WO2006027111A1 - Modified coagulation factor viii with enhanced stability - Google Patents

Abstract

The present invention relates to modified nucleic acid sequences coding for coagulation factors, in particular for human factor VIII and their derivatives with improved stability, to recombinant expression vectors containing such nucleic acid sequences, to host cells transformed with such recombinant expression vectors, to recombinant polypeptides and derivatives which have biological activities of the unmodified wild type protein but have improved stability and to processes for the manufacture of such recombinant proteins and their derivatives. The invention also covers a transfer vector for use in human gene therapy, which comprises modified DNA sequences.

Description

Modified Coagulation Factor VIII with Enhanced Stability

Field of the invention:

The present invention relates to modified nucleic acid sequences coding for coagulation factors, in particular for human Factor VIII and its derivatives with improved stability, recombinant expression vectors containing such nucleic acid sequences, host cells transformed with such recombinant expression vectors, recombinant polypeptides and derivatives which do have biological activities of the unmodified wild type protein but having improved stability and processes for the manufacture of such recombinant proteins and their derivatives. The invention also relates to a transfer vector for use in human gene therapy, which comprises such modified nucleic acid sequences.

Background of the invention:

Classic hemophilia or hemophilia A is an inherited bleeding disorder. It results from a chromosome X-linked deficiency of blood coagulation Factor VIII, and affects almost exclusively males with an incidence of between one and two individuals per 10.000. The X-chromosome defect is transmitted by female carriers who are not themselves hemophiliacs. The clinical manifestation of hemophilia A is an increased bleeding tendency. Before treatment with Factor VIII concentrates was introduced the average life span for a person with severe hemophilia was less than 20 years. The use of concentrates of Factor VIII from plasma has considerably improved the situation for the hemophilia patients increasing the average life span extensively, giving most of them the possibility to live a more or less normal life. However, there have been certain problems with the plasma derived concentrates and their use, the most serious of which has been the transmission of viruses. So far, viruses causing AIDS, hepatitis B, and non-A non-B hepatitis have hit the population seriously. Since then different virus inactivation methods and new highly purified Factor VIII concentrates have recently been developed which established a very high safety standard also for plasma derived Factor VIII.

The cloning of the cDNA for Factor VIII (Wood, W.I., et al. (1984) Nature 312, 330- 336; Vehar, G.A., et al. (1984) Nature 312, 337-342) made it possible to express

Factor VIII recombinantly leading to the development of several recombinant Factor

VIII products, which were approved by the regulatory authorities between 1992 and

2003. The fact that the central B domain of the Factor VIII polypeptide chain residing between amino acids Arg-740 and Glu-1649 does not seem to be necessary for full biological activity has also led to the development of a B domain deleted Factor VII I.

The mature Factor VIII molecule consists of 2332 amino acids which can be grouped into three homologous A domains, two homologous C domains and a B domain which are arranged in the order: A1-A2-B-A3-C1-C2. During its secretion into plasma Factor VIII is processed intracellular^ into a series of metal-ion linked heterodimers as single chain Factor VIII is cleaved at the B-A3 boundary and at different sites within the B-domain. This processing leads to heavy chains consisting of the A1 , the A2 and various parts of the B-domain which have a molecular size ranging from 90 kDa to 200 kDa. The heavy chains are bound via a metal ion to the light chain, which consists of the A3, the C1 and the C2 domain (Saenko et al., 2002, BJH, 119: 323-331). In plasma this heterodimeric Factor VIII binds with high affinity to von Willebrand Factor, which protects it from premature catabolism. The half-life of circulating Factor VIII is about 12 hours in plasma. During coagulation Factor VIII is activated via proteolytic cleavage by FXa and thrombin at amino acids Arg372 and Arg740 within the heavy chain and at Arg1689 in the light chain resulting in the release of von Willebrand Factor and generating the activated Factor VIII heterotrimer which will form the tenase complex on phospholipid surfaces with FIXa and FX provided that Ca²⁺ is present. The heterotrimer consists of the A1 domain, a 50 kDa fragment, the A2 domain a 43 kDa fragment and the light chain (A3-C1-C2), a 73 kDa fragment. Thus the active form of Factor VIII (Factor Villa) consists of an A1-subunit associated through the divalent metal ion linkage to a thrombin-cleaved A3-C1-C2 light chain and a free A2 subunit relatively loosely associated with the A1 and the A3 domain.

To avoid excessive coagulation Factor Villa must be inactivated soon after activation. The inactivation of Factor Villa occurs both via the dissociation of the A2 domain (which is after thrombin activation only loosely attached to the A1 and A3 domain) as well as via activated Protein C (APC) mediated cleavage at Arg336 and Arg562. In addition to the cleavage sites on Factor Villa the binding site of APC to Factor VIII was localized to residues 2009 to 2018 (Walker et al. JBC 1990; 265: 1484-1490), and defined more precisely to residue 2001 to 2022 (Nogami et al, 2002, blood, 99, 3993-3998). In the APC mediated cleavage of FVIIIa Proteins S serves as a cofactor facilitating cleavage by APC.

Whereas in vitro experiments point to that the dissociation of the A2 domain might be rate limiting during inactivation (Saenko et al., Vox Sang. 83: 89-96 (2002)) the situation in vivo is less clear, as on the phospholipid surface FIX and FX in the tenase complex with Factor Villa might exert a stabilizing effect to the trimeric Factor Villa and in this situation APC inactivation might be decisive for inactivation. Hence reducing the susceptibility to APC cleavage on its own or in combination with mutations increasing the stability of the A2 domain might prolong the functional half- life of Factor Villa. - A -

In severe hemophilia A patients undergoing prophylactic treatment Factor VIII has to be administered i.v. about 3 times per week due to the short plasma half life of Factor VIII of about 12 hours. Each i.v. administration is cumbersome, associated with pain and entails the risk of an infection especially as this is mostly done in home treatment by the patients themselves or by the parents of children being diagnosed for hemophilia A.

It would thus be highly desirable to create a Factor VIII with increased functional half-life allowing the manufacturing of pharmaceutical compositions containing Factor VIII, which have to be administered less frequently.

In the literature there are found some publications on Factor VIII variants with resistance to APC mediated cleavage.

Pittman and Kaufman (1988, PNAS, 1988; 85: 2429-33): A modification of Arg336 to lie or Lys resulted in a Factor VIII with increased procoagulant activity. The authors postulated that this increased activity may result from greater resistance to proteolytic inactivation but performed no experiments to verify this hypothesis.

In WO87/07144 the generation of APC resistant Factor VIII molecules by mutating Arg226, Lys325, Arg336, Lys338 and Arg359 and Arg562 is described.

Pipe and Kaufman (1997 PNAS; 94:11851-11856): A Factor VIII variant which is inactivation resistant by covalently attaching the A2 domain to the A3 domain and by mutating the APC cleavage sites was made. This genetic construct was also used to generate transgenic animals as described in WO 02/072023A2.

Amano et al. (1998, Thromb. Haemost.; 79: 557-563): APC cleavage site mutants in Factor VIII were generated in mutating Arg336 to lie, Arg562 to Lys and the double mutant Arg336 to lie plus Arg562 to Lys. All mutants showed resistance to APC cleavage. Oh et al. (2002, Exp. MoI. Med.; 34: 233-238): A baculovirus expressed B-domain deleted Factor VIII with the mutation Arg336Gln showed a slightly slower decrease in activity than wild type plasma derived Factor VIII in the presence of APC.

Jenkins et al. (2002, Blood; 100: 501-508): Factor VIII was mutated at Ser558 to Phe, Val559 to Ala, Asp560 to Ala, Arg562 to Ala and Gln565 to Arg. All mutations except Arg562Ala decreased Factor VIII specific activity to less or equal of 10% of that of wild type Factor VIII. APC resistance was not tested in these mutants.

It has recently been shown that constitutively high levels of Factor VIII might constitute a risk factor for thromboembolism (Kyrle 2003, Hamostasiologie 1 : p. 41- 57). A Factor VIII which is stabilized against complete inactivation entails a potential for thrombotic events as the life saving negative feedback mechanisms of coagulation are compromised and the delicate balance between coagulation and thrombosis might shift unfavourable towards thrombosis.

Hence, there is an ongoing need to develop modified blood coagulation Factors which exhibit favourable properties.

In the present invention it has been surprisingly found that APC resistance of Factor VIII can be fine tuned in a way that reduces susceptibility to APC mediated cleavage without completely abolishing APC mediated cleavage. These mutants are less likely to confer a thrombotic risk than mutations of the cleavage sites at Arg336 and Arg562.

In a first aspect, the invention therefore relates to modulating the amino acids in the vicinity of the cleavage site of APC around Arg336 and Arg562 in order to only partially reduce the cleavage via APC. In a second aspect the invention relates to mutating the binding site for APC in order to reduce the affinity of Factor VIII to APC also leading to a decreased but not completely abolished susceptibility of Factor VIII to cleavage by APC.

One subject of the invention is, therefore a modified blood coagulation Factor VIII with a reduced susceptibility to APC mediated cleavage as compared to wild type Factor VIII a) containing a sequence of amino acids which has been changed in the vicinity of the APC cleavage site at Arg336 (without mutating Arg226, Lys325, Arg336, Lys338 and Arg359) and/or Arg562 (without mutating Arg562 itself) and or within the APC binding sites at amino acids 2001 to 2022 as compared to the sequence of the amino acids of the wild type blood coagulation Factor VIII,

b) exhibiting a specific activity of more than 10% of that of wild type Factor VIII after expression in stable cell lines and

c) which can be activated by thrombin.

The terms "blood coagulation Factor VIII", "Factor VIM" and ¹¹FVIII" are used interchangeably herein. "Blood coagulation Factor VIM" includes derivatives of wild type blood coagulation Factor VIII of human or non-human, for instance porcine origin having the procoagulant activity of wild type blood coagulation Factor VIII. Derivatives may have deletions, insertions and/or additions compared with the amino acid sequence of wild type Factor VIII. As non-limiting examples, Factor VIII molecules include full-length recombinant Factor VIII, B domain deleted Factor VIII (Pittman 1993, Blood 81 :2925-2935), Factor VIII mutants further stabilizing the A2 domain (WO 97/40145), Factor VIII mutants resulting in increased expression (Swaroop et al. 1997, JBC 272:24121-24124), Factor VIII mutants reducing its immunogenicity (Lollar 1999 Thromb. Haemost. 82:505-508), Factor VIII reconstituted from differently expressed heavy and light chains (Oh et al. 1999, Exp. MoI. Med. 31 :95-100), Factor VIII mutants reducing binding to receptors leading to catabolism of Factor VIII like HSPG (heparan sulfate proteoglycans) and/or LRP (low density lipoprotein receptor related protein) (Ananyeva et al. 2001 , TCM, 11 :251-257. Also encompassed are Factor VIII variants in which parts of the Factor VIII are replaced by corresponding parts of Factor V (Marquette et al. 1995, JBC, 270:10297-10303, Oertel et al. 1996, Thromb. Haemost. 75:36-44). A suitable test to determine the procoagulant activity of Factor VIII is the one-stage or the two- stage coagulation assay (Rizza et al. 1982 Coagulation assay of Factor VIIIc and FIXa in Bloom ed. The Hemophilias. NY Churchchill Livingston 1992). The cDNA sequence and the amino acid sequence of the mature wild type form (e.g. human wild type FVIII with its signal sequence cleaved off) of human blood coagulation Factor VIII are well known. The indication of amino acid positions herein allows the identification of the respective amino acid residues within the sequence of a polypeptide. The reference to an amino acid position of a specific sequence does not exclude the presence of mutations, e.g. deletions, insertions and/or substitutions at other positions in the sequence referred to. For example, a mutation in "Glu2004" does not exclude that in the modified homologue one or more amino acids at positions 1 through 2003 are missing.

Generally, the modified Factor VIII has an increased stability compared to the non- modified form and/or compared to the wild type form of Factor VIII. An increase in stability of the modified Factor VIII molecule can be measured as previously described by functional assays such as the determination of Factor VIII half-life by clotting time assays or by the measuring FXa generation following incubation in the presence or not of APC with or without protein S, or after treatment with thrombin by establishing a ratio between the remaining Factor Villa activity treated with APC and the remaining activity of Factor VIII not treated with APC or by Western blotting of Factor VIII before and after cleavage by APC compared to wild type Factor VIII as described in detail in the examples.

The susceptibility to APC is usually decreased by at least 50%, preferably by at least 75%, more preferably by at least 90%, compared to the non-modified form. The mutants as claimed in claims 1-7 may hereby made as a single mutation or as several mutations added so as to achieve at least 50 %, or 90 % decrease of susceptibility to APC without, as stated before, abolishing such susceptibility altogether. At the same time, the functional half life is increased as detailed below.

In another embodiment, the modified Factor VIII of the invention exhibits an increased functional half-life compared to the non-modified form and/or to the wild type form of human Factor VIII. The functional half-life can be determined in vitro as shown in figure 5 of US 2003/0125232 or as published by Sandberg (Thromb. Haemost. 2001 ; 85(1 ):93-100) and Gale (Gale et al., 2003, J. Thromb. Haemost., 1 : p. 1966-1971 ) which basically consists of determining the kinetics of Factor VIII activity after thrombin activation. In vivo one could test the modified Factor VIII in animal models of hemophilia A, like Factor VIII knockout mice, in which one would expect a longer lasting hemostatic effect of a stabilized Factor VIII or a higher hemostatic effect at the same concentration as compared to wild type Factor VIII. The hemostatic effect could be tested for example by determining time to arrest of bleeding after a tail clip.

A preferred embodiment of the invention is the combination of mutations leading to APC resistance with mutations leading to decreased dissociation of the A2 domain after thrombin activation. Modifications of FVIII to decrease dissociation of the A2 domain after thrombin activation comprise but are not limited to amino acid substitutions which lead to a covalent attachment of the A2 domain to other parts of

FVIIIa, or which increase the affinity of the A2 domain to other parts of FVIIIa for example by way of enhancing ionic interactions between the A2 domain and other parts of FVIIIa. Suitable modifications achieving a covalent attachment are for example described in the published US Patent Application with Publication No.:

US2003/0125232 A1. Here there is at positions 664 and 1826 each a Cysteine substituted for the respective wild type amino acid, thus enabling the forming of a disulfide bridge in between A2 and A3. The functional half life is usually increased by at least 50%, preferably by at least 100%, more preferably by at least 200%, even more preferably by at least 500% compared to the non-modified form and/or to the wild type form of human Factor VIII.

The functional half-life of the wild type form of human Factor Villa is 2 minutes. The functional half life of the modified Factor Villa molecule of the invention is usually at least about 3 minutes, preferably at least about 4 minutes, more preferably at least about 6 minutes, most preferably at least about 12 minutes.

The invention further relates to a polynucleotide encoding a modified human Factor VIII as described in this application. The term "polynucleotide(s)" generally refers to any polyribonucleotide or polydeoxyribonucleotide that may be unmodified RNA or DNA or modified RNA or DNA. The polynucleotide may be single- or double- stranded DNA, single or double-stranded RNA. As used herein, the term "polynucleotide(s)" also includes DNAs or RNAs that comprise one or more modified bases and/or unusual bases, such as inosine. It will be appreciated that a variety of modifications may be made to DNA and RNA that serve many useful purposes known to those of skill in the art. The term "polynucleotide(s)" as it is employed herein embraces such chemically, enzymatically or metabolically modified forms of polynucleotides, as well as the chemical forms of DNA and RNA characteristic of viruses and cells, including, for example, simple and complex cells.

The skilled person will understand that, due to the degeneracy of the genetic code, a given polypeptide can be encoded by different polynucleotides. These "variants" are encompassed by this invention.

Preferably, the polynucleotide of the invention is an isolated polynucleotide. The term "isolated" polynucleotide refers to a polynucleotide that is substantially free from other nucleic acid sequences, such as and not limited to other chromosomal and extra-chromosomal DNA and RNA. Isolated polynucleotides may be purified from a host cell. Conventional nucleic acid purification methods known to skilled artisans may be used to obtain isolated polynucleotides. The term also includes recombinant polynucleotides and chemically synthesized polynucleotides.

Yet another aspect of the invention is a plasmid or vector comprising a polynucleotide according to the invention. Preferably, the plasmid or vector is an expression vector. In a particular embodiment, the vector is a transfer vector for use in human gene therapy.

Still another aspect of the invention is a host cell comprising a polynucleotide of the invention or a plasmid or vector of the invention.

The host cells of the invention may be employed in a method of producing a modified homologue of human Factor VIII, which is part of this invention. The method comprises:

a) culturing host cells of the invention under conditions such that the modified homologue of human Factor VIII is expressed; and

b) optionally recovering the modified homologue of human Factor VIII from the host cells or from the culture medium.

It is preferred to purify the modified homologue of the present invention to > 80% purity, more preferably > 95% purity, and particularly preferred is a pharmaceutically pure state that is greater than 99.9% pure with respect to contaminating macromolecules, particularly other proteins and nucleic acids, and free of infectious and pyrogenic agents. Preferably, an isolated or purified modified homologue of the invention is substantially free of other polypeptides. The various products of the invention are useful as medicaments. Accordingly, the invention relates also to a pharmaceutical composition comprising a modified homologue of human Factor VIII as described herein, a polynucleotide of the invention, or a plasmid or vector of the invention.

Another aspect of the invention is the use of a modified homologue of human Factor VIII as described herein, of a polynucleotide of the invention, of a plasmid or vector of the invention, or of a host cell of the invention for the manufacture of a medicament for the treatment or prevention of a blood coagulation disorder. Blood coagulation disorders include but are not limited to hemophilia A. Preferably, the treatment comprises human gene therapy.

The invention also concerns a method of treating an individual suffering from a blood coagulation disorder such as hemophilia A. The method comprises administering to said individual an efficient amount of the modified homologue of human Factor VIII as described herein. In another embodiment, the method comprises administering to the individual an efficient amount of the polynucleotide of the invention or of a plasmid or vector of the invention. Alternatively, the method may comprise administering to the individual an efficient amount of the host cells of the invention described herein.

Description of the tables and drawings:

Table 1: Oligonucleotide pairs for mutagenesis of FVIII-LO. Nucleotides underlined correspond to the mutation introduced. Silent mutations introducing (c) or removing (t) an Ace I site are indicated in bold. Sense (S), Anti-Sense (AS).

Table 2: Activity of FVIII-LO and mutants expressed in BHK cells following transient transfection. Exponentially growing BHK cells were transfected with pD- FVIII-2I-L0 wild-type and mutants as described in material and methods. Two days after transfection the complete medium was replaced by GMEM containing 1% BSA and let for 6 h. The conditioned medium was recovered and centrifuged. The FVIII- δB content was determined using an ELISA kit, and the activity was measured by one-stage and two-stage clotting assays. ^* Standard Deviation of 3 determinations in 3 independent transfections.

Table 3: Expression level and activity of FVIII-LO mutants expressed constitutively in CHO cell lines. CHO cells were electroporated with pD-FVIII-2l- LO wild-type and mutants as described under material and methods. The best expressing clones were selected and spread in a 9.5 cm² well where the complete medium was replaced by IMDM containing 1% BSA and let for 6 h. The conditioned medium was recovered and centrifuged. The best expressing clones were also spread in roller bottles, where the complete medium was also replaced by IMDM containing 1% BSA and let for 16 h. The supematants were recovered and centrifuged. The FVIII- δB contents were determined using an ELISA kit. The supernatant from roller bottles was loaded on a heparin column for partial purification. The activity of the partially purified FVIII-δB was tested using a two- stage assay.

Table 4: Activity of FVIII-L4 and mutants expressed in COS-1 cells following transient transfection. Exponentially growing COS-1 cells were transfected with pD FVIII-2I-L4 wild-type and mutants as described in material and methods. Two days after transfection the complete medium was replaced by IMDM containing 1% BSA and let for 6 h. The conditioned medium was recovered and centrifuged. The FVIII-δB content was determined using an ELISA kit, and the activity by one-stage and two-stage clotting assays. * Standard Deviation of 2 determinations in 1 transfection. $ Standard Deviation of 2 determinations in 2 independent transfections. Table 5: Expression level and activity of FVIII-L4 mutants expressed constitutively in CHO cell lines. CHO cells were electroporated with pD-FVIII-2l- L4 wild-type and mutants as described under material and methods. The best expressing clones were selected and spread in a 9.5 cm² well where the complete medium was replaced by IMDM containing 1% BSA and let for 6 h. The conditioned medium was recovered and centrifuged. The best expressing clones were also spread in roller bottles, where the complete medium was also replaced by IMDM containing 1% BSA and let for 16 h. The supernatants were recovered and centrifuged. The FVIII-δB content was determined using an ELISA kit. The supernatant from roller bottles was loaded on a heparin column for partial purification. The activity of the partially purified FVIII-δB was tested using a two- stage assay.

Figure 1 : Schematic representation of the protein sequence of FVIII-LO and of APC-resistant mutants in the area of the R562 APC cleavage site. The structural domains of FVIII-LO are depicted. The linker RRRR is replacing the B domain. APC cleaves FVIII at residues R336 and R562. The arrows and numbers indicate the positions and residues around APC cleavage. R562 is underlined and the mutants generated are Q561 N-L0 and N564Q-L0.

Figure 2: Thrombin activation of FVIII-LO and mutants. Partially purified FVIII-δB wild-type and mutants (2.5U/ml) were treated with 2.5U/ml thrombin for increasing periods of time and FXa generation was measured.

Figure 3: APC Resistance ratio of FVIII-LO and mutants in the commercialized APC resistance assay Kit. FVIII-δB partially purified on heparin column were assayed using FVIII deficient plasma, in the presence or in the absence of diluted APC/CaCl₂. The APC resistance ratio was calculated by dividing the clotting time in the presence of APC, by the clotting time in the absence of APC. The values are the average of 3 determinations. Figure 4 : APC inactivation of activated FVIII-LO and APC-resistant mutants.

Partially purified FVIII-LO wild-type and mutants (1.25 nM, 1 U/ml) were incubated 2 min with thrombin (1 U/ml), 1 min with hirudin (2U/ml) and with bovine APC (1.25 nM) for increasing periods of time at 37°C. At the indicated times, FIXa, FX and spetrozyme were added, and FXa generation was measured. Each value was adjusted using the control without APC to remove the effect of A2 dissociation and visualize the APC effect only. The FVIIIa activity was expressed as percent activity of the respective control without APC. The values are the average of 2 determinations. Panel A, comparison of the APC resistance of FVIII-LO, R336I/R562K, R562K and R336I. Panel B, comparison of the APC resistance FVIII- LO, Q561 N-L0, R562K-L0 and N564Q-L0.The values are the average of 2 determinations.

Figure 5: lmmunoblot analysis of the APC-mediated cleavage at Arg562 of FVIII-LO wild-type and mutants. Partially purified FVIII-δB wild-type and mutants (29 ng) were treated with 300 ng of bovine APC in the presence of 100 mg/ml inosithin at 37°C for 1.5 h. The samples were subjected to electrophoresis, blotted and incubated with a monoclonal FVIII anti-A2 domain antibody as a primary antibody, and with a peroxidase-coupled anti-mouse antibody as a secondary antibody. A : APC cleavage of control mutants . B : APC cleavage of mutants partially resistant to APC. HC : Heavy Chain, A2 : A2 domain, A2C : C-terminal part of FVIII A2 domain.

Figure 6: Schematic representation of the protein sequence of FVIII-L4 and of APC-resistant mutants in the APC binding site. The structural domains of FVIII- L4 are depicted. The linker RRGRR replaced the B domain of FVIII. The APC- binding site is localized at the end of the A3 domain, from residues 2001 to 2022. The mutant generated was H2009A-L4. Figure 7: Thrombin activation of FVIII-L4 and mutants. Partially purified FVIII-δB wild-type and mutant (2.5U/ml) were treated with 2.5U/ml thrombin for increasing periods of time and FXa generation was measured.

Figure 8: lmmunoblot analysis of the APC-mediated cleavage at Arg562 of FVIII-L4 wild-type and mutants. Partially purified FVIII-δB wild-type and mutants

(29 ng) were treated with 300 ng of bovine APC in the presence of 100 μg/ml inosithin at 37°C for 1.5 h. The samples were subjected to electrophoresis, blotted and incubated with a monoclonal FVIIIanti-A2 domain antibody as a primary antibody, and with a peroxidase-coupled anti-mouse antibody as a secondary antibody. A : APC cleavage of control mutants . B : APC cleavage of mutants partially resistant to APC. HC : Heavy Chain, A2 : A2 domain, A2C : C-terminal parts of FVIII A2 domain.

Figure 9 : APC inactivation of activated FVIII-L4 and APC-resistant mutants. Partially purified FVIII-L4 wild-type and mutants R562K-L4 and H2009A-L4 (1.25 nM, 1 U/ml) were incubated 2 min with thrombin (1 U/ml), 1 min with hirudin (2U/ml) and with bovine APC (1.25 nM) for increasing periods of time at 37⁰C. At the indicated times, FIXa, FX and spetrozyme were added, and FXa generation was measured. Each value was adjusted using the control without APC to remove the effect of A2 dissociation and visualize the APC effect only. The FVIIIa activity was expressed as percent activity of the respective control without APC. The values are the average of 2 determinations.

Detailed description of the invention:

Protein C is part of an anticoagulant pathway to prevent excessive coagulation leading to thrombosis. While thrombin primarily acts to boost coagulation it can also in association with thrombomodulin activate Protein C. Binding of Protein C to an endothelial Protein C receptor (EPCR) seems to enhance the rate of activation. Subsequently activated Protein C (APC) is supposed to dissociate from EPCR and after forming a complex with Protein S to inactivate FVa and FVIIIa. FVIIIa is cleaved by APC at two sites within the heavy chain at Arg336 and at Arg562. For cleavage to occur APC must bind to Factor VIII. As it was shown that a peptide from residue 2009 to 2018 could inhibit binding of APC to Factor VIII (Walker et al. JBC 1990; 265: 1484-1490), and more precisely (Nogami et al, 2002, blood, 99, 3993-3998) as three peptides (residues 2001-2011 , 2005-2014 and 2013-2022) also inhibit the binding of APC to Factor VIII, the APC binding site extends from residue 2001 to 2022 of the amino acid sequence of Factor VIII.

In the present invention amino acids in the vicinity of the APC cleavage sites at Arg336 (without mutating Lys325, Arg336, Lys338 and Arg359) and/or Arg562 (without mutating Arg562 itself) where vicinity means from amino acid 327 to 346 and/or amino acid 552 to 572 and/or within the APC binding domain between amino acids 2001 to 2022 of FVIII were modified by site directed mutation. The references are made to the mature Factor VIII wild type sequence after cleavage of the signal peptide. All mutants were tested for their specific activity after recombinant expression and only followed up if their specific activity was higher than 10% of that of wild type FVIII after stable expression and partial purification over a heparin column. Mutants also only were followed up if shown that they can still be activated by thrombin. These mutant Factor VIII molecules were subsequently tested for their susceptibility to APC. It was surprisingly found that some mutations like Q561 N or N564Q within the vicinity of the APC cleavage sites, or H2009A within the APC binding domain confer a partial resistance to APC to FVIII whereas other mutations like G563A or I566M did not change the susceptibility of FVIII to APC. By following the suggested experiments it is therefore possible to select mutated FVIII molecules which will have after thrombin activation a prolonged functional half-life without completely blocking the negative feedback loop by APC mediated cleavage. Such mutants are less likely to be thrombogenic than mutants destroying the cleavage site completely like R336I or R562K. Expression of the proposed mutants

The production of recombinant mutant proteins at high levels in suitable host cells requires the assembly of the above-mentioned modified cDNAs into efficient transcriptional units together with suitable regulatory elements in a recombinant expression vector that can be propagated in various expression systems according to methods known to those skilled in the art. Efficient transcriptional regulatory elements could be derived from viruses having animal cells as their natural hosts or from the chromosomal DNA of animal cells. Preferably, promoter-enhancer combinations derived from the Simian Virus 40, adenovirus, BK polyoma virus, human cytomegalovirus, or the long terminal repeat of Rous sarcoma virus, or promoter-enhancer combinations including strongly constitutively transcribed genes in animal cells like beta-actin or GRP78 can be used. In order to achieve stable high levels of mRNA transcribed from the cDNAs, the transcriptional unit should contain in its 3'-proximal part a DNA region encoding a transcriptional termination- polyadenylation sequence. Preferably, this sequence is derived from the Simian Virus 40 early transcriptional region, the rabbit beta-globin gene, or the human tissue plasminogen activator gene.

The cDNAs are then integrated into the genome of a suitable host cell line for expression of the Factor VIII proteins. Preferably this cell line should be an animal cell-line of vertebrate origin in order to ensure correct folding, disulfide bond formation, asparagine-linked glycosylation and other post-translational modifications as well as secretion into the cultivation medium. Examples of other post-translational modifications are tyrosine O-sulfation, and proteolytic processing of the nascent polypeptide chain. Examples of cell lines that can be use are monkey COS-cells, mouse L-cells, mouse C127-cells, hamster BHK-21 cells, human embryonic kidney 293 cells, and preferentially hamster CHO-cells.

The recombinant expression vector encoding the corresponding cDNAs can be introduced into an animal cell line in several different ways. For instance, recombinant expression vectors can be created from vectors based on different animal viruses. Examples of these are vectors based on baculovirus, vaccinia virus, adenovirus, and preferably bovine papilloma virus.

The transcription units encoding the corresponding DNA's can also be introduced into animal cells together with another recombinant gene which may function as a dominant selectable marker in these cells in order to facilitate the isolation of specific cell clones which have integrated the recombinant DNA into their genome.

Examples of this type of dominant selectable marker genes are Tn5 amino glycoside phosphotransferase, conferring resistance to geneticin (G418), hygromycin phosphotransferase, conferring resistance to hygromycin, and puromycin acetyl transferase, conferring resistance to puromycin. The recombinant expression vector encoding such a selectable marker can reside either on the same vector as the one encoding the cDNA of the desired protein, or it can be encoded on a separate vector which is simultaneously introduced and integrated to the genome of the host cell, frequently resulting in a tight physical linkage between the different transcription units.

Other types of selectable marker genes which can be used together with the cDNA of the desired protein are based on various transcription units encoding dihydrofolate reductase (dhfr). After introduction of this type of gene into cells lacking endogenous dhfr-activity, preferentially CHO-cells (DUKX-B11 , DG-44) it will enable these to grow in media lacking nucleosides. An example of such a medium is Ham's F12 without hypoxanthine, thymidin, and glycine. These dhfr- genes can be introduced together with the Factor VIII cDNA transcriptional units into CHO-cells of the above type, either linked on the same vector or on different vectors, thus creating dhfr-positive cell lines producing recombinant protein.

If the above cell lines are grown in the presence of the cytotoxic dhfr-inhibitor methotrexate, new cell lines resistant to methotrexate will emerge. These cell lines may produce recombinant protein at an increased rate due to the amplified number of linked dhfr and the desired protein's transcriptional units. When propagating these cell lines in increasing concentrations of methotrexate (1-10000 nM), new cell lines can be obtained which produce the desired protein at a very high rate.

The above cell lines producing the desired protein can be grown on a large scale, either in suspension culture or on various solid supports. Examples of these supports are micro carriers based on dextran or collagen matrices, or solid supports in the form of hollow fibres or various ceramic materials. When grown in cell suspension culture or on micro carriers the culture of the above cell lines can be performed either as a bath culture or as a perfusion culture with continuous production of conditioned medium over extended periods of time. Thus, according to the present invention, the above cell lines are well suited for the development of an industrial process for the production of the desired recombinant mutant proteins

The recombinant mutant protein, which accumulates in the medium of secreting cells of the above types, can be concentrated and purified by a variety of biochemical and chromatographic methods, including methods utilizing differences in size, charge, hydrophobicity, solubility, specific affinity, etc. between the desired protein and other substances in the cell cultivation medium.

An example of such purification is the adsorption of the recombinant mutant protein to a monoclonal antibody which is immobilised on a solid support. After desorption, the protein can be further purified by a variety of chromatographic techniques based on the above properties.

The recombinant proteins described in this invention can be formulated into pharmaceutical preparations for therapeutic use. The purified proteins may be dissolved in conventional physiologically compatible aqueous buffer solutions to which there may be added, optionally, pharmaceutical excipients to provide pharmaceutical preparations. The modified polynucleotides (e.g. DNA) of this invention may also be integrated into a transfer vector for use in the human gene therapy.

The various embodiments described herein may be combined with each other. The present invention will be further described more in detail in the following examples thereof. This description of specific embodiments of the invention will be made in conjunction with the appended figures.

Examples:

Example 1 : APC cleavage site mutants

All Factor VIII cDNAs were obtained by mutating a cDNA previously described (Plantier JL et al. Thromb Haemost, 2001 , 86, 596-603), containing the truncated FIX intron I sequence in 1 and 13 locations, and expressing a B-domain-deleted human Factor VIII (FVIII-δB) (Figure 1), called FVIII-LO. In this construct, the linker R₇₄o-RRRR-Ei₆₄₉ is replacing the B domain. Mutants of the B-domain-deleted human Factor VIII were generated by substitution of one amino acid respectively in the vicinity of the R562 APC cleavage site : Q561 N, N564Q (Figure 1 ). In addition 3 mutants were generated as a control : R336I, R562K and 336I/562K. As described (Amano K et al. Thromb Haemost, 1998, 79, 557-563) these mutants are not cleaved by APC at the respective mutated sites in contrast to wild type Factor VIII.

1.1 Generation of pD-FVIII-L0-2l mutants vectors

The restriction enzymes were from Promega, Charbonnieres, France. The QuickChange XL mutagenesis Kit (Stratagene, La JoIIa, CA) was used to generate the cDNA of the following Factor VIII mutants : Q561 N, N564Q, R336I, R562K and 336I/562K. The plasmid pKS-FVIII-L0-2l (Plantier JL et al. Thromb Haemost, 2001 , 86, 596-603) was used as template together with the oligonucleotides (Eurogentec S.A., Seraing, Belgium) listed in Table 1. These oligonucleotide sequences contain an additional silent mutation creating or destroying an Acc\ restriction site, used for the screening of mutated clones. The Bgl\\/BamH\ fragment of the mutated plasmid obtained was controlled by sequencing and reintroduced into pKS-FVIII-L0-2l. For R336I, the fragment SpeUBglW of the mutated plasmid was cleaved off and reinserted in the vector pKS-FVIII-L0-2l. To build pKS-FVIII-L0-2l-336l/562K, the fragment SpeUBglU of pKS-FVIII-L0-2l-R336l was cloned into the vector pKS-FVIII- L0-2I-R562K. The Not\/Xho\ fragment of the final pKS-FVIII-L0-2l-mutants constructs, containing the FVIII-δB full sequence, was introduced into pcDNA3 expression vector (pD) (Invitrogen, Leek, The Netherlands) opened by Not\ and Xho\ to generate pD-FVIII-L0-2l-mutant vectors.

1.2 Transfection conditions

The BHK (Baby Hamster Kidney) cell line was obtained from the European Collection of Cell Cultures (ECACC 85011433). The cells were maintained at 37°C under 5% CO₂ in GMEM medium (Gibco BRL Life Technologies, Cergy Pontoise, France) supplemented with 10% fetal calf serum (ICN, Orsay, France), 2mM L- Glutamine and 1% penistreptomycin (Gibco BRL Life Technologies). The COS-1 cells where maintained under the same conditions in IMDM with 10% fetal calf serum. BHK cells or COS-1 cells (1.10⁶) were transfected using 5 μl FuGENE™ 6 (Roche Diagnostics, Meylan, France) precomplexed with 1 μg of circular pD-FVIII- L0-2I wild-type or mutants vectors. The medium was removed 6 hours after transfection and replaced by fresh medium. Forty eight hours later, the cells were washed 3 times with PBS and incubated with 500 μl of appropriate SVF-free medium containing 1% BSA (Gibco BRL Life Technologies). After 6 hours, the supernatant was collected, centrifuged and frozen before further analysis.

CHO (Chinese Hamster Ovary) cells (7.10⁶) (ECACC 85050302) were transfected by electroporation with pD-FVIII-L0-2l mutants recombinant vectors (10 μg DNA) linearized by Pvu\. The medium was removed 16 h after transfection and replaced by fresh IMDM medium containing 0.6 mg/ml geneticin (Gibco BRL Life

Technologies). After obtaining geneticin resistant clones, immunoblotting was performed on plate replica, using a polyclonal sheep anti-human factor VIII antibody (Cedarlane, Le Perray-en Yvelynes, France) (see procedure in immunoblot analysis, example 1.8 in this patent application). To obtain isolated clones, picked cells were expanded and subjected to limited dilutions in 96-well plates. The best factor VIII expressing clones were expanded. The cells were spread in a six well plate (1.5.10⁶ / well), and 24 h later were washed 3 times with PBS and incubated with 500 μl of medium supplemented with 1% BSA (Gibco BRL Life Technologies). After 6 hours, the supernatant was collected, centrifuged and frozen before further analysis.

1.3 Large scale production of FVIII-δB

In order to obtain sufficient amount of FVIII-δB for purification, either Nunc triple flask, 500 cm² (Nunc, Glostrup, Denmark) or folded roller bottles, 1700 cm² (Corning, NY, US) were used. The producing clones were spread using 9.10⁶ cells/flask or 28.10⁶ cells/roller. Three days later the cells were rinsed twice with PBS and cultured in 21 or 24 ml of serum-free IMDM medium containing 1% BSA, respectively. The supernatant was collected 16 hours later, centrifuged and frozen at -80°C. The cells were cultured again in medium + 10% SVF in order to allow them to recover during 8 hours. This procedure was repeated three times for each flask or roller bottle. The amount of FVIII-δB produced was determined using an ELISA kit (Asserachrom FVlIkAg, Diagnostica Stago, Asnieres, France).

1.4 Protein purification

FVIII-δB purification was operated with a FPLC™ System (AKTA, Amersham). A 5 ml Hi-Trap heparin column (Amersham Pharmacia Biotech, Orsay, France) was equilibrated with 60 ml of Buffer A (5 mM CaCI2, 20 mM HEPES pH 7.4 and 0.01% Tween). The supernatant was thawed, 0.45 μm filtered (Minisart, Sartorius, Hannover, Germany) and loaded at room temperature on the column, that was then washed with 25 ml of 125 mM NaCI in Buffer A. FVIII-δB was eluted using a 13 ml gradient of NaCI ranging from 125 to 800 mM in Buffer A, and 1 ml fractions were collected. The fractions containing FVIII-δB were identified and quantified using an ELISA kit, aliquoted and stored at - 80⁰C. The column was regenerated using 40 ml of 3M NaCI in Buffer A, and with 12.5 ml of 0.5 M NaOH, and stored at +4°C in 20% ethanol containing 50 mM sodium acetate. 1.5 Factor VIII antigen and coagulant activity measurements

Factor VIII antigen was quantified either in the supernatant of cells producing FVIII- δB, or after purification on heparin column, using an ELISA kit (Asserachrom VIIIC:Ag, Diagnostica Stago) following the recommendations from the manufacturer.

Factor VIII activity was determined either by one-stage clotting assay and/or by two- stage clotting assay (chromogenic assay). The one-stage clotting assay was performed either on a STA Compact coagulometer, or on a KC-10 apparatus, using 50 μl of diluted FVIM-δB sample, 50 μl of plasma depleted in Factor VIII, 50 μl of APTT and 50 μl of CaCI₂ 50 mmol/l (Diagnostica Stago). The two-stage clotting assay was performed using a chromogenic Factor VIII assay (COAMATIC Factor VIII kit, Chromogenix, Milano, Italy) following the recommendations of the manufacturer. The standard curve was obtained using dilutions of ReFacto (WYETH-Lederle Nordiska AB, Solna, Sweden).

1.6 Thrombin activation

Partially purified FVIII-δB (2.5U/ml) was incubated at 37°C in 150 mM NaCI, 20 mM HEPES pH 7.4, 5 mM CaCI₂ and 1% BSA (Sigma-Aldrich, Saint-Quentin Fallavier,

France), with 2 μM PC/PS (75/25 mol/mol) (produced as described by Hope MJ,

Bally MB, Webb G, CuIMs PR. Biochim Biophys Acta. 1985; 812:55-65), . Human-

Thrombin (2.5U/ml) (Synapse b.v., Maastricht, The Netherlands) was added at different time points. The reaction was then stopped by 2.5U/ml hirudin (Diagnostica Stago). and simultaneously, a mixture containing stock solutions of

FX (100 nM final concentration, Diagnostica Stago), FIXa (1 nM final concentration,

Haematologic technologies, Essex Junction, VT, US) and Spectrozyme (0.5 mM final concentration, CBS31.39, Diagnostica Stago) was added. Factor Xa generation was followed by the absorbance variation for 10 minutes at 405 nm. For each time point, the initial rate of FXa generation was established. 1.7 Kit for APC resistance assays

The commercially available APC resistance assay kit (COATEST APC Resistance, Chromogenix) was used according to the manufacturer recommendations with the following modifications. FVIII-ΔB partially purified on a heparin column was diluted with FVIII-deficient human plasma (Diagnostica Stago) to 200 ng/ml. A sample of that dilution (50 μl) was placed in a cuvette into a KC-10 apparatus at 37°C, and incubated for 2 min. Then, 50 μl of pre-warmed CaCI₂ and 50 μl of pre-warmed APTT reagent (Diagnostica Stago) were added. The timing for clot formation began simultaneously. For each sample a second analysis was performed, replacing CaCI₂ with 4 fold diluted APCVCaCI₂ provided by the kit, and the time for clot formation was recorded. The APC resistance ratio was calculated by dividing the clotting time in presence of APC, by the clotting time in absence of APC. To define the appropriate dilution of APC/CaCI₂, different dilutions were tested and the one giving an optimal difference in the ratio of FVIII-LO and R562K-L0 was selected.

1.8 lmmunoblot analysis of APC cleavage of FVIII-δB

FVIII-δB (29 ng) was incubated at 37°C in 150 mM NaCI, 20 mM HEPES pH 7.4, 5 mM CaCI2 and 2.5 % glycerol containing 100 μg/ml inosithin (Asolectin from

Soybean, mixture of phospholipids, Fluka, St Gallen, Switzerland),. An excess of bovine APC (300 ng, molecular ratio APC/Factor VIII 27/1 ) (Haematologic

Technologies) was added for 90 minutes. A control without APC was performed in parallel. The 20 μl reaction was stopped with Laemmli buffer and boiled. Each sample was subjected to electrophoresis either on a 4-15 % polyacrylamide gel

(Bio-Rad, Ivry sur Seine, France) or on a 12% gel prepared with ProSieve 50 gel solution (TEBU, Le Perray-en-Yvelynes, France) and semi-dry blotted onto Hybond-

C Pure membrane (Amersham Pharmacia Biotech). Precision Plus Protein™

Standards (Bio-Rad) were loaded simultaneously with the samples. The blots were blocked with 1 % powder milk in TBS-Tween for 1 h at room temperature and then incubated with a 1 :1000 dilution of a mouse FVIII anti-A2 domain antibody (Green Mountain Antibody, Burlington, VT, US) in TBS-Tween. After three washes, the membrane was incubated with a peroxidase-coupled anti-mouse antibody (Bio- Rad) for further detection of chemiluminescent signals by autoradiography using the ECL System (Amersham).

1.9 APC inactivation of activated FVIII-δB

Inactivation of FVIIIa by APC was performed according to Shen et al. (1997, Thromb Haemost 78, 1030-6) with the following modifications. In a microtiter plate, partially purified FVIII-δB (1.25 nM, 1 U/ml) was incubated at 37°C in 150 mM NaCI, 20 mM HEPES pH 7.4, 5 mM CaCI₂ and 1% BSA (Sigma-Aldrich), with 50 μM PC/PS (75/25 mol/mol), in a total volume of 100 μl. Human-thrombin (1U/ml) (Synapse b.v.) was added for 2 min. The reaction was stopped by 2U/ml hirudin (Diagnostica Stago) for 1 min. Then bovine APC (1.25 nM, Haematologic Technologies) was added for increasing periods of time. A fresh preparation containing stock solutions of FX (100 nM final concentration, Diagnostica Stago), FIXa (1 nM final concentration, Haematologic Technologies) and Spectrozyme (0.5 mM final concentration, CBS31.39, Diagnostica Stago) was then added. The amount of FXa generated was monitored by hydrolysis of CBS31.39 at 405 nm.

1.10 Results

1.10.1 Q561N-L0 and N564Q-L0 are secreted and active

The following constructs were transiently transfected using FuGENE™ 6 in BHK cell line: pD-FVIII-L0-2l as wild-type control, pD-FVIII-L0-2l-Q561 N, pD-FVIII-L0-2l- N564Q, pD-FVIII-L0-2l-R336l, pD-FVIII-L0-2l-R562K and pD-FVIII-L0-2l-336l/562K, the later being the control completely resistant to APC cleavage. Two days following transfection, the cells were incubated for 6 h in IMDM containing 1% BSA. The quantity of FVIII-δB produced in the cell-conditioned medium was quantified using an ELISA. Three independent transfections were performed, except for R336I-L0, R562K-L0 and 336I/562K-L0 for which a single experiment was done. All the mutant proteins were retrieved in the cell culture medium.

Factor VIII activity was determined in the supernatants by one-stage clotting assay and by two-stage clotting assay for each transfection. The supernatants were diluted in order to have the same amount of FVIII-δB in each sample. The values (+/- standard deviation) obtained for mutant are given in Table 2, in comparison to the wild-type FVIII-LO activity (100%).

1.10.2 Characterization of FVIII-δB produced by CHO cell lines

In order to produce sufficient amounts of FVIII-δB for purification and characterization, stable transfections of a CHO cell line were generated by electroporation. For each mutant, the CHO clone that was found to be the best expressing clone is given in Table 3. The level of expression in the supernatant of cells in a well of 9.5 cm² was determined using an ELISA kit (see Table 3). These clones were used for the production of FVIII-δB mutants. Levels of expression obtained in a roller bottle are also given in Table 3.

The supernatant of cells grown either in triple flasks (in case of clone W6) or in roller bottles (for all other clones) was loaded on a heparin column for partial purification. The partially purified FVIII-δB variants were tested for activity using a chromogenic assay (see Table 3). In conclusion, the activities of Q561N-L0 and N564Q-L0 were similar after purification than those found in the supernatants (see Table 2). A discrepancy was found for the control mutants (R336I-L0, R562K-L0, 336I/562K-L0) between purified material and supernatants

The FVIII-δB obtained after partial purification on heparin column was treated with thrombin for increasing periods of time, and FXa generation was measured. The time-course analysis of Factor Villa activity demonstrated peak activity at 30 to 60 seconds (Figure 2), with a 17- to 21-fold increase of Factor VIII activity for FVIII-LO and the mutants Q561N-L0 and N564Q-L0. Therefore each Factor VIII mutant was shown that it can still be activated by thrombin.

1.10.3 Q561N-L0 and N564Q-L0 are partially resistant to APC cleavage

A commercially available APC Resistance assay kit was used to detect APC resistant FVIII-DB. This assay was performed by reconstitution of FVIII deficient human plasma with partially purified FVIII-LO wild-type and mutant proteins (see example 1.7 in this application). Each sample was tested 3 times for its clotting time in the absence or in the presence of diluted APC. The APC resistance ratio was calculated by dividing the clotting time in presence of APC, by the clotting time in the absence of APC (Figure 3). The mutant Q561 N-L0 presented an APC resistance ratio close to the one of R562K-L0, and therefore demonstrated a tendency to resist to APC. N564Q-L0 has a ratio slightly lower than FVIII-LO. In order to confirm these data, the cleavage by APC was followed by immunoblot analysis.

Partially purified ΔB-FVIII wild-type and mutants were activated by thrombin. The thrombin was then inactivated by hirudin, and the FVIIIa produced was incubated with or without bovine APC. At different time points, the activity of the remaining FVIIIa was measured in presence of FIXa, FX and spectrozyme. Two curves were obtained from such values for each mutant (data not shown). The first one, following the incubation without thrombin, showed the diminution of FVIIIa activity due to the dissociation of the A2 domain only. The second curve showed a faster loss of FVIIIa activity due to the combination of A2 dissociation and APC cleavage. The A2 dissociation phenomenon was deduced from the second curve and the residual activity (in %) compared to time zero was plotted against the time (figure 4, panel AJB). The FVIII-LO was rapidly degraded following bovine APC cleavage as well as the mutant R336I. In contrast the double mutant R336I/R562K-L0 remains insensitive to APC whereas the single mutant R562K-L0 showed an intermediate profile (figure 4, panel A). The mutants Q561N-L0 and N564Q-L0 were compared with FVIII-LO and R562K-L0 where they showed an intermediate profile (figure 4, panel B). This result demonstrated that the mutants Q561 N-L0 and N564Q-L0 were partially resistant to bovine APC.

Partially purified FVIII-δB wild-type and mutants were treated with an excess of bovine APC (bAPC) for 90 min. The cleavage was analyzed by a Western-blot performed after electrophoresis, using a Factor VIII anti-A2 domain antibody. Using this antibody, untreated FVIII-LO wild-type is visualized (Figure 5A, lane 1) by a band running at around 80 kDa, representing the heavy chain (HC), and another band running at 37 kDa representing the complete A2 domain. In the presence of bAPC, an additional band appeared at around 18 kDa, being the C-terminal part of the A2 domain (A2C) resulting from the APC cleavage of the A2 domain (Figure 5A, lane 2). The A2C species appeared also for the Factor VIII mutant R336I-L0 (Figure 5A, lanes 5 and 6), because this mutant is unaffected at Arg562. However, the Factor VIII mutants R562K-L0 and 336I/562K-L0 are not cleaved by bAPC at all (Figure 5A, lanes 3 and 4, 7 and 8), and no A2C band could be detected.

Compared to FVIII-LO cleavage by APC (Figure 5B, lanes 1 and 2), the mutants Q561 N-LO and N564Q-L0 are only partially cleaved by APC (Figure 5A, lanes 3 and 4, 5 and 6). Other mutants generated, e.g. G563A-L0 and I566M-L0, showed the same cleavage pattern as FVIII-LO (data not shown), indicating that such assays allowed us to discriminate the behavior between non cleaved molecule and totally cleaved ones.

In conclusion, FVIII-δB mutants Q561N-L0 and N564Q-L0 possessed procoagulant activity, and a profile of thrombin activation identical than the control FVIII-LO. In addition these mutants were shown to be partially resistant to APC cleavage. Example 2 : APC binding site mutants

All Factor VIII cDNAs were obtained by mutating a cDNA containing the truncated FIX intron I sequence in 1 and 13 locations, and expressing a B-domain-deleted human Factor VIII (FVIII-δB) (Figure 6), called FVIII-L4. In these constructs, the linker R740-RRGRR-E1649 is replacing the B domain. Mutants of the B-domain- deleted human Factor VIII were generated by substitution of one amino acid respectively within the APC binding site located at the end of the A3 domain from residue 2001 to residue 2022 (Figure 6) (Walker et al. J. Biol. Chem., 1990, 265, 1484-148, Nogami et al, 2002, blood, 99, 3993-3998). Mutant H2009A was obtained. In addition, the R562K mutant was generated based on the FVIII-L4 backbone to be used as a control.

2.1 Material and Methods To generate FVIII-L4, site-directed mutagenesis was performed with a kit (QuickChange SiteDirected Mutagenesis Kit, Stratagene), on pCR2-FVIII and using the oligonucleotides 5'gaaccaagacgtcgtggtcgacgagaaataac3' (L4.S) (=SEQ ID No.1 ) and 5'gttatttctcgtcgaccacgacgtcttggttc3' (L4.AS) (=SEQ ID No.2), to give pCR2-FVIII-L4. the fragment SaIlIBgIW from pCR2-FVIII-L4 was introduced into pKS-FVIII-L0-2l cut by the same enzymes to obtain pKS-FVIII-L4-2l. Finally, the Not\IXho\ fragment of pKS-FVIII-L4-2l construct, containing the FVIII-δB full sequence, was introduced into pcDNA3 expression vector (pD) (Invitrogen) opened by Not\ and Xho\ to generate pD-FVIII-L4-2l vector.

To generate the H2009A mutant, site-directed mutagenesis was performed on pKS- FVIII-L4-2I with a commercially available mutagenesis kit (QuickChange SiteDirected Mutagenesis Kit, Stratagene) according to the manufacturer's instructions. Oligonucleotides used for mutagenesis were 5'cttattggcgagcatctagctgctgggatgagcacactttttc3' (H2009A S) (=SEQ ID No.3) and 5'gaaaaagtgtgctcatcccagcagctagatgctcgccaataag3' (H2009A AS) (=SEQ ID No.4) DNA sequence of mutant clones was verified between Pf/MI and SspEI restriction sites and mutant PfM\ I BspE\ fragments were transferred into the pcDNA3 expression vector called pD-FVIII-L4-2l, which was previously digested with Pf/MI and SspEI and gel-purified, to generate pD-FVIII-L4-2l-H2009A. The expression vector pD-FVIII-L4-2l-R562K was obtained by subcloning fragment Bgl\\/BamH\ of the vector pKS-FVIII-L0-2l-R562K into pKS-FVIII-L4-2l. The Not\/Xho\ fragment of pKS-FVIII-L4-2l-R562K construct, containing the FVIII-δB full sequence, was introduced into pcDNA3 expression vector (pD) (Invitrogen) opened by Noti and Xho\ to generate pD-FVIII-L4-2l-R562K vector.

All the following material and methods were described in examples 1.2 to 1.9.

2.2 Results

2.2.1 H2009A-L4 is secreted and active

The following constructs were transiently transfected using FuGENE™ 6 in a COS- 1 cell line: pD-FVIII-L4-2l as wild-type control, pD-FVIII-L4-2l-R562K being the control resistant to APC cleavage at the second APC cleavage site and pD-FVIII- L4-2I-H2009A. Two days following transfection, the cells were incubated for 6 h in IMDM containing 1% BSA. The quantity of FVIII-δB produced in the cell conditioned medium was quantified using an ELISA. Two independent transfections were performed. All the mutants were retrieved in the cell culture medium.

Factor VIII activity was determined in the supematants by a chromogenic assay for each transfection. The supematants were diluted in the reaction buffer in order to have the same amount of FVIII-δB from each sample. The mean (+/- standard deviation) of the 2 values obtained per mutant is given in Table 4, in comparison to the wild-type FVIII-L4 activity (100%). 2.2.2 Characterization of FVIII-ffi produced by CHO cell lines

In order to produce sufficient amounts of FVIII-δB for purification and characterization, stable transfections of a CHO cell line were generated by electroporation. For each mutant, the CHO clone that was found to be the best expressing clone is given in Table 5. The level of expression in the supernatant of cells in a 9.5 cm² well was determined using an ELISA kit (see Table 5). These clones were used for the production of FVIII-δB mutants. Levels of expression obtained in a roller bottle are also given Table 5. The clones expressing mutants of FVIII-δB were producing no more than FVIII-L4 wild-type.

The supernatant of cells grown in roller bottles was partially purified on a heparin column. The partially purified FVIII-δB was tested for activity using a chromogenic assay (see Table 5). In conclusion, the activity of H2009A-L4 was found at a similar range than those found in the supernatant (see Table 4).

The FVIII-δB obtained after partial purification on a heparin column was treated with thrombin for increasing periods of time, and FXa generation was measured. The time-course analysis of Factor Villa activity demonstrated peak activity at 30 to 60 seconds (Figure 7), with a 20 to 23- fold increase of Factor VIII activity for FVIII-L4 and the mutant H2009A-L4. Therefore Factor VIII mutant H2009A-L4 was shown that it can still be activated by thrombin.

2.2.3 H2009A-L4 is- partially resistant to APC cleavage

Partially purified DB-FVIII-L4 wild-type, R562K-L4 and H2009A-L4 mutants were activated by thrombin. The thrombin was then inactivated by hirudin, and the FVIIIa produced was incubated with or without bovine APC. At different time points, the activity of the remaining FVIIIa was measured in presence of FIXa, FX and spectrozyme. Two curves were obtained from such values for each mutant. The first one, following the incubation without thrombin, showed the diminution of FVIIIa activity due to the dissociation of the A2 domain only. The second curve showed a faster loss of FVIIIa activity due to the combination of A2 dissociation and APC cleavage. The A2 dissociation phenomenon was deduced from the second curve and the residual activity (in %) compared to time zero was plotted against the time (figure 8). The FVIII-L4 was rapidly degraded following bovine APC cleavage whereas the R562K -L4 showed a partial resistance to cleavage, as already shown. The mutant H2009A-L4 showed an intermediate resistance to APC cleavage suggesting that the binding of APC to FVIIIa was altered.

Partially purified FVIII-δB wild-type and mutants were treated with an excess of bovine APC for 90 min. The cleavage was analyzed by a Western-blot performed after electrophoresis, using a Factor VIII anti-A2 domain antibody. Using this antibody, untreated FVIII-L4 wild-type is visualized (Figure 9A, lane 1) as two predominant bands: one running at around 80 kDa, representing the heavy chain (HC), and another band running at 37 kDa representing the complete A2 domain. In the presence of bAPC, the C-terminal parts of the A2 domain (A2C) appeared as a doublet between 15 and 20 kDa (Figure 9A, lane 2). The A2C species did not appear for the Factor VIII mutant R562K-L4 (Figure 9A, lanes 3 and 4). The A2 domain remained intact because this mutant cannot be cleaved by APC at residue 562.

Compared to FVIII-L4 cleavage by APC (Figure 9B, lanes 1 and 2), the mutant H2009A-L4 is only partially cleaved by APC (Figure 9A, lanes 3 and 4), because the intensities of A2C bands are weaker than those for FVIII-L4. Therefore, FVIII-δB mutant H2009A-L4 was shown to have a procoagulant activity and a profile of thrombin activation identical to that of the control FVIII-L4. In addition this mutant was partially resistant to APC cleavage. SEQUENCE LISTING

<110> ZLB Behring GmbH

<120> Modified Coagulation Factor VIII with Enhanced Stability

<130> C9P94EP

<160> 12

<170> Patentln version 3.1

<210> 1

<211> 32

<212> DNA

<213> Artificial <400> 1 gaaccaagac gtcgtggtcg acgagaaata ac 32

<210> 2

<211> 32

<212> DNA

<213> Artificial

<400> 2 gttatttctc gtcgaccacg acgtcttggt tc 32

<210> 3

<211> 43

<212> DNA

<213> Artificial

<400> 3 cttattggcg agcatctagc tgctgggatg agcacacttt ttc 43

<210> 4

<211> 43

<212> DNA

<213> Artificial

<400> 4 gaaaaagtgt gctcatccca gcagctagat gctcgccaat aag 43

<210> 5

<211> 45

<212> DNA

<213> Artificial

<400> 5 ctgctacaaa gaatctgtag acaacagagg aaaccagata atgtc 45 <210> 6

<211> 45

<212> DNA

<213> Artificial

<400> 6 gacattatct ggtttcctct gttgtctaca gattctttgt agcag 45

<210> 7

<211> 45

<212> DNA

<213> Artificial

<400> 7 gctacaaaga atctgtagac caaagaggac agcagataat gtcag 45

<210> 8 <211> 45

<212> DNA

<213> Artificial

<400> 8 ctgacattat ctgctgtcct ctttggtcta cagattcttt gtagc 45

<210> 9

<211> 45

<212> DNA

<213> Artificial

<400> 9 gctacaaaga atctgtagac caaaaaggaa accagataat gtcag 45

<210> 10

<211> 45 <212> DNA

<213> Artificial

<400> 10 ctgacattat ctggtttcct ttttggtcta cagattcttt gtagc 45

<210> 11

<211> 47

<212> DNA

<213> Artificial

<400> 11 tcaaagtaga tagctgtcca gaggaacccc aactaataat gaaaaat 47

<210> 12

<211> 47

<212> DNA <213> Artificial

<400> 12 atttttcatt attagttggg gttcctctgg acagctatct actttga 47

Table 1:

Table 2:

Table 3:

Table 4:

Table 5:

Claims

Claims:.

1. A modified blood coagulation Factor VIII with a reduced susceptibility to APC mediated cleavage as compared to wild type Factor VIII, characterized in that it

a) contains a sequence of amino acids which has been changed in the vicinity of the APC cleavage site at Arg336 (without mutating Arg226, Lys325, Arg336, Lys338 and Arg359) and/or Arg562 (without mutating Arg562) and/or within the APC binding sites at amino acids 2001 to 2022 as compared with the sequence of the amino acids of the wild type blood coagulation Factor VIII,

b) exhibits a specific activity of more than 10% of that of wild type Factor VIII after expression in stable cell lines and

c) can be activated by thrombin.

2. A Factor VIII as claimed in claim 1 , characterized in that in the sequence of the 9 amino acids aminoterminal and 10 amino acids carboxyterminal of Arg336 substitutions, deletions or insertions have been made leading to a reduced susceptibility to APC mediated cleavage

3. A Factor VIII as claimed in claims 1 or 2, characterized in that in the sequence of the 10 amino acids amino- and carboxyterminal of Arg562 substitutions, deletions or insertions have been made leading to a reduced susceptibility to APC mediated cleavage

4. A Factor VIII as claimed in claims 1 to 3, characterized in that within the APC binding site of the amino acids 2001 to 2022 substitutions, deletions or insertions have been made leading to a reduced susceptibility to APC mediated cleavage

5. A Factor VIII as claimed in claims 1 to 4, characterized by the amino acids sequence FVIII(Q561N) in which at position 561 Glutamine has been replaced by Asparagine.

6. A Factor VIII as claimed in claims 1 to 4, characterized by the amino acids sequence FVIII(N564Q) in which at position 564 Asparagine has been replaced by

Glutamine.

7. A Factor VIII as claimed in claims 1 to 4, characterized by the amino acids sequence FVIII(H2009A) in which at position 2009 Histidine has been replaced by Alanine.

8. A polynucleotide encoding a modified Factor VIII according to any one of claims 1 to 7.

9. A plasmid or vector comprising a nucleic acid according to claim 8.

10. A plasmid or vector according to claim 9 which is an expression vector.

11. A vector according to claim 10 which is a transfer vector for use in human gene therapy.

12. A host cell comprising a polynucleotide according to claim 8 or a plasmid or vector according to any one of claims 9 to 11.

13. A method for producing a modified Factor VIII comprising:

a) culturing host cells according to claim 12 under conditions such that the modified Factor VIII is expressed; and

b) recovering the modified Factor VIII from the host cells or from the culture medium.

14. A pharmaceutical composition comprising a modified Factor VIII according to any one of claims 1 to 7, a polynucleotide according to claim 8, or a plasmid or vector according to any one of claims 9 to 11.

15. A pharmaceutical composition as claimed in claim 14 comprising a modified Factor VIII molecule leading to APC resistance and additionally having a modification of said Factor VIII molecule leading to a decreased dissociation of the A2 domain after thrombin activation.

16. The use of a modified FVIII according to any one of claims 1 to 7, of a polynucleotide according to claim 8, or a plasmid or vector according to any one of claims 9 to 11 , or of a host cell according to claim 12 for the manufacture of a medicament for the treatment or prevention of a blood coagulation disorder.

17. The use according to claim 15, wherein the blood coagulation disorder is hemophilia A.

18. The use according to claim 15 or 16, wherein the treatment comprises human gene therapy.