Human Plasma and Recombinant Factor VI1 - The Journal of ...

8 downloads 0 Views 1MB Size Report
with either purified relipidated tissue factor apopro- tein or tissue factor on the surface of a human bladder carcinoma cell line (582) to activate either factor X or.
THEJOURNALOF BIOLOGICAL CHEMISTRY

Vol. 266, No. 17, Issue of dune 15, pp. 11051-11057,1991 Printed in U,S. A.

0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Human Plasma and Recombinant Factor VI1 CHARACTERIZATIONOF0-GLYCOSYLATIONSATSERINERESIDUES SITE-DIRECTEDMUTAGENESISOFSERINE 52 T O ALANINE*

52 AND 60 ANDEFFECTS

OF

(Received for publication, danuary 2 , 1991)

Soeren BjoernS, DonaldC. Foster$, Lars Thim$, Finn C. Wibergz, Mogens Christensen$, Yutaka Komiyamay, Anders H. Pedersenqll , and WalterKisielV** From $Bioscience Corporate Research and the 11 Biopharmaceuticals Division, Novo Nordisk AIS, Novo Alle, Bagsuaerd, Denmark, SZymoGenetics, Incorporated, Seattle, Washington 98105, and the 7Blood Systems Research Foundation Laboratory, Department of Pathology, University of New Mexico School of Medicine, Albuquerque, New Mexico 87131

factor IX was virtually identical to that observed for Factor VI1 is a multidomain, vitamin K-dependent plasma glycoprotein that participates in the extrinsic wild-type factor VIIa. These results indicate that the pathway of blood coagulation. Earlier studies demon- carbohydrate moiety 0-glycosidically linked to serine strated a novel disaccharide (Xyl-Glc) or trisaccharide 52 does not appear to be involved either in the inter(Xy12-Glc)0-glycosidically linked to serine 52 in hu- action of factor VIIa with tissue factor, or the expresX or factor man plasma factor VI1 (Nishimura, H., Kawabata, S., sion of its proteolytic activity toward factor Kisiel, W., Hase, S., Ikenaka, T., Shimonishi, Y., and IX following complex formation with tissue factor. The reason(s) for the decreased clotting activity of mutant (1989) J. Biol. Chem. 264, 20320Iwanaga,S. factor VIIa remains to be established but may reflect 20325). In the present study, human plasma and reits enhanced interaction with plasma inhibitors. combinant factor VI1 were isolated and subjected to enzymatic fragmentation. Peptides comprising residues 48-62 of the first epidermal growth factor-like domain of each factor VI1 preparation were isolated Factor VI1 is a single-chain, vitamin K-dependent glycofor comparative analysis. Using a combined strategy protein (Mr -50,000) that is synthesizedinthe liver and of amino acid sequencing, carbohydrate and amino secreted acid into the blood as a zymogen of a serine protease, compositionanalysis, and mass spectrometry, three factor VIIa. Single-chain human factor VI1 is converted to different glycan structures consisting of either glucose, by cleavage of an glucose-xylose, or glucose-(xylose)2 were detected 0- two-chainfactor VIIa inthetesttube internal peptide bond located at Arg-152-Ile-153 by one of glycosidically linked to serine 52 in plasma and recombinant factorVII. Approximately equal amountsof the severalcoagulation proteases includingfactor XIIa, factor IXa,factorXa,andthrombin (1-3). Factor VIIa isthus three glycan structures were observed in plasma factor VII, whereas in recombinant factor VI1 the glucose and composed of an amino-terminallight chain of 152 amino acid residues and a carboxyl-terminal heavy chain of 254 amino the glucose-(xylose)z structures predominated. In adacidresiduesheld together by asingledisulfide bond (4). dition to the 0-linked glycan structures observed at Factor VIIa is virtually devoid of proteolytic activity. Howserine 52, a single fucose was found to be covalently linked at serine60 in both human plasma and recom- ever, upon forming a complex with its cell-surface receptor, binant factor VII. Carbohydrate and mass spectrome- tissue factor, factor VIIa readily activates factors X and IX t r y analyses indicated that the fucosylation of serine by limited proteolysis in the presenceof calcium (5). Prior to secretion from the liver, human factor VI1 under6 0 was virtually quantitative.Metabolic labelingstudies using [’4C]fucose confirmedthepresence of 0- goes several post-translational modifications including y-carboxylation of 10 glutamic acid residues located in its aminolinked fucose at serine60. In order to assess whether the carbohydrate moiety terminal region, and N-glycosylation of Asn-145 and Asn-322 a t serine 52 contributes to the biological activity of (6). In addition, recent studieshave identified a unique disacfactor VII, we have constructed a site-specific mutant charide (Xyl-Glc)or trisaccharide (Xy1,-Glc) 0-glycosidically of recombinant factor VI1 in which serine52 has been linkedtoSer-52inthefirstepidermal growth factor-like replaced with an alanine residue. Mutant factor VIIa domain of humanplasmafactor VI1 (7). While it is well exhibited -60% of the coagulant activityof wild-type documented that the y-carboxyglutamicacid residues of facfactor VIIa in a clotting assay. The amidolytic activitytor VII/VIIa are essentialfor its interaction with tissue factor of mutant factor VIIa was indistinguishable from that and theexpression of its proteolytic activity toward factorsX observed for recombinant wild-type factor VIIa. In and IX (5, 8),the function, if any, of the glycosyl moieties in addition, the ability of mutant factor VIIa in complex factor VII/VIIa remains to be established. In order to inveswith either purified relipidated tissue factor apopro- tigate further whether glycosylation of Ser-52 in factor VII/ tein or tissue factor on the surface of a human bladder VIIa plays arole in itsbiological activity, we have constructed X or a mutant factor VI1 molecule by site-directed mutagenesis in carcinoma cell line (582) to activate either factor which Ser-52 has been replaced with an alanine residue. We * This work was supported in part by a research grant from the report here on the amidolytic and proteolytic activities of this mutant factor VI1 molecule, designated S52A factor VII.’ We Blood Systems, Inc. and National Institutes of Health Grant HL 35246. The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore hereby be marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ** To whom correspondence should be addressed.

Mutant factor VI1 is designated according to the notation described by Shapiro and Vallee (9) in which the single-letter code for the original amino acid is followed by its position in thesequence and the single-letter code for the new amino acid.

11051

Mutant Factor VIIa

11052

also report that human plasma factor VII, as well as recombinant human factor VI1 produced in baby hamster kidney cells, contains glucose 0-linked to Ser-52 and fucose 0-linked t o Ser-60 in addition to thepreviously reported disaccharide (Xyl-Glc) or trisaccharide (Xy12-Glc)linked to Ser-52. EXPERIMENTAL PROCEDURES' RESULTS

Characterization of Glycan Structures 0-Glycosidically Linked to Serine Residues 52 and 60 in Human Factor VIIEarlier studies demonstrated that human plasma factor VI1 contains a novel disaccharide (Xyl-Glc) or trisaccharide (Xy1,-Glc) 0-glycosidically linked to serine 52 in the first epidermal growth factor-like domainof the molecule (7). The present study was undertaken to determine whether recombinant factor VI1 synthesized inbaby hamster kidney (BHK)" cells also contained these unique glycan structures at serine 52. Factor VIIa was initially isolated to homogeneity from human plasma and the conditioned medium of BHK cells (6). Eachpreparation was pure as judged by SDS-PAGE and HPLC analysis (Fig. M1, A and B ) . Each factor VIIa preparation was then reduced and alkylated with 4-vinylpyridine and the S-pyridylethylatedheavy and light chains separated by reverse-phase HPLC (Fig. M1, C and D).The isolated Spyridylethylated light chain from each factor VIIa preparation was subjected to endoproteinase Asp-N digestion, and a peptide fragment containing residues 48-62 (fragment 48-62) was isolated by reverse-phase HPLC (Fig. M2, A and B ) . Each isolated fragment 48-62 was then subjected to aminoacid and carbohydrate composition analyses, NH,-terminal sequence analysis before and after deglycosylation with hydrogen fluoride, and analysisby '%f plasma desorption mass spectrometry. Amino acid and carbohydratecomposition analyses of fragment 48-62 indicated 1 mol of glucose and 1.4 mol of xylose in human plasma fragment 48-62, and 0.9 mol glucose and 0.9 mol of xylose in recombinant fragment 48-62 (Table MI). The glucose and xylose content in human plasma fragment 48-62 is inreasonably good agreement with the results of Nishimuraet al. (7) who reported Xyl-Glc and Xy12-Glc structures on serine52 of human plasma factor VII. In addition toglucose and xylose, -1 mol of fucosewas also observed inplasmaandrecombinantfragment 48-62 (TableMI). Amino-terminal amino acid sequence analyses indicated that no amino acid could be assigned to cycle number 5 (residue 52) in fragments obtainedfrom human plasmaor recombinant factor VIIa S-pyridylethylated light chains (Table MII). In addition, serine 60 was obtained in low yields. This finding indicated a post-translational modification of serine 52, which was confirmed by sequence analyses of chemically deglycosylated (HF-treated) fragment 48-62 in which serine 52 was detected in normal yields (Table MII). In order to determine the molecular mass of plasma and recombinant fragment48-62, each glycopeptidewas subjected t o plasma desorption mass spectrometry. Fig. M3A shows a '' Portions of thispaper(including"ExperimentalProcedures," Tables MI-VI, and Figs. M1-8) are presented in miniprintat the end o f this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal thatis available from Waverly Press. " T h e abbreviations used are: BHK, baby hamster kidney; SDS, sodium dodecyl sulfate; HPLC, high performance liquid chromatography; TBS, 50 mM Tris-HC1 (pH 7.5)/ 100 mM NaC1; 582, human bladder carcinoma cell line; PS, phosphatidylserine; PC, phosphatidylcholine; ELISA, enzyme-linked immunosorbent assay; TFA, triiluoroacetic acid PTH, phenylthiohydantoin.

massspectrum of plasmafragment 48-62. Molecular ions were observed whichclosely matched thecalculated molecular masses of peptide fragment 48-62 containing a glycan structure or truncated forms of the glycan (Table MIII). Similar results were obtained with the recombinant fragment 48-62 (Fig. M3B). As no fragmentation is normally seen by plasma desorption mass spectrometry (PDMS), the PDMS analysis was able toconfirm the difference in the amountof xylose in the two glycopeptides. However, due to suppression effects, PDMS analysis isgenerally not a quantitative method. Since the three identified monosaccharides have different masses, we were able to obtain some structural information on the glycan attached to serine 52. The smallest of the assigned molecularmasses corresponds to fragment 48-62 with attached glucose and fucose. Thisresultisin accordwith previous data indicating a glucose proximal to serine 52 (7). However, the position of the fucose residuecould not be ascertained, but in alllikelihood was either linked toglucose or directly to an aminoacid in the peptidebackbone. The location of fucose in fragment 48-62 was determined by a combination of metaboliclabeling and NH2-terminal amino acid sequencing experiments. In these studies, transfected BHK cells were cultured in medium containing ["C] fucose, and the S-pyridylethylated heavy and light chains of purified I4C-labeled factor VIIa were isolated asdescribed for plasma and recombinant factorVIIa. The specific radioactivity of the light chain was approximately twice that of the heavy chain (data not shown), indicating one fucose in the N-linked carbohydrate of the heavy and light chain, respectively, and an additional fucose attached to the light chain. The S-Pyridylethylated heavy and light chains of I4C-factor VIIa were then subjected to deglycosylation with N-glycosidase F, which cleaves N-linked glycans and leaves an aspartic acid residue in the polypeptide chain. SDS-PAGE and autoradiography indicated quantitative removal of the N-glycan from the light chain with retention of radiolabeled fucose, although in reduced amount relative to the untreated light chain (Fig. M4). In contrast, the heavy chain was only partially deglycosylated and theradiolabel lost with the N-linked glycan (Fig. M4). These results indicated thatfucose is built into the N-linked glycans of the heavy and light chains of factor VIIa as well as in a structure on the light chain not liberated by N-glycosidase F treatment. The radiolabeled derivative of the light chainwas digested with trypsinandfragments purified by HPLC(datanot shown). The radiolabeled fragments were sequenced and aliquots of the liberatedP T H amino acidswere analyzed for I4C. As expected,a fragment composed of residues 39-62 was sequenced and theradiolabel was found tobe associated with residue 60 (Table MIV).However, trace amountsof the PTHSer was obtained for residue 60, suggesting that the serinefucose linkage is unstable duringsequencing, or that the PTHderivative of the fucosylated serine residue is eluting with the same retention time as PTH-Ser and is underestimated due to a reduced response factor of the conjugate. We next performed carbohydrate composition analysis on the radiolabeled fragment 39-62 in order to determineif the radiolabel on serine60 could be liberated andidentified chemically as fucose. Perbenzoylated monosaccharide derivatives of the methanolyzed radiolabeled fragment were subjected to HPLC and the radioactivity elution position compared to that observed for a mixture of perbenzoylated fucose isomers obtained from similar methanolysis conditions (Fig. M5). Although we were unable to quantitate the fucose obtained from the radioactive fragment 39-62, three peaks of radioactivity were detected in thecolumn effluent which coincided exactly

Mutant Factor VIIa

11053

to the elution positions of the standard fucose isomers (Fig. highly sequence-specific as the vicinal serine 53 was apparM5). Moreover, theratio of 14C inthethreepeaks also ently not glycosylated. matched the ratio of the three peaks in the fucose standard. We next conducted a series of experiments comparingS52A Collectively, these data provide evidence for the existence of factor VIIa and wild-type factor VIIa to assess whether the a single fucose residue 0-glycosidically linked to serine 60 in glycosylation at serine 52 might be involved in the biological humanfactor VII, inadditiontothe microheterogeneous activity of factor VIIa. In a one-stage clotting assay, S52A glycan structures 0-linked to serine52 (Fig. 1). factor VIIa exhibited a specific activity of 26-28 units/pg, or Physicochemical and Functional Characterization of S52A -60% of that observed for wild-type factor VIIa (-45 units/ FactorVII"S52Afactor VI1 was purified to homogeneity pg). In contrast, the S-2288 amidolytic activity of a S52A from the conditionedmedium of BHK cells stably transfected factor VIIa-tissue factor complex was identical to that obwith a plasmid containing theSer52-Ala factor VI1 sequence served for a complex of wild-type factor VIIa-tissue factor. by a combination of Mono Q fast protein liquid chromatog- The latter finding suggested that the integrity of the S52A raphy and immunoaffinity chromatography usinga calcium- factor VIIa active site was not impaired and that thereduced dependentanti-factor VI1 monoclonal antibody(6). The activity observed in the clotting assay may reflectamore expression level of S52A factor VI1 was essentially the same complex phenomenon. In this regard, it is conceivable that as that observed for wild-type factor VI1 (14), and no differ- S52A factor VIIa exhibits one or more catalytic deficiencies ences were noted in its purification relative to recombinant including 1) a reduced affinity for tissue factor, 2) impaired wild-type factor VII. As with wild-type factor VII, S52A factor catalysis of factor X and/or factor IX activation, or 3) an VI1 was quantitatively converted to factor VIIa during the enhanced association with plasma extrinsic pathway inhibitor purification process (12). Purified S52A factor VIIa exhibited in the clotting assay (35). a single band by SDS-PAGE that migrated with an apparent To determine whether theassociation of S52A factor VIIa molecular weight of 50,000 in the unreduced form (Fig. M6). with tissue factor was normal, we examined the activationof Following reduction of disulfide bonds with 2-mercaptoetha- factor X by a complex of factor VIIa (S52A and wild-type) in nol, S52A factor VIIa migrated as a heavy chain (Mr 34,000) the presenceof the factor Xa-specific chromogenic substrate, and a light chain ( M , 24,000). The migration of S52A factor S-2222. In this experiment, factor X, S-2222, calcium, and VIIa in SDS-PAGE in the unreduced and reduced state was relipidated tissue factor apoprotein were preincubated in a indistinguishable from that observed for wild-typefactor VIIa cuvette for 10 min. At this point, equal amounts of either under identical conditions (Fig. M6). effective wild-type or S52A factor VIIa, equivalenttothe Fragment 48-62 from S52A factor VIIa was obtained by tissue factor concentration (10 PM), was added to the cuvette sequentialS-pyridylethylation,chainseparation, digestion and the absorbance at405 nm continuously recorded. In this with endoproteinase Asp-N, and purification by reverse-phase system,the increasein absorbance at 405 nm is directly HPLC as describedfor plasma and wild-type factor VIIa. relatedtofactorXaformation which, in turn, is directly NH2-terminal amino acid sequencing of the isolated S52A related to the rate of factor VIIa-tissue factor complex forfactor VIIa fragment 48-62 confirmed that theSer-Ala site- mation. As shown in Fig. M8, both wild-type and S52A factor directed mutagenesis of residue 52 had been achieved (Table VIIa generated factor Xa at equal rates under these experiMV). In addition,a normal yield of PTH-Ser was observed at mental conditions. Thus, by this criteria, S52A factor VIIa cycle 6 while a reduced yield of PTH-Ser was obtained in appears to associatewith tissue factor at essentially the same cycle 13 (corresponding to residue 60) consistent with a pu- rate aswild-type factor VIIa and suggests, but does not prove, tative 0-linkedfucose residue attached to serine60 similar to that the affinity of S52A factor VIIafor tissue factor is that observedfor plasma and wild-type factor VIIa. Mass identical to thatobserved for wild-type factor VIIa. spectrometry of the S52A factor VIIa fragment 48-62 (Fig. We next determined kinetic constantsfor the activation of M7) revealed a considerably simpler spectrum in comparison factors X and IX by S52A factor VIIa and wild-type factor to thatobserved for wild-type fragment 48-62 and confirmed VIIa in complex with either purified relipidated tissue factor the presence of fucose linked to fragment 48-62. Predominant apoprotein or tissue factor abundantly expressed on the surpeaks were observed at molecular mass 1932.1 and 1955.0 face of a human bladder carcinoma cell line (582). In these corresponding to MH+ and MNa' ions of fragment 48-62 experiments, the kinetic parameters ( K , and k,,,) were obwith an attached fucose residue(Fig.M7). These assigned tained by linear regression analyses of Hanes-Woolf plots molecular mass values are reasonably close to the calculated molecular mass values (Table MVI). The mass spectrometric essentially as described (34). Table I summarizes the kinetic analysis of S52A factor VIIa fragment 48-62 also confirmed data obtained for S52A factor VIIa and wild-type factor VIIa the absence of glucose and/or xylose in thisglycopeptide and in complex with soluble relipidated tissue factor apoprotein for demonstrates that the enzymaticglycosylation of serine 52 is or 582 cell surface tissue factor. The kinetic constants factor X activation by mutant andwild-type factor VIIa using soluble relipidatedtissuefactorapoprotein were virtually TABLE I Kinetic parameters of factor X and factor I X actiuation by wild-type factor VIIa and S52A factor V I I a Factor X

K,

FIG. 1. Proposed structure of carbohydrate moieties 0linked to serine 52 and serine 60 in human factor VII. The amino acid sequence 48-85 corresponding to thefirstepidermal growth factor-likedomain is shown.-Glc, glicose; Xyl, xylosk; FUC,fucose.

Wild-type VIIa + TF" S52A VIIa + TF Wild-type VIIa + 582 S52A VIIa + 582

kcat min"

PM

0.200 0.196 0.270 0.296

" TF,. rehidated tissuefactor carcinoma cell linetissue factor.

124.8 125.9 20.0 20.4

_ _

Factor IX

K",

k,,,

PM

min"

0.014 0.012 0.308 0.255

16.7 17.9 3.90 3.33

aDoorotein: 582, human bladder

Mutant Factor VIIa

11054

identical: K , = 196 nM and 200 nM, and kc,, = 126 min" and 125 min" for S52A factor VIIa and wild-type factor VIIa, respectively. Similar kinetic constants were also obtained for factorIXactivation by mutant and wild-type factor VIIa where K , = 12 and 14 nM, and kcat = 17.9 min" and 16.7 min" for S52A factor VIIa and wild-type factor VIIa, respectively. In addition, the kinetic constants for the activation of factor X and IX by cell-bound mutant or wild-type factor VIIa were, within experimental error, essentially the same (Table I). These results provide a definitive demonstration that theglycosyl moieties 0-glycosidically linked to serine52 do not contribute to the catalytic efficiency of factor VIIa towards its protein substrates, factorsX and IX. DISCUSSION

Previous studies established that human coagulation factor VI1 contains a novel trisaccharide (Xy1,-Glc) or truncated form (Xyl-Glc) 0-glycosidically linked to serine52 located in itsfirstepidermal growth factor-likedomain(7).Similar structures have previously been reported a t serine 52/serine 53 in bovine factors VI1 and IX (36), human factor IX (7), human protein Z (7), andbovine protein Z (7). Wenow report that recombinant human factor VI1 produced in babyhamster kidney cells also glycosylates this serine residue. In the present study, a glycan structure consisting of 1 glucose and 2 xylose residues was found in human plasma and recombinant factor VI1 at serine 52 where the glucose residue is linked to theserineinthepeptide backbone. Truncatedstructures consisting of a single glucose residue or together with xylose could also be demonstrated at serine52. Approximately equal amounts of the three structures were found in plasma factor VII, whereas in recombinant factor VI1 structures of a single glucose andthe full structurewith two additional xylose residues were predominant. In addition to the microheterogeneous glycan structures found at serine52, a single fucose residue was observed covalently linked to serine 60 in both plasma and recombinant factor VII. Metabolic labeling experiments using [ ' 4 C ] f ~confirmed ~ ~ ~ e the presenceof fucose at serine 60 in the isolated radiolabeled recombinant factor VI1 molecule. Carbohydrate and mass spectrometry analyses indicatedthatthe fucosylation of serine 60 was virtually quantitative.Inourstudies, we observed that the serinefucose bond was labile to acidhydrolysis, similar to that observed for other glycans 0-linked to serine or threonine. However, a reduced but significant yield of PTH-Ser-60 was noted by amino acid sequenceanalysis. This finding suggested that either thelinkage between fucose and serine is unstable, or that the PTH-derivative of the fucosylated serine residue iseluting with PTH-Serbut is underestimateddueto a reduced response factor. The attachmentof a single fucose to the peptidebackbone of factor VI1 represents the secondexample of fucose in this type of linkage reported in the literature. In a recent paper, Kentzer et al. (37)reportedonthepresence of a fucoseprotein linkage in the amino-terminal portion of recombinant human pro-urokinase expressed in a mouse hybridoma cell line Sp 2/0 Ag14. Although not known with certainty, these authors speculated that fucose was 0-linked to a threonine residue at position 18 inpro-urokinase(37).Ourfindings demonstrating fucose 0-linked to serine60 in plasma-derived factor VI1 indicates that afucose-peptidelinkage is not a unique manifestation of the eukaryotic cell line used to synthesize and express the recombinant factor VII. Furthermore, the presence of fucose 0-linked to serine 60 in recombinant factor VII, as well as theglycan structures 0-linked to serine 52 in this molecule, provide evidence that the baby hamster

kidney cells are capable of executing an unusual and probably site-specific 0-glycosylation at these residues in addition to other post or cotranslational modifications (protein folding, disulfide pairing, y-carboxylation, and N-glycosylation) found in authentic human material. In order to determine whether glycan the structure 0-linked to serine52 in factorVI1 contributed to itsbiological activity, we have constructed, by site-specific mutagenesis, a recombinant mutant factor VI1 in which serine 52 is changed to an alanine residue. The sequence of the plasmidencoding for the S52A factor VI1 was confirmed by dideoxy sequencing prior to its expressionin BHK cells. The S52A factor VI1 was subsequently purifiedfrom the cell supernatants of transfected BHK cells by Mono Q fast protein liquid chromatography and immunoaffinity chromatography using a calciumdependent monoclonal antibody column. Sequence analysis of a peptide fragment from S52A factor VIIa consisting of residues 48-62 confirmed at the protein level that serine 52 had indeed been mutated to alanine. Furthermore, mass spectrometric and carbohydrate analyses revealed that the adjacent serine 53 was not glycosylated. In a one-stage clotting assay, purified S52A factor VIIa exhibited a specific activity -60% of that observed for wild-type factor VIIa suggesting that the lack of glycosylation at serine 52 had reduced its biological activity. However, subsequentkineticanalyses usingpurified proteinsandcofactors revealed that S52A factor VIIa was indistinguishable from wild-type factor VIIa with respect to 1) amidolytic activity, 2) apparent rate of association with tissue factor, and 3) kinetic constants (K, and k,,,) for the activation of factors X and IX when complexed to eithersoluble relipidated tissue factor apoprotein or cell-surface tissue factor provided by a human bladder carcinoma cell line. Inasmuch as the kinetic parameters for the activation of factors X and IX by wild-type factor VIIa and S52A factor VIIa were essentiallythesamein apurified system,the reason(s) for the diminished activity of S52A factor VIIa relative to wild-type factor VIIa in a clotting assay remains to be established. One possible explanation for the apparent low specific activity of S52A factor VIIa involves its incipient neutralization by plasma extrinsic pathway inhibitor during the clotting assay. This Kunitz-type plasma inhibitor, also designated as lipoprotein-associated coagulation inhibitor or LACI (38), recognizes a ternary complex composed of factor VIIa, tissue factor and factor Xa (38,39). Recent evidence by Wun et al. (35) demonstratesthat LACI-depleted normal human plasma exhibiteda shorter tissue factor-induced clottingtime relative tountreatedplasma.Furthermore,the clotting time of LACI-depleted plasma increased in a linear fashion following reconstitutionwith purified LACI (35). These results provide a clear demonstration that extrinsic pathway inhibitor (or LACI) regulates coagulation during a tissue factor-induced clotting assay such as the factor VI1 assay. In this regard, it is conceivable that S52A factor VIIa exhibits a tendency to interactwith extrinsic pathway inhibitor at a faster rate thanwild-type factor VIIa thus leading to the apparentlower specific activity of S52A factor VIIa in the factor VI1 clotting assay. This possibility, among others, is currently under active investigation in our laboratory. Acknowledgments-We aregratefulto Drs.Gordon Veharand Eileen Mulvihill for generously providing us with recombinant tissue factor apoprotein and the Zem219b vector, respectively. We would also like to thank Nancy Basore, Inge Skraepgaard, Anni Demandt, and Kirsten Klausen for excellent technical assistance.

Mutant Factor VIIa

22. Sundquist, B. U. R., Kamensky, I., Hlkonsson, O., Kjellberg, J., Salehpour, M., Widdiyas, S., Fohlman, J., Peterson, P. A., and Roepstorff, P. (1984) Biomed. Mass Spectrom. 11,242-257 23. Jonsson, G. P., Hedin, A. B., Hlkonsson, O., Sundquist, B. U. R., Save, B. G. S., Nielsen, P. F., Roepstorff, P., Johansson, K. E., Kamensky, I., and Lindberg, M. S. L. (1986) Anal. Chem. 58,1084-1087 24. Nielsen, P. F., Klarskov, K.,Hojrup, P., and Roepstorff, P. (1988) Biomed. Enuiron. Mass Spectrom. 17, 355-362 25. Roepstorff, P. (1989) J. Pharm. Biomed. Anal. 7, 247-253 26. Maniatis, T., Fritch, E. F., and Sambrook, J. (1982) Molecular Cold Spring Harbor Laboratory Cloning: A Laboratory Manual, Press, Cold Spring Harbor, NY 27. Sanger, F., Necklen, S., and Coulson, A. R. (1977) Proc. Natl. Acad. Sci U. S. A. 74, 5463-5467 28. Brewer, J., Grund, E., Hagerlid, P., Olsson, I., and Lizana, J. (1986) in Electrophoresis '86 (Dunn, M. J., ed) pp. 226-229, VCH Verlagsgesellscaft, New York 29. Pedersen, A. H., Nordfang, O . , Norris, F., Wiberg, F. C., Christensen, P. M., Moeller, K. B., Meidahl-Pedersen, J., Beck, T. C., Norris, K., Hedner, U., and Kisiel, W. (1990) J . Biol. Chem. 265, 16786-16793 30. Petersen. L. C., Boel, E., Johannessen, M., and Foster, D. (1990) Biochemistry 29, 3451-3457 31. Graham, F. L., and van der Eb, A. J. (1973) Virology 52, 456. ~-

REFERENCES 1. Kisiel, W., Fujikawa, K., and Davie, E. W. (1977) Biochemistry 16,4189-4194 2. Wildgoose, P., and Kisiel, W. (1989) Blood 73, 1888-1895 3. Broze, G. J., and Majerus, P. W. (1981) Methods Enzymol 80, 228-237 4. Hagen, F. S., Gray, C. L., O'Hara, P., Grant, F. J., Saari, G. C., Woodbury, R. G., Hart, C. E., Insley, M., Kisiel, W., Kurachi, K., and Davie, E. W. (1986) Proc. Natl Acad. Sci. U. S. A . 83, 2412-2416 5. Nemerson. Y. (1988) Blood 71, 1-8 6. Thim, L., Bjoern, S:, Christensen, M., Nicolaisen, E. M., LundHansen, T., Pedersen, A. H., and Hedner, U. (1988) Biochemistry 27, 7785-7793 7. Nishimura, H., Kawabata, S., Kisiel, W., Hase, S., Ikenaka, T., Takao, T., Shiminoshi, Y., andIwanaga, S. (1989) J . Biol. Chem. 264,20320-20325 8. Sakai, T., Lund-Hansen, T., Thim, L., and Kisiel, W. (1990) J. Biol. Chem. 265, 1890-1894 9. Shapiro, R., and Vallee, B. L. (1989) Biochemistry 28, 7401-7408 10. Kondo, S., and Kisiel, W. (1987) Blood 70, 1947-1954 11. Mahoney, W. C., Kurachi, K., and Hermodson,M. A. (1980) Eur. J. Biochem. 105, 545-552 12. Bioern, S., and Thim, L. (1986) Res. Discl. 269,564-565 13. Engvall, E. (1980) Methods Enzymol. 70, 419-439 14., Wildgoose, P,, Berkner, K, L., and Kisiel, W, (1990) Biochemistry 29,3413-3420 15., Paborsky, L. R., Tate, K. M., Harris, R. J., Yansura, D. G., Band, L., McCray, G., Gorman, C. M., O'Brien, D. P., Chang, J. Y.,

46'l

32. Busby, S. Berkner, K. L., HalfPap, L. M.3 Gambee, J. E., and Kumar, A.A. (1988) in Current Advances in Vitamin K Research (Suttie, J., ed) pp. 173-183, Elsevier Science Publishing, Amsterdam 33. Smith, K. J. (1988) Blood 72, 1269-1277 34. Komiyama, Y., Pedersen, A. H., and Kisiel, W. (1990) Biochemistry 29,9418-9425 3.5, w u n , T, c., H ~M, D,, ~~ ~ ~~K, K,,, palmier, ~ M. D., t D ~ ~~ K. C., Bulock, J . W., Fok, K. F., and Broze, G. J., cJr. (1990) J. Biol. Chem. 265, 16096-16101 36. Hase, S., Kawabata, S., Nishimura, H., Takaya, H., Sueyoshi, T., Miyata, T., Iwanaga, S., Takao, T., Shimonishi, Y., and Ikenaka, T. (1988) J. Biochem. (Tokyo) 104, 867-868 37. Kentzer,E. J., Buko, A,, Menon, G., Sarin, V. K. (1990) Biochem. Biophys. Res. Commun. 171, 401-406 38. Broze, G. J., Girard, T. J., and Novotny, W. F. (1990) Biochemistry 29, 7539-7546 Rapaport, S. I. (1989) Blood 73, 359-365 J.2

R.' Fung' v' "' Thomas' N.' and Vehar' G' A' (1989) Biochemistry 28, 8072-8077 16.. Braze, G. J., Jr., Leykam, J. E., Schwartz, B. J., and Miletich, J. P. (1985) J. Biol. Chem. 260, 10917-10920 17. Bach, R., Gentry, R., and Nemerson, Y. (1986) Biochemistry 25, 4007-4020 18. Barenholz, Y., Gibbes, D., Litman, B. J., Goll, J., Thompson, T. E., and Carlson,F. D. (1977) Biochemistry 16, 2806-2810 19.Thim, L., Hansen, M. T.,andSoerensen, A. R. (1987) Lett. 212, 307-312 20. Jentoft, N. (1985) Anal. Biochem. 148, 424-433 21. Mort, A. J.,andLamport, D. T. A. (1977) Anal.Biochem. 82, 39. 289-309 J'

11055

J'

SUPPLEMENTARY MATERIAL Human Plasma and Kecomhinant Factor VII: Chscartermrionof 0Glycosylations at Serine Residues S2 and 60 and Effects of Site-Directed Muragene\ir of Serine 52 lo Alanine by

S. B~oern.D.C. Fmrer, L. Thim. F.C. Wiherg. M. Chrislenren, Y . Komigsmn. A H . Pedersen and W. Kistel EXPERIMENTAL PROCEDURES

,

~

Mutant Factor VIIa

11056 0

,

,v/,,fj""

"f

-

'

%CH3CN

%CH$N

Chemical deglycosylation of glycopeptides was achieved by treatment with $%us h y d r o g e e ( 2 1 ) . Ethandiol (10 pl) was added as a scavenger to 8 nmol of thoroughly dried glycopeptide. The reaction vessel was evacuated and cooled i n a dry ice/acetone bath -200 lii o f H F WAS distilledfrom a reservoir. Thereaction vessel war warmed to 0'C for 3 h to allow the deglycosylation reaction to proceed. T h e reaction was stopped hy the evaporation of H F at 0'C for 30 mm. and the sample redissolved i n 20 mM H C I (60pl) and acetic acid (10 PI). Extraction was follwved hy a briefcenlrifuaation step. and aminoacid sequence awlysis was w i f o r m e d on the supernatant. Enzymatic deglyco&tion of "C.fucosc laheled h&vy and light chain; o f factor Vlla ( 1 "mol) was performedbyincubationwith N-Glycosidase F (0.4 U) at 37'C i n 0.3 M Tris-HCI (pH 7.5). Following incubation for 16 h. the reaction was terminated bv the addition of 5 UIof 1 M HCI. This acidificalion uep a1w redissolved a precipitate that developed d&ng the deglyeoryidion of the factor V l l n heavy chain. Samples. hefore and after deglymsylation. were analyzed by SDS-polyacrylamide gel electrophoresis (SUSPAGE).Twoidentical gels were run and one was stained withCwmassiebluewhiletheother was suhjectcd 10 autoradiography. I,

M a c j - Molecular weight determination

was ohtained on a BIO.ION 20 plasma desorption mass spectrometer (PDMS) equipped and operated in positive mode with a flight tube of approximately I5 em as previously desoibed (22). Aliquots of the redissolved H P L C fractions were diluted with one volume of ethanol and 5 pl were applied on a nilroccllulose-coated target Of aluminized mylar foil (23). After a few minutes, the matrix was spin-dried, washed with 5-10 PIof 0.1% T F A and finally spin-dried before analysis in the mass spectrometer (24). The accelerating voltage of the PDMS was set at 15 k V and positive ions were collected for 5 million fission events. The time-of-flight spectra was calibrated by use of !he hydrogen and sodium i o m present i n the analyzed sample. The accuracy on assigned molecular ions is approximately 0.1% for well-defined peaks, otherwise somewhat less (25).

. .

SImCNral Chamremaumu$ F a r or Vfl- Human plasma and recombinant factor Vlla were reduced and alkylated with 4-vinylpyridine and the S-pyridylethylated derivatives of the light and heavy chains of factor V i l a separated andpurified by reverse-phase H P L L as described (6). Enzymatic digestion of the derivstizedlight chains (400 p g ) was performed i n 150 gI of 50 m M sodium phosphate ( p H 8.0)by addition o f 2 p g Endoproteinase Asp-N dissolved i n 50 pl o f water. The digestions were performed at 3PC lor 20 h and stopped by the addition of 1 M HCI until the p H was 2.0. The peptides i n the digest mixture were purified on a Vydac 214TP54 reverse-phase H P L C column (250 x 4.6 mm) using 0.1% T F A 8s eluent A and 0.07% T F A in 50% acetonitrile as eluent 8. Equilibration was performed with 5% B at 3U"C at a flow rate of 1.5 ml/min and injection was followed by isouatic elution with 5% B for 5 min A linear gradient from 5.90% B i n 33 min was then pumped through the column. The absorbance of the column effluent was measured at 256 nm and fractions collected that corresponded to the individual perks. Fractions were lyophilized i n a vacuum ecntrifuge. redissolved in 20 m M HCI and suhjected 1 0 N-terminal amino acid requsnce analysis, amino acid and carbohydrate composition awlysir and piasma desorption m a s 5peetrometry. The S-pyridylethylated heavy and light chains o f "C-labeled recombinant factor V i l a s e r e obtained by identieal procedures. "C-labeled hctor VIIa light chain (1.3 mg) was digested i n 400 pi of 0.1 M Tris-HCI (pH 7.5) hy the addition of 25 p g oftrypsin dissolved i n 10 pl of water. T h e digestion was carried out at 3 P C for 16 h, and the reaction sopped by the addition of 10gl uf 4 M IKI. The t w t k peptides were purified by reverse-phase H P L C by procedures previously employed for tryptic mappmg of unlaheled S-pyridylethylnted light chains o f human plasma and recombinant fdnm V l l v (6). Fractions were collected manually and aliquoa were analyzed in order to detect peptides that contained radiolabeled iucuse. which subrequentlywas subjected to modified carbohydrate analysis and NH,-termiiu amino acid q u e n e e andyris including detection of the rrdirrlahel i n the indwidual Edman cycles.

10

Time (mi",

20

Gcnrrnl Mrl/u& Standard D N A techniques were carried out as described (26). Synlhrllc oiiganucleotides were preparedby solid-phase phosphoramidite chemistry on an automated synthesizer (Applied Biosyrtems Model 380A). Nucleotide sequence determinations were performed by the dideoxy SDS-polyacrylamide slab gel electrophoresis (SDS-PAGE) war chiiin termination technique (27). performed on a Phastsystem. an integrated horizontal electrophoresis system, usmg pre-cast (50 x 43 mm) and thm (0.4 mm) IO-lS% gradientpolyacrylamide gels (28). Followingelenrophoresis. the proteins were visualized hy staining with Coomassie Brilliant Blue or autoradiography. The coagulant activities of wild-type factor V l l a and S52A factor V i l a were assessed i n a one-stage clotting assay using heredirory lactar VII-deficient plasma ( < l % factor VI1 antigen) and purified. relipidated human brain t i w e lactor apoprotein (29). The arsay was ealihrated with various dilutions of normal. pooled human p l m m where one ,#nit of factor VI1 activity is arbitrarily defined as that amount of activity i n 1 ml of plasma. factor VI1