Expression and characterization of human factor IX and factor IX-factor X chimeras in mouse C127 cells.

Human blood clotting factor IX, and two chimeric molecules of factor IX, in which the first epidermal growth factor-like domain or both epidermal growth factor-like domains have been replaced by that of human factor X, have been expressed in mouse C127 cells. The recombinants have been purified using a metal ion-dependent monoclonal antibody specific for residues 1-42 of human factor IX. All recombinant molecules are activated normally by human factor XIa in the presence of calcium ion. Activation of the factor IX recombinants by factor VIIa-tissue factor appears to be normal for the epidermal growth factor-1 exchange but considerably reduced for the construction containing both epidermal growth factor-like domains of factor X. The analysis of gamma-carboxyglutamic acid residues reveals that all of the purified recombinants are almost fully carboxylated. The extent of aspartic acid hydroxylation at residue 64 is 60% for all recombinants. The chimeric molecule with both epidermal growth factor-like domains from factor X has about 4% normal activity in the activated partial thromboplastin time assay. In contrast, the construct containing the first epidermal growth factor-like domain of factor X shows essentially normal clotting activity. Thus, it is unlikely that this domain is involved in a unique interaction with factor VIII.


Human
blood clotting factor IX, and two chimeric molecules of factor IX, in which the first epidermal growth factor-like domain or both epidermal growth factor-like domains have been replaced by that of human factor X, have been expressed in mouse Cl27 cells. The recombinants have been purified using a metal ion-dependent monoclonal antibody specific for residues l-42 of human factor IX. All recombinant molecules are activated normally by human factor XIa in the presence of calcium ion. Activation of the factor IX recombinants by factor VIIa-tissue factor appears to be normal for the epidermal growth factor-l exchange but considerably reduced for the construction containing both epidermal growth factor-like domains of factor X. The analysis of y-carboxyglutamic acid residues reveals that all of the purified recombinants are almost fully carboxylated.
The extent of aspartic acid hydroxylation at residue 64 is 60% for all recombinants. The chimeric molecule with both epidermal growth factorlike domains from factor X has about 4% normal activity in the activated partial thromboplastin time assay. In contrast, the construct containing the first epider-ma1 growth factor-like domain of factor X shows essentially normal clotting activity. Thus, it is unlikely that this domain is involved in a unique interaction with factor VIII.
Factor IX (Christmas factor) is the zymogen of a serine protease active in normal hemostasis. The enzymatic activity of factor IX requires carboxylation of specific glutamic acid residues in a vitamin K-dependent post translational modification (l-3). Factors IX, X, VII, and protein C are closely related members of the same family of plasma serine proteases. Evidence supporting the close relationship of this family is the almost complete identity of intron-exon arrangement of the genes coding for these proteins (4, 5) and their high degree of amino acid sequence identity. The apparent functional domains of these related proteins closely parallel the exon structure of their genes.
The most easily definable functional domains of these related proteins from amino to carboxyl terminus, respec- tively, are: the vitamin K-dependent domain containing the modified glutamic acid, y-carboxyglutamic acid (Gla) residues, two epidermal growth factor (EGF)l-like domains, an activation peptide region, and the catalytic domain, which confers the protease function (6,7). The vitamin K-dependent Gla domains consist of approximately the first 40 amino acids of the zymogens. There are 12 Gla residues in factor IX and 11 in factor X. These Gla residues are thought to be responsible for the calcium-mediated binding of these coagulation factors to phospholipids or platelets (8,9). The Gla domains of these proteins are followed by two EGF-like domains (EGF-1 and -2), named for their similarity to repeats of the EGF precursor. In the first EGF-like domain there is an unusual amino acid residue, identified as /3-hydroxyaspartic acid (Hya). At this single aspartic acid residue normal human factor IX is 26% hydroxylated while factor X is essentially 100% hydroxylated at a comparable site (10, 11). Following the EGF-like domains is the activation peptide region, which is cleaved by limited proteolysis to render the zymogen active. Most of the vitamin K-dependent proteins share little amino acid sequence identity in the activation peptide. Following the activation peptide region is the serine protease domain containing the catalytic triad of histidine, aspartic acid, and serine. This region shares remarkable amino acid sequence similarity with trypsin and chymotrypsin. The crystal structures of numerous families of proteins have revealed that evolutionarily related proteins with similar primary structures share a similar three-dimensional structure (12, 13). The serine proteases and their zymogens, of which factor IX is one, comprise such a family and have become textbook models of this hypothesis (14-17). The backbones of the serine proteases are essentially identical; most of the sequence changes reside in surface amino acids. Thus, the rationale of the present study was that there is a likelihood that one can exchange homologous domains between factors IX and X without significantly altering the overall three-dimensional structure. There are several reasons for choosing factors IX and X for exchanging homologous domains. In addition to the structural similarities mentioned above, the physiological reactions in which they participate are very similar. In the intrinsic pathway, factor X is activated by the factor IXa-factor VIIIa complex. This reaction is analogous to that of the conversion of prothrombin to thrombin by the factor Xa-factor Va complex. Both reactions are at least 100 times faster in the presence of phospholipids and calcium (18,19 Factor IX and Factor IX-Factor X Chimeras The EGF-like domains of factor IX are of particular interest. The function of these domains in factor IX are unknown, but EGF and other proteins containing EGF-like regions similar to those in factor IX have been shown to be involved in receptor-ligand interactions (for review, see Ref. 20). Here we report the production of chimeric forms of factor IX with the first EGF-like domain or both EGF-like domains of factor IX replaced by the homologous regions of factor X. The chimeric proteins were expressed in animal tissue cell cultures, isolated, and partially characterized structurally and functionally.
Sal1  (27). Twenty rg of BPV-factor IX DNA and 2 pg of pSV2-neo (28) were used to transfect cells grown in 60-mm dishes. Two days after transfection, cells were passaged, and grown with growth medium containing 600 pg/ml G418 for selection.
Two to 3 weeks later, the surviving foci were screened with a filter-immunoassay (29) FX-Bsl E II ANSF-LEEHKKCHLERECHEETCSYEEAREVFEDSDKTNEFUNKYKDGDOCETSPCGNGG :: ::: : :::::::: :: :::::::: : ::: : :::::: :: : : Full length factor IX cDNA (2.8 kb) is shown in black. The regions which were modified are open and are bounded by the B&E11 and Not1 sites. The expression vector contains the complete BPV genome (7.94 kb) ligated to a fragment of the plasmid pML 2 (2.63 kb), which contains a bacterial origin of replication and an ampicillin resistance gene. The factor IX cDNA is flanked 5' by a mouse metallothionein promoter (0.6 kb) and the Moloney murine sarcoma virus enhancer (0.375 kb), and 3' by the SV40 polyadenylation recognition sequences (0.85 kb). ylated factor IX by immunoradiometric assay using antibody A-7. Clones expressing a high level of carboxylated factor IX were established as cell lines for further analysis.

Polyacrylamide
Gel Electrophoresis and Immunoblotting (Western Blotting)-SDS-polyacrylamide gel electrophoresis (SDS-PAGE) was performed as described (30). Following electrophoresis, the proteins were visualized by staining with Coomassie Brilliant Blue or subjected to immunoblotting (30). When A-7 was used for immunoblotting, 5 mM CaC& and 1 mM MgCI, were included in all the buffers (31). Zodination of Protein-One-hundred pg of monoclonal antibody was labeled with lz51-Na using Iodobeads according to the manufacturer's instructions (Pierce Chemical Co.). The radioactively labeled antibody was separated from free iz51 on a Sephadex G-25 column. ~mmunoradiometric Assay-Three monoclonal antibodies (A-l, A-4, and A-7) were used in this analysis, as described previously (31). These assays show no cross-reactivity with bovine factor IX in culture supernatants. A-4 was used at an appropriate dilution in 50 mM NaHC03, pH 8.5, to coat the 96-well microtiter plate. Samples con- taining factor IX were diluted in 150 mM NaCI, 20 mM Tris-Cl, (pH 7.2), (TBS) and 0.1% ovalbumin, and subsequently added to the wells. After incubation at 4 "C overnight, second antibodies A-l or A-7, labeled with **'I, were added at 1 x lo5 cpm/well. A-7 was diluted in the presence of 5 mM CaCl, and 1 mM MgCI,. Unbound radioactive antibodies were removed after 4 h incubation and the radioactive content of each well was measured.

Purification of Recombinant Factor
IX-Factor IX was purified using batch adsorption to DEAE-Sepharose CL-6B as described (32). Ten to I5 liters of the cultured supernatant was collected, adjusted to pH 6.5 with 0.1 M citric acid, and adsorbed to 100 ml of DEAE-Sepharose CL-6B (preequilibrated in 50 mM MES, pH 6.5) per liter of media supernatant. After extensive washing, factor IX was eluted with 600 mM NaCl, 50 mM MES, pH 6.5. Fractions containing factor IX as determined by immunoradiometric assay with antibody A-7 were pooled, centrifuged at 30,000 rpm for 1 h, and then filtered through a 0.4~pm Millipore nitrocellulose membrane. The filtrate containing factor IX was diluted to 150 mM NaCl, 20 mM MgC12, 20 mM Tris-Cl, pH 7.2, and applied to a column containing the conformation-specific monoclonal antibody A-7 coupled to Affi-Prep 10 at 3-5 mg/ml. After washing the column extensively with 20 mM Tris-Cl, pH 7.2, 0.05% Tween 20, 100 mM NaCI, 20 mM MgCl*, and then washing with 20 mM EDTA in TBS. Peak fractions were assayed by immunoradiometric assay, pooled, and concentrated in either of two ways: with an Amicon microconcentrator (Centricon 30) in the presence of bovine serum albumin or on a DEAE-Sephadex A-50 column in the absence of bovine serum albumin.

Actiuation by Factor
XZa-The activation of 0.6-l fig normal human factor IX or purified recombinant factor IX by human factor XIa in the presence of 5 mM CaC& was performed at an enzyme to substrate ratio of 1:20 (w/w). Aliquots were withdrawn at intervals and subjected to SDS-PAGE analysis. Following electrotransfer of proteins to nitrocelldlose membranes, factor IXa was imaged with A-7.
Actiuation of Recombinunts by Factor VIIa and Tissue Factor-The experiment was carried out as described by Osterud and Rapaport (33). Approximately 1 pg of each recombinant was reacted with purified human factor VIIa at an enzyme to substrate ratio of 1:50. Tissue factor was from a crude human brain extract and was added at 1:lO (v/v) to the reaction mixture. Aliquots were removed at intervals and subjected to SDS-PAGE. After electrotransfer, Western blotting was performed and factor IX was detected with iodinated A-7 as described above.
Clotting Assay-One-stage activated partial thromboplastin time (aPTT) assays were performed as described (30). The ability of the proteins to correct the clotting time of factor IX-deficient plasma was calculated from a standard curve derived with pooled normal human plasma, assuming that the plasma factor IX concentration is 5 pg/ ml. IX and X there is a common Glu-Leu at residues 83-84 and 82-83, respectively.
The codons for these amino acids were converted to a unique Sac1 site. At the carboxyl terminus of the EGF-2 domain a unique Not1 site was generated at residues 132-133 (Cys-Gly) of both factors IX and X. These three restriction sites were used to generate two factor IX cDNA fragments containing EGF-1 from factor X (BstEII-Sac1 fragment) or both EGF-1 and -2 from factor X (BstEII-Not1 fragment). These two modified DNA fragments and the wild-type factor IX were inserted into the BPV vector (Fig. 2) Blue (Fig. 3B) or electrotransferred to nitrocellulose paper followed by probing with antibody A-l (Fig. 3A). The wild-type recombinant migrated as a single band and the mobility is equivalent to that of plasma IX. Analysis of the purified chimeric molecules by Coomassie Blue or silver staining also revealed a single band of the proper size. Amino acid sequence analysis confirmed that greater than 95% of each of the three recombinant proteins had the aminoterminal sequence expected of the zymogen. Less than 5% of the purified recombinants still retained the propeptide. Additionally, we used two other monoclonal antibodies specific for the first or second EGF-like domains of human factor IX to further verify the identity of the purified chimeric proteins. The result is shown in Fig. 4. As recombinant factor IX molecules were recognized by A-l and A-7 (Panels A and B, respectively); only wild-type or plasma factor IX could react with 2D5, a monoclonal antibody recognizing the EGF-1 of human factor IX (22). The failure of the two chimeric proteins IXcxepn, and IXcxepn+2) to be recognized by 2D5 confirms that the two recombinant factor IXs contain the altered EGF-1 domain (panel C). These proteins were also allowed to react with monoclonal antibody 1X-30, which recognizes the carboxyl portion of EGF-2 of human factor IX (23). Although the reactivity was weaker compared with that of the above three antibodies, it is clear that all but recombinant IX~Xegfl+a were able to demonstrate equal intensity (panel D).

Activation of Recombinant
Factor IX-The activation of the various forms of recombinant factor IX by factor XIa and factor VIIa-tissue factor complex was compared with that of plasma factor IX. The time course was followed and the activation products were analyzed by immunoblotting using monoclonal antibody A-7, which recognizes the light chain of factor IX. Fig. 5 ' 6' IO' 20' 60' 0' 3' 6' IO 20' 60' 0' 3,6' IO' 20' 60' 0' 3' 6' IO' 20' 60' Table I indicates that wild-type recombinant factor IX contains 9 Gla residues/m01 of factor IX. This is comparable to plasma factor IX, which has 9.8 Gla/ mol as measured by the same technique (34,41). Chimeric factor IXtx,,c, contained 8.7 while factor IXtxegfli2) contained 8.6. The analysis for /!-hydroxylation demonstrated 0.6 residues of P-hydroxyaspartic acid in the three recombinants in spite of the differences of the EGF-1 domains, the target of hydroxylation.

REFERENCES
Our purpose was to create factor IX molecules altered in their interaction with the other components of the coagulation system. To circumvent the problems inherent in site-directed mutagenesis in the absence of crystal structure, we elected to exchange homologous domains of closely related proteins that interact with different cofactors. It is our assumption that proteins prepared in this manner should fold correctly and have altered substrate and cofactor specificity. Our observations of normal activation by factor XIa and, in one case, normal clotting activity indicate that this was achieved. The existence of a hemophilia B patient whose defect is an apparent deletion of exon d (42) further argues for this approach rather than deletion of domains, as the patient's factor IX, in contrast to the domain exchange experiments reported here, is inactive.
Aspartic acid residue 64 of the EGF-1 domain of factor IX is normally incompletely hydroxylated (10, 11) while that of factor X is completely hydroxylated.
A consensus sequence has been identified in the EGF regions that appears to be necessary for hydroxylation (12). We therefore expected that factor IXcx,pn, might be fully hydroxylated.
This was not the case as all constructions appeared to be hydroxylated to the same extent in this expression system. It will be necessary to express the factor X gene in the same cells and determine its percent hydroxylation to resolve this problem.
The observation that factor IXcXegfl) has approximately normal activity appears to rule out the suggestion that the function of EGF-1 of factor IX is a specific interaction with factor VIII (37,43). The alternative suggestion that the EGF-1 domain of factor IX is involved in binding to the endothelial cell receptor is currently under investigation.