EP4017521A1 - Factor viii protein with increased half-life - Google Patents

Abstract

The present invention relates to recombinant coagulation factors, in particular, recombinant Factor VIII (FVIII) proteins having an increased half-life. They comprise a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least one albumin binding domain is C-terminal to the heavy chain portion and at least one albumin binding domain is C-terminal to the light chain portion. If the protein is a single chain protein, the albumin binding domain(s) C-terminal to the heavy chain portion is/are N-terminal to the light chain portion. The protein of the invention may also be a de-immunized Factor VIII protein comprising specific point mutations at defined positions, which serve to reduce the immunogenicity, wherein the protein substantially retains its coagulant activity. The invention also relates to nucleic acids encoding the proteins of the invention, methods of producing them and pharmaceutical compositions comprising any of these, wherein the pharmaceutical composition preferably is for use in treatment of hemophilia A.

Description

Factor VIII protein with increased half-life

The present invention relates to recombinant coagulation factors, in particular, recombinant Factor VIII (FVIII) proteins having an increased half-life. They comprise a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least one albumin binding domain is C-terminal to the heavy chain portion and at least one albumin binding domain is C-terminal to the light chain portion. If the protein is a single chain protein, the albumin binding domain(s) C-terminal to the heavy chain portion is/are N- terminal to the light chain portion. The protein of the invention may also be a de-immunized recombinant Factor VIII protein comprising specific point mutations at defined positions, which serve to reduce the immunogenicity of said FVIII protein, wherein the Factor VIII protein substantially retains its coagulant activity. The invention also relates to nucleic acids encoding the proteins of the invention, methods of producing them and pharmaceutical compositions comprising any of these, wherein the pharmaceutical composition preferably is for use in treatment of hemophilia A.

FVIII is an important co-factor in the coagulation cascade. Wildtype human FVIII is synthesized as a single chain consisting of 2351 amino acids and comprises three A domains (A1-A3), one B domain and two C domains (C1 and C2), interrupted by short acidic sequences (a1-a3). The first 19 amino acids are the signal sequence, which is cleaved by intracellular proteases, leading to a FVIII molecule of 2332 amino acids. The resulting domain structure is A1-a1-A2-a2-B-a3-A3-C1-C2. During post-translational modification,

FVIII becomes glycosylated, sulfated and proteolytically processed. Sulfation is important for the extracellular interaction with different proteins, especially Thrombin and von Willebrand factor (vWF). It takes place on six tyrosines in the acidic regions a1 , a2 and a3. Intracellular cleavage, by the serine protease furin, divides FVIII into a heavy chain (A1-a1-A2-a2-B) and a light chain (a3-A3-C1-C2). During this cleavage, parts of the B domain can be lost. Therefore, the light chain has a molecular weight of 80 kDa, whereas the heavy chain can be slightly heterogeneous, with a molecular weight around 210 kDa. The binding between heavy and light chain is not covalent, but mediated by the divalent metal ion Cu²⁺ between the A1 and A3 domain.

In the circulation, FVIII is bound to vWF via the a3, C1 and C2 domain, which protects FVIII from early activation as well as degradation.

Upon activation, FVIII is cleaved by Thrombin at three positions, leading to a heterotrimer and loss of the B domain (heterotrimeric FVIIIa). The heterotrimer forms a complex with the activated coagulation Factor IXa and coagulation Factor X, and the light chain binds to a phospholipid bilayer, e.g., the cell membrane of (activated) platelets.

Hemophilia A mainly is a genetic bleeding disorder linked to the X-chromosome, occurring in 1 of 5000 newborn males. However, Hemophilia A can also occur spontaneously due to an auto-immune response against FVIII. Patients with Hemophilia A suffer from longer bleeding durations, spontaneous and internal bleedings, affecting their everyday life.

Hemophilia A patients are generally treated by administration of FVIII. Depending on the severity of the disease (mild, moderate or severe), treatment is on demand or prophylactic. Therapeutic FVIII products are either purified from human plasma (pFVIII) or the products are produced recombinantly in cell culture (rFVIII).

During the development of recombinant FVIII molecules for therapy, B-domain deleted FVIII molecules have been designed, because the B-domain is not important for the functionality of FVIII in clotting. This predominantly leads to a reduction in size. One of the most common B-domain deleted FVIII product is ReFacto® or ReFacto AF® produced by Pfizer. This FVIII variant lacks 894 amino acids of the B domain.

One issue with regard to FVIII substitution therapies is the relatively low in vivo half-life of the plasma-derived or recombinant Factor VIII proteins. There have been different approaches for enhancing FVIII half-life reaching from PEGylation to albumin incorporation, single chain molecules and Fc fusion.

WO 2014/070953 A1 relates to a method of reducing or decreasing the bleeding rate of a hemophilia patient by administering a long acting Factor VIII polypeptide, wherein the long acting Factor VIII polypeptide may be a fusion of the Factor VIII polypeptide and a heterologous moiety, which is an FcRn binding partner and may comprise an Fc region.

WO 91/01743 A1 describes a process for extending the half-life of a biologically active protein or peptide in vivo by covalently coupling said protein or peptide to a polypeptide fragment capable of binding to a serum protein, especially to serum albumin. Preferably, the albumin-binding fragment derives from streptococcal protein G or staphylococcal protein A. The generated fusion protein binds to serum albumin in vivo, and benefits from its longer half-life, thereby increasing the net half-life of the fused therapeutically interesting protein or peptide. WO 2005/097202 A2 generally mentions fusion proteins combining a therapeutic protein and a plasma protein in a single polypeptide chain, wherein such fusion proteins may provide clinical benefits in requiring less frequent injection and higher levels of therapeutic protein in vivo. WO 2009/016043 A2 discloses an albumin binding polypeptide. WO 2010/054699 A1 discloses a capture molecule for modulation of pharmacokinetics and/or pharmacodynamics of a target having a biological function in a mammal, wherein said molecule comprises at least one albumin binding moiety.

WO 2011/101284 A1 describes Factor VIII fusion proteins. Reference to albumin binding moieties is made. Fusion partners are presumed to delay in vivo clearance of FVIII by interaction with serum albumin. Albumin binding polypeptides, such as the ABD1 polypeptide, are mentioned as examples for fusion partners. Suggested fusion proteins comprise four albumin-binding moieties (albumin binding domain - ABD) comprising ABD1 , e.g. 4 x ABD1 in the B domain or 4 x ABD1 at the C-terminus of the FVIII light chain.

WO 2012/004384 A2 discloses an albumin binding sequence. It describes a fusion protein or conjugate with the albumin binding polypeptide, wherein the second moiety may be Factor VIII.

WO 2013/143890 A1 discloses a compound for oral administration comprising a moiety with desired therapeutic activity and a moiety that binds to albumin.

WO 2014/048977 A1 discloses a class of engineered polypeptides having a binding affinity for albumin. WO 2014/064237 A1 provides albumin binding domain binding polypeptides comprising an ABD binding motif. WO 2015/091957 A1 relates to a class of engineered polypeptides having a binding affinity for albumin, wherein the polypeptides have a high resistance to proteolytic cleavage.

WO 2015/023894 A1 provides recombinant FVIII proteins, in which one or more amino acids in at least one permissive loop or a3 domain are substituted or deleted, or replaced with heterologous moieties, while retaining the procoagulant FVIII activity. The generated FVIII proteins are supposed to have, e.g., increased in vivo stability.

Further, up to 30 % of patients with severe Hemophilia A develop inhibitory anti-FVIII antibodies against therapeutic FVIII. This is due to the fact that the immune system of these patients recognizes the applied therapeutic FVIII as foreign, because the patients produce an altered endogenous FVIII variant, which can be mutated or truncated, or no FVIII at all. It is known that the inhibitory antibodies against FVIII have undergone class switching and affinity maturation. This hints towards a T cell-dependent activation of the B cells, which secrete the antibodies. This T cell-dependent B cell activation requires activated T helper cells, which derive from naive T helper cells through interaction with antigen presenting cells (APCs), which present the FVIII antigen and additional co-stimuli. The fully human sequence of FVIII, which is administered as a therapeutic, could be considered a foreign protein by at least some hemophiliacs, because no central tolerance to the protein has developed. Depending on the HLA of the subject, the frequency of dosing and the location and nature of the mutations present in each subject’s FVIII, immune responses to FVIII may be induced by treatment with FVIII. Those antibodies against FVIII, which interfere with the function of FVIII, are designated inhibitory antibodies or inhibitors. In the past, the development of FVIII inhibitors in subjects receiving FVIII therapy has been correlated with more severe mutations or non-expression of FVIII. It is expected that the more “foreign” the replacement therapy the more robust the resulting immune response. Indeed, in hemophiliacs, an anti-therapeutic immune response may be the normal and expected result of interaction between therapeutic FVIII and a healthy functioning immune system.

In the case of inhibitor formation, the patients mostly undergo an immune-tolerance-induction (ITI) therapy. During this therapy, which can take weeks, months or years, very high doses of FVIII are applied to the patients, in order to exhaust the immune system and, accordingly, to induce tolerance. This therapy is very cost-intensive as well as strenuous for the patients and their caregivers. During ITI, FVIII application occurs daily, in some cases even twice a day. In addition to the strenuous therapy, the number of bleeds are increased when inhibitors are present. The aim to protect the patient from disabilities resulting from joint bleeds impairs the social life of the patient as well as of the whole family. Furthermore, in a significant proportion of patients, ITI is not successful.

Recombinant porcine FVIII was approved by the FDA for the treatment of hemophilia A patients who have developed an autoimmune response to human FVIII. WO 99/46274 A1 discloses hybrid FVIII having human and animal FVIII sequences or human FVIII and non- FVIII sequences, including a modified factor VIII in which the amino acid sequence is changed by a substitution at one or more of specific loci, wherein the modified factor VIII is not inhibited by inhibitory antibodies against the A2 or C2 domain epitopes.

WO 2016/123200 A1 also describes recombinant or chimeric FVIII proteins wherein one or more protein domains comprise amino acid sequences that are derived from ancestrally reconstructed amino acid sequences, wherein the resulting FVIII shows reduced binding of inhibitors, i.e., wherein B cell epitopes have been deleted.

As T cells are believed to be involved in the generation of high affinity antibodies to FVIII, the development of recombinant FVIII molecules that do not contain common T cell epitopes and thus do not induce an immune response in patients has been suggested (Scott, 2014, Haemophilia 20 (01):80-86, Tangri et al., 2005, J Immunol.174:3187-3196). Moise et al. (2012, Clin Immunol 142(3):329-331) published de-immunized FVIII peptides, wherein C2 domain T cell epitopes have been identified by in silico approaches, and modified. The modified peptides have been evaluated in an HLA binding assay and were used to immunize mice. Schubert et al. (2018, PLoS Comput Biol 14(3):e1005983) published a similar approach for population-specific design of de-immunized protein biotherapeutics, describing a computational approach for identifying mutations in the C2 domain of FVIII which lead to reduced immunogenicity whilst retaining pharmaceutical activity and protein function. For example, they teach de-immunization results for sequences with up to three simultaneous point mutations (e.g., V2333E, L2321F, Q2335H or V2313M/V2313T). As an experimental validation of the in silico calculations, they measure affinity of 15mer peptides to specific HLA alleles.

WO 2011/060371 A2 discloses a modified FVIII polypeptide comprising at least one amino acid modification within a specific region of the C2 domain of FVIII believed to form a B cell epitope for an inhibitor, and/or at least one amino acid modification within a specific region of the A2 domain of FVIII believed to form a relevant T cell epitope, for preventing or reducing an initial immune response to factor VIII in patients suffering from hemophilia A or for reducing the intensity of an immune response in patients having pre-formed inhibitor antibodies against factor VIII.

In light of the state of the art, the inventors of the present invention addressed the problem of providing recombinant FVIII proteins with increased half-life, which preferably allow a more convenient substitution therapy for the patient, allowing administration intervals of more than three days or even more than one week or more.

This problem is solved by the present invention, e.g., by the claimed subject-matter.

Description of the invention

The inventors have constructed new FVIII variants having a longer in vivo half-life and excellent specific activity as evidenced by different biological activity assays. These proteins further have a high level of expression and a low profile of fragments and side products. Further advantages and preferred embodiments are explained elsewhere in this description. The inventors have found that a particular arrangement of albumin binding domains contributes to an increase in in vivo half-life.

The invention thus provides a recombinant Factor VIII protein comprising a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least one albumin binding domain is C-terminal to the heavy chain portion and at least one albumin binding domain is C-terminal to the light chain portion. If the protein is a single chain protein, the albumin binding domain(s) C-terminal to the heavy chain portion is/are N-terminal to the light chain portion.

Factor VIII in complex with von Willebrand factor (vWF) has an in vivo half-life of about 12 hours. FVIII not associated with vWF is typically degraded much more rapidly.

Albumin has an in vivo half-life of about 19 days. By introducing at least two albumin binding domains in the FVIII sequence, it was possible to obtain a significant half-life prolongation. Different positions and different numbers of the albumin binding moiety have been tested in order to identify the optimal positions and numbers of integrated albumin binding moieties.

Without intending to be bound by the theory, it is believed that albumin binding to the FVIII protein of the invention through the albumin binding domains in the specific positions described herein is particularly effective in inhibiting breakdown of the FVIII protein of the invention. This appears to increase in vivo half-life more than association with vWF associated with native FVIII in blood.

FVIII

The skilled person understands the term FVIII (or Factor VIII) and is aware of the structure and biological functions of wild type FVIII and typical variants thereof. Apart from the features specified herein, the FVIII protein of the invention may be designed as deemed appropriate and advantageous by the skilled person.

In particular, the Factor VIII protein of the invention should typically comprise all necessary portions and domains known to be important for biological function. For example, preferably, the FVIII protein further comprises domains corresponding to, substantially corresponding to, and/or functionally corresponding to the A and C domains of wild type FVIII, especially to A1, A2, A3, C1 and C2 domains. It may further comprise additional portions and domains. For example, preferably, the FVIII protein further comprises an a1 domain between the A1 and the A2 domains and an a2 domain C-terminal to the A2 domain. For a double chain protein, on a separate chain, or for a single chain protein, C-terminal to said domain, and, optionally, C-terminal to the B-domain or a truncated B-domain and to at least one albumin binding domain, the FVIII protein comprises at least a truncated a3 domain. Before processing, the Factor VIII protein of the invention may also comprise a signal sequence. Thus, the heavy chain portion preferably comprises the domains A1 and A2, and typically comprises the domains A1-a1-A2-a2 or A1-a1-A2-a2-B. Preferably, the B-domain of the Factor VIII protein is at least partly deleted. The light chain portion preferably comprises the domains A3 and C1 and C2, and typically comprises the domains a3-A3-C1-C2. Any or all of said domains may be wildtype (wt) FVIII domains, or they may differ from the wildtype domains, e.g., as known in the state of the art or deemed appropriate by the skilled person.

The domains are preferably contained in the protein in that order, i.e. , from N-terminus to C- terminus of the protein.

While parts of the FVIII protein of the invention can be designed as desired by the skilled person, the FVIII preferably maintains a high FVIII biological activity. As shown in the examples, the invention allows generation of a FVIII protein with a high biological activity, as measured e.g. by the chromogenic activity. Therefore, preferably the FVIII protein according to the invention has a chromogenic activity which is at least comparable to the activity of the wt FVIII protein, i.e., it has at least 50% of the specific chromogenic activity of the wt protein (SEQ ID NO: 1). Preferably, the FVIII protein according to the invention has at least 80%, at least 100 % or more than 100% of the specific chromogenic activity of the wt protein. Preferably, the chromogenic activity also is at least 50%, at least 80%, at least 90%, at least 100% or more than 100% of the chromogenic activity of ReFacto AF^® (international non proprietary name: Moroctocog Alfa), a commercially available B-domain deleted FVIII (Pfizer).

A FVIII protein according to the present invention shall have at least one biological activity or function of a wt FVIII protein, in particular with regard to the function in coagulation. The FVIII protein should be cleavable by thrombin, leading to activation. Preferably, the FVIII protein according to the invention comprises at least one thrombin recognition and/or thrombin cleavage site, wherein said thrombin recognition and/or thrombin cleavage sites may correspond to or substantially correspond to those of wild type FVIII. It is then capable of forming a complex with the activated coagulation Factor IXa and coagulation Factor X, and the light chain is capable of binding to a phospholipid bilayer, e.g., the cell membrane of (activated) platelets.

The biological activity of FVIII can be determined by analyzing the chromogenic, the clotting or the coagulant activity of the protein, as described herein. Typically, the chromogenic activity is taken as a measure of biological activity.

An increased in vivo half-life may be achieved for Factor VIII proteins of the invention that are double chain proteins. Double chain proteins which may form a basis for the FVIII proteins of the invention are known in the art, e.g., wt FVIII or B-domain deleted or truncated versions thereof, e.g., ReFacto AF^®.

Moreover, the inventors have found an excellent increase in in vivo half-life with Factor VIII proteins of the invention that are a single chain proteins. Single chain Factor VIII proteins which may form a basis for the FVIII proteins of the invention are known in the art. In general, singe chain FVIII proteins do not comprise a functional furin cleavage site and thus, before activation, remain in the circulation as a single chain. Such proteins are also disclosed in EP application No. 19173440 or taught herein.

As a single chain backbone for the inventive protein, a single chain FVIII molecule has been developed, in which several amino acids including the furin cleavage site (positions R1664 - R1667, wherein the signal peptide is also counted) have been deleted. The B domain is deleted to a large extent, wherein an internal fragment (at least NPP) of the B-domain is maintained and an intact Thrombin cleavage site is preceding the internal fragment. This single chain Factor VIII protein (FVIII-sc) has been shown to be more stable than wt Factor VIII. Thus, in a preferred embodiment, the recombinant Factor VIII protein of the invention comprises, in a single chain, a heavy chain portion comprising an A1 and an A2 domain and a light chain portion comprising an A3, C1 and C2 domain of Factor VIII, wherein a) in said recombinant Factor VIII protein, 894 amino acids corresponding to consecutive amino acids between F761 and P1659 of wild type Factor VIII as defined in SEQ ID NO: 1 are deleted, leading to a first deletion; b) said recombinant Factor VIII protein comprises, spanning the site of the first deletion, a processing sequence comprising SEQ ID NO: 2 or a sequence having at most one amino acid substitution in SEQ ID NO: 2, wherein said processing sequence comprises a first thrombin cleavage site; c) in said recombinant Factor VIII protein, at least the amino acids corresponding to amino acids R1664 to R1667 of wild type Factor VIII are deleted, leading to a second deletion; and d) said recombinant Factor VIII protein comprises, C-terminal to the second deletion and N- terminal of the A3 domain, a second thrombin cleavage site.

As defined in a), in the FVIII of the invention, 894 amino acids corresponding to consecutive amino acids between F761 and P1659 of wild type Factor VIII as defined in SEQ ID NO: 1 are deleted in the Factor VIII protein of the invention, leading to a first deletion. In certain embodiments, in particular, starting from a numbering of amino acids in FVIII without deletions or insertions, the term "corresponding to" should be understood to mean "identical to".

For specific amino acids which may be mutated compared to the wt, an amino acid corresponding to the wild type aa is determined by an alignment e.g. using EMBOSS Needle (based on the Needleman-Wunsch algorithm; settings: MATRIX: “BLOSUM62”, GAP OPEN: “20”, GAP EXTEND:”0.5”, END GAP PENALTY: “false”, END GAP OPEN: “10”, END GAP EXTEND: “0.5”).

In order to assess sequence identities of two polypeptides, such an alignment may be performed in a two-step approach: I. A global protein alignment is performed using EMBOSS Needle (settings: MATRIX: “BLOSUM62”, GAP OPEN: “20”, GAP EXTEND:”0.5”, END GAP PENALTY: “false”, END GAP OPEN: “10”, END GAP EXTEND: “0.5”) to identify a particular region having the highest similarity. II. The exact sequence identity is defined by a second alignment using EMBOSS Needle (settings: MATRIX: “BLOSUM62”, GAP OPEN: “20”, GAP EXTEND:”0.5”, END GAP PENALTY: “false”, END GAP OPEN: “10”, END GAP EXTEND: “0.5”) comparing the fully overlapping polypeptide sequences identified in (I) while excluding non-paired amino acids.

"between" excludes the recited amino acids, e.g., it means that the recited amino acids are maintained "deletion" or "deleted" does not necessitate that the protein was actually prepared by deleting amino acids previously present in a predecessor molecule, but it merely defines that the amino acids are absent, independent from the preparation of the molecule. For example, the protein can be produced based on nucleic acids prepared by de novo synthesis or by genetic engineering techniques.

As defined in b), the recombinant Factor VIII protein comprises, spanning the site of the first deletion, a processing sequence comprising SEQ ID NO: 2 (PRSFSQNPP) or a sequence having at most one amino acid substitution in SEQ ID NO: 2, wherein said processing sequence comprises a first thrombin cleavage site. Accordingly, at least one amino acid of the processing sequence corresponds to an amino acid C-terminal to the deletion and at least one amino acid of the processing sequence corresponds to an amino acid N-terminal to the deletion. The processing sequence comprises SEQ ID NO: 2 or a sequence having at most one amino acid substitution in SEQ ID NO: 2, i.e., the processing sequence can be longer. In particular, the processing sequence is selected from the group comprising SEQ ID NO: 2, 4, 5, 6, 7 or 8. The inventors have found that a processing sequence of the invention enables a particularly good cleavage by thrombin.

In preferred embodiments, the processing sequence is no longer than SEQ ID NO: 4. The processing sequence may be directly C-terminal to sequences from the a2 domain, e.g., wt a2 domain sequences. The first N-terminal two amino acids of the processing sequence may already belong to the a2 domain. Preferably, the amino acid directly N-terminal to the processing sequence is E.

One amino acid in SEQ ID NO: 2 can be substituted, e.g., to reduce immunogenicity. Optionally, the F, the S C-terminal to the F, the Q or the N are substituted.

Preferably, the F is substituted, e.g., to an A or S, leading to a F761A or F761S substitution.

The processing sequence may be SEQ ID NO: 4 (PRSFSQNPPVL) or a sequence having at most one amino acid substitution in said sequence, wherein, optionally, the F, the S C- terminal to the F, the Q or the N are substituted.

Moreover, the present inventors have shown that an L at the C-terminus of the processing sequence, as in SEQ ID NO: 4, 5, 6, 7 or 8, endows the FVIII with particularly good activity. One especially preferred example of a single chain FVIII protein, which may form the backbone for the protein of the invention, is shown in the examples in further detail under the name VO (SEQ ID NO: 16). The processing sequence of the FVIII protein VO, which has been found to be particularly advantageous, consists of SEQ ID NO: 4, which is a specific embodiment of SEQ ID NO: 5-8.

The alternative processing sequences SEQ ID NO: 5 (PRSXSQNPPVL), SEQ ID NO: 6 (PRSFXQNPPVL), SEQ ID NO: 7 (PRSFSXNPPVL) and SEQ ID NO: 8 (PRSFSQXPPVL) are variants in which X can be any naturally occurring amino acid. Optionally, X is a conservative substitution compared to the corresponding amino acid in SEQ ID NO: 4, i.e. a hydrophobic amino acid is substituted by a hydrophobic amino acid, a hydrophilic amino acid is substituted by a hydrophilic amino acid, an aromatic amino acid by an aromatic amino acid, an acid amino acid by an acid amino acid and a basic amino acid by a basic amino acid.

For example, in SEQ ID NO: 5, it was predicted in silico that alternative processing sequences of SEQ ID NO: 5 wherein X is A or S lead to a less immunogenic product (Table 11). This applies, e.g., for variants F01_AD2CD2_SC, F02_AD2CD2_SC analysed herein. A further deimmunized variant of the processing sequence is SEQ ID NO: 132 (PRSFSQNPEVL). Alternatively, the S directly C-terminal to the processing sequence (i.e. , the first amino acid of the linker, e.g., the thrombin cleavable linker) can be substituted to a D, as in SEQ ID NO: 131.

As defined in c), in the FVIII protein of the invention, the amino acids corresponding to amino acids R1664 to R1667 of wild type Factor VIII are deleted, leading to a second deletion. These amino acid correspond to the furin cleavage recognition site of wt FVIII. Accordingly, the protein is essentially not cleaved by furin. In a composition, at least 80%, optionally, at least 90% or at least 95% of the FVIII protein of the invention are present in a single chain form.

As defined in d), the recombinant Factor VIII protein of the invention comprises, C-terminal to the second deletion and N-terminal of the A3 domain, a second thrombin cleavage site. Accordingly, upon activation, the part of the FVIII protein between the thrombin cleavage site in the processing sequence and the second thrombin cleavage site are excised from the activated FVIII protein.

Further, the invention provides a recombinant Factor VIII protein comprising, in a single chain, a heavy chain portion comprising an A1 and an A2 domain and a light chain portion comprising an A3, C1 and C2 domain of Factor VIII, wherein, a) said recombinant Factor VIII protein comprises a processing sequence comprising SEQ ID NO: 2 or a sequence having at most one amino acid substitution in SEQ ID NO: 2, wherein said processing sequence comprises a first thrombin cleavage site; b) directly C-terminal to said processing sequence, said Factor VIII protein comprises a heterologous sequence comprising at least one, preferably, two albumin binding domain(s); c) directly C-terminal to said heterologous sequence, said Factor VIII protein comprises a merging sequence having at least 90% sequence identity to SEQ ID NO: 9 (e.g., SEQ ID NO: 9); and d) said recombinant Factor VIII protein comprises, C-terminal to SEQ ID NO: 9, a second thrombin cleavage site; and e) said recombinant Factor VIII protein comprises, C-terminal to the light chain portion, at least one, preferably, two albumin binding domain(s).

Said recombinant FVIII protein may be a FVIII protein as described above. The FVIII protein typically comprises at least one further thrombin cleavage site. In one embodiment, a FVIII protein of the invention that optionally is a single chain protein comprises a heavy chain portion having at least 90% sequence identity to aa20-aa1667 of SEQ ID NO: 1, and a light chain portion having at least 90% sequence identity to aa1668- aa2351 of SEQ ID NO: 1. Optionally, the respective sequence identity aa20-aa1667 of SEQ ID NO: 1 and sequence identity to aa1668-aa2351 of SEQ ID NO: 1 are at least 95%. The respective sequence identity to aa20-aa1667 of SEQ ID NO: 1 and sequence identity to aa1668-aa2351of SEQ ID NO: 1 may be at least 98%. Optionally, the respective sequence identity to said sequences is at least 99%. The invention also provides a FVIII protein of the invention comprising a heavy chain portion having aa20-aa1667 of SEQ ID NO: 1 and a light chain portion having aa1668-aa2351 of SEQ ID NO: 1.

Several experiments performed by the inventors have been carried out with a single chain FVIII of the invention on the basis of the VO single chain construct (SEQ ID NO: 16) with at least one albumin binding domain introduced C-terminal to the heavy chain portion and C- terminal to the light chain portion, as described herein. Such proteins have shown advantageous characteristics with regard to expression, stability, in vivo half-life and purification. Accordingly, a preferred FVIII protein of the invention, which may be a single chain protein, comprises a heavy chain portion having at least 90% sequence identity to aa20-aa768 of SEQ ID NO: 16, and a light chain portion having at least 90% sequence identity to aa769-aa1445 of SEQ ID NO: 16. Optionally, the respective sequence identity to aa20-aa768 of SEQ ID NO: 16 and sequence identity to aa769-aa1445 of SEQ ID NO: 16 are at least 95%. The respective sequence identity to aa20-aa768 of SEQ ID NO: 16 and sequence identity to aa769-aa1445 of SEQ ID NO: 16 may be at least 98%. Optionally, the respective sequence identity to said sequences is at least 99%. The invention also provides a FVIII protein of the invention comprising a heavy chain portion having aa20-aa768 of SEQ ID NO: 16 and a light chain portion having aa769-aa1445 of SEQ ID NO: 16. wt FVIII typically is bound by vWF. vWF shields FVIII from proteolytic degradation and receptor-mediated clearance, e.g. via low-density lipoprotein (LDL) receptor-related protein (LRP1), LDL-receptor (LDLR) and heparan-sulfate proteoglycans (HSPG), within the liver (Lenting etal . , 2007. J Thromb Haematol 5:1353-60). However, it has been shown that the half-live of vWF is approx. 15 h, thereby limiting the FVIII:vWF complex half-life to the vWF- related clearance pathway. The inventors found that vWF binding potency of the FVIII protein of the invention may be diminished compared to wt FVIII or ReFacto AF^®, which may be explained by sterical hindrance due to albumin binding. For example, FVIII proteins of the invention may have 0%-90%, 10%-80%, 20-70%, 30-60%, or 40-50% of the binding potency of ReFacto AF^® to vWF, which can be determined by an assay as described below. Preferably, in the presence of human serum albumin in physiological concentrations, said binding potency is less than 50% of the binding potency of ReFacto AF^® to vWF. vWF binding is mediated in particular by amino acid positions Y1683 and Y1699. To avoid vWF binding, e.g. amino acids corresponding to Y1683 and/or Y1699 of wt FVIII of SEQ ID NO: 1 may be mutated. For example, the amino acid corresponding to Y1683 and/or Y1699 of wt FVIII of SEQ ID NO: 1 may be mutated to a C or F, e.g., Y1699C or Y1699F. In particular, a mutation of the amino acid corresponding to Y1699 to F and a mutation of the amino acid corresponding to Y1683 to F, both mutations together also designated "b mutation" have been confirmed to further decrease binding of vWF to FVIII proteins of the invention. Beside the “b mutation”, the inventors have additionally tested an “a mutation” comprising the amino acid substitutions Y737F, Y738F, and Y742F of wt FVIII of SEQ ID NO: 1 and a “c mutation” comprising the amino acid substitutions 12117S and R2169H of wt FVIII of SEQ ID NO: 1. Additionally, the inventors have tested combinations of “a mutation” and “b mutation” and further combinations of “a mutation” and “b mutation” and “c mutation”.

It was observed, that the “c mutation” negatively influenced the protein expression and functionality. The “b mutation” either alone or in combination with the “a mutation” did not influence the protein expression and functionality, but strongly decreased the binding to vWF. In comparison, the “a mutation” did not decrease the binding to vWF.

Thus, to further decrease vWF binding, the recombinant Factor VIII protein of the invention may have a suitable mutation as described herein, e.g., a "b mutation", i.e. , a mutation of the amino acid corresponding to Y1699 to F at position 1699 and a mutation of the amino acid corresponding to Y1683 to F at position 1683 in wt Factor VIII protein of SEQ ID NO: 1. For example, the FVIII protein of the invention may comprise a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least two albumin binding domains (e.g., two) are C-terminal to the heavy chain portion and at least two albumin binding domains (e.g., two) are C-terminal to the light chain portion, wherein the FVIII protein further comprises a b mutation. Such a FVIII protein may further comprise linkers, e.g., a thrombin cleavable linker optionally flanked by a glycine-serine linker, between the albumin binding domains and other parts of the protein and between the albumin binding domains. Alternatively, such a FVIII protein does not comprise linkers. In the context of the invention, "flanked" means that the relevant portions are in a close vicinity, preferably, with a distance of at most 10, 5 or 2 amino acids positions. Optionally, the relevant portions are immediately adjacent. Albumin binding domains

The recombinant Factor VIII protein of the invention comprises a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least one albumin binding domain is C-terminal to the heavy chain portion and at least one albumin binding domain is C-terminal to the light chain portion.

Different albumin-binding domains (ABDs) may be employed in the context of the invention. Historically, the ABD typically is a small, three-helical protein domain derived from one of various surface proteins expressed by gram-positive bacteria. For example, domains derived from streptococcal protein G and protein PAB from Finegoldia magna, which share a common origin and therefore represent an interesting evolutionary system, have been thoroughly studied structurally and functionally. Their albumin-binding sites have been mapped and these domains form the basis for a wide range of protein engineering approaches. By substitution-mutagenesis they have been engineered to achieve a broader specificity, an increased stability or an improved binding affinity, respectively.

For example, albumin binding domains disclosed by Nilvebrant et al. (2013, Comput Struct Biotechnol J. 6: e201303009), Johansson et al. (2001, JBC 277: 8114-8120), Jacobs et al. (2015, Protein Engineering, Design and Selection 28 (10); 385-393), WO 91/01743 A1, WO 2009/016043 A2, WO 2010/054699 A1, WO 2012/004384 A2, WO 2014/048977 A1 or WO 2015/091957 A1 may be used.

Preferably, the albumin binding domain comprises a sequence according to SEQ ID NO: 44: LAXsAKXeXyANXn) ELDX₁₄YGVSDFYKRLIX₂₆KAKTVEGVEALKX₃₉X₄₀ILX₄₃X₄₄LP wherein independently of each other

X₃ is selected from E, S, Q and C, preferably, E;

Cb is selected from E, S, C and V, preferably, E;

X₇ is selected from A, S and L, preferably, A;

X₁₀ is selected from A, S and R, preferably, A;

Xi₄ is selected from A, S, C and K, preferably, S;

X₂₆ is selected from D, E and N, preferably, D;

X39 is selected from D, E and L, preferably, D;

X40 is selected from A, E and H, preferably, A;

X43 is selected from A and K, preferably, A;

X44 is selected from A, S and E, preferably, A; L in position 45 is present or absent, preferably, present; and P in position 46 is present or absent, preferably, present.

Alternatively, the albumin binding domain may comprise an amino acid sequence which has at least 95 % identity to the sequence of SEQ ID NO: 44.

The inventors have achieved good results using an albumin binding domain designated ABD1 (SEQ ID NO: 45). It is preferred to use the sequence of ABD2 (SEQ ID NO: 46) that has been de-immunized for the human immune system, i.e., adapted to avoid immune responses in humans. If not otherwise mentioned, said albumin binding domain is used in the experiments shown herein. ABD2 may be encoded by a nucleic acid of SEQ ID NO: 57, which is codon-optimized for expression in human cells.

While it is possible to use different albumin binding domains at different locations in the FVIII protein, typically, all albumin binding domains in the FVIII protein will have the same sequence, preferably ABD2. Alternatively, it is possible to use different albumin binding domains at different locations within the FVIII protein in order to obtain multivalent albumin binding.

For example, in the FVIII protein of the invention, one albumin binding domain may be C- terminal to the heavy chain portion and one albumin binding domain C-terminal to the light chain portion. Alternatively, in one of the two selected positions, there may be one albumin binding domain, and two, three, four or more albumin binding domains in the other. For example, one albumin binding domain may be C-terminal to the heavy chain portion and two albumin binding domains C-terminal to the light chain portion, or one albumin binding domain may be C-terminal to the heavy chain portion and three albumin binding domains C-terminal to the light chain portion, or one albumin binding domain may be C-terminal to the heavy chain portion and four albumin binding domains C-terminal to the light chain portion.

In a FVIII protein of the invention, two albumin binding domains may be C-terminal to the heavy chain portion and one albumin binding domain C-terminal to the light chain portion, or three albumin binding domains may be C-terminal to the heavy chain portion and one albumin binding domain C-terminal to the light chain portion, or four albumin binding domains may be C-terminal to the heavy chain portion and one albumin binding domain C-terminal to the light chain portion.

Preferably, the number of albumin binding domains in each of the two positions is the same.

It has also been shown to be advantageous if the FVIII protein of the invention comprises at least four albumin binding domains. The inventors have found a still better increase in the half-life for Factor VIII protein of the invention comprising at least two albumin binding domains C-terminal to the heavy chain portion and at least two albumin binding domains C- terminal to the light chain portion, preferably, two albumin binding domains C-terminal to the heavy chain portion and two albumin binding domains C-terminal to the light chain portion.

The invention also provides Factor VIII proteins of the invention with two albumin binding domains C-terminal to the heavy chain portion and three albumin binding domains C-terminal to the light chain portion, or with two albumin binding domains C-terminal to the heavy chain portion and four albumin binding domains C-terminal to the light chain portion, or with three albumin binding domains C-terminal to the heavy chain portion and two albumin binding domains C-terminal to the light chain portion, or with four albumin binding domains C- terminal to the heavy chain portion and two albumin binding domains C-terminal to the light chain portion. Optionally, there is an even number of albumin-binding domains both C- terminal to the heavy chain and C-terminal to the light chain.

Linker

While the inventors have shown that linkers are not principally required for activity and stability of the FVIII proteins of the invention, to increase accessibility of all domains of the FVIII of the invention, in particular, access to albumin in blood, linkers were introduced into some FVIII proteins of the invention. The inventors have shown that the linkers, in particular, inclusion of at least glycine-serine linker sections, further improve expression and function. In particular, in addition to access to albumin, access for thrombin also appears to be improved. Accordingly, preferably, albumin binding domains may be separated from the heavy chain portion and/or the light chain portion and/or other albumin-binding domains by a linker, wherein, optionally, albumin-binding domains are separated from the heavy chain portion and the light chain portion and (if directly adjacent otherwise) other albumin-binding domains by a linker. It is also possible that albumin-binding domains are separated from the heavy chain portion and the light chain portion and (if directly adjacent otherwise) other albumin binding domains by a linker, except that there is no linker N-terminal to the light chain, because the a3 domain contains a thrombin cleavage site.

In one preferred embodiment, in the recombinant Factor VIII protein of the invention, the linker comprises a thrombin-cleavable linker section. Optionally, a thrombin cleavable linker has the sequence of SEQ ID NO: 39 (abbr. L). Further thrombin cleavable sites are known in the art, e.g., disclosed in Gallwitz et al. (2012, PLoS ONE 7 (2): e31756). Thrombin cleavable linkers may thus also comprise any of these cleavable sites. Thrombin-cleavable linkers have the advantage that in generation of the active protein, i.e., after activation by thrombin, the linkers may be cleaved and, consequently, the albumin binding domains may be removed from the active protein.

Alternatively, in the recombinant Factor VIII protein of the invention, uncleavable glycine- serine linker sections may be used to introduce flexible, sterical distance between motifs to avoid structural influences. Thus, optionally, the linker comprises a glycine-serine linker section that optionally has the sequence of SEQ ID NO: 40 (abbr. G1, preferred) or SEQ ID NO: 41 (abbr. G2). The linker G1 may, e.g., be encoded by SEQ ID NO: 58. The linker G2 may, e.g., be encoded by SEQ ID NO: 59.

In preferred embodiments, different linker sections are combined. E.g. uncleavable linker sections were used to flank a central thrombin cleavable linker section to maintain thrombin accessibility of the thrombin cleavable linker section. Thus, in some embodiments, the linker comprises a thrombin cleavable linker section flanked on each side by a glycine-serine linker section, wherein said combined linker optionally has the sequence of SEQ ID NO: 42 or SEQ ID NO: 43, preferably, SEQ ID NO: 42.

The polynucleotide sequence for all linkers preferably is codon-optimized for human use. Exemplary codon-optimized sequences are provided herein in the context of FVIII proteins of the invention.

Specific FVIII proteins of the invention

All FVIII proteins of the invention demonstrated good in vitro functionality, wherein the FVIII proteins showed reduced vWF binding in correlation to increasing numbers of albumin binding domains. vWF has a major impact on the half-life of FVIII. It was found that shielding FVIII from vWF by albumin positively influences the half-life of the FVIII protein. A broad distribution of albumin binding domains with one position between heavy chain and light chain and one position at the C-terminus of the protein was shown to enhance the shielding of FVIII from vWF.

Preferably, the recombinant Factor VIII protein the invention comprises one albumin binding domain between the heavy chain portion and the light chain portion and one albumin binding domain C-terminal to the light chain portion, wherein the sequence has at least 70%, optionally, at least 80%, at least 90%, at least 95%, at least 99% or 100% sequence identity sequence identity to SEQ ID NO: 47. Preferably, said protein is a single chain protein. The single chain protein having SEQ ID NO: 47 is designated ADLCLD_SC. It was shown to have an in vivo half-life increased by a factor of about 1.5 compared to ReFacto AF^®. In an especially preferred embodiment, the recombinant Factor VIII protein of the invention comprises at least two albumin binding domains between the heavy chain portion and the light chain portion and at least two albumin binding domain C-terminal to the light chain portion, wherein the protein has at least 80% sequence identity, optionally, at least 90%, at least 95% or at least 99% sequence identity to any of SEQ ID NO: 48, 49, 51. Preferably, the recombinant Factor VIII protein has at least 80% sequence identity, optionally, at least 90%, at least 95%, or at least 99% sequence identity to SEQ ID NO: 48. Preferably, said protein is a single chain protein.

The recombinant Factor VIII protein may also have at least 80% sequence identity, optionally, at least 90%, at least 95%, or at least 99% sequence identity to SEQ ID NO: 49. Preferably, said protein is a single chain protein.

The recombinant Factor VIII protein may also have at least 80% sequence identity, optionally, at least 90%, at least 95%, or at least 99% sequence identity to SEQ ID NO: 51. Preferably, said protein is a single chain protein.

For example, the invention provides a recombinant FVIII protein having SEQ ID NO: 48 (AD2CD2_SC), SEQ ID NO: 49 (AD2CD2woL_SC), or SEQ ID NO: 51 (AbD2CD2_SC).

Such FVIII proteins have been shown to have a particularly extended in vivo half-life, e.g., for AD2CD2_SC, an in vivo half-life extended by a factor of 2.5 has been found in hemophilia A mice and a half-life extension of factor 4 has been found in albumin-deficient, transgenic neonatal Fc-receptor mice (see Examples). For AbD2CD2_SC, an in vivo half-life extended by a factor of 2.2 has been found. An increase in half-life can be analyzed on the level of FVIII antigen or on the level of activity, e.g., chromogenic activity, or both. Preferably, it is analyzed on the level of chromogenic activity.

The invention thus provides FVIII proteins of the invention, wherein the in vivo half-life of the recombinant Factor VIII protein is prolonged (i.e. increased) by a factor of at least 1.2, preferably, by a factor of at least 1.5, optionally, at least 2 or at least 2.5 in comparison to a recombinant Factor VIII protein of SEQ ID NO: 28 (ReFacto AF^®). While the increase in in vivo half-life may be analyzed in model systems, e.g., mice, rats or dogs, such as in hemophilia A mice or albumin-deficient Tg32 mice having a knock-out of murine albumin and expressing human FcRn a-chain instead of the murine one (B6. Cg-Alb^em12 Mvw Fcgri^tm1Dcr Tg(FCGRT)32Dcr/MvwJ), the observed increase in in vivo half-life may be underestimated, because human albumin has a longer half-life than e.g. murine albumin, and it is expected that an increase seen in a murine model will be still more pronounced in humans. A fusion partner may be employed to extend the in vivo plasma half-life of the FVIII protein of the invention. In one embodiment, the recombinant Factor VIII protein of the invention is a fusion protein with at least one further heterologous fusion partner in addition to the albumin binding moiety, preferably with a further fusion partner extending the in vivo plasma half-life of the FVIII protein. The fusion partner may e.g. be selected from the group comprising an Fc region, albumin, PAS polypeptides, HAP polypeptides, the C-terminal peptide of the beta subunit of chorionic gonadotropin, , and combinations thereof. The FVIII protein may alternatively or additionally be covalently linked to non-protein fusion partners such as albumin-binding small molecules (e.g., dabigatran), PEG (polyethylenglycol) and/or HES (hydroxyethyl starch). PAS polypeptides or PAS sequences are polypeptides comprising an amino acid sequence comprising mainly alanine and serine residues or comprising mainly alanine, proline and serine residues, the PAS sequences forming a random coil conformation under physiological conditions, as defined in WO 2015/023894. HAP polypeptides or sequences are homo-amino acid polymer (HAP), comprising e.g., repetitive sequences of glycine or glycine and serine, as defined in WO 2015/023894. Potential fusions, fusion partners and combinations thereof are described in more detail e.g., in WO 2015/023894.

Optionally, for certain therapeutic applications, the recombinant FVIII protein may be fused to an Fc region. A fusion to an Fc region may be used to extend the half-life and reduce immunogenicity.

Optionally, said heterologous fusion partner may be inserted directly N-terminal or directly C- terminal to one of the albumin binding domains, e.g., C-terminal to the heavy chain, and/or C-terminal to the C2-domain, or C-terminal to the albumin binding domain(s) C-terminal to the heavy chain or C-terminal to the albumin binding domain(s) C-terminal to the heavy chain. These locations have been found by the inventors to be advantageous for fusion, while maintaining good biological activity of the FVIII protein. Optionally, the fusion protein further comprises at least one linker.

The protein may further be glycosylated and/or sulfated. Preferably, post-translational modifications such as glycosylation and/or sulfation of the protein occur in a human cell. A particularly suitable profile of post-translational modifications can be achieved using human cell lines for production, e.g. CAP cells, in particular CAP-T cells or CAP-Go cells (WO 2001/36615; WO 2007/056994; WO 2010/094280; WO 2016/110302). CAP cells, available from Cevec Pharmaceuticals GmbH (Cologne, Germany), originate from human amniocytes as they were isolated trans-abdominally during routine amniocentesis. Obtained amniocytes were transformed with adenoviral functions (E1A, E1B, and pIX functions) and subsequently adapted to growth in suspension in serum-free medium. During post-translational modification, the FVIII protein of the invention may be sulfated, e.g., on one, two, three, four, five or six tyrosines in the acidic regions a1, a2 and a3.

De-immunized Factor VIII proteins

Optionally, the recombinant Factor VIII protein of the invention is a de-immunized protein, i.e., the protein has a reduced immunogenicity as compared to wt FVIII in hemophilia patients. For example, certain mutations, preferably, substitutions, have been introduced to avoid the presence of epitopes which can be presented on human HLA molecules, preferably, common human HLA molecules.

The present invention provides a recombinant Factor VIII protein of the invention comprising at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79, I80, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, I632, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, S2125, F2215, K2226, K2258, V2313, S2315, V2333 and Q2335; wherein substitutions of N are independently selected from the group consisting of D, H, S and E; wherein substitution of I are independently selected from the group consisting of T and V; wherein substitutions of S are independently selected from the group consisting of A, N, G, T and E; wherein substitutions of L are independently selected from the group consisting of N, Q, F and S; wherein substitutions of V are independently selected from the group consisting of A and T; wherein substitutions of Y are independently selected from the group consisting of N, H and S; wherein substitutions of F are independently selected from the group consisting of H and S; wherein substitutions of K are independently selected from the group consisting of N, D, E, Q, S and T; wherein substitutions of R are independently selected from the group consisting of Q, H and S; wherein substitutions of M are selected from the group consisting of R, Q, K and T ; and/or wherein substitutions of Q are selected from the group consisting of R, D, E, H and K; wherein the positions are specified in relation to full length human Factor VIII molecule of SEQ ID NO: 1 including numbering of the signal sequence; and wherein the recombinant Factor VIII protein retains at least 50 % coagulant activity, as determined in a chromogenic assay, compared to a Factor VIII protein consisting of SEQ ID NO: 60 (FVI I l-6rs). The invention also provides a fusion protein of said recombinant Factor VIII protein. In one aspect, the invention also provides a recombinant Factor VIII protein of the invention, the recombinant Factor VIII protein comprising at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79, I80, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, F2215, K2226, K2258, V2313, S2315, V2333 and Q2335; wherein substitutions of N are independently selected from the group consisting of D, H, S and E; wherein substitution of I are independently selected from the group consisting of T and V; wherein substitutions of S are independently selected from the group consisting of A, N, G, T and E; wherein substitutions of L are independently selected from the group consisting of N, Q, F and S; wherein substitutions of V are independently selected from the group consisting of A and T; wherein substitutions of Y are independently selected from the group consisting of N, H and S; wherein substitutions of F are independently selected from the group consisting of H and S; wherein substitutions of K are independently selected from the group consisting of N, D, E, Q, S and T; wherein substitutions of R are independently selected from the group consisting of Q, H and S; wherein substitutions of M are selected from the group consisting of R, Q, K and T ; and/or wherein substitutions of Q are selected from the group consisting of R, D, E, H and K; wherein the positions are specified in relation to full length human Factor VIII molecule of SEQ ID NO: 1; and wherein the recombinant Factor VIII protein retains at least 50% coagulant activity, as determined in a chromogenic assay, compared to a Factor VIII protein consisting of SEQ ID NO: 60. The invention also provides a fusion protein of said recombinant Factor VIII protein of the invention.

The present invention further provides a recombinant Factor VIII protein of the invention comprising at least one amino acid substitution at a position selected from the group consisting of Y748, L171, S507, N79, I80, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, I632, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, S2125, F2215, K2226, K2258, V2313, S2315, V2333 and Q2335 (or, preferably, Y748, L171, S507, N79, I80, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, F2215, K2226, K2258, V2313, S2315, V2333 and Q2335); wherein substitutions of N are independently selected from the group consisting of D, H, S and E; wherein substitution of I are independently selected from the group consisting of T and V; wherein substitutions of S are independently selected from the group consisting of A, N, G, T and E; wherein substitutions of L are independently selected from the group consisting of N, Q, F and S; wherein substitutions of V are independently selected from the group consisting of A and T; wherein substitutions of Y are independently selected from the group consisting of N, H and S; wherein substitutions of F are independently selected from the group consisting of H and S; wherein substitutions of K are independently selected from the group consisting of N, D, E, Q, S and T; wherein substitutions of R are independently selected from the group consisting of Q, H and S; wherein substitutions of M are selected from the group consisting of R, Q, K and T ; and/or wherein substitutions of Q are selected from the group consisting of R, D, E, H and K; wherein, if the substitution is at position S507, it is S507E, and if the substitution is at position N616, it is N616E, and if the substitution is at position F2215, it is F2215H; wherein the positions are specified in relation to full length human Factor VIII molecule of SEQ ID NO: 1 including numbering of the signal sequence, and wherein the recombinant Factor VIII protein retains at least 50% coagulant activity, as determined in a chromogenic assay, compared to a Factor VIII protein consisting of SEQ ID NO: 60 (FVI I l-6rs), or a fusion protein of said recombinant Factor VIII protein. Said protein preferably also is a protein comprising at least three of the substitutions defined above.

Preferably, if the substitution is at position K2226, it is K2226Q, and if the substitution is at position Q2335, it is Q2335H. In one embodiment, there is no substitution of Q2335.

The inventors have found that a recombinant Factor VIII protein of the invention having the substitutions as defined herein has a significantly reduced immunogenicity while substantially maintaining coagulant activity. Accordingly, it is useful for treatment of hemophilia A, in particular, to avoid generation and/or further production of anti-FVIII antibodies including FVI 11 inhibitory antibodies.

FVI 11 proteins of the invention have been de-immunized on the level of T cell epitopes. Generally, antigens are presented to T cells as peptides bound to the MHC class II on the surface of APCs. As T cell epitopes relevant for the majority of the human population have been identified and eliminated in the protein of the invention, fewer immunogenic peptides will be presented by antigen-presenting cells (APCs), e.g., dendritic cells (DC) or B cells, to the T cells. This in turn prevents or reduces the activation of naive T cells. Without activated T helper cells, naive B cells are not activated and cannot differentiate into anti-FVI II antibody- secreting plasma and memory B cells.

By the approach of the present invention, the antibody formation is thus reduced or, optimally, prevented at the very beginning of the process, namely by reducing the stimulation of naive T helper cells in response to FVIII antigens. In addition to the reduction of naive T helper cell maturation, restimulation of memory T helper cells against FVIII, which are potentially already present, may also be prevented or reduced due to reduced presentation of the antigen according to the inventive approach.

The positions in the MHC class II binding groove required for peptide binding as well as the amino acids of the peptide important for the binding are known. As a first step, in silico analysis methods have been used to predict which FVIII peptides are most likely bound in common MHC class II complexes, and which of these only occur in FVIII, and not in other human proteins. These peptides are considered as immunogenic. Using further in silico tools and comparisons with both FVIII from other species and non-related human proteins, recommendations for amino acid mutations have been made to prevent FVIII peptide binding to MHC class II complex. As described in PCT/EP2019/059233 filed on 11 April 2019, based on these predictions and experimental tests, mutated FVIII variants have been generated, which are still functional in coagulation, but are considered to no longer elicit the generation of inhibitory antibodies in hemophilia A patients to the same extent. For the present invention, de-immunized FVIII proteins comprising a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least one albumin binding domain is C-terminal to the heavy chain portion and at least one albumin binding domain is C-terminal to the light chain portion, wherein, if the protein is a single chain protein, the albumin binding domain(s) C-terminal to the heavy chain portion is/are N-terminal to the light chain portion, have been generated and tested.

The de-immunized FVIII proteins of the invention have at least 50 % coagulant activity, as determined in a chromogenic assay, compared to a Factor VIII protein consisting of SEQ ID NO: 60 (FVI I l-6rs). FVIII-6rs is a B-domain deleted FVIII protein containing no further mutations, which has substantially the same coagulant activity as wildtype human FVIII. Preferably, the FVIII proteins of the invention have at least 70%, at least 80%, at least 90%, or at least 100% coagulant activity compared to a Factor VIII protein consisting of SEQ ID NO: 60. The coagulant activity may also be higher, e.g., at least 110%, at least 120%, at least 130%, at least 140%, at least 150%, at least 190%, at least 200% or at least 400% of coagulant activity compared to a Factor VIII protein consisting of SEQ ID NO: 60.

Throughout the invention, if not specified otherwise, coagulant activity is determined in a chromogenic assay. The chromogenic assay is carried out according to standard procedures, e.g., as described in detail in the examples below. This assay is preferably carried out with the supernatant of human cells, e.g., HEK293-F cells, transfected with an expression vector, e.g., as described in the examples, and expressing the FVIII variant of interest, in comparison to supernatant of the same cells transfected with the same basic expression vector expressing FVI I l-6rs under the same conditions. Accordingly, relative coagulant activities can be analyzed, wherein the chromogenic coagulant activities of the mutants are standardized to the chromogenic coagulant activities of the molecule without mutations, namely FVIII-6rs. This assay tests both the capability of the mutant protein to be synthesized and secreted by the cells and the coagulant activity of the secreted protein.

In addition, the FVIII of the invention preferably further has a high specific coagulant activity. The specific coagulant activity describes the ratio of FVIII chromogenic coagulant activity as defined above to FVIII antigen concentration, as determined by an FVI I l-specific ELISA (e.g., as described herein). The specific coagulant activity of a FVIII protein of the invention may be, e.g., at least 50%, at least 70%, at least 80%, at least 90%, at least 100%, at least 110%, at least 120%, at least 130%, at least 150%, at least 170% or at least 190%.

Proteins having a low relative coagulant activity in the supernatant, but a high specific coagulant activity can be assumed to have problems with synthesis, folding and/or secretion. This can potentially be improved by expression in specific cells lines, e.g., with overexpression of chaperones.

Factor VIII proteins of the invention may have both a coagulant activity and a specific coagulant activity (both determined by the chromogenic method) of at least 50% compared to FVI I l-6rs, preferably, of at least at least 70%, at least 80%, at least 90%, at least 100%, at least 110%, at least 120%, or at least 130%, respectively.

Coagulant activity can alternatively or additionally be assessed by the one stage clotting method, as also described in the experimental part herein. In a particularly preferred embodiment, both the coagulant activity as determined by the chromogenic method and as determined by the clotting method are at least 50% compared to the coagulant activity of FVI I l-6rs, preferably, at least 60%, at least 70%, at least 80%, at least 90%, at least 100%, or even at least 150%. The inventors found that specific substitutions tested were particularly advantageous with regard both to a reduced immunogenicity and maintenance of functional activity in coagulation. Accordingly, throughout the invention, the amino acid substitutions in the recombinant Factor VIII protein of the invention are preferably selected from the group consisting of Y748S, L171Q, S507E, N79S, I80T, 1105V, S112T, L160S, V184A, N233D, L235F, V257A, I265T, N299D, Y426H, Y430H, L505N, F555H, I610T, N616E, I632T, L706N, N754D, K1837E, R1936Q, S2030A, S2037G, N2038D, S2077G, M2123K, S2125G, F2215H, K2226Q, K2258Q, V2313A, S2315T, V2333A and Q2335H.

The Factor VIII protein of the invention may e.g. comprise 3-38, 3-25, 4-25, 5-24, 6-23, 7-22, 8-21, 9-20, 10-19, 11-18, 12-17, 13-16 or 14-15 of said substitutions. Preferably, the recombinant Factor VIII protein of the invention comprises 3-25 of said substitutions, and the substitutions are located within different immunogenic clusters. An immunogenic cluster is a peptide identified in a protein, which binds to a plurality of HLA-DR supertypes with a high affinity. In other words, immunogenic clusters are clusters of T-cell epitopes for different HLA supertypes identified in a protein, e.g., as described in more detail in the Examples below. Immunogenic clusters of FVIII are defined in SEQ ID NO: 74-108 and 112 (Table 6). Optionally, there is only one of the recited substitutions per immunogenic cluster. Most preferably, the recombinant Factor VIII protein of the invention comprises 15-19 of said substitutions.

The recombinant Factor VIII protein of the invention comprising at least three substitutions located within different immunogenic clusters preferably comprises at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79,

S112, L160, V184, N233, I265, N299, Y426, F555, N616, I632, L706, K1837, R1936, N2038, S2077, S2125, F2215, K2226, K2258, S2315, and V2333. The at least three amino acid substitutions are preferably selected from the group consisting of Y748S, L171Q, S507E, N79S, S112T, L160S, V184A, N233D, I265T, N299D, Y426H, F555H, N616E, I632T, L706N, K1837E, R1936Q, N2038D, S2077G, S2125G, F2215H, K2226Q, K2258Q, S2315T and V2333A.

Preferred FVIII proteins incorporate substitutions at four positions in the A1 region and/or 7 positions in A1 and/or 3 positions in A1A2 and/or 5 positions in A2 and/or 6 positions in A3C1C2, which have more than 100 % specific coagulant activity, e.g., according to the following list:

A1: N79, S112, N233, I265; especially N79S, S112T, N233D, I265T A1: N79, S112, L160, L171, V184, N233, I265; especially N79S, S112T, L160S, L171Q, V184A, N233D, I265T A1A2: N299, Y426, S507; especially N299D, Y426H, S507E

A2: F555, N616, I632, L706, Y748; especially F555H, N616E, I632T, L706N,

Y748S

A3C1C2: N2038, S2077, S2125, K2258, S2315, V2333; especially N2038D, S2077G, S2125G, K2258Q, S2315T, V2333A.

Further preferred FVIII proteins incorporate substitutions at 4 positions in A2 and/or 3 positions in A3C1C2 and/or 4 positions in A3C1C2 and/or 4 positions in A3C1C2 and/or 5 positions in A3C1C2:

A2: F555, N616, L706, Y748; especially F555H, N616E, L706N, Y748S

A3C1C2: S2077, S2315, V2333; especially S2077G, S2315T, V2333A

A3C1C2: N2038, S2077, S2315, V2333; especially N2038D, S2077G, S2315T, V2333A

A3C1C2: S2077, K2258, S2315, V2333; especially S2077G, K2258Q, S2315T, V2333A

A3C1C2: N2038, S2077, K2258, S2315, V2333; especially N2038D, S2077G, K2258Q, S2315T, V2333A.

Preferred combinations with especially good results in coagulant activity (chromogenic coagulant activity, clotting coagulant activity and specific coagulant activity) incorporate substitutions at the following positions:

FVIII-GOF1: L171; S507; Y748; V2333; especially L171Q; S507E; Y748S; V2333A FVIII-GOF2: L171; N299; N616; V2333; especially L171Q; N299D; N616E; V2333A.

Further preferred combinations with especially good results in reducing cluster score, which accordingly are calculated to strongly reduce immunogenicity, are:

FVIII-LS1: S112; S507; Y748; K1837; N2038; especially S112T; S507E; Y748S; K1837E;

N2038D

FVIII-LS2: S112; Y426; N754; K1837; N2038; especially S112T; Y426H; N754D;

K1837E; N2038D

Preferred recombinant Factor VIII proteins of the invention comprise amino acid substitutions at least at positions a. N79S, S112T, N233D, and I265T; and/or b. N79S, S112T, L160S, L171Q, V184A, N233D, and I265T; and/or c. N299D, Y426H, and S507E; and/or d. F555H, N616E, L706N, Y748S; and/or e. F555H, N616E, I632T, L706N, and Y748S; and/or f. S2077G, S2315T, and V2333A; and/or g. N2038D, S2077G, S2315T, and V2333A; and/or h. S2077G, K2258Q, S2315T, and V2333A; and/or i. N2038D, S2077G, K2258Q, S2315T, and V2333A; and/or j. N2038D, S2077G, S2125G, K2258Q, S2315T, and V2333A; and/or k. L171Q, S507E, Y748S and V2333A; and/or

L. L171Q, N299D, N616E and V2333A; and/or m. S112T, S507E, Y748S, K1837E and N2038D; and/or n. S112T, Y426H, N754D, K1837E and N2038D.

Preferred proteins of the invention have the substitutions listed in Fig. 14 herein.

Especially preferred proteins combine at least the substitutions specified under b and c. As shown in Fig. 14, these substitutions in combination lead to high chromogenic and clotting coagulant activities as well as high specific coagulant activities. Further especially preferred proteins of the invention combine at least the substitutions specified under b and c and those specified under d or e and/or f, g, h, i or j and/or K1837E. A protein of the invention may, e.g., include the substitutions specified under b, c and d or e. Other advantageous proteins of the invention comprise the substitutions specified under b, c and d and f, g, h, i or j. Other advantageous proteins of the invention comprise the substitutions specified under b, c and e and f, g, h, i or j.

Optionally, the proteins further comprise K1837E. This substitution has a high effect on the immunogenic score, but appears to have a negative effect on coagulant activity of the protein. Accordingly, it is also envisaged that proteins of the invention do not comprise a substitution at K1837, or do not comprise K1837E.

Optionally, proteins of the invention, e.g., comprising substitutions Y748S, L171Q, S507E, N79S, S112T, L160S, V184A, N233D, I265T, N299D, Y426H, F555H, I632T, L706N,

K1837E, R1936Q, N2038D, S2077G, S2125G, F2215H, K2226Q, K2258Q, S2315T and V2333A do not comprise a substitution at N616 such as N616E. On the other hand, inclusion of this substitution further reduces the immunogenic score, and immunogenicity of the protein, so, in general, inclusion of the substitution is preferred. In one combination, proteins of the invention comprise the substitutions Y748S, L171Q,

N79S, S112T, L160S, V184A, I265T, N299D, Y426H, F555H, I632T, L706N, R1936Q, N2038D, S2077G, S2125G, F2215H, K2226Q, K2258Q, and, optionally, S507E.

The inventors could particularly show advantageous combinations of substitutions of recombinant Factor VIII proteins comprising at least amino acid substitutions at positions N79, S112, L160, L171, V184, N233, I265, N299, Y426, S507, F555, N616, L706, and Y748, wherein preferably the substitutions are N79S, S112T, L160S, L171Q, V184A, N233D,

I265T, N299D, Y426H, S507E, F555H, N616E, L706N, and Y748S (e.g., FVIII-14M). Optionally, the protein further includes a substitution at K1837 such as K1837E.

Preferably, all amino acids selected for substitution in the specified positions reduce the cluster score of the relevant immunogenic cluster.

The de-immunized recombinant Factor VIII proteins of the invention have a reduced immunogenicity compared to a Factor VIII protein consisting of SEQ ID NO: 60 (FVIII-6rs). Immunogenicity may be determined by an immunogenicity score, which may be calculated as described herein. The immunogenicity score of FVIII-6rs is 7.01 , and the immunogenicity score of ReFacto AF is 10.03. Preferably, Factor VIII proteins of the invention have an immunogenicity score, which is reduced by at least 3, by at least 5, by at least 7, by at least 10, by at least 12, by at least 13 or by at least 15 compared to the Factor VIII protein without the recited substitutions, e.g., compared to FVIII-6rs. For example FVIII-19M has an immuno genicity score of -10.55, i.e. , the immunogenicity score is reduced by 17.56 compared to FVI I l-6rs. As further example, SEQ ID NO: 16 (AC_SC) has an immunogenicity score of 11.18, which is reduced by 17.58 when said 19 de-immunizing amino acid substitutions are incorporated (SEQ ID: 113 (AC-19M_SC). Additional incorporation of four albumin-binding domains with linkers in SEQ ID NO: 114 (AD2CD2-19M_SC) result in an even lower immunogenicity score of -14.93.

Preferably, immunogenicity may be determined by an assay comprising co-cultivating dendritic cells incubated with said protein and regulatory T-cell-depleted CD4⁺ T cells of a donor and testing activation of said T cells. Such an assay, provided by the inventors, is described in further detail below. The T cells may be derived from a healthy donor or from a patient, e.g. from a Hemophilia A patient.

In all de-immunized recombinant FVI 11 proteins of the invention, the positions are specified in relation to full length human Factor VIII molecule of SEQ ID NO: 1. In the state of the art, annotation of amino acids in the FVIII molecule differs between authors. This is mainly due to the 19 amino acid signal sequence, which can be included into the amino acid count or can be omitted. This variation of plus or minus 19 amino acids is in general the only difference in numeration for full-length FVIII sequences. For B-domain deleted FVIII sequences, the deletion may also lead to a shift in numeration. For the heavy chain the numeration correlates with the numeration of the full-length FVIII. From the B-domain deletion on the numeration of the light chain is either kept the same as for the full-length FVIII molecule (e.g. Q763 in front of the deletion is followed by D1582 after the deletion) or can be continued as if no deletion has occurred (e.g. Q763 is followed by D764 despite missing amino acids). The continued numeration complicates the comparison of amino acid sequences if it is not known how many amino acids were deleted. The continued numeration is rare and most authors keep the numeration of the full-length FVIII molecule despite B-domain deletion. In accordance with this, in the present invention, the positions of substitutions in the recombinant FVIII protein are specified in relation to full length human FVIII molecule of SEQ ID NO: 1. Nevertheless, the secreted recombinant FVIII protein does not comprise the signal sequence, comprises the albumin-binding domains as specified herein, and typically is a B- domain deleted variant.

It is known in the art that the B-domain is not required for proper coagulant function of FVIII, and therefore, various B-domain deleted FVIII proteins are well known. In the context of the present invention, a B-domain deleted FVIII protein may comprise full or partial deletion(s) of the B-domain. The B-domain deleted FVIII protein may still contain amino-terminal sequences of the B-domain which may e.g. be important for proteolytic processing of the translation product. Moreover, the B-domain deleted FVIII protein may contain one or more fragments of the B-domain in order to retain one or more N-linked glycosylation sites. Preferably, the FVIII protein does not contain any furin cleavage sites, resulting in a single chain protein in which light and heavy chains of the protein are covalently linked.

For example, the B-domain deleted FVIII protein may still comprise 0-200 residues, e.g.,1- 100 residues, preferably 8 to 90 residues of the B-domain. The remaining residues of the B- domain may derive from the N-terminus and/or the C-terminus and/or from internal regions of the B-domain. For example, the remaining residues from the C-terminus of the B-domain may contain 1-100, preferably 20-90, more preferably 86 residues. In other embodiments the remaining residues from the C-terminus may contain 1-20 residues, e.g. 4 residues. For example, the remaining residues from the N-terminus of the B-domain may contain 1-100, preferably 2-20 residues, more preferably 2-10 residues, more preferably 4 residues. For example, the remaining residues from internal regions of the B-domain may contain 2-20, preferably 2-10, more preferably 4 to 8 residues. In a preferred embodiment, the FVIII protein comprises 86 C-terminal residues of the B-domain and 4 residues from the N-terminus of the B-domain, e.g., as in FVIII-19M.

Throughout the invention, the recombinant Factor VIII protein of the invention may have at least 90%, at least 95%, at least 96%, at least 97%, at least 98%, or at least 99% sequence identity to a mature (i.e. , not including the signal sequence) FVIII-19M protein of SEQ ID NO: 63, wherein only the A1 , a1 , A2, a2, a3, A3, C1 and C2 domains (residues 20 - 759 and residues 1668 - 2351) are considered for determination of sequence identity. In other words, for determination of sequence identity, the B-domain (residues 760 - 1667 of the full length human sequence SE ID NO: 1, and the residues corresponding thereto in partially B-domain deleted proteins) and the signal sequence (residues 1-19), as well as the albumin-binding domains and, optionally, linkers or other fusion partners, are not taken into account.

Accordingly, the % sequence identity to a mature full length human Factor VIII protein of SEQ ID NO: 1, or to a B-domain deleted variant thereof, e.g., according to SEQ ID NO: 61, to a FVIII protein of SEQ ID NO: 63 is the same, in particular, it is 98.67%, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity. Preferred FVIII proteins of the invention have a sequence identity to SEQ ID NO: 63 of at least 98.74 %, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity.

For example, for a mature B-domain deleted FVIII protein with only one of the recited substitutions, the % sequence identity to mature FVIII-19M protein of SEQ ID NO: 63 is determined over the A1, a1, A2, a2, a3, A3, C1 and C2 domains, i.e. 18 of 1424 amino acids are substituted, and the protein accordingly has at least 98.74% sequence identity to FVIII- 19M protein of SEQ ID NO: 63. For a mature B-domain deleted FVIII protein with 3 of the recited substitutions also occurring in FVIII-19M, the % sequence identity to mature FVIII- 19M protein of SEQ ID NO: 63, is determined over the A1, a1, A2, a2, a3, A3, C1 and C2 domains, i.e. 16 of 1424 amino acids are substituted, and the protein accordingly has 98.88 % sequence identity to FVIII-19M protein of SEQ ID NO: 63. A mature B-domain deleted FVIII protein of the invention with 4 of the recited substitutions also occurring in FVIII- 19M has 15 of 1424 amino acids substituted, and thus has 98.95% sequence identity. A mature B-domain deleted FVIII protein incorporating all 38 recited substitutions has 19 additional substitutions compared to in FVIII-19M, and thus has 98.67 % sequence identity to FVIII-19M.

If sequence identity is defined by reference to the A1 , a1 , A2, a2, a3, A3, C1 and C2 domains only, sequence identity is furthermore determined for the Factor VIII part (as defined, based on the A1, a1, A2, a2, a3, A3, C1 and C2 domains) of the molecule only, i.e., the albumin-binding domains, and optionally, linkers are not taken into account, or if the protein is a fusion protein with a further fusion partner (for example, contains insertions of any size), fused or inserted parts, protein domains or regions (e.g., as further described herein) are not taken into account. Thus, for the determination of sequence identity, if present, fusion partners are ignored, and the % sequence identity to A1 , a1 , A2, a2, a3, A3, C1 and C2 domains is then calculated. Sequence identity can be calculated as known in the art, e.g., using the Needleman-Wunsch algorithm or, preferably, the Smith- Waterman algorithm (Smith et al., 1981. Identification of Common Molecular Subseqences, J Mol Biol. 147: 195-197).

In one embodiment, all residues of the FVIII protein, in particular, with regard to the A1, a1, A2, a2, a3, A3, C1 and C2 domains, except for the substitutions specified herein, correspond to (i.e., are identical to) residues of human Factor VIII protein of SEQ ID NO: 1. Optionally, this may also apply for the B-domain or those parts of the B-domain which are present.

In another embodiment, the FVIII protein of the invention incorporates further mutations, e.g., mutations known in the art to reduce immunogenicity either with regard to further T cell epitopes and/or B cell epitopes, and/or mutations known in the art to improve serum half-life of the protein and/or mutations facilitating purification of the protein, e.g., leading to a single chain protein. Mutations may also be introduced due to partial deletion of the B-domain and engineering of a single chain protein.

A recombinant Factor VIII protein of the invention may have, e.g., at least 90%, optionally, 100% sequence identity to aa 20-1533 of SEQ ID NO: 65 (FVIII-15M), i.e., the mature protein does not comprise the 19 aa N-terminal signal sequence, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity.

A preferred recombinant Factor VIII protein comprises at least 18 amino acid substitutions at positions N79, S112, L160, L171, V184, N233, I265, N299, Y426, S507, F555, N616, L706, Y748, N2038, S2077, S2315 and V2333, wherein preferably the 18 substitutions are N79S, S112T, L160S, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, F555H, N616E, L706N, Y748S, N2038D, S2077G, S2315T and V2333A. Optionally, the protein has at least 90%, e.g., 100% sequence identity to aa 20-1533 of SEQ ID NO: 64 (FVIII-18M), i.e., the mature protein does not comprise the 19 aa N-terminal signal sequence, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity. Another preferred recombinant Factor VIII protein comprises at least 19 amino acid substitutions at positions N79, S112, L160, L171, V184, N233, I265, N299, Y426, S507, F555, N616, L706, Y748, K1837, N2038, S2077, S2315 and V2333, wherein preferably the 19 substitutions are N79S, S112T, L160S, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, F555H, N616E, L706N, Y748S, K1837E, N2038D, S2077G, S2315T and V2333A. Optionally, the protein has 100% sequence identity to aa 20-1533 of SEQ ID NO: 63 (FVIII- 19M), i.e., the mature protein does not comprise the 19 aa N-terminal signal sequence wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity.

While, overall, the inventors found that introduction of the 19 mutations recited above leads to a reduced number of T cell epitopes, mutations at positions L160, F555 and S2315 were found to lead to potential de novo T cell epitopes. It may thus be advantageous to avoid these mutations. Consequently, another preferred recombinant Factor VIII protein comprises at least 16 amino acid substitutions at positions N79, S112, L171, V184, N233, I265, N299, Y426, S507, N616, L706, Y748, K1837, N2038, S2077, and V2333, wherein preferably the 16 substitutions are N79S, S112T, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, N616E, L706N, Y748S, K1837E, N2038D, S2077G, and V2333A. Preferably, said protein comprises at least one, preferably, all of L160, F555 and S2315, i.e., it does not comprise mutations at these positions. Optionally, the protein has 100% sequence identity to aa 20- 1533 of SEQ ID NO: 134 (FVIII-16M_SC), i.e., the mature protein does not comprise the 19 aa N-terminal signal sequence wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity.

Sequences of further FVIII proteins to which the proteins of the invention may have 100% sequence identity, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity, are provided as SEQ ID NO: 66-73 or 109, wherein, while the sequences all comprise the 19 aa N-terminal signal sequence, the preferred mature FVIII proteins of the invention do not comprise the signal sequence any more. Accordingly, they may have at least 90% or, optionally, 100% sequence identity to aa 20-1533 of SEQ ID NO: 66-73 or 103, respectively, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity.

The invention further provides FVIII proteins of the invention having 100% sequence identity to any of SEQ ID NO: 114-119, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity, wherein, while the sequences all comprise the 19 aa N-terminal signal sequence, the preferred mature FVIII proteins of the invention do not comprise the signal sequence any more. Accordingly, they may have 100% sequence identity to aa 20-1778 of SEQ ID NO: 114, or to aa 20-1594 of SEQ ID NO: 115, or to aa 20-1575 of SEQ ID NO: 116, or to aa 20-1575 of SEQ ID NO: 117, or to aa 20-1686 of SEQ ID NO: 118, or to aa 20-1778 of SEQ ID NO: 119, respectively, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity.

The invention further provides FVIII proteins having at least 80%, preferably, at least 90%, at least 95%, at least 99% or 100% sequence identity to any of SEQ ID NO: 114-119, wherein, while the sequences all comprise the 19 aa N-terminal signal sequence, the preferred mature FVIII proteins of the invention do not comprise the signal sequence any more. These proteins also may comprise the albumin-binding domains and/or linkers as defined in SEQ ID NO:

114-119. Accordingly, they may have at least 80%, preferably, at least 90%, at least 95%, at least 99% or 100% sequence identity to aa 20-1778 of SEQ ID NO: 114. A protein of the invention may have at least 80%, preferably, at least 90%, at least 95%, at least 99% or 100% sequence identity to aa 20-1594 of SEQ ID NO: 115. A protein of the invention may have at least 80%, preferably, at least 90%, at least 95%, at least 99% or 100% sequence identity to aa 20-1575 of SEQ ID NO: 116. A protein of the invention may have at least 80%, preferably, at least 90%, at least 95%, at least 99% or 100% sequence identity to aa 20-1575 of SEQ ID NO: 117. A protein of the invention may have at least 80%, preferably, at least 90%, at least 95%, at least 99% or 100% sequence identity to aa 20-1686 of SEQ ID NO:

118. A protein of the invention may have at least 80%, preferably, at least 90%, at least 95%, at least 99% or 100% sequence identity to aa 20-1778 of SEQ ID NO: 119.

As described above, the FVIII part of the FVIII protein of the invention may be de-immunized. Additionally or alternatively, it is also possible to de-immunize the linker regions between the FVIII parts, the ABDs and the linkers. Substitutions in positions which lead to de immunization of such regions are specified in Table 10 below. FVIII proteins of the invention may comprise at least one, preferably, 2 or more, three or more, for or more, five of more, six or more, seven or more, eight or more, nine or more, ten or more, 11 or more, 12 or more, 13 or more or 14 of the substitutions recited in said table, preferably, in one of the combinations listed. Further, the first amino acid of the thrombin cleavable linker C terminal to the processing sequence may be substituted to a D, as defined, e.g., above.

In addition to the albumin-binding domains contained in the FVIII protein of the invention, the protein may be a fusion protein with another fusion partner, e.g., a fusion protein of a recombinant Factor VIII protein having at least 90%, at least 95%, at least 96%, at least 97%, at least 98% or at least 99% sequence identity to a FVIII-19M as specified in SEQ ID NO: 63, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for calculation of sequence identity. The fusion partner preferably extends the in vivo serum half-life of the FVIII protein of the invention. The fusion partner may be selected from the group comprising an Fc region, albumin, PAS polypeptides, HAP polypeptides, the C-terminal peptide of the beta subunit of chorionic gonadotropin, and combinations thereof. The FVIII protein may alternatively or additionally be covalently linked to non-protein fusion partners such as albumin-binding small molecules, and/or PEG (polyethylenglycol) and/or HES (hydroxyethyl starch). PAS polypeptides or PAS sequences are polypeptides comprising an amino acid sequence comprising mainly alanine and serine residues or comprising mainly alanine, proline and serine residues, the PAS sequences forming a random coil conformation under physiological conditions, as defined in WO 2015/023894. HAP polypeptides or sequences are homo-amino acid polymer (HAP), comprising e.g., repetitive sequences of Glycine or Glycine and Serine, as defined in WO 2015/023894. Potential fusions, fusion partners and combinations thereof are described in more detail e.g., in WO 2015/023894.

Preferably, for therapeutic applications, the recombinant FVIII protein is at least fused to an Fc region. Fusion proteins of FVIII to Fc regions are known in the state of the art to reduce immunogenicity (Krishnamoorthy et al., Recombinant factor VIII Fc (rFVIIIFc) fusion protein reduces immunogenicity and induces tolerance in hemophilia A mice, Cell. Immunol. 2016, http://dx.doi.Org/10.1016/j.cellimm.2015.12.2008; Carcao et al., Recombinant factor VIII Fc fusion protein for immune tolerance induction in patients with severe haemophilia A with inhibitors - A retrospective analysis. Haemophilia 2018:1-8).

Fusion partners may e.g., be linked to the N-terminus or the C-terminus of the FVIII protein of the invention, but they may also be inserted within the FVIII sequence, as long as the FVIII protein remains functional as defined herein. As described above, for determination of sequence identity, insertions of, e.g., one, two, three, four, five, six, seven, eight, nine or ten fusion partners, as defined herein, are not considered to reduce sequence identity when the sequence identity is defined by reference to the A1, a1, A2, a2, a3, A3, C1 and C2 domains.

The inventors found that a high proportion of the FVIII protein of the invention was produced as a single chain protein in the cell lines selected for production. Production of FVIII as a single chain protein is not believed to reduce coagulant activity, but may be beneficial for purification. To simplify purification, the FVIII protein of the invention may be a single chain protein or at least have a proportion of at least 40%, at least 50%, at least 60%, at least 70%, at least 80%, at least 90% or at least 95% single chain protein. Alternatively, the FVIII protein of the invention may be produced as a heterodimeric FVIII protein. Preferably, the FVIII protein of the invention is a single chain B-domain deleted Factor VIII protein. Recombinant single chain FVIII proteins are known in the art, wherein, e.g., at least part of the B-domain and 4 amino acids of the adjacent acidic a3 domain (e.g., residues 784-1671 of full length FVIII) are removed, in particular, removing the furin cleavage-site (EMA/CHM P/699390/2016 - Assessment report AFSTYLA). An exemplary single chain FVIII protein is provided as SEQ ID NO: 62. An exemplary FVIII single chain protein based on SEQ ID NO: 62 which incorporates 19 mutations as specified herein, e.g., the same 19 mutations incorporated in FVIII-19M, lacks 4 amino acids of the a3 domains of FVIII-19M, i.e., it has 99.72 % (at least 99% sequence identity) to SEQ ID NO: 63, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for calculation of sequence identity. Said protein may be B-domain deleted, and it may be a fusion protein, e.g., as described above.

The protein may further be glycosylated and/or sulfated. Preferably, post-translational modifications such as glycosylation and/or sulfation of the protein occur in a human cell.

In one embodiment of the invention, the protein is capable of association with vWF. For example, the binding potency of the FVIII protein of the invention to vWF is 0%-100%, 10%- 90%, 20-80%, 30-70%, 40-60% or 50-60% of the binding potency of ReFacto AF to vWF, which can be determined by an ELISA-based method, e.g., as described herein. As shown herein, the binding capacity of a FVIII protein of the invention comprising several of the recited mutations may be reduced compared to ReFacto AF, e.g., to less than 60%.

The protein of the invention is preferably stable in human plasma in vitro and in vivo, so that it can be pharmaceutically used. The inventors could show that about 83% of chromogenic coagulant activity of FVIII-19M were maintained after in vitro incubation in human plasma at 37°C for 24 hours. For FVI I l-6rs, under the same conditions, about 91% coagulant activity were maintained, for ReFacto AF and Nuwiq, it was 97%.

Preferably, in vivo, the half-life of the FVIII protein of the invention in human serum (in a patient without inhibitors) is about at least 6 hours, preferably, at least 12 hours, at least 18 hours, at least 24 hours, or at least to 30 hours. As defined herein, the FVIII protein may be a FVIII protein without a further fusion partner, or it may be a fusion protein as defined herein. Optionally, the specified half-life is already obtained without fusion partners. In case of the presence of further partners the half-life of the FVIII protein may be the same, or even longer. The combination of substitutions described herein reduces the immunogenicity score for all subjects having at least one of the analyzed HLA-DR supertype alleles (DRB1*0101, DRB1*0301, DRB1*0401, DRB1*0701, DRB1*0801, DRB1*1101, DRB1*1301, DRB1*1501). These subtypes are present in more than 90% of the population (cf. Southwood et al. , J. Immunol. 1998;160;3363-3373). FVIII of the invention can thus advantageously be used for treatment of all patients in need thereof, in particular those having one of the HLA-DR supertype alleles. Table 5 shows, for the Example of FVIII-19M, that immunogenicity is differently reduced for patients with different alleles.

Table 5 A-C: Individual T cell Epitope Measure (ITEM) scores indicating the immunogenicity for FVI I l-6rs (A) and FVIII-19M (B) for different HLA-DR supertypes, and absolute reduction of the Immunogenicity for FVI 11-19M compared to FVIII-6rs for different HLA-DR supertypes (C). The ITEM Score is based on the number and intensity of the EpiMatrix Hits (method see below) for a pair of alleles normalized for the length of the protein. A low ITEM score in Table 5A or B reflects a low immunogenicity. A high reduction in the ITEM score in Table 5C, i.e. , a high positive value in said table, reflects a high benefit from the substitutions introduced.

A: FVIII-6rs

B: FVIII-19M C: Absolute reduction of the immunogenicity score for FVIII-19M compared to FVIII-6rs

For all alleles analysed, there is a reduction in the immunogenicity score, in particular, the reduction in immunogenicity score is more than 13. A particularly high reduction in the immunogenicity score of more than 17 shows that patients having one of the following combination of HLA types can particularly benefit from treatment with the pharmaceutical composition of the invention:

• DRB1*0701 in combination with DRB1*0101, DRB1*0301, DRB1*0401, DRB1*0701, DRB1*0801 , DRB1*1101, or DRB1*1501;

• DRB1*0801 in combination with DRB1*0101, DRB1*0701, DRB1*0801, DRB1*1101, or DRB1*1501 ;

• DRB1*1101 in combination with DRB1*0101, DRB1*0301, DRB1*0401, DRB1*0701, DRB1*0801 , DRB1*1101, DRB1*1301, or DRB1*1501;

• DRB*1301 in combination with DRB1*1101;

• DRB1*1501 in combination with DRB1*0101, DRB1*0301, DRB1*0401, DRB1*0701, DRB1*0801 , DRB1*1101, DRB1*1501

A still higher reduction in the immunogenicity score of more than 20 shows that patients having one of the following combination of HLA types can even more particularly benefit from treatment with the pharmaceutical composition of the invention: DRB1*1101 in combination with DRB1*0701, DRB1*0801, DRB1*1101, or DRB1*1501. Treatment of patients having a particularly high reduction in immunogenic score is preferred.

The invention teaches an in vitro method for preparing a FVIII protein of the invention, comprising culturing a host cell of the invention expressing said FVIII protein under suitable conditions, and isolating said FVIII protein, wherein the protein is optionally formulated as a pharmaceutical composition. As described herein, the host cell preferably is a human cell.

The present invention further discloses an assay for determining immunogenicity of a protein, e.g., a FVIII protein of the invention, comprising co-cultivating dendritic cells incubated with said protein and regulatory T-cell-depleted CD4+ T cells of a donor (e.g., a healthy human or a patient, e.g., with Hemophilia A) and testing activation of said T cells. For such an assay, monocytes may be purified, e.g., from PBMCs, differentiated to immature DCs (iDCs) (e.g., in the presence of IL-4 and GM-CSF) and finally stimulated to become mature DCs (mDCs) (e.g., using LPS or a mixture of cytokines such as IL-1beta, IL-6 and TNF-alpha) and, at the same time, incubated with the antigen, i.e., the protein of interest. CD4+CD25-T cells are purified from PBMCs of the same donor, preferably, from the same batch of PBMC, labeled with CFSE (Carboxyfluorescein diacetate succinimidyl ester) for later detection of proliferation and cultivated prior to co-cultivation, e.g., to provide time for recovery of the cells and removal of not steadily bound CFSE. After co-culture with the DC, which preferably takes place at a ratio of DC:T cells of at least 1:10, e.g., for about 12 h to about 2 weeks, for 1 day to 9 days or 2 days to 7 days, the T cells are analyzed for activation and/or proliferation by flow cytometry. Alternatively or additionally, the supernatant may be analyzed for cytokines. Preferable conditions for the assay are described in the examples below. This assay has the advantage that it allows for assessment of both primary and secondary T-cell- mediated immune responses in the absence of regulatory CD25+ T cells, which facilitates detection of immune responses. The results are expected to correlate with immunogenicity of the protein in vivo.

For the de-immunized FVIII protein FVIII-19M, the assay confirmed that, in the majority of subjects analyzed, there was a reduced T cell proliferation in response to FVIII-19M. It can be concluded that a low immunogenicity score correlated with a low immunogenicity of the protein in the in vitro assay, i.e., the substitutions in the epitopes identified in silico translates into a reduced immunogenicity.

Nucleic acids and host cells

The invention also provides a nucleic acid encoding a recombinant Factor VIII protein of the invention. Said nucleic acid may be an expression vector, e.g., suitable for expression of said recombinant Factor VIII protein in a mammalian cell, such as a human cell, such as a CAP cell. The nucleic acid preferably encodes the FVIII with an N-terminal signal sequence, e.g., the 19 aa signal sequence of SEQ ID NO: 1. Preferred nucleic acids of the invention encode a recombinant FVIII protein having SEQ ID NO: 47 (ADLCLD_SC), SEQ ID NO: 48 (AD2CD2_SC), SEQ ID NO: 49 (AD2CD2woL_SC), SEQ ID NO: 50 (AD2CD2woLG_SC) or SEQ ID NO: 51 (AbD2CD2_SC). They may be SEQ ID NO: 52-56. Nucleic acids of the invention encoding a de-immunized FVIII protein may have, e.g., any of SEQ ID NO: 121- 126. The nucleic acids of the invention may be DNA molecules or RNA molecules. The nucleic acids may be optimized for expression in the respective host cell, e.g., in a human cell, e.g., a CAP cell.

The expression vector comprises the sequence encoding the FVIII protein, preferably, in codon-optimized form, under the functional control of a suitable promoter, which may be a constitutive or an inducible promoter. The promoter may be a promoter not associated with expression of FVIII in nature, e.g., EF-1alpha or a heterologous promoter, e.g., CMV or SV40. It may further comprise pro- and/or eukaryotic selection markers, such as ampicillin resistance and dihydrofolate reductase (dhfr), and origins of replication, e.g., an SV40 origin and/or a pBR322 origin “codon-optimized” means optimized for expression in the host cell, preferably, for expression in a human host cell.

In certain embodiments, the nucleic acid may be a vector suitable for gene therapy, e.g., for gene therapy of a human patient. Vectors suitable for gene therapy are known in the art, e.g., virus-based vectors e.g., based on adenovirus or adeno-associated virus (AAV) or based on retrovirus, such as lentiviral vectors etc., or non virus-based vectors such as but not limited to small plasmids and minicircles or transposon-based vectors. An AAV-based vector of the invention may e.g., be packaged in AAV particles for gene therapy of Hemophilia A patients.

The invention also provides a host cell comprising a nucleic acid of the invention. The host cell may be a bacterial cell, a plant cell, a fungal cell, a yeast cell or an animal cell.

Preferably, the host cell is an animal cell, in particular, a mammalian cell comprising an expression vector suitable for expression of said recombinant Factor VIII protein in said cell. The host cell preferably is a human cell comprising an expression vector suitable for expression of said recombinant Factor VIII protein in said human cell. The cell may be transiently or stably transfected with the nucleic acid of the invention. The cell may be a cell line, a primary cell or a stem cell. For production of the protein, the cell typically is a cell line such as a HEK cell, such as a HEK-293 cell, a CHO cell, a BHK cell, a human embryonic retinal cell such as Crucell's Per.C6 or a human amniocyte cell such as CAP. For treatment of human patients with the protein, the host cell preferably is a human cell, e.g., a HEK293 cell line or a CAP cell line (e.g. a CAP-T cell or a CAP-Go cell). The inventors have found that in a CAP cell line, a particularly high single chain content of FVIII protein of the invention is produced. Among the CAP cells, CAP-T cells are preferred for transient expression, while CAP-Go cells may be used for creation of stable cell lines conveying an advantageous glycosylation profile to the FVIII molecule.

The cell may be an autologous cell of a Hemophilia A patient, in particular, a human Hemophilia A patient, suitable for producing FVIII in the patient after transfection and reintroduction into the patient's body. The cell may be a stem cell, e.g., a hematopoietic stem cell, but preferably it is not an embryonic stem cell, in particular when the patient is a human. The cell may also be hepatocyte, a liver sinusoidal endothelial cell or a thrombocyte.

Cell lines expressing the protein of the invention may also be used in a method of preparing the protein of the invention, comprising cultivating said cells under conditions suitable for expression of the FVIII protein and purifying said protein, e.g., using a plurality of methods known to the skilled person, such as described herein. Such purification methods may comprise standard harvesting procedures for cell removal, e.g. centrifugation, followed by chromatography steps, e.g. affinity chromatography, and methods for exchanging the FVIII proteins into a suitable buffer. The invention thus also provides a method for preparing a Factor VIII protein, comprising culturing the host cell of the invention under conditions suitable for expression of the Factor VIII protein and isolating the Factor VIII protein, wherein the method optionally comprises formulating the Factor VIII protein as a pharmaceutical composition.

Pharmaceutical compositions

The invention provides a pharmaceutical composition comprising the recombinant Factor VIII protein of the invention, the nucleic acid of the invention or the host cell of the invention.

Such pharmaceutical compositions may comprise suitable excipients or carriers, e.g., a buffer, a stabilizing agent, a bulking agent, a preservative, another (e.g., recombinant) protein or combinations thereof. In the context of the invention, if not explicitly stated otherwise, "a" is understood to mean one or more.

A suitable buffer for formulation of the protein of the invention may e.g. contain 205 mM NaCI, 5.3 mM CaCh, 6.7 mM L-Histidine, 1.3 % Sucrose and 0.013 % Tween 20 in distilled water and have a pH of 7.0 (FVIII formulation buffer). Said buffer is used in the experiments described herein if not otherwise stated. Formulations of FVIII may be sterile, e.g., sterile filtered, in particular for in vivo use. Optionally, the pharmaceutical composition of the invention comprising FVIII protein further comprises albumin, preferably, for a human patient, human serum albumin. Albumin may, e.g., be in a concentration of 0.1-10% w/w, such as 1-5% human serum albumin (w/w). Albumin may bind to the ABD of the FVIII proteins of the invention either before administration, or after administration to a human subject.

The pharmaceutical composition may be formulated as desired appropriate by the skilled person, e.g., for intravenous (i.v.) or subcutaneous application, intraperitoneal or intramuscular application. Generally, it is for administration as slow i.v. push bolus injection. Continuous infusion is indicated e.g., for patients requiring admission for severe bleeds or surgical procedures. Oral application, which may contribute to tolerance induction, is also possible, e.g., after expression in plants. The pharmaceutical composition may be for slow release.

Pharmaceutical compositions comprising FVIII can be lyophilized.

Due to the increase in half-life in vivo, the pharmaceutical compositions of the invention may be administered at longer intervals than previous FVIII compositions. For example, they may be for use in administration every 5 to 14 days, preferably, every 7 to 10 days.

Dosages and treatment schemes may be chosen as appropriate, e.g., for prophylaxis of bleeding or with intermittent, on-demand therapy for bleeding events. Decisions on dosing may be made by the physician. Dosing depends on the patent, e.g., weight, FVIII status, severity of disease etc. For example, the FVIII of the invention may be administered in dosages of 0.5 to 250 lU/kg body weight every 0.5 to 14 or every 6-7 days intravenously depending on the severity of the disease, typically, 0.5 to 200 lU/kg body weight.

The invention also provides a pharmaceutical composition comprising the FVIII protein of the invention in combination with an immunosuppressive agent (e.g., methylprednisolone, prednisolone, dexamethasone, cyclophosphamide, rituximab, and/or cyclosporin), and/or it may be for administration at substantially the same time (e.g. within five minutes to within 12 hours) with such an agent. The invention thus also provides a kit comprising, in addition to a FVIII protein of the invention, optionally, combined with albumin, an immunosuppressive agent, e.g., an immunosuppressive agent selected from the group comprising methylprednisolone, prednisolone, dexamethason, cyclophosphamide, rituximab, and/or cyclosporin.

The pharmaceutical composition, e.g., comprising the protein of the invention, may be for use in treating a patient in need thereof, in particular, a Hemophilia A patient, e.g., a patient with acquired hemophilia involving an autoimmune response to FVIII or a congenital Hemophilia A patient. Mammals such as mice or dogs may be treated with the pharmaceutical composition of the invention, but the patient typically is a human patient.

Use of a pharmaceutical composition of the invention comprising a de-immunized FVIII protein or a nucleic acid encoding the same, as described herein, is particularly advantageous in settings wherein a reduced immunogenicity is desired, e.g., for use in treating a patient with Hemophilia A not previously treated with any recombinant or plasmatic Factor VIII protein. According to the invention, the incidence and/or severity of generation of antibodies including inhibitory antibodies in the patient is thus reduced compared to treatment with conventional FVIII, or preferably, the generation of antibodies including inhibitory antibodies is prevented. The pharmaceutical composition of the invention may also be used for treatment of a patient previously already treated with a recombinant and/or plasmatic Factor VIII protein. In a patient who has an antibody including an inhibitory antibody response to a recombinant and/or plasmatic Factor VIII protein, the pharmaceutical compositions may, e.g., be used for immune tolerance induction (ITI) treatment, as it is desired to use a FVIII protein having a low immunogenicity or even tolerogenic characteristics (Carcao et al., Recombinant factor VIII Fc fusion protein for immune tolerance induction in patients with severe haemophilia A with inhibitors - A retrospective analysis. Haemophilia 2018:1-8). The compositions of the invention may thus also be used for rescue ITI. The pharmaceutical compositions may also be advantageously used in a patient who has had an antibody response including an inhibitory antibody response to a recombinant and/or plasmatic Factor VIII protein, e.g., who has been treated by ITI. The pharmaceutical compositions may also be advantageously used in a patient who has had an antibody response including an inhibitory antibody response to a recombinant and/or plasmatic Factor VIII protein, who has not been treated by ITI.

The invention also provides a vial comprising the pharmaceutical composition of the invention, e.g., a syringe. The syringe may be a pre-filled syringe, e.g., a ready-to-use syringe.

The inventors contemplate that, in the context of gene therapy, due to the presence of the albumin binding domains, long-term expression and folding of the FVIII protein of the invention may be improved compared to expression of conventional FVIII.

All publications cited herein are fully incorporated herewith. The invention is further illustrated by the following embodiments, figures and examples, which are not to be understood as limiting the scope of the invention. Embodiments

In summary, in a first embodiment, the invention provides a recombinant Factor VIII protein comprising a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least one albumin binding domain is C-terminal to the heavy chain portion and at least one albumin binding domain is C-terminal to the light chain portion, wherein, if the protein is a single chain protein, the albumin binding domain(s) C- terminal to the heavy chain portion is/are N-terminal to the light chain portion.

In a second embodiment, the recombinant Factor VIII protein of the first embodiment (embodiment 1) is a single chain protein.

In a third embodiment, the recombinant Factor VIII protein of embodiment 1 is a double chain protein.

In a fourth embodiment, in the recombinant Factor VIII protein of any of embodiments 1-3, one albumin binding domain is C-terminal to the heavy chain portion and one albumin binding domain is C-terminal to the light chain portion.

In a fifth embodiment, in the recombinant Factor VIII protein of any of embodiments 1-3, one albumin binding domain is C-terminal to the heavy chain portion and two albumin binding domains are C-terminal to the light chain portion. In a sixth embodiment, in the recombinant Factor VIII protein of any of embodiments 1-3, one albumin binding domain is C-terminal to the heavy chain portion and three albumin binding domains are C-terminal to the light chain portion. In a seventh embodiment, in the recombinant Factor VIII protein of any of embodiments 1-3, one albumin binding domain is C-terminal to the heavy chain portion and four albumin binding domains are C-terminal to the light chain portion.

In an eighth embodiment, in the recombinant Factor VIII protein of any of embodiments 1-3, two albumin binding domains are C-terminal to the heavy chain portion and one albumin binding domain is C-terminal to the light chain portion. In a ninth embodiment, in the recombinant Factor VIII protein of any of embodiments 1-3, three albumin binding domains are C-terminal to the heavy chain portion and one albumin binding domain is C-terminal to the light chain portion. In a tenth embodiment, in the recombinant Factor VIII protein of any of embodiments 1-3, four albumin binding domains are C-terminal to the heavy chain portion and one albumin binding domain is C-terminal to the light chain portion.

In an eleventh embodiment, in the recombinant Factor VIII protein of any of embodiments 1- 3, at least two albumin binding domains are C-terminal to the heavy chain portion and at least two albumin binding domains are C-terminal to the light chain portion, preferably, two albumin binding domains are C-terminal to the heavy chain portion and two albumin binding domains are C-terminal to the light chain portion. In a twelfth embodiment, in the recombinant Factor VIII protein of embodiment 11, two albumin binding domains are C- terminal to the heavy chain portion and three albumin binding domains are C-terminal to the light chain portion. In a thirteenth embodiment, in the recombinant Factor VIII protein of embodiment 11, two albumin binding domains are C-terminal to the heavy chain portion and four albumin binding domains are C-terminal to the light chain portion. In a fourteenth embodiment, in the recombinant Factor VIII protein of embodiment 11, three albumin binding domains are C-terminal to the heavy chain portion and two albumin binding domains are C- terminal to the light chain portion. In a fifteenth embodiment, in the recombinant Factor VIII protein of embodiment 11 , four albumin binding domains are C-terminal to the heavy chain portion and two albumin binding domains are C-terminal to the light chain portion.

In a sixteenth embodiment, in the recombinant Factor VIII protein of any of embodiments 1- 15, albumin-binding domains are separated from the heavy chain portion and/or the light chain portion and/or other albumin-binding domains by a linker, wherein, preferably, albumin binding domains are separated from the heavy chain portion and the light chain portion and other albumin-binding domains by a linker.

In a seventeenth embodiment, in the recombinant Factor VIII protein of embodiment 16, the linker comprises a Thrombin-cleavable linker section that optionally has the sequence of SEQ ID NO: 39.

In an eighteenth embodiment, in the recombinant Factor VIII protein of any of embodiments 16 or 17, the linker comprises a glycine-serine linker section that optionally has the sequence of SEQ ID NO: 40 or SEQ ID NO: 41.

In a nineteenth embodiment, in the recombinant Factor VIII protein of any of embodiments 16-18, said linker is a combination of different linker sections, e.g. the linker comprises a Thrombin-cleavable linker section flanked on each side by a glycine-serine linker section, wherein said linker optionally has the sequence of SEQ ID NO: 42 or SEQ ID NO: 43.

In a twentieth embodiment, in the recombinant Factor VIII protein of any of embodiments 1- 19, the albumin binding domain comprises a sequence according to SEQ ID NO: 44.

In a twenty-first embodiment, in the recombinant Factor VIII protein of embodiment 20, the albumin binding domain comprises a sequence according to SEQ ID NO: 46. In a twenty-second embodiment, in the recombinant Factor VIII protein of any of embodiments 1-21, the heavy chain portion comprises the domains A1 and A2, and optionally comprises the domains A1-a1-A2-a2 or A1-a1-A2-a2-B.

In a twenty-third embodiment, in the recombinant Factor VIII protein of any of embodiments 1-22, the light chain portion comprises the domains A3 and C1 and C2, and optionally comprises the domains a3-A3-C1-C2.

In a twenty-fourth embodiment, in the recombinant Factor VIII protein of any of embodiments 1-23, the B-domain of the Factor VIII protein is at least partly deleted.

In a twenty-fifth embodiment, the recombinant Factor VIII protein of any of embodiments 1-24 comprises, in a single chain, a heavy chain portion comprising an A1 and an A2 domain and a light chain portion comprising an A3, C1 and C2 domain of Factor VIII, wherein a) in said recombinant Factor VIII protein, 894 amino acids corresponding to consecutive amino acids between F761 and P1659 of wild type Factor VIII as defined in SEQ ID NO: 1 are deleted, leading to a first deletion; b) said recombinant Factor VIII protein comprises, spanning the site of the first deletion, a processing sequence comprising SEQ ID NO: 2 or a sequence having at most one amino acid substitution in SEQ ID NO: 2, wherein said processing sequence comprises a first thrombin cleavage site; c) in said recombinant Factor VIII protein, at least the amino acids corresponding to amino acids R1664 to R1667 of wild type Factor VIII are deleted, leading to a second deletion; and d) said recombinant Factor VIII protein comprises, C-terminal to the second deletion and N- terminal of the A3 domain, a second thrombin cleavage site.

In a twenty-sixth embodiment, the recombinant Factor VIII protein of any of embodiments 1- 25 that optionally is a single chain protein comprises a heavy chain portion having at least 90% sequence identity to aa20-aa768 of SEQ ID NO: 16 and a light chain portion having at least 90% sequence identity to aa769-aa1445 of SEQ ID NO: 16, wherein said sequence identities preferably are at least 95%, at least 98% or 100%.

In a twenty-seventh embodiment, the recombinant Factor VIII protein of any of embodiments 1-26 that optionally is a single chain protein comprises a heavy chain portion having at least 90% sequence identity to aa20-aa1667 of SEQ ID NO: 1 and a light chain portion having at least 90% sequence identity to aa1668-aa2351 of SEQ ID NO: 1, wherein said sequence identities optionally are at least 95%, at least 98% or 100%. In a twenty-eigth embodiment, the recombinant Factor VIII protein of any of embodiments 1-4 and 16-27, comprises one albumin binding domain between the heavy chain portion and the light chain portion and one albumin binding domain C-terminal to the light chain portion, wherein the sequence has at least 70% sequence identity to SEQ ID NO: 47.

In a twenty-ninth embodiment, the recombinant Factor VIII protein of any of embodiments 11- 28 is a single chain protein comprising at least two albumin binding domains between the heavy chain portion and the light chain portion and at least two albumin binding domain C- terminal to the light chain portion, wherein the protein has at least 80% sequence identity, optionally, at least 90% sequence identity, at least 95% sequence identity or at least 98% sequence identity to any of SEQ ID NO: 48, 49 or 51. In a thirtieth embodiment, the recombinant Factor VIII protein of embodiment 29 has at least 80% sequence identity to SEQ ID NO: 48, e.g., it has SEQ ID NO: 48 or SEQ ID NO: 51.

In a thirty-first embodiment, the recombinant Factor VIII protein of any one of embodiments 1-30 has a "b mutation", i.e. , a mutation of the amino acid corresponding to Y1699 to F at position 1699 and a mutation of the amino acid corresponding to Y1683 to F at position 1683 in wt Factor VIII protein of SEQ ID NO: 1.

In a thirty-second embodiment, the in vivo half-life of the recombinant Factor VIII protein of any one of embodiments 1-31 in a human subject is prolonged by a factor of at least 1.2, preferably, by a factor of at least 1.5, optionally, at least 2 or at least 2.5 in comparison to a recombinant Factor VIII protein of SEQ ID NO: 28.

In a thirty-third embodiment, the recombinant Factor VIII protein of any one of embodiments 1-32 is a fusion protein with at least one fusion partner selected from the group consisting of an Fc region, albumin, PAS polypeptides, HAP polypeptides, the C-terminal peptide of the beta subunit of chorionic gonadotropin, polyethylenglycol, and hydroxyethyl starch.

In a thirty-fourth embodiment, the recombinant Factor VIII protein of any one of embodiments 1-33 is a de-immunized protein.

The invention provides, as a thirty-fifth embodiment, a recombinant Factor VIII protein of any of embodiments 1-34, comprising at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79, I80, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, I632, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, S2125, F2215, K2226, K2258, V2313,

S2315, V2333 and Q2335; wherein substitutions of N are independently selected from the group consisting of D, H, S and E; wherein substitution of I are independently selected from the group consisting of T and V; wherein substitutions of S are independently selected from the group consisting of A, N, G, T and E; wherein substitutions of L are independently selected from the group consisting of N, Q, F and S; wherein substitutions of V are independently selected from the group consisting of A and T; wherein substitutions of Y are independently selected from the group consisting of N, H and S; wherein substitutions of F are independently selected from the group consisting of H and S; wherein substitutions of K are independently selected from the group consisting of N, D, E, Q, S and T; wherein substitutions of R are independently selected from the group consisting of Q, H and S; wherein substitutions of M are selected from the group consisting of R, Q, K and T ; and/or wherein substitutions of Q are selected from the group consisting of R, D, E, H and K; wherein the positions are specified in relation to full length human Factor VIII molecule of SEQ ID NO: 1; and wherein the recombinant Factor VIII protein retains at least 50% coagulant activity, as determined in a chromogenic assay, compared to a Factor VIII protein consisting of SEQ ID NO: 60. The recombinant Factor VIII protein can be a fusion protein. In a thirty-seventh embodiment, the recombinant Factor VIII protein of embodiment 35 comprises at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79, 180, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, F2215, K2226, K2258, V2313, S2315, V2333 and Q2335.

In a thirty-seventh embodiment, the recombinant Factor VIII protein of any one of embodiments 1-36 comprises at least one amino acid substitution at a position selected from the group consisting of Y748, L171, S507, N79, I80, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, I632, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, S2125, F2215, K2226, K2258, V2313, S2315, V2333 and Q2335; wherein substitutions of N are independently selected from the group consisting of D, H, S and E; wherein substitution of I are independently selected from the group consisting of T and V; wherein substitutions of S are independently selected from the group consisting of A, N, G, T and E; wherein substitutions of L are independently selected from the group consisting of N, Q, F and S; wherein substitutions of V are independently selected from the group consisting of A and T; wherein substitutions of Y are independently selected from the group consisting of N, H and S; wherein substitutions of F are independently selected from the group consisting of H and S; wherein substitutions of K are independently selected from the group consisting of N, D, E, Q, S and T; wherein substitutions of R are independently selected from the group consisting of Q, H and S; wherein substitutions of M are selected from the group consisting of R, Q, K and T ; and/or wherein substitutions of Q are selected from the group consisting of R, D, E, H and K ; wherein, if the mutation is at position S507, it is S507E, and if the mutation is at position N616, it is N616E, and if the mutation is at position F2215, it is F2215H; wherein the positions are specified in relation to full length human Factor VIII molecule of SEQ ID NO: 1, and wherein the recombinant Factor VIII protein retains at least 50% coagulant activity, as determined in a chromogenic assay, compared to a Factor VIII protein consisting of SEQ ID NO: 60. The protein may be a fusion protein.

In a thirty-eighth embodiment, the recombinant Factor VIII protein of any of embodiments 35-

37 may e.g. comprise amino acid substitutions selected from the group consisting of Y748S, L171Q, S507E, N79S, I80T, 1105V, S112T, L160S, V184A, N233D, L235F, V257A, I265T, N299D, Y426H, Y430H, L505N, F555H, I610T, N616E, I632T, L706N, N754D, K1837E, R1936Q, S2030A, S2037G, N2038D, S2077G, M2123K, S2125G, F2215H, K2226Q, K2258Q, V2313A, S2315T, V2333A and Q2335H.

In a thirty-ninth embodiment, the recombinant Factor VIII protein of any of embodiments 35-

38 may e.g. comprise 3-25 of said substitutions and the substitutions may be located within different immunogenic clusters.

In a fortieth embodiment, the recombinant Factor VIII protein of any of embodiments 35-39 may e.g. comprise at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79, S112, L160, V184, N233, I265, N299, Y426, F555, N616, I632, L706, K1837, R1936, N2038, S2077, S2125, F2215, K2226, K2258, S2315, and V2333; wherein the at least three amino acid substitutions are preferably selected from the group consisting of Y748S, L171Q, S507E, N79S, S112T, L160S, V184A, N233D, I265T, N299D, Y426H, F555H, N616E, I632T, L706N, K1837E, R1936Q, N2038D, S2077G, S2125G, F2215H, K2226Q, K2258Q, S2315T and V2333A.

In a forty-first embodiment, the recombinant Factor VIII protein of any of embodiments 35-40 may e.g. comprise amino acid substitutions at least at positions a. N79S, S112T, N233D, and I265T; and/or b. N79S, S112T, L160S, L171Q, V184A, N233D, and I265T; and/or c. N299D, Y426H, and S507E; and/or d. F555H, N616E, L706N, Y748S; and/or e. F555H, N616E, I632T, L706N, and Y748S; and/or f. S2077G, S2315T, and V2333A; and/or g. N2038D, S2077G, S2315T, and V2333A; and/or h. S2077G, K2258Q, S2315T, and V2333A; and/or i. N2038D, S2077G, K2258Q, S2315T, and V2333A; and/or j. N2038D, S2077G, S2125G, K2258Q, S2315T, and V2333A; and/or k. L171Q, S507E, Y748S and V2333A; and/or

L. L171Q, N299D, N616E and V2333A; and/or m. S112T, S507E, Y748S, K1837E and N2038D; and/or n. S112T, Y426H, N754D, K1837E and N2038D preferably, combining at least the substitutions specified under b and c, optionally further including substitutions selected from those specified under d or e and/or f, g, h, I or j and/or K1837E.

In a forty-second embodiment, the recombinant Factor VIII protein of any of embodiments 35-41 may e.g. comprise at least amino acid substitutions at positions N79, S112, L160,

L171, V184, N233, I265, N299, Y426, S507, F555, N616, L706, and Y748, wherein preferably the substitutions are N79S, S112T, L160S, L171Q, V184A, N233D, I265T,

N299D, Y426H, S507E, F555H, N616E, L706N, and Y748S. In a forty-third embodiment, the protein of embodiment 42 further includes K1837E. Optionally, the protein comprises the amino acid sequence according to aa 20-1533 of SEQ ID NO: 119.

In an forty-fourth embodiment, the recombinant Factor VIII protein of any of embodiments 35- 42 may e.g. comprise at least amino acid substitutions at positions N79, S112, L160, L171, V184, N233, I265, N299, Y426, S507, F555, N616, L706, Y748, N2038, S2077, S2315 and V2333, wherein preferably the 18 substitutions are N79S, S112T, L160S, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, F555H, N616E, L706N, Y748S, N2038D, S2077G, S2315T and V2333A. In a forty-fifth embodiment, the protein of embodiment 44 comprises an amino acid sequence having at least 90%, preferably, 95% sequence identity to aa 20- 1533 of SEQ ID NO: 114.

In a 46^th embodiment, the recombinant Factor VIII protein of any of embodiments 35-38 comprises at least one, preferably, all of L160, F555 and S2315, i.e. , it does not comprise mutations at these positions. In a 47^th embodiment, the recombinant Factor VIII protein of any of embodiments 35-38 and 46 may e.g. comprise at least amino acid substitutions at positions N79, S112, L171 , V184, N233, I265, N299, Y426, S507, N616, L706, and Y748, wherein preferably the substitutions are N79S, S112T, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, N616E, L706N, and Y748S. In a 48^th embodiment, the protein of embodiment 47 further includes K1837E.

In an 49^th embodiment, the recombinant Factor VIII protein of any of embodiments 35-38 or 46-48 may e.g. comprise at least amino acid substitutions at positions N79, S112, L171, V184, N233, I265, N299, Y426, S507, N616, L706, Y748, N2038, S2077, and V2333, wherein preferably the 15 substitutions are N79S, S112T, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, N616E, L706N, Y748S, N2038D, S2077G, and V2333A. In a 50^th embodiment, the protein of embodiment 49 comprises an amino acid sequence having at least 90%, preferably, 95% sequence identity to aa 20-1533 of SEQ ID NO: 134 (FVIII- 16M_SC). In a 51^st embodiment, the protein of embodiment 49 comprises an amino acid sequence having at SEQ ID NO: 114 except for positions L160, F555 and S2315 (AD2CD2- 19M_SC -16M).

In a 52^nd embodiment, the recombinant Factor VIII protein of any of embodiments 35-51 may e.g. comprise at least the amino acid substitution at position K1837, wherein preferably said substitution is K1837E. In a 53^rd embodiment, the protein of embodiment 52 comprises the amino acid sequence according to aa 20-1533 of SEQ ID NO: 114.

In a 54^th embodiment, the recombinant Factor VIII protein of any of embodiments 35-52 may e.g. have a reduced immunogenicity compared to a Factor VIII protein consisting of SEQ ID NO: 60 and preferably also compared to a Factor VIII protein consisting of SEQ ID NO: 61

In a 55^th embodiment, in the recombinant Factor VIII protein of embodiment 54, said immunogenicity is determined by an immunogenicity score or an assay comprising co cultivating dendritic cells incubated with said protein and regulatory T-cell-depleted CD4⁺ T cells of a donor and testing activation of said T cells, preferably, by said assay. In a 56^th embodiment, the recombinant Factor VIII protein of any of embodiments 35-55 may e.g. have at least 90 % sequence identity to a Factor VIII protein of SEQ ID NO: 63, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity. It may also be a fusion protein of said recombinant Factor VIII protein.

In a 57^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-56 may e.g. be a single chain Factor VIII protein. In a 58^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-57 may e.g. be a heterodimeric Factor VIII protein.

In a 59^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-57 may e.g. be, a single chain B-domain deleted Factor VIII protein. In a 60^th embodiment, e recombinant Factor VIII protein of embodiment 59 comprises a processing sequence of SEQ ID NO: 2.

In a 61^st embodiment, the recombinant Factor VIII protein of embodiment 59 comprises a processing sequence of SEQ ID NO: 5, wherein X preferably is A or S. In a 62^nd embodiment, the recombinant Factor VIII protein of embodiment 61 comprises a processing sequence of SEQ ID NO: 5, wherein X is A. In a 63^rd embodiment, the recombinant Factor VIII protein of embodiment 61 comprises a processing sequence of SEQ ID NO: 5, wherein X is S. In a 64^th embodiment, the recombinant Factor VIII protein of embodiment 59 comprises a processing sequence of SEQ ID NO: 132. In a 65^th embodiment, the recombinant Factor VIII protein of embodiment 59 comprises a processing sequence of SEQ ID NO: 2, and further comprises a sequence of SEQ ID NO: 131 overlapping with said processing sequence. Embodiments 61 to 65 avoid the generation of potential T cell epitopes that may be introduced by the processing sequence of SEQ ID NO: 2.

In a 66^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-65 may e.g. be a fusion protein, wherein the fusion partner is selected from the group comprising an Fc region, albumin, PAS polypeptides, HAP polypeptides, the C-terminal peptide of the beta subunit of chorionic gonadotropin, albumin-binding small molecules, polyethylenglycol, hydroxyethyl starch, and combinations thereof.

The recombinant Factor VIII protein may also be de-immunized in the junctions generated by fusion between the FVIII, linker and ABD sequences Thus, in a 67^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-66 may comprise one or more, preferably all of substitutions F761G, F779G, F1632G, F858G, F1711G and F936G. In a 68^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-67 may comprise one or more, preferably all of substitutions P766Q and N772D. In a 69^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-66 or 68 may comprise one or more, preferably all of substitutions R784Q, S787G, R1637Q, S1640G, R836Q, S866G, R1716Q, S1719G, R941Q and S944G. In a 70^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-69 may comprise the substitution S926G. In a 71^st embodiment, the recombinant Factor VIII protein of any of embodiments 1-70 may comprise the substitution N1625D. In a 72^nd embodiment, the recombinant Factor VIII protein of any of embodiments 1-70 may comprise the substitution N1625Y. In a 73^rd embodiment, the recombinant Factor VIII protein of any of embodiments 1-67 may comprise all of substitutions F761G, F779G, F1632G, F858G, F1711G, S926G and F936G. In a 74^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-67 or 73 may comprise all of substitutions F761G, F779G, F1632G, F858G, F1711G, S926G, F936G and N1625D. In a 75^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-66 or 68-69 may comprise all of substitutions P766Q, N772D, R784Q, S787G, R1637Q, S1640G,

R863Q, S866G, R1716Q, S1719G, R941Q and S944G. In a 76^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-66 or 68-69 or 75 may comprise all of substitutions P766Q, N772D, R784Q, S787G, R1637Q, S1640G, R863Q, S866G, R17167Q, S1719G, S926G, R941Q and S944G. In a 77^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-66 or 68-69 or 75-76 may comprise at least one, preferably all of substitutions P766Q, N772D, R784Q, S787G, R1637Q, S1640G, R863Q, S866G, R1716Q, S1719G, S926G, R941Q, S944G and N1625D. In a 78^th embodiment, the recombinant Factor VIII protein of any of embodiments 1-66 or 68-69 or 75-76 may comprise at least one, preferably all of substitutions P766Q, N772D, R784Q, S787G, R1637Q, S1640G, R863Q, S866G, R1716Q, S1719G, S926G, R941Q, S944G and N1625Y.

In embodiments 67-78, the substitutions named relate to the positions relative to SEQ ID NO: 48 or 114 (SEQ ID NO: 48 and 114 have the same number of positions, so that is interchangeable). While reference to these sequences for the positions does not necessarily mean that the substitutions need to be introduced FVIII proteins of these sequences, in embodiment 79, the substitutions named in any of embodiments 67-78 are introduced into a FVIII protein of SEQ ID NO: 48. In embodiment 80, the substitutions named in any of embodiments 67-78 are introduced into a FVIII protein of SEQ ID NO: 114. However, the respective substitutions can also be introduced into other AD2CD2_SC proteins. For example, in embodiment 81, the substitutions named in any of embodiments 67-78 are introduced into AD2CD2-16M_SC having to SEQ ID NO: 114, but with L160, F555 and S2315 (these three positions referring to positions relative to SEQ ID NO: 1). In an 82^nd embodiment, the invention provides a nucleic acid encoding a recombinant Factor VIII protein of any one of embodiments 1-81. In an 83^rd embodiment, the nucleic acid of embodiment 82 is an expression vector, preferably, suitable for expression of said recombinant Factor VIII protein in a mammalian cell, preferably, in a human cell, such as a CAP cell.

In an 84^th embodiment, the invention provides a host cell comprising a nucleic acid of any of embodiments 82 or 83. In a 85^th embodiment, the host cell of embodiment 84 is a mammalian cell comprising an expression vector suitable for expression of said recombinant Factor VIII protein in said cell, preferably, a human cell selected from the group comprising a Hek293 cell or a CAP cell (e.g., CAP-T cell or CAP-Go cell).

In an 86^th embodiment, the invention provides a method of preparing a recombinant Factor VIII protein, comprising culturing the host cell of embodiments 84 or 85 under conditions suitable for expression of the Factor VIII protein and isolating the recombinant Factor VIII protein, wherein the method optionally comprises formulating the Factor VIII protein as a pharmaceutical composition.

In an 87^th embodiment, the invention provides a pharmaceutical composition comprising the recombinant Factor VIII protein of any of embodiments 1-81, the nucleic acid of any of embodiments 82-83 or the host cell of any of embodiments 84-85. In an 88^th embodiment, the pharmaceutical composition of embodiment 87 further comprises a biologically acceptable carrier such as water or a buffer, optionally, at a physiologic pH, preferably, FVIII formulation buffer, and/or pharmaceutically acceptable excipients. In an 89^th embodiment, the pharmaceutical composition of any of embodiments 87 or 88 further comprises albumin, e.g., 0.1-5%, such as 1% human serum albumin.

In a 90^th embodiment, the invention provides a pharmaceutical composition of any of embodiments 87-89 or a kit further comprising an immunosuppressive agent, e.g., an immunosuppressive agent selected from the group comprising methylprednisolone, prednisolone, dexamethason, cyclophosphamide, rituximab, and/or cyclosporin.

The invention further provides, as a 91^st embodiment, a pharmaceutical composition of any of embodiments 87-90 for use in treatment of hemophilia A, wherein, optionally, the treatment is immune tolerance induction (ITI). In a 92^nd embodiment, the pharmaceutical composition of any of embodiments 87-91 is for use in treating a patient with Hemophilia A selected from the group comprising a patient not previously treated with any Factor VIII protein, a patient previously treated with a Factor VIII protein, a patient who has an antibody response including an inhibitory antibody response to a Factor VIII protein, and a patient who has had an antibody response including an inhibitory antibody response to a Factor VIII protein who has been treated by ITI, or who has not been treated by ITI.

In a 93^rd embodiment, the pharmaceutical composition of any of embodiments 87-92 is for administration every 5 to 14 days, preferably, every 7 to 10 days.

In a 94^th embodiment, the invention provides a vial, e.g., a prefilled or ready-to use syringe, comprising the pharmaceutical composition of any of embodiments 87-93.

In a 95^th embodiment, the invention provides a method of treatment, comprising administering an effective amount of the pharmaceutical composition of any of embodiments 87-93 to a patient in need thereof, e.g., a patient with hemophilia A, which may be selected from the patient groups defined herein.

Proteins and nucleic acids, e.g., having the following sequences are disclosed herein:

SEQ ID NO: 1 wt human FVIII SEQ ID NO: 2 processing sequence in preferred single chain constructs SEQ ID NO: 4 processing sequence in VO SEQ ID NO: 5 variant processing sequence, X can be varied for de-immunization SEQ ID NO: 6 variant processing sequence, X can be varied for de-immunization SEQ ID NO: 7 variant processing sequence, X can be varied for de-immunization SEQ ID NO: 8 variant processing sequence, X can be varied for de-immunization SEQ ID NO: 9 merging sequence, e.g., in VO SEQ ID NO: 16 single chain FVIII VO (AC_SC) SEQ ID NO: 28 6rs-REF SEQ ID NO: 39 Thrombin-cleavable linker SEQ ID NO: 40 glycine-serine linker G1, G in position 14 is present or absent SEQ ID NO: 41 glycine-serine linker G2 SEQ ID NO: 42 Thrombin-cleavable linker flanked on each side by a glycine-serine linker, strictly repetitive

SEQ ID NO: 43 Thrombin-cleavable linker flanked on each side by a glycine-serine linker, non-repetitive

SEQ ID NO: 44 ABD consensus sequence, see above SEQ ID NO: 45 ABD1 SEQ ID NO: 46 ABD2 SEQ ID NO: 47 ADLCLD SC aa SEQ ID NO: 48 AD2CD2_SC aa SEQ ID NO: 49 AD2CD2woL_SC aa SEQ ID NO: 50 AD2CD2woLG_SC aa SEQ ID NO: 51 AbD2CD2_SC aa SEQ ID NO: 52 ADLCLD_SC na SEQ ID NO: 53 AD2CD2_SC na SEQ ID NO: 54 AD2CD2woL_SC na SEQ ID NO: 55 AD2CD2woLG_SC na SEQ ID NO: 56 AbD2CD2_SC na SEQ ID NO: 57 optimized DNA sequence encoding SEQ ID NO: 46 SEQ ID NO: 58 exemplary DNA encoding Glycine-serine linker G1 of SEQ ID NO: 40 SEQ ID NO: 59 exemplary DNA encoding Glycine-serine linker G2 of SEQ ID NO: 41 SEQ ID NO: 60 FVII l-6rs SEQ ID NO: 61 ReFacto AF SEQ ID NO: 62 B-domain deleted scFVIII SEQ ID NO: 63 FVIII-19M SEQ ID NO: 64 FVIII-18M SEQ ID NO: 65 FVIII-15M SEQ ID NO: 66 FVIII-A1-7M SEQ ID NO: 67 FVIII-A2-4M SEQ ID NO: 68 FVIII-BA3-1M SEQ ID NO: 69 FVIII-A3C2-4M SEQ ID NO: 70 FVIII-GOF1 SEQ ID NO: 71 FVIII-GOF2 SEQ ID NO: 72 FVIII-LS1 SEQ ID NO: 73 FVIII-LS2 SEQ ID NO: 74-108+112 immunogenic clusters SEQ ID NO: 109 FVIII-A1A2-3M SEQ ID NO: 110 provides a nucleic acid sequence encoding FVII 1-19M. SEQ ID NO: 111 provides a nucleic acid sequence encoding FVI I l-6rs. SEQ ID NO: 113 Single Chain V0-19M (AC-19M_SC) aa SEQ ID NO: 114 AD2CD2-19M_SC aa SEQ ID NO: 115 ALDLCLD-19M_SC aa SEQ ID NO: 116 ADLCLD-19M_SC-V1 aa SEQ ID NO: 117 ADLCLD-19M_SC-V2 aa SEQ ID NO: 118 AD2CD-19M_SC aa SEQ ID NO: 119 AD2CD2-15M SC aa SEQ ID NO: 120 Single Chain V0-19M (AC-19M_SC) na SEQ ID NO: 121 AD2CD2-19M_SC na SEQ ID NO: 122 ALDLCLD-19M_SC na SEQ ID NO: 123 ADLCLD-19M_SC-V1 na SEQ ID NO: 124 ADLCLD-19M_SC-V2 na SEQ ID NO: 125 AD2CD-19M_SC na SEQ ID NO: 126 AD2CD2-15M_SC na SEQ ID NO: 127 sequence F761-S769, overlapping with sc processing sequence SEQ ID NO: 2:

SEQ ID NO: 128 de-immunized variant of SEQ ID NO: 127 with F761A SEQ ID NO: 129 de-immunized variant of SEQ ID NO: 127 with F761S SEQ ID NO: 130 de-immunized variant of SEQ ID NO: 127 with P766E SEQ ID NO: 131 de-immunized variant of SEQ ID NO: 127 with S769D SEQ ID NO: 132 de-immunized processing sequence with P766E SEQ ID NO: 133 FVIII-16M SEQ ID NO: 134 Single Chain V0-16M (AC-16M_SC) SEQ ID NO: 135 F01_ AD2CD2-19M_SC SEQ ID NO: 136 F02_ AD2CD2-19M_SC SEQ ID NO: 137 F03_ AD2CD2-19M_SC SEQ ID NO: 138 F04_ AD2CD2-19M_SC SEQ ID NO: 139 F01_ AD2CD2-16M_SC SEQ ID NO: 140 F02_ AD2CD2-16M_SC SEQ ID NO: 141 F01_ AD2CD2_SC SEQ ID NO: 142 F02_ AD2CD2_SC SEQ ID NO: 143 F03_ AD2CD2_SC SEQ ID NO: 144 F04_ AD2CD2_SC SEQ ID NO: 145 D01_AD2CD2_SC SEQ ID NO: 146 D02_AD2CD2_SC SEQ ID NO: 147 D03_AD2CD2_SC SEQ ID NO: 148 D04_AD2CD2_SC SEQ ID NO: 149 D05_AD2CD2_SC SEQ ID NO: 150 D06_AD2CD2_SC SEQ ID NO: 151 D07_AD2CD2_SC SEQ ID NO: 152 D08_AD2CD2_SC SEQ ID NO: 153 D09_AD2CD2_SC SEQ ID NO: 154 D10_AD2CD2_SC SEQ ID NO: 155 D11 AD2CD2 SC SEQ ID NO: 156 E01_AD2CD2-19M_SC SEQ ID NO: 157 E02_AD2CD2-19M_SC SEQ ID NO: 158 E03_AD2CD2-19M_SC SEQ ID NO: 159 E04_AD2CD2-19M_SC SEQ ID NO: 160 E05_AD2CD2-19M_SC SEQ ID NO: 161 E06_AD2CD2-19M_SC SEQ ID NO: 162 E07_AD2CD2-19M_SC SEQ ID NO: 163 E08_AD2CD2-19M_SC SEQ ID NO: 164 E09_AD2CD2-19M_SC SEQ ID NO: 165 E10_AD2CD2- 19M_SC SEQ ID NO: 166 E11_AD2CD2-19M_SC SEQ ID NO: 167 G08_AD2CD2-16M_SC SEQ ID NO: 168 G11 AD2CD2-16M SC

Fiqure Leqends

Fig. 1 shows the human albumin binding of ADLCLD_SC, a FVIII protein of the invention comprising two albumin-binding domains in comparison to FVIII 6rs-Ref, a protein having the ReFacto AF sequence. Both FVIII proteins were tested in the presence and absence of HSA via the albumin binding capacity assay as described.

Fig. 2 shows the von-Willebrand factor (vWF) binding capacity of different FVIII-albumin- binding-domain fusion proteins in relation of ReFacto AF. All FVIII molecules were tested for their vWF binding in either the presence or absence of human albumin. The more albumin binding domains were incorporated into FVIII, the lower was the binding to vWF in general. The presence of human albumin dramatically decreased the binding to vWF.

Fig. 3 Comparison of unpurified FVIII-ABD fusion variants and FVIII controls for their in vitro functionality. Cell culture supernatants of CAP-T cells expressing the double chain FVIII molecule 6rs-REF, the single chain FVIII molecule AC_SC, and the FVIII-ABD fusion molecules AD2CD2_SC, AD2CD2woLG_SC, AD2CD2wL_SC, ACD4woLG_SC, and ACL(GD)4_SC were analyzed for chromogenic FVIII activity (A), FVIII clotting activity induced by Actin FSL (B) and FVIII antigen levels indicating total FVIII protein amount (C). Specific chromogenic activity was calculated as chromogenic FVIII activity to FVIII antigen ratio displayed in % (D). Specific clotting activity was calculated as FVIII clotting activity to FVIII antigen ratio displayed in % (E). n=2. Fig. 4 Western blot analysis of unpurified FVIII-ABD fusion variants and FVIII control proteins for assessing structural properties. Cell culture supernatants of CAP-T cells expressing the double chain FVIII molecule 6rs-REF, the single chain FVIII molecule AC_SC, and the FVIII- ABD fusion molecules AD2CD2_SC, AD2CD2woLG_SC, AD2CD2wL_SC, and ACD4woLG_SC were separated by non-reduced sodium dodecyl sulfate polyacrylamide gel electrophoresis and subsequent blotting onto a PVDF membrane was performed. A purified Sheep anti-Human Factor VIII primary antibody and CF680-conjugated donkey anti-sheep IgG (H&L) antibody was used for detection. For size determination, Precision Plus All Blue was applied as marker.

Fig. 5 demonstrates the in vivo pharmacokinetics of AD2CD2_SC compared to ReFacto AF after a single injection of 200 U/kg FVIII (with 1% human albumin) into mice having a knock out for murine albumin and expressing the a-chain of human instead of murine neonatal Fc- receptor. Determined FVIII antigen values were normalized and are shown in percent over time.

Fig. 6 demonstrates the in vivo pharmacokinetics of AD2CD2_SC compared to ReFacto AF after a single injection of 30 U FVIII Antigen/kg formulated with either 1 or 10% human albumin into Gottingen minipigs. Plasma samples were withdrawn and FVIII antigen levels were measured by ELISA. Mean FVIII antigen levels are shown in U/ml over time in hours. n=3 minipigs per group.

Fig. 7 shows the total bleeding time (first column of each group, left Y-axis) and the total blood loss (second column of each group, right Y-axis) after tail transection of hemophilia A mice which were administered 20 h earlier with either Vehicle Control, ReFacto AF, Eloctate, AD2CD2_SC or ADLCLD_SC. Non-hemophilia C57BL/6NCrl mice were treated with 0.9% NaCI and used as control.

Fig. 8: (A) Structure of FVIII protein and generation of the FVIII-19M protein of the invention in several rounds of selection. (B) FVIII-19M amino acid sequence including the signal sequence. Signal sequence: italics, A1 domain: underline, A2 domain: double underline, B domain: fat underline, A3 domain: dotted underline, C1 domain: dashed underline, C2 domain: wavy underline, intermediate domains a1, a2, a3: not marked, mutations versus FVI I l-6rs are marked by italics, fat and larger type.

Fig. 9: Relative coagulant activities of FVIII variants with single mutations. The FVIII coagulant activity of each single-mutation variant was calculated in relation to the FVIII coagulant activity of the control FVI I l-6rs. The brackets indicate mutations, which belong to one cluster. (A) Mutations in the A1 domain. (B) Mutations in the A2 domain. (C) Mutations in the A3 domain. (D) Mutations in the C1 domain. (E) Mutations in the C2 domain.

Fig. 10: Specific coagulant activities of FVIII variants with single mutations. The relation of FVIII coagulant activity to FVIII antigen was calculated for each single-mutation variant. The brackets indicate mutations, which belong to one cluster. (A) Mutations in the A1 domain. (B) Mutations in the A2 domain. (C) Mutations in the A3 domain. (D) Mutations in the C1 domain. (E) Mutations in the C2 domain.

Fig. 11: Results of the combined mutations in section A1, A1A2, A2 and A3C1C2. (A)

Relative coagulant activities of the FVIII variants. The FVIII coagulant activity of each variant was calculated in relation to the FVIII coagulant activity of the control FVI I l-6rs. (B) Specific coagulant activities of the FVIII variants. The relation of FVIII coagulant activity to FVIII antigen was calculated for each variant.

Fig. 12: Relative coagulant activities of FVIII variants comprising different mutations based on a DOE matrix. The FVIII coagulant activity of each variant was calculated in relation to the FVIII coagulant activity of the control FVI I l-6rs. (A) Results for the variants in the A2 domain. (B) Results for the variants in the A3C1C2 domain.

Fig. 13: Coagulant activities of FVIII with combined mutations in the sections A2 and A3C1C2 after the DOE matrix. (A) Relative coagulant activities of the FVIII variants. The FVIII coagulant activity of each variant was calculated in relation to the FVIII coagulant activity of the control FVIII-6rs. (B) Specific coagulant activities of the FVIII variants. The ratio of chromogenic FVIII coagulant activity to FVIII antigen was calculated for each variant.

Fig. 14: Relative and specific coagulant activities of FVIII variants with specific mutations (A, B). Relative coagulant activities are defined in comparison to coagulant activity of FVI I l-6rs. Specific coagulant activity relates to the ratio of chromogenic coagulant activity to antigen. Coagulant activities of advantageous FVIII proteins having mutations in specific domains of FVIII (A) and FVIII proteins having three mutations (B). Clotting coagulant activity of FVI II- BA3-1M was not determined.

Fig. 15: ROTEM analysis of FVIII-19M, FVIII-6rs, ReFacto AF and Nuwiq analyzing clotting time. Different FVIII concentrations were analyzed. The measurements were performed in duplicates and the mean values are displayed.

Fig. 16: Results of the TGA (Thrombin generation assay) for ReFacto AF, Nuwiq, FVIII-19M and FVI I l-6rs. All products were diluted to 0.25 U/ml, 0.063 U/ml and 0.016 U/ml FVIII coagulant activity. Each point indicates the results from one TGA. The line indicates the median of the four performed assays. Statistical analysis was performed using the Friedman test. (A) Amount of generated peak thrombin for each product at the given concentration based on a thrombin standard. (B) Area under the curve for each product at the given concentration. (C) Time to peak thrombin generation for each product at the given concentration.

Fig. 17: Binding of the different FVIII products to vWF. The potency of ReFacto AF binding to vWF was set to 1 and was the reference for the other products. Each point indicates the results from one ELISA. The line indicates the median of the three performed assays. Statistical analysis was performed using the Friedman test.

Fig. 18: Specific coagulant activities of four independent productions of FVI I l-6rs and FVIII- 19M based on purified protein. (A) Specific coagulant activities based on the chromogenic FVIII coagulant activity measurement. The line indicates the median of the four measurements. (B) Specific coagulant activities based on the clotting FVIII coagulant activity measurement. The line indicates the median of the four measurements.

Fig. 19: Western Blot of FVIII activated by thrombin. Each product was applied in its non- activated and activated form. In the non-activated form, the typical bands for the single chain (« 200 kDa), heavy chain (« 95-110 kDa) and light chain (« 80-90 kDa) were detectable. After thrombin cleavage additional bands for A1A2 (« 90 kDa), A3C1C2 (« 75 kDa), A1 (~ 50 kDa), A2 (~ 40 kDa) and Ba3 (~ 20 kDa) were detectable. FVIII was detected with the primary polyclonal sheep anti-human Factor VIII antibody and the secondary donkey anti-sheep IgG IRDye 800CW.

Fig. 20: In vitro immunogenicity assay. Monocytes are purified from PBMCs, differentiated to iDCs and finally stimulated to become mDCs and incubated with the antigen, i.e. , the protein of interest. CD4+CD25-T cells are also purified from PBMCs and cultivated prior to co cultivation, in order to regenerate. After co-culture, the T cells are analyzed for activation and/or proliferation by flow cytometry. Optionally, the supernatant is analyzed for cytokines.

Fig. 21: Results of the in vitro immunogenicity assay. Difference between the CD4+T cell proliferation against DCs stimulated with IL-Mix plus FVIII-19M and DCs stimulated with IL- Mix plus FVIII-6rs. The bars below 0 indicate a reduced CD4+T cell response to FVIII-19M. The lower T cell response to FVIII-19M compared to FVIII-6rs is significant using the Wilcoxon test (p=0.0371). Fig. 22: Comparison of an unpurified FVIII-ABD fusion variant with (AD2CD2-19M_SC) or without (AD2CD2_SC) 19 de-immunizing amino acid substitutions with a FVIII control in terms of protein expression and in vitro functionality. Cell culture supernatants of CAP-T cells expressing the double chain FVIII molecule 6rs-REF (ReFacto sequence), the FVIII-ABD fusion molecule AD2CD2_SC, and the de-immunized FVIII-ABD fusion molecule AD2CD2- 19M_SC were analyzed for chromogenic FVIII activity (A), and FVIII antigen levels indicating total FVIII protein amount (B). Specific chromogenic activity was calculated as chromogenic FVIII activity to FVIII antigen ratio displayed in % (C). n=2.

Fig. 23 demonstrates the in vivo pharmacokinetics of AD2CD2-19M_SC compared to ReFacto AF after a single intravenous injection of 200 U/kg FVIII into hemophilia A mice.

FVIII antigen values and chromogenic FVIII activity were determined. FVIII antigen values are shown over time. The protein of the invention clearly has a longer half-life in vivo.

Fig. 24 demonstrates the in vivo pharmacokinetics of AD2CD2_SC and AD2CD2-19M_SC compared to ReFacto AF after a single intravenous injection of 30 U FVIII:Ag/kg formulated with 10% human albumin into Gottingen minipigs. Plasma samples were withdrawn and FVIII antigen levels were measured. Mean FVIII antigen levels are shown in U/mL over time in hours. n=3 minipigs per group. The proteins of the invention clearly have a longer half-life in vivo.

Fig. 25 shows the total bleeding time after tail vein transection of hemophilia A mice which were administered 30 min earlier with either Vehicle Control (group 6) or different doses (groups 1 to 5) of AD2CD2-19M_SC (200, 70, 20, 7 or 2 U/kq FVIII) intravenously. In addition, non-hemophilia C57BL/6NCrl mice were treated with Vehicle Control (group 7). N=10 mice per group.

Fig. 26 shows the inhibitory potential of five anti-FVIII antibodies (ESH-8, GMA-8009, GMA- 8015, GMA-8026, CL20035AP) against standard human plasma (SHP), ReFacto AF, AD2CD2_SC, and AD2CD2-19M_SC.

Fig. 27 shows the number of immunogenic clusters identified via MAPPs technology in different FVIII variants, namely, ReFacto AF, AC-19M_SC and AD2CD2-19M_SC. The overall number of clusters is reduced in both AC-19M_SC and AD2CD2-19M_SC compared to ReFacto AF (total No. of clusters: columns 1-3). No. of clusters with a reduced frequency (FRQ) in the donor cohort compared to ReFacto AF are demonstrated in columns 4 and 5, including number of eliminated clusters (crosshatched part of the columns). No of clusters with an increased FRQ compared to ReFacto AF are shown in columns 6 and 7, including clusters not observed for ReFacto AF shown in crosshatched parts of the columns.

Fig. 28 shows the chromogenic (A, C) and the specific chromogenic (B, D) FVIII activity of specific junction deimmunized FVIII proteins of the invention.

Fig. 29 shows the total bleeding time in haemophilia A mice after administration of different amounts of specific junction deimmunized FVIII proteins of the invention.

Fig. 30 shows the normalized chromogenic FVIII activity (A) and normalized FVIII antigen levels (B) of specific FVIII proteins of the invention.

Examples

1. Evaluation of Factor VIII proteins with additions of albumin-binding domains for the development of a new hemophilia A therapeutic

Material and methods

Preparation of constructs

Experiments were performed to find and develop a suitable backbone for integration of the albumin-binding domains. The experiments were done on the basis of a B-domain deleted version of FVIII and single chain variants of FVIII. The basic double chain construct was a codon-optimized sequence of ReFacto AF^® (Pfizer), wherein for simplifying cloning, 6 restriction sites were added through silent mutations, but some of these restriction sites were again excluded due to codon-optimization. The basic double chain sequence is 6rs-REF (SEQ ID NO: 28). The basic single chain construct used was VO (SEQ ID NO: 16,

EP19173440)

Firstly, the ABD protein sequence (Affibody AB, Solna, Sweden) was taken as a basis for design of the DNA sequence. If not mentioned otherwise, the ABD2 sequence was used. Moreover, codon optimized linkers were developed, which are partly cleavable by thrombin.

If not otherwise stated, the glycine-serine linker was G1 and the thrombin-cleavable linker was L. Table 1 below demonstrates structures of fusion proteins with albumin-binding domains (ABD) for single chain molecules.

For the constructs encoding the FVIII of the invention and comparative constructs also analysed in this context, either the complete FVIII sequence or DNA regions carrying approx. 700-1200 bp from the FVIII a2 domain to the A3 domain were synthesized. The synthesized DNA was codon-optimized for the total target gene. The a2 to A3 DNA fragments were 5' terminally flanked by an EcoRV restrictions site, and 3' terminally flanked by an EcoRI restriction site, and these restriction sites were also present in the basic FVIII sequence used. For C-terminal fusion to the light chain, DNA fragments carrying approx. 1500-2100 bp were synthesized also in a codon-optimized form. Such DNA fragments were 5' terminally flanked by an EcoRI restrictions site within the A3 domain, and 3' terminally flanked by an Notl restriction site. Restriction of the DNA inserts and the FVIII backbone plasmid allowed for targeted ligation and generation of FVIII single chain plasmids. Completely synthesized FVIII DNA was 5' terminally flanked by a Hindlll restrictions site, and 3' terminally flanked by a Notl restriction site.

By transformation of E.coli K12 with said plasmids, expansion of transformed bacteria under ampicillin selection and plasmid preparation, large amounts of the plasmids could be prepared. Genetic engineering work was carried out by Thermo Fisher Scientific after design with VectorNTI Software (Thermo Fisher Scientific, Massachusetts, USA).

Cultivation of CAP-T cells

For analyzing the candidates for new recombinant FVIII molecules, the constructs, integrated in expression vectors, were transiently and stably expressed in human cell lines. The preferred cell lines are Hek293 and CAP cells, both of which originate from human amniocytes. Because of higher yields of active FVIII molecules CAP cells, in particular, CAP- T cells were chosen as the preferred expression system for transient transfection and CAP- Go cells for stable expression.

Transient transfection was performed with nucleofection programs. The supernatants were screened for FVIII activity and antigen. Purification of the recombinant proteins from CAP cells was done, including FVIII affinity chromatography.

In detail, CAP-T cells (Cevec Pharmaceuticals, Koln, Germany) were cultured in PEM medium supplemented with 4 mM GlutaMAX (Thermo Fisher Scientific, 35050038) and 5 pg/ml blasticidin (Thermo Fisher Scientific, R21001; complete PEM medium). In order to thaw the cells, the required amount of frozen vials were transferred to a 37 °C water bath. After thawing, each vial was transferred to 10 ml of chilled, complete PEM medium. The cell suspension was centrifuged at 150 x g for 5 minutes. During this washing step the dimethyl sulfoxide (DMSO) used for cryopreservation was removed. The pellet was resuspended in 15 ml warm, complete PEM medium and transferred to a 125 ml shaker flask. The cells were incubated at 37 °C in a humidified incubator with an atmosphere containing 5 % CO2. The flasks were set on a shaking platform, rotating at 185 rpm with an orbit of 50 mm. Subculturing of the cells was performed every 3 to 4 days. The fresh culture was set to 0.5x10⁶ cells/ml by transferring the required amount of cultured cell suspension to a new flask and adding complete PEM medium. In the case that the transferred cell suspension would exceed 20% of the total volume, the suspension was centrifuged at 150 x g for 5 minutes and the pellet was resuspended in fresh complete PEM medium. The volume of cell suspension per shaking flask was 20% of the total flask volume.

A minimum of three subcultures were performed after thawing before transfection experiments were performed.

Protein expression in CAP-T cells by transient transfection

The CAP-T cells were transfected using the 4D-Nucleofector™ (Lonza, Basel, Switzerland). For each transfection 10x10⁶ CAP-T cells were centrifuged at 150 x g for 5 minutes in 15 ml conical tubes. The cells were resuspended in 95 pi supplemented SE Buffer, taking into account the volume of the pellet and the volume of the plasmid solution. Afterwards, 5 pg of the respective plasmid were added to the cell suspension followed by gentle mixing. The solution was transferred to 100 pi Nucleocuvettes. The used transfection program was ED- 100. After the transfection, the cells from one Nucleocuvette were transferred to 125 ml shaker flasks, containing 12.5 ml complete PEM medium. The cells were cultivated for 4 days as described above. At day 4 the cells were harvested by centrifugation at 150 x g for 5 minutes. Larger protein amounts could be produced by combining 12.5 ml approaches as described above.

Supernatants were screened for FVIII activity and antigen directly after harvest.

The recombinant Factor VIII protein was further analyzed. FVIII activity was measured by chromogenic activity assay and clotting activity FSL assay. The antigen was estimated by FVIII antigen ELISA. As a further assay for biological activity, the cleavage of the recombinant proteins by thrombin was analyzed. Moreover, chain distribution and appearance was tested by Western Blots. Further, vWF-binding and albumin binding were tested.

Protein expression in CAP Go cells by stable cell pools

In order to produce large material amounts to conduct minipig studies, stable CAP-Go pools expressing either AD2CD2_SC or AD2CD2-19M_SC were generated at Cevec Pharmaceuticals GmbH (Cologne, Germany). Therefore, the FVIII coding sequences were cloned into CEVEC’s pStbl-bsd-MCS(-) plasmid using SgrD1 and Not1 restriction sites. The fragment was subsequently separated by agarose gel electrophoresis, purified by gel filtration and cloned into the pStbl-bsd-MCS(-), which was previously cut with SgrD1 and Not1 and treated with Calf-Intestine-Phosphatase (CIP). Inserts and vector were ligated using T4-DNA ligase and transformed into chemically competent E. coli cells (XL2-Blue). The plasmid DNA was purified using the Maxi Kit from Machery-Nagel. The whole cloning processes as well as the plasmid purifications were performed in a TSE-free production process.

Prior to nucleofection, circular plasmids were linearized with Seal. Therefore, 20-40 pg plasmid DNA were incubated 5-8 h with 50-200 U of the respective enzyme at 37 °C. Subsequently, the DNA was purified by phenol-chloroform-isoamyl alcohol extraction and phenol was washed away with chloroform-isoamyl alcohol. To purify the DNA by ethanol precipitation, the DNA solution was supplemented with 1/10 volume of 3M NaOAc, pH = 5.2 and 2 volumes of ethanol and incubated overnight at -20 °C. The DNA precipitate was pelleted by centrifugation (30 min, 13 000 rpm, 4 °C), washed with 70 % ethanol, centrifuged again, air-dried, and resuspended in TE buffer. The quality of the linearized DNA was assured by a DNA agarose gel analysis.

For nucleofection, CAP-Go cells were counted by Cedex XS (Roche Applied Science, Innovatis) and viable cell density and viability were determined. For each nucleofection reaction 1 x 10⁷ cells were harvested by centrifugation (150 x g for 5 min). The cells were resuspended in 100 pl_ complete nucleofector solution V (Lonza) and mixed with 5 pg linearized plasmid of the respective construct. The DNA/cell suspension was transferred into a cuvette and the nucleofection was performed using the X001 program on the Nucleofector II (Lonza). After the pulse, cells were recovered by adding 500 pL prewarmed complete PEM medium (= supplemented with 4 mM L-alanyl-L-glutamine) to the cuvette and gently transferred into 11,5 mL complete PEM medium in a 125 mL shaking flask. The cuvette was washed once with 500 pL fresh medium to recover residual cells.

72 h post-nucleofection the cell number and cell viability of the transfected cells were determined. The cells were harvested by centrifugation and resuspended in 20 ml complete PEM medium containing 5 pg/ml basticidin as selection marker. The cells were cultured at 37°C, 5% CO2 at 185 rpm with 5 cm amplitude in a Kuhner shaking incubator. As soon as cells recovered from selection and could be expanded, cells from the stable pools were cryopreserved.

For batch production, the culture was inoculated at a viable cell density of 1 x 10⁶ cells/ml in 800 mL complete PEM medium in a 2 L shake flask. The cells were incubated at 185 rpm, 37°C, 5 % C02 in a Kuhner shaking incubator for 4 days. The cell supernatants containing FVIII were harvested by centrifugation and purified by affinity chromatography as described elsewhere in this document.

FVIII activity - chromogenic activity assay

The activity of FVIII was determined by a chromogenic assay. In this two-step assay, FIXa and FVIIIa activate FX in the first step. In the second step, the activated FX hydrolyses a chromogenic substrate, resulting in a color change, which can be measured at 405 nm. Due to the fact that calcium and phospholipids are present in optimal amounts and an excess of FIXa and FX is available, the activation rate of FX is only dependent on the amount of active FVIII in the sample.

The reagents for this chromogenic FVIII activity assay were taken from the Coatest® SP FVIII Kit. The kit contained phospholipids, calcium chloride (CaCh), trace amounts of thrombin, the substrate S-2765, a mixture of FIXa and FX and the thrombin inhibitor 1-2581. The inhibitor was added, in order to prevent hydrolysis of the substrate by thrombin, which was built during the reaction. All dilutions were performed in distilled water or Tris-BSA (TBSA) Buffer, containing 25 mM Tris, 150 mM sodium chloride (NaCI) and 1 % Bovine serum albumin (BSA), set to pH 7.4. Each sample was diluted at least 1:2 with FVIII-depleted plasma. Further dilutions were performed using the TBSA Buffer.

The assay was performed using the BCS XP (Siemens Healthcare, Erlangen, Germany), a fully automated hemostasis analyzer. All reagents including water, TBSA Buffer and the samples were inserted into the analyzer. For each sample the analyzer mixed 34 pi calcium chloride, 20 mI TBSA Buffer, 10 mI sample, 40 mI water, 11 mI phospholipids and 56 mI FlXa- FX-mixture. This mixture was incubated for 300 seconds. Afterwards, 50 mI of S-2765 + I- 2581 were added to the reaction. Upon addition of the substrate, the absorption at 405 nm was measured for 200 seconds.

In order to calculate the amount of active FVIII, the software of the analyzer evaluated the slope of the measured kinetic between 30 seconds and 190 seconds after starting the reaction. This result was correlated to a calibration curve, generated with a biological reference preparation (BRP) of FVIII. The activity of the BRP is indicated in lU/ml. However, lU/ml can be assumed equivalent to U/ml. The results were indicated as “% of normal”.

These results were converted to U/ml, as 100 % of normal FVIII activity are equivalent to 1 U FVIII activity per ml. Clotting Activity FSL

In addition to the two-stage chromogenic assay (see above), a one-stage clotting assay was also performed in order to determine the amount of active FVIII. During this assay, FVIII- depleted plasma, CaCL, the activator Actin FSL and the FVIII-containing sample are mixed in one step. The activator leads to the generation of FXIa, which activates FIX. FVIIIa, FIXa and FX built the tenase complex and FX becomes activated. Further activation of prothrombin and fibrinogen finally leads to the formation of a fibrin clot. The time needed to form the clot, the activated partial thromboplastin time (aPTT), is measured. The aPTT varies, depending on the amount of FVIII.

The clotting assay was performed using the BCS XP. TBSA Buffer, FVIII-depleted plasma, Actin FSL, CaCL and the sample were inserted into the analyzer. The sample was diluted at least 1:2 with FVIII-depleted plasma. Further dilutions were performed using the TBSA Buffer. For each sample the analyzer mixed 45 pi TBSA Buffer, 5 mI sample, 50 mI FVIII- depleted plasma and 50 mI Actin FSL. The reaction was started by the addition of 50 mI CaCL. The analyzer measured the time needed for clot formation.

In order to calculate the amount of active FVIII, the software of the analyzer evaluated a baseline extinction at 405 nm at the beginning of the reaction. All of the following extinction values, within a time of 200 seconds, were analysed regarding their difference to the baseline extinction. The first time point exceeding a defined threshold was determined as the clotting time. This result was correlated to a calibration curve, generated with a BRP of FVIII.

FVIII antigen ELISA

The amount of FVIII antigen was determined using the Asserachrom® VIILAg ELISA (Diagnostica Stago, Asnieres sur Seine Cedex, France). In this sandwich ELISA, the applied FVIII is bound by mouse monoclonal anti-human FVIII F(ab’)2 fragments, which are coated to the plate by the manufacturer. The detection of the bound FVIII occurs via mouse monoclonal anti-human FVIII antibodies, which are coupled to a peroxidase. In the case that FVIII is present, the peroxidase-coupled antibody binds to FVIII and can be detected by the addition of a tetramethylbenzidine (TMB) solution. TMB turns from a clear to a blue-green solution upon reaction with peroxidase. After a short time, this reaction is stopped by the addition of sulfuric acid (H2SO4), which turns the solution yellow. The amount of bound FVIII correlates with the intensity of the yellow color, which can be measured at 450 nm. The final amounts of FVIII are calculated using a calibration curve generated by the measurement of at least five serial dilutions of a calibrator with a known antigen concentration. The supplied calibrator and control were reconstituted with 500 mI of distilled water, 30 minutes before starting the ELISA. After this incubation time, the calibrator was diluted 1:10 in the supplied phosphate buffer. This represented the starting concentration. The calibrator was further serially diluted 1:2 up to a dilution of 1:64. As the concentration of the calibrator contained approximately 1 U/ml FVIII, depending on the batch, the starting concentration was equivalent to 0.1 U/ml FVIII whereas the last dilution contained approximately 0.0016 U/ml FVIII. The control was diluted 1:10 and 1:20 with the phosphate buffer. All samples were diluted with the phosphate buffer, depending on their previously determined activity (see above) with the aim to be in the middle of the calibration curve. After the dilution of FVIII samples, control and calibrator, 200 mI of each solution were applied per well in duplicates. In addition to that, two wells were filled with 200 mI of phosphate buffer as a blank control. The plate was incubated for 2 hours at room temperature covered with a film. During this time, the peroxidase-coupled anti-human FVIII antibodies were reconstituted with 8 ml phosphate buffer and incubated 30 minutes at room temperature. After the antigen immobilization, the wells were washed five times with the supplied washing solution, which was previously diluted 1:20 with distilled water. Immediately after the washing, 200 mI of the peroxidase-coupled anti-human FVIII antibodies were added to each well and incubated for 2 hours at room temperature covered by a film. Afterwards, the plate was washed five times as before. In order to reveal the amount of bound FVIII, 200 mI of TMB solution were added to each well and incubated for exactly 5 minutes at room temperature. This reaction was stopped by the addition of 50 mI 1 M H2SO4 to each well. After an incubation time of 15 minutes at room temperature, the absorbance of each well was measured at 450 nm using the POLARstar Omega plate reader (BMG LABTECH, Ortenberg, Germany).

The results of the ELISA were calculated using the MARS software (BMG Labtech). In a first step, all wells were blank corrected and the mean of the duplicates was calculated. Afterwards, a 4-parameter fit was applied, in order to calculate the concentrations from the calibration curve. According to this calibration curve the amount of FVIII antigen in each well was determined. In the last step, the values were corrected by the dilution factor, resulting in the FVIII antigen amount of each sample.

Adapted FVIII Antigen ELISA for measuring Gottingen minipig samples

The supplied calibrator and control of the Asserachrom® VIII:Ag ELISA (Diagnostica Stago, Asnieres sur Seine Cedex, France, Cat. No. 00280) were reconstituted with 500 pi of distilled water, 30 minutes before starting the ELISA. After this incubation time, the calibrator was diluted 1:5 in Gottingen minipig plasma further 1:2 within the supplied phosphate buffer. This represented the starting concentration. The calibrator was further serially diluted 1:2 up to cover a concentration range from 96 mU/mL down to 1.5 mU/mL. All samples were diluted with minipig plasma, except for a last dilution step, which was performed 1:2 in the phosphate buffer. All dilutions aimed for the middle of the calibration curve. After the dilution of FVIII samples and calibrator, 100 mI of each solution were applied per well in duplicates (volume reduced by 50% in comparison to the manual). In addition, two wells were filled with 100 mI of phosphate buffer as a blank control. The plate was incubated for 2 hours at room temperature covered with a film. During this time, the peroxidase-coupled anti-human FVIII antibodies were reconstituted with 8 ml phosphate buffer and incubated 30 minutes at room temperature. After the antigen immobilization, the wells were washed five times with the supplied washing solution, which was previously diluted 1:20 with distilled water. Immediately after the washing, 200 mI of the peroxidase-coupled anti-human FVIII antibodies were added to each well and incubated for 2 hours at room temperature covered by a film. Afterwards, the plate was washed five times as before. In order to reveal the amount of bound FVIII, 200 mI of TMB solution were added to each well and incubated for exactly 5 minutes at room temperature. This reaction was stopped by the addition of 50 mI 1 M H₂SO₄ to each well.

After an incubation time of 15 minutes at room temperature, the absorbance of each well was measured at 450 nm using the POLARstar Omega plate reader (BMG LABTECH, Ortenberg, Germany).

The results of the ELISA were calculated using the MARS software (BMG Labtech). In a first step, all wells were blank corrected and the mean of the duplicates was calculated. Afterwards, a 4-parameter fit was applied, in order to calculate the concentrations from the calibration curve. According to this calibration curve the amount of FVIII antigen in each well was determined and the values were corrected by the dilution factor, resulting in the FVIII antigen amount of each sample. Since AD2CD2_SC and AD2CD2-19M_SC detection was reduced in the presence of albumin, a correction factor was determined by spiking the application solution into minipigs plasma and evaluating the decrease in FVIILAg detection. The resulting correction factor was applied to calculate specific concentrations used for further pharmacokinetic evaluation.

Albumin binding capacity assay

20% human serum albumin (HSA) was diluted 1:4000 in PBS. 96-well ELISA plates were filled with 100 mI/well with diluted HSA solution and coated during a 2 h incubation at 37°C and 400 rpm on a thermoshaker. ELISA plates were washed 3-times with 300 pl/well washing buffer. Standard control and FVIII samples either with or without Albumin pre incubation were diluted with Tris/NaCI pH 7.4 to a concentration of 0.5 U/ml chromogenic activity and 100 mI/well were added as 7-step 1:2 serial dilution. Incubation was performed for 1 h at 37°C covered on a thermoshaker. In the meantime, FIXa and FX were resolved together in 10 ml aqua dest., substrate (S-2765 and 1-2581) was solved in 12 ml aqua dest.. After FVIII incubation, plates were washed again 3 times with 300 mI/well washing buffer. Phospholipides and the FIXa/FX solution were mixed 1:5 and subsequently 50 mI/well of this solution were added and incubated for 5 min at 37°C. Without any washing step 25 mI CaCL was added to each well, followed by 5 min incubation at 37°C. Finally, 50 mI/well substrate were added and detection of activated FX-mediated substrate turnover was performed at 405 nm for 25 cycles followed by end point measurement using an ELISA reader. vWF binding capacity assay

1 U/ml of each FVIII molecule was either pre-incubated or not with 40 mg/ l albumin for 30 min at RT to promote ABD-albumin binding and an assay for determining the vWF binding capacity was performed as follows:

Plasma purified vWF (Biotest AG) was diluted with 0.9% NaCI solution to a concentration of 0.1 U/ml. Coating onto 96-well ELISA plates was done by transferring 100 pi of this solution to each well followed by an 2 h incubation at 37°C and 400 rpm. The wells were washed 3 times with 300 mI of washing buffer (8 mM sodium phosphate, 2 mM potassium phosphate, 0.14M NaCI, 10 mM KCI, 0.05% Tween-20, pH 7.4). FVIII standard (commercial rFVIII without vWF) and samples were pre-diluted with dilution buffer (25 mM Tris, 150 mM NaCI, pH 7.4) to a concentration of 0.25 U/ml according to chromogenic activity and transferred as a 7-step, serial 1:2 dilution into each plate well (100 mI/well). Incubation was carried out for 1 h at 37°C and 400 rpm. In the meantime, FIXa and FX were resolved together in 10 ml aqua dest., substrate (S-2765 and 1-2581) was solved in 12 ml aqua dest.. After FVIII incubation, plates were washed again 3 times with 300 mI/well washing buffer.

Phospholipides and the FIXa/FX solution were mixed 1:5 and subsequently 50 mI/well of this solution were added and incubated for 5 min at 37°C. Without any washing step 25 mI CaCL was added to each well, followed by 5 min incubation at 37°C. Finally, 50 mI/well substrate were added and detection of activated FX-mediated substrate turnover was performed at 405 nm for 25 cycles followed by end point measurement using an ELISA reader.

Western Blot

Reducing sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE)

Cell supernatants, cell lysates or purified material of FVIII variants were appropriately diluted with 1x NuPAGE LDS Sample Buffer (4x, Thermo Fisher Scientific, NP0007) and further diluted 1:2 with reducing sample buffer. Reducing sample buffer was produced by combining 2.5 parts of NuPAGE LDS Sample Buffer with 1 part of NuPAGE Sample Reducing Agent (10x, Thermo Fisher Scientific, NP0004). 20 mI of each sample were mixed with 20 mI of reducing sample buffer in a 1.5 ml vial and heated for 10 min at 70°C using a thermoshaker (Eppendorf). A NuPAGE 4-12% Bis-Tris Protein Gel (Thermo Fisher Scientific) was inserted into the XCell SureLock Mini-Cell Electrophoresis System (Thermo Fisher Scientific) and inner and outer chambers were filled with 1x NuPAGE MOPS SDS Running buffer (Thermo Fisher Scientific, NP0001). 500 mI of NuPAGE Antioxidant (Thermo Fisher Scientific) was added to the inner chamber. 10 mI of the each prepared sample and 4 mI of Precision Plus Protein All Blue Standard (Bio-Rad, 161-0373) diluted 1/10 in 1x LDS Sample Buffer were loaded onto the gel. The sample separation was achieved by running the gel at a constant voltage of 200 V for 50-60 min.

Non-reducing SDS-PAGE

20 mI of cell supernatants, cell lysates or purified material of FVIII variants were, either pre diluted or not, appropriately diluted with 10 pi of NuPAGE LDS Sample Buffer (4x, Thermo Fisher Scientific, NP0007) and 10 mI aqua dest.. Samples were heated for 10 min at 70°C using a thermoshaker (Eppendorf). A NuPAGE™ 3-8% Tris-Acetate Protein Gel (Thermo Fisher Scientific) was inserted into the XCell SureLock Mini-Cell Electrophoresis System (Thermo Fisher Scientific) and inner and outer chambers were filled with 1x NuPAGE™ Tris- Acetate SDS Running Buffer (Thermo Fisher Scientific, LA0041). 10 mI of the each prepared sample and 4 mI of Precision Plus Protein All Blue Standard (Bio-Rad, 161-0373) diluted 1/10 in 1x LDS Sample Buffer were loaded onto the gel. The sample separation was achieved by running the gel at a constant voltage of 150 V for 55-70 min.

Western Blotting and Detection

To investigate the separated proteins by immunofluorescence detection, they were transferred onto an Odyssey nitrocellulose membrane (Li-Cor) or an Amersham Hybond Low Fluorescence 0.2 pm polyvinylidene fluoride (PVDF) membrane (GE Healthcare Life Sciences) by using the XCell II blot module (Thermo Fisher Scientific) for semi-wet protein transfer. The PVDF membrane was activated in methanol and then applied to the SDS gel, whereas the nitrocellulose membrane was directly applied to the SDS gel. The system was filled with NuPAGE transfer buffer (20X, Thermo Fisher Scientfic) according to the manufacturer’s instructions. Protein blotting was performed for 1 h at 30 V. After protein transfer, the membrane was blocked over night at 4°C in Odyssey blocking buffer (Li-Cor). Afterwards the membrane was incubated for 1 h at room temperature simultaneously with either a rabbit anti-coagulation factor VIII monoclonal antibody (Sino Biological, 13909-R226, 1:1000) and a mouse anti-human factor VIII monoclonal antibody (Merck, MAB038, 1:2500) or with 0.0004 pg/mI sheep anti-human factor VIII:C polyclonal antibody (Cedarlane, CL20035AP, 1.5000), each diluted in Odyssey Blocking buffer containing 0.05% Tween 20. After incubation, the membrane was washed 4-times for 5 min in 0.1% PBST. For detection of FVIII heavy and light chain the membrane was incubated with 0.067 pg/ml IRDye 800CW donkey anti-mouse (Li-Cor, 926-32212, 1:15000) and 0.067 pg/ml IRDye 680RD donkey anti-rabbit (Li-Cor, 926-68073, 1:15000) diluted in Odyssey blocking buffer containing 0.05% Tween 20 for 1 h at room temperature. Alternatively, the CF680 donkey anti-sheep IgG (H&L) antibody (Biotium, 20062-1) was used in a 1:5000 dilution in Odyssey blocking buffer for binding the respective primary antibody. Finally, the membrane was washed 4 times for 5 min in 0.1% PBST, 2 times for 5 min in PBS and rinsed in water. The membrane was visualized using the Licor Odyssey Imager.

Pharmacodynamic studies

Coagulation factors were administered by a single intravenous tail vein injection into female haemophilia A mice with doses of up to 200 U/kg body weight or respective amounts of a control solution. After either 0.5, 4 or 20 h post dosing, a tail vein transection bleeding assay was performed as follows: The animals were anaesthetised with 5% isoflurane in 30% 02 and 70% N20, and immediately placed in prone position on a heating pad at +37°C. Tail vein transection was performed as described by Johansen et al., 2016. Haemophilia 22(4):625- 631.

Bleeding was monitored for 60 min and bleeding time was determined using a stop clock. Primary bleeding time was noted until first bleeding cessation. After the primary bleeding, the tail was put into a new centrifuge tube filled with pre-warmed saline. If the mouse was not bleeding at 15, 30 and 45 min post injury, the tail was lifted out of the saline and the wound was challenged by gently wiping it twice with a saline wetted gauze swab in the distal direction. Immediately after the challenge, the tail was re-submerged into the saline. The cumulative bleeding time of all following bleeds constitute the secondary bleeding time. The total bleeding time is defined as the sum of the primary and all secondary bleeding times.

Following the determination of bleeding time, the tubes were centrifuged at 4140 g at room temperature for 3 minutes. Apart from 1 ml_, the supernatant was removed. The cell pellet was resuspended and hemoglobin content was determined by using a method similar to that described by Elm et al. (2012). Results and Discussion

Generation and prescreening of several different FVIII-ABD fusion proteins covering FVIII double chain and single chain constructs was very promising in initial experiments and similar in both single chain and double chain backbones.

The formation of single chain FVIII molecule was increased compared to double chain forms when ABD was fused in between heavy chain and light chain (data not shown).

The FVIII proteins further developed and produced shown herein are listed in Table 1.

Table 1 Structure of relevant variants of FVIII with ABD fusion. A= FVIII A1+a1+A2+a2+truncated B domain, C= FVIII a3 (optionally truncated) +A3+C1+C2 domains, L= Thrombin cleavable linker, G= flexible glycine-serine linker 1 (G1), D= ABD2,

Six single chain FVIII-ABD fusion molecules were generated in silico and respective DNA constructs were tested for their expression in either HEK293 or CAP-T cells (cf. Table 2). As all of those FVIII-ABD variants were expressed, secreted and functional, based on results of the chromogenic FVIII activity measurement, all molecules were produced in midi-scale CAP-T cell culture and successfully purified in larger amounts as needed for further characterizations and PK (pharmacokinetic) analysis. Table 2 FVIII-ABD fusion proteins analyzed in supernatants of transfected HEK293 or CAP-T cells, (n/a: not available)

All six purified FVIII-ABD fusion variants were extensively characterized by several methods including determination of FVIII antigen and chromogenic activity, Actin FSL clotting, heavy and light chain detection by western blotting (WB), thrombin-cleavage analysis and binding to vWF and albumin. Table 3 gives an overview of produced FVIII-ABD variants in terms of chromogenic and clotting activity as well as antigen levels in the final solutions.

Measurement of these values indicated that FVIII-ABD fusion proteins are still capable of their biological function: Bridging factor IXa and factor X leading thereby to the activation of the latter one. Comparison of the specific chromogenic activity (chromogenic activity / antigen *100) demonstrates that ADLC_SC and ReFacto AF^® are similar (109% vs 104%). However, the specific chromogenic activities of all other FVIII-ABDs are much better, ranging from 130% to 206%.

Interestingly, the results indicate that increasing numbers of ABD motifs within one FVIII molecule decrease the clotting activity and also the capability of vWF binding. The decrease in clotting activity may be caused by the setup of the assay, which is strictly time-dependent. This may not mirror in vivo clotting activity.

Table 3 Measurements of chromogenic FVIII activity, FSL clotting activity, and antigen levels. The indicated specific activity was calculated by the ratio of chromogenic activity and antigen.

Western blot was used to observe the heavy and light chain patterns of the different FVIII- ABD fusion molecules in comparison to ReFacto AF^® (data not shown). The analysis demonstrated that, in contrast to ReFacto AF^®, the FVIII-ABD variants were mostly expressed as single chain molecules.

Activation of FVIII-ABD variants was investigated by direct incubation with thrombin at 37°C for 8 min and subsequent provision for reducing SDS-PAGE followed by western blotting. Band patterns of thrombin-activated or untreated FVIII-ABD molecules show that all FVIII- ABD molecules were activated by Thrombin in a comparable manner as ReFacto AF^® (data not shown).

Albumin binding of the ADLCLD_SC variant was tested by an assay in comparison to FVIII 6rs-Ref, demonstrating the capability of albumin binding (Fig 1). An excess of unbound soluble albumin inhibited the binding to plate-bound albumin.

The influence of ABD and linker modifications on the binding between FVIII and vWF were investigated in two settings: 1. directly, without the presence of albumin; 2. after a 30 min pre-incubation with physiological concentrations of albumin promoting the ABD-albumin binding. As demonstrated in Fig. 2, the vWF binding of FVIII-ABD fusion proteins directly decreases by an increasing number of ABD motifs. However, only one ABD motif per FVIII does not have an influence on the vWF binding in the absence of albumin, independent from its position within the molecule. When FVIII-ABDs were pre-incubated with albumin, a decreased vWF binding was observed for all FVIII-ABD variants. This reduction of vWF binding was higher the more ABD domains were incorporated into FVIII.

To investigate the impact of linkers on the production and functionality of FVIII-ABD variants, one preferred variant, AD2CD2_SC, was also produced (I) without any linkers between FVIII and ABD-Domains (AD2CD2woLG_SC) and (II) with G1 linkers but without thrombin- cleavable L linkers (AD2CD2woL_SC). These variants were compared to the double chain FVIII 6rs-Ref (ReFacto amino acid sequence), single chain FVIII backbone AC_SC and two FVIII-ABD variants having four C-terminal ABD domains either without any linkers (ACD4woLG_SC) or with one thrombin-cleavable linker followed by four ABD domains separated by G1 linkers (ACL(GD)4_SC).

Respective plasmids encoding the different FVIII variants were nucleofected into CAP-T cells and 4-day cell culture supernatants were tested for chromogenic FVIII activity, FVIII clotting activity and FVIII antigen levels according to the above-described methods. As shown in Fig. 3, AD2CD2woLG_SC, ACD4woLG_SC, and ACL(GD)4_SC were expressed in only low amounts and chromogenic activity was strongly decreased. AD2CD2woLG_SC was not ex pressed in high amounts, but had some specific chromogenic activity. No FVIII clotting activity could be detected for any of these variants. In comparison to all other controls, AD2CD2_SC and AD2CD2woL_SC demonstrated good FVIII antigen levels and great FVIII chromogenic and clotting activities, resulting in superior specific chromogenic activity values of approx. 200% or higher. AD2CD2_SC demonstrate an especially high specific clotting activity.

A western blot analysis based on a non-reduced SDS-PAGE separation of these variants is demonstrated in Fig. 4. Except for FVIII 6rs-Ref, all other variants are mainly present as single chain FVIII molecules. However, AD2CD2woLG_SC and AD2CD2woL_SC tend to form multimers or aggregates, which are not observed for the AD2CD2_SC variant.

3. Pharmacokinetic experiments

Purification of FVIII ABD variants was performed for in vivo experiments, based on supernatants of transfected CAP-T cells by strong anion exchange chromatography and affinity chromatorgraphy.

To investigate the half-life prolongation effect of ABD motifs introduced into the FVIII molecules, two pharmacokinetic (PK) studies were performed in hemophilia A mice. 12 mice per test item were used, 2 or 3 for each time point. All FVIII-ABD molecules were administered in a single dose of 200 U/kg body weight (6 ml/kg) into the tail vein by a single intravenous tail vein injection into female haemophilia A mice (B6, 129S4-F8<tm1Kaz>/J). Plasma samples taken 0.5, 4, 8, 12, and 20h (and 24h) post injection were analyzed regarding FVIII chromogenic activity and antigen levels in citrate plasma which was subsequently extracted by centrifugation. Plasma samples were stored at -80°C and analyzed for FVIII antigen and chromogenic activity. ReFacto AF^® was tested as control beside the FVIII-ABD variants.

Results are shown in Table 4. Table 4 Calculated t1/2 of FVIII-ABD variants. *AD2C_SC data were highly variable. ti/2 is specified in hours

Thus, by pharmacokinetic studies in hemophilia A mice, preferred FVIII proteins of the invention were identified which show a half-life prolonged up to 2.5x (e.g., ADLCLD_SC - about 1.5x; AD2CD2_SC - about 2.5x). Pharmacokinetics of AbD2CD2_SC were tested in a separate study and were similar to AD2CD2_SC.

It is noted that the hemophilia A mouse model may even underestimate half-life extension due to the discrepancy of murine and human albumin (murine albumin only has a half-life of about two days). Nevertheless, the observed relative extended half-life of the FVIII proteins of the invention already allows a potential reduction of intravenous FVIII injection in hemophilia patients from 2 - 3 days to a once weekly dosing.

Moreover, a pharmacokinetic proof of concept study in albumin-deficient Tg32 mice having a knock-out of murine albumin and expressing human FcRn a-chain instead of the murine one (B6.Cg-A/b^emi2M™ Fcg/ ^miDcrTg(FCGRT)32Dcr/MvwJ) was performed. This mouse model (AlbVmFcRnYhFcRn⁺), compared to hemophilia A mice, reveals a situation closer to humans as injected human albumin has a half-life of approx. 20 days which is similar to the half-life in humans. Intraveneous FVIII injection (AD2CD2_SC; ReFacto AF®) was done with 200 U/kg (based on chromogenic activity) plus 1% human albumin.

The results, shown in Fig. 5 demonstrated a half-life extension of AD2CD2_SC in comparison to ReFacto AF® of about 4x, allowing a potential reduction in patients of i.v. FVIII injection from 2 - 3 days to a 8 - 12 days dosing.

In addition, a pharmacokinetic study was performed in Gottingen Minipigs. Three animals per group were injected with 30 U FVIII antigen/kg body weight with either (I) ReFacto AF + 1% human serum Albumin (HSA), (II) ReFacto AF + 10% HSA, (III) AD2CD2_SC + 1% HSA or (IV) AD2CD2_SC + 10% HSA via the ear vein. Blood samples were taken predose, 4, 12, 36, 48, and 120 h post administration and citrate plasma was isolated immediately by centrifugation. Bioanalytical sample measurement was performed by FVIII antigen ELISA, which is specific for human FVIII and does not detect any porcine FVIII. Evaluation by non compartment analysis (Figure 6) obtained half-lives for (I) ReFacto AF + 1% HSA of 7.1 h,

(II) ReFacto AF + 10% HSA of 6.4 h, (III) AD2CD2_SC + 1% HSA of 18.6 h, and (IV) AD2CD2_SC + 10% HSA of 20.7 h. Thus, an half-life extension of approx. 3-fold was observed for AD2CD2_SC compared to ReFacto AF in this model.

Moreover, pharmacodynamics studies have been performed. Hemophilia A mice (Jackson No. B6; 129S4-F8<tm1Kaz>/J) and control mice (Jackson No. C57BL/6NCrl) were intravenously injected with 200 U/kg (based on chromogenic FVIII activity) of each FVIII variant (ReFacto®, Eloctate®, AD2CD2_SC, ADLCLD_SC) or control solutions (Vehicle Control, 0.9% NaCI) and weight loss through bleeding, bleeding time and Hb amount by OD550 have been analyzed. Additional plasma sampling (0.5 h p.a. by retro-orbital withdraw, after experiment) have been done for analysis of FVIII activity. As shown in Fig. 7, all FVIII proteins of the invention diminish total bleeding time and blood loss similar to that of control mice indicating in vivo functionality of AD2CD2_SC and ADLCLD_SC.

4. Generation of de-immmunized FVIII proteins

An initial in silico analysis of peptides in human FVIII binding to the MHC class II (T cell epitopes) was performed with the EpiMatrix tools (Epivax, Providence, Rl, USA), which predicts the binding potential with respect to a panel of eight common Class II supertype alleles (DRB1*0101, DRB1*0301, DRB1*0401, DRB1*0701, DRB1*0801, DRB1*1110, DRB1*1301, DRB1*1501), covering the majority of the human population (>90 %) and the ClustiMer algorithm to identify putative T cell epitope clusters (designated immunogenic clusters). The analysis was run for the FVIII sequence of SEQ ID NO: 60 (FVI I l-6rs), comprising 1514 amino acids, excluding the 19 amino acids of the signal sequence and 818 amino acids of the B domain. The excluded amino acids of the B domain did not interfere with either the furin or the thrombin cleavage sites. The in silico tools revealed a total of 52 immunogenic peptide clusters, with cluster scores ranging between 4 and 34, indicating a very high binding affinity at high values and a lower affinity at low values. The clusters comprised between 14 and 22 amino acids and some clusters were overlapping by a few amino acids.

For 12 of the 52 clusters, amino acid mutations were excluded, either due to interference with regions important for activity, binding or stability or due to the lack of possible exchanges. In order to de-immunize the remaining 40 clusters, 74 mutations were selected. The exchanged amino acids were preferably based on changes naturally occurring in other species. If no natural changes were available, amino acid exchanges were selected from point accepted mutation (PAM) matrices which contain mutations that occurred by natural selection. In particular, based on PAM matrices, substitutions of N are independently selected from the group consisting of D, H, S; wherein substitution of I is T; wherein substitutions of S are independently selected from the group consisting of A, N, G, T; wherein substitutions of L are independently selected from the group consisting of N and Q; wherein substitutions of V are independently selected from the group consisting of A and T; wherein substitutions of Y are independently selected from the group consisting of N and H; wherein substitutions of F are independently selected from the group consisting of H and S; wherein substitutions of K are independently selected from the group consisting of N, D, E,

Q, S and T; wherein substitutions of R are independently selected from the group consisting of Q, H and S; wherein substitutions of M are selected from the group consisting of R, Q, K and T; and/or wherein substitutions of Q are selected from the group consisting of R, D, E, H and K.

For some clusters up to three mutations were indicated, all leading to a strong reduction in the cluster score. In these cases, all mutations were selected for the incorporation. In case an additional mutation only led to a low reduction in the score, this mutation was set aside. Additionally, mutations in five clusters were completely set aside, as the total score of the cluster was already low and the predicted improvement by the mutations was marginal.

These exclusion criteria led to the reduction from 74 to 57 mutations for the incorporation into B-domain deleted (BDD)-FVIII:

N79S, I80T, 1105V, L107N, S112T, L160S, L171Q, V184A, F214H, N233D, L235F, V257A, I265T, N299D, I310T, F312S, Y426H, Y430H, L481N, F484S, L505N, S507E, L548N,

F555H, I610T, N616E, F627H, I632T, Y657D, M701K, L706N, Y748S, N754D, F1710H, F1794H, K1837E, R1936Q, F1937H, L1963Q, S2030A, S2037G, N2038D, S2077G,

M2123K, S2125G, Y2134H, Y2167N, F2215H, K2226Q, F2253H, K2258Q, V2276A,

F2279H, V2313A, S2315T, V2333A, Q2335H.

Table 6: Immunogenic clusters identified in FVIII

The incorporation of the mutations was performed in three rounds. Whereas in the first round, only single mutations were incorporated, the second and third round comprised the combination of the successfully incorporated single mutations from the first round. For each round, the most important readout was the coagulant activity of the mutated FVIII variants in comparison to the non-mutated control FVIII. The approach is laid out in Fig. 8.

The DNA sequence for all FVIII variants was synthesized and cloned into a vector backbone under the control of an EF-1a promoter. In order to reduce the size of the synthesized fragments, three additional restriction sites were integrated into the FVIII sequence by silent mutations. The sequence already had a restriction site at the beginning (Hindi 11) and at the end (Xbal) of the FVIII sequence, for cloning into the backbone. One additional restriction site (BamHI) occurred naturally after the removal of the B domain sequence. This led, in combination with the three restriction sites additionally incorporated (Kpnl, Xmal and EcoRI), to a FVIII molecule with six unique restriction sites. As a result, not only the sequences to be synthesized were shortened, but also a modular system that made the combination of mutations easier was made available. The FVIII molecule, derived from the sequence with the six restriction sites, was the reference molecule for all experiments performed and was called FVI I l-6rs. The amino acid sequence is shown in SEQ ID NO: 60. The selection of base triplets for the new amino acids was based on a human codon usage table. The base triplet most frequently used in the human genome for an amino acid was chosen.

All mutations were tested to see if the single substitutions still lead to functional FVIII molecules. The FVIII variants containing the single mutations were produced in small-scale HEK293-F culture. The HEK293-F cells were transfected in duplicates for each FVIII construct in Nucleocuvettes. The transfected cells were cultured for 4 days. After cultivation, the supernatant, containing the FVIII, was harvested by centrifugation. The FVIII coagulant activity in the supernatant was analyzed with the chromogenic method in duplicates, as described herein. The remaining supernatant was frozen until an FVIII antigen ELISA was performed. In order to compare the coagulant activity results for different constructs from different transfection days, HEK293-F cells were additionally transfected with the reference vector, coding for FVI I l-6rs. The FVIII coagulant activity for each variant was therefore not indicated in U/ml, but the relative coagulant activity was calculated, indicating the coagulant activity of the variant in relation to the FVI I l-6rs of the same transfection day.

In Fig. 9, the relative coagulant activities of the single mutation variants are shown, allocated to the domains of FVIII. The analyses revealed that only eight mutations led to a total loss of FVIII coagulant activity in the cell culture supernatant. One of these, L1963Q, was a control mutation known to lead to severe Hemophilia A. Eleven mutations led to a FVIII coagulant activity in the supernatant which was below 50 % of the coagulant activity of the control.

Thus, in total 19 mutations were excluded from further experiments, due to low or absent FVIII coagulant activity. Nevertheless, although the 19 excluded mutations were spread over 16 immunogenic clusters, only ten immunogenic clusters had to be excluded, as further mutations were successfully incorporated in the other six clusters. The remaining 38 mutations led to FVIII variants with coagulant activities, which were at least equivalent to half of the coagulant activity of the FVI I l-6rs. In addition to the coagulant activity, the antigen values of the FVIII variants and the resulting specific coagulant activities were determined (Fig. 10). As the specific coagulant activity is the ratio of FVIII chromogenic coagulant activity to FVIII antigen, 100 % indicated that the amount of FVIII coagulant activity was equivalent to the amount of FVIII antigen. However, most values were above 100 %. Higher values may indicate an improvement of the coagulant activity of the variants. Of the 38 active FVIII variants, 35 had specific coagulant activities of at least 100 %. The three remaining variants had a specific coagulant activity below 100 % but above 70 %, indicating that a fraction of the produced FVIII was inactive. Five of the excluded FVIII variants revealed specific coagulant activities below 70 %, whereof three had values even below 25 %, indicating that most of the secreted FVIII was inactive. In contrast to that, six of the excluded variants had high specific coagulant activities above 100 %, hinting towards active FVIII but a reduced secretion. All eight variants with no FVIII coagulant activity, which led to a specific coagulant activity of 0 %, also revealed no FVIII antigen. This indicated that the incorporated mutations led either to no production or to no secretion of the FVIII variants. In this first round of screening, 38 single substitutions (N79S, I80T, 1105V, S112T, L160S, L171Q, V184A, N233D, L235F, V257A, I265T, N299D, Y426H, Y430H, L505N, S507E, F555H, I610T, N616E, I632T, L706N, Y748S, N754D, K1837E, R1936Q, S2030A, S2037G, N2038D, S2077G, M2123K, S2125G, F2215H, K2226Q, K2258Q, V2313A, S2315T,

V2333A, Q2335H) led to a functional FVIII molecule with substantial coagulant activity (at least equivalent to half of the coagulant activity of the FVIII-6rs).

Five additional single mutations were tested (S660G, I658T, N1796D, N2137H, I2168T). These mutations were proposed for four of the immunogenic clusters that had to be excluded in the first screening round due to non-functional mutational variants. These five mutations were originally not tested, as they had a lower calculated influence on the reduction of immunogenicity. However, analyses of the variants revealed only low or no FVIII coagulant activity in the supernatant, although the specific coagulant activities were around 100 % for three of the variants. Nevertheless, the mutations were not transferred to the second round of screening, as the coagulant activities were quite low, only exceeding the 50 % limit by about 10 % for I658T and N2137H.

Although all of the 38 successfully incorporated single mutations had the characteristics to be transferred to the second screening round, only one mutation for each immunogenic cluster was chosen, in order to keep the combination of the single mutations feasible. Hence, the mutation resulting in a lower FVIII coagulant activity was excluded. Additionally, mutation S2030A was found not to be part of the cluster comprising S2037G and N2038D but of a preceding cluster. As the calculated score of this cluster was already very low without the mutation, S2030A was also excluded. This led to 25 mutations, which were transferred to the second round of screening.

In a second screening round 25 out of the 38 mutations have been chosen (N79S, S112T, L160S, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, F555H, N616E, I632T,

L706N, Y748S, K1837E, R1936Q, N2038D, S2077G, S2125G, F2215H, K2226Q, K2258Q, S2315T, V2333A).

In silico modelling of FVIII variants incorporating mutations was attempted, but was not successful due to lack of a complete crystal structure of FVIII having a sufficiently high resolution. The inventors chose to combine the functional single mutations in small groups to keep identify mutations, which lead to inactive protein in combination with others. For every section (defined by the restriction sites), one vector was designed containing all mutations, which led to FVIII variants with relative coagulant activities above 50 %. Additionally, one vector was designed containing only the mutations that led to relative coagulant activities above 80 % and reduced the immunogenicity score for the cluster by at least 15 points. For section A1A2 and A2, only one vector was constructed, as all mutations had relative coagulant activities above 80 % and reduced the score by more than 15 points. Based on this, FVIII variants comprising substitutions at the following positions were produced as shown in Table 7.

Table 7

The production of the FVIII variants occurred as described for the first round. After four days of production, the FVIII coagulant activity in the cell culture supernatant was determined. Coagulant activities comparable to or even better than the control FVIII-6rs were achieved in the sections A1 and A1 A2 (Fig. 11A). In particular, the combination of the three mutations in section A1A2 seemed to have a positive effect on production and/or secretion of the FVIII variant, leading to more than twice the amount of secreted, active FVIII-6rs. Due to the good coagulant activities, the seven mutations in A1 and the three mutations in A1A2 were taken to the third round. In section A2, the coagulant activity of the FVIII variant was below 80 % and the two variants of the A3C1C2 section revealed coagulant activities below 40 %. Due to these results, the mutations combined in A2 and A3C1C2 had to be further analyzed. The specific coagulant activities for all combinations were above 100 %, indicating that the produced FVIII variants were functional, except for the variant with nine mutations in the A3C1C2 section (Fig. 11B). The low specific coagulant activity of this variant indicated that mainly inactive FVIII was secreted.

In order to detect the mutations in the A2 and A3C1C2 section which interfered with the coagulant activity of the FVIII variant, two design-of-experiment (DOE) matrices were generated. To avoid synthesis of vectors with every possible combination of the mutations, the five mutations in the A2 section were modelled in a half factorial design, whereas the six mutations in the A3C1C2 section were analyzed in an 8^th fraction fractional design. Setting aside the variants which had already been tested (single-mutant, full-mutant and naive variant), ten vectors were designed for the A2 section and 14 vectors were designed for the A3C1C2 section. As before, the variants were produced in HEK293-F cells and the FVIII coagulant activity in the supernatant was determined. The analysis revealed that mutation I632T in the A2 section probably was responsible for the reduced coagulant activity, as it was incorporated in all variants with coagulant activities below 100 % (Fig. 5A). Said mutant is preferably not included in proteins of the invention. In the A3C1C2 section three mutations, N2038D, S2125G and K2258Q, seemed to decrease the FVIII coagulant activity (Fig. 12B). However, an obvious influence on a decreased coagulant activity was only detectable for mutation S2125G, which is preferably not included in proteins of the invention. For N2038D and K2258Q it was not clearly identifiable whether their influence might only have occurred in combination with each other or S2125G.

The specific coagulant activities of all variants in the A2 and the A3C1C2 section were at least around 100 % (data not shown). This revealed that the reduced FVIII coagulant activities compared to FVIII-6rs were probably due to production or secretion problems and not due to inactive FVIII.

Based on the results from the DOE matrices one vector for the A2 section and four vectors for the A3C1C2 section were designed. The four A3C1C2 vectors omitted either only mutation S2125G, or mutation S2125G in combination with K2258Q or N2038D, or all three mutations. The incorporated mutations in the five vectors are shown in Table 8 below.

Table 8

The measurement of the FVIII coagulant activity in the supernatant of the HEK293-F cells, transfected with the different vectors, revealed that the coagulant activity of the variant with four mutations in the A2 section was comparable to the coagulant activity of FVIII-6rs (Fig. 13A). In contrast to that, although all four A3C1C2 variants were active, the coagulant activity of the variants comprising five mutations and four mutations without mutation N2038D revealed a reduced coagulant activity compared to FVIII-6rs. This indicates that the exclusion of mutation N2038D alone had no influence on the production or secretion of FVIII, as the coagulant activity remained low. In contrast to that, the exclusion of K2258Q led to an increase in FVIII coagulant activity. However, although not effective as a single deletion, re moval of N2038D in combination with K2258Q had an additive effect and further improved the FVIII coagulant activity of the variant. Nevertheless, the combination of four mutations, still containing the N2038D, was transferred to the third round. This was due to the aim to incorporate many mutations, and accordingly, reduce immunogenicity as much as possible. Additionally, the coagulant activity results for this variant were around 100 % and similar to the results for combinations in other sections. The specific coagulant activity for all variants was at least 100 %, indicating that only active FVIII was present.

Finally, the second screening round led to 19 mutations, which could be combined in five sections. For section A1 and A1A2, the combination of all mutations from the first round could be included. This was not possible for the sections A2 and A3C1C2. Based on a DOE matrix one mutation had to be excluded in the A2 section, and five mutations had to be excluded in the A3C1C2 section.

Accordingly, in the second round, six additional mutations have been set aside. The last round of screening comprised a single FVIII molecule containing all 19 mutations (N79S; S112T; L160S; L171Q; V184A; N233D; I265T; N299D; Y426H; S507E; F555H; N616E; L706N; Y748S; K1837E; N2038D; S2077G; S2315T; V2333A) remaining from screening round 1 and 2. This mutated FVIII variant was shown to be functional in coagulation, and comprises a high number of single substitutions which renders the molecule less immunogenic.

Due to the incorporation of the 19 mutations into the FVIII sequence, the initial immunogenicity score of FVI I l-6rs of 7.01was reduced to -10.55 for FVIII-19M. The immunogenicity score indicates the immunogenicity of the protein of interest in relation to a protein with a random sequence. The immunogenicity score of the random protein is set to 0. In order to be able to compare the scores for different proteins of different length, the score is given per 1000 of the 9-mers to which a protein is split for in silico analysis. Exemplary immunogenicity scores of other proteins are about 23 for Tetanus Toxin, 10.03 for Refacto AF, about -10 for albumin, or about -42 for an IgG Fc region.

As the A3C1C2 section revealed to be mostly influenced by the incorporation of mutations, an additional vector was designed with no mutations in section A3C1C2 (FVI 11-15M), in order to compare the coagulant activities. This led to two proteins comprising in total 15 (FVIII- 15M, SEQ ID NO: 65) and 19 mutations (FVIII-19M, SEQ ID NO: 63). For both vectors, the analysis of the FVIII coagulant activity in the supernatant revealed coagulant activities at least comparable to FVII l-6rs. The variant with the 15 mutations was secreted in a higher concentration of active FVII I compared to the one with the 19 mutations. The specific coagulant activities were nearly 100 % for the FVIII-15M and above 100 % for FVIII-19M.

A further variant does not comprise a substitution at position K1837 such as the K1837E substitution, which appears to reduce coagulant activity, but comprises the other substitutions of FVIII-19M. This variant is designated FVIII-18M. It has about the same specific coagulant activity as FVIII-19M, but a higher chromogenic coagulant activity when measured in the supernatant. It can be concluded that the K1837E substitution may reduce production, folding or secretion of FVII I to a certain extent. However, the coagulant activity of FVIII-18M with regard to the clotting assay is also improved, so the substitution may also otherwise reduce coagulant activity. Nevertheless, further assays described below show that FVIII-19M can be therapeutically used.

Further advantageous variants were produced, e.g., a FVIII protein FVIII-GOF1 with the substitutions L171Q, S507E, Y748S and V2333A; and FVIII-GOF2 with the substitutions L171Q, N299D, N616E and V2333A. These variants incorporate substitutions in the different regions showing the best results regarding coagulant activity and specific coagulant activity. The following variants incorporate the substitutions with the best results regarding reduction of the immunogenicity score: FVIII-LS1 with the substitutions S112T, S507E, Y748S,

K1837E and N2038D; and FVIII-LS2 with the substitutions S112T, Y426H, N754D, K1837E and N2038D. Of note, the substitutions S507E and Y748S are highly advantageous with regard to both aspects. Preferred proteins of the invention thus at least comprise said substitutions, as well as optionally, L171Q and V2333A. It is further optimal for reducing immunogenicity to incorporate all seven preferred substitutions in the A1 region.

Fig. 14 summarizes chromogenic coagulant activity, clotting coagulant activity and specific coagulant activity of different Factor VIII proteins, wherein all constructs have been produced in HEK293-F cells, with the analysis performed in the supernatant of the cells. The measured relative coagulant activities are given in percentage based on the coagulant activity of FVIII- 6rs, which was always produced and tested in parallel. Specific coagulant activities are based on the ratio of chromogenic coagulant activity detected to antigen detected in the supernatant by an ELISA as described below.

For further experiments, the constructs FVIII-19M and FVIII-6rs have been produced in CAP- T cells, resulting in higher protein masses. Further on, FVIII-19M and FVIII-6rs were purified from the cell culture supernatant. Different analyses of the FVIII-19M in comparison to FVIII-6rs, ReFacto AF and Nuwiq were performed using SDS-PAGE and Western Blot. The blot revealed that FVIII-19M and FVIII- 6rs were mainly produced as single chain proteins, in comparison to the two commercially available products, which are mainly double chain FVIII. This difference may be due to the different cell lines used for production. Two additional Western Blots revealed that FVIII-19M and FVI I l-6rs are glycosylated and sulfated, confirming that the post-translational modifications took place in the CAP-T cells. However, no detailed conclusion could be drawn on whether all six tyrosines were sulfated and which glycosylation patterns were added.

The clotting time for the different FVIII products was determined using the ROTEM method. During this analysis, the clotting time of plasma is analyzed depending on the amount and the functionality of the applied FVIII. The FVIII products were added to FVI ll-deficient plasma and the clotting was initiated via the intrinsic pathway. The time until a clot was starting to form was measured. By adding different concentrations of FVIII, an increase of the clotting time was detected in correlation to decreasing concentrations of FVIII. When comparing the different products, ReFacto AF (Pfizer Inc., produced in CHO cells) and Nuwiq (Octapharma AG, produced in HEK cells), which are both B-domain deleted, revealed very similar clotting times, whereas the clotting times were slightly prolonged for FVIII-6rs and even more for FVIII-19M (Fig. 15). However, all clotting times only varied between 120 seconds and 160 seconds at 1 U/ml FVIII, which was still in the normal clotting time range of 100-240 seconds in healthy people.

In a thrombin generation assay (Fig. 16), FVIII-6rs resembled ReFacto AF and Nuwiq regarding the amount of generated peak thrombin and the time to peak thrombin generation but showed a slightly reduced area under the curve. FVIII-19M revealed significantly lower results for generated peak thrombin, area under the curve and time needed to reach peak thrombin generation, especially compared to ReFacto AF and Nuwiq. However, these results were comparable with the ROTEM results, revealing a slightly prolonged clotting time for FVI 11-19M.

The binding potency of FVI 11-19M and FVIII-6rs to vWF was determined in an ELISA based approach. ReFacto AF was used as a reference, and the potency of the binding of ReFacto AF to vWF was set to 1. All other potencies were calculated in relation to ReFacto AF. The data revealed that the vWF binding was similar for ReFacto AF and Nuwiq but impaired for FVI I l-6rs and FVI 11-19M (Fig. 17). However, the only significant difference could be detected between Nuwiq and FVIII-19M. The reduced binding, also of FVI ll-6rs, might be due to different post-translational modifications made by the CAP-T cells. This could be correlated with the reduced sulfation detected in the heavy and light chain in the Western Blot. Missing sulfation might have influenced vWF binding, as especially the sulfation at Y1683 is important for the binding of vWF. However, decreased sulfation cannot be the only reason for the reduced vWF- binding, as ReFacto AF revealed nearly no sulfation on the Western Blot, but good vWF- binding. The reduction in the potency of FVIII-19M in comparison to FVIII-6rs might thus be due to additional structural changes, derived from incorporated mutations, influencing the vWF binding.

The slightly lower coagulant activity of FVIII-19M in the experiments done with regard to clotting time, thrombin generation and binding to vWF may at least in part be explained by the lower coagulant activity as measured in the clotting assay compared to the chromogenic assay, as the chromogenic assay may particularly overestimate the coagulant activity of FVIII-19M (Fig. 18).

Immunogenicity of FVIII proteins was analyzed with two different approaches:

Immunogenicity score

The reduced immunogenicity of the FVIII molecule with 19 single point mutations was evaluated by an in silico method of calculation. The initial immunogenicity score of the molecule without mutations of about 7 was reduced to about -11 for the FVIII molecule with 19 single point mutations. The immunogenicity scores were calculated in relation to a protein with a randomized sequence of the same length as the analyzed FVIII.

The immunogenicity score, which indicates the immunogenic potential of a protein, was calculated using EpiVax’s EpiMatrix System. In order to be able to compare the immunogenicity score of the protein of interest to the scores of other proteins, it is correlated to the score of a protein with a randomized amino acid sequence. The immunogenicity score of this “average” protein is set to zero. Additionally, the immunogenicity score is indicated per 1000 peptides, each peptide comprising nine amino acids. Due to this, proteins of various length can be compared. Immunogenicity scores above zero indicate the presence of excess MHC class II ligands and denote a higher potential for immunogenicity while scores below zero indicate the presence of fewer potential MHC class II ligands than in a random protein and a lower potential for immunogenicity. Proteins scoring above +20 are considered to have a significant immunogenic potential. In vitro immunogenicity assay

An in vitro T cell assay for analyzing the immunogenicity of a protein of interest, such as FVIII, was established, which is based on dendritic cells (DC) and regulatory T-cell-depleted CD4⁺ T cells of healthy donors and stimulation with the protein of interest.

The recombinant molecule FVIII-19M was shown to be less immunogenic by the in vitro immunogenicity T cell assay compared to the FVIII molecule without mutations.

The in vitro assay is able to determine whether less T cells become activated, due to a reduced presentation of FVIII-19M peptides on the surface of DCs. The assay includes DCs, derived from monocytes, and CD4⁺CD25^_ T cells. The CD4⁺CD25⁺ T cells were depleted prior to co-cultivation, as this subpopulation mainly comprises regulatory T cells. This was important, because the T cell population derived from healthy donors as well as tolerant Hemophilia A patients was expected to contain regulatory T cells suppressing the FVIII- specific T cells which were not depleted during ontogeny (Kamate, C., Lenting, P. J., van den Berg, H. M. & Mutis, T. Depletion of CD4+/CD25^hi9h regulatory T cells may enhance or uncover factor VII l-specific T-cell responses in healthy individuals. Journal of Thrombosis and Haemostasis 5, 611-613 (2007)). As the aim of the assay was to stimulate the FVIII- specific T cells, the regulatory T cells were depleted. The approach of using DCs as APCs was chosen due to two reasons. On the one hand, the focus was on the activation of CD4⁺ T cells based on the interaction with the presented FVIII epitopes. Influences due to interaction with other immune cells might have distorted the results. On the other hand, the assay may also be performed with cells derived from Hemophilia A patients. These patients, especially previously untreated Hemophilia A patients, may still have naive T cells, and the activation of naive T cells primarily occurs by DCs.

As illustrated in Fig. 20, monocytes were purified from thawed PBMCs and differentiated to immature DCs (iDCs) using IL-4 and GM-CSF, e.g., in 5 days. The cells were stimulated, e.g., for 1 day with cytokines (e.g., an IL-Mix as defined below) and antigen, e.g., the FVIII of interest, in order to obtain mature DCs (mDCs). The CD4⁺CD25^_ T cells were purified from PBMCs two days prior to the co-cultivation with the mDCs. After the purification, the CD4⁺CD25 T cells were labelled with CFSE (Carboxyfluorescein diacetate succinimidyl ester) and cultured for 2 days in presence of IL-2, in order to recover from the purification and labeling process. mDCs and labelled CD4⁺CD25^_ T cells were co-cultivated, e.g., for 9 days. The T cells were harvested and analyzed by flow cytometry. The DC-T cell assay was performed with the FVIII-19M and FVIII-6rs. The cells were purified from PBMCs of healthy donors. The DCs were stimulated after differentiation either with the previously determined IL-Mix alone or with the IL-Mix and additional protein. The additional protein was ReFacto AF as a positive control for FVI I l-specific T cell proliferation and FVI II- 6rs and FVIII-19M as the proteins of interest. The concentration of the FVIII products was 15 U/ml, in order to ensure that enough FVIII was present. The co-cultivation was done in 48- well plates, leading to a DC:T cell ratio of at least 1:10. All cells were analyzed by flow cytometry after their purification, in order to ensure purity, and the T cells were analyzed prior to co-cultivation, in order to exclude pre-activation, and after 9 days of co-cultivation. The results from the flow cytometric analyses of the T cells after 9 days of co-cultivation were further analyzed. The proliferation of all CD4⁺ T cells was determined for every approach.

In total, 23 different healthy donors were analyzed. For the final analysis donors revealing a higher T cell response to DCs only treated with the IL-Mix than with the IL-Mix and FVIII, and donors showing a markedly varying viability of T cells when stimulated with different FVIII proteins were excluded. As not all healthy people possess T cells against FVIII, it was expected that not all healthy donors react to FVIII. Based on this selection, the results of 10 healthy donors were analyzed.

Fig. 21 displays the difference between the CD4⁺ T cell proliferation to DCs stimulated with IL-Mix plus FVIII-19M and the CD4⁺ T cell proliferation to DCs stimulated with IL-Mix plus FVI I l-6rs. Results below 0 indicate a reduced T cell response to FVI 11-19M compared to FVI I l-6rs. This reduced response was detected in most donors. However, results above 0 were detected in a minority of donors. Even though, the differences in proliferation were below 10 % for these donors, whereas higher differences were detected in the group of donors showing a reduced response to FVIII-19M. Altogether, the amount of donors revealing a reduced response to FVIII-19M was larger, leading to a significant reduction in CD4⁺ T cell proliferation to DCs stimulated with IL-Mix and FVIII-19M. These results from the in vitro DC-T cell assay confirmed a reduced CD4⁺ T cell proliferation in response to the de- immunized FVIII variant containing the 19 amino acid mutations.

In vitro stability

The recombinant FVIII protein of the invention was incubated in FVIII-deficient plasma at 37°C, and coagulant activity was analyzed after different time periods, in comparison to ReFacto AF, Nuwiq and FVI I l-6rs. As shown in Table 9, the loss of coagulant activity was acceptable for all analyzed proteins. Table 9: Chromogenic coagulant activity of FVIII proteins incubated in FVIII-deficient plasma at 37°C.

Methods

In silico analyses

In silico T cell epitope-modelling (EpiMatrix tools) was applied in order to identify MHC class ll-binding peptides, cluster the results, to compare the clusters to other proteins and to predict amino acid exchanges.

The modelling tools used for the in silico analyses are commercially available from EpiVax (Providence, Rl, USA). The tools analyze protein sequences, in order to find peptides binding to the MHC class II. These peptides are further analyzed regarding potential amino acid exchanges, in order to reduce this binding.

The FVIII molecule used for the modelling process was a B domain deleted Factor VIII molecule (BDD FVIII) in which 818 amino acids of the B domain are deleted (FVI I l-6rs, SEQ ID NO: 60). The modelling process comprised four steps. In the first step, the EpiMatrix tool split the protein into peptides, consisting of nine amino acids, so-called 9-mers. This is due to the fact that the core binding region of the MHC class II comprises nine amino acids. The sequence of a 9-mer and its following 9-mer overlap by eight amino acids. By building these highly overlapping 9-mers, no potential binding peptides were lost. The binding capacity of all 9-mers was tested against eight common HLA class II super-type alleles (DRB1*0101, DRB1*0301, DRB1*0401, DRB1*0701, DRB1*0801, DRB1*1101, DRB1*1301, DRB1*1501), covering over 90 % of the human population (De Groot, A S, MCMURRY, J. & MOISE, L. Prediction of immunogenicity: in silico paradigms, ex vivo and in vivo correlates. Current Opinion in Pharmacology 8, 620-626 (2008); Moise, L. et al. Effect of HLA DR epitope de immunization of Factor VIII in vitro and in vivo. Clinical Immunology 142, 320-331 (2012)). The potential of each 9-mer to bind a HLA allele was indicated by a “Z score”. This score indicated the strength of binding, normalized to the frequency of the given allele in the population (De Groot, Anne S. & Martin, W. Reducing risk, improving outcomes: Bioengineering less immunogenic protein therapeutics. Clinical Immunology 131, 189-201 (2009)). A 9-mer with a high Z-score for at least four HLA alleles was regarded as highly immunogenic and called an EpiBar (Weber, C. A. et al. T cell epitope: Friend or Foe? Immunogenicity of biologies in context. Advanced Drug Delivery Reviews 61, 965-976 (2009)).

In the following step, overlapping EpiBars were clustered using the program ClustiMer (Moise, L. et al. Effect of HLA DR epitope de-immunization of Factor VIII in vitro and in vivo. Clinical Immunology 142, 320-331 (2012)). This analysis led to EpiBar clusters of up to 25 amino acids, binding to class II MHCs derived from multiple HLA alleles. Previous to the optimization of these clusters, an analysis regarding similarity with other endogenous proteins was performed using the program JanusMatrix. A sequence overlap with at least two other endogenous proteins led to the exclusion of the cluster from further modification, because central tolerance is very likely established to common endogenous peptides. In addition to that, clusters comprising cleavage sites, activation sites or other sites important for the activity of FVIII were set aside and were not altered. In the last step, amino acid exchanges were calculated using OptiMatrix. This tool evaluated the contribution of each amino acid of an identified cluster to the binding to the MHC class II (Moise, L. et al. Effect of HLA DR epitope de-immunization of Factor VIII in vitro and in vivo. Clinical Immunology 142, 320-331 (2012)). Afterwards, OptiMatrix calculated which amino acid substitutions reduced the binding affinity to the MHC class II (De Groot, Anne S. & Moise, L. Prediction of immunognicity for therapeutic proteins: State of the art. Current Opinion in Drug Discovery & Development 10, 1-9 (2007)). These optimizations were based on the principles that the amino acid exchanges had to be conservative, preferably occurred in other species and were not registered in the database comprising all known FVIII mutations leading to Hemophilia A (Kemball-Cook, G., Tuddenham, Edward G. D. & Wacey, A. I. The Factor VIII Structure and Mutation Resource Site: HAMSTeRS Version 4. Nucleic Acids Research 26, 216-219 (1998)). Analysis of mutations

By in silico results recommended amino acid exchanges were incorporated into the FVIII sequence. Each mutation of the factor VIII molecule was screened for FVIII coagulant activity. In the first round single mutations were incorporated. The second and third round comprised the combination of the successfully incorporated single mutations from the first round. Reference molecule was FVIII-6rs with six unique restriction sites. The single mutations (first round) and the combinations of mutations (second and third round) have been analyzed after production (transient transfection) in human cell lines (HEK293-F), wherein the produced and secreted protein was analyzed in the cell culture supernatant.

FVIII coagulant activity was analyzed by chromogenic and clotting methods. Antigen values were analyzed by FVIII antigen ELISA. The specific coagulant activities were calculated as the relation of coagulant activity to antigen.

In the first round, from 57 recommended mutations, 19 mutations were excluded due to low or absent FVIII coagulant activity. The remaining 38 mutations led to FVIII variants with coagulant activities of at least equivalent to half of the coagulant activity of the reference molecule. For the second round, 25 mutations were chosen, only one mutation for each immunogenic cluster. Combinations of mutations combined in sections (defined by restriction sites) were analyzed (relative and specific coagulant activity). Further analysis was done based on DOE matrices.

Large scale production of the recombinant proteins was done in CAP-T cells by transient transfection and subsequent purification (SAEx - strong anion exchange chromatography - alternatively TFF - tangential flow filtration, AC - FVIII affinity chromatography, buffer exchange via SEC - size exclusion chromatography).

Further analysis was done regarding posttranslational modifications and functionality (Western Blots detecting FVIII heavy and light chain, thrombin cleavage, glycosylation and sulfation, 2D-DIGE and functional assays ROTEM, TGA and vWF-FVIII ELISA).

The reduced immunogenicity of the recombinant proteins according to the invention was confirmed in silico and in vitro (DC-T cell assay).

T ransfection

For transient transfection, eukaryotic cell systems, namely CAP-T cells and HEK cells, were used. CAP cells are an immortalized cell line based on primary human amniocytes and grow in suspension. CAP-T cells are based on the original CAP cells and additionally express the large T antigen of simian virus 40. CAP-T cells are especially useful for transient transfection. Moreover, the HEK 293-F cell line was used for transient transfection. The HEK 293-F cell line is derived from the original HEK 293 cell line and is adapted to suspension growth in serum-free medium. The HEK cells were used for the small scale production of various mutated FVIII variants.

Transient transfection was done by electroporation using the commercially available 4D- Nucleofector system (Lonza Group Ltd., Basel). Electroporation was performed with 7-10⁶ HEK293-F cells and 7 pg FVIII plasmid in a volume of 100 pi. T10⁷ CAP-T cells were used for transfection with 5 pg FVIII plasmid in a volume of 100 pi. After transfection, the cells were incubated for 4 days. The cells and the supernatant were used for further analysis.

Protein purification

FVI I l-6rs and FVIII-19M was produced in CAP-T cells in up to 800 ml scales. Purification occurred directly from the cell culture supernatant by FPLC. The first step was either a tangential flow filtration or an ion exchange chromatography, using the strong anion exchange columns HiTrap Capto Q (GE Healthcare Europe GmbH, Freiburg). In this step the sample was concentrated, host cell proteins were lost and the buffer was exchanged. The fractions containing the eluted protein were determined according to the chromatogram. The second step was an affinity chromatography, using a column packed with the commercially available VI I ISelect resin (GE Healthcare Europe GmbH, Freiburg). The fractions containing the eluted FVIII were determined according to the chromatogram. The last step was a buffer exchange to FVIII Formulation Buffer by size exclusion chromatography, using the HiTrap Desalting columns (GE Healthcare Europe GmbH, Freiburg). The fractions containing FVIII were determined according to a high UV peak and a stable conductivity peak in the chromatogram. After purification, the FVIII products were concentrated via spin columns (Merck Millipore, Darmstadt) with a molecular weight cut-off of 10 kDa. All columns were run under the conditions specified by the manufacturer.

Analytics

If not otherwise mentioned, analytics were carried out as described above.

For the detection of sulfotyrosines in Western Blot, a mouse anti-human sulfotyrosine antibody (Merck Millipore, Darmstadt) was used. The secondary antibody was a donkey anti mouse antibody coupled to IRDye 800CW. Preparation was performed as described above. In order to determine whether the FVIII variants can be activated by thrombin, the samples were incubated with 10 U/ml thrombin for 8 minutes at 37 °C prior to the SDS-PAGE and Western Blot. SDS-PAGE and Western Blot were performed as described above. The primary antibody for the detection of FVIII in the Western Blot was the polyclonal sheep anti human Factor VIII antibody (Cedarlane, Burlington) detecting heavy and light chain and a secondary donkey anti-sheep antibody coupled to IRDye 800CW (LI-COR Biotechnology GmbH, Bad Homburg).

Functional Assays

In the Thrombin Generation Assay (TGA), the amount of generated thrombin is measured. The clotting cascade takes place, started via the extrinsic pathway by tissue factor. The thrombin finally generated cleaves a fluorogenic substrate which can be measured at 460 nm. The assay was performed with FVIII diluted in FVI I l-deficient plasma. FVIII concentrations up to 0.25 U/ml were analyzed. TGA reagent C low and TGA substrate, both commercially available by Technoclone (Vienna), were added to each sample well referring to the manufacturer's protocol. TGA reagent low consist of low concentrations of phospholipid micelles containing recombinant human tissue factor, in order to initiate the clotting cascade. The substrate is the fluorogenic substrate finally cleaved by the generated thrombin. The reaction was performed at 37 °C in a plate reader and the development of the fluorogenic substrate was measured for two hours. In addition to the samples, a calibration curve was measured using the TGA Cal Set, also available by Technoclone (Vienna). The amount of generated thrombin was calculated based on the calibration curve. Additionally, the area under the curve and the time to maximum thrombin generation was calculated based on the first deviation of the generated curve.

The Thromboelastometry (TEM), using the ROTEM system (Tern International GmbH, Munich), was also used to determine the functionality of the FVIII variants. In this method, the sample is applied to a cup and a pin is set into the middle of the cup. The sample lies in the space between cup and pin. The pin rotates and its rotation is monitored by a light beam, which is reflected from the pin onto a detector. Upon the onset of coagulation, the generated clot restricts the movement of the pin up to a maximum when the final clot is formed. In contrast to the TGA, the clotting was initiated via the intrinsic pathway in the ROTEM system, using the in-tem reagents, commercially available by Tern International GmbH (Munich).

FVIII concentrations between 1 U/ml and 0.01 U/ml, based on the chromogenic coagulant activity, were analyzed. The reagents were used as described in the manufacturer's protocol. The measurement and calculations were performed fully automated by the ROTEM system. Finally, clotting times were determined. In vitro DC-T cell Assay

The DCs and T cells for the in vitro assay were derived from PBMCs of healthy donors. The PBMCs were purified from either leukapheresis products or whole blood donations of healthy donors via a density gradient using Lymphoflot (Bio-Rad Laboratories GmbH, Munchen). The PBMCs were cryopreserved at -150 °C until used for the assay. Monocytes as well as CD4+CD25- T cells were purified with the MACS technology commercially available from Miltenyi Biotec (Miltenyi Biotec GmbH, Bergisch Gladbach). For the monocyte purification CD14 MicroBeads were used, whereas the CD4⁺CD25⁺ Regulatory T Cell Isolation Kit (Miltenyi Biotec GmbH, Bergisch Gladbach) was used for the T cell purification. For monocytes, purification occurred according to the manufacturer’s protocols. For the purification of the CD4+CD25- T cells, the two-step purification process recommended in the manufacturer’s protocol was combined in one step performing the negative selection of CD4+ T cells and the positive selection of CD25+ cells in parallel and using only one purification column. Used amount of antibodies were according to the protocol and incubation times according to the negative selection step. Monocytes were the first cells to be purified during the assay. After purification, the monocytes were plated at T10⁶ cells/ml in X-VIVO 15 medium (Lonza Group Ltd., Basel). In order to differentiate the monocytes to DCs, a final concentration of 4000 U/ml Granulocyte-macrophage colony-stimulating factor (GM-CSF) and 1250 U/ml Interleukin (IL)-4 (PeproTech, Hamburg) were added to each well. The monocytes were cultured for five days at 37 °C. After 4 days the purification of the CD4+CD25- T cells took place. After purification the T cells were labeled with CFSE (BioLegend, Koblenz) according to Quah et ai, Nature Protocols, 2007. Afterwards the purified T cells were plated in a final concentration of 2-10⁶ cells/ml in X-VIVO- 15. IL-2 (PeproTech, Hamburg) was added to the cell suspension in a final concentration of 20 U/ml. The T cells were cultured at 37 °C for 2 days. 24 hours before starting the co-cultivation of DCs and CD4+CD25- T cells, the DCs were stimulated with an IL-Mix consisting of 10 ng/ml I L- 1 b , 10 ng/ml IL-6 and 10 ng/ml Tumor necrosis factor (TNF)-a (Miltenyi Biotec GmbH, Bergisch Gladbach) with or without 15 U/ml FVIII. The next day, the T cells were harvested and the cell count was determined. The T cell concentration was set to 2-10⁶ cells/ml in fresh X-VIVO 15. The supernatant in the wells containing the DCs was carefully removed in order not to disturb the DCs. T cell suspension was applied to the DC wells, in order to reach a DC:T cell ratio of at least 1:10. The amount of T cell suspension added was dependent on the size of the well in which the DCs were originally plated. No additional cytokines were added to the medium. The cells were co-cultivated for 9 days at 37 °C. Afterwards the T cells were harvested and analyzed by flow cytometry regarding proliferation. 5. De-immmunized FVIII proteins of the invention

The 19 de-immunizing amino acid substitutions of FVIII-19M were incorporated into the FVIII- ABD fusion molecules on the DNA level. The DNA sequences was generated in silico using VectorNTI (Thermo Fisher Scientific, Massachusetts, USA), and afterwards the full FVIII sequence was synthesized and cloned into the target vector. By transformation of E.coli K12 with said plasmids, expansion of transformed bacteria under ampicillin selection and plasmid preparation, large amounts of the plasmids could be prepared. Genetic engineering work was carried out by Thermo Fisher Scientific.

Cultivation of CAP-T cells and expression of the FVIII encoding plasmids by transient transfection was done as described elsewhere in this document. In order to verify expression levels and functionality of de-immunized, FVIII proteins fused with albumin-binding domains, plasmids encoding the de-immunized FVIII-ABD variant AD2CD2-19M_SC, and the fusion molecule AD2CD2_SC, both including 4 albumin-binding domains, and the FVIII control 6rs- Ref (ReFacto sequence) were nucleofected into CAP-T cells. 4-day cell culture supernatants were tested for chromogenic FVIII activity and FVIII antigen levels according to the above- described methods. As shown in Fig. 22, both FVIII chromogenic activity and FVIII antigen levels were at least 3-times higher for AD2CD2-19M_SC compared to 6rs-Ref. Interestingly, AD2CD2-19M_SC resulted in a better chromogenic activity and FVIII antigen levels compared to AD2CD2_SC (chromogenic activity: 2.64 vs 1.90 U/mL and FVIII antigen: 2.00 vs 1.40 U/mL, respectively). The specific chromogenic activity was 113% for 6rs-Ref, while AD2CD2_SC and AD2CD2-19M_SC resulted in 136% and 133%, respectively.

First in vivo pharmacokinetic experiments were performed in hemophilia A mice (B6, 129S4- F8<tm1Kaz>/J) with affinity chromatography-purified FVIII material to investigate the half-life prolongation effect of AD2CD2-19M_SC. 12 mice per test item were used, 3 for each time point. AD2CD2-19M_SC and ReFacto AF (control) were administered in a single dose of 200 U/kg body weight (7.14 ml/kg) into the tail vein by a single intravenous tail vein injection into female mice. Blood samples were taken 0.5, 4, 8, 12, and 20h post injection and citrate plasma was extracted by centrifugation. Plasma samples were stored at -80°C and analyzed for FVIII antigen and chromogenic activity as described. For pharmacokinetic evaluation, a non-compartmental analysis was performed using Phoenix WinNonlin (Certara USA Inc., USA). Mean values of the FVIII antigen levels over time are shown in Fig. 23. For AD2CD2-19M_SC terminal half-lives of 12.45 h and 11.58 h were detected for chromogenic activity and FVIII antigen, respectively and based on the mean of individual animals. In comparison, Refacto AF resulted in terminal half-lives of 6.48 h for chromogenic activity and 6.08 h for FVIII antigen. Thus, a half-life prolongation of approx. 2-times was verified in this model. An additional evaluation using the median instead of mean resulted in half-life extensions of approx. 3-times.

Furthermore, a pharmacokinetic study testing the AD2CD2-19M_SC molecule was performed in Gottingen Minipigs. Three animals per group were injected with 30 U FVIII antigen/kg body weight with either (I) ReFacto AF + 1% human serum Albumin (HSA), (II) ReFacto AF + 10% HSA, (III) AD2CD2_SC + 1% HSA, (IV) AD2CD2_SC + 10% HSA, (V) AD2CD2-19M_SC + 1% HSA, (VI) AD2CD2-19M_SC + 10% HSA via the ear vein. Blood samples were taken predose, 4, 12, 36, 48, and 120 h post administration and citrate plasma was isolated immediately by centrifugation. Bioanalytical sample measurement was performed by an adapted FVIII antigen ELISA as described above. Evaluation by non compartment analysis (Figure 24, showing only groups II, IV, and VI for clearity) obtained half-lives for (I) ReFacto AF + 1% HSA of 7.1 h, (II) ReFacto AF + 10% HSA of 6.4 h, (III) AD2CD2_SC + 1% HSA of 18.6 h, (IV) AD2CD2_SC + 10% HSA of 20.7 h, (V)

AD2CD2-19M_SC + 1% HSA of 19.2 h, and (VI) AD2CD2-19M_SC + 10% HSA of 21.0 h. Thus, besides the half-life extension of approx. 3-fold for AD2CD2_SC over ReFacto AF in this model, a similar or even higher half-life extension was observed for AD2CD2-19M_SC.

AD2CD2-19M_SC was additionally tested for its in vivo functionality using a tail vein transection assay as described for pharmacodynamics studies. Hemophilia A mice (Jackson No. B6; 129S4-F8<tm1Kaz>/J) were intravenously injected with different doses of AD2CD2-19M_SC, covering 200 U/kg (group 1), 70 U/kg (group 2), 20 U/kg (group 3),

7 U/kg (group 4), and 2 U/kg (group 5) (all based on chromogenic FVIII activity) or formulation buffer (group 6) (n=10 mice per group). Non-hemophilia C57BL/6NCrl mice were used as control (group 7). The tail vein transection assay was performed 30 min post test item administration. Weight loss through bleeding, bleeding time and Hb amount by OD550 were analyzed as readouts. Additional plasma sampling (0.25 h p.a. by retro-orbital withdraw and after the experiment) have been done for analysis of FVIII activity. As shown in Fig. 25, AD2CD2-19M_SC reduced the total bleeding time to that of control mice in a dose concentration-dependent manner, clearly indicating its in vivo functionality.

In order to evaluate if de-immunized FVIII-ABD fusion proteins maintain a certain FVIII activity in the presence of inhibitory anti-FVI 11 antibodies originally raised against WT or B- domain truncated FVIII (bypassing activity), a modified Nijmegen-Bethesda assay was performed. The Bethesda assay is widely used to quantitate the concentration of a factor VIII inhibitor (inhibitory antibody). 1 Bethesda Unit (BU) is defined as the amount of an inhibitor that will neutralize 50% of 1 unit of FVIII activity in normal plasma after 120 minutes incubation at 37°C. Therefore, five different anti-FVI 11 antibodies (ESH-8, GMA-8009, GMA- 8015, GMA-8026 and CL20035AP), all having inhibitory probabilities to human FVIII activity, were spiked 1:100 into imidazole buffer (Siemens Healthcare Diagnostics, Germany, #OQAA33), which served as stock solutions. Recombinant FVIII variants ReFacto AF, AD2CD2_SC, and AD2CD2-19M_SC were spiked to a final concentration of 1 U/mL into FVIII-depleted plasma (Siemens Healthcare Diagnostics, Germany, #OTXW17). Standard human serum (Siemens Healthcare Diagnostics, Germany, #ORKL17) was reconstituted in imidazole buffer resulting in a FVIII activity of 1 U/mL serving as further control. Anti-FVI 11 antibody stocks were diluted 1:2 up to 1:1024 (1:2 serial dilutions) in FVIII-depleted plasma containing the FVIII products. Additionally, each FVIII product diluted 1:2 with FVIII-depleted plasma was determined as baseline FVIII activity (should result in approx. 0.5 U/mL). A FVIII- inhibitor plasma standard (Technoclone, Austria, #5159008,16.0 BU/ml) diluted 1:2 to 1:128 (1:2 serial dilution series) with FVIII-depleted plasma was used as positive control. All samples were incubated for 2h at 37°C and the activity was determined by chromogenic FVIII activity measurements. The remaining FVIII activity within each samples was calculated by the following formula:

Chromogenic FVIII activity sample [U/mL] / chromogenic FVIII activity baseline [U/mL] * 100

Subsequently, Bethesda units were calculated in remaining activity ranges of 25 - 75% using the following formula:

(2 - Log(remaining FVIII activity) / 0.30103*dilution factor

Afterwards, the Bethesda units of each sample were divided by the Bethesda units of the positive control of each run for a more stringent comparison.

FVIII bypassing activity results of AD2CD2_SC and AD2CD2-19M_SC in comparison to a standard human plasma (SHP) and ReFacto AF against the five inhibitory anti-FVIII antibodies ESH-8, GMA-8009, GMA-8015, GMA-8026, and CL20035AP are shown in Fig.

26. In general, highest FVIII inactivation was observed for SHP followed by ReFacto AF. In contrast, the FVIII activity of AD2CD2_SC and AD2CD2-19M_SC was affected to a much lower extend by all anti-FVIII inhibitors. Interestingly, the anti-FVIII antibody GMA-8009 had a high inhibitor potential against SHP and ReFacto AF, a medium inhibitory potential to AD2CD2_SC and only a marginally inhibitory potential against AD2CD2-19M_SC indicating the elimination of a B cell epitope by one of the introduced de-immunizing mutations.

Identification of peptides presented on HLA-DR on the surface of antigen presenting cells ReFacto AF, AC-19M_sc and AD2CD2-19M_sc were analysed using the major histocompatibility complex-associated peptide proteomics (MAPPs) technology to identify peptides presented by MoDC via HLA-DR (as described in Webster, C. I. et al. , 2016. mAbs, Vol. 8, No. 2, pp. 253-63). Briefly, MoDC were derived from a cohort of 10 healthy donors and incubated with the test samples. Cells were then lysed, HLA-DR / peptide complexes immunocaptured, peptides eluted and sequenced by mass spectrometry (MS). MS data analysis identified peptides unique to individual donors and common across multiple donors. iTope™ analysis was subsequently used to predict core binding 9mers in the presented peptide clusters.

The frequency of each cluster in the cohort was calculated according to the equation below: number of donors common to a cluster

Cluster frequency within cohort = - - - - - — - x 100 total number of donors

As expected, all peptides identified for ReFacto AF were from wild type FVIII i.e., of self origin with respect to healthy donors, but they can in hemophilia A patients trigger inhibitor development as FVIII peptides are recognized as foreign. In total, for ReFacto AF 40 peptide clusters were observed with a frequency of 10 to 80 % of donors. For AC-19M_SC the total number of clusters was reduced to 37, even though new clusters arose. 15 of these 37 clusters demonstrated a reduced donor frequency compared to ReFacto AF, while 6 out of these 15 clusters were even eliminated in AC-19M_SC compared with ReFacto AF (see Fig. 27). On the contrary, 8 of said 37 clusters identified for AC-19M_SC had a higher frequency compared to ReFacto AF, where 3 of these 8 were neoepitopes. Interestingly, for AD2CD2- 19M_SC an even lower total cluster number of 31 was found. 24 of these 31 clusters demonstrated a reduced donor frequency compared to ReFacto AF, while 11 out of these 24 clusters were totally eliminated. The number of clusters with an increased frequency compared to ReFacto AF was only 3, including 2 neoepitopes.

Interestingly, several of the incorporated deimmunizing mutations eliminated or at least reduced the frequency of occurrence in the donors. However, the analysis further indicates that L160S, F555H and S2315T mutations may lead to a presentation of the respective FVIII epitopes. In one embodiment, the invention thus provides FVIII-ABD proteins of the invention comprising 16 de-immunizing mutations, namely, the 19 preferred mutations (19M) except for the L160S, F555H and S2315T mutations. Preferably, thus, said protein comprises L160, F555 and S2315. This combination of mutations is designated 16M.

6. Deimmunization of junction regions between FVIII. linkers and albumin-binding domains In order to further deimmunize the junction regions generated by fusions between the FVIII, linker and ABD sequences, in silico assessment was performed by two different approaches.

The first approach was utilizing the EpiMatrix tools as described in section 4 above. Therefore, non-FVIII sequences of AD2CD2-19M_SC with an overlap of 8 amino acids into the FVIII sequence were isolated in silico and potentially immunogenic clusters were identified using the EpiMatrix tools. This also included investigations on homology with endogenous proteins applying the JanusMatrix (see above).

Individually selected and most promising deimmunizing mutations from the in silico assessment were incorporated using VectorNTI (Thermo Fisher Scientific, Massachusetts, USA) into the DNA Sequences of AD2CD2_SC and AD2CD2-19M_SC resulting in variants D01-D11 and E01-E11, respectively, carrying mutations as outlined in Table 10. Full FVIII sequences were synthesized and cloned into the target vector. Expansion of respective plasmids was performed in transfected E. coli K12 followed by plasmid preparation.

Table 10: Mutations identified by EpiMatrix analysis and used for deimmunization of junction regions between FVIII, ABD and linker regions for AD2CD2_SC (SEQ ID 48) and AD2CD2- 19M_SC (SEQ ID NO: 114). Positions in this table relate to SEQ ID NO 48 or SEQ ID NO: 114 (wherein this designation is identical). The respective substitutions can also be introduced into other AD2CD2_SC proteins, e.g., AD2CD2-16M_SC corresponding to SEQ ID NO: 114, but with L160, F555 and S2315 (these three positions referring to positions relative to SEQ ID NO:1).

Cultivation of CAP-T cells and expression of the FVIII encoding plasmids by transient transfection was done as described elsewhere in this document. In order to verify expression levels and functionality of said junction de-immunized FVIII proteins fused with albumin binding domains, respective plasmids were nucleofected into CAP-T cells. 4-day cell culture supernatants were tested for chromogenic FVIII activity and FVIII antigen levels according to the above-described methods. Results are demonstrated in Fig. 28. All junction deimmunized AD2CD2_SC variants D01, D02, D03, D04, D05, D07, D08, D09, D10, and D11, were capable of inducing substrate processing in the chromogenic assay, demonstrating expression and secretion of every variant as well as their in vitro functionality (Fig 28A). Specific chromogenic activities (ratio of chromogenic activity and FVIII antigen level) were in the range of 132 to 190 % (Fig 28B). All AD2CD2_SC based junction de-immunized variants were as effective as or even more effective than the initial AD2CD2_SC molecule in this in vitro assessment. All tested supernatants of junction deimmunized AD2CD2-19M_SC variants E01, E02, E04, E05, E06, E07, E08, E10, and E11, were found to provide a chromogenic FVIII activity demonstrating protein expression, secretion and general functionality. Observed FVIII chromogenic ranges were between 1.15 U/mL for E04_AD2CD2-19M_SC and 3.99 U/mL for E11_AD2CD2-19M_SC. For the backbone AD2CD2-19M_SC variant, a concentration of 1.98 U/mL was found. Specific chromogenic FVIII activities were in the rage of 113 to 187 % indicating a favorable protein quality for all of the AD2CD2-19M SC-based FVIII variants.

Even though all tested junction deimmunized AD2CD2_SC and AD2CD2-19M_SC based variants were functional in vitro, variants D08 and D11 as well as E08 and E11 may be preferred in terms of immunological aspects, as the highest number of potentially immunogenic clusters was eliminated in these variants. Thus for further investigations, variants D08_AD2CD2_SC and D11_AD2CD2_SC were produced in larger amounts by stable expressing CAP Go pools and supernatants were affinity purified as described above. In vivo functionality of junction de-immunized FVIII-ABD proteins

In vivo efficacy of favorable junction deimmunized D08_AD2CD2_SC and D11_AD2CD2_SC FVIII variants was assessed by a tail vein transection assay with hemophilia A mice as described above. At least 8 animals per group were dosed with either 20, 5, or 2 U/kg (adjusted to chromogenic FVIII activity) of D08_AD2CD2_SC or D11_AD2CD2_SC. 7 hemophilia A mice and 5 C57BI/6 mice (non-hemophilia) were treated with FVIII formulation buffer as positive and negative control. Tail vein transection was started 0.5 h after administration and results of the total bleeding times are demonstrated in Fig. 29. Hemophilia A mice treated with FVIII formulation buffer had a mean bleeding time of 29 min and 22 sec, while the mean bleeding time for C57BI/6 wild type mice was only 2:30 min. When D08_AD2CD2_SC was administered to hemophilia A mice, observed bleeding times were 3:03, 6:01, and 16:43 min respectively to the used dose of 20, 5, and 2 U/kg bodyweight. The reduction in bleeding time was significant for all doses compared to FVIII formulation buffer treated hemophilia A mice. For animals treated with 20, 5, and 2 U/kg bodyweight of D11_AD2CD2_SC, respective bleeding times of 6:47, 4:41, and 22:04 min were observed. A significant reduction in bleeding time was observed at 20 and 5 U/kg bodyweight. Taken together, this pharmacodynamics bleeding assay demonstrates the in vivo efficacy of both junction deimmunized variants D08_AD2CD2_SC and D11_AD2CD2_SC in a dose dependent manner.

Second approach on the deimmunization the FVIII junction regions

The second approach for deimmunization of the junction regions between FVIII, linkers and ABD was performed by utilizing Abzena’s iTope in silico analysis on the AD2CD2_SC amino acid sequence. In line with the MHC II peptide identification analysis described above, only one epitope was identified for the junction regions spanning the 9mer-core from F761 to S769. To deimmunize this epitope, several mutations were assessed by iTope analysis and considered in terms of maintaining the thrombin-cleavage site according to Gallwitz et. al. (Gallwitz M, Enoksson M, Thorpe M, Heilman L. The extended cleavage specificity of human thrombin. PLoS One. 2012;7(2):e31756. doi:10.1371/journal. pone.0031756). Resulting deimmunizing amino acid substitutions as outlined in Table 11 clearly demonstrated in silico a dramatically reduced binding to all MHC II types.

Table 11: Mutations identified by iTope analysis and used for deimmunization of junction regions between FVIII, ABD and linker regions for AD2CD2_SC (SEQ ID 48). The respective substitutions can also be introduced into other proteins of the invention, e.g., AD2CD2- 16M_SC corresponding to SEQ ID NO: 114, but with L160, F555 and S2315 (these three positions referring to positions relative to SEQ ID NO:1) or AD2CD2-19M_SC (SEQ ID NO: 114).

DNA constructs encoding AD2CD2_SC variants comprising one of the four mutations of Table 11, namely F01_AD2CD2_SC comprising F761A, F02_AD2CD2_SC comprising F761S, F03_AD2CD2_SC comprising P766E, and F04_AD2CD2_SC comprising S769D were generated in silico using VectorNTI. Full FVIII sequences were synthesized and cloned into the target vector. Expansion of respective plasmids was performed in transfected E. coli K12 followed by plasmid preparation.

Plasmids encoding for F01_AD2CD2_SC, F02_AD2CD2_SC, F03_AD2CD2_SC. F04_AD2CD2_SC and AD2CD2_SC were transiently transfected into CAP-T cells for expression and functionality assessment as described above. Supernatants of a 4 days batch culture in a 12.5 ml scale were analyzed for chromogenic FVIII activity and FVIII antigen levels. AD2CD2_SC-normalized results are demonstrated in Fig 30 A and B, respectively. No major differences in chromogenic FVIII concentrations have been found - all junction deimmunized variants were in the range of 101 to 110 % compared to AD2CD2_SC. In comparison, AD2CD2_SC-normalized FVIIII antigen levels were higher for said variants with ranges of 109 - 186 % as of AD2CD2_SC control. This led to a lower specific activity for the junction deimmunized FVIII variants F01_AD2CD2_SC, F02_AD2CD2_SC, and F03_AD2CD2_SC, which, however, still demonstrated remarkably good specific activity values of 240 to 313 %. In comparison, F04_AD2CD2_SC and AD2CD2_SC revealed specific activities of 448 and 439 %. Taken together, all four junction deimmunized FVIII variants of this second approach were expressed, and found to be functional in vitro. According to the in silico prediction of iTope analysis, variants F01_AD2CD2_SC and F02_AD2CD2_SC may be preferred.

Claims

Claims

1. A recombinant Factor VIII protein comprising a heavy chain portion and a light chain portion of Factor VIII and at least two albumin binding domains, wherein at least one albumin binding domain is C-terminal to the heavy chain portion and at least one albumin binding domain is C-terminal to the light chain portion, wherein, if the protein is a single chain protein, the albumin binding domain(s) C- terminal to the heavy chain portion is/are N-terminal to the light chain portion.

2. The recombinant Factor VIII protein of claim 1 that is a single chain protein, wherein the protein preferably is at least partly B domain deleted.

3. The recombinant Factor VIII protein of any one of the preceding claims, wherein two albumin binding domains are C-terminal to the heavy chain portion and two albumin binding domains are C-terminal to the light chain portion.

4. The recombinant Factor VIII protein of any one of the preceding claims, wherein albumin-binding domains are separated from the heavy chain portion and/or the light chain portion and/or other albumin-binding domains by a linker, wherein the linker is selected from the group comprising a) a linker comprising a Thrombin-cleavable linker that optionally has the sequence of SEQ ID NO: 39, and b) a linker comprising a glycine-serine linker that optionally has the sequence of SEQ ID NO: 40 or SEQ ID NO: 41 , and c) a linker comprising a Thrombin-cleavable linker flanked on each side by a glycine-serine linker that optionally has the sequence of SEQ ID NO: 42 or SEQ ID NO: 43.

5. The recombinant Factor VIII protein of any one of the preceding claims, wherein the albumin binding domain comprises a sequence according to SEQ ID NO: 44, wherein preferably the sequence is SEQ ID NO: 46.

6. The recombinant Factor VIII protein of any one of the preceding claims that optionally is a single chain protein, wherein said protein comprises a heavy chain portion having at least 90% sequence identity to aa20-aa768 of SEQ ID NO: 16 and a light chain portion having at least 90% sequence identity to aa769-aa1445 of SEQ ID NO: 16.

7. The recombinant Factor VIII protein of any one of the preceding claims that is a single chain protein comprising at least two albumin binding domains between the heavy chain portion and the light chain portion and at least two albumin binding domain C- terminal to the light chain portion, wherein the protein has at least 80% sequence identity to any of SEQ ID NO: 48, 49 or 51 , wherein the protein preferably has at least 80% sequence identity to SEQ ID NO: 48.

8. The recombinant Factor VIII protein of any one of the preceding claims, wherein the recombinant Factor VIII protein comprises at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79, I80, 1105, S112, L160, V184, N233, L235, V257, I265, N299, Y426, Y430, L505, F555, 1610, N616, L706, N754, K1837, R1936, S2030, S2037, N2038, S2077, M2123, F2215, K2226, K2258, V2313, S2315, V2333 and Q2335; wherein substitutions of N are independently selected from the group consisting of D, H, S and E; wherein substitution of I are independently selected from the group consisting of T and V; wherein substitutions of S are independently selected from the group consisting of A, N, G, T and E; wherein substitutions of L are independently selected from the group consisting of N, Q, F and S; wherein substitutions of V are independently selected from the group consisting of A and T; wherein substitutions of Y are independently selected from the group consisting of N, H and S; wherein substitutions of F are independently selected from the group consisting of H and S; wherein substitutions of K are independently selected from the group consisting of N, D, E, Q, S and T; wherein substitutions of R are independently selected from the group consisting of Q, H and S; wherein substitutions of M are selected from the group consisting of R, Q, K and T ; and/or wherein substitutions of Q are selected from the group consisting of R, D, E, H and K; wherein the positions are specified in relation to full length human Factor VIII molecule of SEQ ID NO: 1 ; and wherein the recombinant Factor VIII protein retains at least 50% coagulant activity, as determined in a chromogenic assay, compared to a Factor VIII protein consisting of SEQ ID NO: 60.

9. The recombinant Factor VIII protein of claim 8, wherein the amino acid substitutions are selected from the group consisting of Y748S, L171Q, S507E, N79S, I80T, 1105V, S112T, L160S, V184A, N233D, L235F, V257A, I265T, N299D, Y426H, Y430H,

L505N, F555H, I610T, N616E, L706N, N754D, K1837E, R1936Q, S2030A, S2037G, N2038D, S2077G, M2123K, F2215H, K2226Q, K2258Q, V2313A, S2315T, V2333A and Q2335H, wherein, preferably, the recombinant Factor VIII protein comprises 3-25 of said substitutions and the substitutions are located within different immunogenic clusters.

10. The recombinant Factor VIII protein of any of claims 8 or 9, comprising at least three amino acid substitutions at positions selected from the group consisting of Y748, L171, S507, N79, S112, L160, V184, N233, I265, N299, Y426, F555, N616, I632, L706, K1837, R1936, N2038, S2077, S2125, F2215, K2226, K2258, S2315, and V2333; wherein the at least three amino acid substitutions are preferably selected from the group consisting of Y748S, L171Q, S507E, N79S, S112T, L160S, V184A, N233D, I265T, N299D, Y426H, F555H, N616E, I632T, L706N, K1837E, R1936Q, N2038D, S2077G, S2125G, F2215H, K2226Q, K2258Q, S2315T and V2333A.

11. The recombinant Factor VIII protein of any of claims 8-10 comprising at least amino acid substitutions at positions N79, S112, L160, L171, V184, N233, I265, N299, Y426, S507, F555, N616, L706, and Y748, wherein preferably the substitutions are N79S, S112T, L160S, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, F555H,

N616E, L706N, and Y748S, optionally further comprising the amino acid substitution at position K1837, such as K1837E.

12. The recombinant Factor VIII protein of any of claims 8-11 comprising at least amino acid substitutions at positions N79, S112, L160, L171, V184, N233, I265, N299, Y426, S507, F555, N616, L706, Y748, N2038, S2077, S2315 and V2333, wherein preferably the substitutions are N79S, S112T, L160S, L171Q, V184A, N233D, I265T, N299D, Y426H, S507E, F555H, N616E, L706N, Y748S, N2038D, S2077G, S2315T and V2333A, optionally further comprising the amino acid substitution at position K1837, such as K1837E, e.g., having at least 90% sequence identity to a Factor VIII protein of SEQ ID NO: 114.

13. The recombinant Factor VIII protein of any of claims 8-10, wherein the protein comprises at least one, preferably all of L160, F555 and S2315, wherein the protein optionally has SEQ ID NO: 114 except for L160, F555 and S2315.

14. The recombinant Factor VIII protein of any of claims 8-10 or 13 comprising at least amino acid substitutions at positions N79, S112, L171, V184, N233, I265, N299, Y426, S507, N616, L706, Y748, N2038, S2077, and V2333, wherein preferably the substitutions are N79S, S112T, L171Q, V184A, N233D, I265T, N299D, Y426H,

S507E, N616E, L706N, Y748S, N2038D, S2077G, and V2333A, optionally further comprising the amino acid substitution at position K1837, such as K1837E, e.g., having at least 90% sequence identity to a Factor VIII protein of SEQ ID NO: 114.

15. The recombinant Factor VIII protein of any of claims 8-14 having at least 90 % sequence identity to a Factor VIII protein of SEQ ID NO: 63, wherein only the A1, a1, A2, a2, a3, A3, C1 and C2 domains are considered for determination of sequence identity, wherein the protein optionally has SEQ ID NO: 114.

16. The recombinant Factor VIII protein of any one of the preceding claims that is a single chain protein that is at least partly B domain deleted and comprises a processing sequence of SEQ ID NO: 5 wherein X is A or S, optionally, A.

17. The recombinant Factor VIII protein of any one of the preceding claims, comprising at least one, preferably all of substitutions F761G, F779G, F1632G, F858G, F1711G, S926G, F936G and N1625D, wherein the substitutions named relate to the positions relative to SEQ ID NO: 48.

18. The recombinant Factor VIII protein of any one of claims 1-16, comprising at least one, preferably all of substitutions P766Q, N772D, R784Q, S787G, R1637Q, S1640G, R863Q, S866G, R1716Q, S1719G, S926G, R941Q, S944G and N1625D, wherein the substitutions named relate to the positions relative to SEQ ID NO: 48.

19. The recombinant Factor VIII protein of any one of the preceding claims, wherein the protein is a fusion protein.

20. A nucleic acid encoding a recombinant Factor VIII protein of any one of the preceding claims, wherein said nucleic acid preferably is an expression vector suitable for expression of said recombinant Factor VIII protein in a mammalian cell, optionally, a human cell, such as a CAP cell.

21. A host cell comprising a nucleic acid of claim 20, wherein preferably the host cell is a mammalian cell comprising an expression vector suitable for expression of said recombinant Factor VIII protein in said cell.

22. A method of preparing a recombinant Factor VIII protein, comprising culturing the host cell of claim 21 under conditions suitable for expression of the Factor VIII protein and isolating the recombinant Factor VIII protein, wherein the method optionally comprises formulating the Factor VIII protein as a pharmaceutical composition.

23. A pharmaceutical composition comprising the recombinant Factor VIII protein of any of claims 1-19, the nucleic acid of claim 20 or the host cell of claim 21, optionally further comprising a biologically acceptable carrier and/or albumin, wherein the pharmaceutical composition preferably is for use in treatment of hemophilia A.

24. A method of treating hemophilia A, comprising administering an effective amount of a composition of claim 23 to a subject in need thereof.