CA2450953A1 - Method of solubilizing, purifying, and refolding protein - Google Patents

Abstract

A method of modifying protein solubility employs polyionic polymers. These facilitate the solubilization, formulation, purification and refolding of proteins especially incorrectly folded proteins and aggregated proteins. Compositions are described that are suitable for formulating TFPI. The compositions allow preparation of pharmaceutically acceptable compositions of TFPI at concentrations above 0.2 mg/mL, and above 10 mg/mL.

Description

WO 961407$4 _ PCT/US96/09980 METHOD OF SOLUBILIZING, PURIFYING, AND REFOLDING PROTEIN
~~chnic_~~ Field o~tl~~,jnvention The invention relates to methods useful for refolding, solubilizing, formulating and purifying proteins. These methods are particularly useful for proteins that have been engineered by genetic recombination and produced in bacterial, yeast or other cells in a form that has a non-native tertiary structure.
~ac~around of the Invention To understand fully the entire process of gene expression, it is as important to understand the process for the folding of the peptide chain into a biologically active protein as it is important to understand the synthesis of the primary sequence. The biological activities of proteins depend not only on their amino acid sequences but also on the discrete conformations of the proteins concerned, and slight disturbances to the conformational integrity of a protein can destroy its activity. Tsou et al.
(1988) Biochemistry 27:1809-1812.
Under the proper conditions, the in vitro refolding of purified, denatured proteins to achieve the native secondary and tertiary structure is a spontaneous process. To avoid formation of stable, but undesired, structures, it is necessary to use the tertiary interactions (which are formed late in folding) with their high degree of selectivity power to select and further stabilize those early local structures that are on the correct folding pathway. Thus, the finite, but very low, stability of local structures could be the kinetic "proofreading" mechanism of protein folding. The activated state of folding with the highest energy is a distorted form of the native protein, and the slowest, rate-limiting step of unfolding and refolding appears to be close to the native state in terms of ordered structure. In addition, the refolding of many proteins is not completely reversible in vitro, and reactivation yields of less than 100% are frequently observed, which holds true in particular for experiments at high protein concentration, and competing aggregation of unfolded or partially refolded protein molecules may be the major reason for a WO 96/40784 PCTNS96~9980 lowered reversib~ity, as described in Fischer and Schmid, (1990) Biochemistry 29:2205-2212.
In the case of sufficiently large protein molecules, the nascent polypepdde chain acquires its native three-dimensional structure by the modular assembly of micro-domains. Variables including temperature, and cosolvents such as polyols, urea, and guanidinium chloride, have been tested to determine their role in stabilizing and destabilizing protein conformations. The action of cosoIvents may be the result of direct binding or the alterations of the physical properties of water, as described in Jaenicke et al. (1991) Biochemistry 30 (13):3147-3161.
Experimental observations of how unfolded proteins refold to their native thcee-dimensional structures contrast with many popular theories of protein folding mechanisms. Under conditions which allow for refolding, unfolded protein molecules rapidly equilibrate between different conformations prior to complete refolding. The rapid prefolding equilibrium favors curtain compact conformations that have somewhat IS lower free energies than the other unfolded conformations. The rate-limiting step occurs late in the pathway and involves a high-energy, distorted form of the native conformation. Thcre appears to be a single transition through which essentially all molecules fold, as described in Creighton et al. (1988) Proc. Nat. Acad Sei.
USA
85:5082-5086.
Various methods of refolding of purified, recombinantly produced proteins have been used. For example, the protease encoded by the human immunode6ciency virus type I (HIV-I) can be produced in Escherichia coli, yielding inclusion bodies harboring the recombinant HIV-I protease as described by Hui et al. (1993) J. Prot.
Chem.12:
323-327. The purified HIV-I protease was refolded into an active enzyme by diluting a solution of the protein in 50% acetic acid with 25 volumes of buffer at pH
5.5. It was found that a higher specific activity of protease was obtained if the purified protein was dissolved at approximately 2 mg/ml in 50~/o acetic acid followed by dilution with 25 volumes of cold 0.1 M sodium acetate, pH 5.5, containing 5% ethylene glycol and 10~/0 glycerol. Exclusion of glycerol and ethylene glycol led to gradual loss of protein due to precipitation. About 85 mg of correctly folded HIV-I protease per Liter of E
coli cell culture was obtained by this method, and the enzyme had a high specific activity.

Another example of refolding a recombinant protein is the isolation and refolding of H-ras from inclusion bodies of E. coli as described by DeLoskey et al., (1994) Arch. Biochem. and Biophys. 311:72-78. In this study, protein concentration, temperature, and the presence of 10% glycerol were varied during refolding.
The yield of correctly folded H-ras was highest when the protein was refolded at concentrations less than or equal 0.1 mg/ml and was independent of the presence of 10% glycerol. The yield was slightly higher at 4° than at 25°C.
The refolding of Tissue Factor Pathway Inhibitor (also known variously as Lipoprotein-Associated Coagulation Inhibitor (LACI), Extrinsic Pathway Inhibitor (EPl) and Tissue Factor Inhibitor (EFI) and hereinafter referred to as "TFPI") produced in a bacterial expression system has been described by Gustafson et al., (1994) Protein Expression and Purification 5: 233-241. In this study, high level expression of TFPI in recombinant E. coil xesulted in the accumulation of TFPI
in 1 S inclusion bodies. Active protein was produced by solubiization of the inclusion bodies in 8M urea, and purification of the fill-length molecule was achieved by cation exchange chromatography and renaturation in 6M urea. The refolded mixture was then fractionated to yield a purified nonglycosylated TFPI possessing in vitro biological activity as measure in the Prothombin clotting time assay comparable to TFPI purified from mammalian cells.
A non-glycosylated form of TFPI has -also been produced and isolated from Escherichia coli (E. coil) cells as disclosed in U.S. Patent No. 5,212,091.
The invention described in U.S. Patent No. 5,212,091 subjected the inclusion bodies containing TFPI to sulfitolysis to form TFPI-S-sulfonate, purified TFPI-S-sulfonate by anion exchange chromatography, refolded TFPI-S-sulfonate by disulfide exchange using cysteine and purified active TFPI by cation exchange chromatography. The form of TFPI described in U.S. Patent No. 5,212,091 has been shown to be active in the inhibition of bovine factor Xa and in the inhibition of human tissue factor-induced coagulation in plasma. In some assays, the E. coil- produced TFPI has been shown to be more active than native TFPI derived from SK hepatoma cells. However, TFPI
produced in E. coli cells is modified in ways that increase heterogeneity of the protein.

A need exists in the art of refolding recombinantly produced proteins to increase the amount of correctly folded TFPI during the refolding process. A need also exists for increasing the solubility of TFPI. Presently the yields of recombinantly produced TFPI
have been lower than desirable, and a need exists in the art of producing correctly folded TFPI. See for example Gustafuson et al. (1994) Protein Expression a»d Purificatio~r S:
233-241.
TFPI inhibits the coagulation cascade in at least two ways: preventing formation of factor VIIa/tissue factor complex and by binding to the active site of factor Xa. The primary sequence of TFPI, deduced from cDNA sequence, indicates that the protein contains three Kunitz-type enzyme inhibitor domains. The first of these domains is required for the inhibition of the factor VIIa/tissue factor complex. The second Kunitz-type domain is needed for the inhibition of factor Xa. The function of the third Kunitz-type domain is unknown. TFPI has no known enzymatic activity and is thought to inhibit its protease targets in a stoichiometric manner; namely, binding of one TFPI
Kunitz-type domain to the active site of one protease molecule. The carboxy-terminal end of TFPI is believed to have a role in cell surface localization via heparin binding and by interaction with phospholipid. TFPi is also known as Lipoprotein Associated Coagulation Inhibitor (LACI), Tissue Factor Inhibitor (TFI), and Extrinsic Pathway Inhibitor (FPI).
Mature TFPI is 276 amino acids in length with a negatively charged amino terminal end and a positively charged carboxy-terminal end. TFPI contains 18 cysteine residues and forms 9 disulphide bridges when correctly folded. The primary sequence also contains three Asn-X-Ser/Thr N-linked glycosylation consensus sites, the asparagine residues located at positions 145, 195 and 256. The carbohydrate component of mature TFPI is approximately 30'/0 of the mass of the protein.
However, data from proteolytic mapping and mass spectral data imply that the carbohydrate moieties are heterogeneous. TFPI is also found to be phosphorylated at the serine residue in position 2 of the protein to varying degrees. The phosphorylation does not appear to aEact TFPI function.
TFPI has been isolated from human plasma and from human tissue culture cells including HepG2, Chang liver and SK hepatoma cells. Recombinant TFPI has been expressed in mouse C 127 cells, baby hamster kidney cells, Chinese hamster ovary cells and hwnan SK hepatoma cells. Recombinant TFPI from the mouse C127 cells has been shown in animal models to inhibit tissue-factor induced coagulation.
A non-glycosylated form of recombinant TFPI has been produced and isolate from Ercherichia coli (~ coli) cells as disclosed in U. S. Pat. No.
5,2'12,091. This form, of TFPI has been shown to be active in the inhibition of bovine factor Xa and in the inhibition of human tissue factor-induced coagulation in plasma. Methods have also been disclosed for purification of TFPI from yeast cell culture medium, such as in Petersen et al, J.Biol.Chern. 18:13344-13351 (1993).
Recently, another protein with a high degree of structural identity to TFPI
has been identified. Sprecher et al, Proc. Nat. Acad. Sci., USA 91:3353-3357 (1994). The predicted secondary structure of this protein, caked TFPI-2, is virtually identical to TFPI
with 3 Kunitz-type domains, 9 cysteine-cysteine linkages, an acidic amino terminus and a basic carboxy-terminal tail. The three Kunitz-type domains of TFPI-2 exhibit 43%, 35% and 53% primary sequence identity with TFPI Kunitz-type domains 1, 2, and 3, respectively. Recombinant TFPI-2 strongly inhibits the amidolytic activity of factor VIIs/tissue factor. Hy contrast, TFPI-2 is a weak inhibitor of factor Xa amidolytic activity.
TFPI has been shown to prevent mortality in a lethal Escherichia rnli (~ coh~
septic shock baboon model. Creasey et al, J. Clin. Invest. 91:2850-2860 (1993), Administration of TFPI at 6 mg/kg body weight shortly after infusion of a lethal dose of E coli resulted in survival in all five TFPI-treated animals with significant improvement in quality of life compared with a mean survival time for the five control animals of 39.9 hours. The administration of TFPI also resulted in significant attenuation of the coagulation response, of various measures of cell injury and significant reduction in pathology normally observed in E. coli sepsis target organs, including kidneys, adrenal glands, and lungs.
Due to its clot-inhibiting properties, TFPI may also be used to prevent thrombosis during microvascular surgery. For example, U.S. 5,276,015 discloses the use of TFPI in a method for reducing thrombogenicity of microvascular anastomoses wherein TFPI is administered at the site of the microvascular anastomoses contemporaneously with microvascular reconstruction.
TFPI is a hydrophobic protein and as such, has very limited solubility in aqueous solutions. This limited solubility has made the preparation of pharmaceutically acceptable formulations of TFPI difficult to manufacture, especially for clinical indications which may benefit from administration of high doses of TFPI.
Thus, a need exists in the art for pharmaceutically acceptable compositions containing concentrations of TFPI which can be administered to patients in acceptable amounts.
Brief Description of the Drawings Figure 1 is a coomassie stained SDS-PAGE analysis of TFPI peak fractions from the Phenyl SepharoseTM HIC refolding procedure.
Figure 2 is a plot of the recovery of native TFPI from the HIC column.
Figure 3 is a plot of the recovery of native TFPI from a second HIC column.
Figure 4 is the amino acid sequence of TFPI.
Figure 5 shows the solubility of TFPI at different pH conditions. About 10 mg/mL TFPI in 2M urea was dialyzed against 20 mM acetate, phosphate, citrate, glycine, L-glutamate and succinate in 150 mM NaCI. The concentration of remaining soluble TFPI after dialysis was measured by LTV absorbance after filtering out the precipitates through 0.22 mm filter units.
Figure 6 shows the solubility of TFPI as a function of concentration of citrate in the presence of 10 mM Na phosphate at pH 7. TFPI solubiity increases with increasing concentration of citrate.
Figure 7 shows the solubility of TFPI as a function of concentration of NaCI .
TFPI solubiity increases with increasing salt concentration, indicating salt promotes solubility of TFPI.
Figure 8 shows effect of pH on the stability of TFPI prepared in 10 mlVl Na phosphate, 150 mM NaCl and 0.005% (w/v) polysorbate-80. Stability samples containing 150 mg/mL TFPI were incubated at 40°C for 20 days. Kinetic rate constant for the remaining soluble TFPI was analyzed by following decrease of the main peak on cation exchange chromatograms.

WO 96/40784 PCfNS96109980 Figure 9 shows the percentage of remaining soluble TFPI measured by canon exchange HPLC (A) and remaining active TFPI by prothrombin time assay (B) as a function of phosphate concentration. The formulation contains 150 mg/mL TFPI
prepared in 150 mM NaCI and 0.005 96 (w/v) polysorbate-80 at pH 7 with varying concentrations of phosphate.
Figure 10 shows loss of soluble TFPI at 40°C measured by both cation-exchange HPLC (triangle) and prothrombin time assay (circle) for 0.5 mg/mL
TFPI
formulated in 10 mM Na citrate, pH 6 and 150 mM NaCI.
Figure 11 shows loss of soluble TFPI at 40°C measured by both ,cation-exchange HPLC (open symbol) and prothrombin time assay (closed symbol) for 0.5 mg/mL TFPI formulated in 10 mM Na phosphate, pH 6 and either 150 mM NaCI
(triangle) or 500 mM NaCI (circle).
Figure 12 shows loss of soluble TFPI at 40°C measured by both cation-exchange HPLC (open symbol) and prothrombin time assay (closed symbol) for 0.5 mg/mL TFPI formulated in 10 mM Na acetate and pH 5.5 containing 150 mM NaCl (triangle) or 896 (w/v) sucrose (square) or 4.596 mannitol (circle).
Figure 13 shows two non-reducing SDS gels for TFPI formulation samples at pH 4 to 9 stored at 40°C for 0 and 20 days.
Figure 14 shows the time course of a polyphosphate-facilitated rhTFPI refold monitored using SDS PAGE.
Figure 15 shows the absorbance at 280 nm during~the loading and elution of the S-Sepharose HP column used to purify rhTFPI from a polyphosphate-facilitated refold.
Figure 16 shows SDS PAGE analysis of fractions collected during elution of the S-Sepharose HP column used to purify rhTFPI from a polyphosphate-facilitated refold: Figure 17 shows the absorbance at 280 nm during the loading .and elution of the Q-Sepharose HP column used to purify rhTFPI from a S-Sepharose pool prepared from a polyphosphate-facilitated refold.

-$_ Figure 18 shows SDS PAGE analysis of fractions collected during elution of the ~Sepharose HP column used to purify rhTFPI from a S-Sephamse pool prepared from a polyphosphate-facilitated refold.
Figure 19 shows the time course of a polyethyleneimine-facilitated rhTFPI
refold monitored using SDS PAGE.
Figure 20 shows the absorbance at 280 nm during the loading and elution of the S-Sepharose HP column used to purify rhTFPI . from a polyethyleneimine-facilitated refold.
Figure 21 shows SDS PAGE analysis of fractions collected during elution of the S-Sephamse HP column used to purify rhTFPI from a polyethyleneimine facilitated refold.
Figure 22 shows the absorbance at 280 nm during the loading and elution of the Q-Sephaiose HP column used to purify rhTFPI from a S-Sepharose pool prepared from a polyethyleneimine-facilitated refold.
Figure 23 shows SDS PAGE analysis of fractions collected during elution of the Q-Sepharose HP column used to purify rhTFPI from a S-Sepharose pool prepared from a polyethyleneimine-facilitated refold.
Figure 24 shows the ration exchange HPLC analysis of a 0.4 %
polyphosphate-facilitated rhTFPI refold in the absence of urea.
Figure 25 shows results of ration exchange HPLC analysis of an evaluation of different levels of cysteine on a rhTFPI refold in 0.4 % polyphosphate, 50 mM
Tris in the absence of urea.
Figure 26 shows the effect of polyphosphate chain length on the course of a polyphosphate facilitated refold of rhTFPI inclusion bodies as monitored by ration exchange HPLC.
Figure 27 shows the effect of concentration of polyphosphate (Glass H) on the refolding of rhTFPI from inclusion bodies as monitored by ration exchange HPLC.
Figure 28 shows the ration exchange HPLC analysis of polyethyleneimine and polyphosphate-facilitated refolding of purified and reduced rhTFPI.

Summary of the Invention It is an object of an aspect of the present invention to describe a method of refolding protein.
It is another object of an aspect of the invention to provide aqueous formulations of TFPI.
It is another object of an aspect of the invention to provide methods for modifying a protein's solubility using charged polymers.
It is still another object of an aspect of the present invention to describe a method of refolding TFPI including the steps of adding charged polymers to a solution of denatured TFPI prior to allowing the TFPI to refold.
Additionally, it is another object of an aspect of the invention to describe a method of refolding TFPI including the step of immobilizing charged polymers on a column and passing a solution of denatured TFPI through the column and eluting the refolded TFPI after the refolding has occurred.
It has now been found that solubility of TFPI is strongly dependent on pH and, surprisingly, that polyanions such as citrate, isocitrate, and sulfate have profound solubilizing effects on TFPI. This finding is surprising in light of the hydrophobic nature of TFPI and the hydrophilic character of these counterions. Thus, citrate, isocitrate, sulfate as well as other solubilizers described hereinbelow can be used to produce pharmaceutically acceptable compositions having TFPI concentrations sufficient for administration to patients. It has also been shown that other organic molecules can act as secondary solubilizers. These secondary solubilizers include PEG, sucrose, mannitol, and sorbitol.
The invention relates to pharmaceutically acceptable compositions wherein TFPI is present in a concentration of more than 0.2 mg/mL solubilizing agents.
The solubilizing agents may be acetate ion, sodium chloride, citrate ion, isocitrate ion, glycine, glutamate, succinate ion, histidine, imidazole and sodium dodecyl sulfate (SDS) as well as charged polymers. In some compositions, TFPI may be present in concentrations of more than 1 mg/mL and more than 10 mg/mL. The composition may also have one or more secondary solubilizers. The secondary solubilizer or solubilizers may be polyethylene glycol (PEG), sucrose, mannitol, or sorbitol.

Finally, the composition may also contain sodium phosphate at a concentration greater than 20 mM.
Although the solubility of TFPI is quite low between pH 5 and 10, it has been found that L-arginine can increase the solubility by a factor of 100. The solubility is very dependent on the concentration of arginine, as 300 mM is about 30 times more effective than 200 mM. Urea also is quite effective in solubilization of TFPI.
Further, it has been found that aggregation of TFPI appears to be the major degradation route at neutral and basic pH conditions and that fragmentation occurs at acidic pH conditions.
It has also been found that active TFPI monomers can be separated away from TFPI oligomers that are produced during the process of folding recombinant TFPI
produced E. coil. Some misfolded/modified monomeric forms of TFPI are also removed during this process. The separation employs hydrophobic interaction chromatography. The oligomeric species of TFPI bind more tightly to a hydrophobic resin than does the active TFPI monomers. Resins such as PharmaciaTM octyl sepharose and ToyopearlTM butyl 650-M have been successful. The process is carried out in the presence of high salt, such as 1 M ammonium sulfate or 0.5 M sodium citrate.
According to an aspect of the invention, there is provided a solution having a pH of from 5 to 10, comprising from 200 mM arginine to 300 mM arginine, and further comprising a polypeptide selected from the group consisting of (i) human TFPI having the amino acid sequence shown in Figure 4, (ii) ala-human TFPI
having the amino acid sequence shown in Figure 4 and having one further amino acid which is an amino-terminal alanine, and (iii) muteins of (i) or (ii) having from 1 to 5 amino acid substitutions.
According to another aspect of the invention, there is provided a solution having a pH of from 5 to 10, comprising more than 0.2 mg/ml of a polypeptide selected from the group consisting of (i) human TFPI having the amino acid sequence shown in Figure 4, (ii) ala-human TFPI having the amino acid sequence shown in Figure 4 and having one further amino acid which is an amino-terminal alanine, and (iii) muteins of (i) or (ii) having from 1 to 5 amino acid substitutions, and further comprising a solubilizer selected from the group consisting of sucrose, mannitol, sorbitol, citrate, isocitrate, succinate, malate, polyphosphate, sodium phosphate, -1 Oa-sodium sulfate, acetate, polysorbate-80, polyethylene glycol, histidine, imidazole, glutamate, glycine, ammonium sulfate, and sodium dodecyl sulfate.
Detailed Description of the Invention It is a discovery of the present inventors that polyionic polymers such as polyethylene imine and polyphosphate, can modify the ionic interactions within proteins. The masking of certain areas of high charge density within proteins using polyions can have numerous effects. Proteins whose solubility is reduced through the infra- and/or inter- molecular neutralization of oppositely charged areas can have their solubility improved by masking one of the charged regions with polycations or polyanions. Barners to conformational flexibility and specific attractive or repulsive forces which interfere with the refolding process can be modulated as well.
Proteins which require strong denaturants such as urea or guanidine hydrochloride to solubilize and maintain solubility during purification operations can be solubilized and processed effectively using polyions.

_11_ Many proteins lacling a clear region of charge localization in the primary sequence can still demonstrate areas of charge localization due to their secondary structure. Thus many proteins can have their solubility, refolding, and purification characteristics modifed through interaction with charged polymeric templates.
The nature of the modification will depend on the specific protein structure, the chain length, charge, and charge density of the ionic polymer.
We have characterized refolding of pure TFPI in a guanidine or a urea based refolding buffer and the results indicate that refolding efficiencies and kinetics can be significantly improved by the addition of charged polymers including heparin, dextran sulfate, polyethyleneimine (PEI) and polyphosphates. These polymers increase TFPI
solubility and enhance refolding through ionic interactions with either the N-terminus or the C-terminus. In addition to the polymer additives, refolding pure TFPI
requires a cysteindcystine redox buffer where the refolding reaction can be completed within 48 hours. Refolding yields are a strong function of pH, redox concentration, and pokyri~er additives; however refolding efficiencies as high as 6096 can be achieved for pure TFPI under optimum refolding conditions.
It has been found by the inventors of this invention that addition of glyeosaminoglycans or sulfated polysaccharides such as, for example, heparin and dextran sulfate, to a solution containing a denatured protein prior to refolding increases the amount of correctly folded, active protein where the protein is capable of binding to the glycosaminoglycan or sulfated polysaccharide and is subjected to renaturing conditions.
Recombinant DNA technology has allowed the high level expression of many proteins that could not normally be isolated from natural sources in any appreciable quantities. In E. coli and several other expression systems, the protein is frequently expressed in an inactive, denatured state where the primary amino acid sequence is correct, but the secondary and tertiary structure and any cysteine disulfide bonds are not present. The denatured protein present in an inclusion body may be in such a WO 96140"i84 t'CT/US96/09980 conformation that charged residues of different parts of the amino acid backbone that are not normally in contact are able to interact and form strong ionic bonds between positively charged and negatively charged amino acid residues. The formation of these ionic bonds may limit the hydration that must occur to effect dissolution of the inclusion body. The protein in an inclusion body may also be complexed with other cellular components such as membrane components and nucleic acid which may also Limit the access of solvent (water) to charged and normally hydrated residues.
Also found in the unfolded state is that hydrophobic residues which are normally found buried in the interior of a protein are more exposed to the polar aqueous environment.
Such occurrences may work to prevent the dissolution of inclusion bodies in solvents other than strong chaotropic agents such as urea or guanidine or detergents such as SDS.
Charged polymers preferably in aqueous solution can interfere with and disrupt the undesirable ionic interactions that occur within a polypeptide chain as found in an inclusion body or other environment. The charged polymers may help disrupt the undesirable ionic interactions and facilitate solvation of ionic and polar residues, promoting dissolution without tht need for strong chaotropes or detergents.
The charge, charge density, and molecular weight (chain length) of the charged polymer may vary depending on the specific protein. Suitable polymers include:
sulfated polysaccharides, heparins, dextran sulfates, agaropectins, carboxylic acid polysaccharides,alginic acids, carboxymethyl celluloses, polyinorganics, polyphosphates, polyaminoacids, polyaspartates, polyglutamates, polyhistidinea, polyorganics, polysaccharides,DEAE Dextrans, polyorganic amines, polyethyleneinimes, polyethyleneinime celluloses, polyamines, polyamino acids, polylysines, and polyarginines.
Proteins with pI greater than 7 may benefit more from interactions with negatively charged polymers, as these proteins will have a positive charge at pH 7.
Proteins with pIs below 7 may interact more strongly with positively charged polymers at neutral pH. Changing the solution pH will modify the total charge and charge distribution of any protein, and is another variable to be evaluated.

WO 96/40784' PC'TNS96/09980 Recombinant DNA technology has allowed the high Ievel expression of many proteins that could not normally be isolated from natural sources in any appreciable quantities. In E. coli and several other expression systems, the protein is frequently expressed in an inactive, denatured state where the primary amino acid sequence is correct, but the secondary and tertiary structure and any cysteine disulfide bonds are not present. The denatured protein must be refolded to the pmper active conformation, which often requires overcoming significant energy barriers imposed by ionic attraction and repulsion, restrictions on bond rotation, and other types of conformationally induced stresses. Specific ionic attraction between opposite charges and/or repulsion between like charges can severely limit the refolding pathways available to the denatured protein and reduce the efficiency of the refolding process.
Some proteins have specific areas of charge whose interactions may limit conformational flexibility or promote aggregation. Many proteins lacking a clear region of charge localization in the primary sequence can still demonstrate areas of charge localization due to their secondary swcture found in refolding intermediates, misfolds and improperly folded protein. Proteins which are particularly suitable acxording to the present invention include but are not limited to TFPI, TFPI
muteins, TFPI-2, tissue plasminogen activator, BST, PST. While it is believed that these methods will be suitable for use with proteins generically, those which are most suitably are those which are improperly folded, aggregated, oligomerized, or inactive.
These will likely be proteins which have at least one highly charged domain, and possibly more, which can interact. In the case of TFPI, as well as other proteins, two oppositely charged domains interact with each other to prohibit proper folding and to cause oligomerization and aggregation. Proteins having many disulfide bonds will also be most likely to benefit from the present methods. Preferably the protein will have at least 2 and more preferably the protein will have at least 4 or 6 disulfide bonds.

WO 96/d0784 PCTNS96/09980 -14~
Charged polymers can be used to modify the charge, charge density, and reduce or eliminate ionically mediated limitations to conformation that may arise in the unfolded state. The juxtaposition of charged groups that are not normally in proximity may have result in dead-end refolding pathways from which the refolding process may never recover.
The introduction of extra positive or negative charges through the complexation with charged polymers may allow refolding to proceed more facilely for several reasons: first different types of charge distributions may better accommodate the refolding process; second, the addition of the charged polymer may enhance the solubility of the unfolded pmtein, reducing or eliminating the need for chaotropes which have a negative effect on pmtein conformation.
Because of the frequently unique swcture associated with most proteins the charged polymer that demonstrates preferred characteristics may vary.
Evaluation of the isoelectric pH (pI) of the protein can serve as a starting point. At neutral pH a protein with a pI less than 7 will possess a net negative charge, and will thus be more likely to bind a positively charged polymer. The protein with a pI greater than 7 will possess a net positive charge at neutral pH , and will have a stronger tendency to bind a negatively charged polymer. However, it is well established that charges are unevenly distributed around a protein, and significant charge localization can occur.
The possibility of localized concentrations of charges reduce the ability to predict which type of charged polymer may be most effective for any application.
Theoretically, for any protein with a specific distribution of interacting charges and conformational requirements, there exists a charged polymer of appropriate compositions in terms of molecular weight, charge, and charge distribution which would maximize refolding efficiency. Other variables, such as pH and solvent ionic stretrgth, would also be evaluated. Initial screening would involve polyethyleneimine, DEAF dextran, dextran sulfate, and polyphosphate at several different concentrations and molecular weights. Work with rhTFPI has demonstrated the significant impact that polyphosphate chain length and concentration can have on the course of the TFPI
refolding reaction. Relatively short chain length (n=5) produces high levels of WO 96/40'184 PC?NS96/09980 aggregate. The optimal polyphosphate chain length for refolding rhTFPI was approximately 25 repeating units. Longer chain length polyphosphates (n~ 75) also produced more aggregate and less properly folded monomer.
Fnmaulatinn_ Proteins consist of chains of amino acids, the exact composition and sequentx of which constitutes one of the primary structure determinants of the protein.
The secondary determinant of protein structure is the result of the conformational guidance that the individual amino acid bonds have on protein conformation.
Thirdly, the specific amino acid sequence directs the formation of tertiary structures such as ~-sheets and a- helices. The three dimensional nature of protein conformation often brings into proximity amino acid residues that are not normally close to each other based on the direct sequence of the polypeptide chain. The functional form of a protein is generally a modestly stable conformation held together by a combination of cystcine disulfide bonds, ionic bonds, and hydrophobic and Van der Waals interactions.
In general, protein solubility can be related to the number of charged and to a lesser extent polar amino acids that make up the protein. These charged and polar groups become solvated with water molecules in the aqueous solution and this interaction kelps the polypeptide chain in solution. Polypepddes with insufficient numbers of positively or negatively charged amino acids can have limited aqueous solubility. In some cases, the positive and negatively charged groups present in a protein can interact with each other, displacing the water of solvation and leading to reduced aqueous solubility. Many proteins lacking a clear region of charge lacali~ation in the primary sequence can still demonstrate areas of charge localization due to their secondary swcture. Thus many proteins can have their solubility modified through interaction with charged polymeric templates. The nature of the modification will depend on the specific protein structure, the chain length, charge, and charge density of the ionic polymer.

7&t PCTNS96/09980 - 1b -Compleaation with charged polymers with relatively high charge density represents one approach to increasing the charge density of any protein. A
protein with a small number of positively charged residues (lysine or arginine) can be complexed with a negatively charged polymer such as polyphosphate. Some of the negatively charged groups of the polymer will interact with the positively charged groups present in the protein. The remaining charged groups on the polymer will be free to interact with the solvent, in most cases water, and effectively increase the charge density and solvation of the protein. Alternatively, a positively charged polymer such as polyethyleneimine can be used to complex with the negatively charged residues of the protein. In some cases both types of charged polymers may work equally effectively, in other cases, one charge type may be more effective than others. The effectiveness of any particular charged polymer, will depend on protein amino acid composition, protein amino acid distribution, protein conformation, charged polymer charge density, charged polymer chain length, solution pH, and other variables. However, it is likely that for any given protein, a complementary charged polymer that will bind to the protein and essentially increase the charge density of protein can be found that will improve the solubility characteristics of that protein in aqueous medium.
1'~niijn~ns The term "processing" as used herein refers to the steps involved in the purification and preparation of pharmaceutically-acceptable amounts of proteins.
Processing may include one or more steps such solubilization, refolding, chromatographic separation, precipitation, and formulation.
The term "charged polymer" and "charged polymeric template" refer to any compound composed of a backbone of repeating structural units linked in linear or non linear fashion, some of which repeating units contain positively or negatively charged chemical groups. The repeating swctural units may be polysaccharide, hydrocarbon, organic, or inorganic in nature. The repeating units may range from n = 2 to n=several million The term "positively charged polymer" as used herein refers to polymers containing chemical groups which carry, can carry, or can be modified to carry a positive charge such as ammonium, alkyl ammonium, dialkylammonium, trialkyl ammonium, and quaternary ammonium.
The term "negatively charged polymer" as used herein refers to polymers containing chemical groups which carry, can carry, or can be modified to carry a negative charge such as derivatives of phosphoric and other phosphorous containing acids, sulfuric and other sulfur containing acids, nitrate and other nitrogen containing acids, formic and other carboxylic acids The term "polyethyleneimine" as used herein refers to polymers consisting of repeating units of ethylene imine (H3N+-(CH2-CH2-NH2+)x-CH2-CH2-NH3+).
The molecular weight can vary from 5,000 to greater than 50,000.
The term "polyphosphate" as used herein refers to polymers consisting of repeating units of orthophosphate linked in a phospho anhydride linkage. The number of repeating units can range from 2 (pyrophosphate) to several thousand.
Polyphosphate is frequently referred to as sodium hexametaphosphate (SHMP).
Other common names include Grahams salt, CalgonTM, phosphate glass, sodium tetrametaphosphate, and Glass HTM.
The term "refold" as used herein refers to the renaturation of protein.
Typically, the goal of refolding is to produce a protein having a higher level of activity than the protein would have if produced without a refolding step. A
folded protein molecule is most stable in the conformation that has the least free energy.
Most water soluble proteins fold so that most of the hydrophobic amino acids are in the interior part of the molecule, away from the water. The weak bonds that hold a protein together can be disrupted by a number of treatments that cause a polypeptide to unfold, i.e. denature. A folded protein is the product of several types of interactions between the amino acids themselves and their environment, including ionic bonds, Van der Waals interactions, hydrogen bonds, disulfide bonds and covalent bonds.

WO 96/40'184 PCT/US96/09980 The term "denature" as used herein refers to the treatment of a protein or polypeptide in which results in the disruption of the ionic and covalent bonds and the Van der Waals interactions which exist in the molecule in its native or renaturnd state. Denatzruation of a protein can be accomplished, for example, by treatment with 8 M urea, reducing agents such as mercaptoethanol; heat, pH, temperature and other chemicals. Reagents such as 8 M urea disrupt both the hydrogen and hydrophobic bonds, and if mercaptoethanol is also added, the disulfide bridges (S-S) which are formed between cysteincs are reduced to two -S-H groups. Refolding of proteins which contain disulfide linkages in their native or refolded state may also involve the oxidation of the -S-H groups present on cysteine residues for the protein to reform the disulfide bonds.
The term "glycosaminoglyean" as used herein refers to polysaccharides containing alternating residues of uronic acid and hexosamine and usually contain sulfate. The binding of a protein in a refolding reaction as described herein to a glycosaminoglycan is through ionic interactions.
The term "dextran sulfate" as used herein refers a polyanionic derivative of dextran, ranging in molecular weight from 8,000 to 500,000 daltons. Dextrans are polymers of glucose in which glucose residues are joined by a1,6 linkages.
The term "heparin" as used herein refers to 2 glucoaminoglycans or heparinoids which are based on a repeating disaccharide (-4DGlcA(p)pl, 4GIcNAcal )e but are subject to extensive modification after assembly. Heparin is stored with histanvne in mast cell granules and is thus found in most connective tissues.
In general heparins have shorter chains than heparin.
The term '~C" as used herein refers to hydrophobic interaction chromatography which employs a hydrophobic interaction between the column and the molecule of interest to separate the sulfated polysaccharides and other contaminants from the refolded product.A
Negatively charged polymers include sulfated polysaccharides, such as heparins, dextran sulfates, and agaropectins, as weU as carboxylic acid polysaccharides such asalginic aads and carboxymethyl celluloses. Polyinorganics such as polyphosphates are WO 96/40'f84 PCTNS96/09980 also included. Polyamino acids such as polyasparatate, polyglutamate, and polyhistidine can also be used.
Positively charged polymers include polysaccharides such as DEAE dextran, polyorgnic amines, such as polyethyleneimines, polyethyleneimine celluloses, and polyamines, as well as the polyamino acids, polylysine and polyarginine.
Combinations of polymers may be used, of either charge polarity. In addition, amphoteric co-polymers may also_be used.
As used herein, "TFPP' refers to mature Tissue Factor Pathway Inhibitor. As noted above, TFPI is also known in the art as Lipoprotein Associated Coagulation Inhibitor (LACI), Extrinsic Pathway Inhibitor (EPI) and Tissue Factor Inhibitor ('TFI).
Muteins of TFPI which retain the biological activity of TFPI are encompassed in this definition. Further, TFPI which has been slightly modified for production in bacterial cells is encompassed in the definition as well. For example, a TFPI analog has an alanine residue at the amino-terminal end of the TFPI polypeptide has been produced in Escherichia coli. See U.S. 5,212,091.
As used herein, "pharmaceutically acceptable composition" refers to a composition that does not negate or reduce the biological activity of formulated TFPI, and that does not have any adverse biological effects when formulated TFPI is administered to a patient.
As used herein, "patient" encompasses human and veterinary patients.
As used herein, the term "solubilizer" refers to salts, ions, carbohydrates, amino acids and other organic molecules which, when present in solution, increase the solubility of TFPI above 0.2 mg/mL. Solubilizers may also raise the concentrations of TFPI above 1 mg/mL and above 10 mg/mL. It should be noted that solubilizers may act as stabilizing agents. Stabilizing agents preserve the unit activity of TFPI
in storage and may act by preventing formation of aggregates, or by preventing degradation of the TFPI molecule (e.g. by acid catalyzed reactions).
As used herein, the term "secondary solubilizers" refers to organic salts, ions, carbohydrates, amino acids and other organic molecules which, when present in solution with a solubilizer, further increase the solubility of TFPI. Secondary solubilizers may have other effects as well. For example, secondary stabilizers may be useful in adjusting tonicity (e.g. to isotonicity).
The amino acid sequence of TFPI is disclosed in U.S. Patent No. 5,106,833 and Figure 4. Muteins of TFPI and TFPI-2 are disclosed in U.S. Patent No.
S 6,103,500. As described in U.S. Patent No.6,103,500, muteins of TFPI and TFPI-2, with single or multiple point mutations, and chimeric molecules of TFPI and can be prepared. For instance, the lysine residue in the P1 site of the first Kunitz-type domain of TFPI may be replaced with arginine. Muteins, containing one to five amino acid substitutions, may be prepared by appropriate mutagenesis of the sequence of the recombinant cloning vehicle encoding TFPI or TFPI-2. Techniques for mutagenesis include, without limitation, site specific mutagenesis: Site specific mutagenesis can be carried out using any number of procedures known in the art.
These techniques are described by Smith (1985) Annual Review of Genetics, 19:423, and modifications of some of the techniques are described in METHODS IN
ENZYMOLOGY, 154, part E, (eds.) Wu and Grossman (1987), chapters 17, 18, 19, and 20. A preferred procedure when using site specific mutagenesis is a modification of the Gapped Duplex site directed mutagenesis method. The general procedure is described by Kramer, et al., in chapter 17 of the Methods in Enzymology, above.
Another technique for generating point mutations in a nucleic acid sequence is overlapping PCR. The procedure for using overlapping PCR to generate point mutations is described by Higuchi in Chapter,22 of PCR PROTOCOLS: A GUIDE
TO METHODS AND APPLICATIONS, (eds.) Innis, Gelfand, Sninsky and White (Academic Press, 1990).
Alternatively, hybrid proteins containing the first Kunitz-type domain of TFPI-2 and the second and third Kunitz-type domains of TFPI could be produced. One skilled in the art of DNA cloning in possession of the DNA encoding TFPI and TFPI-2 would be able to prepare suitable DNA molecules for production of such a chimeric protein using known cloning procedures. Alternatively, synthetic DNA molecules encoding part or all of each Kunitz-type domain and peptide sequences linking the Kurutz-type domains can be prepared. As a further alternative, the overlapping PCR
technique may also be used to prepare DNA encoding chimeric molecules containing TFPI and TFPI-2 sequences.
TFPI can be prepared in yeast expression systems as described in U.S. Patent No. 6,103,500. Methods have also been disclosed for purification of TFPI from yeast S cell culture medium, such as in Petersen et al., J.Biol.Chem. 18:13344-13351 (1993).
In these cases, recombinant TFPI is secreted from the yeast cell. TFPI
recovered in such protocols is also frequently heterogeneous due to N-terminal modification, proteolytic degradation, and variable glycosylation. Therefore, a need exists in the art to produce mature TFPI that is authentic (i.e. having the correct N-terminal amino acid sequence), full-length and homogeneous.
TFPI can be produced in E. coli as described in U.S. Patent No. 5,212,091 which discloses a method of producing TFPI by expression of a non-glycosylated form of TFPI in an E. coli host.
In one aspect of the invention recombinantly produced proteins which have the ability to bind polymers of sulfated polysaccharides such as, for example, heparin or dextran sulfate are refolded. The invention provides a method that facilitates refolding of a denatured recombinantly produced protein product using polymers of sulfated polysaccharides which act as a templates for the refolding protein.
Without being limited to any particular theory, the inventors believe that the interactions between the refolding protein and the polymeric template may minimize aggregation of the refolding intermediates and provide an environment for the protein to refold to its native conformation. The polymer acting as a template may bind a domain or region of protein to stabilize the intermediate and allow further folding to occur without aggregation. The protein aggregates, if formed, are generally less active than non-aggregated refolded protein, and generally result in a reduced overall yield of active refolded protein. The NaCI concentration of the refolding conditions is considered important and is selected to achieve the maximum efficiency of refolding by maximizing the interaction between the template polymer and the refolding protein. For example, it has been found by the inventors that approximately 0.2 M
concentration of NaCI or lower promotes binding of the C-terminal and/or the third Kunitz domain of TFPI to heparin or other sulfated polysaccharide polymer. The binding of polymer to the intermediate is presumed to facilitate the solubility of the intermediate and provide an environment for the rest of the protein to refold by reducing aggregation of the refolding intermediates.
General Methods TFPI may be prepared by recombinant methods as disclosed in U.S. 5,212,091.
Briefly, TFPI is expressed in Escherichia coil cells and the inclusion bodies containing TFPI are isolated from the rest of the cellular material. The inclusion bodies are subjected to sulfitolysis, purified using ion exchange chromatography, refolded by disulfide interchange reaction and the refolded, active TFPI
purified by cation exchange chromatography. TFPI may also be produced in yeast as disclosed in U.S. Patent No. 6,103,500.
TFPI activity may be tested by the prothrombin time assay (PTT assays).
Bioactivity of TFPI was measured by the prothrombin clotting time using a model RA4 Coag-A-MateTM from Organon Teknika Corporation (Oklahoma City, OK).
1 S TFPI samples were first diluted to 9 to 24 ug/mL with a TBSA buffer (50 mM
Tris, 100 mM NaCI, 1 mg/mL BSA, pH 7.5). Then 10 uL of Varify 1TM (pooled normal plasma from Organon Teknika Corp.) was mixed with 90 uL of diluted TFPI
samples in a sample tray and warmed to 37 C in the instrument. Finally SimplastinTM
Excel (Thromboplastin from Organon Teknika Corp.) was added to start the clotting.
The time delay in clotting due to anticoagulant activity of TFPI was measured and converted into TFPI concentration in the measured samples by comparison to a TFPI
standard curve.
The amount of soluble TFPI may also be quantified by measuring the area of the main peak on a cation exchange chromatogram. HPLC analysis of TFPI samples was performed using a Waters 626 LC system (Waters Corporation, Milford, MA) equipped with a Water 717 plus heater/cooler autosampler . Data acquisition was processed by a Turbochrom system from Perkin-Elmer.
The cation exchange (IEX) method used a Pharmacia Mono S HR S/5 glass column. The column was equilibrated in 80% buffer A (20 mM sodium acetate trihydrate:acetonitrile solution (70:3 0 v/v) at pH 5.4) and 20% buffer B (20 mM
sodium acetate trihydrate - 1.0 M ammonium chloride:acetonitrile solution (70:30 v/v) at pH 5.4). After a sample was injected, a gradient was applied to elute the TFPI
at a flow rate of 0.7 mLmin from 20% buffer B to 85% buffer B in 21 minutes. Eluting TFPI
species were detected by absorbance at 214 nm. The main peak (monomer TFPn was found to elute at about 18 minutes. Loss of soluble TFPI was quantified by integating remaining peak area of the main peak.
All reagents are U.S.P. or A.C.S. grade. Suppliers include J.T. Baker and Sigma Co. (St. Louis, MO).
The present invention will now be illustrated by reference to the following examples, which set forth certain embodiments. However, it should be noted that these embodiments are illustrative and are not to be construed as restricting the invention in any way.
EXAMPLES
xam~le 1 - Refolding Denatured TFPI
The following example describes the making of stock solutions, the HIC column preparation, the initial recovery and purification of TFPI prior to refolding, the refolding of TFPI, and the recovery of active TFPI.
The TFPI stock was prepared from refractile bodies resulting from the expression of recombinant TFPI in bacteria. The refractile bodies were solubilized at 10 mg/ml in 8 M urea, 50 mM Tris pH 8.5 containing 10 mM DTT, and this solution was clarified by centrifugation at 10,000 x g for 10 nunutes.
The column preparation for the initial purification of the solubilized TFPI
was prepared with S-Sepharose beads mixed in 7.5 M urea, 10 mM Tris and 10 mM
sodium phosphate (pH 6.5) containing 5 mM DTT and 1 mM EDTA. The solubilized TFPI at a concentration of 5 mg/ml was then run over the S-Sepharose column and eluted with a sodium chloride gradient of 0 to 1 M. The purified TFPI had an absorbency at wavelength 280 nm of 3.2 (which is equivalent to 4.l mg/ml using an extinction coe~cient of 0.78).
The dextran sulfate stock consisted of dextran sulfate of molecular weight daltons available from Sigma, item number D-4911, made up at 50 mg/ml (6.25 ~
in 50 mM Tris (pH 8.8) in 0.1 M sodium chloride, and stored at -20 degrees centigrade between uses.

The heparin stock, if heparin was used to conduct the refolding, was of molecular weight 6000 to 30,000 daltons, (with an average molecular weight of 18,000 daltons) prepared as a sodium salt available from Sigma Co. (St. Louis, MO), item number H-3393, made up at 60 mg/ml (3.33 mM) in 50 mM Tris (pH8.8) in 0.1 sodium chloride, and stored at -20°centigrade between uses.
To the S-Sepharose purified TFPI either dextran sulfate stock solution or heparin stock solution can be added. Dextran or heparin was added to TFPI
under denaturing conditions in 6 to 8 M urea. With 4°C reagents, the denaturing solution containing TFPI was diluted to 3 M urea, 50 mM Tris (pH 8.8), 0.2 M sodium chloride, and 0.5 mg/ml TFPI, and to a final dextran sulfate concentration of 0.6 mglml (75 MM) or a final heparin concentration of 1.5 mg/ml (83 p.M), depending on which was used to facilitate the refolding. Cystine was added to the refolding solution to a final concentration equal to the final DTT concentration. The refolding solution was incubated at 4°C with gentle agitation for from 4 to 6 days, preferably 5 days.
As an illustration of this procedure the following is a detail of a protocol for refolding a 5 ml solution of TFPI in dextran sulfate or heparin.
To 610 p1 of TFPI stock either 60 p,1 of dextran sulfate with 65 p1 of SO mM
Tris (pH 8.8) in 0.1 M NaCl, or 125 p1 of heparin stock solution with 50 mM
Tris (pH 8.8) or 0.1 M NaCI was added. The refolding solution was mixed and allowed to incubate 10 minutes on ice. Next, 4.2 ml of refolding buffer containing 2.5 M
urea, 50 mM Tris (pH 8.8) and 165 mM sodium chloride was added to the refolding solution and mixed. Finally, 61 ~.1 of 50 mM Cystine made up in 120 mM sodium hydroxide was added and the total solution was incubated at 4°C with gentle agitation for 4 days. The free sulfhydryl content was checked with Ellman's reagent (also called DTNB). Idoacetamide was added, to 20 mM, made up at 1 M in 100% ethanol for storage at .20°C.
The hydrophobic interaction column (HIC) was prepared from Butyl-650M
Tosohaas ToyopearlTM resin particle size 40-90, part # 014702. The butyl resin was washed in 3 M urea, 1 M ammonium sulfate, 50 mM Tris, 10 mM sodium phosphate, pH 6.5 and resuspended at a 50% slurry.

gyp g~~g4 PCTNS96109980 The refolding samples, stored at -20oC remained in the standard refolding buffer containing 3 M urea, 50 mM Tris, pH 8.8, 1-4 mM redox, 0.5 mg/ml TFPI, and 0.2-0.6 M NaCI depending on condition. Samples refolded with dextran or heparin had 0.2 M
salt, and samples without dextran or heparin had 0.6 M NaCI.
The following steps were performed at room temperature to effect the further purification of the refolded TFPI. To 300 p1 of refolded sample, an equal volume of 2 M ammonium sulfate, 3 M urea, 50 mM Tris, and 10 mM sodium phosphate (pH 6.5) was added. Next, 100 p1 of washed Butyl-650M beads was added to the diluted refolded sample. The solution with the beads was incubated with gentle rocking or mixing for 30 minutes at room temperature. The rnix was then spun in an ependorf centrifuge for 5 seconds, and put in a rack and allowed to sit for one minute for the beads to settle flat in the tube. The supernatant was aspirated carefully, so as not to disturb the beads.
To wash the TFPI-bound beads, 1 ml of wash buffer composed of 1 M ammonium sulfate, 3 M urea, 50 mM Tria, 10 mM sodium phosphate (pH 6.5) was added to the beads to remove the remaining dextran sulfate or heparin. The washed mixture was re-spun in an ependorf centrifuge for 5 seconds, and allowed to sit for one minute for the beads to settle as before. The supernatant was removed, and the beads then washed with the wash buffer a final time, and spun and allowed to sit as before.
After the final wash and settling, the supernatant was removed with a flame-pulled-tip Pasteur pipette very carefully.
To elute the refolded TFPI, 300 p! of elution buffer composed of 3 M urea, O.I
M
ammonium sulfate, 50 mM Tris and 10 mM sodium phosphate (pH 6.5) was added to the slurry of beads and rocked for more than 10 minutes. The beads were pelleted by spinning in an ependorf centrifuge, and the supernatant containing refolded TFPI was recovered. To avoid contamination of the beads with the product, some of the supernatant was left behind.

~acam~le 2 - HIC of Dextran Sulfate Refold The sample of TFPI was renatured at a concentration of 0.5 mg/ml TFPI, 0.6 mg/ml Dextran sulfate, 3.0 M Urea, 200 mM NaCI and 50 mM Tris (pH 5.5). The HIC
column was prepared from TosoHaas Butyl beads for HIC, 4.6 mmD/100mmL, in a 1.6b ml slurry. The flow rate was set for 1.0 mUmin. Before loading the HIC column, the sample was diluted 2:3 with 3.0 M Urea and 3.0 M NH,SO, at a final pH of 5.68;
2 ml of sample was loaded. The gradient start was 33 mM MES/33 mM HEPES/33 mM
sodium acetate, 1.0 M NH,SO,, and 3.0 M Urea, pH 6.0; the gradient end was 33 mM
MES/33 mM HEPES/33 mM sodium acetate, 3.0 M Urea at pH 6Ø The gadieat volume was 5.0 CV. From this column, the recovery of native TFPI was 68%. The results of this run are shown in Figure 2.
A second HIC column was also run. The sample of denatured TFPI was diluted 2:3 with 3.0 M Urea, 1.5 M NH,SO, and two mi were loaded. The gadient start was 33 mM MES/33 mM HEPES/33 mM sodium acetate, 0.5 M NH,SO,, and 3.0 M Urea, pH
6.0; the gradient end was 33 mM MES/33 mM HEPES/33 mM sodium acetate, 3.0M
Urea at pH 6Ø The gadient volume was 5.0 CV. The recovery of native TFPI
from this second column was 74%. The results of this run are shown in Figure 3.
The samples were analyzed by non-reducing SDS-PAGE as illustrated in Figure 1. Correctly refolded, active TFPI species (major band) are seen on the gel.
Euamo_le 3 About 10 mg/mL TFPI in 2M urea was dialyzed against one of the following: 20 mM acetate, 20 mM phosphate, 20 mM citrate, 20 mM glycine, 20 mM L-glutamate or 20 mM succinate in 150 mM NaCI as described above. 6-10 mg/mL TFPI bulk stock was loaded into SpecIPor 7 dialysis tubings (MW cutoff 3,500). Dialysis was carried out either at 41C or ambient temperature. Three changes of buffer at a protein solution to buffer ratio: 1 to 50-100, were made during course of dialysis over 12 to 24 hr time period. After dialysis, TFPI solution was filtered by Costa4r 0.22 micron filter units to separate precipitated TFPI from soluble TFPI. The solubility of TFPI
was then measured by UV/Vis absorbance assuming an absorptivity 0.68 (mg/mL)'' crti' at 278 nm. The solutions were prepared at various pH levels by titration with HCl or NaOH.

WO 96/40'f84 PCTNS96/09980 After completion of dialysis, the precipitates were filtered through 0.22 ~cm filter units. The concentration of remaining soluble TFPI after dialysis was measurod by W absorbance. Fgure 1 shows the results of these experiments. Solubility of TFPI increased greatly in solutions containing 20 mM acetate, ZO mM phosphate, mM L-glutamate and 20 mM succinate at pH levels below 7 and particularly at or below pH 4.5. Solubility of TFPI was also substantially increased in solutions containing 20 mM glycine above pH 10. Fgure 2 shows the solubility~of TFPI as a function of concentration of citrate ion in the presence of 10 mM Na phosphate at pH
7. TFPI solubility increases with increasing concentration of citrate. Figure 3 shows the solubility of TFPI as a function of concentration of NaCI at pH 7Ø TFPI
solubility increases with increasing salt concentration, indicating salt .
promotes solubility of TFPI.
The solubility of TFPI was studied using a number of different solubilizers and secondary solubilizers. Table 1 shows solubility of TFPI in varying buffer solutions measured by UV absorbance after dialyzing 6 to 10 mg/mL TFPI into these buffer solutions.

Table 1 Salt effect Solubilit Content- c m ml uv lOmM NaP04, 7 0.21 ~

lOmM NaP04, 150mM NaCI 7 0.72 20mM NaP04, 150mM NaCi 7 0.85 20mM NaP04, 0.5M NaCI 7 6.71 20mM NaP04, 1 M NaCI 7 8.24 H effect -Content _ - c m ml uv 20mM NaOAc, 150mM NaCI 3 10.27 20mM NaOAc, 150mM NaCI 3.510.25 20mM NaOAc, 150mM NaG 4 7.54 20mM NaOAc, 150mM NaCI 4.51 .75 20mM NaOAc, 150mM NaC) 5 1.15 20mM NaOAc, 150mM NaG 5.50.85 20mM NaP04, 150mM NaG 5.50.89 20mM NaP04, 150mM NaG 6 0.78 20mM NaP04, 150mM NaG 6.50.79 20mM NaP04, 150mM NaG 7 0.9 5 ' 20mM NaP04, 150mM NaG 7.5_ 0.82 20mM NaP04, 150mM NaCI 8 0.88 20mM NaCitrate, 150mM NaCI 4 2.17 20mM NaCitrate 150mM NaC) 4.51.19 20mM NaCitrate, 150mM NaCI . 5 1.1 20mM NaCitrate, 150mM NaG 5.51.84 20mM NaCitrate, 150mM NaCI 6 2.09 20mM NaCitrate, 150mM NaCI 6.52.12 20mM NaCitrate, 150mM NaG 7 1.92 20mM GI tine, 150mM NaCI 9 0.32 20mM Glycine, 150mM NaG 10 0.9 20mM GI tine, 150mM NaCI 1 13.94 20mM't.-Glutamate, 150mM NaCI 4 9.07 20mM l-Glutamate, 150mM NaCI 5 1 .21 20mM Succinate, 150mM NaCI ~ 8.62 20mM Succinate 150mM NaCI 5 1.21 20mM Succinate, 150mM NaCI 6 1.07 Citrate Content H c m ml uv lOmM NaP04, 20mM NaCitrate 7 1 .16 Tabl~ 1 (cont.) lOmM NaP04, 50mM NaCitrate 7 5.81 lOmM NaP04, 100mM NaCitrate 7 12.7 __ lOmM NaP04, 200mM NaCitrate 7 t 5.9 tOmM NaP04, 300mM NaCitrate 7 8.36 M 2+, Ca2+ and of hos hate Content H c m lml uv lOmM NaP04, 150mM NaCI, 1mM M Cl2 7 0.66 lOmM NaP04, 150mM NaCI, lOmM M CI2 7 1.02 lOmM NaP04, 150mM NaCI, 0.lmM CaCl2 7 0.67 lOmM NaP04, 150mM NaCI, 1 mM CaCl2 ~ 7 0.71 lOmM NaP04, 150mM NaCI, lOmM tri hos hate 7 3.64 lOmM NaP04, 5% PEG-400 I 0.07 t OmM NaP04, lOmM EDTA 7 0.36 lOmM NaP04, 100mM Na2S04 7 5.08 lOmM NaP04, 100mM L-as attic acid 7 0.4 lOmM NaP04, 100mM Succinic acid ' 7 2.33 lOmM NaP04, 100mM Tartaric acid 7 2.56 20mM NaP04, t00mM Malefic acid 7 0.11 20mM NaP04, t00mM Malic acid 7 1.87 lOmM NaP04, 100mM L- lutamic acid 7 0 _l OmM NaP04, 150mM NaCI I 7 0.25 lOmM NaP04, 100mM isocitrate 7 10.83 NaOAe, NaP04 and NaCI

Content c m ml uv lOmM NaOAc, 150mM NaCI , 4.51.76 lOmM NaOAC 4.54.89 IOmM NaOAc 5.54.95 t OmM NaOAc 6.5' S.1 lOmM NaOAc 7 5.87 1 OmM NaP04, 150mM NaCI ' 4.50.14 1 OmM NaP04 4.54.97 lOmM NaP04 5.50.79 lOmM NaP04 6.50.091 i 1 OmM NaP04 7 0.94 SOmM NaOAc 5 5.24 5mM NaOAc 5.54.59 t OmM NaOAc 5. 5.05 20rnM NaOAc 5.55.04 50mM NaOAc ' S.55.71 t00mM NaOAc 5.51.4 200mM NaOAc ( 5.51.32 WO 96/40'f84 PCT/US96/09980 Table 1 (corn. ) 5mM NaOAc. 5mM NaCI ~ 5.5! 4.85 5mM NaOAc, tOmM NaCI I ! 5.04 5.5 5mM NaOAc, SOmM NaCI 1 i 0.56 5.5 5mM NaOAc, 100mM NaCI 5.50.43 5mM NaOAc, 200mM NaCI 5.5! 0.8 5mM NaOAc 4.57.27 lOmM NaOAc ~ ! 6.5 4.5 20mM NaOAc 4.58.32 50mM NaOAc 4.59.17 5mM NaOAc ~ 8.9$
5.5 tOmM NaOAc 5.58.08 20mM NaOAc 1 8.99 5.5 50mM NaOAc ! 5.52.92 !

5mM NaOAc, 150mM. NaC) ~ 4.52.6 lOmM NaOAc, t50mM NaCt 4.52.59 20mM NaOAc, 150mM NaCI 4.52.55 50mM NaOAc, 150mM NaCI j 4.52.1 ;

5mM NaOAc, 150mM NaCI 5.50.65 lOmM NaOAc, 150mM NaCI ' S.50.69 I

20mM NaOAc, t50mM NaCI __ j 5.50.7~t '' 50mM NaOAc, 150mM Nad ~ S.50.91 H dro hobic chain len th Content i c m ml w tOmM NaP04. 50mM Formic acid 7 0.12 tOmM NaP04, 50mM Acetic cdd 7 0.16 t OmM NaP04, 50mM Pro anoic acid 7 0.16 tOmM NaP04, SomM Butanoic acid 7 0.13 ~

lOmM NaP04, SOmM Pentanoic acid 7 0.14 tOmM NaP04, 50mM Hexanoic acid 7 0.11 i Others Convent ~ t c m ml w 20mM NaOAc, 3~ Mannitol. 2~ Sucrose, 5'Ye4 19.9 PEG-400 !

20mM Na Citrate. 3% Mannitol, 2% Sucrose.6.50.72 5~6 PEG-400 20mM Na Citrate, 150mM NaCI, 5% PEG-400 6,52.18 20mM NaOAc, t50mM NaCI, 5% PEG-400 4 t 9.8 20mM Na Citrate, 130mM NaCt, t% GI tine, 6.51.48 0.25% Tween-80, 5% PEG-400 ' 20mM Na Citrate, 130mM NaCI, t /. GI ine,6.51.32 0.25% Tween-80 I

Solubitit Content I H a m Imi ~ ,uv 5mM NaAcetate ! 5.58.9 I

5mM NaAcetate. 8% Sucrose ! 5.5 5mM NaAcetate, 0.01% Polvsorbate-80 ! 5.57 !

Tabl~ 1 (cost.) 5mM NaAcetate, 8% Sucrose, 0.01 /. Polysorbate-805.51 2 lOmM NaAcetate _5.57.6 _ _ 5.51 0 lOmM NaAcetate, 8% Sucrose lOmM NaAcetate, 8% Sucrose, 0.01% Pol 5.512.1 sorbate-80 SmM NaAcetate, 5% Sorbitol 5.57.8 5mM NaAcetate, 4.5% Mannitol 5.59.2 5mM Histidine 6 5.5 5mM Histidine 6.51 5mM NaCitrate 5.50.1 5mM NaCitrate 6 0.1 5mM NaCitrate 6.50.1 5mM NaSuccinate - 5.50.6 5mM NaSuccinate 6 0.3 5mM NaSuccinate 6.50.2 lOmM Imidazole 6.52.5, 10.8 lOmM Imidazole 7 0.8 lOmM Imidazole, 8% Sucrose 6.512.2 5mM NaAcetate 6 8.2 lOmM Imidazde, 5mM NaAcetate 6.512.8 t OmM NaCitrate 6 0.2 t OOmM NaCitrate 6 8.1 t OOmM NaCitrate 7 9.3 _ 6 9.1 lOmM Naphosphate, 260mM Na2S04 tOmM NaPhosphate, 100mM NaCitrate 8 8.8 lOmM NaCitrate, 1 % L- lutamic acid 6 4.6 tOmM NaCitrate 2~ L-I sine 6 1.1 tOmM NaCitrate, 0.5% L-as attic acid 6 0.4 1 OmM NaCitrate, 0.1 /. Phos hate lass 7 5.9 tOmM Tris, t00mM NaCitrate 8 8.5 tOmM NaCitrate, 1M GI cine 6 0.3 lOmM NaCitrate, 300mM G) cite 6 0.3 tOmM NaCitrate, 280mM GI carol 6 0.3 lOmM NaCitrate, 0.5M NH4 2504 6 8.3 lOmM NaCitrate, 120mM NH4 2S04 6 8.8 t OmM NaCitrate, 260mM Na2S04 6 9.4 t OmM NaP04, 0.1 ~ Phos hate lass 7 15.8 lOmM NaCitrate, 0.1 % SDS 6 1 1.2 lOmM NaCitrate, 0.02% SDS 6 7.8 tOmM NaAcetate, 8% PEG-400 5.513.7 lOmM NaAcetate,. lSOrpM NaCI, 8% PEG-400 5.50.6 lOmM NaAcetate, 8% PEG-400 6 16.2 ~

tOmM NaCitrate. 8% PEG-400 ' 6 0.2 WO 96140"!84 PCTNS96/09980 The stability of TFPI stored at various pH conditions was tested. TFPI was prepared by dialysis as above in 10 mM Na phosphate, 150 mM NaCI and 0.005 96 (w/v) polysorbade-80. Stability samples containing 150 mg/mL TFPI were incubated at 401C for 20 days. Kinetic rate constant for the remaining soluble TFPI was analyzed by following decrease of the main peak on ration exchange chromatograms.
As can be seen in Figure 5, the decay rate constant increases at pH above 6.0, indicates more aggregation at higher pH conditions.
TFPI was also formulated at a concentration of 150 mg/mL in 150 mM NaCI
and 0.005 % (w/v) polysorbate-80 at pH 7 with varying concentrations of phosphate.
Figure SA shows the percentage of remaining soluble TFPI measured by the ration exchange HPLC. Increasing concentrations of phosphate ion in solution resulted in higher levels of soluble TFPI remaining after incubation at 40°C.
Higher levels of phosphate ion also resulted in higher levels of active TFPI as assayed by the prothrombin time assay. These results are shown in Figure SB.
Stability of TFPI at a concentration of 0.5 mg/mL and formulated in 10 mM
Na citrate, pH 6 and 150 mM NaCI was also tested at 40°C over a 40 day period.
As seen in Figure 6, ration-exchange HPLC (triangle) shows the presence of soluble TFPI at levels greater than 60 % initial, even after the 40 day incubation. In like manner, the prothrombin time assay (circle) shows the presence of active TFPI
at levels greater than 6096 initial, even after the 40 day incubation.
Figure 7 shows loss of soluble TFPI at 40°C measured by both cation-exchange HPLC (open symbol) and prothrombin time assay (closed symbol) for 0.5 mg/mL TFPI formulated in 10 mM Na phosphate, pH 6 and either 150 mM NaCI
(triangle) or 500 mM NaCI (circle).
Figure 8 shows loss of soluble TFPI at 40 ° C measured by both cation-exchange HPLC (open symbol) and prothrombin time assay (closed symbol) for 0.5 mg/mL TFPI formulated in 10 mM Na acetate and pH 5.5 containing 150 mM NaCI
(triangle) or 896 (w/v) sucrose (square) or 4.596 (w/v) mannitol (circle).

Figure 9 shows two non-reducing SDS gels for TFPI formulation samples in mM NaPO" 150 mM NaCI, and 0.005 % polysorbate-80 at pH 4 to pH 9 stored at 40°C for 0 days power) and 20 days (upper). No loss on TFPI is seen at 0 days.
However, at 20 days cleavage fragments of TFPI may be seen at the lower pH
range 5 (i.e. pH 4 and pH 5). Without being bound to a particular theory, it is believed that these fragments may result from ~an acid catalyzed reaction.
Finally, Table 2 shows the half life of remaining soluble TFPI. at 40°C for various formulations. 0.5 mg/mL TFPI was formulated in these formulation conditions and incubated at 40°C. Samples were withdrawn at predetermined time 10 intervals and loss of soluble and active TFPI were examined by the IEX-HPLC
and the PT assay. Half life for remaining soluble TFPI was then calculated by performing a single exponential fitting to the IFX-HPLG and PT assay results.
Elution of TFPI in displacement mode from chromatography resins using polyionic compounds.
TFPI is first bound to a resin in a low salt buffer. Next a buffer containing the polyionic compound used to elute TFPI in displacement mode, is pumpod through the column. This compound binds stronger to the resin than TFPI and displacxs TFPI. For a positively charged resin (anion exchanger) a negatively charged compound is used and for a negatively charged resin (ration exchanger) a positively chargod compound is usod.
Partially purified TFPI was used as starting material. TFPI; in 6 M urea, 20 mM Tris, pH 8.0 was loaded onto a column packed with an anion exchange resin, Q
Sepharose HP, to 20 mg/mL resin. After loading, the column was washed with 6 M
urea, ~ 20 mM Tris, pH 9Ø TFPI was eluted and 10 mg/m1 of Glass H
(polyphosphate) in 6 M urea, 10 mM Tris, pH 9Ø
Elutiarr of TFPI from chromatography resin in aqueous buffer using polyionic compounds.

WO 96/40784 PCT/US96/099>!~

Tabl~ 2 t1/2 da at 40C

0.5 m ml TFPI formulated in: iDC~IP~C PT assa ~

rnM Na Acetate, 150 mM NaCI, 10.8 17.2 H 5.5 10 mM Na Citrate, 150 mM NaCI, 12.2 24.4 H 5.5 10 mM Na Acetate, 8% w/v Sucrose,43.2 42.2 H 5.5 10 mM Na Acetate, 4.5% Mannitol, 47.7 46.6 H 5.5 10 mM Na Succinate, 150 mM NaCI, 7.8 11.0 pH 6.0 10 mM Na Citrate, 150 mM NaCI-pH.,6.013.0 18.8 10 mM Na Phos hate, 150 mM NaCI, 7.8 11.2 H 6.0 10 mM Na Phos hate, 500 mM NaCi, 52.2 68.9 H 6.0 10 mM Na Citrate i 50 mM NaCI, 10.0 14.8 H 6.5 For a positively charged resin, a positively charged compound is used and for a negatively charged resin, a negatively charged compound is used.
TFPI, in 3.5. M urea, 1 mg/ml polyphosphate, 50 mM Tris, pH 5.9 was loaded onto a cation exchange resin, SP Sepharose HP. After loading the column was washed with a non-urea containing buffer, 10 mg/ml polyphosphate, 10 mM sodium phosphate, pH 5Ø TFPI was eluted in the same buffer at pH 7.5, without urea.
Example 7 Selective elution of TFPI from ion exchange resins using polyionic compounds.
Because of the charged ends of TFPI, oppositely charged polyionic compounds can bind to these ends. When the polyionic compound has a higher strength of binding to TFPI than does the resin, the TFPI may be selectively eluted from the chromatography resin.
TFPI, in 3.5. M urea, 1 mg/ml polyphosphate, 50 mM Tris, pH 5.9 was loaded onto a cation exchange resin, SP Sepharose HP. After loading, the column is washed with 6 M urea, lmg/ml polyphosphate, 10 mM sodium phosphate, pH 5.9. TFPI was eluted in a 25 column volume gradient up to 20 mg/ml of polyphosphate. TFPI
starts to elute at about 2-3 mg/ml of polyphosphate.
Example 8 Neutralization of polyionic compounds prior to chromatographic separation of TFPI. TFPI can interact with charged polymers. This interaction may prevent binding and purification to chromatographic resins. By neutralizing the charged polymer with an oppositely charged polymer, TFPI may bind to the resin.
In a buffer containing polyphosphate (Glass H), TFPI does not bind to Express Ion STM (Whatman) and no purification is achieved. By mixing PEI into the column load, TFPI now binds to the resin and TFPI can be purified.

WO 96/d0984 PCTNS96/09980 Refolding and purification of recombinant human TFPI (rhTFPT) using Polyphosphate (Glass H) Facilitated Refolding Process.
Inclusion bodies containing about 40 g of rhTFPI were thawed by removing the containers from the -ZO°C freezer and incubating them in a cold room at 4-10°C
for approximately 196 hours. The thawed inclusion bodies were then dispersed with a high shear mixer to reduce the clumping that occurs during freezing. The thawed inclusion bodies were added to 80 L of 3 M urea, 50 mM Tris-Cl, pH 10.5 buffer containing 2 g/L Glass H contained in a 100 L polyethylene tank equipped with an overhead stirra. The contents were mixed for approximately 15 minutes, and then the absorbance of the solution is measured at 280 nm. If the absorbance is greater than the mixture was diluted with sufficient dissolution buffer to obtain an absorbanee at 280 nm of 1.0-1.1. The solution was incubated with gentle agitation for 15-minutes, and then sufficient cysteine was added to give a cysteine concxntration of 0.1 mM. The solid L-cysteine was dissolved in approximately 50 ml of purified water and added to the refold mixture. The pH was checked and adjusted to pH
10.2 if necessary. The refold mixture was incubated with gentle agitation for 96-hours.
After approximately 96 h, the refolding process was terminated by adjusting the pH of the refold mixture to pH 5.9 using glacial acetic acid. Stirring was continued for 90 minutes and the pH checked. More acid was added, if necessary to adjust the pH to 5.9 +/- 0.1. A two-step filtration process was used to remove the particulates that formed during previous steps and prepare the acidified refold mixture for SP-Sepharose HP chromatography. First the acidif ed refold mixture is passed through A Cuno 60LP depth filter (filter housing model 8ZP1P) using a peristaltic .
pump ( ~ -'/s inch inner diameter silicon tubing).
The filter system was washed with 8-10 L of deionized 6 M urea before use.
The filtrate was collected in a 100 L polyethylene tank. Back pressure was maintainer) at a constant ZO PSI. Initial flow rate for a new filter was approximately 3.0 5-6 L per minute. Filters were replacxd when the flow rate dropped below 1 L per minute in order to maintain the back pressure at 20 PSI. The second stage. of the filtration used a 0.45 micmn f lter cartridge (Sartorius Sartobran pH or equivalent) with a peristaltic pumping system. After filtration, the pH was checked, and adjusted to pH 5.9 if necessary.
The acidified, filtered refold was loaded onto the equilibrated SP Sepharose HP column at a flov~i rate of approximately 80.0 ml/min. Flow rate was adjusted to accommodate overnight loading of the acidified filtered refold mixture. The column was equilibrated in 6 M urea, 20 mM sodium phosphate buffer pH 5.9 prior to loading. After loading, the column is washed with 2 CF of 6 M urea, 0.3 M
NaCI, 20 mM sodium phosphate buffer, pH 5.9 prior to the gradient elution step. The column flow rate was increased to 190-200 ml/min for the wash step and all subsequent steps (linear velocity =- 47 cm/hr). The product was eluted from the column using a linear salt gradient from 0.3 to 0.5 M NaCI in 6 M urea, 20 mM
sodium phosphate buffer, pH 5.9. The gradient was formed by delivering 6 M
urea, 0.5 M NaCI, 20 mM sodium phosphate buffer into 6 M urea, 0.3 M NaCI, 20 mM
sodium phosphate buffer. Limit buffer was pumped with a Masiertlex pump (model 7553-20) with a Masterflex head (model 7015.21) at a flow rate of approximately 100 ml/min. with vigorous mixing using a Paratrol A mixer from Parametrics (model 250210). The total volume of the gradient was 71.0 liters or 13.0 CV. The pH
of the gradient buffers was 5.92 (+/- .02). Fractions arc evaluated qualitatively using SDS PAGE and pooled based on the content of the correctly folded SC-59735 relative to other misfolds and impurities. After pooling the process stream is referred to as the S pool.
The pH of the S pool was next adjusted to pH 8.0 with 2.5 N NaOH. The S
pool was concxntrated 2-3 fold to approximately 2 L using an Amicon DC-lOL
ultrafiltration unit containing an Amicon YM10 spiral cartridge (10,000 M.W.
cut-off membrane). After concentration, the concentrated S pool was diafiltered against 7 volumes of 6 M urea, 20 mM Tris-HCl buffer, pH 8Ø The diafiltration was considered complete when the conductivity of the reten~te was below 2 mS. The diafiltered concentrate was drained from the ultrafiltration unit and the unit was WO 96J40'!8/ PCTNS96/09980 washed with approximately 1 L of diafiltration buffer. The was is combined with the concxntrate to form the Q-load.
An Amicon column (7.0 cm diameter) was packed with approximately 700 ml of Q-Sepharose high performance medium (Pharmacia Q-Sepharose HP). The column was packed with 20% ethanol at 20 psi. The bed height~after packing was approximately I8 cm. The column was equilibrated with 5 CF of 6 M urea, O.OZ M
Tris/HCl buffer, pH 8. The target for protein loading is 8-10 mg protein/ml Q
Sepharose resin. The Q load was applied to the column at a flow rate 30-35 ml/min (50 cm/hr). After loading, the column was washed with approximately 5 CV of 6 M
urea, 20 mM Tris/HCl buffer, pH 8.0, or until the absorbance at 280 nm returned to baseline. The product was eluted using a sodium chloride gradient from 0-0.15 M
NaCl in 6 M urea, 20 mM Tris/HCl buffer, pH 8.0 over 25 column volumes. The first seven column volumes were collected as a single fraction, followed by 30 fractions of 0.25 column volume each.
Fractions are routinely analyze by reducing and non-reducing SDS-PAGE
and size exclusion chromatography. Fractions are pooled based on aggregate content ( < 5 % by SEC HPLC Method MSL 13929) and qualitative evaluation by SDS PAGE
to assess purity. The fractions are stored frozen at -20°C until pooled.
Acceptable Q Sepharose fractions were pooled, and the pH of the pool was adjusted to 7.2 using 2 M HCI. The pool was then concentrated, approximately 5 fold in an Amicon DC-1 ultrafiltration system containing a S 1 Yl Amicon YM-10 cartridge (10,000 MWCO spiral cartridge membrane). The concentrated Q Pool was then diafiltered against seven column volumes of 2 M urea, 0.15 M NaCI, 20 mM
sodium phosphate buffer, pH 7.2. Following ultrafiltration, the solution was drained from the ultrafiltration system. Approximately I00 ml of 2 M urea, O.15 M
NaCI, 20 mM sodium phosphate buffer, pH 7.2 was circulated through the ultrafiltrati~
system for approximately 5 min. The rinse solution was combined with the original concentrate and the solution was filtered through a 0.45 micron vacuum filter unit (Nalgaie).

Refolding and Purification of rhTFPI using Polyethyleneimine (PF.1) Facilitated Refolding Process.
Inclusion bodies containing about 40 g of rhTFPI were thawed by removing the containers from the -20°C freezer and incubating them in a cold room at 4-10°C
for approximately 96 hours. The thawed inclusion bodies were then dispersed with a high shear mixer to reduce the clumping that occurs during freezing. The inclusion body slurry was vigorously blended for approximately 1 minute using a polytron homogenizer (Brinkman model PT45/80) or until the inclusion bodies were then added to 40 L of 6 M urea 100 mM Tris/HCl buffer pH 9.8 containing 300 mM
NaCI and 0.4 g/L PEI contained in a 100 L polyethylene tank equipped with an overhead stirrer. The mixture was vigorously stirred for 20-30 min. The pH was monitored and adjusted to pH 9.8 as necessary. The absorbance of the dissolved inclusion body mixture was measured at 280 nm, and if the absorbance was greater than 2.1, the sample was diluted with 10 liters of the dissolution buffer described above to obtain an A280 value of 2.0-2.1. Gentle agitation was continued for another 15-30 minutes. Next, the dissolved inclusion body solution was diluted with an equal volume of 1.0 M urea, 300 mM NaCI solution. Finally, L-cysteine was added to give a final concentration of 0.25 mM. The solid L-cysteine was dissolved in 50 ml ZO of WFI and added as a solution to the diluted refold. The pH was checked and adjusted, if necessary. The refold continued with gentle mixing for 96-120 hours with periodic checks of the pH, and adjustment to pH 9.8, if necessary: The progress of the refold was monitored by Mon-S ration exchange and prothrombin time assays.
After approximately 96 h, the refolding process was terminated by adjusting the pH of the refold to pH 5.9 using glacial acxtic acid. Stirring was continued for 90 minutes and the pH checked. More acid was added, if necessary to adjust the pH
to 5.9 /- 0.1.
A two-step filtration process was used to remove the particulates that formed during previous steps and prepare the acidified refold for SP-Sepharose HP
chromatograph. First, the acidified refold is passed through a Cuno 60LP depth filter (filter housing model 8ZP1P) using a peristaltic pump (1/4 - 3/8 inch inner diameter silicon tubing).
The filter system was washed with 8-10 L of deionized 6 M urea before use.
The filtrate was collected in a 100 L polyethylene tank. Back pressure was maintained at a constant 20 PSI. Initial flow rate for a new filter was approximately 5-6 L per minute. Filters were replaced when the flow rate dropped below 1 L
per minute in order to maintain the back pressure at 20 PSI. The second stage of the filtration used a 0.45 micron filter cartridge (Sartorius Sartobran pH or equivalent) with a peristaltic pumping system. After filtration, the pH was checked, and adjusted to pH 5.9, if necessary.
The acidified, filtered refold was loaded onto the equilibrated SP Sepharose HP column at a flow rate of approximately 80.0 ml/mm. Flow rate was adjusted to accommodate overnight loading of the acidified filtered refold. The column was then washed with 5.5 column volumes of 6 M urea, 0.3 M NaCI, 20 mM sodium phosphate buffer, pH 5.9. The column flow rate was increased to 190-200 ml/mm for the wash step and all subsequent steps (linear velocity = ~ 47 cm/hr). The product was eluted from the column using a linear salt gradient from 0.3 to 0.5 M NaCl in 6 M
urea, 20 mM sodium phosphate buffer, pH 5.9. The gradient was formed by delivering 6 M
urea, 0.5 M NaCl, 20 mM sodium phosphate buffer into 6 M urea, 0.3 M NaCl, 20 20~ mM sodium phosphate buffer into 6 M urea, 0.3 M NaCI, 20 mM sodium phosphate buffer. Limit buffer was pumped with a MasterflexTM pump (model 7553-20) with a MasterflexTM head (model 7015.21) at a flow rate of approximately 100 mi/mm with vigorous mixing using a ParatrolTM A mixer from Parametrics (model 250210).
The total volume of the gradient was 71.0 liters or 13.0 CV. The pH of the gradient buffers was 5.92 (+/-.02).
Fraction collection was started when the column inlet conductivity reached 28.0 - 28.5 mS/cm as measured by the in-line Radiometer conductivity meter.
Forty 500 ml fractions (0.1 CV) were collected. A Pharmacia Frac-300 fraction collector was used with numbered, 500 ml polypropylene bottles. When the fraction collection was stopped, the remainder of the gradient was collected as a pool.

Column fractions were assayed by A280, size exclusion HPLC, and in addition, for informational purposes, SDS PAGE, reverse phase HPLC, and PT
assays. Fractions were pooled if they met the pooling criteria of containing 20% of less aggregate as determined by the in process SEC HPLC. Pooled SP Sephamse fractions are referred to as the S Pool.
The pH of the S-pool was next adjusted to pH 8.0 with 2.5 N NaOH. 'The S
Pool was concentrated 2-3 fold to approximately 2 L using an Amicon DC-lOL
ultrafiltration unit containing an Amicon YM10 spiral cartridge (10,000 N.W.
cut-off membrane). After concentration, the concentrated S Pool was diafiltered against 7 volumes of 6 M urea, 20 mM Tris-HCl buffer, pH 8Ø The diafiltration was considered complete when the conductivity of the retentate was below 2 mS. The diafiltered concentrate was drained from the ultrafiltration unit and the unit was washed with approximately 1 L of diafiltration buffer. The was is combined with the concentrate to form the Q-load.
An Amicon column (7.0o cm diameter) was packed with approximately 700 ml of Q-Sepharose high performance medium (Pharmacia Q-Sepharose HP). The column was packed in 2096 ethanol at 20 psi. The bed height after packing was approximately 18 c.m The column was equilibrated with 5 CV of 6 M urea, 0.02 M
Tris/HCl buffer, pH 8. .The target for protein loading is 8-10 mg protein/ml Q
Sepharose resin. The Q load was applied to the column at a flow rate 30-35 ml/min (50 cn~/hr). After loading, the column was washed with approximately 5 CV of 6 M
urea, 20 mM Tris/HCl buffer, pH 8.0, or until the absorbance at 280 nm returned to baseline. The product was eluted using a sodium chloride gradient from 0-0.15 M
NaCI in 6 M urea, 20 mM Tris/HCl buffer, pH 8.0 over 25 column volumes. The first seven column volumes were collected as a single fraction, followed by 30 fractions of 0.25 column volume each.
Fractions are routinely analyzed by reducing and non-reducing SDS-PAGE
and size exclusion chromatography. Fractions are pooled based on aggregate content (596 by SEC HPLC) and qualitative evaluation by SDS PAGE to assess purity. The fractions are stored frozen at -20°C until pooled.

The Q-Sepharose fractions to be pooled were thawed by incubation at 2-8°C, pooled, and the pH of the pool was adjusted to 7.2 using 2 MHCI. The pool was then concentrated approximately 5 fold in an Amicon DC-1 ultrafiltration system containing a S1Y1 Amicon YM-10 cartridge (10,000 MWCO spiral cartridge membrane). The concentrated Q Pool was then diafiltered against seven column volumes of 2 M urea, 0.15 M NaCI, 20 mM sodium phosphate buffer, pH 7.2.
Following ultrafiltration; the solution was drained from the ultrafiltration system.
Approximately 100 ml of 2 M urea, 0.15 M NaCI, 20 mM sodium phosphate buffer, pH 7.2 was circulated through the ultrafiltration system for approximately 5 min.
The rinse solution was combined with the original concentrate and filtered through a 0.45 micron vacuum filter unit (Nalgene).
Solubilization, refolding, and purification of rhTFPI from inclusion bodies using polyphosphate in the absence of chaotropes such as urea (GDS 5327089,92) About 2 g of rhTFPI (43 ml inclusion body slurry containing 46 mg/ml rhTFPT) was dissolved with mixing in 4 L of 50 mM Tris buffer, pH 10.5 containing 4 g/1 polyphosphate (Glass H, FMC Corporation) 2-8°C. Sufficient cysteine and cystine was added to make the solutions 0.1 mM and 0.05 mM respectively. The pH
was maintained at pH 10.5 with 1 N NaOH. The refold solution was incubated at 8 ° C with gentle mixing for 72-96 h.
The refold was next adjusted to pH 6 using glacial acetic acid and then filtered thmugh a 0.2 micron filter. An aliquot of the filtered refold was applied to a 200 ml column of SP-Sepharose HP (Pharmacia) previously equilibrated in 0.4 % Glass H, 20 mM sodium phosphate pH 6 buffer after loading, the column was washed with 4 column volumes of 0.4 % Glass H, 20 mM sodium phosphate pH6 buffer. The column was eluted using a linear pH gradient from 0.4 % Glass H, 20 mM sodium phosphate buffer pH 6 to 0.4 '% Glass H, 50 mM Tris pH 8 buffer. Fractions were .
collected and analyzed by SDS PAGE. Relatively pure rhTFPI could be refolded and purified in this manner.

Improved solubility of rhTFPI in water by formation of a complex between TFPI and polyphosphate (GDS 5327046-4'n About 10 g of purified rhTFPI in about 1 liter of 2 M urea, 125 mM sodium chloride, 20 mM sodium phosphate pH 7.4 buffer was thaws by incubation at 2-8°C
for 18-36 h. Sufficient dry urea was added to make the solution 6 M in urea.
The solution was then filtered through a 0.2 micron filter. Five g of polyphosphate glass (Glass H, FMC) was dissolved in 50 ml of 6 M urea, adjusted to pH 7 with 1 N
NaOH, and added to the protein solution. The solution was then concentrated by ultrafiltration using 1 square foot of membrane (Amicon S 1 Y3) to about 400 ml (-25 mg/ml) and diafiltered against 10 volumes (about 4 liters) of purified water to remove residual urea. After diafiltration, the solution was concentrated to about 250 ml and removed from the ultrafiltration unit. The ultrafiltration unit was washed with about 150 ml of purified water and the was addod to the protein oonantrate. The final protein concentrate contained almost 10 g of protein in 400 ml of water (about rng/ml~protein). The normal solubility of rhTFPI in water is less than 0.5 mg/ml.
Use of cationic polymers for removal of E. coli contaminants from TFPI call lysatcs and rcfractile bodies.
The use of cationic polymers to precipitate and remove E coli contaminants from crude TFPI intermediates (lysates, refractile bodies) can significantly improve subsequence process operations (refolding, chromatography etc.) A random screening of cationic polymers identified candidates which selectively precipitate bacterial contaminants while TFPI remains in solution. Specifically, Betz polymer 624 precipitated substantial amounts of bacterial contaminants, while leaving TFPI in solution in an aqueous environment.
Solubilized TFPI refractile bodies (in 3.5 M guanidine hydrochloride, 2 M
sodium chloride, SO mM TRIS, 50 mM dithiothreitol, pH7.1) was the starting material used for a polymer screening experiment. This material was diluted 10 fold into a 0.5 % solution of various polymers. The precipitates from this experiment WO 96/40'f84 PCTNS96/09980 were analyzed by SDS-PAGE for the presence of TFPI. Betz polymer 624 precipitated substantial amounts of contaminants, no TFPI, and resulted in a clear aqueous solution.
The use of aqueous two phase extraction with a polyethylene glycol (PEG), polyphosphate, urea system offers processing advantages for TFPI purification.
Typical aqueous two phase systems consist of two polymer systems (e.g., PEG
aad dextran) or a polymer and salt (e.g., PEG and sulfate). The system described here has advantages in that the polyphosphate chain length can be optimized for the separation, is inexpensive and is specific in removing problematic contaminants from TFPI refractile bodies known to interfere with refolding and chromatography (native polyphosphate and associated divalent metals).
TFPI refractile bodies were solubilized in 7 M urea, 10 mM CAPS, 196 monothioglycerol pHlO. Polyphosphate and PEG. of different chain lengths were added to form two phases. Upon phases separation, the TFPI partitioned into the PEG rich upper phase, leaving the polyphosphates and associated contaminants in the lower phase. Separation is effected by both PEG and polyphosphate chain length and can be optimized by varying both.
Charged polymer facilitated refolding of recombinant tissue plasminogen activator (t-PA) from E, coli inclusion bodies Five grams (wet weight) of inclusion bodies containing about 2 grams of recombinant tissue plasminogtn activator are added to about 1 liter of 0.5 %
Glass H, 50 mM Tris buffer pH 10.8 containing 1 mM reduced glutathione (GSH) and 0.2 mM glutthione disulfide (GSSG). The mixture is thoroughly blended using a polytron (Brinlanan) homogenizer for 2-3 minutes to thoroughly disperse the inclusion bodies. The mixture is incubated with mixing using an overhead stirrer for 15 minutes while the pH is maintained at 10.5-10.9 using 1 N NaOH. The mixture is then gently mixed for 48-72 hours at 2-8oC.
F~~ample..lb Charged polymer facilitated refolding of bovine somatotropin from E. coli inclusion bodies Ten grams (wet weight) of inclusion bodies containing 5 grams of bovine somatotropin are added to about 1 liter of 1% Glass H, 50 mM Tris buffer pH
10.5.
The mixture is thoroghly blended using a polytron (Brinktnan) homogenizer for minutes to thoroughly disperse the inclusion bodies. The mixture is incubated with mixing using an overhead stirrer for 15 minutes while the pH is maintained at 10.4-10.6 using 1 N NaOH. Solid cysteine (121 mg) is added to make the reaction 1 mM
cysteine, and the refolding reaction is mixed for 48-72 hours The present invention has been described with reference to specific embodiments. However, this application is intended to cover those changes and substitutions which may be made by those skilled in the art without departing from the spirit and the scope of the appended claims.

Claims (40)

1. A solution having a pH of from 5 to 10, comprising from 200 mM arginine to 300 mM arginine, and further comprising a polypeptide selected from the group consisting of (i) human TFPI having the amino acid sequence shown in Figure 4, (ii) ala-human TFPI having the amino acid sequence shown in Figure 4 and having one further amino acid which is an amino-terminal alanine, and (iii) muteins of (i) or (ii) having from 1 to 5 amino acid substitutions.

2. The solution of claim 1 wherein said arginine is L-arginine.

3. The solution of claim 1 comprising from 0.2 to 10 mg/ml of said polypeptide.

4. The solution of claim 2 wherein the polypeptide is ala-human TFPI having the amino acid sequence shown in Figure 4 and having one further amino acid which is an amino-terminal alanine.

5. The solution of claim 1 comprising 300 mM arginine.

6. The solution of claim 4 having a citrate/citric acid buffer at a total buffer concentration of from 5 mM to 300 mM.

7. The solution of claim 1 having a pH of 5.5 and comprising ala-human TFPI
having the amino acid sequence shown in Figure 4 and having one further amino acid which is an amino-terminal alanine, 300 mM L-arginine, and a citrate/citric acid buffer at a total buffer concentration of 20 mM.

8. A solution having a pH of from 5 to 10, comprising more than 0.2 mg/ml of a polypeptide selected from the group consisting of (i) human TFPI having the amino acid sequence shown in Figure 4, (ii) ala-human TFPI having the amino acid sequence shown in Figure 4 and having one further amino acid which is an amino-terminal alanine, and (iii) muteins of (i) or (ii) having from 1 to 5 amino acid substitutions, and further comprising a solubilizer selected from the group consisting of sucrose, mannitol, sorbitol, citrate, isocitrate, succinate, malate, polyphosphate, sodium phosphate, sodium sulfate, acetate, polysorbate-80, polyethylene glycol, histidine, imidazole, glutamate, glycine, ammonium sulfate, and sodium dodecyl sulfate.

9. The solution of claim 8 wherein the polypeptide is present at a concentration of from 1 to 20 mg/ml.

10. The solution of claim 9 wherein the polypeptide is present at a concentration of from 1 to 10 mg/ml.

11. The solution of claim 8 further comprising sodium phosphate at a concentration greater than 20mM.

12. The solution of claim 8 wherein the solution comprises 0.5M sodium phosphate.

13. The solution of claim 8 wherein the solution comprises 0.5M sodium citrate.

14. The solution of claim 8 wherein the solution comprises 0.5M sodium isocitrate.

15. The solution of claim 8 wherein the pH of the solution is below pH 7.0, and wherein the solubilizer is not glycine.

16. The solution of claim 8 wherein the solubilizer is acetate ion and the acetate ion is present in the solution at a concentration of from 5 mM to 200 mM.

17. The solution of claim 8 wherein the solubilizer is citrate ion and the citrate ion is present in the solution as sodium citrate or potassium citrate at a concentration from 50 mM to 500 mM.

18. The solution of claim 8 wherein the solubilizer is isocitrate ion and the isocitrate ion is present in the solution as sodium isocitrate or potassium isocitrate at a concentration from 100 mM to 500 mM.

19. The solution of claim 8 wherein the solubilizer is glycine and the glycine is present in the solution at a concentration from 5 mM to 20 mM.

20. The solution of claim 8 wherein the solubilizer is glutamate and the glutamate is present in the solution at a concentration from 5 mM to 20 mM.

21. The solution of claim 8 wherein the solubilizer is succinate ion and the succinate ion is present in the solution as sodium succinate at a concentration from 5 mM to 20 mM.

22. The solution of claim 8 wherein the solubilizer is histidine and the histidine is present in the solution at a concentration from 5 mM to 20 mM.

23. The solution of claim 8 wherein the solubilizer is imidazole and the imidazole is present in the solution at a concentration from 5 mM to 20 mM.

24. The solution of claim 8 wherein the solubilizer is sodium dodecyl sulfate and the sodium dodecyl sulfate is present in the solution at a concentration of 0.001 % to 0.1 % (weight / volume).

25. The solution of claim 8 wherein the solubilizer is polyethylene glycol and the polyethylene glycol is present in the solution at a concentration of 0.2 % to (weight / volume).

26. The solution of claim 8 wherein the solubilizer is sucrose and the sucrose is present in the solution at a concentration of 0.2 % to 10 % (weight / volume).

27. The solution of claim 8 wherein the solubilizer is mannitol and the mannitol is present in the solution at a concentration of 1.0 % to 5.0 % (weight /
volume).

28. The solution of claim 8 wherein the solubilizer is sorbitol and the sorbitol is present in the solution at a concentration of 0.2 % to 10 % (weight / volume).

29. The solution of claim 8 comprising at least 20 mM citrate.

30. The solution of claim 8 comprising at least 5 mM sodium acetate.

31. The solution of claim 8 comprising at least 0.005% (w/v) polysorbate-80.

32. The solution of claim 8 comprising at least 5 mM histidine.

33. The solution of claim 8 comprising at least 10 mM imidazole.

34. The solution of claim 8 comprising at least 1 % (w/v) glutamate.

35. The solution of claim 8 comprising at least 0.1% (w/v) polyphosphate.

36. The solution of claim 8 comprising at least 120 mM ammonium sulfate.

37. The solution of claim 8 comprising at least 0.02% (w/v) sodium dodecyl sulfate.

38. A solution according to any one of claims 8-37 wherein the polypeptide is ala-human TFPI having the amino acid sequence shown in Figure 4 and having one further amino acid which is an amino-terminal alanine.

39. A solution according to claim 38 which is pharmaceutically acceptable.

40. A solution according to any one of claims 8-37 which is pharmaceutically acceptable.