THE .JOURNAL M 1987 by The American Society of Biological Chemists,

OF BIOLOGICAL CHEMISTRV Inc

Vol . 262, No. 7, Issue of March 5, pp. 3285-3290,1987 Printed Ln U. S.A.

Characterization of Recombinant Human Factor VIII* (Received for publication, June 13, 1986)

Dan L. Eaton, Philip E. Hass, Lavone Riddle$, Jennie Matherg, Mike Wiebes, Timothy Gregory$, and Gordon A. Vehar From the Departments of Molecular Biology, $Process Development, and §Research and Deuelopment, Genentech, Znc. South San Francisco, California 94080

Recently, complete human factor VI11 DNA clones have been obtained and subsequently expressed in baby hamster kidney cells (Wood, W. I., Capon, D. J., Si- monsen, c. c., Eaton, D. L., Gitschier, J., Keyt, B., Seeburg, P. H., Smith, D. H., Hollingshead, P., Wion, K. L., Delwart, E., Tuddenham, E. G. D., Vehar, G. A., and Lawn, R. M. (1984) Nature 312, 330-337). The recombinant factor VI11 (rVIII) protein secreted from these cells has now been purified allowing its struc- tural analysis and comparison to plasma-derived factor VI11 (pdVIII). Analysis of purified rVIII by sodium dodecyl sulfate-polyacrylamide gel electrophoresis shows that it consists of multiple polypeptides with relative mobilities (M,) ranging from 80,000- 210,000. The same pattern of polypeptides is also ob- served for pdVIII resolved by sodium dodecyl sulfate- polyacrylamide gel electrophoresis. The proteins as- sociated with rVIII are recognized by pdVIII antibodies in a Western blot. When rVIII and pdVIII are subjected to isoelectric focusing they are resolved into a similar pattern of protein bands. Thrombin, factor Xa, and activated protein C, which modulate factor VI11 activ- ity by proteolysis, process rVIII in the same manner they do pdVIII. As is the case for pdVIII, thrombin activation of rVIII coagulant activity correlates with the generation of subunits with M, of 73,000, 50,000 and 43,000. These subunits appear to form a metal- (perhaps Ca2+) linked complex. EDTA inactivates thrombin-activated rVIII and pdVIII, with the activity being regenerated after the addition of a molar excess of MnC12. The results suggest that rVIII is structurally and functionally very similar to pdVIII.

~

Factor VI11 (antihemophilia factor) functions in the middle of the coagulation cascade, acting as a cofactor for factor X activation by factor IXa in the presence of calcium ions and phospholipid (1). Because of the low levels of factor VI11 in plasma (200 ng/ml), its apparent instability, and its associa- tion with von Willebrand Factor, attempts to purify and characterize the factor VI11 protein were hindered for many years. However, the cloning of factor VI11 cDNA (2, 4, 15) and the purification and characterization of factor VI11 pro- tein from several species (5-9) has now provided a detailed understanding of the structure of factor VIII.

The amino acid sequence deduced from human factor VI11 cDNA clones predicts a mature single-chain protein (2,332 amino acids) that, after accounting for 25 potential N-linked glycosylation sites, would have a M , of -300,000 (2, 3, 15).

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisernent” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

” __

This single-chain form is readily proteolyzed both in vivo and in uitro. As isolated from plasma or commercial concentrates, human factor VI11 consists of multiple polypeptides with M , -80,000-210,000 (6,8,9). Analysis of these proteins by amino acid sequencing shows that the M, 210,000 and 80,000 proteins represent the amino and carboxyl-terminal regions of factor VIII, respectively (2, 9, 13). Further proteolytic processing within the carboxyl-terminal segment of the M , 210,000 pro- tein yields a series of proteins of M, 90,000-188,000 (9). Each of the M , 90,000-210,000 proteins has been suggested to form a complex with the M , 80,000 protein mediated by a metal ion, perhaps calcium (2, 5, 9).

The availability of the cDNA for human factor VI11 allowed the construction of plasmids which would direct the expres- sion of factor VI11 protein in transfected mammalian cells (2, 3, 15). Factor VI11 secreted from these cells exhibits many of the functional characteristics associated with plasma-derived factor VI11 (pdVIII)’ (2, 3). The development of mammalian cell lines producing recombinant factor VIII (rVIII) has per- mitted the production of highly purified factor VIII. Use of such preparations as therapeutics in the treatment of hemo- philia offers a major advantage over current factor VI11 prep- arations that are used for the treatment of hemophilia. These latter preparations are derived from human plasma, highly impure, and inevitably contaminated with virus particles.

In the present study we report the purification and char- acterization of rVIII obtained from a mammlian cell line transfected with a factor VI11 expression vector (3). Results presented in this paper show that rVIII is functionally and structurally similar to pdVIII.

EXPERIMENTAL PROCEDURES

Factor Xa, thrombin, and activated protein C were all generous gifts of Dr. Walter Kisiel (The University of New Mexico). Factor VI11 concentrates were gifts from Cutter Laboratories. Rabbit brain cephalin and phenylmethanesulfonyl fluoride were from Sigma; pla- telin was from General Diagnostics; factor VIII-deficient and normal human plasmas were from George King Biomedical; factor VI11 chromagenic Coatest assay was from Helena; Dubecco’s modified Eagle’s Medium and Ham’s F-12 medium were obtained from Gibco.

Purification ofpdVZZZ and rVZZZ-Plasma-derived factor VI11 was purified from commercially available factor VI11 concentrates (9). The concentrates were diafiltered into 0.02 M Tris buffer, pH 7.4, containing 135 mM NaCI, 5 mM sodium citrate, 1 mM CaC12, 5% glycerol, and 1 mM phenylmethanesulfonyl fluoride. Subsequently, p - mercaptoethanol was added to solubilized concentrates to 35 mM, and factor VI11 was purified using DEAE-Sepharose and a factor VI11 immunoaffinity column as described previously (9). Recombinant factor VI11 was purified from serum-free media (Dulbecco’s modified Eagle’s medium/F-12, 1:l) conditioned by baby hamster kidney cells transfected with a factor VI11 expression plasmid (3) for 48 h. The media were collected and diluted with distilled H20 to reduce the

The abbreviations used are: pdVIII, plasma-derived factor VIII; rVIII, recombinant factor VIII; SDS-PAGE, sodium dodecyl sulfate- polyacrylamide gel electrophoresis; APC, activated protein C.

3285

3286 Recombinant Factor VIII conductivity below 10 mmho. Purification of rVIII was accomplished using the.same chromatographic steps used to isolate pdVIII (9). We found that the recoveries across these columns for rVIII and pdVIII were very similar. Purified rVIII and pdVIII were stored at -80 "C in 0.05 M Tris buffer, pH 7.5, containing 150 mM NaCl, 2.5 mM CaCl,, 1 mM phenylmethanesulfonyl fluoride, and 5% glycerol. Factor VI11 activity was measured by coagulation analysis or by the factor VI11 Coatest assay as described previously (3). Protein concentration was determined by the method of Bradford (17).

Isolation of Factor VIII Subunits-Separation of the M, 80,000 protein from the M, 90,000-210,000 proteins was accomplished by absorbing rVIII or pdVIII in 0.05 M Tris buffer, pH 7.5, 150 mM NaC1, 2.5 mM CaC12, and 5% glycerol to a factor VI11 monoclonal antibody column that binds the M, 80,000 protein (3, 9). The M, 90,000-210,000 proteins were eluted with the above buffer containing 50 mM EDTA. The M, 80,000 protein was then eluted under the same conditions used to elute intact factor VI11 (9). To obtain the M, 73,000 subunit, the M, 80,000 protein was treated with thrombin (1:50 ratio, w/w) for 1 h at 37 "C at which time hirudin was added to inhibit thrombin.

Cleavage and NH2-terminal Sequence Analysis of rVIII-For NH2- terminal amino acid sequencing, rVIII (0.2-0.5 mg) was incubated with either thrombin (1: lOO ratio, w/w), factor Xa (1:lOO ratio), or activated protein C (APC) (1:25 ratio) for 30-60 min at 37 "C. In the case of factor Xa and APC, l/lOth sample volume of rabbit brain cephalin was included as a source of phospholipid. The reactions were terminated by adding SDS to a final concentration of 0.4% and immediately heating samples to 80 "C. Proteolyzed rVIII was resolved on a 6-12% polyacrylamide gradient gel in the presence of SDS (SDS- PAGE). Electrophoresis was carried out by the method of Laemmli (18). Proteins were detected by staining with Coomassie Blue, excised, and electroeluted according to the method of Hunkapiller et al. (19). Peptides eluted from gels were subjected to NH2-terminal amino acid sequence analysis using an Applied Biosystems gas-phase sequenator (20) modified for on-line phenylthiohydantoin identification (21).

Changes in subunit structure and coagulant activity of rVIII during proteolysis by thrombin, factor Xa, or APC were determined as described previously (9). Briefly, rVIII (2.5 pg) was incubated with either thrombin (0.05 pg), factor Xa (0.05 pg), or APC (0.1 pg) over a 1.5- to 2-h time course. Rabbit brain cephalin was added to factor Xa and APC reactions. At various time points during the incubation, coagulant activity was determined and subunit structure was analyzed by SDS-PAGE. Proteins were visualized by silver staining (22).

Isoelectric Focusing of Factor VIII-Isoelectric focusing gels were carried out according to the method of O'Farrell (16) using 0.4-mm ultrathin horizontal gels in place of tube gels. The gels contained 2.5% pH 3.5-10.0 carrier Ampholines. Proteins to be focused were diafiltered into 20 mM Tris buffer, pH 7.5, containing 4 M urea, 2.5 mM CaC12, and 0.01% Tween 80; electrophoresis was carried out a t 8 watts (1000 V limit) a t 18 "C for 7 h.

RESULTS

Structure of Purified rVIII-As in the case with pdVIII, analysis of purified rVIII by SDS-PAGE shows that it consists of several polypeptide chains with M , 80,000-210,000 (Fig. 1). The pattern of polypeptides observed for rVIII is very similar to that of pdVIII. Analysis by Western blot shows that all the bands associated with rVIII are detected by a polyclonal antibody derived against pdVIII (Fig. 1). That the pattern of polypeptides for rVIII and pdVIII are strikingly similar sug- gests that rVIII and pdVIII are processed in the same manner. This indicates that the initial cleavage sites within the single- chain form of factor VI11 are as accessible to proteolysis in rVIII as they are in pdVIII. I t is not known a t present whether this initial proteolysis is an intracellular or extracellular event, or occurs during the purification of rVIII.

Proteolysis of rVIII by Thrombin, Factor Xu, and Activated Protein C-Factor VI11 coagulant activity is altered by throm- bin, factor Xa, and activated protein C (7-12). These pro- teases, through specific cleavages, process factor VI11 into active and inactive forms (9). The proteins generated after cleavage of rVIII by thrombin, factor Xa, or APC comigrated on SDS-PAGE with those derived from pdVIII (Fig. 2). Fur- thermore, these proteins were found to have the same NH2- terminal sequence as their counterparts derived from pdVIII (Fig. 3). This indicates that the cleavage sites (see "Discus- sion") of thrombin, factor Xa, and APC in rVIII and pdVIII are the same.

Activation and Inactivation of rVIII Coagulant Activity- Correlating changes in factor VI11 subunit structure with

MW x C D M W x

73-

50 - 43 -

.*.-. THROMBIN "- Xa APC I F-"

67-

45- 45- I

PD R PD R PD R FIG. 1. SDS-PAGE and western blotting of rVIII and pdVIII. Approximately 15 pg of rVIII ( B and C ) or

pdVIII (A and D) were resolved on a 6-12% SDS-polyacrylamide gel. The proteins were either visualized by staining with Coomassie Blue ( A and B ) or transferred to nitrocellulose (C and D) for western blotting. For westerns a polyclonal antibody derived against pdVIII was used to detect factor VI11 bound to nitrocellulose.

FIG. 2. Cleavage of rVIII and pdVIII by thrombin, factor Xa, and APC. Recombinant VI11 or pdVIII (-15 pg) were incubated with either thrombin (0.3 pg), factor Xa (0.3 pg), or APC (0.6 pg) for 1 h at 37 "C. Rabbit brain cephalin (l/lOth volume) was included in factor Xa and APC reactions. The reactions were stopped by the addition of SDS to 0.5% and heating samples to 90 "C for 5 min. Proteins were resolved on a 6-12% SDS- polyacrylamide gel and visualized by staining with Coomassie Blue.

Recombinant Factor VIII 3287

Polypeptide Thrombin Factor Xn APC

73,000 rv ln (SIFQKKTRHYFIAAV (SFQKKTRHYFIAAV pdVlll (SIFQKKTRHYFIAAVDERL T6ZKKTRHYRAAV

llpo

67,000 rVll1 pdVlll

50.000 pdVlll

r v m ATRRYYLGAVELSOD ATRRYYLGAVEL-W -yM

43.000 pdVlll (SIVAKK--K- WVI

rvln (SIVAKKHPKTWV

ill

AQSGWFQFKK ?.SGSVFQFKK

ATRRWLGAVEL A’TRRWLGAVELS

A?RRYYLGAVE ATRRYYLGAVEL ATRRYYLGAV ATRRYYLGAVEL

I

SVAKKHPKT SVAKKHPK ,n

~~

FIG. 3. NHz-terminal sequence of rVIII proteins and their comparison to pdVIII proteins. Recombinant factor VI11 was treated with either thrombin, factor Xa, or APC and subsequently resolved by SDS-PAGE. Proteins to be sequenced were electroeluted as described under “Experimental Procedures.” The position of these proteins within factor VI11 were determined by alignment of the obtained sequence to the cDNA sequence of factor VI11 (2, 3). Se- quence shown for DdVIII proteins are taken from Eaton et al. (9).

r

210-

90- 80-

I .

2ooo8 c 0 , 2 5 10 15 20 3” 45 6090l20

x) 40 60 80 1 0 0 IM

IncubationTime (minutes) FIG. 4. Thrombin activation of rVIII. Recombinant factor VI11

(2.5 pg) was incubated with (U) or without (o”--o) thrombin (0.05 pg) for 0-120 min at 37 “C at the time points shown. Both coagulant activity and subunit structure were determined as described under “Experimental Procedures.”

activation by thrombin has been the subject of several recent reports (6-9,12, 14). All of these reports show that thrombin activation is correlated with the proteolytic processing of the M , 80,000-210,000 proteins to lower MI subunits. We have observed that thrombin activation of pdVIII correlates with the generation of M, 73,000, 50,000, and 43,000 subunits (9). Similarly, a time course treatment of rVIII with thrombin resulted in approximately an 80-fold increase in rVIII coagu- lant activity (Fig. 4). Analysis of the changes in subunit structure of rVIII by SDS-PAGE and gel scanning shows that this activation coincided with the processing of the M, 80,000- 210,000 precursor species to the M , 73,000,50,000, and 43,000 active subunits (Figs. 4 and 5). As determined by amino- terminal sequencing, COOH-terminal proteolysis of the Mr 210,000 protein generates the Mr 90,000 protein which is subsequently cleaved at arginine 372 to yield the M, 50,000 and 43,000 proteins (Fig. 3, Refs. 2, 9, 13). Concomitantly, the M , 80,000 protein is cleaved at arginine 1689 to yield the M , 73,000 protein (Fig. 3). Thrombin-activated rVIII re- mained stable for at least 1 h a t 37 “C after reaching the apparent maximum activity (Fig. 4). This stability has also been observed for thrombin-activated pdVIII (9).

The kinetics of inactivation of rVIII by APC is shown in Fig. 6. Similar to pdVIII (9, 11) the inactivation of rVIII by APC is correlated with the proteolysis of the Mr 90,000-

210,000 proteins and the appearance of an M, 45,000 protein (Fig. 6). The M, 45,000 protein is derived from the NHz- terminal portion of the M , 90,000-120,000 proteins (Ref. 9 and Fig. 3). The COOH-terminal portions of the M, 90,000- 210,000 proteins that would arise after cleavage by APC are not apparent by SDS-PAGE (Fig. 6), which has also been observed for pdVIII (9, 11).

Factor Xa cleaves factor VI11 at the same sites as thrombin or APC (Figs. 2 and 3; Ref. 9). Factor Xa has been observed to initially activate and then ultimately inactivate pdVIII (9). Incubation of rVIII with factor Xa over a 90-min time course resulted initially in a &fold activation followed by a decrease in coagulant activity to below control levels (Fig. 7). Analysis of rVIII subunit structure during this time course shows that activation is best correlated with the appearance of the Mr 73,000, 50,000, and 43,000 subunits, while inactivation coin- cides with the degradation of the Mr 73,000 and 50,000 sub- units to fragments of M, 67,000 and 45,000 (Fig. 7, inset). Similar results have been obtained in our laboratory with pdVIII (9).

Subunit Dissociation and Reassociation-It has been sug- gested that each of the M, 90,000-210,000 proteins form a calcium-linked complex with the M, 80,000 protein of factor VI11 (5, 9). When rVIII or pdVIII are adsorbed to a factor VI11 monoclonal antibody column, which is specific for the Mr 80,000 protein (9, 13), the M , 90,000-210,000 proteins can be eluted with EDTA (Fig. 8). The M , 80,000 protein is subsequently eluted under conditions normally used to elute factor VI11 (9). This suggests that pdVIII and rVIII each exist as a metal-linked complex. Neither the isolated M, 80,000 protein nor the M, 90,000-210,000 proteins exhibited coagu- lant activity.

It is likely that the M, 73,000, 50,000, and 43,000 subunits of thrombin-activated factor VI11 (VIIIa) also form a metal- linked complex. Treatment of either pdVIIIa or rVIIIa with EDTA results in inactivation (Fig. 8). Coagulant activity can be subsequently regenerated by quenching the EDTA with excess MnClZ (Fig. 9). Activity was also regenerated, but to a lesser extent, by the addition of excess CaClZ. Incubation of EDTA-inactivated factor VIIIa with 50 mM MnC12 for 1 h resulted in restoration of -80% of the initial activity, whereas only -30% of the activity was regenerated when 50 mM CaClz was used. This is similar to what has been previously observed for factor Va by Esmon (25). When factor VIIIa is subjected to gel permeation chromatography all three subunits comi- grate, while they do not if factor VIIIa is pretreated with EDTA (data not shown). These results suggest that EDTA inactivates both pdVIIIa and rVIIIa by causing subunit dis- sociation, supporting the notion that factor VIIIa exists as a metal- (perhaps Ca2+) linked complex.

Isoelectric Focusing of r VIII and pdVIII-When rVIII and pdVIII were subjected to isoelectric focusing only the MI 80,000 protein was resolved (Fig. 9). The M , 90,000-210,000 proteins of either rVIII or pdVIII could not be focused, even under denaturing conditions, using several isoelectric focusing gel systems (Fig. 10). The isolated Mr 80,000 protein of rVIII or pdVIII resolved into a cluster of four bands with isoelectric points of 6.9-7.2 and a separate band with an isoelectric point of 6.5 (Fig. 9). These bands corresponded to the focused bands of intact rVIII and pdVIII. The M , 73,000 subunit of both rVIIIa and pdVIIIa resolved into a single broad band with an isoelectric point of 8.1 (Fig. 10). At this time we do not have sufficient quantities of the MI 50,000 and 43,000 subunits of activated factor VI11 to perform isoelectric focusing.

3288 Recombinant Factor VIII

I . - 8 0 .

FIG. 5. Quantitation of changes in .C rVIII subunit structure during acti- ea. vation. The various protein species & present a t each time point in the gel of 3 Fig. 4 were quantitated by scanning each 5 '

lane using an LKB laser densitometer $ gel scanner. P.

I o P u) m ea l o o m

0 . . . . . . . . . . . . .

c -80

c " 4 5

1 I I I I I 1 I I I x) 40 60 80 1 0 0

Incubation Time (minutes) FIG. 6. Inactivation of rVIII by APC. rVIII (2.5 pg) was incu-

bated with APC (100 ng) and l/lOth volume of rabbit brain cephalin for 0-90 min a t 37 "C. A t the time points indicated, coagulant activity and subunit structure were determined as described under "Experi- mental Procedures."

. " ""

loo: x) 40 60 80 I00

lncubat ion Time (minutes 1 FIG. 7. Activation of rVIII by factor Xa. rVIII (2.5 pg) was

incubated with (M) or without (W) factor Xa (50 ng) and l/lOth volume of rabbit brain cephalin for 0-120 min a t 37 "C. At the times indicated subunit structure and coagulant activity were determined as described under "Experimental Procedures."

DISCUSSION

The recent cloning and protein characterization studies of factor VI11 have elucidated the structure of factor VI11 (2, 3, 5-14). The DNA sequence of factor VI11 predicts a single- chain glycoprotein with an approximate M, of 300,000. The single-chain form is readily proteolyzed in uiuo (23) and in vitro (8) to yield a form consisting of multiple polypeptides with M, ranging from 80,000-210,000 (6,8,9). The M, 210,000 and 80,000 proteins represent the NH2-terminal and COOH- terminal regions of factor VIII, respectively (2, 13). Proteol- ysis within the COOH-terminal region of the M , 210,000 protein generates a series of proteins with M, ranging from

o zo u) ea ea 1m1a

Incubation time

04 ! . . . . . . . . . . . r 0 P 40 ea ea l o o 1 2 l

90- 80-

FIG. 8. Separation of factor VI11 protein chains. Either pdVIII or rVIII were adsorbed to a factor VI11 monoclonal antibody column and the M, 90,000-210,000 and M, 80,000 proteins were eluted from the column as described under "Experimental Procedures." The isolated subunits were resolved on a ti-12% SDS-polyacrylamide gel, and proteins were visualized by silver staining (18).

r pdVllla r RVllla

- 1400

600

200 I Control EDTA EDTA

MnC12 EDTA EDTA

MnC12

FIG. 9. Regeneration of factor VIIIa coagulant activity after EDTA treatment. Both pdVIII and rVIII in 0.05 M Tris, pH 7.5, 0.15 M NaC1, 2.5 mM CaCI,, and 5% glycerol were activated by incubation with thrombin (1:50 ratio) for 1 h a t 37 "C a t which time hirudin was added. T o pdVIIIa or rVIIIa either no addition was made (control) or EDTA was added to 20 mM. The samples were then incubated for 1 h a t 37°C a t which time coagulant activity was determined. To the EDTA-treated samples, MnCI, was added to 50 mM and incubated an additional 1 h a t 37 "C. Subsequently, coagulant activity was determined.

110,000 to 180,000 (9). As discussed below the M, 80,000 protein forms a metal-linked complex with each of the M, 90,000-210,000 proteins (Fig. 11). Both the single-chain form and the above mentioned multiple polypeptide form of factor VI11 have been suggested to be inactive or less active precur- sors that can be activated by factor Xa or thrombin and inactivated by APC through specific proteolytic processing (8-12, 14). The proteolysis of factor VI11 by thrombin, factor Xa, and APC is summarized in Fig. 11.

Recombinant Factor VIII 3289

1 }Thrombin

x FIG. 10. Isoelectric focusing of pdVIII and rVIII. pdVIII,

rVIII, and the various isolated subunits were diafiltered into 20 m M Tris, pH 7.2, 2.5 mM CaC12, 0.01% Tween 80, and 4 M urea. Approx- imately 10 pg of each were resolved on pH 3.5-10.0 gradient isoelectric focusing gels as described under “Experimental Procedures.” Proteins were visualized by silver staining (22).

Recently the cleavage sites of thrombin, factor Xa, and APC within factor VI11 have been identified (2,9,13). Throm- bin and factor Xa initially cleave the M, 110,000-210,000 proteins a t position 740 to yield the M , 90,000 protein (2, 9), which is subsequently cleaved at position 372 to generate the M , 50,000 and 43,000 subunits of factor VIIIa (Fig. 11, Ref. 9). Concomitantly, the M, 80,000 protein is cleaved at position 1689 to generate the M , 73,000 subunit of factor VIIIa (Fig. 11, Ref. 9). Further proteolysis of the M , 73,000 and 50,000 subunits by factor Xa generates fragments of M, 67,000 and 45,000 (Fig. 11, Ref. 9). One or both of these latter cleavages correlate with inactivation of factor VIIIa (9). APC has been proposed to cleave the M , 90,000-210,000 proteins of factor VI11 at position 336 to generate a M, 45,000 protein (Fig. 11, Ref. 9). This appears to be the same site at which factor Xa cleaves the M, 50,000 subunit of factor VIIIa (Fig. 10, Ref. 9). It has been proposed that cleavage at position 336 by either factor Xa or APC correlates with inactivation (9). Cleavage at position 372 by thrombin or factor Xa is in part responsible for activation (9). This suggests that the region between positions 336-372, which contains 15 aspartic and glutamic acid residues and only 4 lysine or arginine residues, is of functional importance.

Analysis of purified rVIII shows that, like pdVIII, it consists of multiple polypeptides. By SDS-PAGE the pattern of poly- peptides observed for rVIII is strikingly similar to that of pdVIII (Fig. 1). Both are resolved into seven or eight distinct bands with M , ranging from 80,000 to 210,000 (Fig. 1). That the M, 210,000 proteins of rVIII and pdVIII comigrate on SDS-PAGE suggests that these two proteins may be glyco- sylated to a similar extent. The M, 210,000 protein contains 23 potential N-linked glycosylation sites (2, 3, 13). It is also interesting that the M , 80,000 proteins of either rVIII or pdVIII resolve into a doublet on SDS-PAGE (Fig. 1). These observations suggest that the DNA-predicted single-chain form of rVIII and pdVIII are post-translationally processed in a similar manner to yield the multiple polypeptide form.

Treatment of rVIII with thrombin, factor Xa, and APC shows that these proteases process rVIII in the same manner they do pdVIII (Figs. 2 and 3). Activation of rVIII by thrombin or factor Xa correlated with the generation of M , 73,000, 50,000, and 43,000 subunits (Figs. 4, 5, and 7) while inacti- vation of rVIIIa by factor Xa coincided with the degradation of the M, 73,000 and 50,000 subunits (Fig. 7). Inactivation of rVIII by APC resulted in the proteolysis of the M , 90,000- 210,000 proteins (Fig. 6). These results clearly show that rVIII is proteolytically processed by thrombin, factor Xa, and APC to activated and inactivated forms in the same manner as pdVIII.

Initial results of Fass et al. ( 5 ) , using porcine factor VIII, suggested that, like factor V (24, 25), factor VI11 may exist as a Caz+-linked complex. Similarly, results presented here in- dicate that both rVIII and pdVIII exist as metal-linked com- plexes in that treatment of factor VI11 or VIIIa with EDTA causes separation of the various protein chains. We found that when separated the M, 80,000 and M , 90,000-210,000 proteins of factor VI11 and the individual subunits of factor VIIIa did not exhibit coagulant activity even in the presence of excess MnClz or CaC12 (data not shown). However, the coagulant activity of EDTA-treated rVIIIa or pdVIIIa could be restored by the addition of a molar excess of MnClZ (Fig. 9). Since EDTA appears to dissociate factor VIIIa subunits, these results suggest that the subunits of factor VIIIa must be in a complex to exhibit coagulant activity. A t this time it is not known if all three subunits are required for activity or just two.

For a more detailed comparison of rVIII and pdVIII both were analyzed by isoelectric focusing. When rVIII or pdVIII were subjected to isoelectric focusing only the M , 80,000 protein was resolved. The isolated M , 80,000 protein of either

1 740

N 90,000

1649 ca I *

32

I 80,000 C

+ degradation peptides

80,000

FIG. 11. Proteolytic processing of human factor VI11 by thrombin, factor Xa, and APC.

3290 Recombinant Factor VIII

rVIII or pdVIII was resolved into a cluster of four distinct bands with isoelectric points between 6.9-7.2 and a separate band with an isoelectric point of 6.5 (Fig. 10). The M , 90,000- 210,000 proteins of either rVIII or pdVIII were not resolved by a number of isoelectric-focusing gel systems. This probably reflects the heterogeneity of these proteins as well as the potentially high content of carbohydrate associated with these proteins (13). The M, 73,000 subunit of either rVIIIa or pdVIIIa was resolved into a single broad band with an isoe- lectric point of 8.1 (Fig. 9). Cleavage of the M , 80,000 protein by thrombin to generate the M , 73,000 subunit, results in the possible removal of a 44-residue polypeptide that contains 15 aspartic and glutamic acid residues and only 4 lysine or arginine residues (9, 13). Hence, the M , 73,000 subunit has a more basic isoelectric point than that of the Mr 80,000 protein from which it is derived. These results indicate that at least the charge density of the M, 80,000 and 73,000 proteins of rVIII and pdVIII are similar.

The expression and secretion of recombinant factor VI11 from mammalian cells has enabled the purification of factor VI11 from a plasma-free system. This is in marked contrast to initial purifications of pdVIII accomplished only a few years ago that were laborious and required a large quantity of plasma (5-8). The initial characterization studies done here suggest that rVII1 is both structurally and functionally similar to pdVIII. Recent pharmacological data also suggest that, based on a number of pharmacological and physiological parameters, rVIII and pdVIII function similarly in vivo (26). The availability of highly purified factor VI11 will allow for both the development of a safe pharmaceutical for treating hemophilia and pure protein for characterization studies. This latter capability will hopefully better our understanding of the factor VI11 molecule.

Acknowledgments-We thank Richard Lazar of Genentech and Cutter Laboratories for supplying baby hamster kidney cell-condi- tioned media containing rVIII. We also thank Henry Rodriguez for amino acid sequencing of factor VI11 and Dr. Robert Hershberg for critically reviewing the manuscript.

REFERENCES

1. Jackson, C. M., and Nemerson, Y. (1980) Annu. Reu. Biochem. 49,767-811

2. Toole, J. J., Knopf, J . L., Wozney, J. M., Sultzman, L. A., Buecker, J. L., Pittman, D. D., Kaufman, R. J., Brown, E.,

Shoemaker, C., Orr, E. C., Amphlett, G. W., Foster, W. B., Coe, M. L., Knutson, G. J., Fass, D. N., and Hewick, R. M. (1984) Nature 312 , 342-348

3. Wood, W. I., Capon, D. J., Simonsen, C. C., Eaton, D. L., Gitschier, J., Keyt, B., Seeburg. P. H., Smith, D. H., Hollings- head, P., Wion, K. L., Delwart, E., Tuddenham, E. G. D., Yehar, G. A., and Lawn, R. M. (1984) Nature 312,330-337

4. Gitschier, J., Wood, W. I., Goralka, T. M., Wion, R. L., Chen, E. Y., Eaton, D. H., Vehar, G. A., Capon, D. J., and Lawn, R. M. (1984) Nature (London) 312 , 326-330

5. Fass, D. N., Knutson, G. J., and Katzman, J. A. (1982) Blood 59 ,

6. Fulcher, C. A., and Zimmerman, T. S. (1982) Proc. Natl. Acad.

7. Vehar, G. A., and Davie, E. W. (1980) Biochemistry 19,401-410 8. Rotblat, F., O’Brien, D. P., O’Brien, F., Goodall, A. H., and

Tuddenham, E. G. D. (1985) Biochemistry 24 , 4294-4300 9. Eaton, D. L., Rodriguez, H. R., and Vehar, G. A. (1986) Biochem-

10. Fulcher, C. A., Roberts, J. R., and Zimmerman, T. S. (1983) Blood

11. Fulcher, C. A., Gardiner, J. E., Griffin, J. H., and Zimmerman,

12. Lollar, P., Knutson, G., and Fass, D. (1984) Blood 6 3 , 1303-1308 13. Vehar, G. A,, Keyt, B., Eaton, D. E., Rodriguez, H., O’Brien, D.

P., Rotblat, F., Opperman, H., Keck, R., Wood, W. I., Harkins, R. N., Tuddenham, E. G. D., Lawn, R. M., and Capon, D. J. (1984) Nature 312 , 337-342

14. Lollar, P., Knutson, G. J., and Fass, D. M. (1986) Biochemistry

15. Truett, M. A., Blacher, R., Burke, R. L., Caput, D., Chu, C., Dina, D., Hartog, K., Kuo, C. H., Masiarz, F. R., Merryweather, J. P., Najarian, R., Pachl, C., Potter, S. J., Puma, J., Quiroga, M., Rall, L. B., Randolph, A., Urdea, M. S., Valenzuela, P., Dahl, H. H., Favalaro, J., Hansen, J., Nordfang, O., and Ezban, M.

594-600

Sci. U. S. A. 79 , 1648-1652

istv 25,505-512

6 1,807-81 1

T. S. (1984) Blood 63,486-489

24,8056-8064

(1986) DNA 4,333-349 16. OFarrell, P. H. (1975) J. Biol. Chem. 250,4007-4021 17. Bradford, M. M. (1976) Anal. Biochem. 72 , 248-254 18. Laemmli, U. K. (1970) Nature 227 , 680-685 19. Hunkapiller, M. W., Lujan, E., Ostrander, F., and Hood, L. E.

20. Hewick, R. M., Hunkapiller, M. W., Hood, L. E., and Dreyer, W.

21. Rodriguez, H. R. (1986) J. Chromatogr. 350,217-225 22. Morrissey, J. H. (1981) Anal. Biochem. 117,307-310 23. Weinstein, M. J., Fulcher, C. H., Chute, L. E. (1983) Blood 6 2 ,

24. Hibbard, L. S., and Mann, K. G. (1980) J. Biol. Chem. 255,638-

25. Esmon, C. T. (1979) J. Biol. Chem. 2 5 4 , 964-973 26. Fournel, M. A., Pancham, M., Ng, P., Mitra, G., Mozen, M.,

Eaton, D. L., Gregory, T. (1986) Res. Clin. Lab. 16, 112

(1983) Methods Enzymol. 91 , 227-247

J. (1982) J. Biol. Chem. 256 , 7990-7997

1114-1117

645