CA1314011C - Method for the purification of proteins - Google Patents

Abstract

Abstract of the Disclosure The recovery of vitamin X-dependent proteins produced by transformed microorganisms can be effected from the cell culture medium utilizing the changes in the protein which occur in the presence of divalent cations. The present process uses divalent cations to alter the binding affinity of the proteins and thereby selectively elute the proteins away from contaminants in the culture medium using standard chromatography.

Description

METHOD FOR T~E PURIFICATION OF PROTEINS

A large number of human and other mammalian proteins, including, for example, human growth hormone, human protein C and clotting Factor VII, have been produced in host cells by transfecting these cells with DNA encoding these proteins and growing the recombinant cells under conditions favorable for the expression of the protein. Grinnell et al. describe the expression of recombinant human pro~ein C (HPC) by human kidney cells in Biotechnology, 5:1189-1192 (1987). The proteins are secreted by the cells into the cell culture medium, and must be separated from the culture medium and the other components, such as cell waste products, cell debris and proteins or other material, which also has collected in the medium. In addition, the biological activity of the protein must be preserved, so the recovery conditions must be mild enough to preserve the biological activity of the protein, but, at the same time, thorou~h enough to effectively separate the protein from contaminants in the medium. Purity is often an important consideration, especially or pharmaceutical applications.
Recovery of proteins in biologically active form from cell culture medium presents a number of problems. ~or example, the desired protein must be separated from other closely related proteins in the cell culture medium, such as homologous, biologically inactive proteins, which may be associated with the protein. The recovery process should yield the biolo~ically active form of the protein with a high level of purity.

X~7029A -2-Jones et al. describe a method for recovering re~ractile proteins (non-e~ported proteins which form in~oluble protein granul~s within the host cell3 from the cytoplasm o~ a host cell in U.S. ~atent 4,512,922.
Related patent~ describing denaturing-refolding protein recovery ~ystem~ include U.S. Paten~s 4,599,197;
4,518,526; and 4,511,503.
Raush and ~eng in U.S. Patent 4,~77,196 describe recovery of heterogenous proteins ~rom a host cell, which are also in the form of refractile bodies.
~ ung et al. in U.S. Patent 4,734,362 describe a process for recovering recombinant refractil~ protei~s ~rom a host cell, involving dena~uring the protein, and subsequent renaturing to yield ~he desired product.
The recovery and purification of human ~o-agulation Factor VII is described by Brose and Majerus in The Journal of Bioloqical Chemist~y, ?55 :1242-1247 ~1980). They purified Factor VII ~rom human plasma with a yield of about 30% using a process which involved first absorbing ~he proteins to barium citrate, and separ~ting by chromatography.
Vitamin R-dependent proteins are a class of proteins i~volved in ~aintaining hemostasis. The dependency on vitamin K occurs during the biosynthesis of the proteins. ~uman protein C (HPC) is a vitamin K-dependent plasma glycoprotein that plays a key role in maintainin~ hemostasis. C.T. Esmon, Science, 235-1348-1352 (lg87).
The binding of calcium ions (Ca2 ) to HPC
causes a confoxmational chan~e in ~PC ~hat can be measured by fluorescence emission spectro copy. Johnson ~1 et al., J. Biol. Chem., 258:5554-5560 (1983). The conformational change results in a change in surface charge distribution as measured by a difference in the migration pattern of the protein in an electrical field, such as agarose gel electrophoresis. Stenflo, J., J. Biol. Chem., 251:355-363 ~1976).
The invention provides a purification pro-cedure by which an exported vitamin K-depend~nt pxotein produced by a host cell, or produced by a host cell after transformation or transfection with DNA encoding the protein, is recovered from the cell culture medium and purified. Vitamin K-dependent proteins bind divalent cations, such as calcium or barium ions, resulting in conformational changes in the protein, and alteration of the surface charges on the protein. These changes are utilized in the present process to control the binding affinity of the proteins to various sub-strates in the presence of divalent cations. The process uses conventional chromatography to separate the proteins based on the ionically altered binding affinity of the proteins.
In the present process, the cell culture medium containing the protein is treated with a chelating agent to remove endogenous divalent cations.
The medium is contacted with an ion exchange resin, for which it has a strong affinity. The protein is then eluted from the resin with a solution containing divalent cations which bind to the protein which elutes as a protein-cation complex. The protein-cation complex is the~ contacted with a resin which has an immobilized chelating agent which binds the cation. The chelating resin preferentially binds the cationr and th2 protein alone elutes from thi8 resin. The protein i6 then contacted with a second ion ~xchange resin ~or ~urther purification. The protein i~ ~reated with a 6econd cation-containing buffer forming a protein cation complex, and the complex i8 contacted with a hydrophobic resin. The protein-cation comple~ binds ~trongly to the hydrophobic resin. The protein bound resin can then be treated wi~h a chelating agent which binds the cation, and highly pure protein ca~ then be ~luted from the hydrophobic resin. The binding dif~erential between the protein and the protein-cation complex can be utilized to pro~ide an efficient, non-de~aturing pro~ess for recovexing substantially pure, biologically ~ctive protein in yields of over 90% in each ~tep.
Figure 1 show a :Elow chart depicting the present process for purification of a divalent-cation binding protein.
Figure 2 show~ ~ho elutlon profile of human Protein C from a Pharmacia ~'MonoQ"~ ion exchange resin using a NaCl gradientO
Figure 3 shows the elution profile of human Protein C from a Pharmacia 9'MonoQ':* ion exchange resin using a CaCl~ gradient.
Figure 4 shows the elution profiles of human Protein C from a Pharmacia "Fast Flow Q"~ ion exchange resin using both a CaCl2 elution buffer and a high NaCl buffer.
HPC, and most of the other vitamin K-dependent 30 protein~, bi~d divalent cations, such as Ca2~. It is ~Trademark ~' 131~01 1 believed that the majority of binding sites on the proteins are modified glutamic acid residues. Ohlin et al., 1988, J. Biol. Chem., 263:7411-7417. The reaction by which the glutamic acid residues are modified is gamma carboxylation, which is a post-translational modification performed by a microsomal enzyme vitamin K-dependent carboxylase. The gamma carboxylated glutamates (called Gla residues), are necessary for biological activity of ~itamin K~dependent proteins.
For exa~ple, in the case of HPC, the first nine consecutive glutamate residues in the HPC prot~in sequence must be modified by gamma carboxylation for the protein to be biologically active, (e.g., having antithrombotic activity).
For HPC, these Gla residues form most of the binding sites for Ca2 . N.L. Esmon et al., J. Biol.
Chem., 258:5548-5553 (1983). There is a high-affinity Ca2+ binding ~ite that is fo~med between the epidermal growth factor-like domain in the light chain of HPC and the heavy chain of HPC as described by Johnson et al., in J. Biol. Chem., 258:5554-5560 (1983); Ohlin and Stenflo, J. Biol. Chem., 262:I3798-13804 (1987); and Stearns et al., J. Biol. Chem., 269:826-832 (1986). The change in surface charge distribution of the HPC protein is due to the neutralization of the nine Gla residues (2 negative changes per residue) by Ca2+, resulting in a net loss of 18 negative charges. The change in surface charge distribution in HPC caused by Ca2 binding could also be a result of conformational changes. This change in conformation aff2cts its binding profile to con-ventional re~ins such as those used in ion-exchange ~31401 1 chromatography and hydrophobic chromatography. More particularly, this change causes convenkional ion-exchange chromatography resins to behave like "pseudo-affinity" resins.
The method of the invention can selectively separate low specific activity protein from high specific activity protein. This selectivity is based on the number of Gla residues present on the protein. For example, low specific activity proteins, (i.e., proteins having fewer Gla residues), can be separated from higher specific activity proteins (i.e., proteins having a high number of Gla residues), based on the higher affinity of Gla-containing proteins for the resin. Proteins having a higher number of Gla residues will show more pro-nounced conformational and electrical changes uponcomplexing with a divalent cation such as calcium, and these high-activity proteins will therefore elute more readily from the column when an elution buffer con-taining divalent cations is used. This selectivity is extremely powerful and useful. Many mammalian cell lines are not capable of expressing fully biologically active, recombinant vitamin K-dependent proteins due to the lack of the presence of all the Gla residues. The method of the invention can separate the fully active vitamin K-dependent proteins from less active forms of the same protein. This procedure is simple, ine~pensive, and readily set up by any biochemical laboratory.
The invention is based upon the use of con-ventional chromatography resins (such as ion-exchange or hydrophobic) as pseudo-affinity resins. The presence or absence of a low concentration of a divalent cation, X-7029~ -7-specifically Ca2 , affects the elution profile of HPC onconventional chromatography resins. This phenomenon can be extended to all vitamin K-dependent proteins and/or peptides, and potentially to all divalent cation-binding proteins, including Ca2+-binding proteins, peptides or macromolecules. Since Ca2 is the physiologically most abundant effector divalent metal ion for ~inding to the known vitamin K-dependent proteins, it is being used for most of the subseguent experiments. However, other divalent cations such as strontium (Sr2 ), and barium (Ba2 )~ can be substituted for Ca2 . These metal ions achieve the same results.
The present process is effective for all vitamin K-dependent proteins, however produced, including, for example, human protein C (HPC), Factor IX, Factor X, Factor II, Factor VII, human protein S
(~PS), Protein Z, bone Gla protein and bone matrix Gla protein. The present method is effective for both vitamin K-dependent protein zymogens, such as HPC, and for the corresponding activated forms of the serum proteases, such as activated Protein C (APC).
In one embodiment, the invention described herein is directed to procedures which are useful in isolating, purifying, reactiva~ing and using hetero-logous recombinant proteins that, subsequent toexpression in microorganisms (host cells), are secreted from the host cell into the cell culture medium. For purposes of the present invention, proteins which are secreted are referred to as "exported proteins"~ In another embodime.l~, the inventior~ described herein is directed to the isolating, purifying, reactivating and using exported proteins that are produced in non-transformed cell lines.
When recombinant DNA technology is employed to induce host microorganisms to produce foreign proteins, such proteins are often referred to as "het~rologous proteins" or "recombinant proteins". In the present invention, the term "protein" is meant to encompass all divalent cation binding polypeptides and proteins. The terms "heterologous" and "recombinant" are used interchangeably to denote a protein secreted by a host microorganism which binds a divalent cation.
The protein is first cloned according to well-known standard recombinant DNA procedures. The cloning of ~PC has been described by Beckmann et al.
lS in Nucleic_Acids Research, 13:5233 (1985). The expression of recombinant HPC (rHPC) with human kidney 293 cells has been described by Grinnell et al. in Biotechnolo~yy, 5:1189-1192 (1987~.
The culture medium is collected and, option-ally, centrifuged, at about 20,000 times gravity, for about twenty minutes at chill room temperatures (of about 4C) to remove cell debris. The supernatant contains the protein. Aft~r centrifugation, a protease inhibitor, such as benzamidine, and a chelating agent, such as EDTA or EGTA, in a concentration sufficient to remove all divalent cations, can be added to the medium (see Figure l, steps 1-2).
The medium can then be contacted with an ion exchange resin, such as an anionic quarternary or tertiary amine-~ased resin (Figu e 1, step 3~. Some examples of available suitable commercial resins include - t31~01 1 X~7029A _g_ Pharmacia ~Fa~t Flow Q" (FFQ) and ~ono Q," and QAE-A50-120 ~nd DEAE tertiaxy/guaternary amine from Sigma. In one a~pect of the invention, the re~in can be contained in a colu~n. ~owever, the re~in may also be in a bed or o~her configuration as long as the medium i~ able to filter through and ~ontact a sufficient resin ~urface area to ensure adequate ion exchange. This ~tep is carried out ~t chill-room temperatures (between 8-10C).
The resin can be first equilibrated with a neutral pH buffer solution ~ontaining a small amount of proteas~ inhibikor, chelating agent and, optionally,~
a monovalent salt. Any neutral buffer may be used, providing that it doe not react wi~h Ca2+; for example, phosphate buffer forms an insoluble complex with Ca2+, thus cannot be u~ed. A preferr~d equilibrating buffer ~olution can contain about 20 mM Tri~ buffer, 2 ~M EDTA, 2 mM benzamidine and 0.15 M NaCl, having a p~ of about 7.4. The recept~cle, (e.g., a column), can then be packed with the re~in. Bed volume should be su~ficient to provide binding siteC for th2 proteinO The culture medium, which has already been treated with a proteas~
inhibitor and chelating agent, is then loaded onto the column. Flow rate is adjusted so that maximum protein binding o~curs~ In the case of HPC, the li~ear flow rate should be about 40-80 centimeters per hour.
The loaded column can then be washed with about three or more column volumes of a neutral buffer, (e.g., Tris buffer, pH 7.4~, which contains a monovalent salt (e.g., NaCl or KCl), a protease i~hi~itor (e.g., benzamidine~ and a chelating agent Ce g~, EDTA).
Optionally, a second wash wi~h about two column volumes of neutral buffer containing a salt and protease inhibitor can be done. At this point, the desired protein is bound tightly to the ionic resin, as these proteins have a high affinity for the resin. Most of the other proteins and contaminants in the cell culture medium have been washed awayO To remove the protein from the column, an 'ielution" buffer containing the divalent cation, preferably calcium (Ca2 ), is used (Figure 1, step 4). The calcium ions will bind preferentially to the protein forming a Ca-protein complex. This complex has a low affinity for the resin, therefore the Ca-protein-complex will be contained in the eluate. The elution buffer can be a combination of a neutral buffer (e.g., Tris), a monovalent salt (e.g., NaCl), a calcium salt (e.g., CaCl2), and a protease inhibitor (e.g., benzamidine). A preferred elution buffer can contain 20 mM Tris, 0.15 NaCl, 10 mM CaCl2 and 5 mM benzamidine, and have a pH of about 7.4. The protein elutes with the second column volume of the eluant. About ninety (90%) percent of the protein is eluted by the end of the second column volume. Protein recovery after this step is about 80-90%.
The eluate containing the protein can -then be treated with a resin containing an immobilized chelating ayen-t, and then contacted with a second ion-exchange resin (Figure 1, s~eps 5-7). Columns or beds containing these two resins may, optionally, be set up in tandem, so that the eluate from the chelate column flows directly into the ion-exchange column. Alternatively, the eluate from the chelating column can be collected, and then loaded onto the ion-exchange column. A com-13~40~ 1 mercial chelating column containing a resin having an immobilized chelating agent can be u~ed, 6uch a~ ~Chelex 100" (Biorad),which has immobilized EDTA. The pu~pose of this c~lumn is to remove the calcium ~rom the protein. The ion-exchange resin can be the same type as the ion-exchange resin u~ed in the first ~tep. In ~his ~tep, both resin~ are firæt eguilibrated by washing with a neutral p~ buffer, (e.g., Tri6 buffer) containing a low ~oncentration of salt. Th~ capacity of the columns is ~ependen~ upon the sample volume. Bed volume of the chelating ~olumn should preferably be about 20 ml for each 200 ml of sample; and bed volume for the ion exchange column ~hould preferably be about 50 ml for each 0.5-1.0 grams of prot~in. Both columns should be 15 run at a flow rate ~ufficient to remove unbound calcium, and further purify the protein. This step can also be carried out at chill room temperatures. In a preferred method, thP eluate from the first step is loaded on the tandem-linked columns. The loaded chelate column c~n ~hen be washed with two col~mn volumes, ba~ed on ~he chelate-column volume, o a neutral pH buffer having a low concentration of salt. Once the liguid has eluted, the chelate column can then be disconnected. At ~his point, ~he protein is bound to the ion ~xchange column.
It has been found that the protein will bind to the ion exchange column at low salt concentrations, and elute at higher salt concentrations. .To elute the protein, therefore, the column can be treated wi~h a eries of buffers containing a salt gradient (see Figure 1, step 8 and Fiqure 2). For example, a buffer, consisting of pH
7.4 Tris buffer and 1 M NaCl, can be contacted with ~he Trademark X-7029A -12~

column using a series of solutions ~ontaining between 0-50% of this buffer over about twenty column volwmes.
The protein begins to elute with ~he solution containing about 27% buffer, and peaks at about 30% buffer. The protein may al~o be eluted using high salt buffers in lieu of a gradient ~e.g., about 0.4 to 1 M NaCl). The elution is monitored by mea~uring the change in optical density u~ing spectroscopy to measure absorbance at 280 mm a~ described b~ Kisiel and Davie in Meth. in Enzymolo~y, 80:320-332 ~1981). At this point, ~he protein recovery is more than 90~.
The protein-containing eluate fr~ctions are then contacted with a hydrophobic resin i~ order to concentrate and purify ~he protein by removing protein contamina~t6 from the eluate. A hydrophobic resin, su~h as phenyl "Superose,~'~ can be used. Commercially available resins include phenyl ~'Superose"~ HR5/5 and phenyl-"Sepharose"~ CL-4B, both from Pharmacia. The hydrophobic resin can fir~t be eguilibrated with a neutral buf~er containing, optionally, a monovalent salt, and a divale~t cation. A preferred equilibration buffer is 20 m~ Tris, 1 M NaCl, and 10 mM CaCl2, having a pH of about 7.4.
In ~his step, the protein containing ~raction, eluted from the prior step, is treated with a second divalent cation, such as a buffer containing about 10 mM
CaCl2, and loaded onto the hydrophobic r sin and washed with ~he eguilibration buffex (Figure 1, 8tep6 9-10).
It has been ~ound that vitamin K-dependant proteins bind weakly to hydrophobic resins, such as phenyl-"Superose,"
in the abse~ce of Ca~; but have a hiyh affinity ~or the . .. 1 Trademark ~Trademark ~' 1 3 1 40 1 ~

re~in in the presence of Ca2+, a~d c~n ~hus be eluted fr~ the re~in with ~ solution containin~ a chelating ~gent, ~uch ~ ~DT~. The prot~in can be eluted with an elution buffer containing a neutral buffer, a low concentration of monovalent salt, ~nd a chelati~g ~gent.
A pr~ferred elukion buffer can contain about ~0 ~M Tris, 0.15 M NaCl and 1 mM EDTA ~p~ 7.4).
The purity of the protein uæing ~his proc~dure is greater than 98~, as det~rmined by SDS sPAGE
chromatography. Laemmli, Nature, 227:680 685 (1974).
The protein alco retain~ 100% biological activity as determined by functional assays, as described by Grinnell et al., in Biotechnolo~y, 5:1189~ 2 (1987~.
The invention i~ further illustr ted by ~he following exemplification.
.

ExamPle 1 Separation of HPC u~ing Anion-exchanqe column chromato~ra~kY

A quaternary ~mine-based ~trong a~ion exchange re~in (i.e.,"Fast Flow Q" or "~ono-Q" from Pharmacia) are used for ~he ~ollowing experiments.
Quaternary amine based resin from any reputable commercial company should ervice (~.g., QAE~A50-120 from Sigma). Since RPC binds a's~ to ~ertiary amine based resins, such as DEAE-'ISepharos~ CL-6B (Sigma).
These resins ~an al~o be used to obtain the same results.
The results illustrate ~hat HPC binds to the 3^ anion exchange resin in the ab ence of Ca2~.

~Trademark Materials:

Column: :I?ha~macia "Mono-Q, n HR5/5 Instnlment: Pha:rmacia E:PLC LCC~500~ sy3tem to run S the NaCl gradient ESuffer A: 2ûmM Tris, p~ 7.4, 0.15M NaCl 10 Buffer B: 20mM Tris, pH 7.4, lM NaCl Flow rate: 1 ml/min NaCI gradient: 0-100% Buffer B in 20 minute~
The ~olumn was conditioned as ~uggeste~ by the manufacturer. Then the t:olumn (bed volume 1 ml ~ was eguilibrated with Buffer A. ~ sample t:ontai~ing 6 mg of plasma Ecec in 8 . 5 ml o~ Buffer A wels loaded orlto the 15 colwnn, and the column was washed with three column volumes ( 3 ml ) of Buffer A prior to the ~tart o~ the NaCl gradient. As ~hown in Figure 2, all of the ~IPC
bo~and to the resin. The concerltration of EIPC was monitored by optic:al density by m~sa~urinçl absorbanc~ at 20 280~m as described by Ki~iel and Davie in Meth. in Enzymolo~y, 80: 320-332 ( 1981 ) .
It wa~ found that if HPC is in Buffer A conr taining 2mM CaCl2, the ~PC would not bind tv the ~'Mono-Q"
eolumn. 2mM CaCl2 is what is typically present in cell 2~ culture media or in human plasma. ~PC was shown to elute from "Mono-Q"resin with ~ solutiorl containing 0 . 4M
NaCl in 20mM Tris, (p~ 7.4). The amount of NaCl needed to elute H~C is pH dependent. For axample, the lower the pH, the higher is the concentration of NaCl required 30 and the higher the pEI, t:he lower the concentration o~
NaCl res~ired.
.

~Trademark X-7029A ~-15~ 1 3 1 4 O 1 1 ~ Elution_of HPC from an anion exchange .

column with a low concentration of .
CaC1~
The following experiment uses the Pharmacia Mono-Q column and the same protocol described in Example 1.

Materials:

Column: Pharmacia Mono-Q HR 5/5 Instrument: Pharmacia FPLC LCC_500 Buffer A: 20mM Tris, pH 7.4, 0.15M NaCl Buffer B: 20mM Tris, pH 7.4, 0.15M NaCl, 30mM
CaCl2 Flow rate: 1 ml/min NaCl Gradient: 0-50% buffer B in 2 minutes The column was equilibrated with Buffer A. A
sample containing 0.6 mg of HPC dissolved in 0.7 ml of Buffer A was loaded onto the column with Buffer A
prior to the development of the Ca2+ gradient. The HPC was eluded with a gradient of 6-9 mM CaCl2 in 20mM
Tris pH 7.4, 0.15 NaCl. The results, shown in Figure 3, sho~ that HPC elutes with increasing concentrations of CaCl2 .
HPC was quantified by determining optical densit~ by measuring absorbance at 280 nm as described by Kisiel and Davies in Meth. in_Enzymology, 8Q:320-332 30 (1981).

Example 3 Specificity of divalent metal cations for the elution of HPC from an anion e~change column The experiment was set up and run as described in Example 2. It was shown that EPC can be eluted isocratically with various concentrations of CaCl2 in buffer A or buffer C.

Buffer A: 20mM Tris, pH 7.4, 0.15M NaCl Buffer C: 20mM Tris, pH 7.4 The results are shown in Table 1.

TABLE I

divalent cation Buffer_A Buffer C Yield of HPC

5mM CaCl2 + - 80%
lOmM CaC12 +
lOmM caCl2 ~ ~ %
lOmM MgCl2 ~ - 20%

The data indicated that the divalent cation effect of Ca2~ in eluting HPC is ion-specific because magnesium chloride (MgCl2) in the same concentration is much less effective -than CaCl2.
The ionic strength of -the buffer containing the CaCl2 is also important. In the absence of 0.15M
~aCl, CaCl2 at lOmM CaCl2 was ineffective in eluting ~PC from Mono-Q column.

X-7029A wl7_ 1 31 4 0 1 1 Example 4 Selectiveness of using lOmM CaCl2 to lute_HPC instead of _4M NaCl from a Mono-~ column Two percent fetal calf serum (FCS) conditioned media from human kidney 293 cells (Grinnell et al., (1987) Blot chnology, 5:1189-1192) expressing 3.3 ~g/ml of r~PC was used to show the achievement of 240 fold purification in o~e step using an anion exchange column.
lOOml of Pharmacia Fast Flow Q (FFQ) resin was properly prepared as reco~mended by the manufacturer.
The FFQ resin was then equilibrated with a buffer solution containing 20~M Tris, 0.15M NaCl, 2mM EDTA, 2mM
benzamidine, (pH 7.4~. EDT~ and benzamidine were added to the 3.3 liters of 2% FCS conditioned media containing 3.3 ~g/ml of rHPC to a final concentration of 4mM and 5mM respectively. Then the culture media was passed through ~he FFQ column (3 x 16 cm) at a linear flow rate of 20cm.h 1. The column was washed first with 300 ml (3 column volumes) of a solution containing 20mM Tris, 0.15M NaCl, 2mM EDTA, 2mM benzamidine (p~ 7.4), then 300 ~l (3 colu~n volumes~ of a solution containing 20mM Tris, 0.15M NaCl, 2mM benzamidine (p~ 7.4), then 300 ml of a solution containing 20 mM Tris, 0.15 M NaC1, ~ 25 2 mM benzamidine, lOmM CaC12 (pH 7.4). The : column was then further eluked with a solution con-taininq 20~M Tris, 0.4M NaCl, 2mM benzamidine SpH 7.4).
The amount of ~PC was determined by measuring OD~80 as described in Example ~. Specific activity of ~PC was determined according to the procedure described by ~ Grinnell et al in Biotechnolo~y, 5:1189-1192 (1987), as : follows: ~PC was ~irst activated with dn immobilized thrombomodulin-thrombi~ complex ~obtained from Dr. C.T.

Esmon, Oklahoma Medical Research Foundation). The amidolytic activity of the activated protein C (APC) was measured by the hydrolysis of a tripeptide substrate S-2238 (Helena). The anticoagulant activity of HPC
was determined by the prolongation of an activated partial thromboplastin time (APTT) using reagents from Helena. The assays and the definition of a unit of the specific activity of HPC is that described by Grinnell et al. The results are shown in Figure 4, and below in Table II.

TABLE II
Sam~le Total Total Purity Specific protein rHPC of r~PC activity (mg) (mg) [HPC] anti~en (units/
__ mg HPC) starting media 4422 10.9 0.25% 0.074 unbound fraction 4290 0.016 0.0004% --10mM caCl2 fraction 16.2 9.4 58% 17.5 0.4M NaC1 fraction 115.2 0.12 0.1% --The results from this experiment clearly demonstrated that the purity of rHPC was increased from 0.25% in the starting material to about 58% (a to~al increase of 232 fold). By comparison, using the "conventional" mode of eluting rHPC with 0.~M NaCl, the purity of rHPC at that stage is only 7% (a total increase of 28 fold). So ~he present mode gave an additional 8.3 fold of purification.

Example 5 The elution of proteins from anion exchan~e chromatography is specific for Ca2 binding proteins and vitamin ~-dependent proteins.

Two non-Ca2 binding and non-vitamin K-dependent proteins were used in this example. Both proteins normally bind to the Pharmacia Mono-Q column under the conditions specified in Example 1, iOe. 20mM
Tris, 0.15M-NaCl (pH 7.4). The two proteins used were glucose oxidase and amyloglucosidase (Aspergillus niger Cat. ~ G2133 and A3423, respectively, from Sigma). The experiments described in Examples 1 and 2 were repeated for each of the two proteins and the results are shown in Table III.

TABLE III

protein Concentration of Concentration of CaCl2 required NaCl required for elution in for elution in 20mM Tris, 0.15M 20mM Tris (p~ 7.4) MaCl (p~ 7.4~ _ _ glucose oxidase 18mM 0.30M
amyloglucosidase over 20mM O.36 HPC 9mM O.40 E~am~le 6 Selectivity of the "pseudo-affinity"
-mode for removal of non-protein contaminants Conditioned culture media from human kidney 293 cells expressing r~PC was used for this experiment.
Grinnell et al., Biotechnolo~y, 5:1189-1192 (1987).
The culture media contained endotoxin (lipopoly-saccharide A) at 80 endotoxin units/ml (8 ng endo-toxin/ml). Endotoxins are heterogeneous molecules of lipopolysaccharide, negatively charged, and derived from the outer coat of gram-negative bacteria. The experiment was carried out as described in Example 4, except that the endotoxin level was measured in place of total protein concentxation. Endotoxin levels were measured using an Endotoxin assay kit from Whittaker Bioproducts. Starting with a total of 4 X 106 e~dotoxin units, 5.7 x 104 endotoxin units were recovered in the rHPC peak eluted wi$h lOmM CaCl2, 20mM Tris, 0.15M NaCl, pH 7.4. This represents a total removal of 98.5% of the endotoxin from the starting culture media after one step of purification.

Example 7 Selectivity of the "pseudo~affinity"
mode for the removal of contaminatin~
. . . _ or~anisms The experiment was carried out as described in Example 6. 5 x 101 phi-X174 phages (~TCC number 13/~6-sin shiemer-c-vl) were introduced into conditioned culture media from human kidney 293 cells expressing :: .

rEPC. This media was then pa~sed through the FFQ
column. O~ly 1 x 105 phi-X174 phages were recov~red in ~he CaCl2 eluted fraction containing the r~PC, whi~e 2-3 x 106 phi-X174 phage~ were recovered in th~ 0.4M
NaCl eluted fraction. These ~how that the CaCl2 elution (~Ipseudo-affinity~ ~ode) gives 20-30 ~old bett~r selectivity than the 0.4M NaCl elution (conventional mode).

Example 8 Purification o~ reco~b ~ant ~uman ~r~tei~L~L~D~
-A. Purification of rHPS ~roduced b~ AV12 Cells ~PS is a vitamin K-dependent protein con-taininy 11 Gla residues. Conditioned culture media containi~g ~PS was obtain~d by conventionally tra~s-forming Syrian hamster AV12 cells (ATCC number CRL 9595, deposited November 24, 1987) with plasmid pShD, con-structed in substantial accordance with the teaching of European Patent Application EP-A 0247843, publi~hed February 12, 1987~
and was used for the following experiments.
The procedure described in Example 1 was repeated using the pres~nt culture media containing rHPS. rHPS wa~ eluted using the l'conventional" mod~
(described in Example 7) from a Pharmacia FFQ column with a solution of 20mM Tris, 0.33M NaCl, ~pH 7.~).
The CaCl2 elution procedure described in Example 2 : 30 w ~ then used for t~.~ culture media containing rHPS.
rHPS was eluted succe~sfully using ~he ~p~eudo-affinity"
mode from the FFQ column with a ~lu~ion of 20mM Tris, O.lSM NaCl, 3.5mM CaCl2 (pH 7.4).

~Trademark X-7029A ~22-B. Puri~ication of hlqh specific activilt~ r~IPS_ E~ced .
by 293 cells rEPS was also ob~ained by conventionally !5 tran :Eorming human kidney 293 c:ells with plasmid pShD, ~en culturing the cells in serum-free media. Th~
r~S c:ultuxe media was added to Pharmacia "Fast Flow Q'~
resiIl then washed with Buffer A in ~ stanti 1 accordance with the teaching of Example 1. The CaCl2 elution lO proceduxe described in Example 2 was then used for the ~PS culture media, except that the elution buffer contained 20mM Tris, 0.15Pq NaCl, 3.0mM CaCl2 (pH 7.4).
About three col~ volumes were c:ollected, then 'che column was eluted with a buffer containing 20mM Txis, 15 pE~ 7.4, 0.5 M NaCl. The biological activity of the eluted r7~S Irom both elution buffers was then tested using the assay method of Malm, et al ~1987 ) Eur . J .
Biochem. 165: 39-45 .

rHPS obtained from ~V12--trans~ormed cells grown in serum-free media (~s in Example 8A) was al50 loaded onto Pharmacia"Fast Flow Q"resin. The AV12-derived r~PS was then eluted using 3.0~M CaCl2, fol-lowed by a 0.5 M NaCl elution, substantially as described above for the 293~derived r~PS. Bioactivities were then assayed by the method of Malm et al.
Ninety-seven (97%) of the total functional activity of the 293-derived rHPS was eluted with a solution of 20m~ Tris, 0.15M NaCl, 3.0mM ~aCl2, (pH 7.4 (CaCl2 fraction), while the remaining three (3%~ per-cent of the fun~tional activity o~ the 293-derived ~31~01 1 X-7029A -23~

rHPS was eluted with a solution of 20mM Tris 0.5M
NaCl (pH 7.4~, (NaCl fraction). However, only forty-three (43%) percent of the total functional activity of the AV12-derived rHPS was eluted in the CaCl2 fraction, while fifty-three 153%) percent of the functional activity of the AV12-derived rHPS was eluted in the NaCl fraction.
The Gla content and beta-hydroxyaspartate content were measured in both the CaCl2 and NaCl fractions of rHPS as described in Example 9. The rHPS
molecules from the CaCl2 and NaCl fractions displayed no differences in beta-hydroxyaspartate content, molecular weight (reduced and non-reduced SDS-PAGE) and N~terminal protein sequence. However, the rHPS molecules from the two fractions did differ in Gla content, as the molecule from the NaCl fraction has 2 fewer Gla residues than does the molecule from the CaCl2 fraction.
This accounts for the lower specific activity tabout 50%
less) of rHPS derived rom AV12 cells as compared to fully functional rHPS derived from 293 cells.
This experiment demonstrated that the "pseudo-affinity" mode (CaCl2 fxaction) of eluting r~PS using anion exchange chromatography can selectively separate low specific activity rHPS (low Gla content) from high specific activity rHPS (high Gla content).

X~7029A -24 ~ 3 1 4 0 1 1 Exam~le 9 r~PC.

S ~uman Prothrombin protein ha~ 10 Gla residues, which are essential for biological activity. Borowski et al., ~ , 260:9258-9264 (1985). Natural variants of hum~n Prothrombin ~issing two or four Gla re~idue~ retain only 66% and 5% o~ their biological activi~y, respectively. Since prothrombin ~i~5ing 2 ~la out of a total of 10 Gla result~ in a drop of more than 30% of activity, the presence of all Gla residues are e~sential for ~ull activit~.
r~PC that was only partially acti~e (30-60%
anticoagulant activity as compared to a pla~ma EPC
standard) when measured in the crude culture media was obtained by trans~orming Syrian hamæter A~12 cells ~ATCC number CRL 9S95) wi~h plasmid p4-14, constructed in 6ubstantial accordance with ~he teaching of C~nadianPatentApp~ca~onSen~ N~mberS33,716OfBnan W.G~nnell,~led April 2, 1987.
Activity was measured as described for ~PC in Example 4. The rHPC from this culture media was absorbed and ~luted according to the procedure described in Example 4.
Forty-five (45%) percent of the total starting rHPC in the cult~r~ media was eluted with a ~olution of 20mM Tris, 0.15M NaCl, lOmM CaCl2, ~p~ 7.4) (CaCl2 fraction), and 20~ was eluted with 20mM Tris, 0.4M NaCl, (pH 7.4) (NaCl ~raction). The anticoagulant activity of ~he r~PC in ~he CaCl2 fraction and in the Na~l fraction, were 100% and 25% respectively, as compared to a plasma HPC stan~ard. The Gla cont~nt and beta-hydroxy-.'~ ' .

a~partate content were mea~ured in ~he rEPC i~ bo~h the CaCl2 fraction and in the NaCl fraction, u~i~g a procedure adapted ~rom the procedure de~cribed by ~uwanda and Katayama in Anal. Bio~hem., 131:173 179 S (1983): the alkaline hydroly~iæ of the protein prior to the ~mino acid analysis was carried out wi~h a nTeflon"~
vial with miniert valves. (Pierce, Cat. ~ 14005,10130).
The protein ~ample in 2.5N NaOH wa~ evacuated and purged with N2 via the miniert valve u~i~g a Waters picotag work station. After 20 hours of hydrolysis at 110C, the hydrosylate was ~eutralized, extracted and derivatized with o~ph~halaldehyde/ethanethiol as described by Ruwada and Katayama. The HPLC analysis wa~ carried out under th~ followi~g conditions:
column~ ucleosil 5SB"~ (4.6 x 50) (Macherey-Nagel) Isocratic elution: 20 mM Na citrate, p~

4.30 in 50% acetonitrile Flow rate: 1.5 ml/minute.
The following elution times were obtained:
AMINO ACIDS ELUTION TIME
_ non-acidic amino acid~ 6 min Glu 9.5 mi~
10 Asp 13 min erythyro-beta-OH-asp 20 min threo-beta-O~-asp 34 min Gla 44 min cysteic a~id 53 min ___ . _ _ _ _ _ _ _ The CaCl2 fraction and the NaCl fraction were found to co~tain 9 and 6.5 moles of Gla per mole of r~PC, : respectively. .~

Trademark for polytetrafluoroethylene resin Trademark X-7029A -26~ 4 0 1 1 The number of Gla residues present correlates very well with biological activity in rHPC as predicted by what was reported in the literature for other vitamin K-dependent proteins. Borowski et al., J. Biol. Chem., 260:9258-9264 (1985). Other than the difference in Gla content in the rHPC between the CaCl2 fraction and the NaCl fraction, no o-ther difference was detected in beta-hydroxyaspartate content, molecular weights (reduced and non-reduced SDS-PAGE) and N-terminal protein seguence. N-terminal protein sequence analysis was performed by automated Edman degradation chemistry on Applied Biosystem model 470A gas phase se~uenator with on-line HPLC system (model 120A) for the analysis of PT~-amino acids.
This experiment demonstrated that the "pseudo-affinity" mode (CaCl2 fraction3 of eluting rHPC using anion exchange column chromatography can selectively separate low specific activity rHPC (low Gla content) from high specific activity rHPC (high Gla content).
- Example 10 Elution of activated human Protein C
(APC) from an anion exchange column HPC is the zymogen form of the active serine protease, activated human Protein C (APC). The only molecular difference between HPC and APC is that APC
lacks a 12-amino acid peptide at the N-terminus of the heavy chain of the HPC. Thus, there is no dif-ference in the Gla content of ~PC and HPC.

X-7029A -27~ 4 0 1 1 rAPC was prepared from r~PC with immobilized thrombomodulin-~hrombin com~lex as describ~d by Grinnell et al. in iotechnologx, 5:1189-1192 (1987). The ~perimental proto~ol described in Ex~mples 1 and 2 wer~
repeated for rAPC. The results of the elution profiles of rAPc from a Pharmacia "Mono-Q~ column ~ere identical to ~hat of r~PC. The amount of CaCl~ or NaCl reguired for elution of rAPC for either the "pseudo-affinity" mode or the "conventional" ~ode were identical to that of rHPC.
Exam~le 11 aydro~hobic column chromato~raphy ThrPe of the most common conventional types of column chromatographie6 used in bio~hemical research are, ion-excha~ge, hydrophobic/rever~e pha~e and ~ize-exclusion. The ~ormer two types are dependent on ~he surface charge di~tributions of the biochemical compounds of interest, while size-exclusion chroma-tography is not. ~ydrophobic column chromatography was therefore used to illustrate that the l'pseudo-affinity"
vitamin K-dependent proteins can be separated on this typ2 of column using the "pseudo-affinity" mode.
Hydrophobic side chains are linked to a rigid BUpport to creat hydrophobic column resins. Phenyl groups were used for thi~ illustration. Other hydro~
phobic side chains, such as various lengths of aliphatic hydrocarbons, can al~o be u~ed. Two different types of rigid supports wexe used for phenyl "Superose" ~R 5~5 and phenyl "Sepharose" CL-4B, both from Pharmacia.

X-7029A -28- . 1 3 1 ~ O 1 1 (a) Materials Colu~n : Pharmacia phenyl ~Superose" HR 5/5 Buffer A: 20 mM Tris, 2 M NaCl, ~ 7.4 Buffer B: 20 mM Tris, 0.15 M NaCl, p~ 7.4 Buffer ~: 20 ~M Tris, 2 M NaCl, 10 mM CaCl2, pH 7.4 Buffer D: 20 mM Tris, 0.15 ~ NaCl, 10 ~M CaCl2, pX 7.4 Flow rate: 0.5 ml/min.
Chromatography system: Pharmacia FPLC LCC-500 system The ~olumn was prepared a su5~gested by the rltanufacturer and then eguilibrated ~ith buffer A. 1 ~g of rHPC was dissolved in buffer A, and then applied to the column.
The ~oncentration of protein was monitored by measuring the optical density at 280 ~m. The rBPC did not bind to .the column. No further material could be eluted with a gradient of 0-100% buffer B in 40 minutes. rHPC was dissolved in buffer C and then applied ko the column.
All the rH~C bound ~o the pherlyl "Superose" column. Tha only difference between buffer A and C is that buffer C
contained 10 mM CaC12. A gradient of 0-100% buffer D
was developed over 40 minutes. r~PC was eluted at 60%
bufffer D and 40% buffer C, or at 20 mM Tris, O . 9 M
NaCl, 10 mM CaCl2, (pH 7.4).
Thus, it was shown r~PC has a higher affinity to hydrophobic resins in the presence of a low con-centratio~ of Ca2+.

13~401 1 The exper ~ent ~as r~peated u~ing a ph2nyl ~'~epharose" column Sb~ Material 5column: Pharmacia ph~nyl"S~pharo~e" CL-4B
0.5 x 5 cm flow rate: 0~5 ~l/min rHPC was show~ to bind 100% to the column ei~her with buffer A ~20 mM Tris, 2 M NaC1, pH 7.4) or with a solution o~ 20 ~M Tris, 1 M NaCl, 10 mM CaC12, (pH
7.4). ~owever, r~PC would not bind to ~he col~mn in a ~olution of 20 mM Tris, 1 M NaCl, p~ 7.4.

~3~ a~ Usinq "pseudo-a~finit~" chromakoqra~y to ~urify r~PC_from cell culture media The following scheme is an example of a ~uri~
~ication ~cheme for a certain ~et of conditions and variables.
All the following ~teps were carried out at chill room temperature (~-10C~.

_te~ 1. Anion-exchan~ Fast Flow Q" column.

Serum free conditioned culture media from 293 cells expressing rHPC at 5 ~g/ml was u~ed. The serum ~xee culture media contained protein/peptlde ~upplement of insulin, transferrin. The concentration of rHPC
generally comprised 10-15% of the total protein in the conditioned culture medl~. Pharmacia ~Fast Flow Q" resin (FFQ) was cleaned with 1~ ~Cl and 1~ NaO~ in a manner as ,. .

X~7029A -30 1 3 1 4 0 1 1 suggested by ~he manufacturer. The resin was then packed into a 10 x 20 cm column. For every 500 liters of culture media, 1 liter of FFQ resin was needed. The column was packed to flow at a rate of 120 cm.h 1 with 20 mM Tris, 1 M NaCl, (pH 7.4). The column was equili-brated with a solution of 20 mM Tris, 0.15 M NaCl, 2 mM
EDTA, 2 mM benzamidine, (pH 7.4).
Solutions of 0.2 M EDTA, (pH 7.4) and 1 M
benzamidine were addPd to the culture media containing rHPC to a final concentration of 4 mM and 5 mM, respectively. The culture media was then applied to the FFQ column at a flow rate of 80 cm.h 1.
The FFQ column was then washed with a minimum 3 column volumes of a solution containing 20 mM Tris, 0.15 M NaCl, 2 mM EDTA, 5 mM benzamidine, (pH 7.4).
The FFQ column was then further washed with a minimum 3 column volumes of a solution containing 20 mM Tris, 0.15 M NaCl, 5 mM benzamidine, (pH 7.4). The rHPC was eluted with a solution of 20 mM Tris, 0.15 M NaCl, 10 mM
CaCl2, 5 mM benzamidine, (p~ 7.4). The flow rate was 5 cm.h . The rEPC was detected with Bradford protein reagent (M. Bradford, (1976) Anal. Biochem., 72:248-254) or ELISA assay as described by Grinnell et al., Biotechnology, 5:1189-1192 (1987). The rHPC eluted at the beginning of the second column volume using this elution buffer. Ninety (90%) percent of rHPC was eluted in half a column volume.

X-7029A ~31-Ste~e 2 nC;helex 100"_column in ~andem with "~as~
1~
A "Chelex 100" column ~Bio-rad) was used to remove the Ca2 in ~he r~3PC from 6tep 1. The FFQ was run in the conventional mode in this st~p. ~Chelex 100"
resin ~390 ml 3 war- washed with 1 N NaOgI HzO - 1 N ElCl HzO a~ recommended by the manufactuxer. The resin was packed irato a 3 . 2 x 40 cm column and was washed with a 0 60111tiOrl oî` 1 M Tris, (p~ 7.4). The column was equilibrated with an equilibration buffer c:o2ltaining 20 mM Tris, 0.15 rq NaCl, (p~ 7.4). The 1 M Tris wash was nece~sary to achieve fast equilibration o the "Chelex 100" to pH 7.4. The FFQ colulTn (3.2 x 25 cm~ was cleaned as described in Step 1, and eguilibrat~d with a ~olution of 20 mM Tris, 0.15 M NaCl, (pl~ 7.4~. The "Chelex 100" c:olumn was hoo}ced up in ~andem with the TFQ
colu~n ~uch that ~he eluate containing r~PC from Step 1 will pass through the "Chelex 100" first~ and then the 2 0 FFQ .
After all of l:he rHPC ~rom Step 1 had been . loaded, the columns were washed with 1.5 liters of the equilibraltion buffer. Then the ~Chelex 100" oolumn was unhooked from the FFQ.
The FFQ wae; further washed with 600 ml of the es~uilibrating buffer. The FFQ was then washed with 600 ml of a solution of 20 mM Tris, O . 25 M NaCl, (pH
7 . 4 ) . No rHPC was eluted here . ~he r~?C was eluted from the FFQ with a high salt solution o~ 20 mM Tris, 0.4 M NaC~, (pH 7.4). The ~HPC was detected by monitoring absorbance at 280 nm. The yield of r~PC from this step was 90-95%.
., Trademark .~i ~
S~r>

X-702~A -32-~drophobic shenyl-"~iepharose" ~esin A 3 . 2 x 40 cm eolu~nn of phenyl-"Sepharose" CL-4B
(Pharmacia) wa~ packed and then wash~d with 3 column 5 volumes each o~ the following eolution~ at a flow of 20 c:m . h 1 50% methanol; H2O; 196 acetic acid; H2O; 0 .1 M
NaOH; HzO.
The column was then es~uilibrated with an eguilibration buf~er containing 20 mM Tri6, 1 ~ NaCl, 10 10 m~ CaCl2, (pH 7.4). The rE3PC from Step ~ was diluted with an ~sIual volume o~ a ~olution containing 20 mM
Tris, 2 M NaCl, 20 mM t: aCl2, (p~l 7 . 4 ), arld put through the column.
The column was fur~her washed with 1 liter of 15 eguilibration buffer. The r~PC was eluted with a ~olution OI 20 mM Tris, O.15 M NaCl, 1 mM EDTA, (pH
7.4~ .
The recovery of r~PC ~t this step was ab~ut 85%. The purity is greater than 98% as measured by 20 SD~-PAGE (Laemmli, (1974) Nature, 227:680-685) or specific activity as described in Example 4. The level of endotoxin was reduGed 10 ~old after this step.

Equivalents 2~
Those skilled in ~he art will recognize, or be able to ascertain, using no more than routine experi-mentation, numerous equivalents to the specific substances and procedures described herein. Such ~; equivalents are considered to be within the scope of this invention, and are covered by the following claims.

,'~ .

-

Claims (47)

1. A method for recovering and purifying vitamin K-dependent proteins from a cell culture medium of transformed cells which produce recombinant vitamin K-dependent proteins, comprising:
a. removing divalent cations from the medium;
b. contacting the medium with a protein-binding ion-exchange resin under conditions such that the protein is bound to the resin;
c. treating the resin-bound protein with a divalent cation under conditions appropriate to form a cation-protein complex and to thereby dissociate the protein from the resin; and d. treating the cation-protein complex under conditions appropriate to remove the cation to obtain free, biologically active protein.

2. A method of Claim 1, wherein the vitamin K-dependent protein comprises activated human protein c.

3. A method of Claim 1, wherein the vitamin K-dependent protein comprises human protein C zymogen.

4. A method of Claim 1, wherein the vitamin K-dependent protein comprises human protein S.

5. A method of Claim 1, wherein the removal of divalent cations in (a) comprises adding a chelating agent to the medium.

6. A method of Claim 1, wherein the divalent cation is selected from the group consisting of ionic calcium, barium and strontium.

7. A method of Claim 1, wherein the protein-binding ion-exchange resin comprises an anionic amine-based ion-exchange resin.

8. A method of Claim 1, wherein the treatment of the cation-protein complex in (d) comprises combining a chelating agent with the complex.

9. A method for purifying vitamin K-dependent proteins from a cell culture medium of transformed cells which produce recombinant vitamin K-dependent proteins, comprising the steps of:
a. combining the cell culture medium containing the proteins with a chelating agent sufficient to remove endogenous divalent cations from the medium;
b. contacting the mixture from (a) with an ion-exchange material under conditions appropriate to effect binding of the proteins to the ion-exchange material;
c. contacting the protein-bound ion-exchange material from (b) with a source of divalent cations under conditions appropriate to form a cation-protein complex and to thereby dissociate the protein from the ion-exchange material;
d. contacting the cation-protein complex formed in (c) with chelating material under conditions appropriate to remove the cations from the complex thereby obtaining free protein;

e. purifying the protein obtained in (d) by contacting the protein with a second ion exchange material under conditions appropriate to effect binding of the protein to the ion-exchange material;
f. contacting the protein-bound ion-exchange material from (e) with a monovalent salt under conditions appropriate to dissociate the protein from the ion-exchange material;
g. contacting the protein obtained in (f) with a divalent cation sufficient to form a cation-protein complex;
h. contacting the cation-protein complex obtained in (g) with a hydrophobic material under conditions appropriate to effect binding of the cation-protein complex to the hydrophobic material; and i. contacting a chelating agent with the protein-bound hydrophobic material of (h) under conditions appropriate to remove the cations from the cation-protein complex and to thereby dissociate the protein from the hydrophobic material.

10. A method of Claim 9, wherein the divalent cation is selected from the group consisting of ionic calcium, barium and strontium.

11. A method of claim 9, wherein the protein comprises activated human protein C.

12. A method of Claim 9, wherein the protein comprises human protein C zymogen.

13. A method of Claim 9, wherein the protein comprises human protein S.

14. A method of Claim 9, wherein the chelating agent comprises EDTA.

15. A method of claim 9, wherein the ion-exchange material of (b) comprises an anionic amine-based ion-exchange resin.

16. A method of Claim 15, wherein the ion-exchange resin is packed into a column.

17. A method of Claim 9, wherein the chelating material of (d) comprises a resin having EDTA immobilized thereon.

18. A method of Claim 17, wherein the chelating resin is packed into a column.

19. A method of Claim 9, wherein the ion exchange material of (e) comprises an anionic amine-based ion-exchange resin.

20. A method of Claim 19, wherein the ion-exchange resin is packed into a column.

21. A method of Claim 9, wherein the monovalent salt of (f) comprises sodium chloride having a concentration between about 0.4 M to about 1.0 M.

22. A method of Claim 9, wherein the hydrophobic material of (h) is selected from the group consisting of phenyl "Superose" resin and phenyl "Sepharose" resin.

23. A method for separating high-specific-activity vitamin K-dependent proteins from low-specific activity vitamin K-dependent proteins contained in a cell culture medium of transformed cells which produce recombinant vitamin K-dependent proteins, comprising the steps of:
a. combining the cell culture medium containing the proteins with an amount of EDTA sufficient to remove endogenous divalent cations from the medium;
b. contacting the mixture from (a) with an ion-exchange resin under condition appropriate to effect binding of the proteins to the ion-exchange resin;
c. contacting the protein-bound ion-exchange material from (b) with a source of calcium ions under conditions appropriate to form a calcium-protein complex and to thereby dissociate the protein from the ion-exchange material;
d. contacting the calcium-protein complex formed in (c) with a resin material under conditions appropriate to remove the cations from the calcium ions from the complex thereby obtaining free protein;
e. purifying the protein obtained in (d) by contacting the protein with a second ion-exchange resin under conditions appropriate to effect binding of the protein to the ion-exchange resin;
f. contacting the protein-bound ion-exchange material from (e) with a monovalent salt under conditions appropriate to dissociate the protein from the ion-exchange resin;

g. contacting the protein obtained in (e) with a source of calcium ions sufficient to form a calcium-protein complex;
h. contacting the calcium-protein complex obtained in (g) with a hydrophobic resin under conditions appropriate to effect binding of the calcium-protein complex to the hydrophobic resin; and i. contacting the protein-bound hydrophobic material of (h) with an amount of EDTA sufficient to remove the calcium from the calcium-protein complex and to thereby selectively dissociate the high-specific activity protein from the hydrophobic resin.

24. A method of Claim 23, wherein the vitamin K-dependent protein comprises activated human protein C.

25. A method of Claim 23, wherein the vitamin K-dependent protein comprises human protein C zymogen.

26. A method of Claim 23, wherein the vitamin K-dependent protein comprises human protein S.

27. A method of Claim 23, wherein the ion-exchange resin of (b) comprises an anionic amine-based resin.

28. A method of Claim 23, wherein the hydro-phobic resin of (h) is selected from the group consisting of phenyl "Superose" and phenyl "Sepharose."

29. A method of Claim 23, wherein the monovalent salt of (f) comprises sodium chloride having a concentration between about 0.4 to about 1.0M.

30. A method for recovering and purifying vitamin K-dependent proteins from a cell culture medium of cells which produce vitamin K-dependent proteins, comprising:
a. removing divalent cations from the medium;
b. contacting the medium with a protein-binding ion-exchange resin under conditions such that the protein is bound to the resin;
c. treating the resin-bound protein with a divalent cation under conditions appropriate to form a cation-protein complex and to thereby dissociate the protein from the resin; and d. treating the cation-protein complex under conditions appropriate to remove the cation to obtain free, biologically active protein.

31. A method of Claim 30, wherein the protein is selected from the group consisting of activated human protein C, human protein C zymogen, and human protein S.

32. A method for purifying vitamin K-dependent proteins from a cell culture medium of cells which produce vitamin K-dependent proteins, comprising the steps of:
a. combining the cell culture medium containing the proteins with a chelating agent sufficient to remove endogenous divalent cations from the medium;
b. contacting the mixture from (a) with an ion-exchange material under conditions appropriate to effect binding of the proteins to the ion-exchange material;
c. contacting the protein-bound ion-exchange material from (b) with a source of divalent cations under conditions appropriate to form a cation-protein complex and to thereby dissociate the protein from the ion-exchange material;
d. contacting the cation-protein complex formed in (c) with chelating material under conditions appropriate to remove the cations from the complex thereby obtaining free protein;
e. purifying the protein obtained in (d) by contacting the protein with a second ion exchange material under conditions appropriate to effect binding of the protein to the ion-exchange material;
f. contacting the protein-bound ion-exchange material from (e) with a monovalent salt under conditions appropriate to dissociate the protein from the ion-exchange material;
g. contacting the protein obtained in (f) with a divalent cation sufficient to form a cation-protein complex;
h. contacting the cation-protein complex obtained in (g) with a hydrophobic material under conditions appropriate to effect binding of the cation-protein complex to the hydrophobic material; and i. contacting a chelating agent with the protein-bound hydrophobic material of (h) under conditions appropriate to remove the cations from the cation-protein complex and to thereby dissociate the protein from the hydrophobic material.

33. A method of Claim 32, wherein the divalent cation is selected from the group consisting of ionic calcium, barium, and strontium.

34. A method of Claim 32, wherein the protein is selected from the group consisting of activated human protein C, human protein C zymogen, and human protein S.

35. A method for separating high-specific-activity vitamin K-dependent proteins from low-specific-activity vitamin K-dependent proteins contained in a cell culture medium of cells which produce vitamin K-dependent proteins, comprising the steps of:
a. combining the cell culture medium containing the proteins with an amount of EDTA sufficient to remove endogenous divalent cations from the medium;
b. contacting the mixture from (a) with an ion-exchange resin under condition appropriate to effect binding of the proteins to the ion-exchange resin;
c. contacting the protein-bound ion-exchange material from (b) with a source of calcium ions under conditions appropriate to form a calcium-protein complex and to thereby dissociate the protein from the ion-exchange material;
d. contacting the calcium-protein complex formed in (c) with a resin material under conditions appropriate to remove the cations from the calcium ions from the complex thereby obtaining free protein;
e. purifying the protein obtained in (d) by contacting the protein with a second ion-exchange resin under conditions appropriate to effect binding of the protein to the ion-exchange resin;
f. contacting the protein-bound ion-exchange material from (e) with a monovalent salt under conditions appropriate to dissociate the protein from the ion-exchange resin;
g. contacting the protein obtained in (e) with a source of calcium ions sufficient to form a calcium-protein complex;
h. contacting the calcium-protein complex obtained in (g) with a hydrophobic resin under conditions appropriate to effect binding of the calcium-protein complex to the hydrophobic resin; and i. contacting the protein-bound hydrophobic material of (h) with an amount of EDTA sufficient to remove the calcium from the calcium-protein complex and to thereby selectively dissociate the high-specific activity protein from the hydrophobic resin.

36. A method of Claim 35, wherein the vitamin X-dependent protein comprises activated human protein C.

37. A method of Claim 35, wherein the protein is human protein C zymogen.

38. A method of Claim 35, wherein the protein comprises human protein S.

39. A method for removing non-proteinaceous contaminants from a sample of vitamin K-dependent proteins, said method comprising the steps of:
a. removing divalent cations from the medium;
b. contacting the medium with a protein-binding ion-exchange resin under conditions such that the protein is bound to the resin;

c. treating the resin-bound protein with a divalent cation under conditions appropriate to form a cation-protein complex and to thereby dissociate the protein from the resin; and d. treating the cation-protein complex under conditions appropriate to remove the cation to obtain free, biologically active protein.

40. A method of Claim 39, wherein the vitamin K-dependent protein comprises activated human protein C.

41. A method of Claim 39, wherein the vitamin K-dependent protein comprises human protein C zymogen.

42. A method of Claim 39, wherein the divalent cation-binding protein comprises human protein S.

43. A method of Claim 39, wherein the non-proteinaceous contaminant is a bacterial endotoxin.

44. A method for removing viral contaminants from a sample of vitamin K-dependent proteins, said method comprising the steps of:
a. removing divalent cations from the medium;
b. contacting the medium with a protein-binding ion-exchange resin under conditions such that the protein is bound to the resin;
c. treating the resin-bound protein with a divalent cation under conditions appropriate to form a cation-protein complex and to thereby dissociate the protein from the resin; and d. treating the cation-protein complex under conditions appropriate to remove the cation to obtain free, biologically active protein.

45. A method of Claim 44, wherein the divalent cation-binding protein comprises activated human protein C.

46. A method of Claim 44, wherein the divalent cation-binding protein comprises human protein C zymogen.

47. A method of Claim 44, wherein the divalent cation-binding protein comprises human protein S.