding of proteinases, inhibitors, and proteinase-inhibitor complexes was measured using a range of concentrations (10 -0.15-nM). Collected data were analyzed with BIA evaluation 3.0 software (Biacore) using global analysis to fit a 1:1 Langmuire binding model with mass transfer limitation.
Cellular Endocytosis and Degradation Assays of Proteinases-A rat pretype-II pneumocyte cell line (T-II) was a generous gift from Dr. R. K. Mallampalli (University of Iowa College of Medicine, Iowa City, IA) (26). These cells were grown in Dulbecco's modified Eagle's medium (BioWhittaker, Walkersville, MD) supplemented with 10% fetal bovine serum (Hyclone, Logan, UT) containing penicillin and streptomycin. For endocytosis and degradation studies, T-II cells were seeded onto 12-well plates, (1-3 ϫ 10 5 cells/well) coated with 0.1% gelatin and allowed to adhere for 18 h at 37°C, 5% CO 2 in Dulbecco's modified Eagle's medium containing 10% bovine calf serum. Before addition of radiolabeled proteinases, cells were washed twice using serum-free Dulbecco's modified Eagle's medium and incubated for 30 min in serumfree Dulbecco's modified Eagle's medium containing 1% bovine serum albumin before addition of either preformed 125 I-labeled proteinaseinhibitor complexes or 125 I-labeled proteinases. Where indicated, RAP (1 M) was added prior to the 125 I proteinase-inhibitor complexes.
Quantitation of the endocytosed and degraded ligand was done as described (27) with minor modifications. Briefly, cellular degradation of the 125 I-labeled proteinases was determined by removing the medium and precipitating it using 10% trichloroacetic acid and 5% phosphotungstic acid (final concentration). Acid-soluble radioactivity was taken to represent degraded ligands. For quantitation of endocytosis, cell layers were washed twice with serum-free medium and incubated in serum-free medium containing trypsin and proteinase K (0.5 mg/ml) and 0.5 mM EDTA for 2-5 min at 4°C. The cells were then centrifuged at 6000 ϫ g for 2 min, and the radioactivity in the cell pellet was taken to represent the amount of endocytosed ligand.
To evaluate the effects of ␣ 1 PI and PAI-1 AV on the endocytosis and degradation of exogenously added 125 I elastase, cells were grown in medium containing 10% serum as described above. Washed monolayers were then incubated with 125 I elastase (10 nM) for 30 min, after which increasing concentrations of either ␣ 1 PI or PAI-1 AV were added and incubated for 4 h at 37°C to determine endocytosis or 18 h for to determine degradation.
Kinetics of Proteinase Inhibition by ␣ 1 PI, ␣ 1 ACT, and PAI-1 Mutants-Before measuring the second order rate constant for the inhibition of the neutrophil proteinases, the stoichiometry of the inhibitor as a proteinase substrate to a proteinase inhibitor (SI value) was determined. Briefly, a constant concentration of the neutrophil proteinases was incubated with an increasing concentration of the proteinase inhibitors in the presence or absence of heparin or DNA. After 30 min the appropriate chromogenic proteinase substrate was added to the reaction mixture, and the proteinase activity was measured. The results were plotted as activity of the proteinase versus the inhibitor concentration divided by the enzyme concentration. The intercept on the x-axis is therefore an indicator of whether the inhibitor inhibits in a 1:1 molar ratio or whether higher concentrations of the inhibitors are required.
Two methods were used to obtain kinetic parameters for the inhibition of the proteinases. For slow reactions (k i Ͻ 10 5 ), the PAI-1 mutants will be assumed to inhibit target proteinases by the two-step inhibitory mechanism first described for irreversible inhibitors by Kitz and Wilson (28) and Hastings et al. (29).
For fast reactions between the proteinases and the inhibitors (k i Ͼ10 5 ), the scheme described in Lawrence et al. (24) was used. The effects of heparin and DNA on the kinetics of inhibition were determined in the presence of either 0.1 mg/ml heparin or 0.1 mg/ml DNA.
In Vivo Lung Inflammation Assay-In vivo lung inflammation assays were performed essentially as described (31), with some modifications. Briefly, male C57BL/6J mice (age 8 -12 weeks) under anesthesia were intranasally instilled with either PBS or a mixture of endotoxin and fMLF, 100 and 50 g, respectively, in a total volume of 150 l in PBS. PAI-1 AV and PAI-1 F mutants, 50 g/25 g, respectively, or ␣ 1 AP/ ␣ 1 ACT, 50 g/25 g, respectively, in a total volume of 150 l in PBS were administered along with the endotoxin/fMLF instillation mixture. Additionally, PAI-1 AV /PAI-1 F , 100 g/25 g, respectively, or ␣ 1 AP/ ␣ 1 ACT, 100 g/25 g, respectively, in a total volume of 150 l or PBS were injected intraperitonally to both PBS-and endotoxin/fMLFtreated mice immediately following nasal instillation and again 16 -17 h postinstillation. At 24 -26 h postinstillation, animals were anesthetized and perfused with PBS. Lungs were then excised, washed briefly with 0.3% hexadecyl-trimethyl ammonium bromide in Tris-buffered saline, and stored on dry ice, followed by homogenization in Trisbuffered saline containing 1% bovine serum albumin and 0.3% hexadecyl-trimethyl ammonium bromide, followed by centrifugation to remove particulates. Neutrophil elastase and cathepsin G activities were measured in the lung homogenates using the chromogenic substrates described above. The effectiveness of the inhibitors as assessed by the neutrophil proteinase load in the lungs was determined by subtracting the activities of animals treated intranasally with PBS from the endotoxin/fMLF-treated animals, with or without added proteinase inhibitors. (Approved Animal Welfare Assurance A3379 -01 registered with the Public Health Services, Office of Laboratory Animal Welfare).

RESULTS AND DISCUSSION
Generation and Characterization of PAI-1 Mutants That Inhibit Pancreatic and Neutrophil Proteinases-Serpins act as "suicide inhibitors" because when a proteinase cleaves a serpin at a site within the RCL, termed the scissile bond (P1-P1Ј), it traps the proteinase in a non-reversible covalent complex (32,33). In many cases the amino acid composition of the serpin scissile bond reflects the substrate specificity of the target proteinase. The scissile bond of wild-type PAI-1 is Arg-346-Met-347, which enables it to inhibit many serine proteinases with trypsin-like specificities, but kinetic analysis showed that it is the most effective inhibitor of both PAs (34). Because PAI-1 is an efficient inhibitor of its target proteinases both in solution and in solid phase (17)(18)(19)(20) and because it is very efficient at promoting their cellular clearance, (27), we wanted to determine whether we could redirect PAI-1 specificity toward neutrophil elastase and cathepsin G, which wt PAI-1 does not inhibit, (35,36), and still retain these properties.
Mutations in a serpin scissile bond have been shown to redirect it to target another proteinase (37). Therefore we tested whether a PAI-1 mutant with the P1 Arg-346 residue changed to an alanine (PAI-1R346A, referred herein as PAI-1 A ) (38), could inhibit pancreatic elastase, which prefers Ala as the P1 residue. The PAI-1 A has no inhibitory activity against urokinase-type plasminogen activator (data not shown) but retained the ability to bind vitronectin with affinity, equal to wild-type PAI-1 (39). The rate of inhibition (k i ) of pancreatic elastase by PAI A is 4.6 ϫ 10 5 M Ϫ1 s Ϫ1 , which is comparable with the rate obtained by Laurent and Bieth (40) for ␣ 1 PI (4.7 ϫ 10 5 M Ϫ1 s Ϫ1 ). These results demonstrate that the specificity of PAI-1 can be redirected toward pancreatic elastase with a rate of inhibition similar to ␣ 1 PI. To target serine proteinases with chymotrypsin-like specificities, a second PAI-1 mutant was generated with a phenylalanine at the P1 position (PAI-1R346F, referred herein as PAI-1 F ). This mutant was tested against chymotrypsin, and the rate of inhibition of PAI-1 F was 6.1 ϫ 10 4 M Ϫ1 s Ϫ1 compared with 8.1 ϫ 10 5 M Ϫ1 s Ϫ1 for ␣ 1 ACT reported by Rubin et al. (41).
The results reported above demonstrate that the proteinase specificity of PAI-1 can be easily redirected by changing the amino acid composition of the P1 residue without a substantial loss in the rate of inhibition in solution phase. Therefore, these PAI-1 mutants were tested for their ability to inhibit other serine proteinases with similar specificity, specifically neutrophil elastase and cathepsin G. Kinetic analysis of PAI-1 A inhibition of neutrophil elastase yielded a k i of 1.4 ϫ 10 3 M Ϫ1 s Ϫ1 , which is about 300-fold less than the rate of inhibition of pancreatic elastase by PAI-1 A (k i ϭ 4.6 ϫ 10 5 M Ϫ1 s Ϫ1 ) and is ϳ10,000 fold less than ␣ 1 PI inhibition of neutrophil elastase (k i ϭ 1.3 ϫ 10 7 M Ϫ1 s Ϫ1 ). The decrease in the ability of PAI-1 A to inhibit neutrophil elastase was surprising because it efficiently inhibited pancreatic elastase with a rate equal to that of ␣ 1 PI. This discrepancy could be because of the preference of neutrophil elastase for valine over alanine as the P1 residue or because the proteinase was cleaving the RCL at another site than the P1-P1Ј, rendering it inactive as was demonstrated for wtPAI-1 (35). Sequence analysis of the reaction mixture of PAI-1 A with neutrophil elastase showed that PAI-1 A was being cleaved at both the Val-343-Ser-344 position (P3-P4) and at Ala-346-Met-347 (P1-P1Ј, data not shown), indicating that the inefficiency of PAI-1 A might be due in part to this non-productive cleavage at Val-343-Ser-344, coupled with the P1 alanine being a suboptimal residue for neutrophil elastase.
Therefore, to improve PAI-1 A as an inhibitor of neutrophil elastase, the Val-343 residue was replaced with alanine and the Ala-346 was replaced with valine. The resulting mutant, PAI-1V343A A346V (referred herein as PAI-1 AV ), has the nonproductive Val-343 cleavage site changed to alanine, which is a suboptimal cleavage site for neutrophil elastase, and the alanine at the P1 position changed to valine, which is preferred by neutrophil elastase as demonstrated by small peptide substrates (43). Valine was also chosen as the P1 residue because studies by Shubeita et al. (44) demonstrated that replacing the P1-P1Ј residues of wt-PAI-1 with those of ␣ 1 PI (Met-Ser) did not convert PAI-1 into an inhibitor of elastase or trypsin. Additionally, the PAI-1 AV mutant does not alter the length of the PAI-1 RCL, which is critical for its ability to inhibit its target proteinases (45).
Inhibition Kinetics of PAI-1 AV and PAI-1 F Compared with ␣ 1 PI and ␣ 1 ACT in the Presence or Absence of Anionic Surface-Kinetic analysis of the PAI-1 AV inhibition of neutrophil elastase showed an approximate 400-fold improvement in the  Table I). These results demonstrate that the rearrangement of the Ala-346 and Val-343 residues in the RCL dramatically improves the ability of PAI-1AV to inhibit neutrophil elastase.
To determine whether the PAI-1 AV and PAI-1 F mutants retain the ability of wtPAI-1 to inhibit surface-bound proteinases, their rates of inhibition of neutrophil proteinases were determined in the presence or absence of anionic surfaces. In solution phase, ␣ 1 PI is approximately a 20-fold faster inhibitor of neutrophil elastase than PAI-1 AV (Table I). Thus, PAI-1 AV is less effective than ␣ 1 PI in inhibiting neutrophil elastase in solution phase. However, in the presence of heparin, the k i for ␣ 1 PI inhibition of neutrophil elastase is reduced by about 200fold, which is in agreement with previous reports (46). In contrast, the k i for PAI-1 AV inhibition of neutrophil elastase was not as adversely affected by heparin (Table I), demonstrating that PAI-1 AV is as efficient as ␣ 1 PI under these conditions. Our results also demonstrate that DNA reduces the k i of ␣ 1 PI for the inhibition of elastase by about 10-fold (Table I), which agrees with Belorgey et al. (47), whereas DNA did not affect the rate of the inhibition of neutrophil elastase by PAI-1 AV to the same extent.
Anionic surfaces have been shown to have a more profound effect on the rate of inhibition of cathepsin G by ␣ 1 ACT and ␣ 1 PI (47). As is seen in Table I, the presence of heparin decreases the k i of ␣ 1 ACT and ␣ 1 PI inhibition of cathepsin G by a factor of 400 and 500, respectively, which agrees with the results of Ermolieff et al. (11). In solution phase, PAI-1 F is ϳ100-fold less effective at inhibiting cathepsin G than ␣ 1 ACT (Table I), but the presence of heparin did not severely decrease its rate of inhibition. Additionally, Table I shows that in the presence of heparin and DNA, PAI-1 F is about 10-and 200-fold more effective in inhibiting cathepsin G than ␣ 1 ACT and ␣ 1 PI, respectively.
Together these results demonstrate that even though ␣ 1 PI and ␣ 1 ACT are superior inhibitors of the neutrophil proteinases in solution, the presence of anionic surfaces, which are likely present at sites of inflammation in vivo, greatly effect ␣ 1 PI and ␣ 1 ACT inhibition but do not severely affect the PAI-1 AV and PAI-1 F mutants.

Binding of Neutrophil Elastase and Cathepsin G and their Complexes to LRP and Their Clearance by Lung Epithelial
Cells in Vitro-Previously we demonstrated that when PAI-1 is in a covalent complex with a proteinase, it binds LRP with high affinity and is cleared by cells more effectively than proteinases in complex with other serpins or with synthetic inhibitors. This clearance is thought to be mediated by a cryptic high affinity binding site for LRP that is exposed in PAI-1 only when it is in a covalent complex with a proteinase (27). To determine whether PAI-1 AV and PAI-1 F retained the ability of wtPAI-1 to promote cellular clearance when in complex with these proteinases, we measured their affinities for LRP and their cellular endocytosis and degradation compared with ␣ 1 PI, and ␣ 1 ACT.
Before analyzing the affinities of the serpin-proteinase complexes, the equilibrium dissociation constants for each individual serpin and proteinase were measured by their binding to immobilized LRP as measured by surface plasmon resonance. The results demonstrate that the PAI-1 mutants show weak binding to LRP (K D ϳ 0.2 M) and that the binding of ␣ 1 PI, and ␣ 1 ACT was also weak, showing K D s greater than 10 Ϫ4 M. Of the proteinases, neutrophil elastase showed moderate binding affinity to LRP (K D ϳ 0.065 M) with pancreatic elastase, chymotrypsin, and cathepsin G showing a K D Ͼ 10 Ϫ4 M. Fig. 1 shows representative surface plasmon resonance binding curves of neutrophil elastase in complex with either PAI-1 AV or ␣ 1 PI (Fig. 1, A and B, respectively) binding to immobilized LRP. These data show that neutrophil elastase in complex with ␣ 1 PI bound weakly to the LRP sensor chip, whereas neutrophil elastase in complex PAI-1 AV showed high affinity binding. The binding affinities of the proteinase:serpin pairs are summarized in Table II. These data indicate that both pancreatic and neutrophil proteinases in complex with the PAI mutants bind LRP with nanomolar to picomolar affinities, whereas proteinases in complex with either ␣ 1 PI or ␣ 1 ACT exhibited only very weak binding to LRP.
Because previous studies have shown that high affinity binding of PAI-1 in complex with a proteinase to immobilized LD-LRs in vitro can predict efficient cellular clearance, we wanted to determine whether these PAI-1 mutants also promote the cellular clearance of their proteinase-inhibitor complexes. For these studies the pancreatic and neutrophil proteinases were radiolabeled and preincubated with either ␣ 1 PI, ␣ 1 ACT, PAI-1 AV , PAI-1 F , or PMSF. These premade complexes were then added to a lung epithelial cell line generated from rat pre-  type-II pneumocytes (26); the degradation of the proteinase was measured in the presence or absence of RAP, an antagonist of ligand binding to LDLR (48). The cell-mediated degradation of each proteinase in complex with inhibitor indicated that all PAI-1-proteinase complexes are degraded by the pneumocytes ϳ10 -30-fold more efficiently than proteinases in complex with ␣ 1 PI, ␣ 1 ACT, or PMSF (Fig. 2, A-D, respectively). Additionally, the cellular degradation was mediated through internalization by an LDLR family member because the inclusion of RAP blocked the degradation of the PAI-I mutant proteinase com- plexes. Because the above study was performed using preformed complexes of radiolabeled proteinase with inhibitors, we wanted to determine the efficiency of the PAI-1 mutants to inhibit and mediate the cellular clearance of a free proteinase.
For these studies 125 I-labeled neutrophil elastase was added to a cell layer of the type-II pneumocytes followed by the addition of increasing concentrations of either ␣ 1 PI or PAI-1 AV . As is seen in Fig. 3, A and B, increased concentrations of the PAI-1 AV mutant added to the cells showed a concentration-dependent increase in the endocytosis and degradation of the labeled neutrophil elastase. In contrast, increasing concentrations of ␣ 1 PI did not result in an enhancement of neutrophil elastase cellular clearance.
The results shown in Figs. 2 and 3 demonstrate that cellular clearance of neutrophil proteinases in complex with PAI-1 mutants can by mediated by lung epithelial cells because they express members of the LDLR family (49 -51). This indicates that the clearance of PAI-1-proteinase-inhibitor complexes can occur locally rather than through hepatic receptors as is the case for most plasma proteinase-inhibitor complexes (52).
PAI-1 AV and PAI-1 F Reduce the Neutrophil Proteinase Activities in an in Vivo Model of Acute Lung Inflammation-The in vitro studies presented above demonstrate that PAI-1 AV and PAI-1 F , in the presence of anionic polymers such as proteoglycans and DNA, are efficient inhibitors of neutrophil elastase and cathepsin G. In addition, these PAI-1 mutants were much more efficient at mediating the cellular clearance of the neutrophil proteinases than ␣ 1 PI or ␣ 1 ACT. Therefore, to see whether their efficiency in vitro could also be demonstrated in vivo, we examined their ability to reduce neutrophil proteinase activities in a mouse model of acute lung inflammation through intranasally instilled lipopolysaccharide (LPS) and an fMLF peptide.
Instillation of LPS and/or fMLF into lungs has been previously used as a model to study the effects inflammatory cells have on the structure and function of lungs. Both LPS and fMLF are bacterial products that are recognized by many cells as signs of infection, to which many cells respond by secreting chemokines, which attract neutrophils and other defensive cells (53,54). In addition to responding to agents secreted by cells in response to bacterial products, inflammatory cells express receptors that recognize these bacterial products, which stimulates their extravasation from the circulation toward the focus of the infection.
For these studies, mice were intranasally instilled with LPS and fMLF in combination with either ␣ 1 PI and ␣ 1 ACT or PAI-1 AV and PAI-1 F , in conjunction with intraperitoneal injections of these inhibitor pairs. The mice received a second intraperitoneal injection of proteinase inhibitors 16 -17 h later, and the neutrophil proteinase levels in the lungs were determined 24 -26 h after the initial instillations. Fig. 4 shows the effects of human ␣ 1 PI/␣ 1 ACT and PAI-1 AV / PAI-1 F on the neutrophil elastase and cathepsin G activities in the lungs of mice instilled intranasally with endotoxin and fMLF peptide. PAI-1 AV /PAI-1 F significantly decreased proteinase activities (p Ͻ 0.05), whereas there was no significant decrease of proteinase activity in lungs treated with ␣ 1 PI/ ␣ 1 ACT compared with controls (p Ͼ 0.1). These data demonstrate that the PAI-1 mutants are more efficient at reducing the load of neutrophil proteinases in the lungs in vivo than injections of equivalent amounts of ␣ 1 PI and ␣ 1 ACT. This is notable because the mice used in this study have a full complement of endogenous plasma neutrophil proteinase inhibitors (42) and enhancing their levels to those of their human counterparts does not influence the outcome. However, injecting the mice with the engineered PAI-1 variants significantly reduced the proteinase burden in the lungs. Thus, in this study we show that PAI-1 can be engineered to be an effective inhibitor of neutrophil proteinases in environments where the endogenous inhibitors, ␣ 1 PI and ␣ 1 ACT, are impaired. Additionally, our results demonstrate that redirecting the specificity of PAI-1, which has distinct capabilities such as the ability to inhibit surface-bound proteinases and mediate their efficient cellular clearance, can yield potentially superior inhibitors of proteinases involved in pathologies.