Cystine knot domain fusion protein dimers

Abstract

The application relates to a protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a cystine knot domain (CK) fused to a first pharmaceutically active protein and the second fusion protein comprises a CK domain fused to a second pharmaceutically active protein, wherein the two fusion proteins are covalently linked by their CK domains. The application further relates to the individual fusion proteins, to the polynucleotide encoding the fusion proteins, the vector comprising the polynucleotide and the host cell comprising the vector.

Description

[0001] CYSTINE KNOT DOMAIN FUSION PROTEIN DIMERS

[0002] FIELD OF THE INVENTION

[0003] The invention relates to the improvement of the properties, in particular the physicochemical and pharmacokinetic properties of pharmaceutically active proteins, using fusions of cystine knot (CK) domains, in particular the CK domain of von Willebrand factor (CKVWF).

[0004] BACKGROUND OF THE INVENTION

[0005] The purposive arrangement of different IgG heavy and light chains provides bispecific antibodies that allow binding to two different antigens or two different epitopes on the same antigen. The combination of two different Fab arms in one antibody molecule significantly expanded the therapeutical options of monoclonal antibodies historically binding to just one antigen or epitope. Today, bispecific antibodies are largely used in tumor therapy and other indications. Basically, the combination of two arms with different binding epitopes allows (i) the inhibition of two cell surface receptors, (ii) simultaneous blocking of two ligands, (iii) cross-linking of two receptors or (iv) recruitment of immune competent cells to a tumor cell. Such elaborate strategies are not possible with classical antibody formats.

[0006] Inherently, the Fc-domain with its related effector functions is part of bispecific antibodies in the IgG format. The Fc effector functions might be beneficial or may be unwanted depending on the respective target indication. The three-dimensional orientation and distance of the antigen-binding domains of such antibodies are determined by the antibody format and IgG subclass used as a structural basis for the therapeutic drug.

[0007] Like in these bispecific antibodies, the mono-Fc domain may serve as a structural basis for the construction of bi- specific therapeutic drugs. Heterodimers formed by two Fc-fusion protein monomers allow the combination of the different protein moieties fused to the C- and / or N-termini of the two Fc-domain. For example, WO 2019 / 129053 A1 describes a fusion protein dimer using an antibody Fc region as the backbone. Both fusion proteins of the dimer comprise one or more single domain antibodies and optionally a cytokine.

[0008] Likewise, Fc domains have been fused to pharmaceutically active proteins like coagulation Factor VIII (FVIII) for increasing the half-life or adding a different functionality. Such FVIII fusion proteins are suitable for the treatment of Hemophilia A.

[0009] Hemophilia is a group of hereditary genetic disorders that impair the body's ability to control blood clotting or coagulation. In its most common form, Hemophilia A, clotting factor VIII (FVIII) is deficient. Hemophilia A occurs in about 1 in 5,000 to 10,000 male births. The FVIII protein is an essential cofactor in blood coagulation with multifunctional properties. The deficiency of FVIII can be treated with plasma-derived concentrates of FVIII or with recombinantly produced FVIII. The treatment with FVIII concentrates has led to a normalized life of the hemophilia patients.

[0010] Hemophilia A patients are treated with FVIII on demand or as a prophylactic therapy administered several times a week. For prophylactic treatment, 15-25 lll / kg bodyweight of FVIII is administered three times a week, which is necessary due to the constant need of FVIII and its short half-life in the blood system, which in humans is only about 11 hours (Ewenstein et al., 2004). The short circulatory half-life of FVIII and the associated frequent need for FVIII concentrate infusion is a major challenge in Hemophilia A therapy.

[0011] In the blood, under normal conditions, the FVIII molecule is associated with its cofactor von Willebrand factor (VWF), which stabilizes the FVIII molecule from different forms of degeneration. The non-covalent complex of FVIII and VWF has a high binding affinity of 0.2-0.3 nM (Vlot et al., 1996).

[0012] Historically, Hemophilia A has been treated with FVIII originating from human blood plasma. Moreover, since the 1990s, different recombinantly produced FVIII (rFVIll) proteins were marketed. However, neither the plasma-derived nor the recombinant produced FVIII proteins have optimal pharmacokinetic properties. Like many other therapeutic proteins, they are subject to metabolic turnover by peptidases and other clearance mechanisms, which significantly limit their in vivo half-life. Attempts for prolonging FVIII half-life include immunoglobulin Fc-fusion (efmoroctocog alfa, Eloctate), addition of polyethylene glycol (turoctocog alfa pegol, Esperoct; damoctocog alfa pegol, Jivi; rurioctocog alfa pegol, Adynovate), and a single-chain construct (lonoctocog alfa, Afstyla). All these technologies result in approximate 1.5 times half-life prolonged FVIII (reviewed by Tiede 2015). It is well documented that the FVIII molecule circulates in complex with VWF and both molecules are cleared simultaneously, predominantly via the VWF clearance pathways. Therefore, the half-life of FVIII is mainly determined by the half-life of VWF.

[0013] VWF and VWF fragments, containing FVIII binding sites, are known to stabilize FVIII against rapid clearance, proteolytic digestion, uptake by antigen presenting cells and to facilitate higher bioavailability after subcutaneous administration. As shown by Yee et al. (2014), the human VWF D'D3 domain is sufficient to stabilize FVIII in plasma. However, a D'D3-Fc fusion protein is able to extend FVIII half-life only in VWF- / - mice. In Hemophilia A mice, the D’D3-Fc construct does not result in FVIII half-life prolongation due to ineffective competition of the VWF protein fragments with endogenous VWF for FVIII binding.

[0014] WO 2014 / 011819 A2 describes successful half-life prolongation of a FVIII construct containing the D'D3 domain of VWF, the Fc domain of IgG and XTEN. Since this construct does not bind to endogenous VWF, the same half-life prolonging effect is seen in both, VWF / FVI Il-double knock-out and Hemophilia A mice.

[0015] Other approaches for increasing the half-life of therapeutic proteins include the genetic fusion of the therapeutic protein to a protein with naturally long half-life such as transferrin and albumin, or to protein domains such as the C-terminal peptide (CTP) of chorionic gonadotropin. As reviewed in Strohl et al (2015), different fusion proteins of a therapeutic protein and CTP have been developed and are presently in clinical trials. The therapeutic proteins include FSH (Elonva®), FVIIa, FIX, IFN-|3 and oxyntomodulin.

[0016] WO 2017 / 198435 A1 describes fusion proteins comprising a main protein, which is a mammalian protein, such as human VWF, or a fragment thereof and one or more extension peptides. The extension peptide contains a cluster of O-glycosylation sites with at least two O-glycosylated amino acids and may be derived from human VWF. Due to the extension peptides, the fusion protein has an increased half-life as compared to the main protein by itself, i.e. the mammalian protein or fragment thereof. The fusion protein may be used to increase the half-life of a binding partner, e.g. FVIII. The OCTA12 molecule, falling under the definition of WO 2017 / 198435 A1 , is a fusion of a fragment of VWF able to bind FVIII with high affinity. It has a markedly prolonged half-life in comparison to full-length VWF (terminal half-life up to approx. 200 h after subcutaneous administration in humans vs. approx. 18 h for full-length VWF) due to the absence of certain domains that are recognized by clearance receptors (e.g. A1 and D4 domains recognized by the SR-AI receptor). In addition, it contains three repeats of extension peptides, i.e. of a VWF-derived four-fold O-glycosylated 31 amino acid-long sequence.

[0017] In addition to creating fusion proteins by joining of two or more genes that originally coded for separate proteins and translation of the resulting fusion gene, other approaches for obtaining multi-functional proteins include the combination of proteins by dimerization. For example, WO 2013 / 156054A1 describes fusion proteins comprising a pharmaceutically active moiety and a heavy chain domain 2 (HD2) from IgM or IgE, which can be combined by dimerization of the HD2 domains.

[0018] SUMMARY OF THE INVENTION

[0019] The present invention is inter alia based on the finding that the cystine knot domain of human VWF (CKVWF), when isolated from the rest of the protein and fused to a pharmaceutically active protein, may be used to combine this pharmaceutically active protein with a second pharmaceutically active protein also fused to a CKVWF by dimerization of the fused CKVWF domains. The CKVWF domains not only lead to strong intermolecular binding, but also aid in the folding and increase the stability of the pharmaceutically active proteins. The dimerization is independent of the type of pharmaceutically active proteins fused to the CKVWF domains. Moreover, the inventors could show that not only CKVWF can be used for dimerization of pharmaceutically active proteins but also other cystine knot domains such as the CKNDP. This means that the domains can be used universally for the dimerization of proteins, in particular of pharmaceutically active proteins. Thus, according to a first aspect, the invention provides a protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a CKVWF fused to a first pharmaceutically active protein and the second fusion protein comprises a CKVWF fused to a second pharmaceutically active protein, wherein the two fusion proteins are covalently linked by their CKVWF domains.

[0020] For example, via CKVWF domain dimerization, it was possible to prepare novel heterodimers of a human VWF fragment fused to a first CKVWF (VWF-CKVWF) and human FVIII fused to a second CKVWF (FVI I I-CKVWF).

[0021] According to one embodiment of the first aspect the invention provides a protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a FVIII protein, preferably comprising a first linker, fused to a von Willebrand Factor cystine knot domain (CKVWF) by a second linker and the second fusion protein comprises a von Willebrand Factor (VWF) fragment fused to a CKVWF domain by a third linker, wherein the third linker is an engineered peptide, and wherein the two fusion proteins are covalently linked by their CKVWF domains.

[0022] The heterodimer contained highly O-glycosylated extension peptides (EPs) added to either one of the fusion proteins forming the dimer, such as into the linkers connecting FVIII or VWF with their respective CKVWF domains, into the linker connecting the FVIII heavy and light chains, or to the C- or N-terminus of any of the fusion proteins.

[0023] The resulting FVI I I-CKVWF + VWF-CKVWF (FVI I I-CKVWF-VWF) heterodimers show improved properties, i.e. increased expression levels, a higher storage stability and a prolonged circulatory half-life compared to FVIII alone. Thus, according to one embodiment of the first aspect, the first pharmaceutically active protein is a FVIII protein and the second pharmaceutically active protein is a VWF fragment.

[0024] According to an alternative embodiment of the first aspect the invention provides a protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a cystine knot domain (CK) fused to a first pharmaceutically active protein and the second fusion protein comprises a CK domain fused to a second pharmaceutically active protein, wherein the two fusion proteins are covalently linked by their CK domains, provided that the first pharmaceutically active protein and the second pharmaceutically active protein are not selected from the group consisting of VWF, FVIII and fragments thereof.

[0025] For example, the CK fusion protein dimer, in particular CKVWF fusion protein dimer formation was confirmed with a completely different pair of pharmaceutically active proteins. In this example, the first fusion protein comprises a CKVWF or CKNDP fused to a first pharmaceutically active single domain antibody (VHH1 -CKVWF) and the second fusion protein comprises a CKVWF fused to a second pharmaceutically active single domain antibody (VHH2-CKVWF).

[0026] The resulting VHH1 -CK-VHH2 dimers show improved properties, i.e. increased shear stress resistance and thermal stability in comparison to dimers formed by other dimerization domains such as immunoglobulin Fc domains. Thus, according to one embodiment of the first aspect, the first pharmaceutically active protein is a first VHH protein and the second pharmaceutically active protein is a second VHH protein. The single chain antibodies may bind to coagulation factors FIX and FX and may be able to generate mimetic FVIII activity.

[0027] In a second aspect, the invention relates to the use of a CKVWF domain pair to combine two pharmaceutically active proteins by dimer formation.

[0028] In a third aspect, the invention provides a fusion protein comprising a CKVWF domain, a linker and a pharmaceutically active protein, wherein the CKVWF domain is capable of covalently binding a second CKVWF domain.

[0029] In a fourth aspect, the invention relates to a polynucleotide encoding the first or second fusion protein according to the first aspect.

[0030] According to the fifth aspect, the invention relates to a vector containing the polynucleotide according to the fourth aspect.

[0031] In a sixth aspect, the invention relates to a host cell containing the polynucleotide according to the fourth aspect or the vector according to the fifth aspect, wherein the host cell is a mammalian cell. Finally, in a seventh aspect, the invention also relates to a pharmaceutical composition comprising the protein dimer according to the first aspect for use in the treatment or prevention of a bleeding disorder.

[0032] According to an eighth aspect, the invention relates to the use of one or more EPs for decreasing the aggregation tendency of a target protein, wherein the one or more EPs are fused to or inserted into the target protein.

[0033] FIGURES

[0034] Fig. 1 shows a schematic representation of the overall structure of the FVIII- CKVWF-VWF (FVIII-CKVWF + VWF-CKVWF) protein heterodimers.

[0035] Fig. 2 shows a schematic representation of FVIII-CKVWF-VWF heterodimers C9+C15, C10+C15, C11 +C15, C12+C15, C13+C15, C9+C14, C10+C14 and C11 +C14.

[0036] Linker 1 : Non-marked - contains furin cleavage site 1 (SEQ ID NO: 20); *- contains furin cleavage site 2 (SEQ ID NO: 101 ); **- Furin cleavage site deleted. C14: The discontinued linker between the VWF-propeptide and the VWF-fragment illustrates the post-translational processing by furin. EP indicates the human VWF-derived EP with SEQ ID NO: 2.

[0037] Fig. 3 shows a schematic representation of the FVIII-CKVWF-VWF heterodimers C12+C14, C13+C14, C11 +C25, C11 +C26, C13+C26, C27+C15, C27+C25 and C27+C26.

[0038] Linker 1 : Non-marked - contains furin cleavage site 1 (SEQ ID NO: 20); *- contains furin cleavage site 2 (SEQ ID NO: 101 ); **- furin cleavage site deleted. C14: The discontinued linker between the VWF-propeptide and the VWF fragment illustrates the post-translational processing by furin. ABV indicates the albumin binding VHH moiety with SEQ ID NO: 36.

[0039] Fig. 4 shows a schematic representation of the FVIII-CKVWF-VWF heterodimers C28+C15, C28+C25, C28+C26, C29+C15, C29+C25 and C29+C26. Linker 1 : Contains furin cleavage site 1 (SEQ ID NO: 20). ABV indicates the albumin binding VHH moiety with SEQ ID NO: 36. EP indicates the human VWF-derived EP with SEQ ID NO: 2.

[0040] Fig. 5 shows a schematic representation of the FVI I I-CKVWF-VWF heterodimers C30+C15, C30+C25, C30+C26, C31 +C15, C31 +C25, and C31 +C26. Linker 1 : Contains furin cleavage site 1 (SEQ ID NO: 20). ABV indicates the albumin binding VHH with SEQ ID NO: 36. EP indicates the human VWF- derived EP with SEQ ID NO: 2.

[0041] Fig. 6 (A) shows a column diagram indicating the FVIII activity (FVIII:C) of FVIII-

[0042] CKVWF-VWF heterodimers detected in culture medium supernatant after transient expression in Expi293F cells. (B), mean FVIII:C activity of FVIII- CKVWF fusion proteins C11 and C27 to C31 (differing in the number and position of EPs and in the presence and placement of an ABV), combined with three different VWF-CKVWF molecules (C15, C25 and C26, also differing in the presence and position of the ABV). Bars represent the arithmetic mean and SD of values obtained from triplicates.

[0043] Fig. 7 shows western blot and non-reducing SDS-PAGE analyses of the purified heterodimers C9+C15 and C11 +C15. A, B and C show FVI I I-CKVWF-VWF heterodimers after western blot with FVIII detection (A), VWF detection (B) or PAGE after Coomassie staining (C). M, molecular weight marker. Molecular weights are indicated in kDa.

[0044] Fig. 8 shows western blot and non-reducing SDS-PAGE analyses of purified heterodimeric fusion proteins. (A) to (C) show FVI I I-CKVWF-VWF heterodimers after western blot with FVIII detection (A), VWF detection (B) or PAGE after Coomassie staining (C). Numbers above each lane represent the combinations of fusion proteins: 1 - C27+C25, 2 - C27+C26, 3 - C30+C25, 4 - C31 +C25, 5 - C30+C26, 6 - C31 +C26, 7 - C31 +C15, 8 - C30+C15, 9 - C11 +C25, 10 - C11 +C26, 11 - C13+C26, 12 - C27+C15, 13 - C28+C26. M, molecular weight marker. Molecular weights are indicated in kDa. Fig. 9 shows a column diagram indicating the normalized binding levels of purified heterodimers to full-length human VWF (fIVWF) detected by surface plasmon resonance (SPR). FIVWF was coated on a CM5 sensor chip, followed by injection of the purified FVIII-CKVWF-VWF heterodimers C9+C15, C11 +C15 or control proteins (OCTA12). The SPR signal detected 30s after injection stop was normalized to the binding of rFVH I (NUWIQ, set to 100 %). OCTA12 is a negative control that does not bind to fIVWF.

[0045] Fig. 10 shows a diagram of the FVIII activity (FVIII:C) detected in HemA mouse plasma after single IV administration of FVIII-CKVWF-VWF heterodimers C9+C15 and C11 +C15 according to the invention. rFVIll (NUWIQ) is used as reference. Data is presented as mean + / - SD obtained from 5 animals per timepoint.

[0046] Fig. 11 shows the results of pharmacokinetic (PK) experiments in WT mice. The diagram in (A) shows the level of FVIII activity (FVIII:C) 96 h after IV administration of 200 lU / kg bw of the different FVIII-CKVWF I VWF-CKVWF protein combinations (FVIII-CKVWF-VWF dimers). Each data point represents the FVIII:C level of one animal. Bars represent the geometrical mean and SD of values obtained from five animals. (B) shows the terminal half-life of the FVIII-CKVWF-VWF heterodimers formed. Each data point represents the plasma half-life calculated for one mouse cohort. Bars represent the geometrical mean and SD of all 5 cohorts.

[0047] Fig. 12 shows the impact of selective deletion of linkers and protease cleavage sites in the FVIII-CKVWF fusion proteins C32, C35, C42 and C43 on heterodimer activity. (A) shows the schematic structure of the expressed heterodimers. Linker 1 : **- furin cleavage site deleted. (B) shows the FVIII:C activity detected in the culture medium supernatant after transient expression of the heterodimers indicated.

[0048] Fig. 13 shows the impact of the selective deletion of linkers in the FVIII-CKVWF fusion proteins C27, C44, and C45 on heterodimer activity. (A) shows the schematic structure of the heterodimers expressed. Linker 1 : Non-marked - contains furin cleavage site 1 (SEQ ID NO: 20). (B) shows the FVI 11 : C activity in culture medium supernatant after transient expression of the heterodimers indicated.

[0049] Fig. 14 shows the impact of selective deletion of linkers in the VWF-CKVWF protein portion on heterodimer activity. (A) shows the schematic structure of the expressed heterodimers. C14 and C41 : The discontinued linker between the VWF-propeptide and the VWF fragment illustrates the post-translational processing by furin. Linker 1 : Non-marked - contains furin cleavage site 1 (SEQ ID NO: 20). (B) shows the FVIII:C activity in culture medium supernatant after transient expression of the heterodimers indicated. Left panel: comparison of VWF propeptide-containing VWF-CKVWF fusion proteins C14 and C41 expressed together with C27. Right panel: comparison of VWF propeptide-deleted VWF-CKVWF fusion proteins C15 and C40 expressed with C27.

[0050] Fig. 15 (A) shows a schematic representation of the FVI I I-CKVWF-VWF heterodimers C32+C15 and C37+C15. EP indicates the human VWF- derived EP with SEQ ID NO: 2. C32 and C37 differ only in the presence (C32) or absence (C37) of three EP repeats in linker 2. Linker 1 : **- furin cleavage site deleted. (B) shows the results of thrombin generation assays (TGA) of purified C32+C15 and C37+C15 after transient expression in 293F cells.

[0051] Fig. 16 shows the results of a tail-clip bleeding assay in Hemophilia A mice. Comparison of the total blood loss (calculated from photometric measurements of hemoglobin loss) after IV administration of 75 lU / kg bw of the FVI I I-CKVWF-VWF heterodimers C32+C15 and C37+C15 vs. recombinant B-domain deleted FVIII (BDD FVIII). Formulation buffer (without protein component) was used as negative control.

[0052] Fig. 17 shows a schematic representation of the configuration of heterodimers H1 , H2, H3, H5, H6 and H40 formed by the indicated fusion proteins comprising the single-domain antibodies VHH1 and VHH2 and the CK domains of human VWF (CKVWF) or Norrin (CKNDP). EP indicates the human VWF- derived EP with SEQ ID NO: 2. Fig. 18 shows non-reducing SDS-PAGE and western blot analyses of the VHH1 - CKVWF-VHH2 heterodimers H1 , H2, H3, H5 and H6 expressed in HEK293 cells. (A) Coomassie staining; (B) Western blot with His-Tag detection; (C) Western blot with Strep-Tag detection. Numbers below each lane represent the heterodimeric fusion proteins: 1 - H1 , 2 - H2, 3 - H3, 5 - H5, 6 - H6. Molecular weights are expressed in kDa.

[0053] Fig. 19 shows western blot analyses of the VHH1 -CKNDP-VHH2 heterodimer H40 expressed in HEK293 cells. Western blot analysis under non-reducing (lanes 1 to 3) and reducing conditions (lanes 4 to 6) with (A) His-Tag detection and (B) with Strep-Tag detection.

[0054] The positions of the H40 dimer in lanes 1 and of the H40 monomers (FP(FIX)20 and FP(FX)14) in lanes 6 is indicated by arrows. Lanes 2 to 5 show irrelevant proteins. Molecular weights are expressed in kDa.

[0055] Fig. 20 shows reducing and non-reducing SDS-PAGE analyses of purified VHH1 - CKVWF-VHH2 heterodimers. (A) SDS-PAGE analysis of heterodimeric fusion proteins H1 , H4 and H5 under non-reducing (lanes 1 to 3) and reducing conditions (lanes 5 to 7). Numbers below each lane represent the dimeric proteins: H1 - lanes 1 and 5, H4 - lanes 2 and 6, H5 - lanes 3 and 7. (B) SDS-PAGE analysis of the dimeric fusion proteins H2 and H3 under non- reducing (lanes 1 and 2) and reducing conditions (lanes 4 and 5). Lanes 1 and 4 show protein H2, lanes 2 and 5 show protein H3. Molecular weights are expressed in kDa.

[0056] Fig. 21 shows a schematic representation of the configuration of heterodimers H4 and TPP349. H4 is formed by the VHH-Fc fusion proteins VHH1 -lgFck-Twin- Strep and VHH2-lgFch-His. TPP349 is formed by the VHH-Fc fusion proteins VHH1 -lgFck and VHH2-lgFch.

[0057] Fig. 22 shows the dynamic light scattering (DLS) analyses of the VHH1 -CKVWF- VHH2 heterodimers H1 , H2, H3, H5, and H6. An overlay of the distribution of hydrodynamic radii of the heterodimers is depicted. Graph identities are indicated next to the peaks. Each graph represents the mean of triplicate measurements with the standard deviation visualized as shaded area around the graph line.

[0058] Fig. 23 shows the temperature induced un- and refolding of (A) the VHH1 -CKVWF- VHH2 dimers H1 , H5 and H6 and (B) the VHH1 -lgFc-VHH2 heterodimers H4 and TPP349 analyzed by DLS. The unfolding of the molecules is demonstrated in 3 panels for each molecule class. The upper panel shows the 350 / 330 nm ratio (graphs represent the mean of triplicate measurements with the standard deviation visualized as shaded area around the graph line), the middle panel shows the first derivative of the 350 / 330 nm ratio, and the bottom panel shows the cumulant radius. The temperature scale at the bottom is divided into two sections: the left part shows the ascending temperatures (up to 95°C) for testing protein unfolding, the right part represents descending temperatures where refolding may be observed.

[0059] Fig. 24 shows a comparison of the shaking stress stability of (A) TPP349 (VHH1 - lgFc-VHH2 heterodimer) and (B) of H1 (VHH1 -CKVWF-VHH2 heterodimer). On the left side of each panel, a SEC chromatogram overlay of the intact product treated under different shaking stress conditions is shown. On the right side of each panel the integrated area under the SEC peak as percentage of the unstressed sample is depicted.

[0060] Fig. 25 shows a comparison of the pH stability of VHH1 -CKVWF-VHH2 and VHH1 - lgFc-VHH2 heterodimers. (A) SEC chromatogram overlays of the unstressed (black) and stressed (treated, grey) sample is shown for the VHH1 -CKVWF-VHH2 dimer H5 (top) and for the VHH1 -lgFc-VHH2 dimer TPP349 (bottom). (B) The relative area of high molecular weight compounds (HMWC) comparing unstressed (black bars) and stressed samples (grey bars) is depicted.

[0061] Fig. 26 shows the analyses of FVIII:C activity of the VHH1 -CKVWF-VHH2 heterodimers H1 , H2, H3, H5, and H6 and of the VHH1 -lgFc-VHH2 dimers H4 and TPP349. The FVIII activity is expressed as the absorbance at 405 nm after baseline correction. The grey bars in the left part represent signals obtained with a human plasma FVIII standard (SHP). Black, white and grey bars in the right part represent signals obtained with the heterodimers at the protein concentrations indicated.

[0062] Fig. 27 shows a schematic representation of the configuration of heterodimers H17, H18, H19, H20, H22, H28, and H30 formed by the indicated fusion proteins of the single-domain antibodies VHH1 and VHH2 with the CK domain of human VWF (CKVWF) and half-life extension modules. ABV2 indicates the anti-RSA single domain antibody described in (van Faassen et al., 2020). 3xEP, 6xEP and 9xEP indicate three, six and nine consecutive repeats of the human VWF-derived EP with SEQ ID NO: 2.

[0063] Fig. 28 shows a schematic representation of the configuration of heterodimers H31 , H32, H33, H34, H35, and H36 formed by the indicated fusion proteins. Individual domains include the single-domain antibodies VHH1 and VHH2, the CK domain of human VWF (CKVWF) and the half-life extension modules ABV2 (anti-RSA single domain antibody), 1xEP, and 3xEP. 1xEP and 3xEP indicate one and three consecutive repeats of the human VWF-derived EP with SEQ ID NO: 2, respectively.

[0064] Fig. 29 shows reducing and non-reducing SDS-PAGE analyses of purified CKVWF- mediated heterodimers H3, H17, H18, H19, H20, H22, H28, H30 and H31 (lanes 1 to 9). Analyses in (A) were performed under non-reducing conditions and in (B) under reducing conditions. (C) SDS-PAGE analysis of H32, H33, H34, and H36 under reducing (lanes 1 to 4) and non-reducing conditions (lanes 5 to 8). Molecular weights are expressed in kDa.

[0065] Fig. 30 shows the results of pharmacokinetic (PK) experiments of the purified heterodimers H3, H5, H17, H18, H19, H20, H22, H28, H30, H31 , H32, H33, H34, and H36 in Sprague-Dawley / CD rats. The diagram shows the fusion protein dimer concentration in rat plasma (ng / mL) over time after IV administration of equimolar doses. Data are presented as mean levels + / - SD obtained from 3 animals per timepoint.

[0066] Fig. 31 shows the results of pharmacokinetic (PK) experiments of the purified heterodimers H5, H17, H20, H31 , H33, and H36 in Sprague-Dawley / CD rats. The diagram shows the dimer concentration in rat plasma (ng / mL) over time after SC administration of equimolar doses. Data are presented as mean levels + / - SD obtained from 3 animals per timepoint.

[0067] Fig. 32 shows the abundance of initial HMWC present in the untreated samples. The average values for different subgroups of heterodimers, grouped by the total number of EPs in the molecule, are indicated by light grey bars. HMWC content of dimers containing ABV2 is indicated by the dark grey bar. Error bars depict the standard deviation of all samples within the group.

[0068] DETAILED DESCRIPTION OF THE INVENTION

[0069] In order to provide a clear and consistent understanding of the specification and claims, and the scope to be given to such terms, the following definitions are provided.

[0070] Definitions

[0071] A "peptide" as used herein may be composed of any number of amino acids of any type, preferably naturally occurring amino acids, which, preferably, are linked by peptide bonds. In particular, a peptide comprises at least 3 amino acids, preferably at least 5, at least 7, at least 9, at least 12, or at least 15 amino acids. Furthermore, there is no upper limit for the length of a peptide. However, preferably, a peptide according to the invention does not exceed a length of 500 amino acids, more preferably it does not exceed a length of 300 amino acids; even more preferably it is not longer than 250 amino acids.

[0072] Thus, the term “peptide” includes “oligopeptides”, which usually refers to peptides with a length of 2 to 10 amino acids, and “polypeptides” which usually refers to peptides with a length of more than 10 amino acids.

[0073] A “protein” as used herein may contain one or more polypeptide chains and optionally additional components due to posttranslational modification or chemical modification in vitro. Proteins with more than one polypeptide chain are often expressed as one polypeptide chain from one gene and cleaved post-translationally. Thus, the terms “polypeptide” and “protein” are used interchangeably. The polypeptides and proteins as used herein include chemically synthesized proteins as well as naturally synthesized proteins which are encoded by genes. The polypeptides or proteins may be obtained from a natural source, such as human blood or produced in cell culture as recombinant proteins.

[0074] A “post-translational modification” (PTM) refers to the biochemical modification of a protein after its synthesis, i.e., after the protein has been translated from messenger RNA (mRNA) by a ribosome. These modifications typically occur in the endoplasmic reticulum or Golgi apparatus in eukaryotic cells and can alter a protein's function, stability, location, or interactions with other molecules. Examples of PTMs are glycosylation, ubiquitination, sumoylation, prenylation, sulfation, myristoylation, palmitoylation, attachment of a glycosylphosphatidylinositol (GPI) anchor and proteolytic cleavage.

[0075] The term “fusion protein” as used herein refers to proteins that are created through the joining of two or more genes that originally coded for separate proteins or protein fragments, wherein the components of the fusion protein are linked to each other by peptide bonds, either directly or through peptide linkers. The term “fused” as used herein refers to components that are linked by peptide bonds, either directly or via one or more peptide linkers.

[0076] A ’’pharmaceutically active protein" as used herein is a protein that demonstrates a specific biological, pharmacological, or therapeutic effect on a target, tissue, or system within a living organism. The protein exhibits its activity by modulating, interacting with, or regulating one or more molecular, cellular, or physiological process(es) associated with a particular disease, condition, or biological function. In the context of the invention a pharmaceutically active protein may also fulfil its functions by modifying the physicochemical or pharmacokinetic properties of a second pharmaceutically active protein. Such pharmaceutically active protein is also referred to as modulating protein.

[0077] According to the invention a “peptide linker” is a peptide connecting two protein elements of the fusion protein, in particular the FVIII heavy chain with the FVIII light chain or the FVIII light chain with the VWF portion. The peptide linkers are also referred to just as “linkers”. Peptide linkers contain structural amino acids to permit important domain interactions, reinforce stability, and reduce steric hindrance. In addition to the structural amino acids, the linkers may contain functional motifs. EPs may be considered as a protein element or as a part of a linker. The linkers may have a length of 2 to 200 amino acids.

[0078] An “engineered peptide” according to the invention is a peptide not derived from an organism but containing an artificial amino acid sequence, in particular a modular amino acid sequence obtained by combining amino acid sequences of different proteins, such as a flexibility motif or an EP.

[0079] According to the invention, a “flexibility motif” is a motif that enhances the flexibility of a peptide linker.

[0080] The term “therapeutic protein” as used herein relates to a subgroup of pharmaceutically active proteins, namely to proteins with a therapeutic effect, i.e. proteins used as active pharmaceutical ingredient.

[0081] According to the invention the terms “protein precursor” and “pro-protein” relate to an inactive protein (or peptide) that can be turned into an active form by post- translational modification, e.g. by enzymatic cleavage of a portion of the amino acid sequence.

[0082] The relatedness between two amino acid sequences or between two nucleotide sequences is described by the parameter "sequence identity". For purposes of the present invention, the degree of sequence identity between two amino acid sequences is determined using the Needleman-Wunsch algorithm (Needleman and Wunsch, 1970, J. Mol. Biol. 48: 443-453) as implemented in the Needle program of the EMBOSS package (EMBOSS: The European Molecular Biology Open Software Suite, Rice et al., 2000, Trends Genet. 16: 276-277), preferably version 3.0.0 or later. The optional parameters used are: a gap open penalty of 10, a gap extension penalty of 0.5, and the EBLOSUM62 (EMBOSS version of BLOSUM62) substitution matrix. The output of Needle labeled "longest identity" (obtained using the no brief option) is used as the percent identity and is calculated as follows:

[0083] (Identical Residues x 100) / (Length of Alignment - Total Number of Gaps in Alignment). For purposes of the present invention, the degree of sequence identity between two nucleotide sequences is determined using the Needleman-Wunsch algorithm (Needleman and Wunsch, 1970, supra) as implemented in the Needle program of the EMBOSS package (EMBOSS: The European Molecular Biology Open Software Suite, Rice et a / ., 2000, supra), preferably version 3.0.0 or later. The optional parameters used are: a gap open penalty of 10, gap extension penalty of 0.5, and the EDNAFULL (EMBOSS version of NCBI NUC4.4) substitution matrix. The output of Needle labeled "longest identity" (obtained using the nobrief option) is used as the percent identity and is calculated as follows:

[0084] (Identical Desoxyribonucleotides x 100) / (Length of Alignment - Total Number of Gaps in Alignment)

[0085] The term "recombinant" when used in reference to a cell, nucleic acid, protein or vector, indicates that the cell, nucleic acid, protein or vector has been modified by the introduction of a heterologous nucleic acid or protein or the alteration of a native nucleic acid or protein, or that the cell is derived from a cell so modified. Thus, for example, recombinant cells express genes that are not found within the native (non- recombinant) form of the cell, or express native genes at different levels or under different conditions than found in nature.

[0086] The term “half-life” as used herein is the time required for plasma / blood concentration to decrease by 50% after pseudo-equilibrium of distribution has been reached (in accordance with the definition in Toutain et al., 2005). The term “half-life” is also referred to as “circulatory half-life”, “terminal half-life” or “elimination half-life”.

[0087] As used herein, the terms "transformed”, "stably transformed", and "transgenic", used with reference to a cell means that the cell contains a non-native (e.g. heterologous) nucleic acid sequence integrated into its genome or carried as an episome that is maintained through multiple generations of progeny.

[0088] The term “fragment” as used herein refers to a polypeptide that has an amino- terminal and / or carboxy-terminal deletion of one or more amino acids as compared to the native or wild-type protein, but where the remaining amino acid sequence is identical to the corresponding positions in the amino acid sequence deduced from a full-length cDNA. Fragments are typically at least 50 amino acids in length.

[0089] The term "glycosylation" as used herein refers to the attachment of glycans to molecules, for example to proteins. Glycosylation may be an enzymatic reaction. The attachment formed may be through covalent bonds. Accordingly, a glycosylated polypeptide as used herein is a polypeptide to which one or multiple glycans are attached. The phrase "highly glycosylated" refers to a molecule such as an enzyme which is glycosylated at all or nearly all of the available glycosylation sites, for instance O-linked or N-linked glycosylation sites.

[0090] The term "glycan" as used herein refers to a polysaccharide or oligosaccharide, or the carbohydrate section of a glycoprotein or glycosylated polypeptide. Glycans may be homo- or heteropolymers of monosaccharide residues. Glycans typically contain at least three sugars and can be linear or branched. A glycan may include neutral sugar residues (e.g., glucose, N-acetylglucosamine (GIcNAc), N-acetylgalactosamine (GalNAc), galactose, mannose, fucose, arabinose, ribose, xylose, etc.), charged sugars (e.g. N-acetylneuraminic acid (sialic acid, NeuAc)), and / or modified sugars (e.g., 2'-fluororibose, 2'-deoxyribose, phosphomannose, 6'-sulfo-N-acetyl- glucosamine, etc.).

[0091] The term “O-glycans” as used herein refers to glycans that are generally found covalently linked to serine and threonine residues of mammalian glycoproteins. O-glycans may be a-linked via a GalNAc moiety to the -OH of serine or threonine by an O-glycosidic bond. Other linkages include a-linked O-fucose, |3-l inked O-xylose, a- linked 0-mannose, [3-linked O-GIcNAc, a- or [3-linked O-galactose, and a- or [3-linked O-glucose glycans.

[0092] According to the invention, the terms “O-glycosylation cluster”, “O-glycan cluster” and “cluster of O-glycosylated amino acids” are used interchangeably and relate to two or more of O-glycosylated amino acids.

[0093] The term “sialylated” as used herein refers to molecules in particular glycans that have been reacted with sialic acid or its derivatives. The transitional term “comprising”, which is synonymous with “including,” “containing”, or “characterized by”, is inclusive or open-ended and does not exclude additional, unrecited elements or method steps. The transitional phrase “consisting of” excludes any element, step, or ingredient not specified in the claim, except for impurities ordinarily associated therewith. When the phrase “consists of” appears in a clause of the body of a claim, rather than immediately following the preamble, it limits only the element set forth in that clause; other elements are not excluded from the claim as a whole. The transitional phrase “consisting essentially of” limits the scope of a claim to the specified materials or steps “and those that do not materially affect the basic and novel characteristic(s)” of the claimed invention. “A ‘consisting essentially of claim occupies a middle ground between closed claims that are written in a ‘consisting of’ format and fully open claims that are drafted in a ‘comprising’ format.”

[0094] In the context of the invention for practical reasons the term “glycosylated protein” such as the fusion protein is used in the singular form. Generally, in praxis, proteins occur in a composition of protein molecules of the same type. However, in the case of glycosylated proteins, glycosylation will not be identical in every molecule of the composition. For example, not all of the individual molecules of the composition may be glycosylated to 100 %. Moreover, differences in the glycans bound to a specific O- glycosylation site may arise. Accordingly, in the present application a reference to the “fusion protein” also relates to a composition of fusion protein molecules with identical amino acid sequences but variations in the O-glycan structure.

[0095] The terms “binding affinity” or “affinity” as used herein indicate the strength of the binding between two molecules, in particular a ligand and a protein target. Binding affinities are influenced by non-covalent intermolecular interactions between the two molecules such as hydrogen bonding, electrostatic interactions, hydrophobic interactions, and van der Waals forces.

[0096] An “immune response” as used herein relates to adaptive or innate immune response. The innate immune response refers to nonspecific defense mechanisms that are activated immediately or within hours of an antigen's appearance in the body. These mechanisms include physical barriers such as skin, chemicals in the blood, and immune system cells that attack foreign cells in the body. The innate immune response is activated by chemical properties of the antigen. The adaptive immune response refers to antigen-specific immune response. For this, the antigen first must be processed and recognized. Once an antigen has been recognized, the adaptive immune system creates a large number of immune cells specifically designed to attack that antigen.

[0097] The term “VWF” refers to the blood protein von Willebrand Factor.

[0098] An “extension peptide” or “EP” as used herein refers to a glycosylated peptide based on the O-glycosylation cluster 1 of VWF (SEQ ID NO:2).

[0099] The term “EP assembly” refers to two or more consecutive extension peptides.

[0100] “CKVWF” as used herein refers to a domain based on the cystine knot domain of VWF (SEQ ID NO: 1 ).

[0101] A “VHH domain” or “VHH fragment” or “VHH” as used herein refers to single-domain antibodies, which were engineered from heavy-chain antibodies found in camelids.

[0102] The terms "mammal" and "mammalian", as used herein, refer to any vertebrate animal, including monotremes, marsupials and placental, that suckle their young and either give birth to living young (eutharian or placental mammals) or are egg-laying (metatharian or nonplacental mammals). Examples of mammalian species include humans and other primates (e.g., monkeys, chimpanzees), rodents (e.g., rats, mice, guinea pigs) and ruminants (e.g., cows, pigs, horses).

[0103] The expressions “C-terminal” and “N-terminal" as used herein, either define the part of the protein close to the C-terminus and N-terminus or the position of one part of the fusion relative to the other. A first part that is “N-terminal of” a second part means that the first part is located closer to the N-terminus than the second part. A first part that is “C-terminal of’ a second part means that the first part is located closer to the C- terminus than the second part. Protein Dimers and CK Fusion Proteins

[0104] According to a first aspect, the invention provides a protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a cystine knot (CK) domain fused to a pharmaceutically active protein and the second fusion protein comprises a CK domain fused to a pharmaceutically active protein, wherein the two fusion proteins are covalently linked by their CK domains.

[0105] The cystine knot (CK) is a protein structural motif containing three disulfide bridges formed by pairs of cysteine residues. The sections of a polypeptide that occur between two of them form a loop through which a third disulfide bond passes, thereby forming a rotaxane substructure. The CK motif stabilizes the protein structure and is conserved in proteins across various species (Vitt et al. 2001 ; Sherbet 2011 ). There are three types of CK motifs, which differ in the topology of the disulfide bonds (Daly and Craik 2011 ): the Growth Factor CK (GF-CK), the Inhibitor CK (l-CK) common in spider and snail toxins and the Cyclic CK (C-CK), or cyclotide.

[0106] Accordingly, the CK domain may be a Growth Factor CK, an Inhibitor CK or a Cyclic CK. The GF-CK is found in TGF-[3, BMPs, VEGF, NGF, and similar growth factors. It is characterized by a ring formed by two disulfide bonds and the peptide backbone, with a third disulfide bond penetrating the ring. The l-CK is found in spider and snail toxins such as knottins, as well as some protease inhibitors. The C-CKs are found in plant-derived peptides and are characterized by a cyclic backbone. Due to their cyclic nature, they form extremely stable and knotted disulfide bonds.

[0107] VWF is a multimeric adhesive glycoprotein present in the plasma of mammals, which has multiple physiological functions. During primary hemostasis, VWF acts as a mediator between specific receptors on the platelet surface and components of the extracellular matrix such as collagen. Moreover, VWF serves as a carrier and stabilizing protein for pro-coagulant Factor VIII. VWF is synthesized in endothelial cells and megakaryocytes as a 2813 amino acid precursor molecule. Upon secretion into plasma, VWF circulates in the form of various species with different molecular sizes. These VWF molecules consist of oligo- and multimers of the mature subunit of 2050 amino acid residues. VWF can be found in plasma as multimers ranging in size approximately from 500 to 20.000 kDa (Furlan 1996). The precursor polypeptide, pre-pro-VWF, consists of a 22 -residue signal peptide, a 741 residue pro-peptide (domains D1 -D2) and the 2050-residue polypeptide found in mature plasma Von Willebrand Factor (Fischer et al., 1994). Full-length VWF is identified by entry P04275 of UniprotKB (entry version 224 of April 12, 2017). The domain organization of VWF is typically characterized as D3-TIL4-A1 -A2-A3-D4-C1 - C2-C3-CK. In VWF, the cystine knot domain is located at the C-terminal end of the protein.

[0108] The primary function of the cystine knot domain in VWF (CKVWF) is to provide structural stability to the protein. The cystine knot is formed by a unique arrangement of three disulfide bonds and a series of interconnected loops. This configuration creates a highly stable, compact, and rigid structure, contributing to the overall stability of the VWF protein. The CKVWF mediates dimerization of proVWF in the endoplasmic reticulum and is essential for long multimers required for hemostatic function.

[0109] Norrin is a secreted signalling protein encoded by the NDP (Norrie Disease Protein) gene. It plays a critical role in vascular and neural development, particularly in the eye, ear, and central nervous system. Unlike many other growth factors, Norrin functions primarily by activating the Wnt / [3-catenin signalling pathway. The CK domain of Norrin CKNDP has the amino acid sequence of SEQ ID NO: 100.

[0110] The inventors were able to extract the CKVWF domain from the surrounding structure of VWF and use this CKVWF domain to create a variety of fusion proteins including a VWF fragment with EPs fused to one CKVWF and a B-domain deleted FVIII (BDD- FVIII) fused to another CKVWF. The fused CKVWF domains form stable dimers that form covalent bonds with each other. Hence, the VWF fragment and the FVIII fusion proteins stably dimerize via the three disulfide bonds of their CKVWF domains. This provides a very strong connection of the two fusion proteins.

[0111] Disulfide bridges, also known as disulfide bonds, are covalent bonds formed between two cysteine residues within a protein or between two separate protein chains. These bridges play a crucial role in stabilizing the tertiary and quaternary structures of proteins, thereby contributing to their overall function and stability. The bond energy, or bond dissociation energy, is the energy required to break a specific covalent bond, in this case, a disulfide bond. For disulfide bonds, the bond energy typically ranges from 50 to 70 kcal / mol (210 to 290 kJ / mol). In the case of three disulfide bridges connecting two proteins, the total bond energy can be estimated by summing up the bond energies of each individual bridge. Assuming an average bond energy of 60 kcal / mol (251 kJ / mol) for each disulfide bond, the total bond energy for the three disulfide bridges of the CKVWF dimer would be approximately 180 kcal / mol (753 kJ / mol).

[0112] Accordingly, the bond energy of the fusion proteins may be in the range of 400 kJ / mol to 1150 kJ / mol, in the range of 450 kJ / mol to 1100 kJ / mol, in the range of 500 kJ / mol to 1050 kJ / mol, in the range of 550 kJ / mol to 1000 kJ / mol, in the range of 600 kJ / mol to 950 kJ / mol, in the range of 650 kJ / mol to 900 kJ / mol, in the range of 700 kJ / mol to 850 kJ / mol.

[0113] As shown in the examples, the cystine knot domain of VWF forms stable dimers in the context of fusion proteins. This is confirmed in Example 10 below with the CK domain of Norrin (CKNDP).

[0114] Through dimerization mediated by the fused CKVWF domains, it was possible to prepare novel heterodimers of VHH-CK fusion proteins in various layouts. These include dimer configurations with the CKVWF domain placed on the C-terminus or at the N-terminus of both VHH-CKVWF fusion proteins, or with the CKVWF domain placed C-term inally in the first and N-term inally in the second fusion protein, or vice versa. Moreover, the inventors have also isolated the CKNDP and confirmed that stable dimers of CK domain fusion proteins, in particular VHH-CKNDP fusion proteins can be formed.

[0115] The CKVWF and the CKNDP are both of the GF-CK type. According to one embodiment, the CK domains are GF-CK domains. According to a preferred embodiment, the CK domains are selected from CKVWF and CKNDP. According to another preferred embodiment, the CK domains are CKVWF domains.

[0116] According to one embodiment, the CKVWF domains have an amino acid sequence with an identity of at least 90 % to SEQ ID NO: 1. The identity may be 90%, 91 %, 92%, 93%, 94%, 95%, 95.5%, 96%, 96.5%, 97%, 97.5%, 98%, 98.5%, 99%, 99.5%, or 100%. According to one embodiment, the CKVWF domains have an amino acid sequence with an identity of at least 95 % to SEQ ID NO: 1. According to one embodiment, the CKVWF domains have an amino acid sequence with an identity of at least 98 % to SEQ ID NO: 1. Example 6 shows a heterodimer with a CKVWF domain variant, namely SEQ ID NO: 37 which has the following amino acid substitutions: A2785I, V2794A, and L2799A.

[0117] According to one embodiment, the CKNDP domains have an amino acid sequence with an identity of at least 90 % to SEQ ID NO: 100. The identity may be 90%, 91 %, 92%, 93%, 94%, 95%, 95.5%, 96%, 96.5%, 97%, 97.5%, 98%, 98.5%, 99%, 99.5%, or 100%. According to one embodiment, the CKNDP domains have an amino acid sequence with an identity of at least 95 % to SEQ ID NO: 100. According to one embodiment, the CKNDP domains have an amino acid sequence with an identity of at least 98 % to SEQ ID NO: 100.

[0118] In addition to the strong binding, the CKVWF domain provides several other functions that are beneficial for the fusion partner and, consequently, for the protein dimer according to the invention. For example, the protein dimers according to the invention (FVI I I-CKVWF + VWF-CKVWF (FVI I I-CKVWF-VWF)) show increased expression levels compared to BDD-FVIII.

[0119] The CK domain, in particular the CKVWF domain, has the advantage that it provides an additive stabilizing effect on its fusion partner. A comparison of pharmaceutically active proteins (VHH domains) connected to CKVWF domains with identical proteins bound to Fc domains revealed higher thermal stabilities and higher stabilities against shaking stress during culture and against pH changes during purification (see Example 12). In addition, the CK domain, in particular the CKVWF domain stabilizes the fusion partner during folding. When heated beyond the melting temperature, the proteins bound to CKVWF refolded (while identical VHHs bound to Fc domains did not) and showed practically no aggregation (see Example 12). Moreover, due to its small size, it allows efficient expression of the fusion protein.

[0120] Thus, dimerization of two jointly acting pharmaceutically active proteins, such as a single domain antibody binding to coagulation factor IX (FIX) and a second single domain antibody binding coagulation factor X (FX) or FVIII and a FVIII binding VWF fragment, via the CK domain, in particular the CK WF domain, has the advantage that the pharmaceutically active proteins form stable covalently linked dimers in the cell which stabilize one or both partners and allow effective expression and transport to the cell supernatant.

[0121] The CK domain, in particular the CKVWF, serves as a structural basis for the construction of bi- up to tetraspecific therapeutic drugs. Heterodimers formed by two CK-fusion protein monomers, in particular CKvwF-fusion protein monomers allow the combination of up to four different protein moieties fused to the C- and / or N-termini of the two CK domains, in particular two CKvwF-domains. The fusion of proteins to either end, i.e. to the N- or C-terminus of the CK domain (CKvwF-domain), allows different three-dimensional orientations and distances, which can be further varied by the introduction of protein linkers with individual properties. The fused proteins might have different functional properties, e.g. provide binding to specific targets, enzymatic activity, inhibitory properties, half-life-prolongation, receptor activation or clustering etc. The availability of four fusion sites within the CK domain pair, in particular the CKvwF-domain pair allows the construction of two fusion proteins with up to four functionally different protein moieties.

[0122] Accordingly, the CKVWF domain fusion protein dimers, in particular the CKVWF domain fusion protein dimers are a highly modular system allowing the combination of several pharmaceutically active proteins. Thereby, several therapeutic functions can be added, including modulating proteins modifying the function of other pharmaceutically active proteins. However, the protein dimer according to the invention may not only include two or more pharmaceutically active proteins. Due to the modular set up also other functional moieties may be introduced into each of the fusion proteins and therefore into the protein dimer according to the invention. As shown in the examples, in particular an EP or other half-life prolonging moieties may be introduced.

[0123] For example, the insertion of an EP increases storage stability and circulatory half- life. Hence, according to one embodiment, the first fusion protein and / or the second fusion protein comprise(s) at least one copy of an EP. As shown in the examples, EPs with the sequence QEPGGLVVPPTDAPVSPTTLYVEDISEPPLH (SEQ ID NO: 2), i.e. the O-glycosylation cluster 1 of VWF (amino acids 1238-1268 of SEQ ID NO: 6) confer increased expression levels, an improved stability and a reduced tendency for aggregation to the fusion protein according to the invention. Thus, the EPs according to the invention have an amino acid sequence with an identity of at least 90 % to SEQ ID NO: 2. The identity may be 90%, 91 %, 92%, 93%, 94%, 95%, 95.5%, 96%, 96.5%, 97%, 97.5%, 98%, 98.5%, 99%, 99.5%, or 100%. According to one embodiment, the sequence identity to SEQ ID NO: 2 is preferably at least 95 %. According to one embodiment, the sequence identity of the EP to SEQ ID NO: 2 is at least 98 %. Example 6 shows a fusion protein with the sequence variant of SEQ ID NO: 2, namely SEQ ID NO: 42, which has the following amino acid substitution: G5A. According to one embodiment, the two or more EPs have a sequence identity to SEQ ID NO: 2 of 100 %. According to one embodiment, the first and / or second fusion protein contain(s) at least one copy of the EP. For example, the first and / or second fusion protein contains 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, 15, 16, 17, 18, 19, or 20 copies of the EP. As shown in Example 14, by increasing the number of EPs, the tendency of the fusion proteins to aggregate, i.e. forming High Molecular Weight Components (HMWC) is reduced. In particular, the level of HMWC decreases proportionally to the increase in number of EPs in the fusion protein or fusion protein dimer. A fusion protein with 1 copy of the EP aggregates less than a fusion protein without EPs. A fusion protein with 2 copies of the EP aggregates less than a fusion protein with 1 copy. According to one embodiment, the first and / or second fusion protein contain(s) at least 2 copies of the EP. A fusion protein with 4 copies of the EP aggregates less than a fusion protein with 2 copies. According to one embodiment, the first and / or second fusion protein contain(s) at least 4 copies of the EP. A fusion protein with 6 copies of the EP aggregates less than a fusion protein with 4 copies. According to one embodiment, the first and / or second fusion protein contain(s) at least 6 copies of the EP. A fusion protein with 9 copies of the EP aggregates less than a fusion protein with 6 copies. According to one embodiment, the first and / or second fusion protein contain(s) at least 9 copies of the EP.

[0124] According to one embodiment, the first fusion protein and / or the second fusion protein comprise(s) at least one half-life prolonging moiety. There is a variety of half-life prolonging moieties known in the art. The half-life prolonging moiety may be selected from an immunoglobulin Fc-domain, serum albumin or parts thereof, or an albumin binding molecule. The Fc-domain is the fragment crystallizable (Fc) region of the tail region of an IgG antibody that interacts with cell surface receptors called Fc receptors. This interaction as well as the slower renal clearance for larger sized molecules increases the half-life of the protein to which it is attached. An exemplary Fc domain is the Fc domain of IgG 1 . Several fusion proteins containing albumin to increase the half-life of a therapeutic protein have been described including Factor VII, FVIII and Factor IX fusions with albumin. According to one embodiment, a full-length human serum albumin (HSA) is added to the fusion protein. The half-life of albumin is also regulated by an Fc receptor, i.e. the neonatal Fc Receptor (FcRn). The HSA sequence added to the fusion proteins has preferably the sequence according to Uniprot entry P02768.

[0125] The albumin-binding molecule may be selected from the group consisting of antibodies, antibody fragments, antibody mimetics, and other albumin binding proteins or parts thereof. Albumin binding proteins and protein domains may include the native or engineered albuminbinding domain of streptococcal protein G, albumin binding fibronectin type III (Fn3) domains, albumin binding single-domain antibodies and engineered lipocalins.

[0126] According to one embodiment, the antibody fragment is selected from the group consisting of Fab fragments, F(ab’)2 fragments, Fab’ fragments and single domain antibodies, in particular VHH domains. Fab fragments are antibody fragments that consist of the variable regions of the heavy and light chains of an antibody, as well as the first constant region of the heavy chain. Fab fragments can be produced by enzymatic digestion of full-length antibodies with papain. F(ab')2 fragments are antibody fragments that consist of two Fab fragments linked together by a disulfide bond. F(ab')2 fragments can be produced by enzymatic digestion of full-length antibodies with pepsin. Fab’ fragments are antibody fragments that consist of the variable regions of the heavy and light chains of an antibody, as well as a portion of the constant region of the heavy chain. Fab’ fragments can be produced by enzymatic digestion of full-length antibodies with papain followed by reduction of the disulfide bonds connecting the heavy chains. These fragments are commonly used in research and diagnostic applications and methods for the generation of antibody fragments are not particularly limited and are known in the art.

[0127] A single-domain antibody (sdAb) is an antibody fragment consisting of a single monomeric variable antibody domain. Like a whole antibody, it is able to selectively bind to a specific antigen. With a molecular weight of only 12-15 kDa, single-domain antibodies are much smaller than common antibodies (150-160 kDa). The first single- domain antibodies were engineered from heavy-chain antibodies found in camelids and are known as VHH fragments or VHH domains (Hamers-Casterman et al., 1993). An alternative approach is to split the dimeric variable domains from common immunoglobulin G (IgG) from humans or mice into monomers. Although most research into single-domain antibodies is currently based on heavy chain variable domains, single-domain antibodies derived from light chains have also been shown to bind specifically to target epitopes.

[0128] The antibody mimetic according to the invention may be selected from the group consisting of single-chain variable fragments (scFv), affibodies, affilins, affimers, affitins, anticalins, DARPins, monobodies, and peptide aptamers.

[0129] Single-chain variable fragments (scFv) are a type of antibody fragment that consist of the variable domains of the heavy and light chains of an antibody linked together by a short peptide linker.

[0130] Affibodies are small protein scaffolds that are engineered to bind to specific targets with high affinity and specificity. They are based on the binding domain of protein A, which is a natural ligand for the Fc region of antibodies. Affilins are a type of small protein scaffold that are engineered to bind to specific targets with high affinity and specificity. They are based on the ubiquitin protein family. Affimers are a type of protein scaffold that are engineered to bind to specific targets with high affinity and specificity. They are based on the cysteine protease inhibitor family of cystatins, which function in nature as cysteine protease inhibitor. Affitins are a type of protein scaffold that are engineered to bind to specific targets with high affinity and specificity. They are derived from the DNA binding protein Sac7d. Anticalins are a type of protein scaffold that are engineered to bind to specific targets with high affinity and specificity. They are based on a protein called lipocalin, which is involved in the transport of small hydrophobic molecules. DARPins, or designed ankyrin repeat proteins, are a type of protein scaffold that are engineered to bind to specific targets with high affinity and specificity. They are based on the ankyrin repeat protein family. Monobodies are a type of antibody mimetic that consist of a single protein domain engineered to bind to specific targets with high affinity and specificity. They are based on the fibronectin type III domain. Peptide aptamers are a type of protein scaffold that consist of a short peptide sequence that is engineered to bind to specific targets with high affinity and specificity. They are often generated using phage display or other selection methods. Respective antibody mimetics and methods for producing the same are not particularly limited and are known in the art.

[0131] According to one embodiment, the half-life prolonging moiety is an albumin binding VHH domain (ABV). Albumin binding VHH domains are known in the art. One example of an albumin binding VHH domain is MSA21 described in EP 2316852 B1. According to one embodiment, the albumin binding VHH domain is the ABV shown in the examples with the SEQ ID NO: 36. The albumin binding VHH domain according to the invention may have an amino acid sequence with an identity of at least 90 % to SEQ ID NO: 36. The identity may be 90%, 91 %, 92%, 93%, 94%, 95%, 95.5%, 96%, 96.5%, 97%, 97.5%, 98%, 98.5%, 99%, 99.5%, or 100%. According to one embodiment, the sequence identity to SEQ ID NO: 36 is preferably at least 95 %. According to one embodiment, the sequence identity of the albumin binding VHH domain to SEQ ID NO: 36 is at least 98 %.

[0132] A half-life prolonging moiety may be fused to the N-terminus and / or the C-terminus of the pharmaceutically active protein in the first fusion protein. A half-life prolonging moiety may be fused to the N-terminus and / or the C-terminus of the pharmaceutically active protein in the second fusion protein. According to one embodiment, the half-life prolonging moiety is fused to the N-terminus and / or the C-terminus of the first pharmaceutically active protein and / or the second pharmaceutically active protein. According to one embodiment, the half-life prolonging moiety is fused to the N- terminus and / or the C-terminus of the CKVWF domain in the first fusion protein and / or the second fusion protein. Another option is that the half-life prolonging moiety forms part of a linker in the first fusion protein and / or the second fusion protein.

[0133] The components of the fusion protein may be connected by peptide linkers. The linkers preferably have a length in the range from 2 to 200 amino acids. The linker length may be 2 amino acids, 4 amino acids, 6 amino acids, 8 amino acids, 10 amino acids, 12 amino acids, 14 amino acids, 16 amino acids, 18 amino acids, 20 amino acids, 25 amino acids, 30 amino acids, 35 amino acids, 40 amino acids, 45 amino acids, 50 amino acids, 55 amino acids, 60 amino acids, 65 amino acids, 70 amino acids, 75 amino acids, 80 amino acids, 85 amino acids, 90 amino acids, 95 amino acids, 100 amino acids, 105 amino acids, 110 amino acids, 115 amino acids, 120 amino acids, 125 amino acids, 130 amino acids, 135 amino acids, 140 amino acids,

[0134] 145 amino acids, 150 amino acids, 155 amino acids, 160 amino acids, 165 amino acids, 170 amino acids, 175 amino acids, 180 amino acids, 185 amino acids, 190 amino acids, 195 amino acids, or 200 amino acids.

[0135] The peptide linkers may be rigid or flexible. The linkers may comprise flexibility motifs that give the linkers flexibility to allow for sufficient mobility. The most flexible amino acids are the [3-tum-forming amino acids glycine and serine (Huang and Nau, 2003). Hence, flexibility motifs may be glycine-serine repeats. Glycine-serine repeats are repeating units of glycine and serine, such as (GGS)n or (GGGGS)n. They are common in flexible linkers. The glycine provides flexibility because of its small size, while serine adds stability to the linker in aqueous solutions by forming hydrogen bonds with the water molecules, thereby reducing unfavorable interactions with other protein moieties (Chen et al., 2013). Glycine-serine repeats may be selected from (GS)n, (GGS)n, (GGGS)n, and (GGGGS)n. n may be an integer in the range of 1 and 10. n may be 1 , 2, 3, 4, 5, 6, 7, 8, 9, or 10. In agreement with the one letter amino acid code, G represents glycine and S represents serine. GGGS is SEQ ID NO: 75 and GGGGS is SEQ ID NO: 76.

[0136] Glycine-serine repeats are examples of flexibility repeats. However, other suitable flexibility motifs such as (GGGX)n, (GXG)n, Elastin-like motifs and (SG)n or (TG)n repeats are also suitable for the fusion proteins of the protein dimer of the invention.

[0137] (GGGX)n: Variations of the GGGGS motif with the fourth position occupied by a different amino acid (X), which can be serine, threonine, or another small, polar amino acid. This variation maintains flexibility while possibly altering the hydrophilicity, steric hindrance, or conformational preferences of the linker.

[0138] (GXG)n: This motif includes glycine residues separated by a different amino acid (X), which can be serine, alanine, or another small residue. The presence of glycine ensures flexibility, while the intervening amino acid can modulate the linker's properties. Elastin-like motifs: Sequences such as "(VPGXG)n", where V is valine, P is proline, G is glycine, and X can be any amino acid (often a small one like alanine), are inspired by the elastin protein and can confer elasticity and flexibility. These motifs can also undergo reversible phase transitions in response to changes in temperature, which can be useful in protein purification and in designing responsive materials. VPGXG is SEQ ID NO: 77.

[0139] (Ser-Gly)n or (Thr-Gly)n repeats: These motifs, with repeating units of serine or threonine followed by glycine, provide flexibility and a higher degree of hydrophilicity due to the presence of serine or threonine. They can be useful in increasing the solubility of the protein constructs.

[0140] To provide rigidity and a defined spatial orientation, other motifs such as Polyproline (PPI I) helices can be used.

[0141] Polyproline (PPII) helices: Sequences rich in proline can form a type of secondary structure known as a polyproline helix. While not as flexible as glycine-rich linkers, polyproline helices provide a rigid and extended structure that can space apart functional domains. They are often used when a more defined distance or orientation between domains is desired.

[0142] Moreover, the peptide-linkers may contain one or more copies of the EP. The peptide linkers may contain a half-life prolonging moiety, preferably an albumin binding VHH (ABV).

[0143] According to one embodiment, one of the pharmaceutically active proteins The FVIII protein contains a first linker (Linker 1 in Figure 1 ). According to one embodiment, the CKVWF domain of the first fusion protein is fused to the pharmaceutically active protein by a second linker (Linker 2 in Figure 1 ). The second linker may be a flexible linker or a rigid linker. According to one embodiment the linker length is in the range from 5 to 180 amino acids. According to one embodiment, the linker length is in the range from 10 to 160 amino acids. According to one embodiment, the linker length is in the range from 12 to 140 amino acids. Preferably, the second linker is a flexible linker. According to one embodiment, the second linker comprises a flexibility motif. The flexibility motif is selected from (GGS)n, (GGGS)n, and (GGGGS)n. In this embodiment, n is an integer in the range of 1 and 10. n may be 1 , 2, 3, 4, 5, 6, 7, 8, 9, or 10. In agreement with the one letter amino acid code, G represents glycine and S represents serine. These motifs give the second linker flexibility to allow sufficient interactions and in particular binding to the VWF and FVIII binding domains. According to one embodiment, the second linker contains a (GGGGS)2, and / or a (GGGGS)4 motif.

[0144] According to one embodiment, two consecutive copies of the GGGGS motif are located at the N-terminus and / or the C-terminus of the second linker. According to one embodiment, a (GGGGS)2 motif is located at the N-terminus. According to one embodiment, a (GGGGS)2 motif is located at the C-terminus. According to one embodiment, a (GGGGS)n motif with n > 2 is located at the C-terminus. According to one embodiment, (GGGGS)2(SEQ ID NO: 78), (GGGGS)4(SEQ ID NO: 79), (GGGGS)6(SEQ ID NO: 80) or (GGGGS)s (SEQ ID NO: 81 ) is located at the C- terminus of the second linker. Longer flexible linkers facilitate the interaction between the first pharmaceutically active protein and its binding partner or with the second pharmaceutically active protein. Thus, (GGGGS)n linkers with n > 2 are preferred. According to one embodiment, a (GGGGS)2 motif is located at each of the N-terminus and the C-terminus.

[0145] According to one embodiment, the second linker is cleavable. According to one embodiment, the second linker comprises a thrombin cleavage site. The advantage of the presence of a protease cleavage site is the possibility of separating the first pharmaceutical protein, the FVIII protein, from the second pharmaceutically active protein, e.g. the VWF fragment. The thrombin cleavage site may be defined by SEQ ID NO: 19. A thrombin cleavage site is chosen because it is part of the native FVIII sequence.

[0146] According to one embodiment, the second linker comprises at least one copy of the EP. As shown in Example 2, the FVI I I-CKVWF-VWF fusion protein dimers with EPs in the second linker, i.e. C27, C30 and C31 , show increased expression levels and prolonged half-lives. In addition, as shown in Example 9, such FVI I I-CKVWF-VWF fusion protein dimers comprising EPs in the second linker (i.e. C32) exhibit improved thrombin generation kinetics in vitro and in vivo (Fig. 15). The second linker may for example contain one, two, three, four, five, six, seven, or eight copies of the EP. The FVI I I-CKVWF proteins shown in the examples C27, C30, C31 and C32 have three EPs in the second linker. Thus, according to one embodiment, the second linker comprises at least two copies of the EP. According to one embodiment, the second linker comprises at least three copies of the EP. The EPs may be distributed over the length of the linker, with structural amino acids of the linker or other elements between them. Alternatively, two or more of the EPs may be assembled adjacently, i.e. in a consecutive order. According to one embodiment, all EPs in the second linker are assembled in a consecutive order. According to one embodiment, the second linker is formed by an assembly of EPs only. According to one embodiment, the second linker contains at least two copies of a flexibility motif, in particular the GGS, GGGS, or GGGGS motif C-terminal and / or N-terminal of an EP assembly. According to one embodiment, the second linker contains at least two copies of the GGGGS motif C- terminal of an EP assembly.

[0147] According to a preferred embodiment, the half-life prolonging moiety is part of the second linker. According to a preferred embodiment, the half-life prolonging moiety is an ABV and is part of the second linker. According to one embodiment, the first linker contains at least two copies of a flexibility motif, in particular the GGS, GGGS, or GGGGS motif C-terminal and / or N-terminal of an ABV. According to one embodiment, the first linker contains at least two copies of the GGGGS motif C-terminal and N- terminal of the ABV.

[0148] According to one embodiment, the CKVWF domain of the second fusion protein is fused to the pharmaceutically active protein by a third linker. The third linker preferably has a length in the range from 2 to 200 amino acids. According to one embodiment the linker length is in the range from 10 to 160 amino acids. According to one embodiment the linker length is in the range from 12 to 140 amino acids.

[0149] The third linker may be a flexible linker or a rigid linker. Preferably, the third linker is a flexible linker. According to one embodiment, the third linker comprises a flexibility motif, preferably selected from (GGS)n, (GGGS)n, and (GGGGS)n. In this embodiment, n is an integer in the range of 1 and 10. n may be 1 , 2, 3, 4, 5, 6, 7, 8, 9, or 10. In agreement with the one letter amino acid code, G represents glycine and S represents serine. These motifs give the third linker flexibility to allow sufficient interactions between the second pharmaceutically protein and its binding partner or with the first pharmaceutically active protein, and in particular binding to the VWF and FVIII binding domains. According to one embodiment, the third linker contains a (GGGGS)2, a (GGGGS)4 and / or a (GGGGS)e motif.

[0150] According to one embodiment, the third linker comprises at least one copy of the EP. The constructs used in the examples contain two EPs fused to the C-terminus of the VWF fragment, which therefore can be considered as part of the linker. The third linker may for example contain one, two, three, four, five, six, seven, or eight copies of the EP. According to one embodiment, the third linker comprises at least two copies of the EP. According to one embodiment, the third linker comprises at least three copies of the EP. The EPs may be distributed over the length of the linker, with structural amino acids of the linker or other elements between them. Alternatively, two or more of the EPs may be assembled adjacently, i.e. in a consecutive order. According to one embodiment, all EPs in the third linker are assembled in a consecutive order. According to one embodiment, the third linker is formed by an EP assembly only. According to one embodiment, the third linker contains at least two copies of a flexibility motif, in particular a GGS, GGGS, or GGGGS motif C-terminal and / or N- terminal of an EP assembly. According to one embodiment, the third linker contains at least two copies of the GGGGS motif C-terminal of an EP assembly.

[0151] According to one embodiment, the third linker does not contain any full domain of VWF and preferably does not contain any other part of the D’ and D3 domain than the EP. According to a preferred embodiment, the half-life prolonging moiety is part of the third linker. According to a preferred embodiment, the half-life prolonging moiety is ABV and is part of the third linker. According to one embodiment, the third linker contains at least two copies of the GGS, GGGS, or GGGGS motif C-terminal and / or N-terminal of an ABV. According to one embodiment, the third linker contains at least two copies of the GGGGS motif C-terminal and N-terminal of the ABV.

[0152] The first pharmaceutically active protein or the second pharmaceutically active protein may individually be a synthetic protein, a naturally occurring protein or a fragment of the latter. According to one embodiment, the first pharmaceutically active protein and the second pharmaceutically active protein are mammalian proteins or fragments thereof. According to one embodiment, the first pharmaceutically active protein or second pharmaceutically active protein are preferably individually a human protein or fragment thereof.

[0153] According to one embodiment, the pharmaceutically active proteins of the first and second fusion protein are identical or different. The pharmaceutically active proteins may individually be a blood clotting factor, a transport protein, a protease inhibitor, an immunoglobulin, a cell-related plasma protein, an apolipoprotein, a complement factor, a growth factor, an antiangiogenic protein, a highly glycosylated protein, a blood factor or another blood protein.

[0154] The blood clotting factor, in particular human blood clotting factor, is preferably selected from the group consisting of fibrinogen (Fl), prothrombin (Fll), tissue factor (Fill), Factor V (FV), Factor VII (FVII), Factor VIII (FVIII), Factor IX (FIX), Factor X (FX), Factor XI (FXI), Factor XII (FXII), and Factor XI 11 (FXIII), VWF, and ADAMTS13.

[0155] It is appreciated that the clotting factors Fl, Fll, FV, FVII, FVIII, FIX, FX, FXI, FXII, and FXIII can be in a non-active or an activated form. Thus, in the context of the invention, a reference to Fl, Fll, FV, FVII, FVIII, FIX, FX, FXI, FXII, and FXIII includes the activated forms Fla (fibrin), Flla (thrombin), FVa, FVIIa, FVIIIa, FIXa, FXa, FXIa, FXI la, and FXI I la, respectively, unless explicitly stated otherwise or if the activated form is logically excluded from the context. Thus, e.g., in this context Fl, Fll, FV, FVII, FVIII, FIX, FX, FXI, FXII, and FXIII may be read as Fl / Fla, Fll / Flla, FV / FVa, FVII / FVIla, FVIII / FVIlla, FIX / FIXa, FX / FXa, FXI / FXIa, FXII / FXIIa, and FXIII / FXIIIa.

[0156] The transport protein, in particular human transport protein, may be selected from albumin, transferrin, ceruloplasmin, haptoglobin, hemoglobin, and hemopexin.

[0157] According to one embodiment the mammalian protein is a protease inhibitor, in particular human protease inhibitor. Examples of such protease inhibitors are I3>- antithrombin, a-antithrombin, pre-latent-antithrombin, oxidized-antithrombin, 2- macroglobulin, C1 -inhibitor, tissue factor pathway inhibitor (TFPI), heparin cofactor II, protein C inhibitor (PAI-3), Protein C, Protein S, and Protein Z. Examples of immunoglobulins such as polyclonal antibodies and monoclonal antibodies are lgG1 , lgG2, lgG3, lgG-4, IgA, lgA1 , lgA2, IgM, IgE, IgD, and Bence Jones protein.

[0158] The cell-related plasma protein may be for example, fibronectin, thromboglobulin, or platelet factor 4. Examples of apolipoproteins are apo A-l, apo A-l I, and apo E.

[0159] Complement factors according to the invention are e.g., Factor B, Factor D, Factor H, Factor I, C3b-lnactivator, properdin, C4-binding protein and the like.

[0160] Examples of growth factors include Platelet-derived growth factor (PDGF), Epidermal growth factor (EGF), Transforming growth factor alfa (TGF-a), Transforming growth factor beta (TGF-[3), Fibroblast growth factor (FGF) and Hepatocyte growth factor (HGF).

[0161] Antiangionetic proteins include latent antithrombin, prelatent antithrombin, oxidized antithrombin and plasminogen.

[0162] Examples of highly glycosylated proteins are alfa-1 -acid glycoprotein, antichymotrypsin, inter-a-trypsin inhibitor, a-2-HS glycoprotein, or C-reactive protein. Blood factors may be, e.g., erythropoeitin, interferon, tumor factors, tPA, or G-CSF.

[0163] Other human blood proteins include histidine-rich glycoprotein, mannan binding lectin, C4-binding protein, fibronectin, GC-globulin, plasminogen / plasmin, a-1 microglobulin, C-reactive protein.

[0164] The pharmaceutically active proteins are, in particular, selected from VWF, prothrombin, fibrinogen, Fill, FV, FVII, FVIII, FIX, FX, FXI, FXII, FXIII, ADAMTS13, antithrombin, alpha-1 antitrypsin, C1 -inhibitor, anti-chymotrypsin, PAI-1 , PAI-3, a2- macroglobulin, TFPI, heparin cofactor II, protein C, protein S, protein Z, and fragments of the aforementioned proteins.

[0165] According to one embodiment, the first pharmaceutical protein and / or the second pharmaceutical protein is an antigen binding molecule. According to one embodiment, the antigen binding molecule binds to any one of FIX, FIXa, FX, VWF, activated protein C, protease nexin-1 , protein Z dependent protease, antithrombin, protein S, ADAMTS13, platelet Gpllb / ll la receptor, complement C5, complement C3, P-selectin, human serum albumin or FcRn.

[0166] The antigen binding molecule may be selected from the group consisting of an antibody, an antibody fragment, and an antibody mimetic. According to one embodiment, the antibody fragment is selected from the group consisting of Fab fragments, F(ab’)2 fragments, Fab’ fragments and single domain antibodies, in particular VHH domains.

[0167] The antibody mimetic according to the invention may be selected from the group consisting of single-chain variable fragments (scFv), affibodies, affilins, affimers, affitins, anticalins, DARPins, monobodies, and peptide aptamers.

[0168] In this regard, according to one embodiment, it is provided that the first pharmaceutically active protein and the second pharmaceutically active protein are not selected from the group consisting of VWF, FVIII and fragments thereof. Preferably, the CK domains are CKVWF domains as defined above.

[0169] According to one embodiment, the first pharmaceutical protein and / or the second pharmaceutical protein is a VHH domain. According to one embodiment, the VHH domain binds to any one of FIX, FIXa, FX, VWF, activated protein C, protease nexin 1 , protein Z dependent protease, antithrombin, protein S, ADAMTS13, platelet Gpllb / llla receptor, complement C5, complement C3, P-selectin, human serum albumin or FcRn.

[0170] According to one embodiment, the first pharmaceutically active protein is a VHH domain binding to FIX (VHH1 ). According to one embodiment, the VHH domain binding to FIX comprises an amino acid sequence with an identity of at least 90 %, The identity may be 90%, 91 %, 92%, 93%, 94%, 95%, 95.5%, 96%, 96.5%, 97%, 97.5%, 98%, 98.5%, 99%, 99.5%, or 100%. According to a preferred embodiment, the VHH domain binding to FIX comprises an amino acid sequence with an identity of at least 95 % to SEQ ID NO: 72. According to a more preferred embodiment, the VHH domain binding to FIX comprises an amino acid sequence with an identity of at least 98 % to SEQ ID NO: 72. According to one embodiment of the protein dimer the second pharmaceutically active protein is a VHH domain binding to FX (VHH2). According to one embodiment, the VHH domain binding to FX comprises an amino acid sequence with an identity of at least 90 % to SEQ ID NO: 73. The identity may be 90%, 91 %, 92%, 93%, 94%, 95%, 95.5%, 96%, 96.5%, 97%, 97.5%, 98%, 98.5%, 99%, 99.5%, or 100%. According to a preferred embodiment, the VHH domain binding to FX comprises an amino acid sequence with an identity of at least 95 % to SEQ ID NO: 73. According to a more preferred embodiment, the VHH domain binding to FX comprises an amino acid sequence with an identity of at least 98 % to SEQ ID NO: 73.

[0171] The heterodimer VHH1 -CKVWF-VHH2 forms a factor Villa-mimetic bispecific molecule, comparable to the factor Villa-mimetic bispecific antibody constituting the API of Hemlibra.

[0172] A further example is the combination of an immunoglobulin targeting the vascular endothelial growth factor A (VEGF-A) together with an immunoglobulin targeting delta-like ligand 4 (DLL4). By simultaneously targeting VEGF-A and DLL4, the heterodimer can aim to inhibit tumor angiogenesis and enhance the efficacy of anti- VEGF-A therapies. A known example combining these pharmaceutical activities is ABT-165, which is a dual variable domain immunoglobulin (DVD-lg) fusion protein that targets both VEGF-A and DLL4. According to one embodiment the first pharmaceutically active protein is a VHH domain binding to VEGF-A. According to one embodiment the second pharmaceutically active protein is a VHH domain binding to DLL4.

[0173] A further example is the combination of a first single-chain variable fragment (scFv) binding specifically to CD19, a cell surface protein found on B-cell lymphoblasts and a second scFv binding to CD3, a protein on the surface of T-cells. By bridging these two immune cell populations, the dimer facilitates the formation of an immunological synapse between T-cells and B-cell lymphoblasts, ultimately leading to the targeted killing of the cancerous B-cells. An example of a fusion protein with these two activities is Blinatumomab (Blincyto) a bispecific T-cell engager (BiTE) antibody construct used for the treatment of acute lymphoblastic leukemia (ALL). A further example is Ozoralizumab. Ozoralizumab is a 38 kDa humanized trivalent bispecific construct consisting of two anti-TNFa single domain antibodies and anti- HSA single domain antibodies for the treatment of inflammatory diseases.

[0174] Exemplary second linkers (with the corresponding sequence IDs), in particular for VHH-CKVWF protein dimers are Linker 10-1 (SEQ ID NO: 56), Linker 10-2 (SEQ ID NO: 57), Linker 10-3 (SEQ ID NO: 58) and Linker 10-4 (SEQ ID NO: 59). The amino acid sequences of these linkers are shown in Table 19 (below). A variant of Linker 2- 1 has the sequence of SEQ ID NO: 46. According to one embodiment, the amino acid sequence of the first linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15, SEQ ID NO: 16, SEQ ID NO: 45, SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58 and SEQ ID NO: 59. According to one embodiment, the amino acid sequence of the first linker is selected from SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15, SEQ ID NO: 16, SEQ ID NO: 45, SEQ ID NO: 46, SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58 and SEQ ID NO: 59.

[0175] Exemplary third linkers (with the corresponding sequence IDs) are Linker 10-1 (SEQ ID NO: 56), Linker 10-2 (SEQ ID NO: 57), Linker 10-3 (SEQ ID NO: 58), Linker 10-4 (SEQ ID NO: 59). The amino acid sequences of these linkers are shown in Tables 18 and 20 (below). According to one embodiment, the amino acid sequence of the third linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 17, SEQ ID NO: 18, SEQ ID NO: 55, SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58, and SEQ ID NO: 59. According to one embodiment, the amino acid sequence of the third linker is selected from SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58, and SEQ ID NO: 59.

[0176] A further example is the combination FVIII with a VWF fragment as shown in the examples below.

[0177] According to one embodiment of the first aspect the invention provides a protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a FVIII protein, preferably comprising a first linker, fused to a von Willebrand Factor cystine knot domain (CKVWF) by an second linker and the second fusion protein comprises a von Willebrand Factor (VWF) fragment fused to a CKVWF domain by a third linker, wherein the third linker is an engineered peptide and wherein the two fusion proteins are covalently linked by their CKVWF domains.

[0178] With the CKVWF domain-mediated dimerization, it was possible to prepare novel variants of VWF and FVIII, exploiting the stabilizing effects of VWF on FVIII upon binding. According to one embodiment, the first pharmaceutically active protein is an FVIII protein.

[0179] Factor VIII in humans is encoded by the F8 gene, which comprises 187.000 base pairs in six exons. The transcribed mRNA has a length of 9.029 base pairs and is translated into a protein of 2.351 amino acids from which 19 amino acids are removed. The FVIII molecule in humans is glycosylated with 25 N-glycosylation chains and 6 O-glycans on 31 amino acids (Kannicht et al., 2013).

[0180] After translation, the amino acid chain is cleaved by specific proteases leading to the formation of a heavy chain with about 200 kDa and a light chain with about 80 kDa. The domain organization is typically characterized as A1 -A2-B-A3-C1 -C2. The light chain is a composition of domains A3-C1 -C2. The heavy chain is composed of the domains A1 -A2-B. According to one embodiment, the FVIII protein comprises an FVIII heavy chain and an FVIII light chain. FVIII heavy chains found in plasma have a heterogeneous composition with molecular weights varying from 90 to 200 kDa. The reasons for this size variation are the heterogeneity in its glycosylation, the existence of splice variants and proteolytic products such as the B-domain depleted heavy chain A1 -A2. The amino acid sequence of the full-length FVIII is identified by amino acids 20 to 2.351 of P00451 of UniProtKB, sequence version 1 of July 21 , 1986.

[0181] The human FVIII heavy chain according to the invention contains at least the domains A1 and A2 and may further contain parts of the B-domain or the entire B-domain. The amino acid sequence of the human FVIII heavy chain without the B-domain is identified by SEQ ID NO: 3. The amino acid sequence of the human FVIII heavy chain including the B-domain is identified by SEQ ID NO: 4.

[0182] According to one embodiment, the FVIII heavy chain does not contain the FVIII B- domain (BDD-FVIII). In this case, the FVIII heavy chain in the first fusion protein has an amino acid sequence similar or identical to SEQ ID NO: 3. The heavy chain without the FVIII B-domain may comprise an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to SEQ ID NO: 3. Example 6 shows a fusion protein with the FVIII heavy chain sequence variant of SEQ ID NO: 38, which has the following amino acid substitution: V592A. According to one embodiment, the heavy chain is at least 95 % identical to SEQ ID NO: 3. According to one embodiment the heavy chain is at least 98 % identical to SEQ ID NO: 3.

[0183] According to one embodiment, the FVIII heavy chain contains the FVIII B-domain. In this case, the FVIII heavy chain of the fusion protein has an amino acid sequence similar or identical to SEQ ID NO: 4. The heavy chain with the FVIII B-domain may comprise an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to SEQ ID NO: 4. According to one embodiment the heavy chain is at least 95 % identical to SEQ ID NO: 4. According to one embodiment the heavy chain is at least 98 % identical to SEQ ID NO: 4.

[0184] According to one embodiment, the FVIII light chain comprises the domain organization A3-C1 -C2. The human FVIII light chain with the domain organization AS- CI -C2 has the sequence of SEQ ID NO: 5. The FVIII light chain of the fusion protein may have an amino acid sequence similar or identical to SEQ ID NO: 5. According to one embodiment, the FVIII light chain comprises an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to SEQ ID NO: 5. According to one embodiment, the light chain is at least 95 % identical to SEQ ID NO: 5. According to one embodiment, the light chain is at least 98 % identical to SEQ ID NO: 5. Example 6 shows a fusion protein with the FVIII light chain sequence variant of SEQ ID NO: 39 which has the following amino acid substitution: S1732T.

[0185] According to one embodiment, the C-terminus of the FVIII heavy chain is fused to the N-terminus of the FVIII light chain by a first linker. Linkers connecting the heavy and the light chain are known in the art. One example is the linker in NUWIQ®, i.e. SFSQNSRHQAYRYRRG (SEQ ID NO: 21 ). This linker comprises a sequence derived from the B-domain of FVIII. According to one embodiment, the first linker preferably comprises a sequence derived from the B-domain of FVIII.

[0186] The first linker may be a flexible linker or a rigid linker. According to one embodiment the linker length is in the range from 5 to 180 amino acids. According to one embodiment the linker length is in the range from 10 to 160 amino acids. According to one embodiment the linker length is in the range from 12 to 140 amino acids.

[0187] Preferably, the first linker is a flexible linker. According to one embodiment, the first linker comprises a flexibility motif. The flexibility motif may be selected from (GGS)n, (GGGS)n, and (GGGGS)n. In this embodiment, n is an integer in the range of 1 and 10. n may be 1 , 2, 3, 4, 5, 6, 7, 8, 9, or 10. In agreement with the one letter amino acid code, G represents glycine and S represents serine. These motifs give the first linker flexibility to allow sufficient interactions between the FVIII heavy and light chain.

[0188] According to one embodiment, the first linker is a cleavable linker, i.e. it contains a protease cleavage site. The advantage of the presence of a protease cleavage site is the possibility of intracellular processing of FVIII to its native two-chain conformation. According to one embodiment, the first linker comprises a furin cleavage site. A furin cleavage site is chosen because it is a naturally occurring cleavage site in the wild type FVIII. The furin cleavage site may have the amino acid sequence of SEQ ID NO: 20 or SEQ ID NO: 1.

[0189] According to one embodiment, the first linker of the FVI I I-CKVWF-VWF heterodimers is an engineered peptide.

[0190] According to one embodiment, the first linker comprises at least one copy of the EP. The first linker may for example contain, one, two, three, four, five, six, seven, or eight copies of the EP. The FVI I I-CKVWF-VWF heterodimers shown in the examples have three EPs in the first linker. Thus, according to one embodiment, the first linker comprises at least two copies of the EP. According to one embodiment, the first linker comprises at least three copies of the EP. The EPs may be distributed over the length of the linker, with structural amino acids of the linker or other elements between them. Alternatively, two or more of the EPs may be assembled adjacent, i.e. in a consecutive order. According to one embodiment, all EPs in the first linker are assembled in a consecutive order. According to one embodiment, the first linker is formed by an EP assembly only. According to one embodiment, the third linker contains at least two copies of a flexibility motif, in particular a GGS, GGGS, or GGGGS motif C-terminal and / or N-terminal of an EP assembly. According to one embodiment, the third linker contains at least two copies of the GGGGS motif C-terminal of an EP assembly.

[0191] According to one embodiment of the protein dimer, the first linker, the second linker and / or the third linker contain(s) at least two copies of the GGS, GGGS, or GGGGS motif on either side of an EP assembly and / or on either side of the half-life prolonging moiety.

[0192] Exemplary first linkers (with the corresponding sequence IDs) are Linker 1 -1 (SEQ ID NO: 8), Linker 1 -2 (SEQ ID NO: 9), Linker 1 -3 (SEQ ID NO: 10), Linker 1 -4 (SEQ ID NO: 11 ), and Linker 1 -5 (SEQ ID NO: 12), Linker 1 -6 (SEQ ID NO: 43), and Linker 1 - 7 (SEQ ID NO: 44). The amino acid sequences of these linkers are shown in Table 1 (below). A variant of Linker 1 -3 has the SEQ ID NO: 41. According to one embodiment, the amino acid sequence of the first linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 8, SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 11 , SEQ ID NO: 12, SEQ ID NO: 43, and SEQ ID NO: 44. According to one embodiment, the amino acid sequence of the first linker is identical to a sequence selected from SEQ ID NO: 8, SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 11 , SEQ ID NO: 12, SEQ ID NO: 41 , SEQ ID NO: 43, and SEQ ID NO: 44.

[0193] According to one embodiment, the second linker of the FVIII-CKVWF-VWF heterodimers is an engineered peptide. Exemplary second linkers (with the corresponding sequence IDs), in particular for FVIII-CKVWF-VWF protein dimers, are Linker 2-1 (SEQ ID NO: 13), Linker 2-2 (SEQ ID NO: 14), Linker 2-3 (SEQ ID NO: 15), Linker 2-4 (SEQ ID NO: 16), and Linker 2-5 (SEQ ID NO: 45). The amino acid sequences of these linkers are shown in Tables 2 and 14 (below). A variant of Linker 2-1 has the sequence of SEQ ID NO: 46. According to one embodiment, the amino acid sequence of the first linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15, SEQ ID NO: 16, SEQ ID NO: 45. According to one embodiment, the amino acid sequence of the first linker is selected from SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15, SEQ ID NO: 16, SEQ ID NO: 45, SEQ ID NO: 46, SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58 and SEQ ID NO: 59.

[0194] According to one embodiment, the third linker of the FVI I I-CKVWF-VWF heterodimers, i.e. the linker linking the VWF fragment and the CKVWF domain, does not contain the C3 domain of VWF. Preferably, the third linker does not contain the C3 and the C2 domain of VWF. According to one embodiment, the third linker does not contain the C3, C2 and the C1 domain. According to one embodiment, the third linker does not contain any full domain of VWF. According to one embodiment, the third linker does not contain any other part of the D’ and D3 domain than the EP. Exemplary third linkers (with the corresponding sequence IDs) in particular for FVI I I-CKVWF-VWF protein dimers, are Linker 3-1 (SEQ ID NO: 17), Linker 3-2 (SEQ ID NO: 18), and Linker 3-3 (SEQ ID NO: 55). The amino acid sequences of these linkers are shown in Tables 4 and 16 (below). A variant of Linker 3-1 has the sequence of SEQ ID NO: 74. According to one embodiment, the amino acid sequence of the third linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 17, SEQ ID NO: 18, and SEQ ID NO: 55. According to one embodiment, the amino acid sequence of the third linker is selected from SEQ ID NO: 17, SEQ ID NO: 18, SEQ ID NO: 55 and SEQ ID NO: 74.

[0195] The half-life prolonging moiety may be connected to the rest of the first or second fusion protein, i.e. to the first or second pharmaceutically active protein or one of the two CKVWF domains by a fourth linker. The fourth linker may be a flexible linker or a rigid linker. According to one embodiment the linker length is in the range from 5 to 180 amino acids. According to one embodiment the linker length is in the range from 10 to 160 amino acids. According to one embodiment the linker length is in the range from 12 to 140 amino acids.

[0196] According to one embodiment, the second pharmaceutically active protein is a VWF fragment. Thus, according to one embodiment of the first aspect, the pharmaceutically active protein of the first fusion protein is a FVIII protein and the pharmaceutically active protein of the second fusion protein is a VWF fragment. The resulting FVIII-CKVWF-VWF fusion protein heterodimers show improved properties, i.e. increased expression levels, a higher storage stability and a prolonged circulatory half-life compared to FVIII alone.

[0197] The human VWF according to the present invention has an amino acid sequence of any of the sequences of UniprotKB P04275, in particular SEQ ID NO: 6 (isoform 1 ). VWF contains two clusters of O-glycosylated amino acids. The first cluster of 0- glycosylated amino acids is found between amino acids 1238 to 1268 of SEQ ID NO: 6. The second cluster includes amino acids 1468 to 1487 of SEQ ID NO: 6.

[0198] According to one embodiment, the VWF fragment comprises an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to a section of SEQ ID NO: 6. According to one embodiment the VWF fragment is at least 95 % identical to a section of SEQ ID NO: 6. According to one embodiment the light chain is at least 98 % to a section of SEQ ID NO: 6.

[0199] In the fragment of human VWF does not contain any of the domains C3 and CK. Moreover, one or more of the domains A1 , A2, A3, D4, C1 , and C2, may be missing relative to the human mature VWF (TIL3-D3-TIL4-A1 -A2-A3-D4-C1 -C2-C3-CK). The VWF fragment may, for example, have a domain organization selected from the following group consisting of TIL3-D3-TIL4-A1 , TIL3-D3-TIL4-A1 -A2, TIL3-D3-TIL4- A1 -A2-A3, TIL3-D3-TIL4-A1 -A2-A3-D4, TIL3-D3-TIL4-A1 -A2-A3-D4-C1 , and TIL3- D3-TI L4-A1 -A2-A3-D4-C 1 -C2..

[0200] In this regard, the section of SEQ ID NO: 6 is, in particular, a section starting with amino acid 764 of SEQ ID NO: 6. Amino acids 764 to 1035 of SEQ ID NO: 6 contain the FVIII binding domain of VWF. The section may, for example, be a section as defined in WO 2015 / 185758 A2. As shown in WO 2015 / 185758 A2, the complex of FVIII and the VWF fragments as defined therein exhibit a reduced binding to phospholipid membranes compared to FVIII alone as well as a reduced binding to collagen III and heparin compared to the complex of FVIII and full-length VWF. The section of VWF preferably starting with amino acid 764 of SEQ ID NO: 6, preferably ends with an amino acid of SEQ ID NO: 6 in the range from 1905 to 2153. According to one embodiment the VWF fragment ends with an amino acid of VWF in the range from 2030 to 2153 of SEQ ID NO: 6. According to a further embodiment the VWF fragment ends with an amino acid of SEQ ID NO: 6 in the range from 2100 to 2153.

[0201] According to one embodiment, the VWF fragment comprises an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to SEQ ID NO: 7. The VWF fragment with the amino acid sequence of SEQ ID NO: 7 is based on the section of amino acids 764 to 1268 of SEQ ID NO: 6, with two amino acid substitutions, namely C1099A and C1142A. The replacement of the two cysteines by alanines abolishes the ability of the VWF fragment to form multimers. This modified fragment of VWF is used in the fusion proteins in the examples. According to one embodiment, the VWF fragment is at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % identical to SEQ ID NO: 7. Example 6 shows a fusion protein with the sequence variant of SEQ ID NO: 7, namely SEQ ID NO: 40 which has the following additional amino acid substitution: A1164V.

[0202] According to one embodiment, at least one copy of the EP is fused directly to the C- terminus of the VWF fragment. One, two, three, four, five, or six copies of the EP may be fused to the C-terminus of the VWF fragment. The FVI I I-CKVWF I VWF-CKVWF heterodimers shown in the examples have three EPs at the C-terminus of the VWF- fragment, one as part of the fragment and two fused to it. The VWF protein with the amino acid sequence of SEQ ID NO: 7 joined with two EP copies with the amino acid sequence of SEQ ID NO: 2 added to the C-terminus is a sequence modified derivative of OCTA12 described in WO 2017 / 198435 A1 .

[0203] According to one embodiment, at least one copy of the EP forms the N-terminus and / or the C-terminus of the first fusion protein. For example, 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, or 15 copies of the EP may form the N-terminus and / or the C- terminus of the first fusion protein. According to one embodiment, at least three copies of the EP form the N-terminus and / or the C-terminus of the first fusion protein.

[0204] According to one embodiment, at least one copy of the EP forms the N-terminus and / or the C-terminus of the second fusion protein. For example, 1 , 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, or 15 copies of the EP form the N-terminus and / or the C-terminus of the second fusion protein. According to one embodiment, at least three copies of the EP form the N-terminus and / or the C-terminus of the first fusion protein. The FVIII- heterodimers shown in the examples have three EPs at the C- terminus of the VWF-fragment, also forming the C-terminus of the second fusion protein.

[0205] According to one embodiment, at least one of the N-terminus of the first fusion protein, the C-terminus of the first fusion protein, the N-terminus of the second fusion protein, and the C-terminus of the second fusion protein contains at least one EP.

[0206] According to one embodiment, at least two of the N-terminus of the first fusion protein, the C-terminus of the first fusion protein, the N-terminus of the second fusion protein, and the C-terminus of the second fusion protein contains at least one EP.

[0207] According to one embodiment, in the first fusion protein, the first pharmaceutically active protein is located N-terminal of the CKvwF-domain and optionally there is a third pharmaceutically active protein located C-terminal of the CKvwF-domain. According to one embodiment, in the first fusion protein, the first pharmaceutically active protein is located C-terminal of the CKvwF-domain and optionally there is a third pharmaceutically active protein located N-terminal of the CKvwF-domain. Preferably, the third pharmaceutically active protein is connected to the CKvwF-domain by a further linker.

[0208] According to one embodiment, in the second fusion protein, the second pharmaceutically active protein is located at the N-terminus of the CKvwF-domain and optionally there is a fourth pharmaceutically active protein located at the C-terminus of the CKvwF-domain. According to one embodiment, in the second fusion protein, the second pharmaceutically active protein is located at the C-terminus of the CKVWF- domain and optionally there is a fourth pharmaceutically active protein located at the N-terminus of the CKvwF-domain. Preferably, the fourth pharmaceutically active protein is connected to the CKvwF-domain by a peptide linker.

[0209] Of course, further protein moieties, in particular further pharmaceutically active proteins can be added to each fusion protein. The first fusion protein and the second fusion protein may each comprise one, two, three, four or more pharmaceutically active proteins. With more than two pharmaceutically active proteins per fusion protein, the further pharmaceutically active proteins are not directly connected to CKVWF, but connected to the other pharmaceutically active proteins, in particular by peptide linkers.

[0210] Moreover, it could make sense to have more than one copy of a pharmaceutically active protein in the fusion proteins forming the protein dimer according to the invention. According to one embodiment, the first fusion protein contains more than one copy of the first pharmaceutically active protein. For example, the first fusion protein contains at least two, at least three, at least four copies of the first pharmaceutically active protein. The copies of the first pharmaceutically active protein may be directly linked to each other or indirectly by a peptide linker. According to one embodiment, the copies of the second pharmaceutically active protein are linked to each other by copies of the second linker.

[0211] According to one embodiment, the second fusion protein contains more than one copy of the second pharmaceutically active protein. For example, the second fusion protein contains at least two, at least three, at least four copies of the second pharmaceutically active protein. The copies of the second pharmaceutically active protein may be directly linked to each other or indirectly by a peptide linker. According to one embodiment, the copies of the second pharmaceutically active protein are linked to each other by copies of the third linker.

[0212] The protein dimer FVI I I-CKVWF-VWF according to the invention has an increased half- life in comparison to FVIII alone. According to one embodiment, the half-life prolongation of the fusion protein is at least 20 %. According to one embodiment, the half-life prolongation of the fusion protein is at least 30 %. According to one embodiment, the half-life prolongation of the fusion protein is at least 40 %. According to one embodiment, the half-life prolongation of the fusion protein is at least 50 %. According to one embodiment, the half-life prolongation of the fusion protein is at least 60 %.

[0213] The protein dimer FVI I I-CKVWF-VWF according to the invention shows a reduced binding to endogenous VWF upon administration to a patient. According to one embodiment, the binding to VWF is at most 11 % of the binding level of FVIII alone. The binding is determined by surface plasmon resonance (SPR).

[0214] The half-life (ti / 2) may be calculated by linear regression analysis of the log-linear portion of the individual plasma concentration-time curves or by non-linear regression using one-phase exponential decay model. Exemplary softwares for calculation are GraphPad Prism version 6.07 (La Jolla, CA 92037 USA) and WinNonlin, version 6.4 (Pharsight Corporation, Mountain View, CA, USA).

[0215] The calculations are based on the following equations:

[0216] Kei = elimination rate constant t / 2= elimination half-life c = concentration t = time

[0217] In case of the FVIII-CKVWF-VWF heterodimer, according to one embodiment, the FVIII protein is located N-terminal of the CKVWF domain in the first fusion protein. According to a preferred embodiment, the FVIII is linked via the second linker to the N-terminus of the CKVWF domain. According to one embodiment, the VWF fragment is located N- terminal of the CKVWF domain in the second fusion protein. According to a preferred embodiment, the VWF fragment is linked via the third linker to the N-terminus of the CKVWF domain. the first fusion protein of the protein dimer is, for example, selected from the proteins shown in the examples: C9 (SEQ ID NO: 22), C10 (SEQ ID NO: 23), C11 (SEQ ID NO: 24), C12 (SEQ ID NO: 25), C13 (SEQ ID NO: 26), C27 (SEQ ID NO: 27), C28 (SEQ ID NO: 28), C29 (SEQ ID NO: 29), C30 (SEQ ID NO: 30), C31 (SEQ ID NO: 31 ), C32 (SEQ ID NO: 47), C35 (SEQ ID NO: 48), C42 (SEQ ID NO: 51 ), C43 (SEQ ID NO: 52), C44 (SEQ ID NO: 53), and C45 (SEQ ID NO: 54). Preferred embodiments of the first fusion protein are C11 , C13, and C27. Tables 3 and 15 below show the components (and their sequence IDs) forming these fusion proteins. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to a sequence selected from SEQ ID NO: 22, SEQ ID NO:

[0218] 23, SEQ ID NO: 24, SEQ ID NO: 25, SEQ ID NO: 26, SEQ ID NO: 27, SEQ ID NO:

[0219] 28, SEQ ID NO: 29, SEQ ID NO: 30, SEQ ID NO: 31 , SEQ ID NO: 47, SEQ ID NO:

[0220] 48, SEQ ID NO: 51 , SEQ ID NO: 52, SEQ ID NO: 53, and SEQ ID NO: 54.

[0221] According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 22. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 23. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 24. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 25. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 26. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 27. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 28. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 29. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 30. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 31. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 47. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 48. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 51. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 52. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 53. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 54.

[0222] In case of the FVIII-CKVWF-VWF heterodimer, the second fusion protein is, for example, selected from the proteins shown in the examples: C14 (SEQ ID NO: 32), C15 (SEQ ID NO: 33), C25 (SEQ ID NO: 34), C26 (SEQ ID NO: 35), C40 (SEQ ID NO: 49), and C41 (SEQ ID NO: 50). Tables 5 and 17 below show the components (and sequence IDs) forming these fusion proteins. Preferred embodiments of the second fusion protein are C15, C25 and C26.

[0223] According to one embodiment, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to a sequence selected from SEQ ID NO: 32, SEQ ID NO: 33, SEQ ID NO: 34, SEQ ID NO: 35, SEQ ID NO: 49, and SEQ ID NO: 50.

[0224] According to one embodiment of the protein dimer, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 32. According to one embodiment of the protein dimer, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 33. According to one embodiment of the protein dimer, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 34. According to one embodiment of the protein dimer, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 35. According to one embodiment of the protein dimer, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 49. According to one embodiment of the protein dimer, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 50.

[0225] According to one embodiment, the protein dimer is formed by a combination of first and second fusion protein selected from: C11 plus C15, C11 plus C25, C11 plus C26, C13 plus C26, C27 plus C15, C27 plus C25, C27 plus C26, C28 plus C26, C30 plus C15, C30 plus C25, C30 plus C26, C31 plus C15, C31 plus C25, and C31 plus C26.

[0226] In case of the VHH1 -CKVWF-VHH2 heterodimer, the first fusion protein of the protein dimer is, for example, selected from the proteins shown in the examples: FP(FIX)1 (SEQ ID NO: 60), FP(FIX)2 (SEQ ID NO: 62), FP(FIX)3 (SEQ ID NO: 64) FP(FIX)4, (SEQ ID NO: 68), FP(FIX)5 (SEQ ID NO: 82), FP(FIX)6 (SEQ ID NO: 83), FP(FIX)7 (SEQ ID NO: 84), FP(FIX)8 (SEQ ID NO: 86), FP(FIX)9 (SEQ ID NO: 88), FP(FIX)10 (SEQ ID NO: 90), FP(FIX)11 (SEQ ID NO: 93), FP(FIX)12 (SEQ ID NO: 95), and FP(FIX)20 (SEQ ID NO: 98). It is noted that FP(FIX) binds to both FIX and FX. Table 19 below shows the components (and their sequence IDs) forming these fusion proteins. According to one embodiment of the protein dimer, the first fusion protein comprises an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to a sequence selected from SEQ ID NO: 60, SEQ ID NO: 62, SEQ ID NO: 64, SEQ ID NO: 68, SEQ ID NO: 82, SEQ ID NO: 83, SEQ ID NO: 84, SEQ ID NO: 86, SEQ ID NO: 88, SEQ ID NO: 90, SEQ ID NO: 93, SEQ ID NO: 95, and SEQ ID NO: 98.

[0227] In case of the VHH1 -CKVWF-VHH2 heterodimer, the second fusion protein is, for example, selected from the proteins shown in the examples: FP(FX)1 (SEQ ID NO: 61 ), FP(FX)2 (SEQ ID NO: 63), FP(FX)3 (SEQ ID NO: 65) FP(FX)4 (SEQ ID NO: 69), FP(FX)5 (SEQ ID NO: 85), FP(FX)6 (SEQ ID NO: 87), FP(FX)7 (SEQ ID NO: 89), FP(FX)8 (SEQ ID NO: 91 ), FP(FX)9 (SEQ ID NO: 92), FP(FX)10 (SEQ ID NO: 94), FP(FX)11 (SEQ ID NO: 96), FP(FX)12 (SEQ ID NO: 97), and FP(FX)14 (SEQ ID NO: 99). Table 19 below shows the components (and their sequences) forming these fusion proteins. According to one embodiment of the protein dimer, the second fusion protein comprises an amino acid sequence with an identity of at least 90 %, at least 91 %, at least 92 %, at least 93 %, at least 94 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99% or 100 % to a sequence selected from SEQ ID NO: 61 , SEQ ID NO: 63, SEQ ID NO: 65, SEQ ID NO: 69, SEQ ID NO: 85, SEQ ID NO: 87, SEQ ID NO: 89, SEQ ID NO: 91 , SEQ ID NO: 92, SEQ ID NO: 94, SEQ ID NO: 96, SEQ ID NO: 97, and SEQ ID NO: 99.

[0228] According to one embodiment, the protein dimer is formed by a combination of first and second fusion protein selected from: H1 (FP(FIX)1 plus FP(FX)1 ), H2 (FP(FIX)2 plus FP(FX)2), H3 (FP(FIX)3 plus FP(FX)3), H5 (FP(FIX)4 plus FP(FX)4), H6 (FP(FIX)4 plus FP(FX)1 , H17 (FP(FIX)5 plus FP(FX)4), H18 (FP(FIX)6 plus FP(FX)4), H19 (FP(FIX)7 plus FP(FX)4), H20 (FP(FIX)5 plus FP(FX)5), H22 (FP(FIX)8 plus FP(FX)6), H28(FP(FIX)5 plus FP(FX)3), H30 (FP(FIX)9 plus FP(FX)4), H31 (FP(FIX)4 plus FP(FX)7), H32 (FP(FIX)10 plus FP(FX)8), H33 (FP(FIX)4 plus FP(FX)4), H34 (FP(FIX)11 plus FP(FX)6), H35 (FP(FIX)11 plus FP(FX)10), and H36 (FP(FIX)12 plus FP(FX)11 ).

[0229] In a second aspect, the invention relates to the use of a CK domain pair, in particular a CKVWF domain pair, to combine two pharmaceutically active proteins by dimer formation, wherein the one CK domain, in particular a CKVWF domain, is fused to a pharmaceutically active protein to form a first fusion protein and the other CK domain, in particular a CKVWF domain is fused to a pharmaceutically active protein to form a second fusion protein.

[0230] In the use according to the second aspect, the CK domain, the pharmaceutically active protein as well as any element of the first fusion protein and the second fusion protein are defined as described above for the first aspect. In particular, if the second fusion protein contains a fragment of VWF, the fragment does not contain the CKVWF and the C3 domain of VWF and the third linker is an engineered linker and does not contain the C3 domain of VWF.

[0231] In a third aspect, the invention provides a fusion protein comprising a CKVWF domain of VWF, a linker and a pharmaceutically active protein, wherein the CKVWF domain is able to covalently bind a second CKVWF domain.

[0232] In the third aspect, the fusion protein is defined like the first or second fusion protein as described in the first aspect. Accordingly, the pharmaceutically active protein of the second aspect as well as any element of the first fusion protein and the second fusion protein are defined as described above for the first aspect. Moreover, the linker may be defined according to the definition of the second and third linkers above.

[0233] Exemplary fusion proteins are the fusion proteins (with the corresponding sequence IDs) are: C9 (SEQ ID NO: 22), C10 (SEQ ID NO: 23), C11 (SEQ ID NO: 24), C12 (SEQ ID NO: 25), C13 (SEQ ID NO: 26), C27 (SEQ ID NO: 27), C28 (SEQ ID NO: 28), C29 (SEQ ID NO: 29), C30 (SEQ ID NO: 30), C31 (SEQ ID NO: 31 ), C14 (SEQ ID NO: 32), C15 (SEQ ID NO: 33), C25 (SEQ ID NO: 34), and C26 (SEQ ID NO: 35) C32 (SEQ ID NO: 47), C35 (SEQ ID NO: 48), C40 (SEQ ID NO: 49), C41 (SEQ ID

[0234] NO: 50), C42 (SEQ ID NO: 51 ), C43 (SEQ ID NO: 52), C44 (SEQ ID NO: 53), C45

[0235] (SEQ ID NO: 54), FP(FIX)1 (SEQ ID NO: 60), FP(FIX)2 (SEQ ID NO: 62), FP(FIX)3

[0236] (SEQ ID NO: 64) FP(FIX)4 (SEQ ID NO: 68), FP(FX)1 (SEQ IDNO: 61 ), FP(FX)2

[0237] (SEQ ID NO: 63), FP(FX)3 (SEQ ID NO: 65), FP(FX)4 (SEQ ID NO: 69), FP(FIX)5 (SEQ ID NO: 82), FP(FIX)6 (SEQ ID NO: 83), FP(FIX)7 (SEQ ID NO: 84), FP(FX)5 (SEQ ID NO: 85), FP(FIX)8 (SEQ ID NO: 86), FP(FX)6 (SEQ ID NO: 87), FP(FIX)9 (SEQ ID NO: 88), FP(FX)7 (SEQ ID NO: 89), FP(FIX)10 (SEQ ID NO: 90), FP(FX)8 (SEQ ID NO: 91 ), FP(FX)9 (SEQ ID NO: 92), FP(FIX)11 (SEQ ID NO: 93), FP(FX)10 (SEQ ID NO: 94), FP(FIX)12 (SEQ ID NO: 95), FP(FX)11 (SEQ ID NO: 96), FP(FX)12 (SEQ ID NO: 97), FP(FIX)20 (SEQ ID NO: 98), and FP(FX)14 (SEQ ID NO: 99). Tables 3, 5, 15, 17, and 19 below show the components (and their sequence IDs) forming these fusion proteins.

[0238] According to one embodiment the amino acid sequence of the fusion protein has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 22 , SEQ ID NO: 23 , SEQ ID NO: 24, SEQ ID NO: 25, SEQ ID NO: 26, SEQ ID NO: 27, SEQ ID NO: 28, SEQ ID NO: 29, SEQ ID NO: 30, SEQ ID NO: 31 , SEQ ID NO: 32, SEQ ID NO: 33, SEQ ID NO: 34, SEQ ID NO: 35, , SEQ ID NO: 47, SEQ ID NO: 48, SEQ ID NO: 49, SEQ ID NO: 50, SEQ ID NO: 51 , SEQ ID NO: 52, SEQ ID NO: 53, and SEQ ID NO: 54.

[0239] According to one embodiment the amino acid sequence of the fusion protein is identical to a sequence selected from SEQ ID NO: 22, SEQ ID NO: 23, SEQ ID NO:

[0240] 24, SEQ ID NO: 25, SEQ ID NO: 26, SEQ ID NO: 27, SEQ ID NO: 28, SEQ ID NO:

[0241] 29, SEQ ID NO: 30, SEQ ID NO: 31 , SEQ ID NO: 32, SEQ ID NO: 33, SEQ ID NO:

[0242] 34, SEQ ID NO: 35, SEQ ID NO: 47, SEQ ID NO: 48, SEQ ID NO: 49, and SEQ ID

[0243] NO: 50, SEQ ID NO: 51 , SEQ ID NO: 52, SEQ ID NO: 53, and SEQ ID NO: 54.

[0244] According to one embodiment the amino acid sequence of the fusion protein has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 60, SEQ ID NO: 62, SEQ ID NO: 64, SEQ ID NO: 68, SEQ IDNO: 61 , SEQ ID NO: 63, SEQ ID NO: 65, SEQ ID NO: 69, SEQ ID NO: 82, SEQ ID NO: 83,

[0245] SEQ ID NO: 84, SEQ ID NO: 85, SEQ ID NO: 86, SEQ ID NO: 87, SEQ ID NO: 88,

[0246] SEQ ID NO: 89, SEQ ID NO: 90, SEQ ID NO: 91 , SEQ ID NO: 92, SEQ ID NO: 93,

[0247] SEQ ID NO: 94, SEQ ID NO: 95, SEQ ID NO: 96, SEQ ID NO: 97, SEQ ID NO: 98, and SEQ ID NO: 99.

[0248] According to one embodiment the amino acid sequence of the fusion protein is identical to a sequence selected from SEQ ID NO: 60, SEQ ID NO: 62, SEQ ID NO: 64, SEQ ID NO: 68, SEQ ID NO: 61 , SEQ ID NO: 63, SEQ ID NO: 65, SEQ ID

[0249] NO: 69, SEQ ID NO: 82, SEQ ID NO: 83, SEQ ID NO: 84, SEQ ID NO: 85, SEQ ID

[0250] NO: 86, SEQ ID NO: 87, SEQ ID NO: 88, SEQ ID NO: 89, SEQ ID NO: 90, SEQ ID

[0251] NO: 91 , SEQ ID NO: 92, SEQ ID NO: 93, SEQ ID NO: 94, SEQ ID NO: 95, SEQ ID

[0252] NO: 96, SEQ ID NO: 97, SEQ ID NO: 98, and SEQ ID NO: 99.

[0253] Polynucleotide

[0254] According to a fourth aspect, the invention provides an isolated polynucleotide that comprises a nucleic acid sequence encoding a fusion protein according to the third aspect of the invention. The isolated polynucleotide may be a DNA molecule or an RNA molecule. The isolated polynucleotide is preferably a DNA molecule, in particular a cDNA molecule. The techniques used to isolate or clone a polynucleotide encoding a peptide are known in the art and include isolation from genomic DNA, preparation from cDNA, or a combination thereof. The cloning of the polynucleotides from such genomic DNA can be effected, e.g., by using the well-known polymerase chain reaction (PCR) or antibody screening of expression libraries to detect cloned DNA fragments with shared structural features (see, e.g., Innis et al. 1990). Other nucleic acid amplification procedures such as ligase chain reaction (LCR), ligation activated transcription (LAT) and polynucleotide-based amplification (NASBA) may be used.

[0255] The sequence of the isolated polynucleotide may comprise a first part encoding a FVIII heavy chain, a second part encoding the first linker, a third part encoding a FVIII light chain, a fourth part encoding a second linker and a fifth part encoding a CKVWF.

[0256] According to one embodiment, the first part encodes a FVIII heavy chain with a sequence identity to SEQ ID NO: 3 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0257] According to one embodiment, the second part encodes a first linker with a sequence identity of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 % to a sequence selected from the group consisting of SEQ ID NO: 8, SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 11 , SEQ ID NO: 12, SEQ ID NO: 41 , SEQ ID NO: 43, and SEQ ID NO: 44.

[0258] According to one embodiment, the third part encodes a FVIII light chain with a sequence identity to SEQ ID NO: 5 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0259] According to one embodiment, the fourth part encodes a second linker with a sequence identity of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 % to a sequence selected from the group consisting of SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15, SEQ ID NO: 16, SEQ ID NO: 45, SEQ ID NO: 46, SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58, and SEQ ID NO: 59. According to one embodiment, the fifth part encodes a CKVWF with a sequence identity to SEQ ID NO: 1 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0260] According to one embodiment, the sixth part encodes three consecutive EPs with a sequence identity to SEQ ID NO: 2 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0261] According to one embodiment of the polynucleotide, the sequence of the isolated polynucleotide may comprise a first part encoding a VWF fragment, a second part encoding the third linker, and a third part encoding a CKVWF.

[0262] According to an alternative embodiment of the polynucleotide the sequence of the isolated polynucleotide encodes a VHH, a second part encoding the third linker, and a third part encoding a CKVWF.

[0263] According to a further alternative embodiment of the polynucleotide the sequence of the isolated polynucleotide encodes a VHH, a second part encoding the third linker, and a third part encoding a CKNDP.

[0264] According to one embodiment, the first part encodes a VWF fragment with a sequence identity to SEQ ID NO: 6 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0265] According to one embodiment, the first part encodes a VHH with a sequence identity to SEQ ID NO: 72 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least

[0266] 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0267] According to one embodiment, the first part encodes a VHH with a sequence identity to SEQ ID NO: 72 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least

[0268] 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0269] According to one embodiment, the second part encodes a third linker with a sequence identity of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 % to a sequence selected from the group consisting of SEQ ID NO: 17,SEQ ID NO: 18, SEQ ID NO: 55, SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58, SEQ ID NO: 59, and SEQ ID NO: 74.

[0270] According to one embodiment, the third part encodes a CKVWF with a sequence identity to SEQ ID NO: 1 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0271] According to one embodiment, the third part encodes a CKNDP with a sequence identity to SEQ ID NO: 100 of at least 80 %, at least 85 %, at least 90 %, at least 95 %, at least 96 %, at least 97 %, at least 98 %, at least 99 %, or 100 %.

[0272] According to one embodiment, the polynucleotide encodes a fusion protein selected from C9 (SEQ ID NO: 22), C10 (SEQ ID NO: 23), C11 (SEQ ID NO: 24), C12 (SEQ ID NO: 25), C13 (SEQ ID NO: 26), C27 (SEQ ID NO: 27), C28 (SEQ ID NO: 28), C29 (SEQ ID NO: 29), C30 (SEQ ID NO: 30), C31 (SEQ ID NO: 31 ), C14 (SEQ ID NO: 32), C15 (SEQ ID NO: 33), C25 (SEQ ID NO: 34), C26 (SEQ ID NO: 35), , C32 (SEQ ID NO: 47), C35 (SEQ ID NO: 48), C40 (SEQ ID NO: 49), C41 (SEQ ID NO: 50), C42 (SEQ ID NO: 51 ), C43 (SEQ ID NO: 52), C44 (SEQ ID NO: 53), and C45 (SEQ ID NO: 54).

[0273] According to one embodiment, the polynucleotide encodes a fusion protein with a sequence identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 22 , SEQ ID NO: 23 , SEQ ID NO: 24, SEQ ID NO: 25, SEQ ID NO: 26, SEQ ID NO: 27, SEQ ID NO: 28, SEQ ID NO: 29, SEQ ID NO: 30,

[0274] SEQ ID NO: 31 , SEQ ID NO: 32, SEQ ID NO: 33, SEQ ID NO: 34 SEQ ID NO: 35, ,

[0275] SEQ ID NO: 47, SEQ ID NO: 48, SEQ ID NO: 49, SEQ ID NO: 50, SEQ ID NO: 51 ,

[0276] SEQ ID NO: 52, SEQ ID NO: 53, and SEQ ID NO: 54. .

[0277] According to one embodiment, the polynucleotide encodes a fusion protein selected from FP(FIX)1 (SEQ ID NO: 60), FP(FIX)2 (SEQ ID NO: 62), FP(FIX)3 (SEQ ID NO: 64) FP(FIX)4 (SEQ ID NO: 68), FP(FIX)1 (SEQ IDNO: 61 ), FP(FIX)2 (SEQ ID NO: 63), FP(FIX)3 (SEQ ID NO: 65), and FP(FIX)4 (SEQ ID NO: 69), FP(FIX)5 (SEQ ID NO: 82), ,FP(FIX)6 (SEQ ID NO: 83), FP(FIX)7 (SEQ ID NO: 84), FP(FX)5 (SEQ ID NO: 85), FP(FIX)8 (SEQ ID NO: 86), FP(FX)6 (SEQ ID NO: 87), FP(FIX)9 (SEQ ID NO: 88), FP(FX)7 (SEQ ID NO: 89), FP(FIX)10 (SEQ ID NO: 90), FP(FX)8 (SEQ ID NO: 91 ), FP(FX)9 (SEQ ID NO: 92), FP(FIX)11 (SEQ ID NO: 93), FP(FX)10 (SEQ ID NO: 94), FP(FIX)12 (SEQ ID NO: 95), FP(FX)11 (SEQ ID NO: 96), FP(FX)12 (SEQ ID NO: 97), FP(FIX)20 (SEQ ID NO: 98), and FP(FX)14 (SEQ ID NO: 99).

[0278] According to one embodiment, the polynucleotide encodes a fusion protein with a sequence identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 60, SEQ ID NO: 62, SEQ ID NO: 64, SEQ ID NO: 68, SEQ IDNO: 61 , SEQ ID NO: 63, SEQ ID NO: 65, SEQ ID NO: 69, SEQ ID NO: 82, SEQ ID

[0279] NO: 83, SEQ ID NO: 84, SEQ ID NO: 85, SEQ ID NO: 86, SEQ ID NO: 87, SEQ ID

[0280] NO: 88, SEQ ID NO: 89, SEQ ID NO: 90, SEQ ID NO: 91 , SEQ ID NO: 92, SEQ ID

[0281] NO: 93, SEQ ID NO: 94, SEQ ID NO: 95, SEQ ID NO: 96, SEQ ID NO: 97, SEQ ID

[0282] NO: 98, and SEQ ID NO: 99. Vector

[0283] In a fifth aspect, the invention also relates to expression vectors comprising a polynucleotide according to the fourth aspect of the invention.

[0284] The expression vector further preferably comprises control elements such as a promoter, and transcriptional and translational stop signals. The polynucleotides according to the second aspect and of the control elements may be joined together to produce a recombinant expression vector that may include one or more restriction sites to allow for insertion or substitution of the polynucleotide encoding the polypeptide at such sites. The polynucleotide may be inserted into an appropriate expression vector for expression. In creating the expression vector, the coding sequence is located in the expression vector so that the coding sequence is operably linked with the appropriate control sequences for expression.

[0285] The recombinant expression vector may be any vector (e.g., a plasmid or a virus) that can be conveniently subjected to recombinant DNA procedures and can bring about expression of the polynucleotide of the fourth aspect of the invention. The choice of the expression vector will typically depend on the compatibility of the expression vector with the host cell into which the expression vector is to be introduced. The expression vectors may be linear or a closed circular plasmid. The expression vector is preferably adapted to expression in mammalian cells. The expression vector may be an autonomously replicating vector, i.e., a vector that exists as an extrachromosomal entity, the replication of which is independent of chromosomal replication, e.g., a plasmid, an extrachromosomal element, a minichromosome, or an artificial chromosome. For autonomous replication, the vector may further comprise an origin of replication enabling the vector to replicate autonomously in the host cell in question. The origin of replication may be any plasmid replicator mediating autonomous replication that functions in a cell. The term "origin of replication" or "plasmid replicator" means a polynucleotide that enables a plasmid or vector to replicate in vivo.

[0286] The vector is preferably one that, when introduced into the host cell, is integrated into the genome and replicated together with the chromosome(s) into which it has been integrated. For integration into the host cell genome, the expression vector may rely on any other element of the expression vector for integration into the genome by homologous or non-homologous recombination. Alternatively, the vector may contain additional polynucleotides for directing integration by homologous recombination into the genome of the host cell at a precise location in the chromosome.

[0287] The vectors of the present invention preferably contain one or more (e.g., several) selectable markers that permit easy selection of transformed, transfected, transduced, or the like cells. A selectable marker is a gene the product of which provides for biocide or viral resistance, resistance to heavy metals, prototrophy to auxotrophs, and the like.

[0288] The procedures used to ligate the elements described above to construct the recombinant expression vectors of the present invention are well known to one skilled in the art (see Green and Sambrook 2012; Chapter 3).

[0289] According to one embodiment the backbone of the vector according to the fifth aspect is selected from pCDNA3, pCDNA3.1 , pCDNA3.2, pCDNA3.3, pCDNA3.4, pCDNA4, pCDNA5, pCDNA6, pCEP4, pCEP-puro, pCET1019, pCMV, pEF1 , pEF4, pEF5, pEF6, pExchange, pEXPR, pIRES, and pSCAS. Host Cell

[0290] According to a sixth aspect the invention provides a host cell, comprising the polynucleotide according to the fourth aspect or the expression vector according to the fifth aspect of the invention. The expression vector according to the fifth aspect is introduced into a host cell so that the expression vector is maintained as a chromosomal integrant or as a self-replicating extra-chromosomal vector as described earlier. The choice of a host cell will to a large extent depend upon the gene encoding the polypeptide and its source.

[0291] According to one embodiment the fusion protein is produced by expression in a mammalian host cell line. The fusion protein is preferably produced in a human host cell line. Generally, any human host cell line is suitable for expression of the fusion protein. The host cell is preferably of human origin in order to ensure that the fusion protein is properly processed during folding and receives the proper post-translational modifications (e.g. glycosylation, hydroxylation, phosphorylation and sulfation). A favourable glycosylation profile of the fusion protein is particularly obtained with human kidney cell lines. Preferred human kidney cell lines are HEK cell lines, in particular HEK 293 cell lines.

[0292] Examples of HEK cell lines for production of the glycosylated polypeptide are HEK 293 F, Expi293F (Thermo Scientific A14527), Flp-ln™-293 (Invitrogen R75007), 293 (ATCC® CRL-1573), 293 EBNA, 293H (Thermo Scientific 11631017), 293S, 293T (ATCC® CRL-3216™), 293T / 17 (ATCC® CRL11268™), 293T / 17 SF (ATCC® ACS4500™), HEK 293 STF (ATCC® CRL 3249™), HEK-293.2sus (ATCC® CRL- 1573™). A preferred cell line for production of the polypeptide is the HEK 293 F cell line.

[0293] Other human cell lines suitable as host cells for expression include, without limitation cell lines derived from myeloid leukemia cells. Specific examples of host cells are K562, NM-F9, NM-D4, NM-H9D8, NM-H9D8-E6, NM H9D8-E6Q12, GT-2X, GT-5s and cells derived from anyone of said host cells. K562 is a human myeloid leukemia cell line present in the American Type Culture Collection (ATCC CCL-243). The remaining cell lines are derived from K562 cells and have been selected for specific glycosylation features. Further mammalian host cell lines suitable for producing fusion proteins according to the invention include cell lines of hamster, mouse, and monkey origin. Suitable host cells include Chinese hamster ovary cells (CHO cells, e.g., DG44, DXB11 , and K1 [ATCC CCL-61 , including its glutamine auxotroph derivative CHOZn, SAFC CHOGS]) and baby hamster kidney (BHK) cells.

[0294] Pharmaceutical and medical use

[0295] Therefore, the fusion proteins according to the first aspect are in particular useful as active ingredients for medical treatment. Preferably, they are useful for treatment or prevention of a bleeding disorder. The fusion proteins according to the first aspect described herein can be administered alone or in the form of pharmaceutical compositions.

[0296] Thus, according to a seventh aspect, the invention provides the fusion protein according to the first aspect for use in the treatment of a bleeding disorder.

[0297] According to one embodiment, the fusion protein may be formulated with at least one pharmaceutically acceptable carrier. Pharmaceutical compositions based on the fusion protein can be prepared and administered to a subject by any method well known in the art of pharmacy. See, e. g, Goodman & Gilman's The Pharmacological Basis of Therapeutics, Hardman et al., eds., McGraw-Hill Professional (10th ed., 2001 ); Remington: The Science and Practice of Pharmacy, Gennaro, ed., Lippincott Williams & Wilkins (20th ed., 2003); and Pharmaceutical Dosage Forms and Drug Delivery Systems, Ansel et al. (eds), Lippincott Williams & Wilkins (7th ed., 1999). In addition, the pharmaceutical compositions of the embodiments may also be formulated to include other medically useful drugs or biological agents. The pharmaceutical composition typically comprises a therapeutically effective amount of the fusion protein combined with a pharmaceutically acceptable carrier. The pharmaceutically acceptable carrier is any carrier known or established in the art. Exemplary pharmaceutically acceptable carriers include sterile pyrogen-free water and sterile pyrogen-free saline solution. Other forms of pharmaceutically acceptable carriers that can be utilized for the present embodiments include binders, disintegrants, surfactants, absorption accelerators, moisture retention agents, absorbers, lubricants, fillers, extenders, moisture imparting agents, preservatives, stabilizers, emulsifiers, solubilising agents, salts which control osmotic pressure, diluting agents such as buffers, and excipients usually used depending on the use form of the formulation. These are optionally selected and used depending on the unit dosage of the resulting formulation.

[0298] Thus, the invention also relates to a method of treatment or prevention of a bleeding disorder of a patient, said method comprising administering to said patient a pharmaceutical composition according to the seventh aspect.

[0299] As used herein “bleeding disorder” refers to a disease or condition that impairs normal hemostasis. The bleeding disorder can be, for example, Hemophilia A, Hemophilia B, Factor VIII deficiency, Factor XI deficiency, von Willebrand Disease, Glanzmann's Thrombasthenia, Bernard-Soulier Syndrome, idiopathic thrombocytopenic purpura, intracerebral hemorrhage, and the like.

[0300] As used herein, “hemophilia” refers to a group of bleeding disorders associated with increased blood clot formation time as compared to blood clot formation time in healthy individuals without hemophilia. Hemophilia includes Hemophilia A, which is a disorder that leads to the production of defective Factor VIII, Hemophilia B, which is a disorder that leads to the production of defective Factor IX and acquired Hemophilia A, a rare bleeding disorder caused by autoantibodies to FVIII.

[0301] The bleeding disorder is preferably Hemophilia A or B. The treatment may for example be the hemophilia treatment of previously untreated patients (PUPS) or an immune tolerance induction (ITI) treatment and / or other related treatments of hemophilia disorders.

[0302] For in vivo applications, pharmaceutical compositions can be administered to the patient by any customary administration route, e.g., orally, parenterally or by inhalation. Parenteral administration includes intravenous injection, subcutaneous injection, intraperitoneal injection, intramuscular injection, liquid agents, suspensions, emulsions and dripping agents. For parenteral administration the pharmaceutical composition should be an injectable agent such as a liquid agent or a suspension. In other embodiments, the pharmaceutical composition is administered orally to a patient. In these embodiments, a form of the drug includes solid formulations such as tablets, coated tablets, powdered agents, granules, capsules and pills, liquid formulations such as liquid agents (e.g., eye drops, nose drops), suspension, emulsion and syrup, inhales such as aerosol agents, atomizers and nebulizers, and liposome inclusion agents. In still some other embodiments, the glycosylated polypeptide, protein complex or pharmaceutical composition is administered by inhalation to the respiratory tract of a patient to target the trachea and / or the lung of a subject.

[0303] According to one embodiment of the seventh aspect, the use comprises an intravenous or non-intravenous injection. The non-intravenous injection preferably is a subcutaneous injection.

[0304] Use of the EPs

[0305] According to an eighth aspect, the invention relates to the use of one or more EPs for decreasing the aggregation tendency of a target protein, wherein the one or more EPs are fused to or inserted into the target protein. As shown in Example 14, with the number of EPs, the tendency of the fusion proteins to aggregate, i.e. forming High Molecular Weight Components (HMWC) is reduced by the EPs. In particular, the number of HMWC reduces proportionally to the increase in number of EPs fused to the target protein. A fusion protein with 1 copy of the EP aggregates less than a fusion protein without EPs.

[0306] According to one embodiment, the target protein is a pharmaceutically active protein as defined above. According to one embodiment, the target protein is a CK fusion protein as defined above. According to one embodiment, the target protein is defined according to the first and second fusion protein according to the first aspect. According to one embodiment, the target protein is a protein dimer according to the first aspect.

[0307] According to one embodiment, the use comprises fusing at least 2 copies of the EP to the target protein. For example, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 , 12, 13, 14, 15, 16, 17, 18, 19, or 20 copies of the EP are added to the target protein. A fusion protein with 2 copies of the EP aggregates less than a fusion protein with 1 copy. According to one embodiment, the use comprises fusing at least 2 copies of the EP to the target protein. According to one embodiment, the use comprises inserting at least 2 copies of the EP into the target protein. A fusion protein with 4 copies of the EP aggregates less than a fusion protein with 2 copies. According to one embodiment, the use comprises fusing at least 4 copies of the EP to the target protein. According to one embodiment, the use comprises inserting at least 4 copies of the EP into the target protein. A fusion protein with 6 copies of the EP aggregates less than a fusion protein with 4 copies. According to one embodiment, the use comprises fusing at least 6 copies of the EP to the target protein. According to one embodiment, the use comprises inserting at least 6 copies of the EP into the target protein. A fusion protein with 9 copies of the EP aggregates less than a fusion protein with 6 copies. According to one embodiment, the use comprises fusing at least 9 copies of the EP to the target protein. According to one embodiment, the use comprises inserting at least 9 copies of the EP into the target protein.

[0308] EXAMPLES

[0309] Example 1 - Design, cloning and expression of FVIII-CKVWF and VWF-CKVWF fusion proteins

[0310] Aims of the

[0311] Design and generation of cDNAs encoding various FVI I I-CKVWF and VWF-CKVWF fusion proteins and cloning into expression vectors. Recombinant expression of FVIII- CKVWF and VWF-CKVWF fusion proteins and combinations thereof in human cells.

[0312] Methods cDNA design, Codon optimization, Gene synthesis and cloning

[0313] For generating expression vector constructs encoding the FVI I I-CKVWF and VWF- CKVWF protein variants, the Golden Gate cloning technology was used (Engler et al., 2008). To this end, cDNA fragments encoding the CKVWF fusion variant were synthesized at Twist Bioscience and cloned into donor vectors compatible for Golden Gate cloning. The donor vectors containing the desired cDNA construct were used in a Golden Gate assembly reaction, together with a proprietary acceptor backbone and discrete donor vectors containing regulatory elements. This strategy was used to generate vectors for transient mammalian expression containing a gpCMV promoter 5’ of the desired cDNA construct and a woodchuck hepatitis virus post-transcriptional regulatory element (WPRE) 3’ of the cDNA construct.

[0314] NEB5-alpha E. coli cells were transformed with the vector constructs and single clones were selected following an overnight incubation at 37 °C on ampicillin- containing LB agar plates.

[0315] Plasmid DNA preparations were performed using the QIAprep DNA Mini kit (Qiagen) or the NucleoBond® Xtra Maxi Plus EF kit (Macherey-Nagel) according to the manufacturer's recommendations. The integrity of the constructs was verified by sequencing with particular focus on the correct orientation and integrity of the gene encoding the desired fusion protein variant.

[0316] Protein expression

[0317] The respective FVI I I-CKVWF and VWF-CKVWF fusion construct pairs were concomitantly transfected and transiently expressed in Expi293F cells (Thermo Fisher Scientific) in 500-1000 mL scale according to the manufacturer's recommendations. Product containing cell culture supernatant was harvested 4-5 days post transfection by centrifugation at 2000 x g for 20 min.

[0318] Results

[0319] The FVI I I-CKVWF fusion proteins encoded by the cloned cDNA constructs together with the linkers used are shown in Tables 1 to 3. With regard to the first linker, the FVI I I-CKVWF molecules were produced in a “NUWIQ-like” and in a “ReFacto-like” version (fusion proteins C9 and C10, respectively). C9 and C10 comprise the linker sequences 1 -1 and 1 -2 flanking the deleted FVIII B-domain as found in the marketed rFVIll substitution products NUWIQ (Octapharma AG) and ReFacto (Pfizer AG), respectively. Based on the “NUWIQ-like” and “ReFacto-like” sequences, the first linker variants 1 -3, 1 -4 and 1 -5 were designed that comprise combinations of GS linkers and EPs inserted into the FVIII B-domain remainder sequence: Linker 1 -3: three EPs and flanking G4S-linkers inserted into the NUWIQ-like B-domain sequence, Linker 1 -4: ReFacto-like B-domain sequence with three EPs and G4S-linkers inserted, and Linker 1 -5: similar to Linker 1 -4 with the furin cleavage site removed (amino acids RHQR deleted, Table 1 ).

[0320] Table 1 : First linkers used for the FVIII-CKVWF fusion proteins.

[0321] The first linker is the linker connecting the FVIII heavy chain to the FVIII light chain (Fig. 1). EP denotes the extension peptide with the SEQ ID NO: 2. The lower-case numbers indicate the number of repetitions of the sequence elements in brackets.

[0322] The above variants were used in combinations with four variants of the second linker that comprise the thrombin cleavage sequence IEPRSFS. The sequence variants differ in the presence and position of the EPs and in the placement of a half-life extending moiety, i.e. an albumin binding VHH domain (ABV, Table 2).

[0323] Table 2: Second linkers used for the FVIII-CKVWF fusion proteins.

[0324] The second linker is a cleavable linker connecting the FVIII light chain to the CKVWF domain of the first fusion protein (Fig. 1). EP represents the extension peptide with the SEQ ID NO:

[0325] 2. ABV indicates the albumin binding VHH with the SEQ ID NO: 36. The lower-case numbers represent the number of repetitions of the sequence elements in brackets.

[0326] Table 3: Overview of the FVIII-CKVWF fusion proteins and their components. ABV indicates the albumin binding VHH moiety with SEQ ID NO: 36. Four different VWF-CKVWF molecules with two different third linker sequences were designed. Linker 3-1 comprises two consecutive EPs followed by a long flexible G4S- linker. The use of Linker 3-2 adds an albumin binding VHH (ABV) before the CKVWF domain (Table 4). As a result, the ABV becomes a sequence constituent flanked by GS-linkers (in fusion protein C26). Alternatively, the ABV is fused to the C-terminus of the VWF-CKVWF molecule (in C25, Table 5).

[0327] Table 4: Third linkers used for the connection of the VWF fragment with the CKVWF domain.

[0328] The third linker is the linker connecting the VWF fragment of the second fusion protein to the CKVWF domain (Fig. 1). EP stands for the extension peptide with the SEQ ID NO: 2. ABV indicates the albumin binding VHH with SEQ ID NO: 36. The lower-case numbers indicate the number of repetitions of the sequence elements in brackets.

[0329] The layout of the VWF-CKVWF fusion proteins is shown in Table 5.

[0330] Table 5: Overview of the VWF-CKVWF fusion proteins and their components.

[0331] ABV denotes the albumin binding VHH (SEQ ID NO: 36).

[0332] The results of the protein expression experiments are presented in Example 2. Example 2 - Dimer formation and factor VIII activity (FVIII :C) of FVIII-CKVWF + VWF-CKVWF fusion proteins

[0333] Aims of the

[0334] Testing of FVI I I-CKVWF and VWF-CKVWF fusion proteins for their ability to dimerize. Characterization of the formed dimers by chromogenic factor VIII activity (FVIII:C) analysis. Assessing the impact of the FVIII B-domain remainder sequence (“NUWIQ- like” or “ReFacto-like”) and of the presence of EPs with flanking GS-linkers in the B- domain sequence on the FVIII activity in the expression supernatant.

[0335] Methods

[0336] The FVI I I-CKVWF and VWF-CKVWF fusion constructs were concomitantly transfected and transiently expressed in Expi293F cells (Thermo Fisher Scientific) in 3 mL scale. Experiments were performed in triplicates. The cell culture supernatant was harvested 4 days post transfection by centrifugation at 4800 x g for 30 min. FVI 11: C activity was assessed by the FVIII chromogenic assay kit (Siemens) on a BCS XP system (Siemens). The expression experiments were performed in two separate rounds. The schematic structure of all heterodimers expressed in both expression rounds is depicted in Figures 2 to 5.

[0337] Results

[0338] The first round of expression experiments included five different FVI I I-CKVWF fusion proteins (C9 to C13 in Table 3), each expressed together with two different VWF- CKVWF fusion proteins of Table 5 (i.e. C14 and C15). Analyses of the FVIII activity of the 10 heterodimers formed resulted in FVIII:C levels ranging from 1.05 to 5.36 lll / ml in the expression supernatant (Fig. 6A). The FVIII activity levels of the dimers formed with the VWF fragment variant containing the propeptide (C14) were between 1.1 lll / ml (for C12+C14) to 2.09 lll / ml (for C11 +C14). In contrast, significantly higher activities were measurable for the dimers formed with C15 (ranging from 2.87 lll / ml for C10+C15 to 5.36 lll / ml for C11 +C15). Among the FVI I I-CKVWF constructs co- expressed with either C14 or C15, the highest expression levels were achieved for heterodimers with the C11 construct comprising three EPs with flanking GS-linkers inserted in the “NUWIQ-like” B-domain remainder sequence (2.09 lll / ml and 5.36 lll / ml for C11 +C14 and C11 +C15, respectively).

[0339] The FVI II-CKVWF fusion proteins C9 (“NUWIQ-like”) and C10 (“ReFacto-like”) differ in a short stretch of amino acids in Linker 1 (Table 3).

[0340] To further evaluate if the insertion of EPs has an effect on the expression levels, fusion proteins that contain the same furin cleavage site in Linker 1 and differ only by the presence of EPs in the same linker were compared: C9+C14 / C15 vs. C11 +C14 / C15 and C10+C14 / C15 vs. C12+C14 / C15. Higher FVIII activity levels with C11 , when compared to the values of the C9 and the C10 heterodimers (C9 and C10 represent the “NUWIQ-like” and “ReFacto-like” sequences without additional modules in the first linker), demonstrate that increased expression levels may be mediated by the addition of EPs in the FVI I I-CKVWF portion (Fig. 6A). Specifically, the data demonstrate a positive impact of EPs on expression levels of the molecules containing the “NUWIQ- like” Linker 1 , but not of those containing the “ReFacto-like” Linker 1 .

[0341] The two different FVI I I-CKVWF proteins comprising variants of the “ReFacto-like” B- domain remainder sequence (C12 and C13) were combined with the VWF-CKVWF fusion proteins C14 and C15, resulting in the formation of the four different heterodimers C12+C14 / C15 and C13+C14 / C15. Among them, the highest FVIII:C activity levels were achieved for heterodimers with the C13 fusion protein that contains the GS linker set and the EPs replacing the FVIII B-domain but lacks the cleavage site for the furin protease (4.36 lU / mL for C13+C15, Fig. 6A).

[0342] In summary, the combination of flexible GS linkers and EPs inserted in the first linker of the FVI I I-CKVWF protein combined with the VWF-CKVWF variant without the propeptide resulted in the highest FVIII activity levels, which implicate highest protein stability and correct folding.

[0343] In the second part of the experiment, another six FVI I I-CKVWF fusion proteins differing in number and position of the EPs and the presence and placement of a half-life extending moiety (ABV, Table 3) were combined with three different VWF-CKVWF molecules also differing in the presence and position of the albumin binding single domain antibody (ABV, Table 5). To this end, the 18 combinations of FVI I I-CKVWF and VWF-CKVWF fusion proteins were co-expressed in the human embryonic kidney 293 cell line and their FVIII activity compared to monomeric rFVIll (BDD-FVIII; NUWIQ). All 18 fusion protein combinations showed FVI 11: C activity. Their activity levels (3.4 to 12.8 lU / ml) are significantly higher than the expression level of the rFVIll control (0.22 lU / ml) (Fig. 6B).

[0344] The data indicate that heterodimers formed by FVIII-CKVWF constructs with EPs inserted into both linkers (Linker 1 and Linker 2 in C27, C30 and C31 ) show higher expression levels than those formed by fusion proteins with EPs in Linker 1 only (C11 , C28 and C29). Co-expression of the FVIII-CKVWF portions with VWF-CKVWF chains containing the albumin binding VHH moiety (ABV in C25 and C26) in most cases mediated higher expression levels than their counterparts lacking this modification (C15, Fig. 5B).

[0345] Example 3 - Purification and structural characterization of the FVIII-CKVWF-VWF (FVIII-CKVWF + VWF-CKVWF) protein dimers

[0346] Aims of the

[0347] To assess the size, integrity and the heterodimerization of the expressed fusion proteins, selected constructs were purified. SDS-PAGE analysis was performed to control homogeneity and purity of the expressed constructs and of the dimers formed by their combination. Western Blot analysis was performed to confirm identity and integrity of the formed heterodimers.

[0348] Methods

[0349] Purification of the FVIII-CKvwF-VWF-heterodimers was achieved by applying two consecutive affinity chromatography steps: The first affinity chromatography against VWF was followed by affinity chromatography to FVIII, with intermediate and final rebuffering steps.

[0350] Briefly, the harvested cell culture supernatant was filtered through a 0.2 pm PES filter and loaded directly onto a VWF affinity resin (VOLT select affinity resin, Thermo Fisher Scientific) at ~ 100 CV loading volume and at least 3 min contact time. The column was equilibrated with 0.05 M Tris, 0.1 M NaCI, 0.02% polysorbate 80 pH 7.0 and eluted using 0.05 M Tris, 0.1 M NaCI, 1 M MgCl2 pH 7.0. The eluate was rebuffered into the formulation matrix (171.1 mM NaCI, 7.1 mM L-Arginine, 26.3 mM Sucrose, 3.4 mM tri-sodium citrate, 1.7 mM CaCl2, 0.1 mM Poloxamer 188 pH 7.0) using a Cytiva Sephadex G-25 desalting column for matrix exchange, before loading it onto the FVIII Select affinity resin. The FVIII Select column was equilibrated with 0.3 M NaCI, 0.02 M CaCl2, 0.02 M L-Histidine, 0.02% polysorbate 80 pH 6.5, and eluted with 1.5 M NaCI, 0.02 M CaCl2, 0.02 M L-Histidine, 50% ethylene glycol, 0.02 % polysorbate 80 at pH 6.5.

[0351] Finally, the resulting eluate was rebuffered into the formulation buffer as above, using a Sephadex G-25 desalting column (Cytiva).

[0352] SDS-PAGE

[0353] Samples from different purification steps were analyzed via non-reducing SDS- PAGE. Samples were denatured by incubation with LDS sample buffer. Samples were run on a 4-12 % BisTris gel (Invitrogen, NuPage) for 70 min at 175 V.

[0354] Coomassie staining was conducted using a ready-to-use Coomassie stain (Thermo Scientific, Page Blue Protein staining solution), stained for 3h at RT, washed and subsequently destained in MilliQ water until background cleared.

[0355] Western Blot

[0356] The proteins were transferred to a PVDF membrane using iBIot 2 Transfer stacks (Invitrogen) in an iBIot 2 Gel Dry transfer device according to protocols of the manufacturer (Thermo Fisher Scientific).

[0357] After blotting, the membrane was blocked with TBS Superblock (Thermo Fisher Scientific) and proteins detected by the antibodies SAF8C-AP (to human FVIII, Coachrom) or 19818 / ABIN190120 (to human VWF, QED Bioscience). Blots were further washed and incubated with the corresponding HRP-conjugated secondary antibodies (anti-sheep, Invitrogen, for FVIII, and anti-mouse, BioRad, for VWF). Signals were detected using the Super signal West Pico kit (Thermo Scientific). Results

[0358] The selected FVIII-CKWF-VWF heterodimers from Example 2 were successfully purified to > 90% purity based on the SDS-PAGE analysis. Product related impurities such as FVIII or VWF homodimers were effectively removed in the two consecutive affinity purification steps.

[0359] The results of the analyses are summarized in Figure 7 and 8. Figure 7 shows the purified C9+C15 and C11 +C15 heterodimers in a western blot with FVIII detection (A), VWF detection (B), and SDS-PAGE with Coomassie staining (C). The heterodimers were formed, as indicated by one single band in the Coomassie gel (Fig. 7C). The bands just above the 250 kDa marker band correspond to the FVIII- CKVWF-VWF heterodimers C9+C15 and C11 +C15, since similar signals were detected in western blot analyses with FVIII (Fig. 7A) and VWF staining (Fig. 7B). Minor bands below 250 kDa represent furin cleavage products. Interestingly, the furin cleavage site, although the same in both FVIII-CKVWF molecules (C9 and C11 ) is less efficiently processed in C11 than in C9. This is visible in stronger furin cleavage product bands of the C9+C15 heterodimer at ~75 kDa on the anti-FVIll blot in Figure 7A and at >150 kDa on the anti-VWF blot in Figure 7B. The reduced processing of C11 (in complex with C15) may be attributed to the presence of three EPs in the first linker of the fusion protein and explained by the strong negative charge on the EPs which in turn may prevent the interaction of furin with its cleavage site.

[0360] Figure 8 shows the purified heterodimers from the second expression round from Example 2. These molecules contain an additional half-life extension moiety in the format of an albumin binding single domain antibody. After purification the molecules were analyzed in a western blot with FVIII detection (A), VWF detection (B), and SDS- PAGE with Coomassie staining (C). The following protein dimers are described by the lane numbers in Figure 8: 1 - dimer formed by C27 and C25, 2 - C27+C26, 3 - C30+C25, 4 - C31 +C25, 5 - C30+C26, 6 - C31 +C26, 7 - C31 +C15, 8 - C30+C15, 9 - C11 +C25, 10 - C11 +C26, 11 - C13+C26, 12 - C27+C15, 13 - C28+C26.

[0361] All proteins showed the characteristic band of the heterodimer at >250 kDa, and some minor bands indicating furin cleavage products. This data demonstrates that the CK domain was able to dimerize the monomers irrespective of its position in the fusion protein: at the C-terminus (in constructs C15, C26, C11 , C27, C29 and C31 ) as well as N-terminal to the albumin binding VHH (in constructs C25, C28, C30).

[0362] Example 4 - Binding of the CKvwF-mediated protein dimers to full-length VWF

[0363] Aim of the experiment

[0364] Assessing the capacity of FVIII-CKVWF-VWF heterodimers C9+C15 and C11 +C15 to bind to human full-length VWF.

[0365] Method

[0366] Binding to full-length VWF (fIVWF) was tested by surface plasmon resonance (SPR) on a Biacore T200 instrument (Cytiva). Purified fIVWF (Sekisui) was coated on a CM5 Chip via amine coupling using an amine coupling kit (Cytiva) according to the manufacturer’s instructions. FIVWF was immobilized in three different flow cells at approximately 1000 response units (RU). The running buffer was 20 mM HEPES, 150 mM NaCI, 5 mM CaCl2, 0.05% Tween 20. After each analyte injection, the surface was regenerated with a regeneration buffer (20 mM HEPES, 600 mM NaCI, 350 mM CaCl2, 0.05% Tween 20). The purified FVIII-CKVWF-VWF fusion protein dimers were injected over three different flow cells, in triplicates in a random order at a fixed concentration of 8.5 lll / ml FVIII:C. The binding level, measured 30 sec after end of the analyte injection, was normalized by dividing the RUs by the molecular weight of the respective protein and expressed in % binding of rFVIll set to 100 %.

[0367] Results

[0368] The purified FVIII-CKVWF-VWF heterodimers C9+C15 and C11 +C15 were further analyzed by SPR for their binding to fIVWF. Both heterodimers showed very low fIVWF binding levels of 1.8% and 1.2%, respectively. Such values are comparable to a negative control - the dimeric VWF fragment OCTA12 (0.6% binding, Fig. 9). Example 5 - Pharmacokinetics of the FVIII-CKVWF-VWF protein dimers C9+C15 and C11+C15

[0369] Aims of the experiment

[0370] To explore the impact of the dimerization of the FVIII and VWF fusion proteins and of the presence of the EPs in Linker 1 , the two purified heterodimeric proteins C9+C15 and C11 +C15 were tested in a Hemophilia A (HemA) mice pharmacokinetic study against BDD-FVIII.

[0371] Methods

[0372] 5 to 8-week-old, male B6;129S-F8tm1 Kaz / J (F8- / -) mice were obtained from the Jackson Laboratory (Bar Harbor, Maine, USA). The animals were treated by a tail vein injection of the purified FVIII-CKVWF-VWF protein dimer preparations or rFVIll at a dose of 200 lU / kg bodyweight based on FVIII:C activity. The study outline is summarized in Table 6. The blood samples were collected at the indicated time points. Five animals were used for blood sampling at each time point of each group. Each mouse was used for two sampling points. Blood was collected in tubes with 3.8% Na-citrate solution. Blood samples were placed on crushed ice immediately after collecting and plasma was separated within 1 hour of sampling by centrifugation at 3350 x g, 4°C for 15 min. Plasma samples were stored at -80°C until analysis by FVIII:C assay (Coamatic Factor VIII Assay Kit; Chromogenix, Bedford, MA, USA).

[0373] Table 6: Study outline Results

[0374] The results of the study are presented in Fig. 10 and Table 7. rFVIll showed, as expected, a half-life (T1 / 2) of 7.61 h. Both CKVWF mediated heterodimers, C9+C15 and C11 +C15, showed significantly different PK profiles. With a value of 9.01 h, C9+C15 had only slightly longer T1 / 2 than monomeric rFVIll. C11 +C15 showed the most preferable PK profile with a Ti / 20f 28.21 h (3.7-fold longer than rFVIll), the highest Cmax of 363.6 % (1.2-fold higher than rFVIll) and the highest AUC of 7947.72 h*% (2.3-fold higher than rFVIll).

[0375] Table 7: PK analysis of FVIII :C data measured in HemA mouse plasma

[0376] These results demonstrate that the covalent linking of a FVIII fusion protein to a VWF fragment fusion protein via their CKVWF domains and the addition of EPs by genetic fusion results in the improvement of the heterodimer’s PK parameters in Hemophilia A mice. The most significant improvement comes from the combination of the formation of heterodimers with the specific presence of EPs in the Linker 1 of the FVI I I-CKVWF fusion protein part. Example 6 - Confirmation of the results with FVI 11 and VWF sequence variants

[0377] Aim of the experiment

[0378] To confirm that minor amino acid sequence variations in the CK domain of VWF, in the FVIII heavy chain, the FVIII light chain, in the EPs of the connecting linkers, and in the VWF fragment have no impact on the biological properties of the protein heterodimers.

[0379] Results

[0380] Additional FVI I I-CKVWF fusion proteins based on C11 with sequence variations in one or all of the FVI 11 heavy chain, the EPs (in Linker 1 ), the FVI 11 light chain, Linker 2 and the CKVWF are produced as shown in Table 8. Furthermore, VWF-CKVWF fusion proteins based on C15 with sequence variations in one or all of the VWF fragment and CKVWF, as shown in Table 9, are produced.

[0381] Table 8: Additional FVIII-CKVWF fusion proteins with C11 sequence variants.

[0382] Table 9: Additional VWF-CKVWF fusion proteins with C15 sequence variants.

[0383] With the above FVIII-CKVWF and the VWF-CKVWF protein sequence variants the following heterodimers are produced: Table 10: Composition of the FVIII-CKVWF-VWF heterodimer sequence variants.

[0384] In addition, the experiments according to Examples 2 to 4 are repeated with the sequence variants C11 a to C11f in combination with C15a to C15e.

[0385] Example 7 - Influence of sequence modifications, albumin binding VHHs and additional extension peptides on the pharmacokinetics of FVIII-CKVWF-VWF dimers

[0386] Aims of the experiment

[0387] To explore the impact of a) deletion of the furin cleavage site in the FVIII sequence, b) the insertion of additional EPs into the FVI I I-CKVWF protein portion, and c) a fusion of an albumin binding VHH moiety as complementary parts of the fusion proteins on the pharmacokinetics of the FVI I I-CKVWF-VWF heterodimers formed.

[0388] Methods

[0389] 6 to 8-week-old, male C57BL / 6J mice were obtained from the Janvier Labs (France). The animals were treated by tail vein injection of the purified FVI I I-CKVWF-VWF heterodimers at a dose of 200 lll / kg bodyweight based on FVIII:C activity. The study outline is summarized in Table 11 . The blood samples were collected at the indicated time points. Five animals were used for blood sampling at each time point of each group. Each mouse was used for two sampling points. Blood was collected in tubes with 3.8% Na-citrate solution. Blood samples were placed on crushed ice immediately after collecting and plasma was separated within 1 hour of sampling by centrifugation at 3350 x g, 4°C for 15 min. Plasma samples were stored at -80°C until analysis. The mouse plasma aliquots were analyzed for FVIII activity (FVIII:C). FVIII:C was measured using a capture assay. Briefly, FVIII-CKVWF-VWF protein dimers were captured from mice plasma by an anti-VWF single domain antibody, specific for the human VWF-D'D3 domain (custom made by Thermo Scientific). After blocking and washing steps, FVIII activity was assessed by the Biophen FVIII:C assay (Hyphen Biomed).

[0390] Table 11 : Pharmacokinetic study outline

[0391] Results

[0392] In a pharmacokinetic study in WT mice, 13 heterodimers formed by the FVIII-CKVWF + VWF-CKVWF fusion protein combinations listed in Table 11 were tested in comparison to the C11 +C15 dimer (see Example 5). The protein modules and structures of the various fusion proteins are listed in Tables 3 and 5 (Example 1 ) and illustrated in Figs. 2 to 5. The results of the study are presented in Fig. 11 and Table 12. The heterodimer C11 +C15 showed a 1.8-fold shorter half-life in WT mice (15.42 h) compared to the half-life detected in Hemophilia A mice (28.21 h - Example 5, Table 7). These data are in agreement with the results published by Dumont et al. 2012, who observed similar differences in the half-life of a FVIII-Fc fusion protein in Hemophilia A mice in comparison to WT mouse strains.

[0393] The VWF-CKVWF fusions C25 and C26 contain an albumin binding VHH moiety at the C-terminus of the protein (C25) or inserted into the fusion protein sequence before of the CKVWF domain (C26). In a heterodimer together with the FVIII fusion protein C11 , C25 and C26 show improvement in the terminal half-life compared to their unmodified counterpart C15 (18.3 h and 17.35 h for C11 +C25 and C11 +C26, respectively vs. 15.42 h for C11 +C15, Fig. 11 B. In addition, C11 +C25 and C11 +C26 show a higher protein recovery in circulation (Cmax of 3.15 ll / rnl and 3.47 ll / rnl for C11 +C25 and C11 +C26, respectively vs. 2.79 ll / rnl for C11 +C15). These improvements also result in higher FVI 11 : C plasma levels of C11 +C25 and C11 +C26 detected 96 h after protein administration (Fig. 11 A). This demonstrated that the introduction of an albumin binding VHH into the VWF-CKVWF sequence in positions both C-terminal (in C25) or N-terminal to the CKVWF domain (in C26) leads to significant improvement of the pharmacokinetic properties of the FVIII-CKVWF-VWF heterodimer.

[0394] Further improvement in the pharmacokinetic properties could be observed for heterodimers with the FVIII-CKVWF construct C13, which is a modified version of the C11 molecule devoid of the furin cleavage site contained within the WT amino acid sequence of FVIII. In a heterodimer together with the VWF-CKVWF protein C26, the C13 chain mediated a terminal half-life of 19.43 h and the highest FVIII:C plasma levels detected in the WT mice (0.15 U / mL for C13+C26, Fig. 11 ).

[0395] The FVIII fusion protein C27 contains three additional EPs in the linker between the FVIII-LC and the CKVWF domain (Linker 2) compared to C11 . In combination with C15, this modification results in a 1.15-fold improvement of the terminal half-life of C27+C15 compared to C11 +C15. The insertion of the additional EPs leads to almost two-fold higher FVIII:C levels of the C27+C15 heterodimer in WT mice plasma 96 h after protein administration (Fig. 11 ). The addition of an albumin binding VHH moiety to C15 (i.e. fusion to the C-terminus of the VWF fusion protein after CKVWF, or insertion of the VHH just before the CKVWF domain) yielded C25 and C26, respectively (see above, Figs. 2 to 5). In combination with the FVI 11 fusion construct C27, this modification did not further improve the half- life beyond the values of the heterodimer C27+C15 but had a positive effect on the recovery of the molecule (3.45 and 3.33 ll / rnl for C27+C25 and C27+C26 vs. Cmax of 2.58 ll / rnl for C27+C15, respectively).

[0396] FVI I I-CKVWF fusion proteins with the ABV added to the C-terminus (i.e. placed C- terminal of the CKVWF domain, C28 and C30) mediated an improvement in the half- life of the heterodimer. In comparison to C27+C15 (without ABV) (Ti / 2=17.86h) the half-life of C30+C15 was significantly extended (Ti / 2=25.81 h). However, a second ABV inserted or fused to the VWF-CKVWF binding partner (C26 and C25, respectively) did not facilitate further improvement in the half-life of the dimer, but improved recovery levels (Cmax differences between C30+C15 vs. C30+C25 and C30+C26, Table 12).

[0397] Fusion proteins with the ABV placed N-term inally to the CK domain of the FVI I I-CKVWF part of the heterodimer (C31 ) mediated lower T1 / 2 values and lower FVI 11 activity levels than the molecules with the ABV fused to the C-terminus (C30, Fig. 11 and Table 12).

[0398] Together, these data show that all modifications, the deletion of the FVI 11 cleavage site for furin, the addition of EPs as well as the insertion of a single domain antibody sequence able to bind to albumin into the fusion proteins, can lead to improvement of the pharmacokinetic properties of the heterodimeric FVI I I-CKVWF-VWF molecules. Importantly, the position of the EPs and ABVs in the heterodimers plays a crucial role. The data indicate that placement of the ABV on the C-terminus of the FVI I I-CKVWF part of the heterodimer has the largest potential for improving the half-life of the molecules. However, fusion of the ABV moiety to the VWF-CKVWF part of the heterodimer improves the recovery of FVI 11 activity in mouse plasma. Table 12 - PK analysis of FVIII:C data

[0399] Example 8 - Impact of flexible linkers in the FVIII-CKVWF and VWF-CKVWF fusion proteins on the activity of FVIII-CKVWF-VWF protein dimers

[0400] Aim of the

[0401] To assess the effects of using flexible G4S linkers in positions 1 and 2 of the FVIII- CKVWF fusion proteins and of their usage as Linker 3 in the VWF-CKVWF fusion proteins on the activity of the FVI I I-CKVWF-VWF protein dimers formed.

[0402] Methods

[0403] The FVI I I-CKVWF and VWF-CKVWF fusion constructs were cloned as described in Example 1 and transiently expressed in Expi293F cells in 3 mL format according to the protocols of the manufacturer (Thermo Fisher Scientific). The cell culture supernatant was harvested 4 days post transfection by centrifugation at 4800 x g for 30 min. FVI I l:C activity was assessed by the FVIII chromogenic assay kit (Siemens) on a BCS XP system (Siemens). All expression experiments were performed in two independent replicates.

[0404] Results

[0405] We investigated the impact of additional combinations of G4S linkers and VWF extension peptides (EPs) replacing the FVIII B-domain in the FVII I-CKVWF fusion proteins (Linker 1 ). Effects of the remaining FVIII B-domain amino acids located between the FVIII heavy and light chain (either of “NUWIQ-like” or “ReFacto-like” amino acid sequence, see example 1 for details) on fusion protein function were tested. We further evaluated the functional influence of linkers and EPs present between the FVIII light chain and the CK module of the FVI I I-CKVWF fusion proteins (Linker 2).

[0406] The initial scaffold molecule for the selective deletions of Linker 1 was fusion protein C27. To assure that the findings universally apply to different B-domain remainder sequences, similar deletions of the G4S linkers were performed on C27 and the FVIII- CKVWF fusion protein C32, which differs from C27 in the sequence of Linker 1 (including the “ReFacto-like” Linker 1 -5 instead of “NUWIQ-like” Linker 1 -3). All FVIII- CKVWF versions were co-expressed with the VWF-CKVWF fusion protein C15.

[0407] Three variants of the C32 fusion protein were created: C43 with a deletion of the G4S linkers flanking the EP repeats in Linker 1 (connecting the LC and HC of FVIII), C42 with a deletion of the linkers flanking the EPs in Linker 2 (connecting the FVIII-LC and the CK domain) and C35 with a deletion of all G4S linkers from both, Linker 1 and Linker 2. The fusion proteins are schematically shown in Figure 12A.

[0408] Two variants of the C27 molecule were created: C44 with a deletion of the G4S linkers flanking the EP repeats in Linker 1 (connecting the LC and HC of FVIII) and C45 with a deletion of the linkers flanking the EPs in Linker 2 (connecting the FVIII-LC and the CK domain). The fusion proteins are schematically shown in Figure 13A.

[0409] The FVIII-CKVWF fusion proteins were designed and produced as shown in Tables 13 to 15 (for reference to the linker nomenclature see also Tables 1 and 2).

[0410] Table 13: First linkers used for the FVIII-CKVWF fusion proteins.

[0411] The first linker is the linker connecting the FVIII heavy chain to the FVIII light chain (Fig. 1). EP indicates the VWF EP with the SEQ ID NO: 2. Lower-case numbers indicate a number of repetitions of sequence elements in brackets. For the properties of Linkers 1-1 to 1-5 see Table 1.

[0412] Table 14: Second linkers used for the FVIII-CKVWF fusion proteins.

[0413] The second linker is the linker connecting FVIII light chain to the CKVWF domain of the first fusion protein (Fig. 1). EP indicates the VWF EP with the SEQ ID NO: 2. Lower-case numbers represent the number of repetitions of sequence elements in brackets. For the properties of Linkers 2-1 to 2-4 see Table 2.

[0414] Table 15: Overview of the FVIII-CKVWF fusion proteins and their components.

[0415] In order to explore if the flexible G4S linkers connecting the VWF fragment and the CK domain are necessary for mediating correct folding and heterodimer activity, new molecules were designed. C40 is a version of C15 lacking Linker 3. C41 is a version of C14 lacking Linker 3. The VWF-CKVWF fusion proteins C40 and C41 were designed and produced as shown in Tables 16 and 17. All co-expressions were performed together with the FVIII-CKVWF fusion protein C27. The designs of the generated heterodimers are schematically presented in Figure 14A.

[0416] Table 16: Third linker used for the connection of the VWF fragment with the CKVWF domain.

[0417] The third linker is the linker connecting the CKVWF domain of the second fusion protein to the VWF fragment (Fig. 1). EP stands for the extension peptide with the SEQ ID NO: 2. The lower-case numbers indicate the number of repetitions of the sequence elements in brackets.

[0418] The layout of the VWF-CKVWF fusion proteins is shown in Table 17.

[0419] Table 17: Overview of the VWF-CKVWF fusion proteins and their components.

[0420] In the first round of expression, four different versions of FVI I I-CKVWF molecules harboring different combinations of GS linkers and EPs inserted in the “ReFacto-like” B-domain remainder sequence were produced (i.e. C32, C35, C42 and C43) (Figure 12A). The different FVI I I-CKVWF proteins were combined with the same VWF-CKVWF fusion protein C15, resulting in the formation of the four different heterodimers C32+C15, C35+C15, C42+C15 and C43+C15. The C32+C15 heterodimer achieved FVI 11 : C activity levels of 8.01 ILI / mL in the expression supernatant. For C35 containing both EP modules, but lacking the GS linkers flanking them, or C42, harbouring only the GS linker set and the EPs replacing the FVI 11 B-domain, the FVI 11 activity of heterodimers together with C15 was significantly reduced to 2.97 ILI / mL and 2.9 ILI / mL, respectively. In contrast, C43, that lacks the GS linker set flanking the EPs in the B-domain but contains the C-terminal GS linker set embracing the EPs in front of the CKVWF domain, achieved FVIII:C levels of 7.22 ILI / mL in combination with C15, an activity level just slightly lower than the FVI 11 activity of the heterodimer C32+C15 that comprises both entire EP and GS linker sets (Fig. 12B).

[0421] In the second round of expression, three versions of FVI II-CKVWF molecules having a “NUWIQ-like” remainder sequence of the B-domain were studied in combination with C15. The combinations resulted in the formation of the three different FVI I I-CKVWF- VWF heterodimers C27+C15, C44+C15, and C45+C15 (Fig. 13A). The heterodimer formed by C27 (comprising both entire EP plus flanking GS linker sets) and C15 achieved FVIII:C levels of 3.46 ILI / mL in the expression supernatant. For the heterodimer of C15 and C45, which lacks the C-terminal GS linker set flanking the EPs in front of the CKVWF domain, the FVIII activity was significantly reduced to 0.66 ILI / mL. The heterodimer of C15 with C44, devoid of the GS linker set flanking the EPs in the B-domain but having the linker set flanking the EPs upstream of the CKVWF domain, achieved FVI ll:C activities of 3.26 ILI / mL. This activity is just slightly lower than the activity of the FVIII-CKVWF-VWF heterodimer C27+C15 (3.46 ILI / mL), harbouring both GS linker sets (Fig.13B).

[0422] In summary, for FVIII-CKVWF molecules independent of the B-domain remainder sequence, the G4S linker set flanking the EPs in the B-domain is dispensable for the correct folding, stability and FVIII activity of the FVIII-CKVWF- VWF-CKVWF heterodimers. In contrast, the G4S linkers flanking the EPs placed in front of the C- terminal CKVWF domain (i.e., between the FVIII light chain and the CK domain) are essential for correct folding and FVIII activity of the FVIII-CKVWF- VWF-CKVWF heterodimers.

[0423] For the VWF-CKVWF part of the heterodimer, four molecules were compared, all co- expressed with the same FVIII-CKVWF partner C27 (schematically shown in Figure 14A). The results of the FVIII:C activity measurements of the formed heterodimers are presented in Figure 14B. The results show that for both VWF-CKVWF molecule variants (C14 and C15) the flexible linker between the VWF and the CK domain is dispensable and does not have any significant impact on the heterodimer folding and activity.

[0424] Example 9 - Impact of extension peptides in the second linker on the thrombin generation kinetics and in vivo efficacy of FVIII-CKVWF-VWF heterodimers

[0425] Aim of the experiment

[0426] To assess the impact of the extension peptide repeats in the second linker (specified in Table 2). Methods

[0427] Two FVIII-CKVWF-VWF dimers, C32+C15 and C37+C15, differing only by the presence of three extension peptide repeats in linker 2 of the FVIII-CKVWF part (C32 containing linker 2-2 with the EPs and C37 containing linker 2-1 without the EPs), were compared in a thrombin generation assay (TGA) as well as in a Tail-Cut bleeding Assay in Hemophilia A (HemA) mice.

[0428] A thrombin generation assay (TGA), developed by Thrombinoscope BV (Netherland) was used to monitor the thrombin generation in real time. The method is based on fluorometric detection of a peptide substrate, which undergoes thrombin-mediated conversion into a fluorescent compound that can be quantified using a photometer. The assay is a general physiologic function test of the thrombotic hemostatic system. The assay is used to measure kinetics of thrombin generation in FVIII-deficient plasma after addition of FVIII or FVIII-containing samples. The measurement is carried out at 37°C in 96-well plates using a Fluoroskan Ascent FL fluorometer. Thrombin generation is monitored by thrombin-mediated activation of a fluorogenic substrate. Four key parameters indicating the activity of a FVIII construct were monitored by TGA: Lag time (time from activation of the coagulation cascade until thrombin is first detected), ETP (edogenous thrombin potential - area under the thrombin signal curve), peak thrombin (maximal thrombin concentration) and ttp (time to peak - time until the maximal thrombin concentration is reached). All parameters are presented as percent of the value obtained for BDD-FVI 11, which was used as a control in each experiment.

[0429] The in vivo performance of both constructs in comparison to a BDD-FVI 11 was assessed by a Tail-cut bleeding study in HemA mice. After an acclimatization period of at least 10 days, the male B6;129S-F8tm1 Kaz / J (F8- / -) mice were treated by a single i.v. injection of BDD-FVIII, C32+C15 of C37+C15, with 75 lU / kg bw. A total of 10 mice per group were treated. Every protein dimer was administered in an application volume of 5 ml / kg bw. One minute after administration, the animals were anaesthetized (80 mg / kg, 10 ml / kg pentobarbital sodium i.p.) and it was ensured that tolerance stage II had been reached. After 6 minutes, animals were subjected to a tail-clip, where 3 mm from the tip of the tail was amputated with a sharp scalpel blade. Immediately afterwards, the animals were placed on a temperature-controlled bleeding rack and the tail was hung in a sample tube filled with 0.9% saline solution and kept at 37°C. The blood was collected over a period of 30 minutes and total blood loss determined by spectrophotometric hemoglobin measurement at 405 nm. Using hemoglobin standards prepared from defined blood volumes, blood loss was calculated and expressed as volume of lost blood (pl).

[0430] Results

[0431] The functional comparison of two constructs differing only by the presence or absence of EP repeats in linker 2 of the FVIII part (C32 vs. C27, schematically depicted in Fig. 15A) revealed functional differences in construct performance. In the TGA assay, both constructs did not differ in the ETP and peak thrombin parameters. However, and surprisingly, both parameters connected to the kinetics of thrombin generation differed between the two molecules tested. Specifically, C32+C15, which contains three EP repeats in linker 2, showed shorter lag times as well as shorter time to peak when compared to C37+C15 without these EPs (Fig. 15B).

[0432] HemA mice treated with C37+C15 lost more blood when compared to mice treated with the same dose of C32+C15, which further supports a significant impact of the presence of EPs in vivo (Fig. 16).

[0433] Overall, these findings demonstrate the functional importance of EP-repeats in linker 2. Since linker 2 contains a thrombin cleavage site, adjacent EP-repeats appear to have a positive effect on the activation of FVIII and thereby on the thrombin generation kinetics mediated by the construct.

[0434] Example 10 - Design, cloning, expression, and CKVWF- and CKNDP-mediated dimerization of single domain antibody fusion proteins

[0435] Aims of the

[0436] Identification and cloning of single domain antibodies to human coagulation factors FIX and FX. Generation of cDNAs encoding different single domain antibody fusions with the CK domains of human Norrin (VH H-CKNDP fusion proteins) or human VWF (VH H-CKVWF fusion proteins). Cloning of cDNAs into expression vectors, recombinant expression of selected VHH-CKVWF and VHH-CKNDP fusion proteins, purification and testing of their ability to dimerize. Methods

[0437] The sequences of VHH1 and VHH2 were obtained from a camelid immune library after immunization of llamas and alpacas with native human plasma-derived coagulation factors FIXa and FX. Immunization, isolation of peripheral blood mononuclear cells (PBMCs) and phage display were performed according to standard protocols used by the service provider (Abcore Inc., USA). Briefly, multiple production bleeds were collected throughout animal immunization. Anti-FIXa and anti-FX serum titers were monitored by ELISA. At the end of the immunization protocol, PBMCs were isolated from the production bleeds with the highest target-specific antibody titers, followed by RNA extraction. The extracted RNA was reversely transcribed and the resulting cDNAs used for construction of the VHH-encoding DNA libraries. Both libraries were displayed on the surface of phages and subjected to multiple rounds of affinity-based selection against the relevant target protein (FIXa and FX). From the resulting phage pools enriched in target-specific binders, individual clones were tested by ELISA. The genes of the positive clones were analyzed by Sanger sequencing and their target-binding properties confirmed using Octet (BLI) technology.

[0438] Gene synthesis, cloning and protein expression

[0439] Codon-optimized cDNAs encoding the fusion proteins were ordered from IDT (Integrated DNA Technologies Inc.). The plasmids were constructed through one-step cloning as described in Example 1 , resulting in the respective cDNA cassette inserted into the expression vectors. The plasmids’ identities were verified by sequencing. The proteins were transiently produced in Expi293F cells. Expression and secretion of the proteins was monitored by Western Blot after SDS-PAGE under non-reducing conditions with anti-His tag (GenScript A00186-100) and anti-Strep tag (Abeam ab97023) antibody detection. Western blot analyses of cell supernatants were performed 48 hours after transfection. On day 7, cell culture supernatants were harvested and either purified immediately, or frozen at -70 °C. Protein containing supernatants were analyzed by SDS-PAGE with Coomassie staining under reducing and non-reducing conditions. Purification and analysis of heterodimers of the VHH-CKVWF fusion proteins

[0440] Purification of the VHH1 -CKVWF-VHH2 dimers was achieved by applying two consecutive affinity chromatography steps: Affinity chromatography against the Twin- Strep-Tag was followed by affinity chromatography towards the His-Tag, with a final rebuffering step.

[0441] Briefly, the harvested cell culture supernatant was loaded onto a Strep-Tactin XT 4Flow resin (IBA Lifesciences). The column was equilibrated with 100 mM Tris / HCI pH 8, 150 mM NaCI, 1 mM EDTA and eluted using 100 mM Tris / HCI, 150 mM NaCI 1 mM EDTA, 50 mM biotin, pH 8.0. For the second affinity step, a HisTrap Excel column (Cytiva) was equilibrated with PBS 350 mM NaCI, and eluted with PBS 350 mM NaCI, 0.5 M imidazole, pH 7.5. Finally, the resulting eluate was rebuffered into PBS (for heterodimers H2 and H3) or 150 mM sodium phosphate buffer, 150 mM NaCI pH 6.5 (for H1 , H5 and H6) using a HiTrap desalting column (Cytiva).

[0442] Results

[0443] The composition of the fusion proteins is indicated in Table 19. A schematic overview of their structure is shown in Figure 17.

[0444] Table 18: Linkers used for the VHH-CKVWF and VHH-CKNDP fusion proteins

[0445] The position of the linkers in the fusion proteins is indicated in Fig. 17. EP indicates the VWF- derived extension peptide with SEQ ID NO: 2. The lower-case numbers indicate the number of repetitions of the sequence elements in brackets. Table 19: Overview of the VHH-CK fusion proteins and their components

[0446] For all fusion protein combinations analyzed by Western Blot, the signals of both tags were detected at the molecular weights expected for a dimeric protein. Hence, all tested combinations of the VHH-CKVWF and the VHH-CKNDP fusion proteins formed dimers (Figs. 18 and 19). All the CKvwF-mediated heterodimers listed in Table 19 (i.e. H1 , H2, H3, H5, and H6) could be purified via the two-step affinity chromatography purification protocol described above. The purified heterodimers were analyzed by SDS-PAGE and displayed the expected molecular weights under reducing and under non-reducing conditions (Fig. 20).

[0447] The results demonstrate the ability of the CK domains of human Norrin (CKNDP) and of human VWF (CKVWF) to mediate dimerization, independent of the position of the CK domain in the fusion protein. VHH 1 -CKVWF-VHH2 heterodimers in various layouts, with the CKVWF domain placed either on the C-terminus or at the N-terminus of both fusion proteins, or with the CKVWF domain placed C-terminally in the first and N- terminally in the second fusion protein, were successfully formed. These configurations offer a variety of options for the assembly of the CK domain fusion partners as depicted in Fig. 17. Moreover, not only dimerization, but also tri- and tetramerization of the fusion proteins is possible. The CK domains of human VWF and Norrin (NDP) may thus provide a universal tool for the di-, and potentially also for the tri- and tetramerization of any combination of soluble proteins.

[0448] Furthermore, EPs were successfully included in the fusion proteins, which represents one possible mechanism for the half-life prolongation of the dimeric single domain antibody molecules.

[0449] Example 11 - Stability and functionality of heterodimers formed by VHH-CKVWF fusion proteins

[0450] Aims of the

[0451] To investigate the stability of the VHH-CKVWF protein dimers in comparison to heterodimers obtained by dimerization of the same VHH domains in fusions to the Fc domain of lgG1. To analyze the biological functionality of the heterodimers, i.e. simultaneous binding to activated coagulation factor IX (FIXa) and coagulation factor X (FX) as well as testing their mimetic FVIII activity.

[0452] Methods

[0453] Further to the VHH-CKVWF fusion proteins of Example 10, four additional cDNAs encoding single domain antibody fusions with the Fc domain of lgG1 were generated, cloned into the respective expression vectors, and expressed in HEK293 cells. The single domain antibody-lgFc fusions used the same two VHH sequences as described above for the CKVWF fusions (i.e. VHH1 against human coagulation factor FIX and VHH2 against human FX). Using the knob-into-hole technology for efficient Fc chain pairing (Ridgway et al., 1996), two different heterodimers, H4 and TPP349, were successfully formed by the four different VHH-lgFc fusion proteins. H4 is formed by the fusion proteins VHH1 -lgFck-Twin-Strep (SEQ ID NO: 66) and VHH2-lgFch-His (SEQ ID NO: 67) and TPP349 is formed by the fusion proteins VHH1 -lgFck (SEQ ID NO: 70) and VHH2-lgFch(SEQ ID NO: 71 ). H4 and TPP349 differ in the absence (H4) or presence (TPP349) of the lgG1 hinge region, in the presence (H4) or absence (TPP349) of C-terminal His- and Strep-tags and in the sequence of the linkers connecting the Fc scaffold with VHH1 and VHH2. The layouts of the Fc-dimerized proteins are schematically shown in Figure 21 and described in Tables 20 and 21 .

[0454] Expression and purification of the Fc-dimerized H4 was performed as described for the CK dimers in Example 10. The second Fc-mediated dimer TPP349 was formed by fusion proteins expressed without His- and Strep-tags and purified by protein A (Amsphere A3) affinity chromatography. Capture and subsequent elution with 100 mM glycin, pH 2.7 was followed by immediate neutralization and buffer exchange to PBS. Mass spectroscopy analysis of the final protein sample indicated that the content of mispaired and monomeric species was less than 10% (data not shown).

[0455] Dynamic light scattering and melting temperature analysis

[0456] Dynamic light scattering (DLS) and melting temperature (Tm) analysis was performed with the Prometheus Panta Instrument from Nanotemper. Samples were diluted to 0.2 mg / ml in buffer (1x PBS pH 7.4 for H2, H3, TPP349 and H4; 150 mM Na- Phosphate / 150 mM NaCI pH 6.5 for H1 , H5 and H6) and transferred to standard capillaries. Each sample was measured in two separate experiments in triplicates. DLS measurements were carried out using the high sensitivity mode with 10 acquisitions for each capillary at 100% laser intensity and 25°C. Viscosity and refractive index of 1x PBS pH 7.4 and 150 mM Na-Phosphate / 150 mM NaCI pH 6.5 were calculated by the BufferBuilder-Tool of the PR. Panta Control Software v1 .6.3 for 20°C and a laser wavelength of 405 nm. For 1x PBS pH 7.4, a viscosity of approximately 1.020 mPa*s and a refractive index of 1.332 was calculated. The values for 150 mM Na-Phosphate / 150 mM NaCI pH 6.5 were 1 .082 mPa*s and 1 .338, respectively. Size distribution analysis was performed by the PR. Panta Analysis Software v1 .6.3.

[0457] Thermal unfolding and refolding were carried out at a thermal ramp from 25°C to 95°C and back to 25°C with a heating and cooling rate of 1 °C / min and at 100% laser intensity by simultaneous acquisition of nano differential scanning fluorimetry (nanoDSF) and DLS. Tm determination was carried out by nanoDSF using the first derivative of the ratio of light emission at 350 nm and 330 nm. Aggregation was monitored by the increase in the hydrodynamic radius (Cumulant radius) measured by DLS. All evaluations were performed with the PR. Panta Analysis Software v1 .6.3. Shaking stress

[0458] Shaking stress was performed by orbital shaking using a ThermoMixer® C from Eppendorf. 1 .5 mL Eppendorf tubes were filled with 60 pL of sample (0.2 mg / mL; 150 mM sodium phosphate, 150 mM NaCI pH 6.5) and shaken at 1850 and 2000 rpm for 3h or at 1850 rpm for 3, 5 or 7h. The temperature during shaking stress was adjusted to 20°C. Stressed and unstressed samples were analyzed by size exclusion chromatography on a HPLC UltiMate 3000 system (Thermo Fisher Scientific) equipped with a Superdex 200 Increase 10 / 300 GL column (Cytiva). The analysis was performed with a flow rate of 0.56 mL / min, 1x PBS pH 7.4 as a mobile phase and an injection volume of 45 pL. Chromatograms were collected at 280 nm and integration of the absorption trace was performed using Chromeleon 7.2 (Thermo Fisher Scientific). The area of the main peak recorded at 280 nm was used to calculate the recovery of the protein after shake stress. pH shift stability

[0459] The pH shift to pH 3.5 was induced by adding 100 pL of sample (0.2 mg / mL) to a defined volume of HCI (0.6 pL of 1 .5 M HCI for 1X PBS, pH 7.4 (H2, H3 and TPP349) and 2.34 pL of 3 M HCI for 150 mM sodium phosphate, 150 mM NaCI, pH 6.5 (H1 , H5 and H6)), followed by rapid mixing to ensure homogenous pH lowering within the entire sample volume. After incubation for 100 min, the pH was neutralized by transferring the sample to a new tube containing the same volume of equimolar NaOH as the previously added HCI. Unstressed and stressed samples were analyzed via SE-HPLC (see shaking stress). Due to the introduction of an inconsistent volumetric dilution, by addition of HCI and NaOH, the relative area of high molecular weight compounds (HMWC) was compared prior to and after the pH shift.

[0460] Bio-Layer Interferometry (BLI)

[0461] Simultaneous binding to FIXa and FX was assessed by Bio-Layer Interferometry (BLI) using the Octet BLI Discovery V13.0 apparatus and data analyzed using Octet Analysis Studio V13.0. The VHH1 -CKVWF-VHH2 and the VHH1 -Fc-VHH2 heterodimers were loaded in a concentration of 2 pg / mL onto AR2G (Amine Reactive 2ndGen) sensors in 10 mM acetate buffer at pH 5.5. A reference sensor loaded with Efgartigimod at 2 pg / mL was used (ArgenX, Fc-receptor antagonist drug that does not bind to any of the coagulation factors FIXa and FX).

[0462] Association of activated human coagulation factor FIXa and FX (Haematologic Technologies Inc.) was performed in CAB-T buffer (10 mM HEPES, 100 mM NaCI, 2.5 mM CaCl2 with 0.05% (w / v) Tween-20 pH 7.4) at concentrations of 60 nM and 100 nM, respectively. All association steps were recorded at 30 °C and at 100 rpm shake speed. Assays were performed in the order of A) association of FIXa first, then FX, or B) association of FX first, then FIXa. In the second step of the association experiments the concentration of the first coagulation factor was kept constant for the recorded signal to only be dependent on the newly added coagulation factor. Amine coupling to the BLI sensor did not work out for heterodimers H2 and H3. Hence, these constructs as well as heterodimer H4 were biotinylated using EZ-Link NHS-PEG4- Biotin according to the manufacturer’s instructions (Thermo Fisher Scientific) in a 3:1 molar coupling ratio. In a loading scout, ideal loading concentrations for H2 and H3 were determined at 2 pg / mL and at 1 pg / ml for heterodimer H4. These concentrations were loaded onto Streptavidin sensors (SA sensors) and the same association steps as above were applied for FIXa and FX. Analysis included subtraction of the unspecific signal, recorded by the reference sensor loaded with Efgartigimod.

[0463] The mimetic FVIII activity of the heterodimers was assessed using the BIOPHEN FVI 11: C kit (Hyphen Biomed) according to protocols provided by the manufacturer.

[0464] Results

[0465] The molecular architecture of the VHH1 -lgFc-VHH2 heterodimers is displayed in Fig. 21 and described in Tables 20 and 21 .

[0466] Table 20: Linkers used for the VHH-Fc fusion proteins.

[0467] The position of the linkers in the fusion proteins is indicated in Fig. 21. The lower-case numbers indicate the number of repetitions of the sequence elements in brackets.

[0468] Table 21 : Overview of the VHH-Fc fusion proteins and their components.

[0469] In the following experiments, the VHH1 -Fc-VHH2 dimers were compared to the VHH1 -CKVWF-VHH2 dimers described in Example 10.

[0470] Size distribution and thermal stability

[0471] The hydrodynamic radii (Rh) of the VHH1 -CKVWF-VHH2 and the VHH1 -Fc-VHH2 heterodimers experimentally determined by DLS are presented in Table 22. Among the VHH1 -CK-VHH2 dimers analyzed, only H2 and H3 (molecules bearing the EPs) showed a single particle population, which is reflected in only one peak detected in the DLS analyses at a hydrodynamic radius of approximately 6 nm. All other VHH-CK dimers (H1 , H5 and H6) showed the presence of two particle populations; one in the range of 4 to 5 nm and a second population with a hydrodynamic radius of approximately 30 nm (Fig. 22). This indicates a lower tendency of H2 and H3 to form higher molecular weight compounds, most probably due to the multiple negative charges provided by the sialylated O-glycans of the EPs. Thermal stability was investigated by recording the changes in the intrinsic fluorescence of the proteins upon thermal unfolding via nanoDSF. The melting temperature (Tm) analysis revealed higher thermal stability for all the VHH-CK molecules (Tm ranges from 71.2°C to 73.1 °C) in comparison to the VHH-Fc molecules (Tms of 64.7°C and 68.1 °C, Table 22). Table 22. DLS analysis of the size (intensity distribution of radius) and the melting temperature (Tm) of the VHH1-CK-VHH2 and the VHH1-Fc-VHH2 dimers.

[0472] After heating to 95°C and then cooling again to 25°C, a refolding process was observed for H1 , H5 and H6, visible by the appearance of an inflection point in the 350nm / 330nm graph during cooling of the capillary (Fig. 23A, upper panel, parameter Ratio 350nm / 330nm and middle panel, parameter Ratio First Derivative, in the right part of the diagram). In addition, no protein aggregation was registered for the VHH1 - CKVWF-VHH2 dimers (Fig. 23A, lower panel, parameter Cumulant Radius). In contrast, protein refolding was not observed for the IgFc-based dimers H4 and TPP349, where (after initial protein unfolding) no inflection point in the 350nm / 330nm graph during cooling of the sample could be detected (Fig. 23B, upper and middle panel). In addition, protein aggregation was observed during unfolding analysis of the TPP349 molecule, visible by the increase of the cumulant radius at higher temperatures (Fig. 23B, lower panel).

[0473] Shaking stress stability

[0474] The shaking stress stability of the molecules was investigated in two separate experiments. In the first experiment, one representative of the group of the IgFc- and the CKvwF-based heterodimers was subjected to shaking stress at 1850 and 2000 rpm for 3h in solution. The results of the SEC analyses of TPP349 (IgFc-based) and H1 (CKvwF-based) after shaking stress are presented in Fig. 24. The SEC analysis shows how much intact product remains after shaking stress was applied. The portion of the investigated protein not visible in the SEC analysis is assumed to be aggregated and / or adsorbed to surfaces via exposed hydrophobic patches. ForTPP349, less than 40% of intact heterodimer was detected after 3h shaking at 1850 rpm, and almost no intact product was observed after 3h shaking at 2000 rpm. In contrast, more than 60% of H1 were still intact after both shaking stress conditions. To understand the fluid dynamics and to identify the threshold shaking speed crucial for potential protein unfolding, the motion of the liquid as a function of shaking speed was evaluated by video recording. Shaking up to ~1750 rpm led to a rotation of the fluid along the tube wall (data not shown). Above 1750 rpm a marked change in the behavior of the liquid was observed, resulting in a continuous turn-over of the air-liquid interface and hence in an increased interfacial stress. Based on these results, a shaking speed of 1850 rpm and a duration of 3, 5 and 7h were selected for the second experiment, where the complete set of the CKvwr-based heterodimers was compared to TPP349. For TPP349, less than 47% of intact heterodimer was detected after 3h shaking at 1850 rpm. In contrast, more than 75% of the VHH-CK heterodimers (H1 , H2 H3, H5 and H6) were still intact after shaking for 3h, this did not change significantly by extending the shaking period to 5 or 7 hours (Table 23).

[0475] Table 23: Shaking stress stability of VHH-Fc (TPP349) and VHH-CK constructs shown as the recovery of the main peak area after shaking at 1850 rpm for 3, 5 and 7 hours.

[0476] The data demonstrate a superior stability of the CKvwF-mediated heterodimers against shaking stress compared to the IgFc-based dimers. Improved stability may represent a major advantage in biotechnological production, where proteins are subjected to constant agitation during mixing, ultrafiltration / diafiltration, pumping, filling and shipping of the final product. pH shift stability

[0477] The production of biopharmaceuticals requires virus inactivation, typically including steps of prolonged incubation at a low pH followed by adjustment to the formulation pH.

[0478] The ability of the dimeric proteins to withstand pH-shifts that occur in the manufacturing process steps (e.g. elution at low pH) was assessed by the formation of high molecular weight compounds (HMWC). The formation of HMWC before and after pH-shift treatment was analyzed via size-exclusion chromatography (SEC). The CKvwF-based heterodimers H1 , H2, H3, H5 and H6 were compared to the IgFc-based heterodimer TPP349 carrying the identical VHHs. pH-shift treatment caused HMWC formation of the IgFc-based dimer TPP349, but not of the CKvwF-dimer sample H5. This difference in stability is portraited by comparing the SE chromatograms of H5 and TPP349 before and after the pH shift. The Fc-fusion construct exhibits multiple distinct species of HMWC at different retention times, indicating more than one mode of aggregation (Fig. 25A). Furthermore, TPP349 is the only construct that additionally exhibits a small portion of low molecular weight species after treatment, indicating fragmentation of the protein. In contrast, pH-shift treatment of the CKvwF-based heterodimers did not significantly change the fraction of HMWC with marginal difference between the constructs (Fig. 25B). A slight decrease in HMWC can be explained by loss of protein due to adsorption to surfaces during handling (e.g. adsorption to container walls, pipette tips, HPLC tubings, and column frits).

[0479] The findings of this experiment underline the superior stability of CKvwF-mediated dimers over the IgFc-fusion protein dimers under manufacturing process conditions.

[0480] Binding to targets

[0481] All single domain antibody fusion proteins, with heterodimer formation either mediated by CKVWF or by IgFc fusion, were analyzed by Bio-Layer Interferometry in both orders of binding to FIX and to FX. Based on the data presented in Table 24, all VHH heterodimers were able to simultaneously bind to both coagulation factors, FIX and FX. The data demonstrate that, independent of the position of the VHH portion relative to the CK domain (i.e. fusion to the N- or the C-terminus), both single domain antibodies retain their functionality, i.e. the ability of binding to their targets FIX and FX. Table 24. Binding of FIX and FX to the VHH1 -CKVWF-VHH2 and VHH1-FC-VHH2 heterodimers

[0482] Mimetic FVIII activity

[0483] We investigated if the VHH1 -CKVWF-VHH2 dimers are not only able to bind FIX and FX but may also be able to generate (mimetic) FVIII activity. To this end, the purified heterodimers H1 , H2, H3, H5 and H6 were analyzed in a chromogenic FVIII activity (FVIII:C) assay. The results of the measurements are depicted in Fig. 26. The FVIII activity of the heterodimers and the FVIII activity of standard human plasma (SHP) are expressed as absorbance at 405 nm (baseline corrected mean), which represents the amount of active FX generated in the assay. The heterodimers were measured in the protein concentrations indicated on the X-axis. In contrast to H1 , H2, H3, H4 and TPP349 that showed no activity, heterodimers H5 and H6 displayed significant mimetic FVIII activity. Interestingly, and contrasting the situation in H1 to H3, in H5 and H6 the FIXa-binding VHH arm of the heterodimer (fusion protein CKVWF-VHH1 ) is fused to the C-terminus of the CK domain. Hence, the unique properties of the C- terminus of the CKVWF domain dimer allows to place the VHH-bound coagulation factors FIXa and FX in a favorable orientation and results in a mimetic FVIII activity of the heterodimers H5 and H6. This demonstrates that the CKVWF domain-based dimerization offers a variety of possible relative arrangements of molecules which are not possible with other state-of-the-art dimerization techniques.

[0484] Example 12 - Cloning, expression, and purification of tri- and tetravalent heterodimers of VHH-CKVWF fusion proteins

[0485] Aims of the experiment

[0486] Design and generation of cDNAs encoding mono- and bivalent VHH-CKVWF fusion protein monomers. Simultaneous expression of the cDNAs and testing the ability of the secreted fusion proteins to dimerize to tri- and tetravalent heterodimers. Purification of the secreted heterodimers from the cell culture supernatant.

[0487] Methods

[0488] Gene synthesis, cloning and protein expression

[0489] Codon-optimized cDNAs were ordered from IDT (Integrated DNA Technologies Inc.). The plasmids for protein expression were constructed through one-step cloning, resulting in vectors encoding secreted proteins. Their coding regions were verified by sequencing. The expression vectors were chemically transfected, and proteins transiently produced in HEK 293 ALL cells. The cells were counted, and their viability measured at the time of transfection and at production end. The plasmids’ identities were verified by sequencing at 48 h post-transfection. The expression and secretion of the proteins were monitored by Western Blot after SDS-PAGE with anti-His tag (GenScript, A00186-100; 1 :5000) or anti-Strep tag (Abeam, ab97023; 1 :8000) detection. Western blot analyses were run 48 hours post-transfection. At production end, the supernatants were analyzed by SDS-PAGE with Coomassie staining under reducing and non-reducing conditions. On day 7, the cell culture supernatant was harvested and either purified right after harvest, or frozen at -70°C.

[0490] Purification and analysis of the heterodimeric nanobody-CKvwp fusion proteins

[0491] Purification of the nanobody-CKvwF-fusion proteins was achieved by applying two consecutive affinity chromatography steps: Affinity chromatography against Twin- Strep-Tag was followed by affinity chromatography to the His-Tag, with a final rebuffering step. Briefly, the harvested cell culture supernatant was loaded onto a Strep-TactinXT 4Flow resin (IBA Lifesciences). The column was equilibrated with 100 mM Tris / HCI pH 8, 150 mM NaCI, 1 mM EDTA and eluted using 100 mM Tris / HCI, 150 mM NaCI, 1 mM EDTA, 50 mM biotin, pH 8.0. For the second affinity step a HisTrap Excel column (Cytiva) was equilibrated with PBS, 350 mM NaCI, and eluted with PBS, 350 mM NaCI, 0.5 M imidazole, pH 7.5. Finally, the resulting eluate was rebuffered into PBS using a HiTrap desalting column (Cytiva).

[0492] Results

[0493] The components and structure of the fusion proteins and of the heterodimers are shown in Table 25 and Figures 27 and 28.

[0494] Table 25: Overview of the nanobody CKVWF fusion proteins and their components.

[0495] The above molecules were expressed with the aim to increase the stability, prolong the half-life and to attach functional groups to the termini of the simple protein dimers described in Example 10. An anti-albumin single domain antibody was selected as a functional group, enabling the dimeric complex to bind to albumin in the circulation.

[0496] Since rats were used for the in vivo testing of the new molecules, the anti-rat serum albumin VHH R28 was chosen as ABV2 (van Faassen et al., 2020).

[0497] All nanobody-CKvwF dimers were expressed and secreted as protein dimers into the cell supernatant, as signals of both tags were detected at the expected molecular weight of a respective dimer (data not shown). The proteins were purified via the two- step affinity chromatography purification protocol described in Example 10. The purified heterodimers were analyzed by SDS-PAGE under reducing and non-reducing conditions (Fig. 29). The results demonstrate the versatility of our fusion protein dimerization platform offering up to four termini for the fusion of functional protein groups.

[0498] Example 13 - Pharmacokinetic properties of the CKvwF-mediated VHH heterodimers in rats

[0499] Aim of the experiment:

[0500] Appropriate pharmacokinetic properties (PK) are required for successful implementation of the CK-mediated anti-FIX / anti-FX VHH heterodimers in therapeutic molecules that mimic FVIII function. We investigated if the addition of half-life extension modules is required to improve these properties, in order to accomplish sufficiently long dosing intervals.

[0501] The experiments were conducted

[0502] - to discover the pharmacokinetics (PK) of the fusion protein dimers after intravenous (IV) administration in rats.

[0503] - to investigate the impact of the fusion of the anti-rat serum albumin ABV2 to the fusion protein dimers on their PK after IV administration.

[0504] - to examine the impact of a fusion of different numbers of EPs in different positions of the mimetic FVIII proteins on the PK after IV administration.

[0505] - to investigate the PK characteristics of different heterodimers with or without ABV2 and / or added EPs after subcutaneous (SC) administration.

[0506] Methods

[0507] Study design

[0508] The study was conducted in Sprague-Dawley / CD rats. The animals were distributed in 20 groups (3 animals per group). 14 groups received IV administration of the tested molecules, and 6 groups obtained SC administration as described in Table 26. The purified heterodimeric fusion proteins (H3, H5, H17, H18, H19, H20, H22, H28, H30, H31 , H32, H33, H34, and H36 - see examples 10 and 11 ) were quantified by pBCA assay (Thermo Fisher Scientific) for dose adjustment. The smallest protein (H5) was dosed at 1 mg / kg body weight (bw) for IV administration and 2 mg / kg bw for SC administration. All other FVIII mimetic proteins were dosed at the same molar dose as H5, adjusted based on their respective molecular weight (Table 26). At each timepoint, samples of 200 pl blood were gained by tail-vein bleeding in Na-Citrate tubes and stored at -80°C until analysis. The following timepoints were collected after IV administration: 0 min (predose), 5 min, 1 h, 6h, 12h, 24h, 48h, 60h, 72h, 96h, 120h, 192h. After SC administration, samples were collected at the timepoints 0 min (predose), 1 h, 12h, 24h, 48h, 60h, 72h, 96h, 120h, and 192h.

[0509] Quantification of the fusion protein dimers in rat plasma

[0510] The plasma concentration of the protein dimers was determined via immunoassay using a Gyrolab xPlore instrument (Gyros Protein Technologies, Uppsala, Sweden). The assay setup was as follows: A polyclonal anti-VWF antibody (Dako A / S) was biotinylated using the EZ-Link Sulfo-NHS-LC-Biotin reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions, then diluted to 100 pg / ml in PBS-T and used as the capturing reagent. The standard and the rat plasma samples were diluted 1 : 10 in RexxipA buffer (Gyros Protein Technologies). An anti-His antibody conjugate (Anti-His Tag Alexa Fluor® 647 antibody, Bio-techne) was diluted to 25nM in RexxipF and used for detection. The quantification was performed on a Gyrolab Bioaffy 1000 CD appliance (Gyros Protein Technologies).

[0511] Results

[0512] The tested fusion protein dimers showed a wide range of PK properties after IV administration, depending on the individual design of the molecule:

[0513] The unmodified VHH-CK dimer H5 showed the shortest half-life (3.19 h). The protein dimer’s half-life increased gradually with increasing numbers of EPs added (e.g. H33 with 1 EP shows a T1 / 2 of 10.9 h, H17 with 3 EPs has a T1 / 2 of 19.2 h, and H18 with 6 EPs has a T1 / 2 of 25.15 h). The most pronounced effect was achieved by the addition of EPs in C-terminal positions of the molecules (Table 26, Figs. 27, 28 and 30). The added EPs had a variable impact on the half-life of the different variants. Interestingly, H30 with 9 consecutive EPs at the N-terminus shows a shorter T1 / 2 (22.74 h) than H34 (T1 / 2 of 40.3 h) with the same overall number of EPs, but with the EPs distributed on three termini of the heterodimer (i.e. 3 EPs each, Figs. 27, 28 and 30). In addition, H36 equipped with just one EP on each of the four termini of the dimer (4 EPs in total), had a significantly longer T1 / 2 compared to H30 (32, 1 h vs. 22, 7h; Table 26). The two fusion protein dimers H19 and H31 , containing the single domain antibody ABV2 in different locations, showed different PK properties depending on its position within the molecule (Fig. 27, 28 and 30 and Table 26). This indicates a large impact of the position of the VHH moiety ABV2 on the half-life of the dimer.

[0514] The impact of EPs and ABV2 on the PK of the VHH-CK dimers after SC administration resembled the same pattern as after IV administration (Fig. 31 ). In addition, the fusion of VWF-derived EPs or of the anti-RSA single domain antibody ABV2 resulted in improved bioavailability of the protein dimers after subcutaneous administration. This was visible by increases in the Cmax and AUC of all constructs tested in comparison to H5 (Table 26).

[0515] In summary, fusion protein dimers with a large range of different PK properties were achieved by variation of the number and position of EPs and / or of ABV2, representing a full range of options that may ultimately be used to adjust the half-life and bioavailability of tailored therapeutic molecules.

[0516] Table 26: PK parameters of the VHH fusion protein dimers in CD rats

[0517] Many modifications and other embodiments of the invention set forth herein will come to mind of the one skilled in the art to which the invention pertains, having the benefit of the teachings presented in the foregoing description and the associated drawings. Therefore, it is to be understood that the invention is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims. Although specific terms are employed herein, they are used in a generic and descriptive sense only and not for purposes of limitation.

[0518] REFERENCES

[0519] Chen X, Zaro J, Shen WC. Fusion Protein Linkers: Property, Design and Functionality. Adv Drug Deliv Rev. 2013, 65(10): 1357-1369.

[0520] Daly NL, Craik DJ. Bioactive cystine knot proteins. Current Opinion in Chemical Biology 2011 , 15 (3): 362-368.

[0521] Dumont JA, Liu T, Low SC, Zhang X, Kamphaus G, et al. Prolonged activity of a recombinant factor Vlll-Fc fusion protein in hemophilia A mice and dogs. Blood 2012, 119(13): 3024-3030.

[0522] Engler C, Kandzia R, Marillonnet S. A one pot, one step, precision cloning method with high throughput capability. PLoS ONE 2008, 3(11): e3647

[0523] Ewenstein BM, Collins P, Tarantino MD, Negrier C, Blanchette V, Shapiro AD, Baker D, Spotts G, Sensei M, Yi SE, Gomperts ED. Hemophilia therapy innovation development of an advanced category recombinant factor VIII by a plasma / albumin-free method. Proceedings of a Special Symposium at the XIXth Congress of the International Society on Thrombosis and Haemostasis. 2004, Vol. 41 : 1-16.

[0524] Fischer B, Mitterer A, Schlokat U, Den Bouwmeester R, Dorner F. Structural analysis of recombinant von Willebrand factor: identification of hetero- and homo-dimers. FEES Lett. 1994, 351(3): 345-348. Erratum in: FEES Lett. 1994, 353(3): 337.

[0525] Furlan M. Von Willebrand factor: molecular size and functional activity. Ann Hematol. 1996, 72(6): 341-348.

[0526] Innis M, Gelfand D, Sninsky J, White T. PCR: A Guide to Methods and Application. Academic Press 1990, New York.

[0527] Green MR and Sambrook J. Molecular Cloning: A Laboratory Manual. 4th Ed., Cold Spring Harbor Laboratory Press 2012, Cold Spring Harbor, NY.

[0528] Hamers-Casterman C, Atarhouch T, Muyldermans S, Robinson G, Hamers C, Songa EB, Bendahman N, Hamers R. Naturally occurring antibodies devoid of light chains. Nature 1993, 363(6428): 446-448.

[0529] Huang F and Nau WM. A Conformational Flexibility Scale for Amino Acids in Peptides. Angew Chem Int Ed. 2003, 42: 2269 - 2272.

[0530] Kannicht C, Ramstrom M, Kohla G, Tiemeyer M, Casademunt E, Walter O, Sandberg H. Characterisation of the post-translational modifications of a novel, human cell line-derived recombinant human factor VIII. Thromb Res. 2013, 131 (1): 78-88. Krishnamurthy A and Jimeno A. Bispecific antibodies for cancer therapy: A review. Pharmacol Then 2018, 185: 122-134.

[0531] Needleman SB, Wunsch CD. A general method applicable to the search for similarities in the amino acid sequence of two proteins. J Mol Biol. 1970, 48(3): 443-453.

[0532] Ridgway JB, Presta LG, Carter P. 'Knobs-into-holes' engineering of antibody CHS domains for heavy chain heterodimerization. Protein Eng. 1996, 9(7): 617-621.

[0533] Runcie K, Budman DR, John V, Seetharamu N. Bi-specific and tri-specific antibodies - the next big thing in solid tumor therapeutics. Mol. Med. 2018, 24(1): 50.

[0534] Sherbet, GV. Growth Factor Families, Growth Factors and Their Receptors in Cell Differentiation. Cancer and Cancer Therapy 2011 , 3-5.

[0535] Strohl WR. Fusion Proteins for Half-Life Extension of Biologies as a Strategy to Make Biobetters. BioDrugs. 2015, 29(4): 215-239.

[0536] Tiede A. Half-life extended factor VIII for the treatment of hemophilia A. J Thromb Haemost.

[0537] 2015, Vol. 13 Suppl 1 : S176-179. van Faassen H, Ryan S, Henry KA, Raphael S, Yang Q, et al. Serum albumin-binding VHHs with variable pH sensitivities enable tailored half-life extension of biologies. FASEB J. 2020, 34(6):8155-8171.

[0538] Vlot AJ, Koppelman SJ, Meijers JC, Dama C, van den Berg HM, Bouma BN, Sixma JJ, Willems GM. Kinetics of factor Vlll-von Willebrand factor association. Blood. 1996, 87(5): 1809-1816.

[0539] Vitt, UA, Hsu, Y, Hsueh, AJW. Evolution and Classification of Cystine Knot-Containing Hormones and Related Extracellular Signaling Molecules. Molecular Endocrinology. 2001 , 15 (5): 681-694.

[0540] Vollack-Hesse N, Oleshko O, Werwitzke S, Solecka-Witulska B, Kannicht C, Tiede A. Recombinant VWF fragments improve bioavailability of subcutaneous factor VIII in hemophilia A mice. Blood. 2021 , 137(8): 1072-1081.

[0541] Yee A, Gildersleeve RD, Gu S, Kretz CA, McGee BM, Carr KM, Pipe SW, Ginsburg D. A von Willebrand factor fragment containing the D'D3 domains is sufficient to stabilize coagulation factor VIII in mice. Blood. 2014, 124(3): 445-452. SEQUENCE OVERVIEW

Claims

C l a i m s1 . A protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a FVIII protein, preferably comprising a first linker, fused to a von Willebrand Factor cystine knot domain (CKVWF) by a second linker and the second fusion protein comprises a von Willebrand Factor (VWF) fragment fused to a CKVWF domain by a third linker, wherein the third linker is an engineered peptide, and wherein the two fusion proteins are covalently linked by their CKVWF domains.

2. The protein dimer according to claim 1 , wherein the CKVWF domains have an amino acid sequence with an identity of at least 90 %, more preferably at least 95 %, most preferably at least 98 % to SEQ ID NO: 1 .

3. The protein dimer according to claim 1 or 2, wherein the first fusion protein and / or the second fusion protein comprise(s) at least one copy of an extension peptide (EP), wherein the EP has an amino acid sequence with an identity of at least 90 % to SEQ ID NO: 2, and contains a cluster of O-glycosylation sites, wherein the cluster contains at least two O-glycosylated amino acids.

4. The protein dimer according to claims 1 to 3, wherein the first fusion protein and / or the second fusion protein comprise(s) at least one half-life prolonging moiety, preferably selected from an immunoglobulin Fc-domain, serum albumin or parts thereof, an albumin binding moiety, the most preferred half-life prolonging moiety being an albumin binding VHH domain.

5. The protein dimer according to claim 4, wherein the first pharmaceutically active protein is a FVIII protein, wherein the FVIII protein preferably comprises a FVIII heavy chain, fused by a first linker to a FVIII light chain, wherein more preferably the FVIII heavy chain lacks the B-domain, and wherein most preferably the first linker preferably comprises a sequence derived from the B-domain of FVIII.

6. The protein dimer according to claim 5, wherein the FVIII heavy chain comprises an amino acid sequence with an identity of at least 90 %, more preferably at least 95 %, most preferably at least 98 % to SEQ ID NO: 3 and / or the FVIII lightchain comprises an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 % to SEQ ID NO: 5.

7. The protein dimer according to any one of claims 4 to 6, wherein the second pharmaceutically active protein is a fragment of VWF, wherein the VWF fragment preferably has an amino acid sequence with an identity of at least 90 %, more preferably at least 95 %, most preferably at least 98 % to SEQ ID NO: 7.

8. The protein dimer according to any one of the preceding claims, wherein the first and the second linker are engineered peptides and / or wherein the third linker does not contain the C3 domain of VWF.

9. The protein dimer according to any one of the preceding claims, wherein the first linker, the second linker and / or the third linker comprise(s) at least one copy, preferably at least two copies, more preferably at least three copies of the EP, wherein the EPs are preferably assembled in a consecutive order.

10. The protein dimer according to any one of the preceding claims, wherein the half-life prolonging moiety a) is fused to the C-terminus of the first fusion protein and / or second fusion protein by a fourth linker, or b) forms part of the third linker.11 . The protein dimer according to any one of the preceding claims, wherein the first linker, the second linker and / or the third linker comprises a flexibility motif, in particular selected from (GGS)n, (GGGS)n, or (GGGGS)n, wherein n is an integer in the range of 1 and 10 and wherein G represents glycine and 8 represents serine.

12. The protein dimer according to any one of claims any one of the preceding claims, wherein the amino acid sequence of the first linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 8, SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 11 , SEQ ID NO: 12, SEQ ID NO: 43, and SEQ ID NO: 44.

13. The protein dimer according to any of claims any one of the preceding claims, wherein the amino acid sequence of the second linker has an identity of at least 90%,at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 13, SEQ ID NO: 14, SEQ ID NO: 15, SEQ ID NO: 16, and SEQ ID NO: 45.

14. The protein dimer according to any one of the preceding claims, wherein the amino acid sequence of the third linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 17, SEQ ID NO: 18, and SEQ ID NO: 55.

15. The protein dimer according to any one of the preceding claims, wherein the first fusion protein is an FVIII fusion protein comprising an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 %, most preferably 100 % to a sequence selected from SEQ ID NO: 22, SEQ ID NO: 23, SEQ ID NO: 24, SEQ ID NO: 25, SEQ ID NO: 26, SEQ ID NO: 27, SEQ ID NO: 28, SEQ ID NO: 29, SEQ ID NO: 30, SEQ ID NO: 31 , SEQ ID NO: 47, SEQ ID NO: 48, SEQ ID NO: 51 , SEQ ID NO: 52, SEQ ID NO: 53, and SEQ ID NO: 54.

16. The protein dimer according to any one of the preceding claims, wherein the second fusion protein is a VWF fusion protein comprising an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 %, most preferably 100 % to a sequence selected from SEQ ID NO: 32, SEQ ID NO: 33, SEQ ID NO: 34, SEQ ID NO: 35, SEQ ID NO: 49, and SEQ ID NO: 50.

17. A protein dimer formed by a first and a second fusion protein, wherein the first fusion protein comprises a cystine knot domain (CK) fused to a first pharmaceutically active protein and the second fusion protein comprises a CK domain fused to a second pharmaceutically active protein, wherein the two fusion proteins are covalently linked by their CK domains, provided that the first pharmaceutically active protein and the second pharmaceutically active protein are not selected from the group consisting of VWF, FVIII and fragments thereof.

18. The protein dimer according to claim 17, wherein the CK domains are selected from CKVWF domains and CKNDP domains, preferably the CK domains are CKVWF domains.

19. The protein dimer according to claim 17 or 18, wherein the CKVWF domains have an amino acid sequence with an identity of at least 90 %, more preferably at least 95 %, most preferably at least 98 % to SEQ ID NO: 1 .

20. The protein dimer according to any one of claims 17 to 19, wherein the first fusion protein and / or the second fusion protein comprise(s) at least one copy of an extension peptide (EP), wherein the EP has an amino acid sequence with an identity of at least 90 % to SEQ ID NO: 2, and contains a cluster of O-glycosylation sites, wherein the cluster contains at least two O-glycosylated amino acids.

21. The protein dimer according to any one of claims 17 to 20, wherein the first fusion protein and / or the second fusion protein comprise(s) at least one half-life prolonging moiety, preferably selected from an immunoglobulin Fc-domain, serum albumin or parts thereof, an albumin binding moiety, the most preferred half-life prolonging moiety being an albumin binding VHH domain.

22. The protein dimer according to any one of claims 17 to 21 , wherein the first pharmaceutically active protein and the second pharmaceutically active protein are identical or different and are preferably selected from the group consisting of, prothrombin, fibrinogen, Fill, FV, FVII, FIX, FX, FXI, FXII, FXIII, ADAMTS13, antithrombin, alpha-1 antitrypsin, C1 -inhibitor, anti-chymotrypsin, PAI-1 , PAI-3, a2- macroglobulin, TFPI, heparin cofactor II, protein C, protein S, protein Z, fragments of the aforementioned proteins and a VHH domain, wherein the VHH domain preferably binds to FIX, FIXa, FX, activated protein C, protease nexin-1 , protein Z dependent protease, antithrombin, protein S, ADAMTS13, platelet Gpllb / llla receptor, complement C5, complement C3, P -selectin, human serum albumin or FcRn.

23. The protein dimer according to any of claims 17 to 22, wherein the first pharmaceutically active protein is a VHH domain, in particular binding to FIX, wherein the VHH domain binding to FIX preferably comprises an amino acid sequence with an identity of at least 90 %, more preferably at least 95 %, most preferably at least 98 % to SEQ ID NO: 72.

24. The protein dimer according to claim 23, wherein the second pharmaceutically active protein is a VHH domain, in particular binding to FX, wherein the VHH domainbinding to FX preferably comprises an amino acid sequence with an identity of at least 90 %, more preferably at least 95 %, most preferably at least 98 % to SEQ ID NO: 73.

25. The protein dimer according to any of claims 17 to 24, wherein the amino acid sequence of the first linker and / or the second linker has an identity of at least 90%, at least 95 %, or at least 98 % to a sequence selected from SEQ ID NO: 56, SEQ ID NO: 57, SEQ ID NO: 58, and SEQ ID NO: 59.

26. The protein dimer according to any one of claims 17 to 25, wherein the first fusion protein is a VHH fusion protein comprising an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 %, most preferably 100 % to a sequence selected from SEQ ID NO: 60, SEQ ID NO: 62, SEQ ID NO: 64, SEQ ID NO: 68, SEQ ID NO: 82, SEQ ID NO: 83, SEQ ID NO: 84, SEQ ID NO: 86, SEQ ID NO: 88, SEQ ID NO: 90, SEQ ID NO: 93, SEQ ID NO: 95, and SEQ ID NO: 98.

27. The protein dimer according to claim 26, wherein the second fusion protein is a VHH fusion protein comprising an amino acid sequence with an identity of at least 90 %, preferably at least 95 %, more preferably at least 98 %, most preferably 100 % to a sequence selected from SEQ ID NO: 61 , SEQ ID NO: 63, SEQ ID NO: 65, SEQ ID NO: 69, SEQ ID NO: 85, SEQ ID NO: 87, SEQ ID NO: 89, SEQ ID NO: 91 , SEQ ID NO: 92, SEQ ID NO: 94, SEQ ID NO: 96, SEQ ID NO: 97, and SEQ ID NO: 99

Images