EP1173602A1 - Ecotin derivatives - Google Patents

Abstract

This invention provides a class binding proteins that specifically bind to and modulate (e.g. enhance) the activity of polypeptides having a chymotrypsin fold (e.g. serine proteases). The binding proteins are based on the structure of ecotin. It was discovered that modification of the amino or carboxyl terminus and/or randomization of one or more of loops 50s, 60s, 80s or 100s will provide an ecotin variant library from which can be selected binding molecules (e.g. protease modulators) specific to virtually any serine protease. Depending on the ecotin variant and the target serine protease, the modulator can act as a serine protease inhibitor or as a serine protease activator. Specific agonists (enhancers) of Factor IXa are disclosed.

Description

ECOTINE DERIVATIVES

CROSS-REFERENCE TO RELATED APPLICATIONS

This application is related to an application entitled Engineering Ecotin- Narian Modulators of Serine Proteases, filed on April 12, 1999, naming Charles S. Craik, and Robert J. Fletterick, and which is herein incorporated by reference in its entirety for all purposes.

STATEMENT AS TO RIGHTS TO INVENTIONS MADE UNDER FEDERALLY SPONSORED RESEARCH AND DEVELOPMENT

This invention was made with work was supported in part by Grant Number DK 39304 from the National Institutes of Health and by Grant Number MCB-9604379 from the National Science Foundation. The Government ofthe United States of America may have certain rights in this invention.

FIELD OF THE INVENTION

This invention relates the field of protein engineering. In particular, this invention teaches the production of a wide variety of protease binding proteins (e.g., activity modulators) based on the structure ofthe bacterial protease inhibitor ecotin.

BACKGROUND OF THE INVENTION

The serine proteases are a large family of enzymes involved in a wide variety of vital biological processes. The crucial physiological functions of these enzymes in blood coagulation, fibrinolysis, complement pathways, viral maturation, apoptosis, and cancer make them important targets for efforts to design and engineer potent and specific inhibitors. A highly selective protease inhibitor can serve as a powerful tool to block key proteolytic activities for dissecting proteolytic pathways and cascades and elucidating the in vivo roles of particular proteases in complex biological processes. Ultimately, this may lead to the development of innovative therapies for life-threatening diseases. Previous studies have established that, for the chymotrypsin family of serine proteases, the electrostatic and hydrophobic characteristics of the substrate binding pocket are critical factors determining substrate specificity (Perona and Craik (1995) Protein Set, 4(3): 337-360). This region has been the major focus of structural, functional, and mechanistic studies as well as inhibitor design efforts. Since many mammalian serine proteases have the same primary specificity, a small molecule inhibitor that only exploits the interactions at the primary binding pocket may not have sufficient discrimination towards similar enzymes. Macromolecular substrate-like serine protease inhibitors, such as bovine pancreatic trypsin inhibitor (BPTI), bind to the target serine protease through a single surface loop that includes the critical PI residue. This residue fits into the binding pocket ofthe target protease in a substrate-like conformation to lock the enzyme in a complex formed between the protease and the inhibitor. The common contact area between the protease and the inhibitor is small and clustered around the active site ofthe enzyme. The highly homologous structures ofthe different enzyme active sites make the design of a^* highly selective and potent inhibitor for a particular protease a challenging task.

SUMMARY OF THE INVENTION This invention pertains to the discovery that native ecotin or ecotin variants of this invention can activate (increase the activity) of a serine protease. This invention thus provides methods of enhancing (increasing) the activity of a serine protease having a chymotrypsin fold. The methods involve contacting the serine protease with a native ecotin or an ecotin variant as described herein. Serine proteases whose activity is thus enhanced include, but are not limited to, plasma kallikrein, Factor XLIa, Factor Xla, Factor IXa, Factor Vila, Factor Xa, Factor Ila (thrombin), Factor Clr, Factor Cls, Factor D, Factor B, C3 convertase, trypsin, chymotrypsin, elastinase, enterokinase, urokinase plasminogen activator, tissue plasminogen activator, plasmin, tissue kallikrein, acrosin, α-subunit nerve growth factor, γ-subunit nerve growth factor, granulocyte elastase, cathepsin G, mast cell chymase, mast cell tryptase. In a particularly preferred embodiment, the serine protease is Factor IXa. This invention also provides methods of identifying a protein that activates (e.g. enhances/increases the activity of) a serine protease. The methods involve contacting the serine protease with a binding protein library (e.g. an ecotin variant library as described herein) and selecting one or more members ofthe protein binding library that specifically activate the serine protease.

In still another embodiment, this invention provides affinity matrices for the isolation of polypeptides characterized by a chymotrypsin fold. The affinity matrices comprise a solid surface to which is attached one or more ecotin variants of this invention.

.?- In a preferred embodiment, the surface can be a surface of a planar solid, or a bead and can be suspended in solution or packed into a "chromatography" column. In a particularly preferred embodiment, the ecotin variant is attached to a polyHis tag (e.g. His₆) that is bound to an Ni-NTA substrate. The affinity matrices of this invention can be used to isolate polypeptides characterized by a chymotrypsin fold from a mixture of molecules. The methods involve contacting an affinity matrix of this invention with the "sample" mixture under conditions in which the ecotin variant can bind to its target polypeptide(s) to produce a bound polypeptide; and then separating the ecotin variant bound to the polypeptide from the mixture. In addition, kits for the purification of a polypeptide characterized by a chymotrypsin fold (e.g. a serine protease) are provided comprising a container containing one or more ofthe affinity matrices of this invention. The kits optionally include instructional materials teaching the use ofthe affinity matrix for the isolation of a polypeptide characterized by a chymotrypsin fold. In certain preferred embodiments, the ecotin variants of this invention do not include the serine protease inhibitors disclosed or claimed in U.S. Patent 5,719,041 and/or native ecotin.

DEFINITIONS

The terms "isolated" "purified" or "biologically pure" refer to material which is substantially or essentially free from components which normally accompany it as found in its native state.

The term "nucleic acid" refers to a deoxyribonucleotide or ribonucleotide polymer in either single- or double-stranded form, and unless otherwise limited, encompasses known analogs of natural nucleotides that can function in a similar manner as naturally occurring nucleotides.

The terms "polypeptide", "peptide" and "protein" are used interchangeably herein to refer to a polymer of amino acid residues. The terms apply to amino acid polymers in which one or more amino acid residue is an artificial chemical analogue of a corresponding naturally occurring amino acid, as well as to naturally occurring amino acid polymers.

An amino acid, identified by name herein "e.g., arginine" or "arginine residue" as used herein refers to natural, synthetic, or version ofthe amino acids Thus, for example, an arginine can also include arginine analogs that offer the same or similar functionality as natural arginine with respect to their ability of be incorporated into a polypeptide, effect folding of that polypeptide and effect interactions of that polypeptide with other polypeptide(s).

The phrase "nucleic acid encoding" or "nucleic acid sequence encoding" refers to a nucleic acid that directs the expression of a specific protein or peptide. The nucleic acid sequences include both the DNA strand sequence that is transcribed into RNA and the RNA sequence that is translated into protein. The nucleic acid sequences include both full-length nucleic acid sequences as well as shorter sequences derived from the full- length sequences. It is understood that a particular nucleic acid sequence includes the degenerate codons ofthe native sequence or sequences which may be introduced to provide codon preference in a specific host cell. The nucleic acid includes both the sense and antisense strands as either individual single strands or in the duplex form.

The term "mutation", when used in reference to a polypeptide refers to the change of one or more amino acid residues in a polypeptide to residues other than those found in the "native" or "reference (pre-mutation) form of that polypeptide. Mutations include amino acid substitutions as well as insertions and/or deletions. A mutation does not require that the particular amino acid substitution or deletion be made to an already formed polypeptide, but contemplates that the "mutated" polypeptide can be synthesized de novo, e.g. through chemical synthesis or recombinant means. It will be appreciated that the mutation can include replacement of a natural amino acid with an "unnatural" amino acid. A "protease" is a polypeptide that cleaves another polypeptide at a particular site (amino acid sequence). The protease can also be self-cleaving.

A protease is said to be "specific" for another polypeptide when it characteristically cleaves the other "substrate" polypeptide at a particular amino acid sequence. The specificity can be absolute or partial (i.e., a preference for a particular amino acid or amino acid sequence).

The term "specifically binds" when used to refer to binding proteins herein indicates that the binding preference (e.g., affinity for the target molecule/sequence is at least 2 fold, more preferably at least 5 fold, and most preferably at least 10 or 20 fold over a non- specific (e.g. randomly generated molecule lacking the specifically recognized amino acid or amino acid sequence) target molecule.

The term "phage", when used in the context of polypeptide display, includes bacteriophage as well as other "infective viruses", e.g. viruses capable of infecting a mammalian, or other, cell. The term "chymotrypsin fold" refers to the anti-parallel beta barrel protein "fold" characteristic of trypsin, chymotrypsin, elastase, and related serine proteases (see, e.g., Branden and Tooze (1991) Introduction to Protein Structure, Garland Publishing, New York; Creighton (1993) Proteins, 2nd edition, W.H. Freeman & Co., New York; Schulz and Schirmer (1979) Principles of Protein Structure, Springer-Nerlag, New York; Perutz (1992) Protein Structure - New Approaches to Disease and Therapy, W.H. Freeman & Co., New York; Fersht (1976) Enzyme Structure and Mechanism, 2nd ed., W.H.Freeman & Co., New York).

A "protease substrate" is a polypeptide that is specifically recognized and cleaved by a protease.

The term "randomized" when referring to a polypeptide indicates that a collection of polypeptides contains members differing in amino acid compositien at the randomized site(s). When the polypeptide is fully randomized, the collection contains a representative polypeptide for every possible natural amino acid at each randomized site . The term "randomized" when referring to a nucleic acid refers to a collection of nucleic acids that encode a randomized collection of polypeptides.

The term "modulate" when used with respect to protease activity refers to an alteration in the rate of reaction (protein hydrolysis) catalyzed by a protease. An increase in protease activity results in an increase in the rate of substrate hydrolysis at a particular protease concentration and a protease modulator that produces such an increase in protease activity is referred to as an "activator" or "protease agonist". The terms "activator" or "agonist" are thus used synonymously. A decrease in protease activity refers to a decrease in the rate of substrate hydrolysis at a particular protease concentration. Such a decrease may involve total elimination of protease activity. A protease modulator that produces a decrease in protease activity is referred to as a "protease inhibitor". It will be appreciated that generally the increase or decrease is as compared to the protease absent the protease modulator.

The term "detectable label" refers to any material having a detectable physical or chemical property. Such detectable labels have been well-developed in the field of immunoassays and, in general, any label useful in such methods can be applied to the present invention. Thus, a label is any composition detectable by spectroscopic, photochemical, biochemical, immunochemical, electrical, optical or chemical means. Useful labels in the present invention include magnetic beads (e.g. DynabeadsTM), fluorescent dyes (e.g., fluorescein isothiocyanate, Texas red, rhodamine, and the like), radiolabels (e.g., ³H, ¹²ⁱI, ³ⁱS, ¹⁴C, or ³²P), enzymes (e.g., LacZ, CAT, horse radish peroxidase, alkaline phosphatase and others, commonly used as detectable enzymes, either as marker gene products or in an ELISA), and colorimetric labels such as colloidal gold or colored glass or plastic (e.g. polystyrene, polypropylene, latex, etc.) beads. It will be recognized that fluorescent labels are not to be limited to single species organic molecules, but include inorganic molecules, multi-molecular mixtures of organic and/or inorganic molecules, crystals, heteropolymers, and the like. Thus, for example, CdSe-CdS core-shell nanocrystals enclosed in a silica shell can be easily derivatized for coupling to a biological molecule (Bruchez et al. (1998) Science, 281: 2013-2016). Similarly, highly fluorescent quantum dots (zinc sulfide-capped cadmium selenide) have been covalently coupled to biomolecules for use in ultrasensitive biological detection (Warren and Nie (1998) Science, 281: 2016-2018).

The label may be coupled directly or indirectly to the ecotin variant to methods well known in the art. As indicated above, a wide variety of labels may be used, with the choice of label depending on the sensitivity required, ease of conjugation of the compound, stability requirements, available instrumentation, and disposal provisions.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 illustrates the tetrameric complex of ecotin-trypsin. The high- resolution crystal structure of rat anionic trypsin-ecotin complex illustrated a network of interactions between ecotin-ecotin monomers, between ecotin-trypsin at the primary binding site, and between ecotin-trypsin at the secondary binding site.

Figure 2(a) shows an SDS-PAGE gel of ecotin primary site and dimer Interface variants. Figure 2(b) shows an SDS-PAGE gel of ecotin secondary site variants.

Figure 3 shows a K_t plot of ecotin truncation variants. The K s of ecotin truncation variants are plotted in log scale along Y-axis.

Figure 4(a) shows K_t data of ecotin 60s loop variants. Figure 4(b) shows K_t data of ecotin 100s loop variants. Figure 4(c) shows Kj data of ecotin 60s and 100s loop variants.

Figure 5 shows a comparison of liquid and plate amplification. The total phage yields from the liquid and plate amplification procedures are compared. The panning experiment is conducted with ecotin M84R+60A4 library panning against uPA for four rounds. Figure 6 shows the modeled electrostatic interaction at the 60s loop of ecotin. The crystal structure of uPA was superimposed to one trypsin molecule in the ecotin-trypsin tetrameric complex by matching the residues Serl95, His57, Asp 102, and Asp 189 of each enzyme using the program MidasPlus (Computer Graphic Laboratory, UCSF) with RMS deviation of 0.66 A.

Figure 7 shows the rate of hydrolysis ofthe peptide nitroanilide substrate by the parent human factor IXa.

Figure 8 shows the rate of hydrolysis ofthe peptide nitroanilide substrate by factor IXa passed over an antifactor XI immunoaffinity column. Figure 9 shows the rate of hydrolysis of the peptide nitroanilide substrate by factor IXa absorbed to and eluted from an anti-factor IX immunoaffinity

DETAILED DESCRIPTION

This invention pertains to the discovery that binding proteins, particularly binding proteins based on the structure of ecotin (e.g., ecotin and ecotin-variants), are capable of enhancing the activity of a serine protease (e.g. Factor 9). Thus, in one embodiment, this invention provides binding proteins that specifically bind and modulate (enhance or inhibit activity) polypeptides having a chymotrypsin fold (e.g. serine proteases) and methods of making such binding proteins.

The crucial physiological functions of serine proteases, e.g., in blood coagulation, fibrinolysis, complement pathways, viral maturation, apoptosis, and cancer make them important targets for efforts to design and engineer potent and specific activators or inhibitors. A highly selective protease modulator can serve as a powerful tool to regulate key proteolytic activities, for dissecting proteolytic pathways and cascades and for elucidating the in vivo roles of particular proteases in complex biological processes. The ecotin-variants used as serine protease modulators (e.g. activators, synergists) in this invention are engineered based on the discovery that ecotin, a macromolecular serine protease inhibitor found in the periplasm of Escheήchia coli, offers a unique platform to develop a wide variety of binding proteins. The binding proteins are based on the structure of ecotin and thus referred to as ecotin-derived binding proteins or ecotin variants. This invention exploits the structure of ecotin to act as a scaffold that orients the domains comprising a primary and secondary binding site that mediate ecotin/ecotin and ecotin/substrate interactions. Using the ecotin backbone as a scaffold the domains can be varied according to the methods of this invention to produce new modulators (e.g. inhibitors) of serine proteases. Moreover, the binding proteins do not only bind serine proteases, but are capable of specifically binding polypeptides characterized by the presence of a chymotrypsin fold.

The binding proteins of this invention are created by modifying the ecotin- protease interactions, particularly those that are distal from the ecotin reactive site described by Chung et al. (1983) J. Biol Chem., 258(18): 11032-11038. Five sites are important for binding activity and/or modulation: the N-terminus, the C-terminus, the reactive site (primary site), the secondary site, and the dimer interface.

Ecotin is a competitive serine protease inhibitor that strongly inhibits trypsin, chymotrypsin and elastase and many other serine proteases with comparable potencies

(Chung et al. (1983) supra.). The inhibitor was purified and its reactive site was determined to be Met 84 which lies within a disulfide bonded protein segment (McGrath et al. (1991) J. Biol. Chem., 266(10): 6620-6625). The gene encoding ecotin was cloned and expressed recombinantly in E. coli. (McGrath et al (1991) supra.; McGrath et al. (1991) /. Mol Biol, 222(2): 139-142). The 142 amino acid monomeric ecotin, with a PI Met at residue 84, dimerizes in solution at submicromolar concentrations. The high-resolution crystal structure of ecotin complexed with rat anionic trypsin shows two ecotin monomers form a dimer through interactions at their C-termini (McGrath et al. (1994) EMBOJ., 13(7): 1502-1507) Figure 1). As explained in detail in Example 1 , the ecotin dimer binds to two trypsin molecules at opposite ends to form a heterotetramer with a two-fold symmetry axis. The crystal structure also reveals a network of interactions between ecotin and trypsin. The protein-protein interaction surface between the inhibitor and the protease consists of two distinct areas, each provided by one ofthe two ecotin molecules. The first area, known as the "primary binding site", involves the reactive site loop of ecotin, i.e. the 80s loop

(residues 81-86), the 50s loop(residues 52-54), and the active site of trypsin. The second area, known as the "secondary binding site" 25 A away (± 3-5 A depending upon the ligand), includes two surface loops of ecotin, the 60s loop(residues 66-70) and 100s loop(residues 108-113), and the C-terminal region of the protease (including part ofthe C-terminal helix and part ofthe 90s loop of trypsin). The dimer interface, primarily amino acids 130 to 142 is important for intersite binding interactions. In some cases, the N- and C-terminus may also bind to the proteinase. The N terminus includes amino acids 1-7 (or an insertion therein) while the carboxyl terminus includes amino acids 132-142 (or an insertion therein). Considering just the primary and secondary sites, ecotin's four loops form two interface regions between ecotin and the protease resulting in a combined surface area of 2800 A². The enormous buried interface area between ecotin and its target protease is far greater than that of most other protease-inhibitor complexes. In addition, ecotin's unique secondary binding site plays a major role in determining the strength of interaction between ecotin and the protease. Through systematic mutagenesis at the 60s and 100s loops (see Example 1, herein), we have demonstrated the significance of this secondary binding site for selective inhibition against several proteases, such as rat trypsin and urokinase-type plasminogen activator (uPA) (see, also Yang and Craik (1998) J. Mo I. Biol, 279: 1001- 1011).

It was a discovery of this invention that ecotin can be randomized in one or more ofthe above-described five domains to generate a library of ecotin variants (ecotin-like molecules) that are specific inhibitors or agonists of serine proteases typically not targeted by native ecotin. In embodiment, it was a discovery of this invention that mutation (e.g. via randomization) ofthe 50s loop, and/or the 60s loop and/or the 80s loop and/or the 100s loop will produce ecotin variants having widely differing specificity and activity. Such ecotin variants are useful as serine protease binding proteins and/or for modulating the activity of a wide variety of serine proteases.

Thus, for example, in one embodiment, ecotin was modified in the 60s loop to produce variants having a wide range of activity against rat trypsin and uPA.

In another embodiment, it was a discovery of this invention that native ecotin and ecotin variants act synergistically on Factor D a to increase the Factor IXa protease activity. Without being bound to a particular theory it is believed that ecotin acts to restructure the conformation of Factor IXa, and by implication other serine proteases, to convert it to a more active enzyme by direct binding. Thus, this invention also provides methods of modulating (e.g. inhibiting or activating) a serine protease, particular a serine protease characterized by a chymotrypsin fold, or analogous folds such as found in viral proteases ofthe NS3, 2C, and 3C classes (see, e.g., Bazan and Fletterick (1989) Virology, 171(2): 637-639, Bazan and Fletterick (1989) FEBS Letts., 249(1): 5-7, and Bazan and Fletterick (1988) Proc. Natl. Acad. Sci. USA, 85(21): 7872-7876). The methods involve contacting the serine protease with native ecotin or an ecotin variant of this invention. I. Polypeptide binding proteins and serine protease modulators.

A) Polypeptide binding proteins.

In one embodiment, the ecotin variants of this invention can be used simply as binding proteins. In this context, the ecotin variants act in a manner analogous to antibodies in that they specifically bind to a target molecule. Preferred ecotin variants specifically bind to polypeptides characterized by the presence of a chymotrypsin fold. It will be appreciated that many serine proteases are characterized by a chymotrypsin fold (e.g. chymotrypsin, elastase, thrombin, urokinase type plasminogen activator, factor IXa, factor Xa, etc.). In addition, however, there are a number of polypeptides characterized by a chymotrypsin fold that are not serine proteases (e.g., the 3C viral protease that has a cysteine in place of the active site serine), often not even proteases, and yet are still specifically bound by the ecotin variant binding proteins of this invention. Such proteins include, but are not limited to haptoglobin, EGF binding protein, etc. Other target polypeptides (particularly those that are modulated by ecotin variants) include, but are not limited to, the 3C proteases, e.g., the 3C proteases from poliovirus, rhinovirus, and encephalovirus.

The ecotin variants of this invention can be used as binding proteins in a wide variety of contexts analogous to the use of antibodies. Thus, for example, they can be labeled with a detectable label and used to probe for the target polypeptide(s) to which they specifically bind, they can be used as binding agents in "immunoassays" (e.g. sandwich assays, lateral flow assays, etc.), and they can be used as binding partners in purification systems to selectively isolate their target (cognate) polypeptide from a mixture of molecules.

Thus, in one particularly preferred embodiment, the binding proteins of this invention are used as affinity chromatography reagents. In this embodiment, one or more ecotin variants of this invention is attached to a solid substrate or an isolatable label (e.g. a magnetic bead, fluorescent moiety separable in a FACs system, etc.). The ecotin variant is contacted with the mixture from which the target polypeptide(s) is to be isolated under conditions that permit protein recognition and binding. The ecotin variant and its bound protein are separated from the mixture and the bound protein is then optionally separated from the ecotin complex (e.g. by high salt, high pH, low pH, temperature change, organic solvents, chaotropic agent, denaturing agents (e.g. urea, guanadinium salts, etc.).

A wide variety of formats for such separations are well known to those of skill in the art. Thus, for example, the ecotin variant can be attached to a regular or irregular, planar or non-planar, solid - r«porous surface. Such surfaces can include, but are not limited to the surfaces of beads, pores, planar surfaces, microchannels, capillaries, and the like. Thus, for example, the ecotin variant can be coupled to particles that are packed into a column permitting the sample mixture, buffers or other reagents to be flowed past the binding protein, or conversely, ecotin variant-bound particles can be suspended in a solution containing the polypeptide that is to be separated. After binding occurs, the bound particles can be separated from the mixture (e.g. via cenrrifugation, the use of magnetic particles, etc.).

The manner of linking a wide variety of compounds to various surfaces is well known and is amply illustrated in the literature Means of coupling the ecotin variants to a surface are well known to those of skill in the art. Linkage ofthe ecotin variant to a surface can be covalent, or by charge or other non-covalent interactions. The surface may be specifically derivatized to provide convenient linking groups (e.g. cystein, hydroxyl, amino, etc.). Covalent linkage ofthe ecotin variant to the surface may be direct or through a covalent linker. Proteins contain a variety of functional groups; e.g., carboxylic acid (COOH) or free amine (-NH2) groups, which are available for reaction with a suitable functional group on either the surface or on a linker attached to the surface. . Proteins, for example, may be joined to linkers or to functional groups coupling through their amino or carboxyl termini, or through side groups of various constituent amino acids. Thus, coupling through a disulfide linkage to a cystein is common. Generally linkers are either hetero- or homo-bifunctional molecules that contain two or more reactive sites that may each form a covalent bond with the respective binding partner (i.e. surface or ecotin variant). Linkers suitable for joining biological binding partners are well known to those of skill in the art. For example, a protein molecule may be linked by any of a variety of linkers including, but not limited to a peptide linker, a straight or branched chain carbon chain linker, or by a heterocyclic carbon linker.

Heterobifunctional cross linking reagents such as active esters of N-ethylmaleimide have been widely used. See, for example, Lemer et al. (1981) Proc. Nat. Acad. Sci. (USA), 78: 3403-3407 and Kitagawa et al. (1976) J. Biochem., 79: 233-236, and Birch and Lennox (1995) Chapter 4 in Monoclonal Antibodies: Principles and Applications, Wiley-Liss, N.Y.).

In one embodiment, where the linker is itself a polypeptide, it can be expressed as a fusion with the ecotin variant. Thus, for example, in one prefeπed embodiment, the ecotin variant is expressed with a poly-Histidine (e.g. His₆) tag that in turn binds to a Ni-NTA substrate, e.g. a NiNTA-column). Of course, the ecotin variant can be bonded to the surface by any of a variety of other well-known chemical procedures. For example, the linkage may be by way of heterobifunctional cross-linkers, e.g. SPDP, carbodiimide, glutaraldehyde, or the like. In another preferred embodiment the linkage is achieved using cyanogen bromide. Virtually any surface that is resistant to reagents used in binding and/or eluting the captured polypeptide and that does not substantially interfere with the ecotin variant/target polypeptide binding interaction is suitable for use as a matrix (surface) material. Particularly preferred matrix materials include glass beads, controlled pore glass, magnetic beads, various membranes or rigid various polymeric resins such as polystyrene, polystyrene/latex, and other organic and inorganic polymers, both natural and synthetic. Illustrative polymers include polyethylene, polypropylene, poly(4-methylbutene), polystyrene, polymethacrylate, poly(ethylene terephthalate), rayon, nylon, poly(vinyl butyrate), polyvinylidene difluoride (PNDF), silicones, polyformaldehyde, cellulose, cellulose acetate, nitrocellulose, and the like. Other materials which may be employed, include paper, glasses, ceramics, metals, metalloids, semiconductive materials, cements or the like. In addition, substances that form gels, such as proteins (e.g., gelatins), lipopolysaccharides, silicates, agarose and polyacrylamides can be used. Polymers which form several aqueous phases, such as dextrans, polyalkylene glycols or surfactants, such as phospholipids, long chain (12-24 carbon atoms) alkyl ammonium salts and the like are also suitable. Where the solid surface is porous, various pore sizes may be employed depending upon the nature ofthe system.

In preparing the surface, a plurality of different materials may be employed, e.g., as laminates, to obtain various properties. For example, protein coatings, such as gelatin can be used to avoid non-specific binding, simplify covalent conjugation, enhance signal detection or the like.

If covalent bonding between a compound and the surface is desired, the surface will usually be polyfunctional or be capable of being polyfunctionalized. Functional groups which may be present on the surface and used for linking can include carboxylic acids, aldehydes, amino groups, cyano groups, ethylenic groups, hydroxyl groups, mercapto groups and the like.

B) Serine protease modulators.

Serine proteases are a family of enzymes that utilize a uniquely activates serine residue in the substrate binding site to catalytically hydrolyze peptide bonds. The active site serine can be characterized by the irreversible reaction of its side chain hydroxyl group with diisopropylfluorophosphate (DFP). Of all the serines in the protein, DFP only reacts with the catalytically active serine to form a phosphate diester. The target ofthe serine proteases are specific peptide bonds in proteins and often their substrates are other serine proteases that are activated from an inactive precursor form (a zymogen) by the catalytic cleavage of a specific peptide bond in their structure.

Serine proteases are implicated in a wide variety of physiological processes including, but not limited to blood coagulation, fϊbrinolysis, complement activation, fertilization, hormone production, tumor cell metastasis, emphysema, arthritis, thrombosis, and hemostasis. In case of viral infection, the viral proteases have been identified in infected cells. Such viral proteases include, for example, HIV protease associated with AIDS and NS3 protease associated with Hepatitis C, and the like. These viral proteases play a critical role in the virus life cycle.

Proteases have also been implicated in cancer metastasis. For example, increased synthesis ofthe protease urokinase has been correlated with an increased ability to metastasize in many cancers. Urokinase activates plasmin from plasminogen which is ubiquitously located in the extracellular space and its activation can cause the degradation of the proteins in the extracellular matrix through which the metastasizing tumor cells invade. Plasmin can also activate the collagenases thus promoting the degradation ofthe collagen in the basement membrane surrounding the capillaries and lymph system thereby allowing tumor cells to invade into the target tissues (Dano, et al (1985) Adv. Cancer. Res., 44: 139). A number of other pathological conditions are associated with altered serine protease regulation. For example, cerebral infarction (stroke), coronary infarction, thrombosis, and bleeding disorders are associated with abnormal regulation of plasma kallikrein, Factor XIIA, Factor Xla, Factor LXa, Factor Vila, Factor Xa and Factor Ila

(thrombin). Inflammation, rheumatoid arthritis and autoimmune disease are associated with abnormal regulation of Factor Clr, Factor Cls, Factor D, and Factor B. Digestive disorders (e.g., pancreatitis) are associated with altered regulation of trypsin, chymotrypsin, elastase and enterokinase. Clotting disorders are associated with urokinase plasminogen activator (uPA), tissue plasminogen activator, or plasmin. Infertility is associated with abnormal regulation of acrosin. Inflammation and allergic response is associated with granulocyte elastase activity, cathepsin G, mast cell chymases, and mast cell tryptases. Tumor invasiveness is associated with urokinase plasminogen activator and elastase activity.

-ι: Thus, specific modulators (e.g. activators or inhibitors) of serine protease activity are expected to prove useful in the treatment and/or mitigation of symptoms associated with these conditions. For example, regulation of Factor IXa, will be useful in the development of anti-thrombotics that are not hemoragic and which could be used for deep vein thromboses, while modulation of collagenases is expected to be useful in the treatment of metastatic disease and the retardation of tumor invasiveness.

It is noted that, in addition to the usually inhibitory activity, it was a discovery that native ecotin and ecotin variants can act as serine protease activators enhancing serine protease activity. Thus, for example, it was observed that native bacterial ecotin, whose typical cognate protease is presently unknown will act as a significant agonist on mammalian (e.g. human) Factor IXa. Other ecotin variants (e.g. M84R) will act as potent Factor IXa inhibitors. Thus the compounds of this invention can be used to increase or decrease (modulate) serine protease activity.

II. Ecotin variants: modifications of ecotin. As indicated above, it was a discovery of this invention that ecotin variants can be used to specifically target and regulate a wide variety of serine proteases. The ecotin variants in some instances can inhibit target serine proteases, while in other instances can act agonistically with serine proteases to increase activity and/or binding specificity or avidity. In a preferred embodiment, the ecotin variants of this invention substantially comprise a native ecotin backbone, but contain one or more mutations in particular regions. In a particularly preferred embodiment, the mutations are in one or more of the following regions: the primary binding site, including, but not limited to the 50s loop (amino acids 52- 54 of native ecotin) and the 80s loop (amino acids 81 to 86 of native ecotin), and the secondary binding site including, but not limited to, the 60s loop (amino acids 67-70 of native ecotin), and the 100s loop (amino acids 108-113 of native ecotin). Mutations adjacent to the loops are also desirable and thus 50s loop mutations include mutations in amino acids 50-56, more preferably 51-55 and most preferably 52-54, 80s loop mutations include mutations of amino acids 79-88, more preferably 80-87, and most preferably 81-86, 60s mutations include mutations of amino acids 65-72, more preferably 66-71, and most preferably 67-70, and 100s loop mutations include mutations of amino acids 106-115, more preferably 107-114, and most preferably 108-113. Additions and the C-terminus and C- terminus may also be effective in binding and modulating protease function. Alterations in the ecotin interface can be made to enhance or limit binding interactions as well. Suitable mutations include replacement of one naturally occurring amino acid with another different naturally occurring amino acid, replacement of an amino acid with a non-naturally occurring amino acid (e.g. an amino acid analogue). The mutations can also include deletions or insertions of one or more amino acids. Similar modifications can be made at the ecotin carboxyl and amino termini. Thus, for example, up to 10 amino acids, more preferably up to 8 amino acids, and most preferably up to 7 amino acids can be deleted from either or both termini.

Ecotin molecules modified as described above can be represented by the generic formula:

T'i-X'- ^ x'-L^k-X'-L^ x'- ¹⁰⁰⁵,-^-^ (I)

where X¹ is a polypeptide having the sequence of amino acids 8 through 50 of native ecotin (SEQ ID NO: 1), X² is a polypeptide having the sequence of amino acids 56 through 65 of native ecotin, X³ is a polypeptide having the sequence of amino acids 72 through 78 of native ecotin, X⁴ is a polypeptide having the sequence of amino acids 88 through 106 of native ecotin, X⁵ is a polypeptide having the sequence of amino acids 115 through 135 of native ecotin, L^50s, L°°^s, L^80s, and, L^100s are independently an amino acid or a polypeptide consisting of 2 to about 15 amino acids, more preferably 2 to about 7 amino acids, T¹ and T² are independently an amino acid or a polypeptide consisting of 2 to about 120 amino acids, more preferably 2 to about 50 amino acids, and most preferably 2 to about 15 amino acids, and i, j, k, m, n, and p are independently 0 or 1.

In one preferred embodiment, the ecotin variant can be represented by the formula: T₁ ¹-X¹-aa⁵¹-aa⁵²-aa⁵³-aa⁵⁴-aa⁵⁵-X²-aa⁶⁶-aa⁶⁷-aa⁶⁸-aa⁶⁹-aa⁷⁰-aa⁷¹-X -aa⁷⁹-aa⁸⁰-aa⁸¹-aa⁸²- aa⁸³-aa⁸⁴-aa⁸⁵-aa⁸⁶-aa⁸⁷-X⁴-aa¹⁰⁷-aa¹⁰⁸-aa^,O9-aa^π0-aa^{ι π}-aa¹¹²-aa¹¹³-aa^{1 I4}-X⁵-T_p ²

where T¹ is a polypeptide having the formula aa¹-aa²-aa³-aa⁴-aa⁵-aa⁶-aa⁷-, T² is a polypeptide having the formula aa¹³⁶-aa¹³⁷-aa¹³⁸-aa¹³⁹-aa¹⁴⁰-aa^14I-aa¹⁴², and aa¹, aa², aa³, aa⁴, aa⁷, aa⁶, aa⁷, aa⁵¹, aa⁵², aa⁵³, aa⁵⁴, aa⁵⁵, aa⁶⁶, aa⁶⁷, aa⁶⁸, aa⁶⁹, aa⁷⁰, aa⁷¹, aa⁷⁹, aa⁸⁰, aa⁸¹, aa⁸², aa^S3, aa⁸⁴, aa⁸⁵, aa⁸⁶, aa⁸⁷, aa¹⁰⁷, aa¹⁰⁸, aa¹⁰⁹, aa¹¹⁰, aa¹¹¹, aa¹¹², aa^{1 13}, aa^{1 14}, aa¹³⁶, aa¹³⁷, aa¹³⁸, aa¹³⁹, aa¹⁴⁰, aa¹⁴¹, and aa¹⁴² are optionally present amino acids that, when present, are independently selected from the group consisting of alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, lysinε, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, and valine. In a particularly preferred embodiment, Tj'-X¹ is a polypeptide having the sequence of amino acids 1 through 50 of native ecotin and X -T is a polypeptide having the sequence of amino acids 115 through 142 of native ecotin.

It will be appreciated that in particularly preferred embodiments, the ecotin is a variant and not a native ecotin and is capable of said specifically binding to and altering the activity of a serine protease.

Amino acids aa⁵¹-aa⁵²-aa -aa⁵⁴-aa⁵⁵- correspond substantially to the 50s loop, amino acids -aa⁶⁶-aa^D7-aa⁶⁸-aa⁶⁹-aa⁷⁰-aa⁷¹- correspond substantially to the 70s loop, -aa⁷⁹- aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵-aa⁸⁶-aa⁸⁷- correspond substantially to the 80s loop and -aa¹⁰⁷- aa¹⁰⁸-aa¹⁰⁹-aa^u0-aa^{l u}-aa^U2-aa¹¹³-aa¹¹⁴-correspond substantially to the 100s loop. It will be appreciated that the carboxyl and/or amino terminus and/or any number of loops can be randomized simultaneously. Thus preferred variants include ecotin having mutations in the following combinations of loops: 50s, 60s, 80s, 100s; 50s and 60s, 50s and 80s, 50s and

100s, 60s and 80s, 60s and 100s, 80s and 100s; 50s and 60s and 80s, 50s and 80s and 100s, 50s and 60s and 100s, 50s and 80s and 100s; and 50s and 60s and 80s and 100s. Any of these mutations can be combined with mutations in the amino or carboxyl terminus as indicated above. Insertions or deletions into one or more of these regions are encompassed by this invention. In a preferred embodiment, an insertion into a loop (50s, 60s, 80s, or 100s) will typically comprise no more than about 15 amino acids, more preferably no more than about 8 amino acids, and most preferably no more than about 4 amino acids.

Similarly, insertions at either terminus will typically comprise no more than about 120, preferably no more than about 50 amino acids, more preferably no more than about 10 amino acids and most preferably no more than about 4 amino acids. In some cases, a complete domain (e.g. a fibronectin type III domain) can be added to the N- or C-termini. Other modifications of ecotin or the ecotin variants that do not significantly alter the activity (e.g. specificity, avidity, or modulatory activity) of the molecule(s) are also contemplated. Such modifications include, but are not limited to, conservative amino acid residue substitutions, the attachment of label(s) or linker(s) to the molecule, minor alterations to facilitate cloning and expression (e.g. addition of a methionine at the amino terminus to provide an initiation site or introduction of anAatli site into the nucleic acid vector to facilitate cloning) and the like. Such modifications that do not alter activity are routine and well known to those of skill in the art.

III. Identification/selection of ecotin variants.

In a preferred embodiment ecotin variants having a particular binding specificity and/or avidity and/or a particular modulatory activity (e.g. protease inhibition or agonism) are identified by providing a library (a collection) comprising a number of different ecotin variants. The library is then screened against one or more target serine proteases and members ofthe library having the desired avidity and/or specificity and/or modulatory activity are selected. Methods of screening polypeptide libraries to select members having particular binding specificity and/or avidity and/or modulatory activity are well-known to those of skill in the art. For example, binding specificity and avidity can be determined using simple binding assays ofthe type generally used for measuring antibody binding avidity or specificity (e.g. competitive binding assays, or direct K measurements using, for example, a BiaCore). Modulatory activity assays involves contacting the ecotin or ecotin variants to be screened, with one or more "target" serine protease(s) under conditions in which the serine protease is normally capable of cleaving its substrate. The effect ofthe ecotin variant on the protease activity can then be assayed according to standard methods. Ecotin variant production and assays are described in more detail below.

A Production of libraries of ecotin variants.

Libraries of ecotin variants can be produced by any of a wide number of methods including, but not limited to chemical syntheses of each individual variant, combinatorial based syntheses, array-based combinatorial syntheses, recombinant expression of each individual variant, recombinant expression of randomized libraries, bacterial display systems, and phage display systems.

1) Phage display

The ability to express polypeptides on the surface of bacteria or of viruses that infect bacteria (bacteriophage or phage) makes it possible to isolate a single binding polypeptide or a single polypeptide having a particular activity from libraries of greater than 10¹⁰ nonbinding clones. To express polypeptides on the surface of phage (phage display), a nucleic acid encoding the polypeptide is inserted into the gene encoding a phage surface protein (e.g., pill) and the polypeptide-surface fusion protein is displayed on the phage surface (McCafferty et al. (1990) Nature, 348: 552-554; Hoogenboom et al. (1991) Nucleic Acids Res. 19: 4133-4137). Since the polypeptides on the surface of the phage are functional, phage bearing binding polypeptides can be separated from non-binding phage by antigen affinity chromatography (McCafferty et al. (1990) Nature, 348: 552-554). Depending on the affinity ofthe antibody fragment, enrichment factors of 20 fold -

1,000,000 fold are obtained for a single round of affinity selection. By infecting bacteria with the eluted phage, however, more phage can be grown and subjected to another round of selection. In this way, an enrichment of 1000 fold in one round can become 1,000,000 fold in two rounds of selection (McCafferty et al. (1990) Nature, 348: 552-554). Thus even when enrichments are low (Marks et al. (1991) J. Mol. Biol. 222: 581-597), multiple rounds of affinity selection can lead to the isolation of rare phage.

Phage display has been successfully applied to a wide range of peptides and proteins, including antibodies McCafferty et al (1990) Nature, 348: 552-554), growth hormone (Bass et al. (1990) Proteins: Struct. Fund. Genet. 8(4): 309-314), DNA binding proteins (Jamieson et al (1994) Biochem., 33(19): 5689-5695), enzymes (McCafferty et al. (1991) Protein Eng., 4(8): 955-961); Corey et al. (1993) Gene, 128(1): 129-134); Soumillion et al. (1994) J. Mol. biol, 237(4): 415-422), and macromolecular protease inhibitors (Roberts et al. (1992) Proc. Natl. Acad. Sci. USA, 89(6): 2429-2433); Pannekoek et al (1993) Gene, 128(1): 135-140; Wang et al. (1995) J. Biol. Chem., 270(20): 12250- 12256); Markland et al. (1996) Biochem., 35: 8058-8067; Markland et al. (1996) Biochem., 35: 8045-8057).

We have developed an ecotin phage display system that presents fully functional ecotin on the surface of filamentous phage (Wang et al. (1995) J. Biol. Chem., 270: 12250-12256) and used the system to investigate the macromolecular recognition between uPA and ecotin. uPA is a serine protease (collagenase activator) that plays an active role in extracellular proteolysis, cell migration, and tissue remodeling processes (Fazioli et al. (1994) Trends Pharmacol. Sci., 15(1): 25-29). Because of its implication in cancer metastasis and tumor invasion, uPA has become an important target for drug design and inhibitor development efforts. We have shown previously that a high affinity inhibitor of uPA can be isolated from a library of phage-displayed ecotin variants at the PI and PI' positions (Wang et al. (1995) supra.). In addition, phage display techniques are described herein to modify ecotin's secondary and other sites and thereby modify potency and specificity against target proteases. In a particularly preferred embodiment, phage display libraries are created that express ecotin, but are "randomized", or contain deletions or insertions, in particular regions (e.g. the 50s loop, the 80s loop, the 60s loop or the 100s loop). Nucleic acids encoding all possible amino acid variants at particular sites, can be prepared and inserted into the vectors comprising the phage display library.

The "randomized" nucleic acids are made according to methods well known to those of skill in the art. For example, in one approach, the nucleic acids can be chemically synthesized using "doped" nucleotide reagents during the coupling steps forming the "randomized" codons. However, in a particularly preferred embodiment, however, the randomized nucleic acids are created using amplification (e.g., PCR) cloning with degenerate primers. In this embodiment, degenerate primers are used to amplify ecotin templates where the primers are degenerate in regions expressing the domain ofthe ecotin it is desired to randomize. In one embodiment, such primers contain N at the desired position, or more preferably introduce codons of the form NNS, where N is A/C/G/T and S is C/G. Detailed protocols for the production of ecotin variant libraries using phage display technology are provided in Example 2.

2) Site-directed mutagenesis

In addition to or as an alternative, ecotin variants can also be expressed from nucleic acids that are modified by site-directed mutagenesis. Methods of site directed mutagenesis are well known to those of skill in the art. In a preferred embodiment, site- directed mutagenesis is performed by the method of Kunkel (1985) Proc. Natl. Acad. Sci. USA, 82(2): 488-492 as described in Example 1.

3) Array-based approaches.

In another embodiment, ecotin variant libraries can be created using combinatorial-based polypeptide synthesis techniques. Array-based combinatorial synthesis techniques are well known to those of skill in the art (see, e.g. See, e.g., U.S. Patent No. 5,143,854; PCT Publication Nos. WO 90/15070, WO 92/10092 and WO 93/09668; and Fodor et al (1991) Science, 251, 767-77). Briefly, in this approach, photolithographic methods are used to selectively couple derivatized amino acids at discrete locations in a solid-phase synthesis system. Highly complex arrays can be produced in relatively few coupling steps (Id.). It is recognized that chemical synthesis of long polypeptides is difficult. However, many of the difficulties are overcome by providing the "common" domains of the ecotin variants as pre-formed polypeptides derivatized for further coupling. Single residue additions are then only carried on for the residues that are to be randomized.

B Screening for desired activity.

Once a library of ecotin variants is created, the library, or a subset thereof, is screened for the desired binding specificity and or avidity, and/or serine protease modulatory activity. Screens for binding avidity and/or specificity and for effect on serine protease activity are well known to those of skill in the art.

1) Screens for binding avidity and/or specificity.

a) Direct binding assays. In direct binding assays the ability of one or more ecotin variants to bind to a serine protease (e.g. trypsin, uPA, etc.) is assayed. Simple binding assays are well known to those of skill in the art. In one embodiment, either the ecotin variant or the serine protease is labeled, the ecotin variant and the serine protease are contacted with each other and the association of the labeled moiety its respective binding partner is detected and/or quantified. Alternatively, both the ecotin variant and the protease can both be labeled and the association ofthe labels then indicates binding. The use of fluorescent labels capable of fluorescence resonance energy transfer (FRET) greatly facilitates the detection of such an association (e.g., the fluorescence ofthe labels is typically quenched when they are brought into proximity to each other, see, e.g., Stryer (1978) Ann. Rev. Biochem., 47: 819-846.). Direct binding assays can also be performed in solid phase where either the ecotin variant or protease is immobilized on a solid support. When the ecotin variant is immobilized it is contacted with the protease (optionally labeled) and conversely where the protease is immobilized it is contacted with the ecotin variant(s) (optionally labeled). After washing away unbound material, remaining bound ecotin variant/protease complexes indicate binding of the ecotin variant (s) to the protease. Where the non-immobilized ecotin variant or protease was labeled, detection of the label associated with the solid support provides a measure of ecotin variant/protease binding. Fluorescence resonance energy transfer systems (FRET) are suitable for use in the solid phase as well.

Neither ecotin variant nor protease need be labeled prior to the assay. An "indirect" subsequently applied label (e.g. a labeled antibody specific for the ecotin variant or protease) can be used to detect the ecotin variant or protease in the ecotin variant/protease complex. Alternatively, no label need be used. The bound polypeptides (e.g. bound phage) can simply be recovered, and optionally expanded, for another round of selection or other uses.

It will be appreciated that neither ecotin variant nor protease need be labeled in the assays. Other means of detecting complex formation are known to those of skill in the art (e.g. electrophoresis, density gradient centrifugation, etc.).

b) BiaCore assays.

Selection for increased avidity involves measuring the affinity of an ecotin variant for one or more target serine proteases. Methods of making such measurements are well known to those of skill in the art. In one preferred embodiment, for example, the K of a ecotin variant and the kinetics of binding to a target protease inhibitor are measured using a BIAcore, a biosensor based on surface plasmon resonance. For this technique,^_serine protease is coupled to a derivatized sensor chip capable of detecting changes in mass. When ecotin is passed over the sensor chip, particularly when additional protease is available in solution to form the tetrameric complex, the ecotin variant binds to the serine protease resulting in an increase in mass that is quantifiable. Measurement ofthe rate of association as a function of ecotin variant concentration can be used to calculate the association rate constant (k_on). After the association phase, buffer is passed over the chip and the rate of dissociation of ecotin variant (k_off) determined. The equilibrium constant _d is then calculated as k_ofi k_on and thus is typically measured in the range 10^"5 to 10^"12. Affinities measured in this manner correlate well with affinities measured in solution by fluorescence quench titration.

2) FRET assays.

As indicated above, fluorescent resonance energy transfer (FRET) systems can also be used to assay protein-protein interactions. In FRET-based assays, both components (e.g. both the ecotin variant and the serine protease) are labeled with fluorescent labels. The absorption and emission spectra ofthe labels are selected such that one label emits at a wavelength that the other absorbs. When the labels are brought into proximity to each other (e.g., by formation ofthe tetrahedral complex) they quench thereby decreasing the fluorescence ofthe mixture. FRET is a powerful technique for measuring protein-protein associations and has been used previously to measure the polymerization of monomeric actin into a polymer (Taylor et al. (1981) J. Cell Biol, 89: 362-367) and actin filament disassembly by severing (Yamamoto et al. (1982) J. Cell Biol, 95: 711-749). 3) Liquid crystal assay systems.

In still another embodiment, binding of ecotin variant to the serine protease can be detected by the use of liquid crystals. Liquid crystals have been used to amplify and transduce receptor-mediated binding of proteins at surfaces into optical outputs. Spontaneously organized surfaces can be designed so that formation ofthe ecotin variant serine protease tetrahedral complex on these surfaces, trigger changes in the orientations of 1- to 20-micrometer-thick films of supported liquid crystals, thus corresponding to a reorientation of ~10⁵ to 10⁶ mesogens per protein. Binding-induced changes in the intensity of light transmitted through the liquid crystal are easily seen with the naked eye and can be further amplified by using surfaces designed so that protein-ligand recognition causes twisted nematic liquid crystals to untwist (see, e.g., Gupta et al. (1998) Science, 279: 2077- 2080). This approach to the detection of protein/protein interactions does not require labeling ofthe analyte, does not require the use of electroanalytical apparatus, provides a spatial resolution of micrometers, and is sufficiently simple that it is useful in biochemical assays and imaging of spatially resolved chemical libraries.

4) Screens for modulatory activity.

As indicated above, in one embodiment native ecotin, or the ecotin variants of this invention are be screened for enzymatic (e.g. protease inhibitory or agonistic) activity. Typically such screens involve combining the ecotin or ecotin variant(s) together with a protease of interest and a substrate for that protease under conditions in which the protease typically has proteolytic activity and determining the effect ofthe presence absence or amount of ecotin or ecotin variant on the proteolytic activity ofthe protease.

Assays for protease activity are well known to those of skill in the art (see, e.g. Haugland (1996) Handbook of Fluorescent Probes and Research Chemicals, Molecular Probes, Eugene OR). Typically the protease activity is assayed with respect to its "native" target polypeptide. However, in many embodiments, the protease activity can be assayed by using an "indicator" substrate that provides a signal indicative of hydrolysis (see, e.g., chromogenic substrate Nα-benzyloxy-carbonyl-L-glyylproylarginine-7-amino-4- methylcoumarin (Z-GPR-AMC) available from Bachem Biosciences, Inc (King of Prussia, PA), chromogenic substrate Z-γ-Glu(α-t-butoxy)-Gly-Arg- 7-nitroanilide (Spectrozyme UK), etc.). A large number of such indicators are well known to those of skill in the art and are often sold in kits for determining protease activity (see, e.g., EnzChek Protease Assay Kits, from Molecular Probes, OR). In a preferred embodiment, the rate of hydrolysis of a target substrate by a particular protease is assayed. For example, as illustrated in Examples 1 and 2, the rate of hydrolysis of Z-GPR-AMC substrate by of rat and bovine trypsin was assayed (e.g. by the change in emission to 460 n ) in the presence of various concentrations of ecotin variants. The resulting data can be fit to the equation derived for kinetics of reversible tight-biding inhibitors, e.g. by non-linear regression analysis to determine the values for apparent K_- and true K_ (see, Examples 1 and 2).

5) Substrates for screening systems.

The activity of native ecotin or ecotin variants can be determined against virtually any protease. Serine proteases, however, are preferred. Suitable serine proteases include, but are not limited to plasma kallikrein, Factor XLIa, Factor Xla, Factor IXa, Factor Vila, Factor Xa, Factor Ila (thrombin), Factor Clr, Factor Cls, Factor D, Factor B, C3 convertase, trypsin, chymotrypsin, elastinase, enterokinase, urokinase plasminogen activator, tissue plasminogen activator, plasmin, tissue kallikrein, acrosin, α-subunit nerve growth factor, γ-subunit nerve growth factor, granulocyte elastase, cathepsin G, mast cell chymase, mast cell tryptase

6) High-throughput screens.

In one embodiment, the assays ofthe present invention offer the advantage that many samples can be processed in a short period of time. For example, plates having 96 or as many wells as are commercially available can be used. In addition, the serine protease or ecotin variants can be attached to solid supports and spatially arranged to form distinct arrays, such as rows of dots or squares, or lines. This, coupled to sophisticated masking, assay and readout machines greatly increase the efficiency of performing each assay and detecting and quantifying the results. It is possible with current technologies to efficiently make vast numbers (10⁶ or more) of peptides having specified sequences and array them at distinct locations in a chip, and then to detect fluorescent associated with each position ofthe chip. See, e.g., U.S. Patent No. 5,143,854; PCT Publication Nos. WO 90/15070, WO 92/10092 and WO 93/09668; and Fodor et al. (1991) Science, 251, 767-77.

Conventionally, new chemical entities with useful properties (e.g., inhibition of myosin tail interactions) are generated by identifying a chemical compound (called a "lead compound") with some desirable property or activity, creating variants ofthe lead compound, and evaluating the property and activity of those variant compounds. However, The current trend is to shorten the time scale for all aspects of drug discovery. Because of the ability to test large numbers quickly and efficiently, high throughput screening (HTS) methods are replacing conventional lead compound identification methods.

In one preferred embodiment, high throughput screening methods involve providing a library containing a large number of compounds (test compounds) potentially having the desired activity. Such "combinatorial chemical libraries" are then screened in one or more assays, as described herein, to identify those library members (particular chemical species or subclasses) that display the desired specificity, avidity or activity. The compounds thus identified can serve as conventional "lead compounds" or can themselves be used as potential or actual therapeutics. U.S. Patent 5,559,410 discloses high throughput screening methods for proteins, U.S. Patent 5,585,639 discloses high throughput screening methods for nucleic acid binding (i.e., in arrays), while U.S. Patents 5,576,220 and 5,541,061 disclose methods of screening for ligand/antibody binding.

In addition, high throughput screening systems are commercially available (see, e.g., Zymark Corp., Hopkinton, MA; Air Technical Industries, Mentor, OH; Beckman Instruments, Inc., Fullerton, CA; Precision Systems, Inc., Natick, MA, etc.) These systems typically automate entire procedures including all sample and reagent pipetting, liquid dispensing, timed incubations, and final readings ofthe microplate in detector(s) appropriate for the assay. These configurable systems provide high throughput and rapid start up as well as a high degree of flexibility and customization. The manufacturers of such systems provide detailed protocols for the various high throughput assays. Thus, for example, Zymark Corp. provides technical bulletins describing screening systems for detecting the modulation of gene transcription, ligand binding, and the like.

1) Assaying multiple agents. While each assay mixture can be utilized to assay the effect of a ecotin variant or the effect on a single protease, it will be recognizes that multiple ecotin variants or target proteases can also be screened in a single assay mixture. In such a multi-agent assay embodiment, two or more, preferably 4 or more, more preferably 16 or more and most preferably 32, 64, 128, 256, or even 512 or more ecotin variants or target proteases are screened in a single assay reaction mixture. A positive result in that assay indicates that one or more ofthe ecotin variants are modulators ofthe target protease, or that one or more of the target proteases are acted upon by the ecotin variant. In this instance, in a preferred embodiment, the method is repeated wherein the candidate agents are separated out to identify the modulator individually, or to verify that the agents work in conjunction to provide the difference in binding specificity, affinity, or avidity. Thus, for example, an assay originally run with 16 ecotin variants (e.g. per well) may be re-run as four assays each containing four ofthe original 16 test agents. Again those assays testing positive can be divided and re-run until the ecotin variant or variant(s) responsible for the positive assay result are identified.

It is also noted that multiple test agents can be assayed together to identify ecotin variants that are additive or even synergistic in their effect on a protease, or conversely, to identify test agent(s) that are antagonistic in their effects on a protease. In one embodiment, the method further comprises the step of entering the identity of an ecotin variant that has been identified to modulate activity of a protease in accordance with the present invention into a database of therapeutic, diagnostie or bio agricultural lead compounds. In some cases it may be desirable to perform further assays on the compounds which have been identified herein. For example, activity ofthe identified compounds can be further assessed in areas other than their ability to modulate protease activity. For example, their ability to affect growth or proliferation of cells, particularly tumor cells, etc. can be assessed.

IV. Production of identified ecotin variant.

The native ecotin or ecotin variants of this invention can be isolated (purified) from bacterial sources or alternatively can be recombinantly expressed. Alternatively, the polypeptides can be chemically synthesized according to standard methods.

A) De novo chemical synthesis.

The native ecotin or ecotin variants of this invention may be synthesized using standard chemical peptide synthesis techniques. Where the sequence is amenable, the molecule may be synthesized as a single contiguous polypeptide. Where larger molecules are desired, subsequences can be synthesized separately (in one or more units) and then fused by condensation ofthe amino terminus of one molecule with the carboxyl terminus of the other molecule thereby forming a peptide bond.

It will be noted that de novo chemical synthesis can be used to incorporate non-natural amino acids as well as natural amino acids into polypeptides.

Solid phase synthesis in which the C-terminal amino acid ofthe sequence is attached to an insoluble support followed by sequential addition ofthe remaining amino acids in the sequence is the preferred method for the chemical synthesis of the polypeptides of this invention. Techniques for solid phase synthesis are described by Barany and Merrifield, Solid-Phase Peptide Synthesis; pp. 3-284 in The Peptides: Analysis, Synthesis, Biology. Vol. 2: Special Methods in Peptide Synthesis, Part A., Merrifield et al (1963) J. Am. Chem. Soc, 85: 2149-2156, and Stewart et al, Solid Phase Peptide Synthesis, 2nd ed. Pierce Chem. Co., Rockford, 111. (1984).

B Recombinant expression.

In a preferred embodiment, the ecotin variants of this invention are synthesized using recombinant DNA methodology. Generally this involves creating a DNA sequence that encodes polypeptide, placing the DNA in an expression cassette under the control of a particular promoter, expressing the protein in a host, isolating the expressed protein and, if required, renaturing the protein.

DNA encoding native ecotin or ecotin variants of this invention may be prepared by any suitable method as described above, including, for example, cloning and restriction of appropriate sequences, amplification techniques, or direct chemical synthesis. Chemical synthesis produces a single stranded oligonucleotide. This may be converted into double stranded DNA by hybridization with a complementary sequence, or by polymerization with a DNA polymerase using the single strand as a template. One of skill would recognize that while chemical synthesis of DNA is limited to sequences of about 100 bases, longer sequences may be obtained by the ligation of shorter sequences.

Alternatively, subsequences may be cloned and the appropriate subsequences cleaved using appropriate restriction enzymes. The fragments may then be ligated to produce the desired DNA sequence.

In a preferred embodiment, the ecotin or ecotin variants of this invention may be cloned using DNA amplification methods such as polymerase chain reaction (PCR). Thus, for example, the nucleic acid sequence is PCR amplified, using a sense primer containing one restriction site (e.g., Ndel) and an antisense primer containing another restriction site (e.g., Hindlll). This will produce a nucleic acid encoding the ecotin or ecotin variant and having terminal restriction sites. This nucleic acid can then be easily ligated into a vector containing a nucleic acid encoding the second molecule and having the appropriate corresponding restriction sites. Suitable PCR primers are provided in Examples 1 and 2, and others can be determined by one of skill in the art using the sequence information provided in SEQ ID No: 1. Typical vectors for use in this invention contain transcription and translation terminators, transcription and translation initiation sequences, and promoters useful for regulation of the expression ofthe particular nucleic acid. The vectors optionally comprise generic expression cassettes containing at least one independent terminator sequence, sequences permitting replication ofthe cassette in eukaryotes, or prokaryotes, or both, (e.g., shuttle vectors) and selection markers for both prokaryotic and eukaryotic systems. Vectors are suitable for replication and integration in prokaryotes, eukaryotes, or preferably both. See, Giliman and Smith (1979), Gene, 8:81-97; Roberts et al (1987), Nature, 328:731-734; Berger and Kirnmel, Guide to Molecular cloning Techniques, Methods in Enzymology, Vol 152m Academic Press, Inc., San Diego, CA (Berger); Sambrook and Ausubel. Product information from manufacturers of biological reagents and experimental equipment also provide information useful in known biological methods. Such manufacturers include the SIGMA chemical company (Saint Louis, MO), R&D systems (Minneapolis, MΝ), Pharmacia LKB Biotechnology (Piscataway, ΝJ), Clontech Laboratories, Inc. (Palo Alto, CA), Chem Genes Corp., Aldrich Chemical Company (Milwaukee, WI), Glen Research, Inc., Gibco BRL Life Technologies, Inc. (Gaithersberg, MD), Fluka Chemica-Biochemika Analytika (Fluka Chemie AG, Buchs, Switzerland), and Applied Biosystems (Foster City, CA), as well as many other commercial sources known to one of skill.

The nucleic acid sequences encoding the ecotin or ecotin variants may be expressed in a variety of host cells, including E. coli, other bacterial hosts, yeast, and various higher eukaryotic cells such as the COS, CHO and HeLa cells lines and myeloma cell lines.

The recombinant protein gene will be operably linked to appropriate expression control sequences for each host (e.g., E. coli, or Staphylococcus). For E. coli this includes a promoter such as the T7, trp, or lambda promoters, a ribosome binding site and preferably a transcription termination signal. For eukaryotic cells, the control sequences will include a promoter and preferably an enhancer derived from immuno globulin genes, SV40, cytomegalovirus, etc., and a polyadenylation sequence, and may include splice donor and acceptor sequences.

The plasmids ofthe invention can be transfected into the chosen host cell by well-known methods such as calcium phosphate transfection, electroporation, fusion ofthe recipient cells with bacterial protoplasts containing the DΝA, treatment of the recipient cells with liposomes containing the DΝA, DΕAΕ dextran, receptor-mediated endocytosis, electroporation, micro-injection ofthe DΝA directly into the cells, infection with viral vectors, etc. Cells transformed by the plasmids can be selected by resistance to antibiotics conferred by genes contained on the plasmids, such as the amp, gpt, neo and hyg genes.

Once expressed, the recombinant ecotin or ecotin variant polypeptides can be purified according to standard procedures ofthe art, including ammonium sulfate precipitation, affinity columns, column chromatography, gel electrophoresis and the like (see, generally, R. Scopes, Protein Purification, Springer- Verlag, N.Y. (1982), Deutscher, Methods in Enzymology Vol. 182: Guide to Protein Purification., Academic Press, Inc. N.Y. (1990)). Substantially pure compositions of at least about 90 to 95% homogeneity are preferred, and 98 to 99% or more homogeneity are most preferred. Once purified, partially or to homogeneity as desired, the polypeptides may then be used (e.g., as protease modulators).

One of skill in the art would recognize that after chemical synthesis, biological expression, or purification, the ecotin or ecotin variant (s) may possess a conformation substantially different than the native conformations ofthe constituent polypeptides. In this case, it may be necessary to denature and reduce the polypeptide and then to cause the polypeptide to re-fold into the preferred conformation. Methods of reducing and denaturing proteins and inducing re-folding are well known to those of skill in the art (See, Debinski et al. (1993) J. Biol. Chem., 268: 14065-14070; Kreirman and Pastan (1993) Bioconjug. Chem., 4: 581-585; and Buchner, et al, (1992) Anal. Biochem., 205: 263- 270). Debinski et al, for example, describe the denaturation and reduction of inclusion body proteins in guanidine-DTE. The protein is then refolded in a redox buffer containing oxidized glutathione and L-arginine.

One of skill would recognize that modifications can be made to the ecotin or ecotin variants without diminishing their biological activity. Some modifications may be made to facilitate the cloning, expression, or purification ofthe protein. Such modifications are well known to those of skill in the art and include, for example, a methionine added at the amino terminus to provide an initiation site, or additional amino acids (e.g., poly-His) placed on either terminus to create conveniently located restriction sites or termination codons or purification sequences. Detailed protocols for ecotin or ecotin variant expression and purification are provided in Example 1.

In addition to the expression of polypeptides encoding naturally-occurring amino acids, recombinant expression systems can be used to express polypeptides encoding unnatural amino acids. Methods for the incorporation of non-natural amino acids (e.g. amino acid analogues) are well known to those of skill in the art (see, e.g., Koh (1997) Biochem., 36(38): 1 1314-11322, Liu et al. (1991) Proc. Natl. Acad. Sci. USA, 94(19): 10092-10097, Liu et al. (1997) Chem. and Biol, 4(9): 685-691, Cload et al. (1996) Chelm. and Biol, 3(12): 1033-1038, Moran et α/. (1995) Biopolymers, 37(3): 213-219, Cornish et al. (1994) Biochem., 33(40): 12022-12031, Cornish et al. (1994) Proc. Natl. Acad. Sci. USA, 91(8): 2910).

V. Identification of lead compounds.

Conventionally, new chemical entities with useful properties (e.g., modulation of various serine proteases) are generated by identifying a chemical compound (called a "lead compound") with some desirable property or activity, creating variants of the lead compound, and evaluating the property and activity of those variant compounds. Lead compounds thus provide a convenient starting point or baseline for developing second or third generation variants.

In one embodiment, this invention provides the identification of characteristic motifs that enhance the activity ofthe ecotin variant against a particular serine protease. Typically this is accomplished by screening an ecotin variant library for inhibitory or agonistic activity on a particular serine protease. Members ofthe library that display the desired activity can be isolated and their sequence determined. Identification of a consensus sequence provides a good lead compound for the desired activity on the subject protease. Having identified a consensus sequence in one region ofthe ecotin variant, other regions can then be varied to provide new variants of potentially greater activity and/or specificity.

VI. Pharmacological formulations and delivery.

The ecotin variants of this invention can be combined with a pharmaceutically acceptable carrier (excipient) to form a pharmacological composition. Pharmaceutically acceptable carriers can contain a physiologically acceptable compound that acts, for example, to stabilize the composition or to increase or decrease the absorption ofthe active agent. Physiologically acceptable compounds can include, for example, carbohydrates, such as glucose, sucrose, or dextrans, antioxidants, such as ascorbic acid or glutathione, chelating agents, low molecular weight proteins, compositions that reduce the clearance or hydrolysis of the protease modulator(s), or excipients or other stabilizers and/or buffers. Other physiologically acceptable compounds include wetting agents, emulsifying agents, dispersing agents or preservatives that are particularly useful for preventing the growth or action of microorganisms. Various preservatives are well known and include, for example, phenol and ascorbic acid. One skilled in the art would appreciate that the choice of a pharmaceutically acceptable carrier, including a physiologically acceptable compound depends, for example, on the rout of administration ofthe protease modulator and on the particular physio-chemical characteristics ofthe modulator.

The pharmaceutical compositions can be administered in a variety of unit dosage forms depending upon the method of administration. For example, unit dosage forms suitable for oral administration include powder, tablets, pills, capsules and lozenges. It is recognized that the protease modulators, when administered orally, must be protected from digestion. This is typically accomplished either by complexing the modulator with a composition to render it resistant to acidic and enzymatic hydrolysis or by packaging the modulator in an appropriately resistant carrier such as a liposome. Means of protecting compounds from digestion are well known in the art (see, e.g., U.S. Patent 5,391,377 describing lipid compositions for oral delivery of therapeutic agents).

Effective systems for the sustained delivery of therapeutic proteins preferably involve (i) processing and formulating the protein and delivery system so that the protein's fragile conformation and biological activity are maintained throughout processing and during prolonged release in the body. In addition such systems control the protein release so that therapeutic levels are maintained for the desired time.

Sustained protein delivery can be achieved with a variety of microsphere delivery systems. In one preferred embodiment, the ProLease biodegradable microsphere delivery system for proteins and peptides (Tracy (1998) Biotechnol Prog. 14: 108; Johnson et al. (1996), Nature Med. 2: 795; Herbert et al. (1998), Pharmaceut. Res. 15, 357) a dry powder composed of biodegradable polymeric microspheres containing a protein in a polymer matrix that can be administered by injection in an aqueous diluent through a narrow-gauge needle.

The ProLease microsphere fabrication process was specifically designed to achieve a high protein encapsulation efficiency while maintaining protein integrity

(Gombotz, et al. (1991) U.S. Patent 5,019,400). The process consists of (i) preparation of freeze-dried protein particles from bulk protein by spray freeze-drying the drug solution with stabilizing excipients, (ii) preparation of a drug-polymer suspension followed by sonication or homogenization to reduce the drug particle size, (iii) production of frozen drug-polymer microspheres by atomization into liquid nitrogen, (iv) extraction of the polymer solvent with ethanol, and (v) filtration and vacuum drying to produce the final dry-powder product. The resulting powder contains the solid form of the protein, which is homogeneously and rigidly dispersed within porous polymer particles. The polymer most commonly used in the process, poly(lactide-co-glycolide) (PLG), is both biocompatible and biodegradable.

Encapsulation can be achieved at low temperatures (e.g., -40°C). During encapsulation, the protein is maintained in the solid state in the absence of water, thus minimizing water-induced conformational mobility ofthe protein, preventing protein degradation reactions that include water as a reactant, and avoiding organic-aqueous interfaces where proteins may undergo denaturation. A preferred process uses solvents in which most proteins are insoluble, thus yielding high encapsulation efficiencies, (e.g., greater than 95%).

Maintaining stability ofthe protein following injection of a sustained release formulation poses can be a concern because proteins in microsphere formulations remain in a concentrated, hydrated state at physiological temperatures for prolonged periods after injection. These conditions are conducive to protein degradation reactions, including aggregation (covalent and noncovalent), deamidation, and oxidation. However, several stabilization strategies can be used to maintain protein integrity under these conditions (Putney and Burke (1998) Nature Biotechnol. 16: 1; Schwendeman et al. (1996) Stability of Proteins and Their Delivery from Biodegradable Microspheres in

Microspheres/Microparticles, R. S. Cohen and H. Bernstein, Eds.(Dekker, New York, pp.l- 49). The choice of one or more stabilizing agents is determined empirically. One effective approach is to form a complex with a divalent metal cation before encapsulation. Zinc has beea-employed in this manner to stabilize recombinant human growth hormone (rhGH) and recombinant α-interferon (α-LFN) in microspheres (Tracy (1998) Biotechnol. Prog. 14: 108, Johnson et al. (1997) Pharmacol. Res. 14: 730; U.S. Patent 5,711,968 (1998)). Also, protein stability in hydrated microspheres can be improved by using certain salts. For example, ammonium sulfate has been shown to stabilize erythropoietin during release (U.S. Patent 5,711,968 ). In addition to maintaining protein stability during processing and release, the microsphere formulation should preferably display the release kinetics required to achieve a sustained therapeutic effect. Following injection ofthe microspheres into the body, the encapsulated protein is released by a complex process involving hydration of the particles, dissolution of the drug, drug diffusion through water-filled pores within the particles, and polymer erosion (Langer and Folkman (1976) Nature 263: 797; Bawa et al, (1985) J. Controlled Release 1 : 259; Saltzman and Langer (1989) Biophys. J. 55: 163). Two primary considerations are minimizing how much protein is released immediately (called the burst) and achieving the desired duration and rate of protein release. The duration of release is governed by the type of PLG polymer used and the addition of release modifying excipients such as zinc carbonate (Saltzman and Langer (1989) Biophys. J. 55: 163).

Advantages inherent in sustained delivery of proteins are likely to include improved patient compliance (by reducing the need for self-injection), potentially lower costs (by reducing the frequency of visits to a caregiver's office), greater usage of a drug (through new indications and ease of use), and improved safety and efficacy (by reducing variability inherent in frequent injections). For certain proteins, it may also be possible to reduce the total dose per month, thereby reducing the cost to patients. Nevertheless, microsphere-based sustained delivery systems may be limited by the daily dose of protein needed for a therapeutic effect. Alternative approaches for sustained delivery of therapeutic proteins are also known. An implantable osmotic pump system delivers peptide drugs at a constant rate for up to 1 year (Wright, et al. (1997), Proc. Int. Symp. Controlled Release Bioact. Mater. 24: 59). Use of liposomes and other multivesicle systems to deliver proteins in a sustained manner and to reduce immunogenicity is also known (see, e.g., (Liu et al. (1995) J. Biol, Chem. 270: 24864-24870, Solodin et al. (1995) Biochem. 34: 13537-13544, Mayhew et al. (1979) Cancer Treat. Rep. 63: 1923-1928, etc.). Systems wherein therapeutic proteins are expressed and secreted from an implantable device containing recombinant cells are also known. Gene therapy provides an alternative approach for sustained delivery of proteins to both localized and widely dispersed sites by reprogramming cells ofthe recipient to express the protein.

Pulmonary delivery of proteins in the form of aerosols may provide a less invasive route of administration compared to injection (Wall (1995) Drug Delivery 2: 1). Injection frequency can also be decreased by increasing plasma half-life. For example, chemical modification with polyethylene glycol has been reported to extend the plasma half- life of therapeutic proteins such as α-IFN (Nieforth et al, (1996) Clin. Pharmacol. Ther. 59:

636) resulting in a reduced injection frequency.

Other potential applications may demand more sophisticated release profiles than those achievable with simple microsphere systems, wherein drug release is primarily dictated by polymer degradation and erosion. Systems have been described where controlled release occurs in response to external stimuli such as an electric or magnetic field or to changes in the microspheres' biological milieu (Langer (1990) Science 249: 1527). Microspheres might be engineered to provide pulsatile drug release in response to relevant biofeedback (Id.) or to normal cyclical rhythms of the body. Additionally, formulations that contain multiple drugs and whose release profiles are tailored to changing physiological needs as treatment progresses represent embodiment. Examples of these indications are the dynamic cascade associated with wound healing and the degeneration, apoptosis, and regeneration sequence that occurs following spinal cord injury. This is accomplished by blending microspheres with different proteins and release characteristics. Finally, microelectronic chips can be interfaced with the injected polymer mass to provide programmed control of protein release, thus offering far greater moment-to-moment flexibility and precision in the release characteristics.

VII. Kits for screening, treatmen or affinity chromatographv.

A) Screening kits. In one embodiment this invention provides kits for screening for ecotin variants having a particular activity against one or more proteases. The kits preferably include an ecotin variant library and/or a nucleic acid library encoding an ecotin variant library. The library preferably includes any of the ecotin variants described herein. The kits may optionally contain any ofthe buffers, reagents, and/or media that are useful for the practice ofthe methods of this invention.

In addition, the kits may include instructional materials containing directions (i.e., protocols) for the practice ofthe methods of this invention. While the instructional materials typically comprise written or printed materials they are not limited to such. Any medium capable of storing such instructions and communicating them to an end user is contemplated by this invention. Such media include, but are not limited to electronic storage media (e.g., magnetic discs, tapes, cartridges, chips), optical media (e.g., CD ROM), and the like. Such media may include addresses to internet sites that provide such instructional materials.

B Therapeutic kits. In another embodiment, this invention provides for therapeutic kits. The kits include, but are not limited to a native ecotin and/or an ecotin variant of this invention or a pharmaceutical composition thereof. The various protease modulators may be provided in separate containers for individual administration or for combination before administration. Alternatively the various compositions may be provided in a single container. The kits may also include various devices, buffers, assay reagents and the like for practice ofthe methods of this invention. In addition, the kits may contain instructional materials teaching the use of the ecotin or ecotin variant in the various methods of this invention (e.g., in the modulation of one or more serine proteases, in the prophylaxis and/or treatment of diseases, and the like).

O Affinity chromatographv kits.

In another embodiment, this invention provides for kits for performing affinity chromatography to isolate one or more polypeptides having a chymotrypsin fold (e.g. serine proteases). The kits include, but are not limited to a container containing an affinity matrix that is a solid support (surface) (e.g. a resin or glass particle, a membrane, a surface of a slide or other solid object, a gel, a porous or non-porous bead, a magnetic particle, a surface bearing one or more channels or microchannels or capillaries, etc. ) having attached thereto one or more ecotin variants of this invention. The affinity matrix can be further package (e.g. in a column or other flow-through device) to facilitate ease of use. In addition the kits may also include various devices, buffers, labels, assay reagents and the like for practice ofthe affinity chromatography or other "immunoassay" methods of this invention. In addition, the kits may include instructional materials containing directions (i.e., protocols) for the practice ofthe methods of this invention.

EXAMPLES

The following examples are offered to illustrate, but not to limit the claimed invention.

Example 1 : Ecotin, A Serine Protease Inhibitor with Two Distinct and Interacting Binding Sites

A Introduction

Serine protease inhibitors are one ofthe most diverse families of macromolecules that achieve similar biological functions with entirely different scaffolds. They have been studied extensively due to their ubiquitous presence in numerous biological functions. The rich mechanistic and structural database available on the interactions between serine protease inhibitors and their target enzymes make these inhibitors an excellent model system to investigate the fundamental biochemical and biophysical principles of protein-protein recognition. Serine protease inhibitors have been classified into at least twenty sub-families based on amino acid sequence and mechanism of interaction (Laskowski and Kato (1980) Annu. Rev. Biochem., 49(593): 593-626). Four common reaction mechanisms have been postulated to describe the different chemical and kinetic pathways ofthe inhibition.

The substrate-like standard mechanism accounts for the inhibition by a large number of canonical small macromolecular inhibitors with less than 200 amino acids. A classic example of this type of inhibition is that achieved by the binding of Bovine Pancreatic Trypsin Inhibitor (BPTI) to trypsin. In this case, the PI f residue ofthe inhibitor occupies the SI binding pocket ofthe protease and the amino acid residues flanking PI bind to the enzyme in a substrate-like conformation.

Serpins, a family of serine protease inhibitors consisting of plasma proteins of more than 400 amino acids, inhibit their target enzymes through a loop insertion followed by a conformafional change (Bode and Huber (1992) Eur. J. Biochem., 204(2): 433-451; Engh et al (1995) Trends Biotechnol, 13(12): 503-510).

In contrast, the other two serine protease inhibition mechanisms are more dependent on steric occlusion ofthe enzyme substrate binding pocket by the inhibitor. The non-specific "molecular trap" inhibition by α2-macroglobulin involves breaking a thiol ester bond and "engulfinent" ofthe target protease (Barrett et al. (1981) Meth. Enzymol, 80 Pt C, 737:754). Finally, the inhibition of thrombin by hirudin involves an extended binding site that bridges the active site nucleophile ofthe enzyme and blocks it from productive substrate binding.

A common structural theme among these reaction mechanisms is that the protease and the inhibitor form a one-to-one complex. The majority of these inhibitors bind to the target protease at a single site. This binding site is usually well-defined structurally and confined to a specific surface of the inhibitor. In the case of substrate-like serine protease inhibitors, an extended surface loop ofthe inhibitor binds to the active site ofthe protease with a substrate-like conformation. This so-called "reactive site loop" provides all the binding interactions between the inhibitor and protease. The rest ofthe inhibitor plays the structural role of maintaining the proper conformation ofthe reactive site loop. Studies using solution kinetics studies have shown that, in many cases, the amino acid side chain at the PI position of this reactive site loop is the primary determinant ofthe inhibitory specificity (Kossiakoff et al. (1993) Biochem. Soc. Trans., 21(pt 3) 614-618). Ecotin, a serine protease inhibitor found in the periplasm of Escherichia coli, is a noticeable exception to the examples outlined above. Ecotin is a competitive inhibitor that strongly inhibits trypsin, chymotrypsin and elastase and many other serine proteases with comparable potencies (Chung et al. (1983) 258(18): 11032-11038). The inhibitor was purified and its reactive site was determined to be Met 84 which lies within a disulfide bonded protein segment (McGrath et al. (1991) J. Biol. Chem., 266(10): 6620-6625). The gene encoding ecotin was cloned and expressed recombinantly in E. coli. (McGrath et al. (1991) J. Biol. Chem., 266(10): 6620-6625; McGrath et al. (1991) J. Mol. Biol, 222(2): 139- 142). The 142 amino acid monomeric ecotin, with a PI Met at residue 84, dimerizes in solution at submicromolar concentrations. The high-resolution crystal structure of ecotin complexed with rat anionic trypsin shows two ecotin monomers form a dimer through interactions at their C-termini (McGrath et al. (1994) EMBO J., 13(7): 1502-15O7) (Figure 1). The ecotin dimer binds to two trypsin molecules at opposite ends to form a heterotetramer with a two-fold symmetry axis. The crystal structure also reveals a network of interactions between ecotin and trypsin. The protein-protein interaction surface between the inhibitor and the protease consists of two distinct areas, each provided by one ofthe two ecotin molecules. The first area, known as the "primary binding site", involves the reactive site loop of ecotin, i.e. the 80s loop (residues 81-86), the 50s loop(residues 52-54), and the active site of trypsin. The second area, known as the "secondary binding site" 25 A away (± 3-5 A depending upon the ligand), includes two surface loops of ecotin, the 60s loop(residues 66-70) and 100s loop(residues 108-113), and the C-terminal region of the protease (including part ofthe C-terminal helix and part ofthe 90s loop of trypsin).

The heterotetramer formed by two ecotin monomers and two trypsins is unique among the known structures of serine protease and inhibitor complexes. From a mechanistic point of view, ecotin's mode of inhibition against its target protease shows novel characteristics in addition to a substrate-like mechanism. The interaction between ecotin's primary binding site loop and trypsin's active site follows the classic model. The conformation ofthe 80s loops is superimposable on that ofthe reactive site loop of BPTI and other canonical macromolecular inhibitors. The inhibitor displays competitive inhibition kinetics and biochemical characterization identified Met84 as the PI residue with direct evidence of bond cleavage by trypsin, chymotrypsin or elastase between Met84 and Met85. However, the distinct secondary binding site observed in ecotin has not been observed in protease-inhibitor complexes. The broad specificity and dual binding mode of ecotin led us to investigate the binding interactions between ecotin and target proteases. Although the three- dimensional structure ofthe ecotin-trypsin complex provides insight regarding the mechanism of inhibition and details ofthe interface between the protease and its inhibitor, this structure does not offer critical functional information relating to the activity of ecotin. Therefore, the contribution ofthe various interactions ofthe protease-inhibitor complex must be determined through other means. We have used a combination of site-directed and region- specific mutagenesis to address the relationship between the structure and function of this unique inhibitor and to ask how the inhibitor uses two distinct binding sites to recognize its target.

B) Materials and Methods

Materials

The E. coli strain JM101, XL-1 Blue F^* and the VCSM13 helper phage were from Stratagene (La Jolla, CA). The E. coli ecotin gene deletion strain IMΔecoJ was derived from JM 101. Enzymes and reagents for the manipulation of DNA were purchased from Promega(Madison, WI) or New England Biolabs (Beverly, MA) and were used following the manufacturer's instructions. Low molecular weight uPA (LMuPA) was obtained from American Diagnostica (Greenwich, CT). Bovine trypsin was from Sigma (St. Louis, MO). Rat trypsin was expressed in E. coli using the expression vector pZ3 and purified as described(Higaki et al. (1989)Biochem., 28(24): 9256-9263). Wild-type and variant forms of ecotin were produced in bacteria from the expression vector pTacTacEcotin following protocol described below. The chromogenic substrate Nα-benzyloxy-carbonyl-L- glyylproylarginine-7-amino-4-methylcoumarin (Z-GPR-AMC) was purchased from Bachem Biosciences, Inc (King of Prussia, PA). The chromogenic substrate Z-γ-Glu(α-t-butoxy)- Gly-Arg-p-nifroanilide (Spectrozyme UK) used for LMuPA kinetic analysis was from

American Diagnostica (Greenwich, CT). 4-Methylumbelliferyl j-guanidinobenzoate was from Sigma (St. Louis, MO). DNA oligonucleotides were synthesized on a Perkin Elmer/ Applied Biosystems 391 DNA synthesizer (Foster City, CA) by using the phosphoramidite method and reagents from the same company. Synthetic oligonucleotides were purified with NENSORB cartridge from DuPont NEN (Boston, MA). DNA sequencing was performed with a Sequenase 2.0 kit from U.S. Biochemical Corp. (Cleveland, OH). α-³⁵S-dATP was from DuPont NEN (Boston, MA). Geneclean® was from Bio 101, Inc (La Jolla, CA). Spectra/Por® molecularporo s dialysis membrane was from Spectrum (Laguna Hills, CA). Amicon Centriprep-10 and Centricon-10 were from Amicon (Beverly, MA). All other chemicals were of reagent grade or better and were used without fuπher purification.

Computer Modeling of Ecotin

Ecotin WTΔ (133-142): The ecotin-ecotin interface was determined by using the program Insightll (Dayringer et al. (1986) Mol. Graph., 5: 82-87) to search for all ecotin residues in one monomer that were within 4J A of the other monomer . In the 2.4 A X-ray structure of ecotin bound to D102N trypsin (McGrath et al. (1994) EMBO J., 13(7): 1502- 1507), the ecotin-ecotin dimer interface buries 3000 A² of surface area as calculated with the program Access using a 1.4 A probe size. In addition it makes 10 monomer-monomer hydrogen bonds as calculated using Insightll with a distance cut off of 3.0 A and an angle cut-off of 120°. Removal of residues 133-142 was calculated to remove 2000 A² buried surface area in the dimer interface leaving a possible dimer interface of only 1000 A² with only 4 ofthe 10 original monomer-monomer hydrogen bonds.

Ecotin 5OA3+8OA5 and Ecotin 6OA₄+IOOA4: Residues in the ecotin-trypsin interface were determined from the 2.4 A X-ray structure of ecotin-D102N trypsin (McGrath et al. (1994) EMBOJ., 13(7): 1502-1507) using Insightll by finding all residues from one molecule within 42 A ofthe other molecule and vice versa. Area buried was computed according to Lee and Richards as implemented by the program Access using a probe size of 1.4 A Lee and Richards (1971) J. Mol. Biol, 55(3): 379-400). Candidate residues for mutation to alanine were all ecotin residues whose solvent accessible surface decreased 50% or more upon binding to trypsin in either the primary binding site (ecotin 5OA3+8OA5) or secondary binding site (ecotin 6OA₄+IOOA₄) with the exception of Gly66 which had an unusual ph ipsi angle such that its changing to another residue would destroy the secondary structure. This resulted in 8 residues in the primary site which were changed to alanine. This group of residues is comprised of Leu52, His53, Arg54, Val81, Ser82, Thr83, Met84, Met85, and Ala86 (left unchanged). The resulting construct was designated ecotin 5OA3+8OA5. In the secondary site residues Trp67, Gly68, Tyr69, Asp70, Arg 108, Asnl 10, Lysl 12 and Leul 13, were all substituted with alanines. This variant was referred to as ecotin 60 A₄+100 A₄. Ecotin Mutagenesis

Site-directed mutagenesis was performed by the method of Kunkel (1985) Proc. Natl. Acad. Sci. USA, 82(2): 488-492. All mutations have been confirmed by DNA sequencing of the mutation region.

Plasmids for the production of M84X (where X=R. K. F. W)

The codon in plasmid pBS Ecotin that codes for Met84 was changed to code for Arg84, Lys84, Phe84 or Trp84 by site-directed mutagenesis using oligonucleotides M84R, M84K, M84F, or M84W respectively. The following is a list of these primers (underlined nucleotides were base changed, italics codons were residues changed).

1) Primer M84R : 5'-GT TCC CCG GTT ACT ACT AGG ATG GCC TGC C-3' (SEQ JD NO. 2) (with a unique Seal site);

2) Primer M84K : 5'-GT TCC CCG GTT ACT ACT AAG ATG GCC TGC C-3' (SEQ ID NOJ); 3) Primer M84F: 5'-GT TCC CCG GTT AGT ACT TTC ATG GCC TGC C-3' (SEQ ID NO. 4); and 4) Primer M84W : 5'-GT TCC CCG GTT AGT ACT TGG ATG GCC TGC C-3' (SEQ ID NO. 5).

Plasmids for the production of the C-terminal deletion variants WTΔ and M84RΔ pBS Ecotin and pBS Ecotin M84R were used as templates in a PCR reaction with oligonucleotides EcoN and EcoC to generate C-terminal deletion variants at residues 133-142. The amplified region was digested with BamHI and Hindlll, purified and ligated to the BamHI/Hindlll fragment of pTacTacEcotin . The PCR condition used were: 94°C for 1 min, 40°C for 2 min, 72°C for 3 min, 35 cycles. The sequences ofthe two primers were:

1) Primer EcoN: 5'-AAATTAACCCTCACTAAAGGG-3' (SEQ ID NO. 6); and

2) Primer EcoC: 5'-TTGTCAATTTAAGCTTACGCCTTCCAGACGCGG-3' (SEQ ID NO. 7).

Plasmids for the production of multiple alanine substitutions at the secondary binding site. The DNA in pBS Ecotin was changed to code for alanines at positions 52-54, 67-70, 81-85, 108,110, 112-113 by site-directed mutagenesis using oligonucleotides 5OA₃, 60 A₄, 80 A5, or IOOA₄ respectively. Each of these primers carries a unique restriction site, their sequences are as follows:

1) Primer 50A₃ : 5'-CTG GAA GTC GAT TGC AAT GCG GCTGCC TTG GGC GGG AAG CTG GAA AAC-3' (SEQ LD NO. 8) (unique Styl site);

2) Primer 60A4: 5'-aac aaa acg ctg gaa ggg gcc gcc gcg gcc tat tat gtc ttt gat aaa gtc-3' (SEQ ID NO. 9) (unique Sfil site); and 3) Primer 80A5: 5'- AAA GTC AGT TCC CCG GCT GCA GCG GCG GCG GCA TGC CCG GAT GGC AAG-3' (SEQ LD NO. 10) (unique Sphl site); 4) Primer 100A4: 5'-GC GAT GCT GGA ATG CTG GCT TAC GCT AGC GCG GCG CCG ATC GTG GTG TAT AC-3' (SEQ LD NO. 11) (unique Nhel site).

The region in the vector containing the desired mutation was digested with

BamHI and Hindlll and ligated to the BamHI/HindLII fragment of expression vector pTacTacEcotin.

Expression and Purification of Recombinant Ecotin

IMΔecoJ cells were transformed with pTacTac Ecotin and transformants were selected by plating on LB/ampicillin plates. A single colony was used to inoculate 3 ml of LB containing 60 μg ml ampicillin. The cultures were grown at 37°C for 9 hours and diluted to 1 liter of LB/ampicillin. Following growth at 37°C for 1 hour, PTG was added to the cultures to a final concentration of 0.2 mM, and continued to grow for 12 hours at 37°C. Cells were harvested and treated with lysozyme in a solution containing 25% sucrose/10 mM Tris, pH 8.0. The periplasmic fraction was dialyzed against 10 mM Tris, pH 8.0 with a Spectra/Por molecularporous dialysis membrane of 12-14 K molecular weight cut-off. Following dialysis, the supernatant was acidified to pH 2.8 by addition of 1M HC1 to the sample. The sample was incubated on ice for 30 minutes and the acid-insoluble material was removed by centrifugation. The supernatant was neutralized to pH 7.4 with 1 M Tris, pH 8.0, and adjusted to a NaCl concentration of 0J M. The solution was heated at 100°C for 10 minutes, and then cooled to room temperature. The precipitate was removed by centrifugation, and the supernatant was dialyzed against deionized distilled water overnight at 4°C. The ecotin sample was concentrated with a 10 K molecular weight cut-off concentrator Amicon Centriprep-10). The concentrated sample was loaded onto a Nydac C4 reverse-phase high performance liquid chromatography column (2.2 x 25 cm) that had been equilibrated with 0.1% trifluoroacetic acid. The column was washed and ecotin was eluted with a linear gradient of 34-31% acetonitrile/0.1% trifluoroacetic acid at a flow rate of 10 ml/minute over 30 minutes. Fractions were analyzed individually by SDS-PAGE and those containing pure ecotin fractions were pooled, and lyophilized. Purified ecotin was redissolved in water and stored at 4°C. The concentrations of ecotin and ecotin variants were determined using a calculated molar extinction coefficient of 14890 cm^-1 M"¹ for variants 60 A₄, M84R+6OA4, 6OA₄+IOOA₄, WTΔ, M84RΔ, 21860 cm"¹ M"¹ for wild-type (WT) and variants IOOA4, M84R+IOOA₄, M84R, M84K, and M84F, and 27550 cm"¹ M"¹ for M84W.

Determination of Ecotin Ki's against Various Proteases

Rat and bovine trypsin activity assays were using the same substrate and procedure. Trypsin was titrated with 4-methylumbelliferyl _/r»-guanidinobenzoate to obtain an accurate concentration ofthe enzyme's active sites. Various concentrations of ecotin or ecotin variants were incubated with trypsin in a total volume of 990 μl of buffer containing 50 mM ΝaCl/50 mM Tris/10 mM CaCtø, pH 8.0. Following a 10-minute equilibration at room temperature, 10 μl of 2.5 mM Z-GPR-AMC substrate was added and the rate of substrate hydrolysis was measured by monitoring the change of emission at 460 nm in a 2- minute period at 25°C. LMuPA assays were performed as described (Wang et al. (1995) J. Biol. Chem., 270(20): 12250-12256). The assays were repeated five to seven times under different inhibitor concentrations. All the data were fit to the equation derived for kinetics of reversible tight-binding inhibitors (Morrison et al. (1969) Biochim. Biophys. Ada, 185(2): 269-286) by nonlinear regression analysis using the program Kaleidagraph. The values for apparent K_- and true K_. , as well as the standard deviations ofthe were determined. The equation is:

where V- V₀ is the ratio ofthe inhibited rate vs. the uninhibited rate, [E₀] is the total enzyme concentration, [I₀] is the total inhibitor concentration. Usually, the [E₀] ranges from 50 pM to 500 pM for rat and bovine trypsin, 1 to 2 nM for LMuPA.

The P and S nomenclature of Schechter and Berger (1967) Biochem. Biophys. Res. Commun., 27: 159, is used herein. PI is the substrate (or inhibitor) residue before the scissile bond, where PI -PI' is the scissile bond, SI, S2, etc. are the coπesponding binding subsites on the enzyme.

Q Results

Ecotin Variants Preparation and Characterization In this study, ecotin variants with amino acid substitutions and deletions at the primary or secondary binding site were generated by site-directed mutagenesis (Kunkel, et al. (1985) Proc. Natl. Acad. Sci. USA, 82(2): 488-492), expressed to high-levels in E. coli and subsequently purified to homogeneity. The final yield ranged from 20 to 100 mg protein per liter of liquid culture. The same expression system and purification protocol using heat treatment and reverse phase HPLC steps were applied to both primary and secondary site ecotin variants. Activity assays were performed at each purification step to monitor for activity loss ofthe samples. All variants were expressed in E. coli at comparable levels and were stable under the purification conditions. Table 1 is a summary ofthe nomenclature of ecotin variants that appear in this study. The SDS-PAGE analysis of purified aliquots of ecotin variants are shown in Figures 2(a)and 2(b). Ecotin WT and six variants at the PI position and dimer interface (WT, M84R, WTΔ, M84RΔ, M84K, M84F, M84W) are shown in Figure 2(a), while Figure 2(b) shows ecotin WT and seven variants at the secondary site loops (WT, 6OA₄, IOOA₄, 6OA4+IOOA₄, 5OA3+8OA5, M84R+60A₄, M84R+IOOA4, M84R+6OA₄+IOOA₄). Each variant exhibited single-band homogeneity on SDS- polyacrylamide gels and migrated at the expected molecular weight. All variants cross- reacted with anti-ecotin polyclonal antibodies on immunoblots. Selected variants were analyzed with electrospray mass spectroscopy to confirm their compositions. Ecotin variants 50A3+80A5_! 6OA₄+IOOA₄, and M84R+6OA₄ had molecular weights of 15713.40±2.14 (calculated 15714.15), 15636.67±3.22 (calculated 15637.19), 15886.95±1.79 (calculated 15887.27) respectively. Table 1. Definition of ecotin variants

Variant Symbol Mutations

M84X (X = R, K, F, W) Single residue substitution at position 84

WTΔ Deletion of residues 133-142

M84RΔ Deletion of residues 133-142 combined with Arg substitution at position 84 50A₃ Multiple alanine substitutions at residues 52-54

6OA Multiple alanine substitutions at residues 67-70

80A₅ Multiple alanine substitutions at residues 81-85

IOOA Multiple alanine substitutions at residues 108, 110, 112, 113

The Role of the Primary Binding Site

Since the PI amino acid side chain is often a major determinant ofthe inhibitory specificity of many substrate-like serine protease inhibitors, the role of Met 84 of ecotin was explored by substituting this position with Arg, Lys, Phe and Trp. Met is not a common residue at PI for serine protease inhibitors that target enzymes with Arg or Lys specificity. For classic canonical inhibitors of trypsin, Arg and Lys are preferred at the PI site, while both Phe and Trp are not well accommodated in the Si pocket of trypsin. The inhibition constants (K_t ) of ecotin PI substitution variants against rat trypsin were measured and are listed in Table 2. Significant changes ofthe electrostatic and hydrophobic properties ofthe PI residue of ecotin had relatively little effect on inhibition of rat trypsin by these variants. Although the favorable substitutions M84R and M84K had lower K_t values compared to that of WT ecotin, the magnitude ofthe change was only two-fold. Similarly, ecotin variants with unfavorable mutations M84F and M84W were still potent inhibitors of rat trypsin with the greatest effect being five-fold weaker association between rat trypsin and ecotin M84F compared to WT ecotin. Table 2. Kj of ecotin PI variants at position 84 against rat trypsin (unit: nM )

WT M48F M84W M84K M84R

0.93 ± 0.16 4.47 ± 0.72 2.20 ± 0.28 0.55 ± 0.16 0.38 ± 0.10

The Role of the C-terminal Dimer Interface

The C-terminal dimer interface between the two ecotin monomers is a key structural element ofthe tetrameric complex as it forms one of the three types of protein- protein interfaces in the tetramer. The other two contact regions are the primary and secondary binding sites that interact with the protease. The dimerization of ecotin not only dramatically increases the contact region between the inhibitor and the protease, but also adapts to maintain the proper orientation of the two ecotins as a molecular "hinge" to permit binding to different proteases which have a chymotrypsin fold structure. The combined surface areas ofthe primary and secondary sites of ecotin that become buried upon binding to trypsin exceed 2800 A². This area is substantially greater than most ofthe other protease- inhibitor interfaces (e.g. BPTI-trypsin interface is only 1390 A²). If the hinge region is disrupted or destroyed, the relative positions ofthe primary and secondary binding sites may change since each binding site results from different surfaces ofthe two contralateral ecotin monomers (Figure 1). We investigated the role ofthe dimer interface on ecotin's unique broad specificities by truncating the C-termini ofthe ecotin monomer.

Based on the crystal structure ofthe ecotin-trypsin complex, residues 133- 142, which consist of about one half of the contact amino acids ofthe C-terminal arm, were deleted while still maintaining a stub of two amino acids. The ecotin variants with residues 133-142 truncated was denoted ecotin WTΔ. The other variant with residues 133-142 truncated and the PI Met at position 84 replaced with an Arg was denoted ecotin M84RΔ. These two variants were designed to probe the importance ofthe dimer interface under the context of "neutral" or favorable PI interactions. The monomer-dimer dissociation constant, K_d , was measured by fluorescence following the protocol of Seymour et al (1994)

Biochem., 33(13): 3949-3958. The KJs of ecotin WT and several truncation variants are shown in Table 3. The K_d of ecotin WT was close to the value of 390±150 nM measured previously (Seymour et al (1994) supra.). However, ecotin WTΔ and M84RΔ had K values of only 2.5 and 3J-fold higher respectively, suggesting that the truncation variants of ecotin still formed dimers, even though their dimer interactions were weaker, as indicated by the modest increase ofthe K_d's. Interestingly, replacement of Met at position 84 with an Arg also increased the K_d slightly relative to WT ecotin (1.7-fold). The subtle change at one surface loop is conveyed to the C-terminal of the same molecule, implying ecotin has extensive structural flexibility.

Table 3. K of ecotin C-terminal truncation variants (unit: nM)

WT 220 ± 130 WTΔ 540 ± 210 M84R 380 ± 170 M84RΔ 820 ± 310 To probe the inhibitory activities of the ecotin variants, three homologous serine proteases, bovine cationic trypsin, rat anionic trypsin, and the catalytic domain of human urokinase-type plasminogen activator (uPA) were chosen as model targets. These enzymes share the same three-dimensional fold and have very similar substrate binding pockets. However, trypsin and uPA have different in vivo physiological functions and catalytic efficiencies. Together, they form a sample set that represents the large family of serine proteases recognizing Arg/Lys at the PI position. The K_t values ofthe ecotin C- terminal truncation variants against the three serine proteases are shown in Table 4. These data are also plotted in Figure 3 to illustrate the relative contributions from the PI residue and the dimer interface. The juxtaposition ofthe K_t data of the same group of ecotin variants against the three serine proteases provides a clear and direct graphic comparison ofthe inhibitory activity ofthe ecotin variants under different circumstances. We used these plots to visualize the distinct patterns of molecular recognition between ecotin variants and their target proteases.

Table 4. RT,- of ecotin C-terminal truncation variants (unit: nM)

Variant Bovine trypsin Rat trypsin uPA

WT 0.31 ± 0.06 0.93 ± 0.16 2800 ± 160

WTΔ 0.72 ± 0.05 181.3 ± 11.9 434,000 ± 72,000 M84R 0.09 ± 0.02 0.38 ± 0.10 3.6 ± 0.6

M84RΔ 0.04 ± 0.01 0.09 ± 0.02 2700 ± 130

One immediate and striking implication from the K measurements was that deletion of ten amino acids at the C-terminus of ecotin has minimal impact on its structural and functional integrity. This observation is seen in the case of bovine trypsin where the C- terminal truncations in either WT or M84R variant only had small effects on Ki , suggesting that the deletion did not destablize the ecotin molecules. In contrast, the inhibition of rat trypsin by WTΔ was almost two hundred-fold weaker, indicating that the truncation could have dramatic consequences. This surprising result also demonstrated that a change distal from the primary binding site of ecotin could transmit through the molecule and affect inhibition. However, once the M84R substitution was introduced and combined with the C- terminal deletion, ecotin M84RΔ not only regained nanomolar potency against rat trypsin, but also became an even better inhibitor of this enzyme (90 pM vs. 930 pM for WT). This result suggests that the favorable electrostatic interaction at the PI position could compensate for the weakened interactions due to the C-terminal truncation. Finally, the Ki data of uPA inhibition revealed yet another pattern. In this case, the M84R substitution only partially compensated for the unfavorable interactions created by the C-terminal deletion. However, the Ki of the double variant, ecotin M84RΔ against uPA remained at micromolar levels (2.1 μM). In contrast to that of rat trypsin, the favorable electrostatic interaction at the PI position alone was not sufficient for tight binding with uPA. Overall, these results show that the inhibition of rat trypsin and uPA by ecotin are very sensitive to the perturbation of the dimer interface. However, the inhibition of bovine trypsin, an enzyme that has 74% sequence identity with rat trypsin, was insensitive to mutations at the PI and C-terminal region.

Region-Specific Mutagenesis at the Primary and Secondary Binding Sites

Although investigations ofthe PI site and dimer interface had revealed the importance ofthe secondary site, more direct evidence is necessary to establish its role in protease binding. A structure-based approach was adopted to dissect and isolate the relative contributions from ecotin's two binding sites. By examining the ecotin-trypsin interface in Table 5. Identification of ecotin secondary site residues.

Hydrogen bond

Residue A • ^a "complex - -tmried Percentage distance⁶

(A) (A) (A) buried^d (A)

Leu52 146 50 96 66

His53 38 22 16 42 3.1

Arg54 133 8 125 94

Val81 83 6 77 92

Ser82 57 17 40 70 2.9

Thr83 62 0.3 61.5 99.5 3.0

Met84 209 0.5 208.5 99.8 3.0

Met85 133 9 124 93

Ala86 100 8 92 92 3.0

Trp67 117 13 104 89 2.8

Gly68 60 3 57 95 3.1

Tyr69 51 14 37 72

Asp70 71 43 28 40 2.9

Argl08 180 87 93 52 2.7

Asnl lO 24 13 11 47 2.9

Lysl l2 161 129 32 20

Leul l3 44 17 28 62

Auncomplex Solvent accessible surface area in uncomplexed ecotin dimer

Acompiex ^: Solvent accessible surface area in ecotin-trypsin tetrameric complex

Aburied^: Calculated as the difference between A_uncompi_eχ and A_compi_ex

Percentage Buried : Calculated as A uried I A_uncomplex * 100° 7o Hydrogen Bond Distance : Putative hydrogen bind distance with closest side chain or main chain atoms in trypsin. the crystal structure of the tetrameric complex, a series of candidate residues was generated and validated through interactive computer modeling (Table 5). We focused on residues that are within 4J A ofthe protease, have large buried surface areas (over 50 A²) or large percentage change of buried surface area upon binding (over 50% buried), and residues with right geometry and location for intermolecular hydrogen bonding. These residues could be divided into two distinct groups. The primary site group includes residues 81-86 and 52-54 located on two surface loops (50s and 80s loops). The secondary site group consists of residues 67-70, 108, 110, 112, and 113, also from two surface loops (60s and 100s loops).

To investigate the roles of these residues in binding, we used a method, by which the amino acids within a specific region ofthe molecule of interest are substituted by alanines simultaneously. The multiple alanine substitutions may reduce the hydrophobic and electrostatic interactions provided by the side chains to a minimum if the conformation of the polypeptide backbone remains unaltered. We made two multiple alanine substitution variants of ecotin, designated as variant 5OA3+8OA5 and 6OA₄+IOOA₄. Ecotin 5OA3+8OA₅ has eight alanine substitutions at residues 81-85 (residue 86 is already an alanine) and 52-54. Variant 6OA4+IOOA₄ also has eight alanine substitutions at residues 67-70, 108, 110, 112 and 113. These two multiple alanine substitution variants were investigated to define the roles of the primary and secondary binding sites of ecotin. Their inhibition constants against bovine and rat trypsins are shown in Table 6.

Table 6. K_t of ecotin multiple alanine substitution variants (unit: nM)

Variant Bovine trypsin Rat trypsin

WT 0.31 ± 0.06 0.93 ± 0.16

50A₃ + 80A₅ 98.3 ± 14.9 27,900 ± 5600 όOA^ ± 100A4 1.59 ± 0.26 15,300 ± 3100

The K_. of ecotin 5OA3+8OA₅ against rat trypsin increased 30, 000-fold compared to that of WT ecotin, from 0.93 nM to 27.9 μM; its K_t against bovine trypsin also increased significantly (about 300-fold). These results confirmed the essential role of the reactive site loop as a major determinant of the strength of protease inhibition. At the secondary binding site, ecotin 60A₄+100A₄'s Ki for bovine trypsin increased slightly (about

5-fold) compared to that of the WT ecotin. The remarkable potency of this variant demonstrated that ecotin remains competent as an inhibitor even after a significant part of the molecule was replaced with alanines. Nevertheless, the inhibition towards rat trypsin by variant 6OA₄+IOOA₄ decreased over 16,000- fold (153 μM), indicating that this loop contributes greatly to binding. WT ecotin inhibits both rat and bovine trypsin equally well, but ecotin 6OA₄+IOOA4 has acquired near 10,000-fold greater potency for bovine trypsin (1.59 nM) over rat trypsin (153 μM).

The dramatic contrast ofthe inhibition of ecotin 6OA₄+IOOA₄ for bovine and rat trypsin highlighted the importance ofthe 60s and 100s loops for different proteases. To ensure that the loss of inhibition was not due to its intrinsic destablization, ecotin 6OA4+IOOA₄ was incubated with rat and bovine trypsin at pH 5J and 8.0, 37°C for up to 16 hours at 1 to 600 enzyme : inhibitor concentration. The reaction mixture was analyzed by SDS-PAGE (data not shown). The inhibitor was stable to proteolysis for the length of time for the Ki measurement (20 minutes). However, partial proteolysis of ecotin is ^"eventually seen by rat trypsin. Similarly, ecotin 5OA3+8OA5 (27.9 μM vs. rat trypsin and 98.3 nM for bovine trypsin) exhibits partial proteolysis by rat and bovine trypsin. Not surprisingly, ecotin WT is table to proteolysis by both enzymes at both conditions. Therefore the results ofthe over digestion assays are consistent with the results ofthe Ki measurement.

The Role of the 60s Loop and 100s Loop

As shown in the case of rat β, the secondary binding site in ecotin plays a major role in the inhibition of particular serine proteases. The substitution ofthe eight residues in the 60s and 100s loops was sufficient to weaken the binding interactions between ecotin and one target protease by at least four orders of magnitude. Since these residues are located at two discontinuous regions of the surface loops, their contributions can be further dissected and analyzed. Ecotin 6OA4, ecotin M84R+6OA₄, ecotin IOOA₄, ecotin M84R+100A₄, and ecotin M84R+6OA₄+IOOA₄ were designed to address this question. Their Ki values against the three serine proteases are listed in Table 7 and are also plotted in Figures 4(a)and 4(b) and 4(c) for ecotin variants with alanine mutations at the 60s loop, the 100s loop, and both loops respectively. Together with ecotin M84R and ecotin 6OA₄+IOOA₄ generated from the previous experiments, they formed a complete set of variants that allowed the separation and definition of the individual role ofthe 60s and 100s loops. Table 7. K, of ecotin combinatorial variants (unit : nM).

Variant Bovine trypsin Rat trypsin uPA

^~WT 0.31 ± 0.06 0.93 ± 0.16 2,800 ± 160

M84R 0.09 ± 0.02 0.38 ± 0.10 3.6 ± 0.6

60A* 0.26 ± 0.05 44.2 ± 250 579,400 94,500±

M84R + 60A4 0.14 ± 0.02 0.18 ± 0.03 1470 ± 150

100A4 0.04 ± 0.01 0.75 ± 0.16 77,800 ± 12,700

M84R + 100A₄ 0.03 ± 0.01 0.04 ± 0.01 4040 ± 350

60^ + 100^ 1.59 ± 0.06 15,300 ± 3100 99,100 ± 16,500

M84R + 60A₄ + 0J6 ± 0.02 0.50 ± 0.07 28,200 ± 1300

100A₄

60 A4: multiple alanine substitutions at positions 67-70. IOOA₄: multiple alanine substitutions at positions 108, 110, 112, 113. 6OA4+IOOA_{4 :} multiple alanine substitutions at positions 67-70, 108, 110, 112, .113.

Figures 4(a)and 4(b) and 4(c) provide an overview ofthe interactions among ecotin's two surface loops and various proteases. The K, plots were used both horizontally and vertically to evaluate either the impact of specific surface loops on different enzymes, or the inhibition of a particular enzyme by different variants. Similar to the results from the previous experiment, the inhibition of bovine trypsin by ecotin was not affected by drastic changes to the inhibitor. The K,'s of all variants were approximately 1 nM. Ecotin IOOA₄ bound even slightly tighter than ecotin M84R (40 pM vs. 90 pM)_. Thus, the only dominant factor for bovine trypsin binding was the proper conformation ofthe primary site reactive loop. In the case of inhibition of uPA, it was clear that both the primary site (PI) and the two secondary site loops were critical for binding. The difference between the K,'s ofthe alanine substitution variants and the Ki of ecotin WT against uPA were significant, ranging from 75 μM to 576 μM. The three double variants which simultaneously carry a favorable M84R mutation and unfavorable multiple alanine substitution variants failed to achieve the low nanomolar binding of ecotin M84R. Their Ki values range from 1.5 to 28.2 μM, strongly suggesting that the proper amino acid sequences at the 60s and 100s loops were essential for optimal interaction between uPA and ecotin. The K, plot of rat trypsin indicated yet another pattern of secondary site inhibition. In this case, the 60s loop was responsible for providing most of the binding energy as indicated by the over 4,000-fold increase ofK, with ecotin 6OA₄, while the effect of the 100s loop was minimal. In fact, for all three enzymes, the 60s loop played a more predominant role in determining the strength of the inhibition at the secondary site. It was remarkable that the inhibition of each enzyme was achieved through the differential contributions from these two surface loops. It was also intriguing that the contributions from these loops were dependent on the identity of the PI residue and of the target enzyme. For example, if the target enzyme was rat trypsin, the favorable substitution M84R alone was sufficient to compensate for the adverse effects ofthe alanine substitution at the secondary binding site. Thus, the 60s loop was only important when PI 84 was a Met. However, the inhibition of uPA required not only a favorable PI residue (Arg) at 84 position, but also key residues at both 60s and 100s loops.

Nonadditivity among Ecotin Variants The Kj data set of the three groups of ecotin variants allow us to analyze the interrelationships among the key determinants of ecotin's potency and specificity. In particular, we determined whether the separate mutations are additive by comparing the Kj values of ecotin 6OA₄+IOOA₄ versus those of ecotin 6OA₄ or ecotin IOOA₄ against one particular protease. In the case of rat trypsin inhibition, the ecotin variant with alanine- substitutions at two loops bound more weakly than either ofthe two single alanine- substituted loop variants. For uPA binding, the 60s loop had a more significant effect. The Ki of ecotin 6OA₄ against uPA was even higher than the Ki ofthe double mutant ecotin 6OA₄+IOOA₄. It is clear that the contributions of these two loops were not independent of each other. The K_t plot was effective in directly comparing the strength of inhibition by ecotin variants and illustrating the relationships between the PI mutation and the secondary site multiple alanine substitutions (or C-terminal truncation). Alternatively, we can define additivity as the absence of interaction energy term ΔGi' in:

ΔΔG(_A,B)' = ΔΔG°_(A) + ΔΔG°(_B) + ΔGι° (1), ΔΔG°(_A) = ΔG°(_A) - ΔG°_(WT) = - RTln^,^*(_A) + -RTlnK,(wτ) (²)>

ΔG°_(X) = -KTlnKi (3)

where ΔΔG°(A,B) is the ΔG difference between double mutant A, B and WT, and ΔG°(A) +

ΔΔG°₍B₎ are the differences between the single site variant A or B and the wild-type, respectively (Wells et al (1990)B iochem., 29(37): 8509-8517)). By converting K_t to free energy using equation 3, the ΔG°ι term for mutations at different parts of ecotin can be calculated. Fifteen double mutation cycles were constructed from our original Kj data. These cycles were based on three different target proteases and between ecotin M84R and ecotin WTΔ, between ecotin M84R and ecotin 6OA₄, between ecotin M84R and ecotin IOOA₄, and between ecotin 6OA₄ and ecotin IOOA₄. The calculated ΔG°ι values are listed in Table 8. Mutational additivity has been seen in other model systems such as growth hormone and its receptor (Wells et al. (1996) Proc. Natl. Acad. Sci. USA, 93(10: 1-6), as well as subtilisin (Wells et al. (1987) Proc. Natl. Acad. Sci. USA, 84(15): 5167-5171) and the third domain of turkey ovomucoid (Qasim et al. (1997) Biochem., 36(7): 1598-1607; Lu et al. (1997) J. Mol. Biol, 266: 441-461). In those cases, the interaction energy ΔG°ι was shown to be small, usually less than 0.4 kcal/mol. In our system, however, the ΔG°ι was much larger. Half of the ecotin interaction energy values were greater than 1 kcal mol. The nonadditivity was most pronounced between ecotin M84R and ecotin 6OA4 mutations when the target enzyme was rat trypsin. The ΔG°ι for this particular mutation cycle was 5.46 kcal/mol. The large interaction energy highlighted the dynamic linkage among the key structural and functional elements of ecotin. These data provide evidence for cooperative and synergistic interactions among the PI residue of ecotin, the two surface loops and the dimer interface. It also suggests that the unique flexibility and adaptability of ecotin are the structural basis for the observed broad specificity.

Table 8. Relative change in interaction energy among ecotin's primary, secondary sites and dimer interface (unit: Kcal /mol)

Mutations Bovine trypsin Rat trypsin uPA

A M84R 0.98 3.98 -0.93

B Δ

A M84R -0.37 5.46 -0.4-

B 60A₄

A M84R -0.56 1.21 -2.19

B 100A₄

A M84R 0.63 5.59 -3.20

B 6OA₁ + 100A₄

A 60A₄ -2J9 -0.86 3.02

B 100A₄ D Discussion

Ecotin's Secondary Site as a Key Element in Macromolecular Recognition

We have used several approaches to establish the role ofthe secondary site of ecotin in protease recognition and inhibition. The experimental results can be interpreted based on the unique network of interactions between ecotin and its target protease. The impact of PI substitutions in ecotin was very different from that ofthe same substitutions in canonical serine protease inhibitors such as BPTI or turkey ovomucoid third domain (Lu et al (1997) J. Mol. Biol, 266: 441-461). For these familiar serine protease inhibitors, there is usually a 100 to 1000-fold difference in Ki's against trypsin between variants carrying a positively-charged and an uncharged apolar residue at the PI position. However, the Kj of ecotin against rat trypsin was insensitive to alterations at this position. Other research groups have also obtained similar results using ecotin against other serine proteases (Seymour et al. (1994) Biochem., 33(13): 3949-3958; Pal et al. (1996) FEBS Letts., 385(3): 165-170; Seong t α/. (1994) 7. Biol. Chem., 269(34) 21915-21918). These results demonstrate that the side chain ofthe PI residue does not play a predominant and independent role in determining the inhibitory specificity of ecotin. The binding energy of ecotin must come from other parts ofthe molecular interface. Structural analysis ofthe trypsin-ecotin complex suggests that a logical candidate is the secondary binding site which is composed ofthe two surface loops covering residues 66-70 and 108-113. These loops derive from the contralateral ecotin monomers in the tetramer with respect to the primary loops (Figure 1).

The C-terminal truncation mutagenesis experiment which affects the hinge between the ecotin monomer reinforced the importance ofthe secondary site and its alignment. The differential inhibition ofthe three proteases by ecotin WTΔ was a novel characteristic of this inhibitor. A plausible explanation could be made based on two assumptions. First, the secondary binding site might play different roles in ecotin binding to different enzymes. Second, the perturbation ofthe dimer interface changed the positions and/or conformations ofthe secondary binding site relative to other parts ofthe molecule. Therefore, if rat trypsin (or uPA) required the secondary site for tight binding, and if the C- terminal truncation dislocated the secondary binding site, the interactions between ecotin WTΔ and rat trypsin (or uPA ) would be severely weakened. In the network of interactions between the protease and ecotin, ecotin's C-terminal hinge region may affect the binding by modulating the relative contributions from the primary and secondary binding sites. The dramatic increase of K_t values of ecotin WTΔ against rat trypsin and uPA offered indirect evidence of the critical roles of the secondary site surface loops.

Finally, the multiple alanine substitution experiments strongly support the notion that the 60s and 100s loops could provide the critical binding energy necessary for protease inhibition. The loss of function of the variants not only established the essential role ofthe secondary site in rat trypsin and uPA binding, but also revealed an interdependent relationship between the 60s and 100s loops, suggesting the dynamic nature ofthe secondary site interaction. The Ki plot ofthe 60s loop variants was almost superimposable with that of the truncation variants. These two different large scale modifications of ecotin have actually generated strikingly similar consequences in terms ofthe inhibitory activity. It implies a fundamental connection between the dimer interface and the secondary binding site, two regions structurally far from each other. These two experiments further support the hypothesis that the dimer interface affected binding through dislocating the secondary site loops at the protease inhibitor interface.

Monomer Dimer Equilibrium

The interaction ofthe protease with the secondary site of ecotin requires the formation of a stable tetrameric complex. Although the K of ecotin's monomer-dimer equilibrium is much higher than K_t values against most target proteases, the K_d of ecotin dimerization could be dramatically perturbed in the presence of proteases. Proteases may serve as templates to facilitate the dimerization of ecotin. If there are strong cooperative interactions upon the binding ofthe second inhibitor and the second enzyme molecule, the reaction pathway may proceed from I -> El -> EI2 -> E₂I2, where the last two steps are rapid and drive the tetrameric complex formation to completion. Analytical centrifugation has been used to pursue this question. Preliminary data show that trypsin monomer, ecotin dimer, and ecotin-trypsin heterotetramer were the only three coexisting homogenous species in solution when a mixture of trypsin and ecotin was subjected to velocity sedimentation. Recently, a C-terminal truncation ecotin variant was shown to be monomeric by gel filtration (Pal et α/. (1996) FEBS Letts., 385(3): 165-170). This discrepancy with our results indicates that the solution state of ecotin truncation variants in the absence of protease may vary with buffer conditions and other variables. However, that study also showed that upon the addition ofthe protease, the tetrameric complex was formed. All the data in our study support the conclusion that the ecotin protease complex is tetrameric under a wide range of concentrations and that the secondary binding site has crucial roles in the formation and stabilization ofthe complex. Clearly the tetrameric nature of the ecotin-protease complex makes the mechanism of ecotin inhibition complicated due to the multiple equilibrium of the ecotin dimers and the two bound proteases monomers. In general all of these equilibrium might be expected to be affected by mutations ofthe inhibitor-protease interface. In addition, there is also the possibility of a folding related destablization for some ofthe variants which would affect their potency of inhibition. Further studies are under way to address these issues.

Protease Inhibitors with Multiple Binding Sites Ecotin offers unique opportunities to study the complex network of interactions between serine proteases and bi-dentate macromolecular inhibitors. It is also an ideal scaffold to design and engineer protease inhibitors. Potency and specificity toward target serine proteases may be introduced through the secondary binding site, a special structural feature of ecotin that does not resemble any previously known binding motifs within the families of macromolecular serine protease inhibitors. Although Streptomyces subtilisin inhibitor (SSI), which uses a substrate-like competitive inhibition mechanism, forms a tetramer with subtilisin, each subtilisin is bound to one SSI in a "chain-like" configuration (Takeuchi et al. (1991) J. Mol. Biol, 221(1): 309-325) which is as strong as the weakest link. In contrast, as shown in this study, in the case of ecotin, each trypsin makes direct contacts with two ecotin monomers in a "matrix" configuration that is stronger than the strongest link.

There are other serine protease inhibitors with multiple binding sites for one protease molecule. Hirudin, rhodniin and omithodorin are three examples of bi-dentate inhibitors against thrombin (Rydel et al. (1990) Science, 249: 277-280; van de Locht et al. (1995) EMBO J., 14(21): 4149-4157, van de Locht et al. (1996) EMBO J. 15(22): 6011- 6017). They bind to thrombin at both the active site and a highly charged exosite in a one- to-one complex. Unlike the predominant electrostatic interaction between these three inhibitors and the thrombin exosite, the recognition between the secondary binding site of ecotin and its target protease is driven by a combination of hydrophobic, electrostatic, and hydrogen bond forces. Ecotin is also different from hirudin, rhodniin and omithodorin because ecotin's two binding sites are provided by two monomers through dimerization. This unique bi-dentate binding mode has two advantages for macromolecular recognition for proteases with a chymotrypsin fold structure. First, it creates a buried surface area that is nearly 50 percent larger than the hirudin/rhodniin ornithodorin-thrombin interface (about 1900 A²), allowing a large variety of interdependent factors to contribute to the formation of the ecotin-protease complex. The macromolecular recognition is very sensitive to the specific residues located at the binding interface. Second, the tetrameric network of interactions creates greater flexibility to modulate the strength of inhibition by introducing new controlling elements such as the "hinge" region at the dimer interface. In this aspect, the ecotin-protease interaction is reminiscent ofthe antibody- antigen interaction, in which the hyper-variable regions ofthe six CDR loops of immunoglobin provide all the possible surface landscapes to recognize any given antigen through an astronomical number of combinatorial side chain conformations. Ecotin's four surface loops, the 50s, 60s, 80s, and 100s loops, have great potential to be tailored to provide a complementary fit with different protease surfaces that are in direct contact with both the primary and secondary binding site.

The dramatic Kj differences of ecotin variants containing multiple alanine substitutions at the secondary site against bovine trypsin, rat trypsin, and uPA highlight the novel structural and functional features of ecotin. Rat and bovine trypsins are closely related enzymes with the same substrate specificity. Their crystal structures are very similar. Although rat trypsin is anionic, while bovine trypsin is cationic, it is still not clear whether the electrostatic interaction is solely responsible for the selective inhibition of rat trypsin by the ecotin 60s loop variants. Other crucial structural factors may contribute to the binding energy of complex formation such as hydrophobic packing ofthe interface, or the flexibility and rigidity ofthe surface loops. For example, structural studies suggest that the relative position ofthe two ecotin monomers can shift upon binding to rat trypsin when binding determinants in the 60's loop are removed (unpublished results).

A systematic approach has been taken to investigate the dynamic relationships among the four major determinants ofthe interactions between ecotin and its target protease, the primary reactive site loop, the 60s and 100s loops, and the dimer interface at the C-terminal hinge region of ecotin. Our data suggest that the mutations at these four regions are nonadditive. Furthermore, the importance of ecotin's unique secondary binding site in recognizing and differentiating between different target proteases has been firmly established. A comprehensive set of ecotin variants at the primary and secondary sites has been generated to probe the molecular recognition between ecotin and target serine proteases. Ecotin binds to proteases through distinct combinations of interactions between the primary and secondary site surface loops. The primary and secondary sites are adapted to the surface features by their intrinsic flexibility and by the C- ine αiscπminating power ot the individual surface loops at the secondary site of ecotin suggests a novel opportunity to exploit the subtle difference among proteolytic enzymes with identical primary substrate specificity and to design selective and potent macromolecular inhibitors against these enzymes.

Example 2: Engineering Bidentate Macromolecular Inhibitors for Trypsin and Urokinase-type Plasminogen Activator

A) Materials and Methods

Materials Enzymes and reagents for the manipulation of DNA were purchased from

Promega (Madison, WI) or New England Biolabs (Beverly, MA) and were used following the manufacturer's instructions. The E. coli strain JM101, XL-1 Blue F' and the VCSM13 helper phage were from Stratagene (La Jolla, CA). The E. coli ecotin gene deletion strain IMΔecoJ was derived from JM101. Low molecular weight uPA (LMuPA) was obtained from American Diagnostica (Greenwich, CT). Rat trypsin was expressed in E. coli using the expression vector pZβ and purified as described (Higaki et /.(1989) Biochem., 28(24):

9256-9263). Bovine trypsin was from Sigma (St. Louis, MO). Δ-^-^S-dATP was from DuPont NEN (Boston, MA). Sequenase Version 2.0 sequencing kit was from U.S.

Biochemical Corp (Cleveland, OH). GeneClean® was from BiolOl, Inc (La Jolla, CA). Oligonucleotides were synthesized with an Applied Biosystems 391 DNA synthesizer (Foster City, CA). Falcon polystyrene petri dishes were from Becton Dickinson Labware (Lincoln Park, NJ). All other chemicals were of reagent grade or better and were used without further purification.

Plasmid and Library Constructions Mutagenesis was performed by the method of Kunkel fKunkel, 1985 #lθ3.

All mutations have been confirmed at the DNA level by sequencing. The vector pBS eco- glll was used to construct phage libraries 6OX4 and M84R+6OX4. A deletion and frameshift mutation was introduced at residues 67-70 of ecotin by primer 5'-C AAA ACG CTG GAA GG TAT TAT GTC TTT GAT-3' (SEQ ID NO.^' 12) to make pBS eco-gIIIΔ60. This construct was used as template to make library 6OX4 by primer 5'-AAC AAA ACG CTG GAA GGC NNS NNS NNS NNS TAT TAT GTC TTT GAT AAA GTC AG-3' (Ν=A/C/G/T, S=C/G) (SEQ ID NO. 13). Primer 5'-GT TCC CCG GTT AGT ACT AGG ATG GCC TGC C-3' (SEQ ID NO. 14) was used to introduce an M84R mutation in pBSeco-gIIIΔ60 to generate pBSeco-glll M84RΔ60. This vector was then used to generate library M84R+60X4 by the same 60s loop library primer 5'-AAC AAA ACG CTG GAA GGC NNS NNS NNS NNS TAT TAT GTC TTT GAT AAA GTC AG-3' (Ν-A/C/G/T, S=C/G) (SEQ LD NO. 15). Both libraries of ecotin phage had four positions 67-70 randomized. The ecotin phage display vector pBSeco-glll and expression vector pTacTacEcotin were mutated to carry an Aatll site by primer 5'-CA GAC AAT GTA GAC GTC AAG TAC CGC GTC-3* (SEQ ID NO. 15) at amino acid 125 of ecotin to facilitate the cloning between the two vectors. All ecotin variants obtained from panning experiments could be directly cloned into the expression vector pTacTacEcotin. For library phage preparation, the mutagenesis reaction mixture was purified by Geneclean, redissolved in water, electroporated into F' XL-1 Blue cells, and grow in 100 ml of 2YT/Ampicillin for 1 to 2 hours. The culture was then divided into two portions. One portion was transferred into fresh 2YT/ Ampicillin medium and grown for 8-12 hours. The cells were harvested by centrifugation and the double strand plasmid was prepared using a Promega Midiprep Kit. This DNA sample was the ecotin phage library stock. The other portion ofthe culture was subsequently diluted to an OD600 = 0J5, and then infected with VCSM13 helper phage. This infected culture was grown for 6 hours at 37°C with rigorous shaking, and the phage were harvested as described in the section of ecotin phage preparation below.

Ecotin Mutagenesis, Expression and Purification

Ecotin and ecotin variants were expressed and purified as described in Example 1.

Determination of Ecotin Ki Values against Target Proteases

Ecotin and ecotin variants Ki values against bovine, rat trypsin and uPA were determined as described in Example 1.

Ecotin Phage Preparation

For the preparation of pBSeco-glll ecotin bacteriophage, plasmid DNAs were transformed into a male strain (F') of IMΔecoJ. A single colony selected on ampicillin plates was grown in 3 ml 2YT medium containing 60 μg/ml ampicillin at 37°C for 8 hours. The culture was diluted into 100 ml of 2 YT/ ampicillin, grown to OD600 = 0.25, and infected with the helper phage VCSM13 at a multiplicity of infection of approximately 100 helper phage per cell. The infected culture was allowed to grow at 37°C with shaking for approximately 6 hours. Phage particles were harvested by precipitation with one fifth volume of 20 % polyethylene glycol 8000, 2.5 M NaCl at 4°C overnight, centrifugation at 6000 g for 40 min, and resuspended in 1 ml TE buffer. Phage stocks were stored at 4°C for up to six months. Phage titers typically ranged from 10^ to 10* *-• cfu μl culture and were stable within six months.

Ecotin Library Panning Polystyrene petri dishes (35 mm, Falcon 1008) were coated with 1 ml of 10 μg/ml bovine trypsin, or rat trypsin, or LMuPA in PBS (137 mM NaCl/2J mM-KCl/10 mM Na₂HPO₄/l-8 mM K₂HPO₄ pH 7.5) overnight at 4°C, and excess binding sites were blocked with 5% non-fat dry milk PBS solution for 2 hours. Phage were added to the dishes in buffer containing 1 ml PBS/0.5% Tween 20 and were incubated for 2 to 24 hours with gentle agitation at 4°C. Solutions containing the phage were then removed and the dishes were washed 9 times with 5 ml PBS/0.5% Tween 20. Bound phage were serially eluted by incubation with 900 μl of 0J N HCl Glycine solution (pH 2.2) with gentle shaking for 15 minutes at room temperature. Three elutions were performed. The eluates were neutralized with 167 μl 1 M Tris/HCl, pH 8.8.

Ecotin Library Amplification

Equal volumes of three eluates were pooled for subsequent amplification and characterization. Two amplification protocols were used in this experiment. In the liquid amplification procedure, 900 μl of phage elution pool was incubated with 9 ml of fresh grown LMΔecoJ lawn cells for 30 min at 37°C with gentle shaking. The infected lawn cells were transferred to 200 ml of 2YT/ampicillin liquid medium and continued to grow to an

OD600 = 0J5, and infected with VCSM13 helper phage. The infected culture was grown for 6 to 10 hours at 37°C with shaking, then harvested and precipitated as described above. In the plate amplification procedure, 9 ml of fresh grown IMΔecoJ lawn cells was incubated with 900 μl of phage elution pool at 37°C for 30 min with gentle shaking. The cells were then plated on multiple LB/ampicillin plates (polystyrene petri dishes, 150 mm, Falcon

3025) and grown for 6 to 10 hours. Cells from these plates were recovered by soaking each plate with 5 ml of 2YT and scraping the cells into the medium. The 2YT medium containing cells from plates were collected in 100 ml of 2YT/ ampicillin and continue to grow for one to two hours. This cell culture was then diluted in 100 ml 2YT/ampicillin to an OD600 = 0.25, and infected with VCSM13 helper phage. The infected culture was grown for 6 hours at 37°C with shaking, and the phage were harvested as described above.

B Results

Identifying Key Residues at the 60s loop of the Secondary Binding Site

Taking advantage of its large binding interface and the unique bidentate mode of interaction, ecotin effectively recognizes a broad range of serine proteases with different substrate specificities. In the accompanying paper (Yang et al. (1998) J. Mol. Biol, 279: 945-957), the roles of the four surface loops (the 50s, 60s, 80s and 100s loops) have been systematically dissected and the key elements responsible for the binding energy ofthe complex formation have been identified. The results of this region-specific mutagenesis study firmly established the significance ofthe secondary binding site and showed that this site played different roles with different target enzymes. The complete experimental results and detailed analysis are described in Example 1. The sharp contrasts ofthe K^*'s ofthe two ecotin alanine substitution variants (6OA4 and M84R+60A-4) towards bovine trypsin, rat trypsin and uPA illustrated that there were substantial differences among the coπesponding C-terminal surface regions ofthe proteases, where the 60s loop bound to the target enzyme. These experiments confirmed the potential of residues 67-70 to serve as novel determinants of specificity and potency of protease inhibition.

Through site-directed and region-specific mutagenesis, a four amino acid residue cluster at the 60s loop has been located as the starting point for engineering ecotin to be a highly specific and potent inhibitor. With limited understanding ofthe nature of the interactions at this region, it was unrealistic to conduct de novo design of a specific ecotin variant that would enhance its binding towards a target protease. On the other hand, it was equally difficult to test all the amino acid combinations on this surface loop individually since the total number of variants is 160,000 (20^), beyond the limit of our current assay methodology. A combinatorial method was developed to approach this problem by taking advantage of the in vitro selection power of phage display, to directly isolate strong binding ecotin variants from a comprehensive 60s loop library. Designing and Constructing Ecotin Phage Display Libraries at the 60s loop

The εcotin phage display vector pBSeco-glll expresses the fusion protein of full length ecotin connected to the C-terminal domain of filamentous phage minor coat protein pIII via a GlyGlyGly linker. With the addition of VCSM13 helper phage, the ecotin- pIII fusion protein is assembled onto phage particles. Phage carrying this fusion protein has ecotin activity and can bind to the immobilized protease on the solid surface. The three enzymes, bovine trypsin, rat trypsin, and uPA were coated onto polystyrene petri dishes and remained active as monitored by 7-nitroanilide release of Z-GPR-p -Na substrate after 30 min incubation at room temperature (25°C).

Table 9. The consensus sequence from library 60x4 panning against rat trypsin.

Position: 67 68 69 70

Y G F I w G I Q w G F T w G L P w Q L P w G L P w G W G w G F N

R G Y P w G F s w G Y D w G Y D w G F P w G Y D w G L W w G M P w G M P w G I P

Consensus w G F P

Occurrence 16 17 5 8

(PozPe)l 20.9 15.5 6.0 6.7

Consensus L

Occuπence 4

(P_OIP_e)/σ 1.9

Wt W G Y D

^aPe, the expected frequency of possible NNS (N=A,C,G,T, S=C,G) codons. Po, the observed frequency of codons in the clones sequenced, n, number of clones sequenced, σ [Pe(l-Pe)/n]^I/2. Two different ecotin libraries were designed, taking advantage of three- dimensional structure information to randomize key residues thereby permitting isolation of the optimal cognate inhibitor for a target protease. The first library, 6OX4, had four residues randomized at positions 67-70 ofthe 60s loop. This library was used to pan against bovine trypsin and rat trypsin separately. The results ofthe multiple alanine substitution experiments showed that the 60s loop was not a determinant of binding for bovine trypsin (Table 9). This suggests that bovine trypsin would bind to many ecotin 60s loop variants equally well regardless ofthe identities of the residues at positions 67-70. In contrast, only a small subset of ecotin variants within the 60s loop library should inhibit rat trypsin with high affinity. For rat trypsin inhibition, the side chain contribution from the 60s loop was important in the absence of strong favorable electrostatic interactions at the PI position. In this case, selective binding towards rat trypsin would certainly put greater constraints upon the nature ofthe amino acid moieties at positions 67-70. Panning the ecotin 6OX4 library with rat trypsin would likely produce one or a few strong binders from the millions of library members.

The second library, ecotin M84R+60X4, combined a favorable PI Arg residue with the randomized 60s loop. This library was designed to encode ecotin variants that inhibit uPA with high potency. The Ki of ecotin M84R+60A-4 against uPA is 1470 nM, several hundred fold higher than the K_ of ecotin M84R, suggesting that four amino acid substitutions 25 A away from the active site were sufficient to induce a dramatic change in the strength ofthe interaction. Since the inhibition of uPA was so sensitive to the residue substitutions at the 60s loop even in the presence of M84R, we reasoned that the uPA panning experiment with the ecotin M84R+60X4 library had a high probability of generating a subset of ecotin variants that recognize uPA with increased potency and specificity. The two ecotin libraries were constructed through Kunkel mutagenesis

(Kunkel (1985) Proc. Natl. Acad. Sci. USA, 82(2): 488-492). The NNS (N=A,C,G,T,

S=C,G) encoding scheme allowed a maximum of 1.0 x 10^ possible DNA sequences that represent 1.6 x 10-5 possible amino acid residue combinations. The ecotin 6OX4 library contained approximately 5 million individual clones; the ecotin M84R+60X4 library contained approximately 50 million individual clones. The completeness ofthe library was calculated using the equation:

N = ln(l-p) / ln[l-(l/n)] (1) where N is the number of total individual clones in the library, n is the number of possible combinations, and p is the probability that any clone can be found in the library given library size N. In both cases, the sizes of the final libraries indicated that these libraries were well over 99% complete in representing all the possible four amino acid sequence combinations at positions 67-70. Both libraries were characterized for completeness using the same procedure. Random individual clones were isolated from the libraries. Their plasmid DNA and phage were purified using standard procedures. A BamHI/Hindlll restriction digest ofthe sample plasmid DNA was used to monitor the total size ofthe ecotin phage display vector and the size of the DNA fragment containing the ecotin-glll fusion. All the individual clones characterized contained a full length insert. Double-strand nucleotide sequencing ofthe ecotin M84R+60X4 library members revealed that 60% ofthe library members were variants. The heterologous nucleotide sequences within the correct reading frame at the designed positions from 67 to 70 encoded random amino acid residues without significant deviation from the expected frequency distributions. An immuno-blot assay of the library phage sample using rabbit anti-ecotin antibodies detected that ecotin was expressed on phage. Bovine trypsin activity inhibition assays ofthe phage library mixture further assured that the ecotin variants displayed on phage were still active. Selected ecotin variants were over-expressed, purified and characterized kinetically (see below).

Panning of the ecotin 60X4 Library with Bovine and Rat Trypsin Throughout this study, two criteria have been adopted to assess the effectiveness of a panning experiment. One was the total recovery of phage eluted from the plates. A rigorous panning protocol should allow the enrichment of strong binding clones over background in the course of iterative panning and amplification. The enrichment could be indirectly measured by comparing the phage recovery in the library panning experiment to the recovery in a control experiment where no protease was coated onto the polystyrene plate solid support. The number of phage recovered from nonspecific background binding with the plate and the blocking agent was less than 10-5 under our normal experimental conditions. A typical round of ecotin phage library panning with significant enrichment would yield 10^ to 10^*3 phage from the elution, given that the input phage was in the range of 10^0 to 10*** *^■*. Although an elevated recovery usually suggested an increase of positive clones in the pool of panning intermediates, other factors that were not directly related to the in vitro binding between ecotin and the immobilized protease might also lead to an artificially high recovery. These factors included the emergence of certain deletion phage variants that possessed a significant growth advantage in the liquid amplification procedure, or a sub population ofthe phage library that bound to the solid support or the blocking agent. To assure the in vitro selection was based solely on the strength ofthe interactions between ecotin phage and immobilized protease, another more stringent criteria was also been applied to evaluate the outcome of a panning experiment: the emergence of one or a few consensus sequences in the final population ofthe ecotin phage library. Ecotin variants encoded by these consensus sequences were then made and characterized kinetically for their affinities towards the immobilized protease used in the panning experiment. The ecotin 6OX4 library was panned against two serine proteases, bovine and rat trypsin, which were coated onto separate polystyrene plates. After the first round of panning, the elution fraction from the two plates were pooled and amplified separately on multiple LB/ampicillin plates to generate two intermediate libraries for the next round of panning. The two panning experiments were then carried out in parallel for the subsequent rounds. Careful precautions were taken to avoid cross-contamination between the bovine trypsin and rat trypsin panning. Four rounds of panning were completed before the final sequencing of the individual clones. For both ligands, there was significant increase of both total recovery of phage eluted and the percentage of phage recovered after each round (data not shown), suggesting the enrichment of phage carrying ecotin variants that bind to the two immobilized proteases. The individual clones from the final round of bovine trypsin and rat trypsin panning were sequenced.

The DNA sequencing results showed that panning the ecotin 6OX4 library against bovine trypsin resulted in a heterologous population of sequences. It confirmed that the impact of 60s loop side chains on binding was minimal under such conditions. On the other hand, panning with the ecotin 6OA4 library against rat trypsin generated a clear consensus sequence W, G, F/L, P for positions 67-70. The strong selection for W and G at positions 67 and 68 were highly significant. At position 69, all 18 clones contained large, hydrophobic residues such as Phe, Leu, He, Tyr, Trp and Met. Phe and Leu were the most predominant residues ofthe consensus with five and four occurrences, respectively. Pro emerged from the library as a predominant residue at position 70. Not surprisingly, the wild- type sequence WGYD appeared three times, confirming that this sequence was also a good solution for rat trypsin binding. Another result that further substantiated the consensus sequence was that the selection operated at the amino acid level instead of nucleotide level. Since the library was encoded by a mixture of NNS codons, amino acids, such as G, L and P, could be encoded by two or three codons. Indeed, among the 18 sequences, different codons were found with comparable frequencies for Gly (12 GGG and 5 GGC), Leu (2 CTG, 2 CTC) and Pro (4 CCC and 4 CCG).

Inhibition Kinetics of Ecotin Variants Y69F+D70P and Y69L+D70P

To show that the panning experiments selected for strong binders, the variant ecotins encoding the consensus sequences WGFP and WGLP at positions 67-70 were cloned into the expression vector pTacTacEcotin (McGrath et al. (1991) J. Biol. Chem., 266(10): 6620-6625). Ecotin Y69F+D70P and Y69L+D70P were expressed and purified. Their R^" s against bovine trypsin, rat trypsin and human uPA are listed in Table 10. The kinetic data shows that for both trypsins, ecotin Y69F+D70P and ecotin Y69L+D70P bound tighter than WT ecotin. Ecotin Y69F+D70P was the stronger binder with a lower Kj (bovine trypsin 30 pM; rat trypsin 80 pM). The similar increase in affinity by ecotin Y69F+D70P for both trypsins suggested a general optimization ofthe hydrophobic packing at the secondary binding site. For uPA, ecotin Y69F+D70P bound better (670 pM) than WT (2800 pM) and ecotin Y69L+D70P bound worse (20200 pM), implying that the inhibition against uPA was extremely sensitive to changes at the 60s loop and that the uPA-ecotin interface differed from the trypsin-ecotin interface at the secondary site. Clearly, high affinity ecotin variants modified at the secondary site can be selected from a phage display library.

Table 10. K* of ecotin variants from library 60X .

Variant Bovine Trypsin Rat Trypsin UPA (unit : μM) (unit : μM) (unit : μM)

WT 310 ± 60 930 ± 160 2.8 ± OJ

Y69F + D70P 30 ± 10 80 + 10 0.67 ±0.09

Y69L + D70P 190 ± 30 460 ± 50 20.2 ± 1.8

Liquid vs, . Plate Amplification of Library M84R+60X/ \

The isolation of an inhibitor with improved binding towards rat trypsin suggested that this approach could be used to identify a high affinity inhibitor for uPA. To reduce the panning cycle time and simplify the experimental procedure, the phage amplification process was changed from plate amplification to liquid amplification. Although panning of the ecotin 6OX4 library was conducted with plate amplification, no serious consequences were expected in switching the amplification protocols. The ecotin M84R+60X4 library' was panned with uPA through liquid amplification for four rounds. However, there was no significant increase of phage recovery in each round of the experiment. When samples from the final round were characterized, a significant fraction of the clones were deletions ofthe full-length ecotin phage clones. In fact, the outcome ofthe panning process was very sensitive to both the fraction of initial positive clones in the library and the panning protocol. Due to their selective growth advantage, the spontaneously occurring deletion phage grew faster in liquid culture, thereby gaining a significant advantage in competition with the full-length ecotin phage. By iterative enrichment, the clones that carry the deletion sequences could be amplified at the early stage of library propagation and dominated the final population.

Since the amplification procedure was actually a critical factor for the success ofthe panning experiments, the plate amplification protocol was further refined to eliminate the nutrient selection pressure for phage growth in liquid culture. A "nursing protocol" was developed to ensure that the small fraction of positive clones in the initial input library are not lost during amplification. This protocol used plate amplification to minimize the selective pressure for nutrients and short incubation time to limit growth, preventing certain clones from dominating the pool of selected variants. The low density growth on solid media and limited propagation times were essential in the first round of library amplification for subsequent selection of desired variant clones. Specifically, E. coli colonies infected with phage were grown on multiple large LB/ampicillin plates with appropriate density (2.5 to 6.0 x 10^) to ensure the separation of individual colonies. The same ecotin M84R+60X4 library was panned with uPA again using the new method. The plate amplification significantly increased the number of phage recovered from the elution, indicating the enrichment of clones that bind to uPA. The comparison ofthe phage yields from the two methods is shown in Figure 5. By adopting this "nursing" protocol, we have not only obtained a final phage sample that was highly enriched over the background (lO^'*⁷ to 10^ in total vs. less than 10-5), but also isolated a clear consensus sequence (see the following section). These results strongly show that the conditions of the amplification protocol had a direct impact on the outcome of the in vitro selection process.

Panning of the Ecotin M8 R+60X4_library with Rat Trypsin and uPA

The ecotin M84R+60X4 library was panned against rat trypsin and uPA in parallel with intermediate plate amplification steps. Similar to the results from panning the ecotin 6OA4 library, a significant increase of phage recovery was observed in the eluates in each of the four rounds of panning for both ligands. Again, the panning results confirmed our observations from prior mutagenesis experiments. In rat trypsin binding, the dominant role ofthe electrostatic interaction with Arg84 completely masked the impact ofthe 60s loop. Thus panning against rat trypsin did not generate a consensus sequence (data not shown), even though in the final round of panning, the phage recovery from the acid elution exceeded 10^*^. As anticipated, a consensus sequence was found in panning against uPA (Table 11). Furthermore, this consensus sequence WGY(R/P) was different from the previous consensus WG (ΕfL) P from panning the ecotin 6OX4 library against rat trypsin, suggesting that the in vitro selection process was indeed dependent on the nature of the

Table 11. The consensus sequence from library M84R + 60X₄ panning againsfuPA.

Position: 67 68 69 70 Kj (pM)

W G Y D

W G H R

M G Y P

W G Y R

Y G Y R

W G H R

W E F P

W G N R

W G Y R

W G Y Q

N G Y R

W G Y P

W G F G w G Y P w G Y P

F G Y K

W G Y G

W G Y W

Consensus W G Y R 50 ± 40

Occurrence 14 17 13 8

(P₀-Pe)/σ 18.2 15.5 6.8 5.1

Consensus W P 80 ± 20

Occurrence 4

(P₀-Pe)/σ 2.8

M84R W G Y D 3600 ± 630 immobilized protease. At positions 67 and 68, the same residues Trp and Gly were selected from the random sequence pool with high efficiency. At position 69, the wild-type residue Tyr occurred in nearly 75% of the sequences. The Asp at position 70 was replaced with Arg (eight times) and Pro (four times). This was the only residue that differed from the wild-type sequence. Again, the nucleotide sequences of the 18 samples from uPA panning showed a mixture of codons encoding the selected amino acids such as Gly (14 GGG and 3 GGC), Pro (3 CCG and 1 CCC) and Arg (5 AGG and 3 CGG), strengthening the conclusion that these residues were selected based on their contributions to increase the affinity towards uPA.

Inhibition Kinetics of Ecotin M84R+D70R and Ecotin M84R+D70P

Ecotin M84R+D70R and ecotin M84R+D70P were constructed by cloning the specific variant sequences from the phage clone into the expression vector pTacTacEcotin, taking advantage of a pair of common restriction sites BamHI/Aatll that flanked 90% ofthe ecotin gene. The two variants were purified to homogeneity via reverse phase HPLC. Their R.,*'s against uPA were determined and are listed in Table 12. The K_t of ecotin M84R+D70R was lower (50 pM) than that of ecotin M84R+D70P (80 pM), mirroring their relative occurrence in the consensus sequences. The D70R substitution alone reduced the Ki 72-fold, resulting in a 50 pM inhibitor of uPA, the tightest binding competitive inhibitor of uPA reported to date. The one amino acid substitution at position 70 generated a significant impact on the affinity of these two ecotin variants.

Table 12. Designing specificity and potency of ecotin.

WT M84R M84R + D70R

Ki (rat trypsin) (nM) 0.93 ± 0.16 0.38 ± 0.10 0.22 ± 0.04

K_t (uPA) (nM) 2800 ± 160 3.6 ± 0.6 0.05 ± 0.04

K (xL?A)IKi (rat trypsin) 3010 10 0.22

Preference value versus 1 300 13680 ecotin WT Preference value is calculated by dividing the ratio of Ki (uPA/AT,* (rat trypsin) of ecotin WT by that of ecotin M84R or M84R + D70R. This value reflects the fold change in preference ofthe variant ecotin compared to ecotin WT for a given protease.

The increased potency and specificity of ecotin M84R+D70R validates the strategy to optimize the affinity of ecotin towards uPA through a stepwise approach. The inhibition of rat trypsin can be used as a benchmark to assess the effectiveness of this methodology. Ecotin M84R+D70R, with a Ki against rat trypsin of 220 pM, was a better inhibitor of uPA than rat trypsin. This result was in sharp contrast with the preference of WT ecotin for rat trypsin over uPA by over 3, 000-fold. Table 12 summarizes the ecotin AT s towards rat trypsin and uPA generated through the combination of region-specific mutagenesis and phage display. The specificity of ecotin has been successfully converted from one serine protease to the other with a significant increase in potency at the same time. The overall specificity preference was 13, 680-fold.

C) Discussion

The consensus sequences generated from rat trypsin and uPA panning revealed that residues 67 and 68 of ecotin were essential in high affinity binding against the target serine protease. In the tetrameric complex of ecotin-trypsin, Trp67 of ecotin provided the majority ofthe hydrophobic interactions at the binding interface. In fact, this tryptophan side chain was within direct van der Waals distance of residues 91, 233, 234, 236, 237 and 240 (chymotrypsin numbering system) ofthe protease. In all three proteases, amino acids 234 and 237 were aromatic side chains. Residue 234 is either Tyr (rat and bovine trypsin) or Phe (uPA); residue 237 is a conserved Trp. The stacking of aromatic rings between ecotin and the protease provides a tightly-packed hydrophobic interface. Due to tryptophan's large side chain volume and buried surface area, substituting Trp67 of ecotin to any other amino acid might create an unfilled cavity that destabilizes the ecotin-protease complex. Thus Trp67 appears to be an integral part ofthe hydrophobic "core" ofthe secondary binding site. On the other hand, Gly68 was also selected from both panning experiments. This residue probably plays a more structural role to maintain the proper flexibility and main chain conformation ofthe 60s loop. Since Trp67 and Gly68 were conserved at the 60s loop, the other two residues, Tyr69 and Asp70 were the only candidates to provide differential recognition towards target proteases. The ecotin consensus sequence WG (F/L) P at positions 67-70 that resulted when the ecotin library was panned against rat trypsin represented an overall improvement ofthe hydrophobicity at positions 69 and 70. Both Phe and Leu are less apolar than Tyr, which has a hydroxyl group. In addition to its hydrophobicity, proline was commonly found at various types of turns to lock the surface loops into stable conformations. Pro70 was presumably selected for this reason. The ten-fold increase in affinity of ecotin Y69F+D90P towards both bovine and rat trypsin illustrated the modest improvement of side chain packing in the vicinity of amino acids 69 and 70. In the case of uPA inhibition, the charge reversion substitution of D70R suggests that an electrostatic interaction could also be important for this position. In the crystal structure of the trypsin-ecotin complex, Asp70 of ecotin is very close to Asn93 of rat trypsin. In fact, the O^δ of Asp70 is within hydrogen bonding distance (2.94 A) of the N^δ of Asn93. Although the crystal structure of the ecotin- uPA complex is not yet available, it is possible to build a complex model based on the high resolution structures ofthe rat trypsin-ecotin complex and the catalytic domain of human uPA. The recently published 2.5 A structure ofthe uPA protease domain has a similar topology to trypsin (Spraggon et al. (1995) Structure, 3(7): 681-691). By superimposing the catalytic triad of uPA and rat trypsin, a model ofthe ecotin-uPA complex was constructed. The polypeptide backbone conformation of uPA overlaid with that of rat trypsin very well with an r.m.s. of 0.66 A. Most ofthe differences occurred in several surface loops.

However, the region near the secondary binding site and close to position 93 of the protease were highly conserved (Figure 6). Residue 70 of ecotin is close to residue 93 of uPA in the model ofthe ecotin-uPA complex. In this context, the results ofthe uPA panning are easily interpreted. In uPA, residue 93 was an Asp instead of Asn. The consensus sequence Arg70 from the ecotin M84R+60X4 library panning against uPA was indeed an ideal choice to provide a counter charge to stabilize Asp93 by forming a salt bridge between these two side chains. Thus, we have not only shown affinity improvement through the creation of alternative intermolecular electrostatic interaction, but also refined the structural model of the ecotin-uPA complex. Since the D70R substitution decreased the K_t of ecotin variant 72- fold from 3.6 nM to 50 pM, this favorable charge substitution alone was responsible for about 2.5 kcal /mol ( RT ln72, where T=298 K) of binding energy improvement.

The agreement between the results from the multiple alanine substitution experiments and the phage display experiments suggests that mutagenesis data of the 60s loop can be used to design ecotin phage libraries and predict the possibility of a consensus sequence. To extend the scope of these efforts to other parts ofthe secondary site (e.g. 100's loop), the impact of a particular surface loop under different contexts was analyzed. The free energy differences of binding for the different ecotin variants were calculated and compared. The K,-'s of all ecotin variants were converted to free energy terms with the equation ΔG •= -RTlnK. The impact of the 60s loop when the PI residue is an Arg can be calculated as the ΔΔG = ΔG(M84R+60A4) - ΔG(M84R Based on our data in the previous study, the ΔΔG values of the 60s and or 100s loop in the presence of the M84R substitution were calculated and listed in Table 13. A high absolute value of ΔΔG will be a good Table 13. Ecotin variants ΔΔG (kcal/mol). ΔΔG Bovine Rat Trypsin uPA

Trypsin

ΔG(60A₄) - ΔG(WT) 0.10 -5.02 -3.16

ΔG(M84R + 60A₄) - Δ(M84R) -0.26 0.44 -3.56

ΔG(M84R + IOOA4) - Δ(M84R) 0.65 1.33 -4.16

ΔG(M84R + 6OA₄ + 100A₄) - Δ(M84R) -0.34 -0.16 -5.31

ΔG was calculated by converting Ki to RTln( T,-), where T = 298 K (25°C). The relative free energy was the ΔG(M84R + 1 OOA-_t) - Δ(M84R)G difference between corresponding variants. According to this scale, a 100-fold difference in Ki will be equal to -2.73 kcal/mol.

indication that the specific surface loop is a key contributor to protease recognition. The outcome ofthe ecotin phage display experiment is summarized in Table 14 for the comparison between the free energy difference and the in vitro selection process. For ecotin 60 A4 binding to rat trypsin, the ΔΔG was - 5.02 kcal/mol and a consensus sequence was observed when the ecotin 6OX4 library was panned against rat trypsin. Similarly, for ecotin M84R+60A4 against uPA, the ΔΔG was - 3.56 kcal mol and a consensus sequence was observed when the ecotin M84R+60X4 library was panned against uPA. Since the high ΔΔG values ofthe ecotin mutated in the binding determinants ofthe 60s loop (Table 13) correlate with the emergence of a consensus sequence (Table 14), we feel that this analysis can be sued to identify areas that are important for high affinity binding. Since the ΔΔG of the ecotins mutated in the binding determinants of the 100s loop(Table 13) are comprable to that ofthe 60s loop for uPA binding, the 100s loop should have a major impact on the strength of ecotin-τrPA complex formation. Panning a phage display library at this region will likely generate novel consensus sequences encoding ecotin variants with high affinity towards uPA. The multiple alanine substitution experiments establish a rigorous strategy from which future combinatorial selection or screening approaches might be rationally designed and developed. Combining region-specific mutagenesis with the comprehesive scope of phage display, this affinity optimization procedure has the potential to be applied to other systems of macromolecular recognition as well. Table 14. Summary of ecotin phage library panning results.

Library Bovine Trypsin Rat Trypsin uPA

Ecotin 6OA4 High Recovery High Recover '

No Consensus Consensus

Ecotin High Recovery High Recovery

M84R+60A-4 No Consensus No Consensus

The highly differentiated and specialized roles ofthe mammalian serine proteases demand inhibitors that interact with their target enzymes with strong selectivity. Both conventional small molecule and macromolecular inhibitor design approaches are constrained by the distinctive structural features at the proximity ofthe active site and binding pocket of the target enzyme. This limitation is partially due to our current knowledge on serine proteases on the reactive mechanism and substrate specificity of model enzymes such as trypsin and chymotrypsin. However, under physiological conditions, many mammalian serine proteases take advantage of a large array of cofactors to develop sophisticated mechanisms to control and regulate the potency and specificity of precise proteolytic events. For example, thrombin, a key enzyme that cleaves fibrinogen and forms fibrin clots in the blood coagulation pathway, is regulated by heparin, α.2-macroglobulin, antithrombin III, thrombomodulin (Stubbs et al. 1995 Trends in Biochem. Sci., 20(3): 131; Stubbs et al. (1995) Trends in Biochem. Sci., 20(1): 23-281) and monovalent ions such as

Na⁺ (Dang et al. (1996) Proc. Natl. Acad. Sci. USA, 93(20): 10653-10656). Through several cofactor interactions that are distal to the active site, thrombin achieves a high level of fine tuning and balance between its coagulation and anticoagulation activities in an intertwined web of biological pathways in haemostasis, platelet aggregation, tissue remodeling, mitosis and chemotaxis. In the case of uPA, the serine protease domain can form a high-affinity complex with several key partners such as PAI-1, PAI-2, protease nexin-1, and α - macroglobulin receptor (Fazioli et al. (1994) Trends Pharmacol. Sci., 15(1): 25-29; Andreasen et al. (1991) FEBS Letts., 33892): 239-245). These examples illustrate the variety of structural features that are involved in the macromolecular recognition between serine proteases and other macromolecules in vivo that takes advantage of binding sites other than the primary recognition pocket. It also suggests unexploited opportunities of using other protease surface regions as the basis of inhibitor design and engineering.

Ecotin offers a unique platform to investigate and utilize the contribution from a binding region distal to the primary binding site for protease inhibition. The dimeric macromolecular inhibition has special structural features for innovative methods of inhibitor design and engineering. By modulating the amino acid residues at the 60s loop, another level of control has been achieved in designing the specificity and potency of ecotin variants. The secondary binding site of ecotin not only facilitated the fine-tuning of the molecular recognition towards many known homologous enzymes, but also provided additional side chain conformational flexibility to accommodate other serine proteases with similar scaffolds. These experimental results demonstrated that ecotin had several distinct advantages as a generic starting point for inhibitor design. The first advantage is the electrostatic and hydrophobic surface diversity available in the contact regions between the ecotin dimer and two protease molecules. Secondly, the combinatorial approach of phage display makes it feasible and highly efficient to search and sort the large repertoires of ecotin surface loop variants. Finally, the crystal structures of ecotin-protease complexes can serve as a framework for designing inhibitors against enzymes with unknown structures. By taking advantage ofthe unique secondary binding site of ecotin, it is possible to improve the affinity against a target protease through stepwise optimization of various surface loops involved in the binding interaction. This approach will be especially powerful for enzymes that share the same primary substrate specificity.

A combination of site-directed mutagenesis and phage display approaches were taken to study the interactions between ecotin and several serine proteases. The secondary binding site of ecotin was shown to play a critical role for certain proteases. Phage display libraries of ecotin variants were then made at these surface loops and used for panning against the target proteases. A protocol was developed that permitted identification of two distinctive consensus sequences from panning the ecotin variant phage libraries with rat trypsin and uPA. In both cases, the consensus sequence encoded ecotin variants with higher affinity for the target protease. This study provided a general strategy to engineer potency and specificity of a macromolecular serine protease inhibitor by modulating various components ofthe network of extended interactions between the inhibitor and the protease.

Example 3: Ecotin is a Factor IXa Agonist

A) Ecotin and ecotin variant preparation.

Three preparations of Factor IXa were evaluated in this study. Preparation #1, used in the experiments described in Figure 7, was the same lot of Factor IXa (L0430) used in our previous studies. This material was prepared from immunoaffinity purified human factor IXa by activation with human factor Xla. The reaction mixture was then passed over a gel filtration column to separate the factor IXa (M_r - 45,000( from factor Xla (M_r = 160,000). Preparation #2, used in the experiments described in Figure 8, was derived from the parent human factor IXa (Lot L0430) by passage over an anti-factor XI immunoaffinity column. Preparation #4, used in the experiments described in Figure 9 was derived from the parent human factor IXa (Lot L0430) by adsorption to and subsequent elution from an anti-factor IX immunoaffinity column.

The various factor IXa preparations were reconstituted to a final concentration of 2.0 μM in 0.05 M Tris-0.1 M NaCL,-0.005 M CaCl₂, pH 7.4 and kept on ice. A portion was taken from each of these solutions and diluted in the same buffer to a concentration of 0J μM for the assay reaction.

Ecotin and the M84R variant was obtained as described above. The concentration of ecotin was 15.4 μM and the M84R variant was 5.6 μM. Both proteins were in 0.05 M Tris-0.1 M NaCl-0.005 M CaCl₂, pH 7.4.

B Assays. To perform the assays, 20 μL factor IXa, 20 μL ecotin, M84R variant or 0.06

M Tris-0.1 M NaCL-0.005 M CaCl₂, pH 7.4 were added to a microtitre plate well slowly to avoid mixing before the addition ofthe assay buffer. A buffer (200 μL 0.05 M Tris-OJ M NaCl-0.005 M CaCl₂, pH 7.4 containing 40% (V/V) ethylene glycol) was added t the well. The assay reaction was started with the addition of 25 μL 1.0 mM nitroanilide substrate (CH₃SO₂-CHG-Gly-Arg-pHA-AcOH in H₂O). The microtiter plate was immediately placed in a Molecular Devices Microplate reader with the chamber maintained at 37°C. Absorbance at 405 nm was measured at various time intervals and the data calculated with the Softmax computer program.

C) Results As illustrated by Figures 7, 8, and 9 all three ecotins markedly stimulated nitroanilide hydrolysis by factor IXa preparations. Conversely, as expected, the ecotin M84R variant was a potent inhibitor of all three factor IXa preparations. While inhibition of factor IXa by the M84R variant was not unexpected stimulation by native ecotin was surprising. The similar effects on all three different factor IXa preparations indicates that the agonistic activity of ecotin was not due to the presence of a contaminant present in the factor IXa preparations. Example 4: Use of Ecotin Variants to Bind Proteins other than Serine proteases -- Purification of haptoglobin.

In this example, haptoglobin was purified using an ecotin M84R, M85R affinity column. The ecotin variant(s) were coupled to an affigel matrix using standard conditions to produce an Ecotin M84R, M85R affinity column. The column (4 ml) was then equilibrated with phosphate buffered saline (PBS) and then incubated with bovine serum (-35 mL) for 2 h to bind the haptoglobin that is present in the serum. The column was then washed with PBS (50 mL) and the OD280nm was obtained to be sure that the column was clean of non-retained material. The retained material was then eluted using 50 mM glycine, 150 mM NaCl, pH 3.0 (--30 mL) and was immediately neutralized using 1 M Tris base (10 μL/1 mL of retained material). The OD 280 nm was monitored continuously to permit collection ofthe ecotin bound protein.

The retained material represented haptoglobin that was visualized by Coomassie brilliant blue staining. Haptoglobin run as an approximately 83 kD band that is recognized by monoclonal antibodies raised against haptoglobin.

The material was dialized and concentrated from the ecotin affinity column and then applied to a small Mono-Q column (buffer A-20 mM Tris, buffer B - Buffer A + 1 M NaCl, gradient 0-1 M in -20 min) and the haptoglobin was eluted in approximately 35%- 38%> NaCl. The N terminal 10 amino acids ofthe purified haptoglobin were sequenced to verify its identity.

In addition to haptoglobin, two unidentified proteases were also retained by the ecotin column. The proteolytic activity of these proteases was observed by running the samples eluted from the ecotin affinity column on an activity gel containing gelatin.

Example 5: Modification of carboxyl or amino termini.

The carboxyl and amino termini of ecotin or ecotin variants of this invention can be modified by insertions or deletions of one or more amino acids. Thus, for example, we have observed that a 17 megadalton bacteriophage particle can be added to the c- terminus of ecotin without affecting its function. We have also shown that adding a (His)₆ tag at the c terminus does not affect its function. In addition, deletions are made at the amino and at the carboxyl terminus without affecting the ecotin activity.

It is understood that the examples and embodiments described herein are for illustrative purposes only and that various modifications or changes in light thereof will be suggested to persons skilled in the art and are to be included within the spirit and purview oi this application and scope of the appended claims. All publications, patents, and patent applications cited herein are hereby incorporated by reference in their entirety for all purposes.

Claims

CLAIMS What is claimed is: 1. A method of enhancing the activity of a serine protease having a chymotrypsin fold, said method comprising contacting said serine protease with a native ecotin or an ecotin variant.

2. The method of claim 1 , wherein said ecotin variant is a serine protease activator having the formula: TJ-X¹ T ^50s-X²-I °°^s -X³ T ^80s X⁴ T ^100s X⁵ T² wherein X¹ is a polypeptide having the sequence of amino acids 8 through 50 of native ecotin (SEQ LD NO: 1); X² is a polypeptide having the sequence of amino acids 56 through 65 of native ecotin (SEQ LD NO: 1); X³ is a polypeptide having the sequence of amino acids 72 through 78 of native ecotin (SEQ ID NO: 1); X⁴ is a polypeptide having the sequence of amino acids 88 through 106 of native ecotin (SEQ ID NO: 1); X⁵ is a polypeptide having the sequence of amino acids 115 through 135 of native ecotin (SEQ ID NO: 1); L^50s, L°°^s, L^80s, L^100s are independently an amino acid, or a polypeptide consisting of 2 to 15 amino acids; T¹ and T² are independently an amino acid, or a polypeptide consisting of 2 to 50 amino acids; i, j, k, m, n, and p are independently 0 or 1 ; and said modulator specifically binds to and increases the activity of a serine protease.

3. The method of claim 2, wherein: T¹ is a polypeptide having the formula aa -aa -aa -aa -aa -aa -aa⁷-; T² is a polypeptide having the sequence of amino acids aa¹³ -aa¹³⁷- aa¹³⁸-aa¹³⁹-aa^,40-aa^,41-aa¹⁴²; and aa^!, aa², aa³, aa⁴, aa⁵, aa⁶, aa⁷, aa¹³⁶, aa¹³⁷, aa¹³⁸, aa¹³⁹, aa¹⁴⁰, aa¹⁴¹, and aa¹⁴² are optionally present amino acids that when present are independently selected from the group consisting of alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, and valine.

4. The method of claim 3, wherein said serine protease activator has the formula: T^X'-aa^-aa^-aa^-aa^-aa^-^-aa^-aa^-aa^-aa^-a^^a^'-^-aa⁷⁹- aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵-a^ wherein T¹ is a polypeptide having the sequence of amino acids 1 through 7 of native ecotin (SEQ LD NO: 1); T is a polypeptide having the sequence of amino acids 136 through 142 of native ecotin (SEQ ID NO: 1); and aa⁵¹, aa⁵², aa⁵³, aa⁵⁴, aa⁵⁵, aa⁶⁶, aa⁶⁷, aa⁶⁸, aa⁶⁹, aa⁷⁰, aa⁷¹, aa⁷⁹, aa⁸⁰, aa⁸¹, aa⁸², aa⁸³, aa⁸⁴, aa⁸⁵, aa⁸⁶, aa⁸⁷, aa¹⁰⁷, aa¹⁰⁸, aa¹⁰⁹, aa¹¹⁰, aa¹¹¹, aa¹¹², aa¹¹³, and aa¹¹⁴ are optionally present amino acids that, when present, are independently selected from the group consisting of alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, and valine.

5. The method of claim 4, wherein -aa^5I-aa⁵²-aa⁵³-aa⁵⁴-aa⁵⁵ is a polypeptide having the sequence of amino acids 51-55 of native ecotin (SEQ ID NO: 1).

6. The method of claim 4, wherein - aa⁶⁶-aa⁶⁷-aa^o8-aa⁶⁹-aa⁷⁰-aa⁷¹- is a polypeptide having the sequence of amino acids 66-71 of native ecotin (SEQ ID NO: 1).

7. The method of claim 4, wherein -aa⁷⁹-aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵- aa⁸⁶-aa⁸⁷- is a polypeptide having the sequence of amino acids 79-87 of native ecotin (SEQ LD NO: 1).

8. The method of claim 4, wherein -aa¹⁰⁷-aa¹⁰⁸-aa¹⁰⁹-aa^{I 10}-aa^{i n}-aa¹¹²- aa¹¹³-aa¹¹⁴- is a polypeptide having the sequence of amino acids 107-114 of native ecotin (SEQ ID NO: 1).

9. The method of claim 4, wherein -aa⁸⁴ is Arg.

10. The method of claim 4, wherein -said serine protease activator has the amino acid sequence of a native ecotin (SEQ ID NO: 1) with aa⁸⁴ replaced by Arg.

11. The method of claim 4, wherein: -aa⁵¹-aa ²-aa ³-aa ⁴-aa⁵⁵- is a polypeptide having the sequence of amino acids 51-55 of native ecotin (SEQ ID NO: 1); and -aa⁶⁶-aa^o7-aa⁶⁸-aa^o9-aa⁷⁰-aa⁷¹- is a polypeptide having the sequence of amino acids 66-71 of native ecotin (SEQ ID NO: 1).

12. The method of claim 4, wherein: -aa^5l-aa⁵²-aa⁵³-aa⁵⁴-aa^5:ι- is a polypeptide having the sequence of amino acids 51-55 of native ecotin (SEQ ID NO: 1); and -aa 79 -aa 80 -aa 81 -aa 82 -aa 83 -aa 84 -aa 8₅ -aa 86 -aa 87 - i •s a pol iypept Jid,e h , aving the sequence of amino acids 79-87 of native ecotin (SEQ ID NO: 1).

13. The method of claim 4, wherein: -aa⁵¹-aa⁵²-aa⁵³-aa⁵⁴-aa⁵⁵- is a polypeptide having the sequence of amino acids 51-55 of native ecotin (SEQ ID NO: 1); and -aa¹⁰⁷-aa¹⁰⁸-aa¹⁰⁹-aa¹¹⁰-aa^{1 H}-aa¹¹²-aa¹¹³-aa¹¹⁴- is a polypeptide having the sequence of amino acids 107-114 of native ecotin (SEQ ID NO: 1).

14. The method of claim 4, wherein: -aa⁶⁶-aa⁶⁷-aa⁶⁸-aa⁶⁹-aa⁷⁰-aa⁷¹- is a polypeptide having the sequence of amino acids 66-71 of native ecotin (SEQ ID NO: 1); and -aa⁷⁹-aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵-aa⁸⁶-aa⁸⁷- is a polypeptide having the sequence of amino acids 79-87 of native ecotin (SEQ ID NO: 1).

15. The method of claim 4, wherein: -aa⁶⁶-aa⁶⁷-aa⁶⁸-aa⁶⁹-aa⁷⁰-aa⁷¹- is a polypeptide having the sequence of amino acids 66-71 of native ecotin (SEQ ID NO: 1); and -aa^,07-aa¹⁰⁸-aa¹⁰⁹-aa^uo-aa^{n ι}-aa¹¹²-aa¹¹³-aa¹¹⁴- is a polypeptide having the sequence of amino acids 107-114 of native ecotin (SEQ ID NO: 1).

16. The method of claim 4, wherein: -aa⁷⁹-aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵-aa⁸⁶-aa⁸⁷- is a polypeptide having the sequence of amino acids 79-87 of native ecotin (SEQ ID NO: 1); and -aa^,07-aa¹⁰⁸-aa¹⁰⁹-aa^{1 10}-aa^{1 M}-aa^{1 12}-aa^{1 13}-aa¹¹⁴- is a polypeptide having the sequence of amino acids 107-114 of native'ecotin (SEQ ID NO: 1).

17. The method of claim 4, wherein: -aa⁵ -aa -aa -aa⁵⁴-aa ⁵- is a polypeptide having the sequence of amino acids 51-55 of native ecotin (SEQ ID NO: 1); -aa -aa -aa -aa⁶ -aa⁷ -aa⁷¹- is a polypeptide having the sequence of amino acids 66-71 of native ecotin (SEQ ID NO: 1); and

7 RO 81 R7 R^ R4 R R-fi R7 -aa -aa -aa -aa -aa -aa -aa -aa -aa - is a polypeptide having the sequence of amino acids 79-87 of native ecotin (SEQ ID NO: 1).

18. The method of claim 4, wherein: -aa ⁶-aa⁶⁷-aa -aa⁶⁹-aa⁷⁰-aa⁷¹- is a polypeptide having the sequence of amino acids 66-71 of native ecotin (SEQ LD NO: 1); and -aa 79 -aa 80 -aa 81 -aa 82 -aa 83 -aa 84 -aa 85 -aa 86 -aa 87 - i •s a po ilypep .ti-dje ι h_avi ^■ng the sequence of amino acids 79-87 of native ecotin (SEQ ID NO: 1); and -aa¹⁰⁷-aa^I08-aa¹⁰⁹-aa^uo-aa^{1 ! I}-aa^{, I2}-aa^{I I3}-aa¹¹⁴- is a polypeptide having the sequence of amino acids 107-114 of native ecotin (SEQ ID NO: 1).

19. The method of claim 4, wherein: -aa⁵¹-aa⁵²-aa⁵³-aa⁵⁴-aa⁵⁵- is a polypeptide having the sequence of amino acids 51 -55 of native ecotin (SEQ ID NO : 1 ) ; -aa ⁶-aa -aa -aa ⁹-aa⁷ -aa '- is a polypeptide having the sequence of amino acids 66-71 of native ecotin (SEQ ID NO: 1); and -aa¹⁰⁷-aa^,08-aa¹⁰⁹-aa^no-aa^{ι π}-aa¹¹²-aa¹¹³-aa¹¹⁴- is a polypeptide having the sequence of amino acids 107-114 of native ecotin (SEQ ID NO: 1).

20. The method of claim 4, wherein: -aa^il-aa⁵²-aa⁵³-aa^:,4-aa⁵⁵- is a polypeptide having the sequence of amino acids 51-55 of native ecotin (SEQ LD NO: 1); -aa⁷⁹-aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵-aa⁸⁶-aa⁸⁷- is a polypeptide having the sequence of amino acids 79-87 of native ecotin (SEQ ID NO: 1); and -aa^-aa^-aa'^-aa* ¹⁰-aa' ' '-aa^{1 12}-aa^{! , 3}-aa¹¹⁴- is a polypeptide having the sequence of amino acids 107-114 of native ecotin (SEQ ID NO: 1).

21. The method of claim 4, wherein: -aa³ -aa ^"-aa² -aa³ -aa - is a polypeptide having the sequence of amino acids 51-55 of native ecotin (SEQ ID NO: 1); -aa⁷⁹-aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵-aa⁸⁶-aa⁸⁷- is a polypeptide having the sequence of amino acids 79-87 of native ecotin (SEQ ID NO: 1) with a Met to Arg substitution at aa84; and -aa¹⁰⁷-aa¹⁰⁸-aa¹⁰⁹-aa¹¹⁰-aa^{l u}-aa^U2-aa^U3-aa^U4- is a polypeptide having the sequence of amino acids 107-114 of native ecotin (SEQ ID NO: 1).

22. The method of claim 4, wherein said serine protease is selected from the group consisting of plasma kallikrein, Factor Xlla, Factor Xla, Factor IXa, Factor Vila, Factor Xa, Factor Ila (thrombin), Factor Clr, Factor Cls, Factor D, Factor B, C3 convertase, trypsin, chymotrypsin, elastinase, enterokinase, urokinase plasminogen activator, tissue plasminogen activator, plasmin, tissue kallikrein, acrosin, α-subunit nerve growth factor, γ- subunit nerve growth factor, granulocyte elastase, cathepsin G, mast cell chymase, mast cell tryptase.

23. The method of claim 22, wherein said serine protease is Factor IXa.

24. A method of identifying a protein that enhances the activity of a serine protease, said method comprising: (i) contacting said serine protease with a binding protein library; and (ii) selecting members of said protein binding library that specifically enhance the activity of said serine protease; wherein said protein binding library protein library comprising ecotin variants that are polypeptides ofthe formula: T¹ ₁-X¹-L^30s _rX²-L⁶⁰VX³-L^80s _m-X⁴-L^100s„-X⁵-T² _p wherein X¹ is a polypeptide having the sequence of amino acids 8 through 50 of native ecotin (SEQ ID NO: 1); X² is a polypeptide having the sequence of amino acids 56 through 65 of native ecotin (SEQ ID NO: 1); X^J is a polypeptide having the sequence of amino acids 72 through 78 of native ecotin (SEQ ID NO: 1); X⁴ is a polypeptide having the sequence of amino acids 88 through 106 of native ecotin (SEQ ID NO: 1); X⁵ is a polypeptide having the sequence of amino acids 115 through 135 of native ecotin (SEQ ID NO: 1); L^50s, L^60s, L^80s, and L^100s are independently an amino acid or a polypeptide consisting of 2 to 15 amino acids; T¹ and T² are independently an amino acid or a polypeptide consisting of 2 to 50 amino acids; i, j, k, m, n, and p are independently 0 or 1.

25. The method of claim 24, wherein said ecotin variants are polypeptides ofthe formula: T¹-X^I-aa⁵¹-aa⁵²-aa⁵³-aa⁵⁴-aa⁵⁵-X²-aa⁶⁶-aa⁶⁷-aa⁶⁸-aa⁶⁹-aa⁷⁰-aa⁷¹-X³-aa⁷⁹- aa⁸⁰-aa⁸¹-aa⁸²-aa⁸³-aa⁸⁴-aa⁸⁵-aa⁸⁶-aa⁸⁷-X⁴-aa^,07-aa¹⁰⁸-aa¹⁰⁹-aa¹¹⁰-aa¹¹¹-aa¹¹²-aa¹¹³-aa¹¹⁴-X⁵-T² wherein T¹ is a polypeptide having the sequence of amino acids 1 through 7 of native ecotin (SEQ ID NO: 1); T² is a polypeptide having the sequence of amino acids 136 through 142 of native ecotin (SEQ ID NO: 1); and aa⁵¹, aa⁵², aa⁵³, aa⁵⁴, aa⁵⁵, aa⁶⁶, aa⁶⁷, aa⁶⁸, aa⁶⁹, aa⁷⁰, aa⁷¹, aa⁷⁹, aa⁸⁰, aa⁸¹, aa⁸², aa⁸³, aa⁸⁴, aa⁸⁵, aa⁸⁶, aa⁸⁷, aa¹⁰⁷, aa'⁰⁸, aa¹⁰⁹, aa¹¹⁰, aa¹¹¹, aa¹¹², aa¹¹³, and aa^{1 14} are optionally present amino acids that, when present, are independently selected from the group consisting of alanine, arginine, asparagine, aspartic acid, cysteine, glutamine, glutamic acid, glycine, histidine, isoleucine, leucine, lysine, methionine, phenylalanine, proline, serine, threonine, tryptophan, tyrosine, and valine.

26. The method of claim 24, wherein said library comprises at least 100 different polypeptide species.

27. The method of claim 24, wherein said ecotin variants are polypeptides displayed on the surface of bacteria or phage.

28. The method of claim 24, wherein said ecotin variants are polypeptides expressed as fusion proteins with the c-terminal domain of a filamentous phage minor coat protein