Direct Expression of Recombinant Activated Human Protein C, a ...

THEJOURNALOF BIOLOGICAL CHEMISTRY

Vol. 264, No. 24, Issue of August 25, pp. 14298-14304, 1989 Printed in U.S.A.

0 1989 hy The American Society for Biochemistry and Molecular Biology, Inc.

Direct Expression of Recombinant Activated HumanProtein C, a Serine Protease* (Received for publication, March 15, 1989)

Hartmut J. EhrlichSBlI, S. Richard JaskunasS, Brian W. GrinnellS, S. Betty YanS, and Nils U. BangSjII From the SLilly Research Laboratories, Indianapolis,Indiana 46285 and the §Departmentof Medicine, Indiana University School of Medicine, Indianapolis,Indiana 46223

Human protein C, like other serine proteases, is nor- animals from a lethal Escherichia coli infusion (9); and ( d ) mally secreted as an inactive zymogen. It is converted activated human protein C markedly delayed platelet accuto its activeform extracellularly bylimited proteolysis mulation and thromboticocclusion in a baboon arterial shunt with the thrombin-thrombomodulincomplex. This ac- model (10). These observationssuggest apotential therapeutic tivation results from the removal of a 12-residue acti- utility in macro- and microvascular thrombosis. vation peptide from the NH2 terminus of the heavy The structure of human protein C has been established in (COOH-terminal) chain. We report here a successful recent years by characterizing the protein purified from hustrategy for the activation of human protein C during post-translational cellular processing, resulting in the man plasma and by cDNA cloning (11, 12). As depicted in secretion of activated protein C from transfected mam- Fig. l A , the proteinC precursor molecule as deduced from the malian cells. Deletion of the nucleotides encoding the cDNA sequence starts at its NHn-terminal end with a 42activation peptide resulted in the expression of a pro- residue leader sequence composed of a signal sequence and a tease with lessthan 5%of the expected activity. How- propeptide; the latter is thought to direct y-carboxylationof ever, thereplacement of the activationpeptide with an specific glutamic acid residues (13). The leader sequence is 9 ylight chain containing the 8-residue sequence (Pro-Arg-Pro-Ser-Arg-Lys-Arg- followed bya155-residue thatarerequiredfor Arg) involved in the proteolytic processing of the hu- carboxylated glutamicacidresidues man insulin receptor precursor resulted in the direct biological activity (5). In addition, the light chain contains expression of fully activated protein C. The mutant two domains homologous to epidermal growth factor. In the protein was shown to be correctly processed by NH2- human as in the bovine protein C precursor, the light chain terminal sequence analysis. This strategy for success- is connected to the heavy chain by a Lys-Arg sequence. The ful expression of an activated form of protein C may 262-residue heavy chain comprises the COOH-terminal porapply to the expression of active forms of other pro- tion of the protein, startingat its NH2 terminus with the 12teases which are naturally expressed as zymogens. residue activation peptide followed by the COOH-terminal serine protease domain. Upon cellular processing, the leader sequence is removed, and further intrachain cleavage generHuman protein C circulates in the blood as an inactive ates a disulfide-linked heterodimer composed of a light ( M , = zymogen main!y in a two-chain form linked by one or more 21,000) and a heavy (Mr = 41,000) chain. It has been shown disulfide bonds (1).The zymogen is activatedby a complex of that the Lys-Arg connecting sequence is removed from the a-thrombin and its endothelialcell receptor thrombomodulin light chain of bovine protein C (12), andit is speculated that through the proteolyticremoval of a12-residue activation it is removed also from the light chain of mature human peptide from the NH2 terminusof the protein C heavy chain. protein C. Using oligonucleotide-directed mutagenesis, we constructed Activated protein C, a serine protease, functions as an antitwo derivatives of the human proteinC cDNA gene whichwe coagulant by proteolytically degrading coagulation cofactors VIIIa andVa (2-4). It plays a critical role in theregulation of hypothesized could be used for the direct expression of actithrombin generation (for review, see Ref. 5 ) . The importance vated protein C in mammaliancells. Fig. 1B shows the differof human protein C in the prevention of micro- and macro- ent strategiesemployed. Derivative PC(AAP)’ had thecoding vascular thrombosis is supported by the following evidence: sequence forthe 12-residue activation peptidedeleted, enjoin( a ) purpura fulminans and/ordeep veinthrombosis have been ingresidues Arg157and Led7’. DerivativePC(1RAAP) was reported to result fromhomozygous (6) andheterozygous (7) constructed using an 8-residue sequence (Pro-Arg-Pro-Serprotein C deficiencies; ( b ) the intravenous administration of Arg-Lys-Arg-Arg) substituting for the activation peptide. This activated human protein C prevents the extension of venous sequence is derived from the COOH terminus of the a-chain of the mature human insulin receptor (14).The human insulin thrombi in dogs and rhesus monkeys (8); ( c ) in ababoon receptor precursor is processed into the insulin receptor a@ model of septicshock,activatedprotein C protectedthe * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ‘‘advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ll Present address: Dept. of Molecular Biology, Central Laboratory of the Netherlands Red Cross Blood Transfusion Service,Amsterdam 1066 CX, The Netherlands. 11 To whom reprint requests should be addressed Lilly Research Laboratories, 1001 W. 10th St., Indianapolis, IN 46202.

The abbreviations used are: PCAAP, human protein C mutant lacking the activation peptide; PC(1RAAP); human protein C mutant in which the activation peptide is replaced by a specific 8-residue sequence from the a-subunit of the insulin receptor; AV12 cells, adenovirus type12-inducedSyrian hamster tumor cells; PTH, phenylthiohydantoin; SDS, sodium dodecyl sulfate; HPC, human protein C; APC, activated protein C, HEPES, 4-(2-hydroxyethyl)-lpiperazineethanesulfonic acid; HPLC, high pressure liquid chromatography.

14298

14299

Expression of Activated Human Protein C 8.

FIG. 1. Processing of the human protein C precursor and activation of human protein C ( A ) and schematic representation of the constructed mutants ( B ) .LS, 42-residue preproleadersequence; LC, 155-residue light, chain; KR, Lys-Arg dipeptide; AP, 12-residue activation peptide; aHC, heavy chain (remaining after activation); ZR, cleavage sequence Pro-Arg-Pro-SerArg-Lys-Arg-Arg derived from the insulin receptor. The dotted line represents (a) disulfide link(s) between thelight chain and theheavy chain.

~

LS

KR AP

I I

aHC

1

Cellular processing ALS, 1 K R

aHC

PC(IR3AP) I

r

heterodimer by cleavage at the a-chain COOH terminus of the 8-residue sequence. Since the insulin receptor is a molecule ubiquitous among mammalian cells, we assumed that the inclusion of this sequence in the human protein C cDNA might result in intracellular cleavage of the human protein C zymogen molecule at the desired position, leading to thedirect expression of activated protein C. In this study, we provide evidence that the expression of PC(1RAAP) in adenovirus type 12-induced Syrian hamster tumor cells (AV12 cells) results in the direct expression of activated protein C possessing full biological (amidolytic and anticoagulant) activity. EXPERIMENTALPROCEDURES

LC

I

HPC

I

LS

I

LC

aHC

KR IR

I I

I

3 aHC

C (12) was inserted into M I 3 mp I8 RF. Site-directed mutagenesis on the single-stranded template was carried out according to Zoller and Smith (18).T o construct PC(AAP), the synthetic oligonucleontide S‘GCGCAGTCACCTGAAACGACTCATTGATGGGAAGATGA 3’ was used to prime the synthesis of a DNA strand encoding a human protein C derivative lacking the 12 residues of the activation peptide. ToconstructPC(IRAAP),thesynthetic oligonucleotide

5’CGCAGTCACCTGAAACGACCCCGCCCCAGCCGCAAGCGGCGCCTCATTGATGGGAAGATG 3‘ was used to generate a gene for a human protein C derivative in which the activation peptidewas replaced by the insulin receptor sequence. The correct sequences of the mutations were verified by DNA sequencing. Construction of Expression Plasmids (Fig. 2)”The mammalian cell expression plasmid pLPCwas constructed from the following DNAs beginning at the EcoRI site and proceedingcounterclockwise: an EcoRI-PuuII fragment derived from pBR322 containing the ampicillin resistancegene and origin of replication; a PuuII-Hind111fragment containing the SV40 early promoter (SVep); a HindIII-Hind111 fragment of plasmid pBL-Cat(19) containing the BK virus enhancer and AD2 major late promoter ( B L ) ;a HindIII-Ban1 linker; theBanI-Pstl fragment of the cDNA-coding sequence for human protein C (12); a linker containing PstI-BclI-BglII restriction sites; a BglII linker ligated to the 610-base pair MboI fragment of SV40, which contains the SV40-small t splicejunction (SVTag), and the 988-base pair BclIEcoRI fragment of SV40, which contains the polyadenylationsignal (PA) and the BarnHI restriction site shown in Fig. 2. The mutated sequences were introduced into pLPC using the mutated Sstl-Sal1 fragment from the M13 RF clones resulting in plasmids pLP(AAP) and ~LPC(IRAAP). Cell Culture. Transfection. and Screening-The AV12-644 (AV12) (ATCC CRL9595) ceil line was derived from an adenovirus t h e 12: inducedtumorin a Syrianhamster.Thetumor wasinduced by subcutaneous injection of adenovirus type 12 into the scruff of the neck of a newborn hamster in 1966. The cell line derived from the tumorisepithelioidin morphology, but the cell type of origin is unknown. The 293 cell line (ATCC CRL1573) is an adenovirus type 5-transformed human embryonic kidney line. AV12 cells as well as 293 cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum, 50 pg/ml gentamicin, and 10 pg/ ml AquaMEPHYTON as an additional source of vitamin K,. One day before transfection, thecells were plated a t a density of lo6 cells/ 100-cm2culturedish. Calcium phosphate DNA precipitates were

Materials-Restriction endonucleases were purchased from New England BioLabs or Bethesda Research Laboratories. T4 DNA ligase was purchased from New England BioLabs. T4 polynucleotide kinase was from Pharmacia LKBBiotechnology Inc., and the Klenow fragment of DNA polymerase I was from Boehringer Mannheim. M13 mp 18RF and a 15-base pair Ml3-sequencing primer (No. 1200) were obtained from New England BioLabs. Dulbecco’s modified Eagle’s medium, Ham’s F-12 medium, and gentamicin were fromGIBCO; hygromycin was from Lilly, fetal calf serum was from HyClone Laboratories, and bovine serum albumin (low endotoxin) was from Miles Laboratories Inc. Assayed reference plasma, activated partial thromboplastin time reagent, and the substrate H-D-Phe-Pip-Arg-pnitroanilide (S2238) (KabiVitrum) were purchasedfromHelena Laboratories. Q-Sepharose FastFlow and QAE-SephadexA-50 were from Pharmacia. Aquamephyton, a n aqueous colloidal solution of vitamin K,,was obtained fromMerck Sharp andDohme. Biotinylated horse anti-mouse IgG and horseradish peroxidase-conjugated avidin D were from Vector Laboratories, Inc., and horseradish peroxidase color developmentreagent was from Bio-Rad. Sodium carbonate, sodium selenite, 2-aminoethanol, bovine insulin, and human transferrin were from Sigma. Murine monoclonal antibodies against humanprotein C were obtainedfrom Drs. C. T . Esmon, Oklahoma Medical Research Foundation (a-HPC-2, directed against the heavy chain of protein C, and a-HPC-4, a Ca2+-dependent antibody described indetailunder“ProteinPurification”)and J. H. Griffin, Scipps Institute (a-HPC-1, directed against theheavy chain of protein C, and CY-HPC-3,directed against the light chain of protein C). A polyclonal antibody (goatIgG) against human proteinC heavy and Eco RI light chains used in Western blot experiments was obtained from &Barn HI AmericanDiagnostica. Purifiedplasma-derivedhumanprotein C, rabbit thrombomodulin, bovine thrombin, bovine antithrombin 111, and polyclonal goat anti-rabbit thrombomodulin antibody covalently coupled to Affi-Gel 10 beads were gifts from Drs. C. T. Esmon and N. L. Esmon. The adenovirus type 12-induced Syrian hamster tumor line AV12-664 was obtained from Dr. 3. P. Burnett, Lilly Research Laboratories. General Methods-Plasmid DNA was isolated by a modification of the Triton X-100 lysis procedure (15) followed by two consecutive CsCl equilibrium centrifugations. Enzyme reactions were carried out using standard conditions (16). Nucleotide sequence determinations . 111 I Hind were performedusing the dideoxy method of Sangeret al. (17). Synthetic oligonucleotides weresynthesized by solid-phase phosphorFIG. 2. Vector for the expression of protein C mutants. For amidite chemistry on an automated synthesizer(Applied Biosystems details, see “Experimental Procedures.” kb, kilobases; SVTag, SV40model 380B). small t splice junction; PA, polyadenylation signal; Amp, ampicillin Oligonucleotide-directed Mutagenesis-A 710-base pairSstI-Sal1 resistance gene; ori, origin; SVep, SV40 early promoter;BL, BK virus restriction fragmentfrom a fulllength cDNA clone for human protein enhancer sndAD2 major late promoter. ”

/‘

14300

Expression of Activated Human ProteinC

prepared (20) with 25 pg of expression plasmid and no carrier DNA. In transient expression experiments, serum-free medium (3 parts of Dulbecco's modified Eagle's medium and 1 part of Ham's F-12 medium containing 1 mM sodium selenite, 100 p~ 2-aminoethanol, 20 mM HEPES, 2.4 g/liter sodium bicarbonate, 1 pg/ml transferrin, 1 pg/ml insulin, 100 pg/ml bovine serum albumin, 50pg/ml gentamicin, and 10 pg/ml vitamin K,) was added to the cells 24 h after transfection. After an additional 48 h, the conditioned serum-free medium was collected and 10-fold concentrated using CentriconTM(Amicon Corp.) cartridges. To establish stable clones producing human protein C derivatives, 5 pg of plasmid pSV2hyg (21) was used for cotransfection, and drugselection (200 pg/ml hygromycin) wasapplied starting 2-3 days after transfection. Upcoming clones resistant tohygromycin were isolated 10-14 days after initiating the drugselection, grown in 24-well plates, and monitoredfor production of protein C antigen. A sandwich type enzyme-linked immunosorbent assay with a sensitivity of 5ng/ml was used to measure the level of recombinant human protein C derivatives in culture media (2 liters).Monoclonal antibody a-HPC-3 directed against the light chain of human protein C was used for coating flat bottomed microplate wells. After incubating a HPC-3 with the sample, biotinylated monoclonal antibody a-HPC-1 directed against the heavy chain of human protein C was added for detection by horseradish peroxidase-conjugatedavidin D. Purified plasma-derived human protein C and activated protein C from the same source served as standards. Protein Purification-Twenty mg of human plasma protein C was purified from 26 liters of fresh frozen plasma essentially as described elsewhere (20). Briefly, the plasma was rapidly thawed and adjusted to 20 mM Tris-HC1, pH 7.4, 5 mM benzamidine, and 2 mM EDTA. After batchwise adsorption onto 50 ml of QAE-Sephadex A-50, the supernatant was decanted, and the gel was packed into a column. The column was washed with 5 column volumes of 20 mM Tris-HC1, p H 7.4, containing 150 mM NaCl and 5 mM benzamidine before the bound human protein C was eluted with 20 mM Tris-HC1, pH 7.4, containing 0.5 M NaCland 5 mM benzamidine. Theeluate was adjusted to10 mM CaC12and 10 units/ml heparin, and human protein C was further purified by immunoaffinity chromatography using the immobilized murine monoclonal antibody a-HPC-4. a-HPC-4 is a Ca2+-dependentconformation-specific antibody recognizing an epitope comprising the activation peptide. Detailed characterization of a-HPC-4 is published elsewhere (22). Human protein C was eluted with 20 mM Tris-HC1, pH 7.4, containing 150 mM NaCl and 4 mM EDTA. Recombinanthumanprotein C andPC(IRAAP), produced by AV12 cells, were purifiedfrom serum-free media as follows. The culture media (21) were centrifuged (10,000 X g, 20 min, 4 "C) to remove cellular debris, then adjusted to 4 mM EDTA and adsorbed batchwise (2 h, 4"C) onto2 ml of Q-Sepharose FastFlow, equilibrated with 20 mM Tris-HC1, pH 7.4, 150 mM NaC1, 4 mM EDTA. After decanting the supernatant, the matrixwas packed into a column (1.0 X 5.0 cm), washed with 5 column volumns of equilibration buffer, and eluted with 20 mM Tris-HC1, pH 7.4, containing 0.5 M NaCl and 10 mM CaC12.Recombinant human protein C was purified to homogeneity using immobilized cu-HPC-4 under the conditions described forhumanprotein C fromplasma except thatadjustment with heparin and CaCI2 was not necessary. Since PC(1RAAP) is lacking the activationpeptide, immunoaffinity purificationwith immobilized e-HPC-4 was not possible. Therefore, all functional activity measurements were carried out with the partially purified material eluted from Q-Sepharose FastFlow. For NH2-terminal sequence analysis, PC(1RAAP) was further purified by bariumcitrateadsorptionandelution(23) followed by reverse-phase HPLC. 550 p1 of 250 mM sodium citrate and 1100 p1 of 450 mM BaC12 were added to 5 ml of eluate from Q-Sepharose Fast Flow, and the mixture was rotated end over end for 1 h a t 37 " c . After centrifugation (7000 X g, 10 min), the pellet was washed twice with 5 ml of 0.05 M sodium acetate before it was resuspended in 1.6 ml of 150 mM sodium citrate. This solution was concentrated and buffer exchanged to 50 mM NH,HCO, in Centricon, 10 microconcentrators (Amicon), and then loaded onto a C3 reverse-phase HPLC column (Beckman). The protein was eluted with a linear trifluoroacetic acid/acetonitrilegradient.AutomatedEdmandegradation chemistry was performed with a gas-phase sequenator (Applied Biosystems model 470A) with an on-line HPLC system (model 120) for the analysis of the phenylthiohydantoins (PTH). Western Blot Analysis-Human protein C and PC(1RAAP) (0.5 pg/lane) were subjected to SDS-polyacrylamide gel electrophoresis using a Laemmli system (24) with a 10% polyacrylamide (acrylfiis,

30:0.8) separating gel. After the proteins had been transferred to nitrocellulose, the filter was blocked for 1 h with washing buffer (50 mM Tris-HC1,150 mM NaCl, p H 7.4) containing 5% nonfat dried milk. The filter was incubated for 1h with twomonoclonal antibodies added together (a-HPC-1 and a-HPC-2, both exhibiting human protein C heavy chain specificity), then with biotinylated horse antimouse IgG, and finally developed with horseradish peroxidase-conjugated avidin D. The light chainof human protein C or PC(1RAAP) could not be detected by this procedure. We were also not able to detect the light chain using apolyclonal antibody against human protein C or the light chain-specific monoclonal antibody a-HPC-3 unless very high concentrations (>2 pgllane) of human protein C were run in the SDS-polyacrylamide gel electrophoresis (data not shown). This is consistent with a previous report (25), suggesting a weak antibodyreactivitytothedenaturedand immobilized light chain. Activation of Human Protein C-The activation of plasma-derived and recombinant human protein C zymogen followed the same protocol and was accomplished by immobilized thrombomodulin-thrombin asdescribed (26). Briefly,noninhibitory murine monoclonal antirabbit thrombomodulin antibody broadly cross-reacting with human thrombomodulin was coupled to Affi-Gel 10 ( 5 mg of antibody/ml of gel). Five ml of gel was washed extensively and incubated ona rocker platform for 60 min a t 37 "C with 700 pg of purified rabbit lung thrombomodulin in 5 ml of 20 mM Tris-HC1, p H 7.4, 150 mM NaCl, 0.01% Lubrol P X (w/v). More than 90%of thethrombomodulin bound to theimmobilized antibody. After washing, the gel was incubated for 1 h a t 37 "C with bovine thrombin (168 pg) in 5 ml of 20 mM Tris-HC1, p H 7.4, 150 mM NaC1, 1mg/ml bovine serum albumin. The added thrombin bound quantitatively to thegel (>95%). Again, the beadswere washed extensively and then stored at4 "C in 20 mM Tris-HC1, pH 7.4, 150 mM NaC1. Activation was achieved by adding 25 p1 of sedimented beads to 500 plof sample (thrombomodulinthrombin complex concentration, 50 nM) in 20 mM Tris-HC1, p H 7.4, 150 mM NaC1, 3 mM CaC12, 1 mg/mlbovine serum albumin and incubating 2 h a t 37 "C on a rocker platform. After centrifugation, the supernatant was removed and incubated for 15 mina t 20 "C with bovine antithrombin I11 (4 units/ml) prior to assayingfor functional activity. Determination of the Functional Activities of Human Protein CThe amidolytic activities of activated protein C and PC(1RAAP) were determined by hydrolysis of the synthetic substrate H-o-Phe-PipArg-p-nitroanilide (S2238) as follows. Duplicate samples (100 pl) were diluted with 600 p1 of 20 mM Tris-HC1, p H 7.4, containing 100 mM NaCland 1 mg/mlbovine serumalbumin. 300 pl of a 2.25 mM solution of S2238 (saturatingconcentration) wasadded, and the release of p-nitroaniline was determined at 25 "C by recording the increase in absorbance at 405 nm over 5 min on a Gilson response spectrophotometer. One unit of activated protein C was defined as theamount of enzyme required forthe release of 1 pmol of p nitroaniline in 1 min a t 25 "C, pH 7.4, using an extinction coefficient for p-nitroaniline at 405 nm of 9620 M" Cm" (27). The anticoagulant activities of activated proteinC and PC(1RAAP) were determined by measuring the prolongationof the clotting time in the activated partial thromboplastin time clotting assay. 100 p1 of assayed reference plasma was incubated with 100 pl of reconstituted activated partial thromboplastin time reagent a t 37 "C for 5min. Then 100 p1 of buffer (20 mM Tris-HC1, p H 7.4, containing 150 mM NaC1, 3 mM CaC12,1 mg/ml bovine serum albumin) with or without activated protein C was added, immediately followed by the addition of 100 p1 of a 25 mM CaCl? solution. The timer of the fibrometer (BBL Fibrosystem) was started upon CaCL addition, and the time requiredforclot formation was recorded.A standard curve was prepared using serial dilutions of activated protein C from plasma. Duplicate determinationswere run for each test sample and converted to the concentrationof functionally active protein.Activated protein C from plasma and AV12 cells as well as PC(1RAAP) followed the same dosedependence. The equivalent of4 pgof plasma human protein C, present normally in 1 ml of plasma (28), was defined as 1 unit of anticoagulant activity, thereby defining 250 units/mg for the plasma standard. RESULTS

Transient Expression-Expression plasmids were constructed for h u m a n protein C zymogen,PC(AAP), and P C ( 1 R A A P ) and designated pLPC, p L P C ( A A P ) , and

Expression of Activated Human Protein C

14301

containing pLPC(AAP) secreted protein C antigen up to 1.0 pg/ml following growthin roller bottles,butthe specific amidolytic activity (hydrolysisof S2238) was less than 5% of the plasma-derived activated proteinC. The human protein C derivative produced by these cells contained two chains linked by disulfide(s) which had approximately the same size as the two chains of the plasma human protein C zymogen (not shown). These results indicated the protein was secreted in largely inactive form that 30 60 90 120 could not be activated. Therefore, we concentrated on the Time (min) characterization of PC(IRAAP), which was expressed as an FIG.3. Expression of amidolytic activitybyAV12 cells activated serine protease in boththe 293 and AV12 cell lines. transiently transfected with pLPC(humanprotein C zymogen) Because of higher expressionlevels, PC(1RAAP) was purified (0).pLPC(AAP) (A), and pLPC(1RAAP) (0).The conditioned from recombinant AV12 cells for further characterization. serum-free media were mixed directly ( A ) or after incubation with Characterization of pLPCfIRAAP)-PC(IRAAP) was purithrombomodulin-thrombin ( B )with S2238,and the increaseinAlos., fied from conditioned serum-free media on Q-Sepharose Fast was observedover2 h. A,a5., readingsobtainedwithserum-free Flow with a recovery of a t least 70%. Western blot analysis media from MOCK-transfected cells were subtracted from all data. comparing PC(1RAAP) with human protein C zymogen and pLPC(IRAAP), respectively (see Fig. 1B). AV12 as well as activated protein C is depicted in Fig. 4. Under nonreducing 293 cells were used for the expression of these derivatives. conditions (Fig. 4A), protein C zymogens from plasma (lane These cell lines were chosen because we had found that they 1 ) and AV12 cells transfected with pLPC (lane 3 ) migrated slightly slower in the SDS-polyacrylamidegel electrophoresis could be efficiently transfected with pLPC, resulting in the C following secretion of fully biologically active recombinant human pro- than plasma-derived and recombinant protein activation with thrombomodulin-thrombin (lanes 2 and 4 ) . tein C zymogen from either 293 cells (21) orAV12 cells. The ability to produce human protein C zymogen, which, Purified PC(1RAAP) (lane 5 ) exhibited a mobility identical C from AV12 cells (lane 4 ) . The presence upon activation, expresses full biological (anticoagulant) ac- to activated protein mobility of tivity, indicatesthat bothcell lines are capableof performing of bovineserumalbumininterferedwiththe (lane 6). Under all the required post-translational modifications of human PC(1RAAP) from crude serum-free media reducing conditions (Fig. 4B),the heavy chain of plasma protein C such as y-carboxylation. T o study the properties of these PC derivatives, 293 and human protein C zymogen migrated as a doublet (lane 1) as AV12 cells were transfected with plasmids, and conditioned reported previously (1).Following activation by thrombomodserum-freemedium wascollected after 72 h. Analysis of ulin-thrombin, theheavy chain doublet had a faster mobility human protein C antigen in the medium with an enzyme- (lane 2). Identical results were found for human protein C linked immunosorbent assay demonstrated that the level of zymogen and activated protein C from recombinant AV12 expression for each plasmid was about 10-50 ng/ml. In con- cells transfectedwithpLPC (lanes 3 and 4 ) . The slower trast, no antigen was detected when cells the weretransfected mobilities of the heavy chains from the zymogens are due with a control plasmid that lacked the human protein C cDNA presumably to the 12 residues still being intact on the NH, terminus.The heavy chaindoublet of PC(1RAAP) before gene. The amidolytic activities of the protein C derivatives se- (lane 5 ) and after (lune 6 ) purification displayed a mobility creted by AV12 cells before and after incubation with throm- similar to theheavy chains of activated protein C from AV12 of human protein bomodulin-thrombin are shown in Fig. 3, A and B, respec- cells (lane 4 ) . In contrast to the native form tively. Cells transfected with pLPCproduced a zymogen that C (lanes 1 and 3 ) , no single-chain material could be detected was only active after incubation with thrombomodulin-thromCrude CNd8 bin, as is human plasma protein C. Cells transfected with Purif. Purif. Purif. Media Purif. Purif. Purif. Media 1 n --n r.' pLPC(AAP) produced a protein with onlya trace of amidol- + - + + + ytic activity, with or without incubation with thrombomodun n n n n n nnnnn T lin-thrombin. In contrast, AV12 cells transfected with LDm kDm pLPC(1RAAP) produced anactiveproteasethatsponta6868neously hydrolyzed S2238 (Fig. 3 A ) . Thisenzymehad a specific activity comparable to the thrombomodulin4343 thrombin-activated enzyme from pLPC-transfectedcells and was not further activated by incubation with thrombomodulin-thrombin (Fig. 3 B ) . Similar results were obtained with 2626293 cells transfected with pLPC, pLPC(AAP), and pLPC(1RAAP) (data not shown). Stable Expression-To obtain enough recombinant protein for further biochemical characterization, stable AV12 or 293 18 18 cell lineswere established by cotransfection of the HPC expression plasmids with pSV2hyg coding for resistance to 1 2 3 4 5 6 1 2 3 4 5 6 hygromycin as the selectable marker.Among the cells transfected with pLPC, pLPC(AAP), and pLPC(IRAAP), approx- FIG.4. Western blot analysis of human protein C, activated imately 40-50% of the hygromycin-resistant clonesproduced protein C, and PC(1RAAP).A, nonreducing conditions;B, reducing human protein C derivatives (20-300 ng/ml in 24-well plates). conditions; -; no activation, +, following activation with thrombomodulin-thrombin. Lanes I and 2, purified recombinant human proThe amidolytic activities of the proteins secreted by these tein Cfrom AV12 cells transfectedwithpLPC; lane 5, partially stable transformants closely resembled the results from the purified PC(1RAAP) from AV12 cells; lane 6, PC(1RAAP) from crude transient expression experiments shown in Fig. 3. The cells conditioned serum-free media of AV12 cells. P

c

1.6

A

-1

C

Expression of Activated Protein Human

14302

with the derivative (lunes 5 and 6) based on carbohydrate analyses. The slight difference in migration between plasma and recombinant HPC or APC appears to be due to differences in glycosylation (data not shown). The functional properties of Q-Sepharose Fast Flow-purified PC(1RAAP) are compared with reference human plasmaderived activated protein Cenzyme in Table I. The activated protein C anticoagulant activity was measured by prolongation of the activated partial thromboplastin time which results from inactivation of coagulation cofactors Va and VIIIa. PC(1RAAP) was fully functional without priorthrombomodulin-thrombin activationas determined by both its amidolytic and anticoagulant activities. The specific anticoagulant activity of both AV12 cell-derived thrombomodulin-thrombin-activated human protein (280 C t 40 units/mg) and PC(1RAAP) (240 k 40 units/mg) was comparable to purified thrombomodulin-thrombin-activated plasma human protein C reported here (250 units/mg) and in the literature (21). The ability to secrete fully functional protein C from 293 and AV12 cells was dependent on the addition of vitamin K to the culture medium. After three to four passages of either AV12 or 293 cell lines in the absence of vitamin K, the anticoagulant activity of recombinant HPC decreased sharply. Similarly, treatment of the cultures with the inhibitors 2-chloro-3-phy) tyl-1,4-naphthoquinone (10 pg/ml) or warfarin (10 p ~ resulted inthe disappearance of functional anticoagulant activity (data notshown). PC(1RAAP) was purified further by barium citrate absorption/elution followed by reverse-phase HPLC and subjected to NH2-terminal sequence analysis. Two residues were released (Table Ii) in each of the first five cycles in approximately equimolar amounts, taking into account the known recoveries of PTH derivatives. These results indicated the presence of two NH2 termini, as expected for a two-chain molecule. The order of release of the residues corresponded to theknown NH2-terminalsequences of the activatedprotein C heavy and light chains. Only 1 residue was detected in cycles 6, 7, 9, 10, and 12, which can be explained as foHows. The light and heavy chains of activated protein C contain the same residues in positions9 and 12(arginine and serine, respectively); the sequenator does not recover the y-carboxylated PTH-Glu residues in positions 6 and 7 of the light

chain; the PTH-His from position 10 of the heavy chain was obscured by background. The observed sequence could not have been generated by proteolytic cleavage at any other position in PC(1RAAP). Thus,the existence of two NH2 termini identical to activatedproteinC was verified for PC(IRAAP), which indicates that cleavage had occurred at the desired position. DISCUSSION

Many mammalian serine proteases suchas human protein C areexpressed as inactive zymogens. The activation of these zymogens through limited proteolysis characteristically occurs in theextracellular milieu. In thisreport, we investigated whether activation of a zymogen by post-translational cellular processing is possible. To thisend, we constructed two human protein C derivatives that were designed to produce directly activated protein C prior to or during secretion from mammalian cells. PC(AAP) exhibited very little protease activity before and after treatmentwith thrombomodulin-thrombin. The reason why this protein was inactive and could not be activitated even though it contained two chains of about the correct size is not known. However, the PC(1RAAP) derivative was expressed as a molecule with the size, the two NH2 termini, and the functional characteristics of activated protein C. Without activation, the specific amidolytic activity of PC(1RAAP) was comparable to that of thrombomodulin-thrombin-activatedhuman protein C purified from plasma or conditioned medium from recombinant AV12 cells. Furthermore, the anticoagulant activities of PC(1RAAP) and native human protein C produced by AV12 cells were similar. This, to our knowledge, represents the first successful attempt to express directly an activated serine protease in a mammaliancell system. The biochemical characterization of PC(1RAAP) revealed no detectable differences from activated human protein C from the recombinant protein Czymogen. As indicated above, the mobility differences between the recombinant and plasma-derived zymogens and activated protein C molecules appear to be due to differences in carbohydrate content. The heavy chain of PC(1RAAP) appeared to have the same mobility as the heavy chains of the AV12-derived control-activated protein C, which has a faster mobility in the gel comTABLE I pared with the heavy chain of nonactivatedrecombinant human protein C. This observation is consistent with the Functional actiuity of human protein C and PC(ZRMP) insulin receptor sequence being released from the heavy chain Activation withFunctionalactivity of the molecule. Furthermore, sequence analysis of Protein thrombomodulipthrombin Amidolytic AnticoagPC(1RAAP) demonstrated that theheavy chain hadthe same ulant NH2-terminal end as the heavy chain of APC, which indicates units/pg that cleavage of the insulin receptor sequence had occurred Yes 25 k 3 250 Human protein C, plasma at the desired position at its COOH-terminal end. COOHYes 24 ? 3 280 f 40 Human protein C, AV12 cells terminal sequence analysis willbe necessary to determine No 27 f 3 240 f 40 PC(IRAAP), AV12 cells whether the Lys-Arg dipeptide and the insulin receptor seRefers to thebiological activity of the oneplasma-derived purified quence are removed from the COOH terminus of the light reference standard APC preparation used throughout thestudies. TABLEI1 Sequence analysis of purified PC(IRaAP) compared with the sequence described for APC (1 1)

light APC chain APC heavy chain -

-

Cycle

Ala Leu Ile

Asn

1

2

Leu PTH-derivative Leu Phe Ser 1 Asn Ala Yield 1 (pmol) 58 PTH-derivative 2 Arg Thr Met Lys Gly Asp Ile Leu Yield 2 (Dmol) 51 51

His

Ser Phe Leu Gla Leu Arg Asp

Gly Lys Met Thr Arg Gly

3

4

5

6

l

a

9 Arg

28

9

26

27

35

* Same PTH-derivatives in both cycles.

34

16

2616

16

Ser Pro

11

12

13

6

21 Pro 20

Leu Ser Ser

25

* 20

10

Ser Leu Asp

3

Asp Gly 22

21

Expression of Activated Human Protein C

14303

chain. In addition, the NH2-terminalsequence analysis indi- quantities of) human protein C with immobilized thrombocated that AV12 cells are able to cleave the preproleader modulin-thrombin is cumbersome. However, in vivo studies necessary to determine the bioequivalence of sequence at the correct position,which is not always the case will be in the expressionof recombinant proteins (29). The fact that PC(1RAAP) and activated proteinC. residues 6 and 7 of the light chain were not seen in the sequence analysis suggests that the glutamate residues were Acknowledgments-We thank Richard M. VanFrank and Mark L. Slisz for the NHZ-terminal sequence analysis; Dr. G. Sitta-Sittampost-translationallyy-carboxylatedandthereforenotrepalam, David R. Smith, and Joni Heller for performing the enzymecovered during the cycles, which was expected since linked immunosorbent assay; and Dr. Hansen Hsiung for preparing PC(1RAAP) exhibited full biological activity. synthetic oligonucleotides. We thank Drs.Charles T. Esmon and As indicated above, the functional anticoagulant activity of Naomi L. Esmon forgenerously providing monoclonal and polyclonal human protein C depends on the correct post-translational antibodies, purified rabbit thrombomodulin, bovine thrombin, and processing, especially on thelevel of vitamin K-dependenty- antithrombin 111; and Dr. John H. Griffin for providing the monocarboxylation,and few cell linesappearto be capable of clonal antibodies used in this study. We gratefully acknowledge Dr. J. Paul Burnett for supplying the AV12-664 cell line used in these producing fully carboxylated proteins. Attempts to express functional recombinant human proteinC in Chinese hamster studies. Wethank Dr.Donald B. McClure for performing experiments demonstrating the requirement for vitamin K for expression of full ovary (30) and mouse C127 (31) cells, and human factor IX, anticoagulant activity of recombinant HPC. The expert technical another vitamin K-dependent protein, in rat (32) and human assistance of Leslie L. Daugherty, Jenna Walls, and David T. Berg is (33) hepatoma cells,mouse fibroblasts(33), baby hamster gratefully acknowledged as is the help of Sharon Jackson in typing kidney (34), and Chinese hamster ovary (35) cells have re- the manuscript. sulted in the secretion of either partially carboxylated (and REFERENCES therefore only partially active) or a mixture of partially active and fully active protein molecules. Jorgensen et al. (36) re1. Kisiel, W. (2979) J. Clin. Inuest. 6 4 , 761-769 ported the expression of fully y-carboxylated prothrombin in 2. Walker, F. J., Sexton, P. A. & Esmon, C. T. (1979) Biochim. Chinese hamster ovary cells but not above expression levels Biophys. Acta 571, 333-342 3. Suzuki, K., Stenflo, J., Dahlback, B. & Teodorsson, B. (1983) J. of 100ng/106cells/day. We reportedpreviously that Bid. Chem. 258,1914-1920 recombinant human protein C expressed in the adenovirus4. Fulcher, C. A., Gardiner, J . E., Griffin, J. H. & Zimmerman, T. transformed 293 cell line was fully carboxylated and hadfull S. (1984) Blood 63,486-489 anticoagulant activity (21).We also have found that AV12 the 5. Esmon, C. T. (1987) Science 2 3 5 , 1348-1352 cell line iscapable of secreting fully carboxylated recombinant 6. Seligsohn, U., Berger, A., Abend, M., Rubin, L., Attias, D., Zivelin, human protein C.* Thus, these two cell lines were used for A. & Rapaport, S.I. (1984) N. Engl. J. Med. 310,559-562 the expression studies to assure that the recombinant human 7. Griffin, J. H., Evatt, B., Zimmerman, T. S., Kleiss, A. J. & Wideman, C. (1981) J. Clin. Znuest. 68, 1370-1373 protein C produced would be adequately carboxylated. S. C., Murayama, H., Yan, S. B., Long, G . L., Harms, At present, the nature of the putative change in the acti- 8. Emerick, C. S., Marks, C. A., Mattler, L. E., Huss, C.A., Comp, P. C., vated protein C serine protease domain resulting from the Esmon, N. L., Esrnon, C. T. & Bang, N. U. (1987) in UCLA release of the activation peptide is notknown. However, the Symposia on Molecular and Cellular Biology (Holcenberg, J. S. presence of the normal activation peptideis not required for & Winkelhake, J. L., eds) Vol. 65, pp. 351-367, Alan R. Liss, Inc., New York the molecule to fold in such a way that it finally will activate since the &residue sequence derived fromthe insulin receptor 9. Taylor, F. B., Chang, A,, Esmon, C. T., D’Angelo,A., ViganoDAngelo, S. & Blick, K. E. (1987) J . Clin. Inuest. 79, 918-925 could replace theactivationpeptide.Theinsulinreceptor 10. Gruber, A., Griffin, J . H., Harker, L. A. & Hanson, S.R. (1989) sequence, in contrast to the normal activation peptide, was Blood 73,639-642 released from the NH2 terminus of the heavy chain during 11. Foster, D. & Davie, E. W. (1984) Proc. Natl. Acad. Sci. U. S. A. post-translational cellular processing.Plasma factorswere not 81,4766-4770 required for this reaction since the activatedform was found 12. Beckmann, R. J., Schmidt, R. J., Santerre, R. F., Plutzky, J., Crabtree, G. R. & Long, G. L. (1985) Nucleic AcidsRes. 13, in conditioned serum-freemedia. 5233-5247 This is consistentwiththehypothesisthattheinsulin 13. Suttie, J. W., Hoskins, J. A., Engelke, J., Hopfgartner,A., Ehrlich, receptor precursor and PC(1RAAP) are being cleaved by the H., Bang, N. U., Belagaje, R. M., Schoner, B. & Long, G. L. same trypsin-like protease. This enzyme has not yet been (1987) Proc. Natl. Acad. S c i U. S. A. 8 4 , 634-638 identified. However, it seems to be very specific since in a 14. Ebina, Y., Ellis, L., Jarnagin, K., Edery, M., Graf, L., Clauso, E., Ou, J., Masiarz, F., Kan, Y. W., Goldfine, I. D., Roth, R. A. & patient carryinga point mutation in the insulin receptor gene Rutter, W. J. (1985) Cell 40, 747-758 which alters the residue sequence of the processing site from Arg-Lys-Arg-Arg to Arg-Lys-Arg-Ser, cleavage of the prore- 15. Kahn, M., Kolter, R., Thomas, C., Figurski, D., Meyer, R., Remaut, E. & Helinski, D. R. (1979) Methods Enzymol. 6 8 , 268ceptor does not occur, resulting in insulin-resistant diabetes 280 (37). 16. Maniatis, T.,Fritsch,E. F. & Sambrook, J . (1982) Molecular Our experiments to express directly activated protein C Cloning: A Laboratory Manual, Cold Spring HarborLaboratory, Cold Spring Harbor, New York show that enzymerecognitionsequences suchastheone derived fromthe insulin receptor precursor may betransferred 17. Sanger, F., Necklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci. U. s.A. 74,5463-5467 from one protein to another. This is in agreement with the 18. Zoller, M. J. & Smith, M. (1984) DNA (NY) 3 , 479-489 model proposing that the specificity of a cleavage site is 19. Grinnell, B. W., Berg, D. T. & Walls, J. (1986) Mol. Cell. Biol. 6, primarily determined by the primary structure immediately 3596-3605 preceding this site. Our strategy may be employed to express 20. Wigler, M., Pellicer, A., Silverstein, S., Axel, R., Urlaub, G. & directly activated forms of other proteases. PC(1RAAP) exChasin, L. (1979) Proc. Natl. Acad. Sci. U. S. A. 76, 1373-1376 pressed by mammalian cells could be used as a pharmaceutical 21. Grinnell, B. W., Berg, D. T., Walls, J. & Yan, S. B. (1987) Bio/ Technology 5 , 1189-1192 for the prevention and treatment of micro- and macrovascUlar D. J., Kurosawa, S., Sims, P. J., Esmon,N. L. & Esmon, thrombosis, especially since the in vitro activation of (large 22. Steams, C. T. (1988) J. Biol. Chen. 263,826-832 23. Fujikawa, K. & Davie, E. W. (1976) Methods Enzymol. 4 5 , 89H. J. Ehrlich, S. R. Jaskunas, B. W. Grinnell, S. B. Yan, and N. U. Bang, unpublished observation.

95

24. Laemmli, U. K. (1970) Nature 227, 680-685

14304

Expression of Activated Human ProteinC

25. Foster, D. C., Rudinski, M. S., Schach, B. G., Berkner, K. L., Kumar, A. A., Hagen, F. S., Sprecher, C.A., Insley, M. Y. & Davie, E. W. (1987) Biochemistry 26, 7003-7011 26. Vigano-D’Angelo, S., Comp, P. C., Esmon, C. T. & D’Angelo, A. (1986) J. Clin. Znuest. 77,416-425 27. Pfleiderer, G. (1970) Methods Enzymol. 19, 514-521 28. Miletich, J., Sherman, L. & Broze, G. (1987) N. Engl. J. Med. 317,991-996 29. Holmes, W. E., Lijnen, H. R. & Collen, D. (1987) Biochemistry 26, 5133-5140 30. Suttie, J . W. (1986) Thromb. Res. 44, 129-134 31. Oppenheimer, C. & Wydro, R. (1988) in Current Aduances in Vitamin K Research (Suttie, J. W., ed) pp. 165-171, Elsevier

Science Publishing Co., New York

32. Anson, D. S., Austen, D. E. G. & Brownlee, G. G. (1985) Nature 3 15,683-685 33. Busby, S.,Kumar, A., Joseph, M., Halfpap, L., Insley, M., Berkner, K., Kurachi, K. & Woodbury, P. (1985) Nature 316, 271273 34. de la Salle, H., Altenburger, W., Elkaim, R., Dott, K., DieterlC, A., Drillien, R., Cazenave, J.-P., Tolstoshev, P. & Lecocq, J.-P. (1985) Nature 316, 268-270 35. Kaufman, R. J., Wasley, L. C., Furie, B. C., Furie, B. & Shoemaker, C. B. (1986) J.Biol. Chem. 261, 9622-9628 36. Jorgensen, M. J., Cantor, A. B., Fury, B. C. & Fury, B. (1987) J. Biol. Chem. 262,6729-6734 37. Yoshimasa, Y., Seino, S., Whittaker, J., Kakehi, T., Kosaki, A., Kuzuya, H., Imura, H., Bell, G. I. & Steiner, D. F. (1988) Science 240, 784-787