Anti-VWF D'D3 single-domain antibodies fuse to clotting factors

Abstract

The invention relates to isolated single-domain antibodies (sdAb) directed against von Willebrand Factor (VWF) D′D3 domain and chimeric polypeptides comprising thereof such as blood clotting factors and their uses in therapy such as in the prevention and treatment of hemostatic disorders. The invention also relates to a method of extending or increasing half-life of a therapeutic polypeptide comprising a step of adding to the polypeptide sequence of said therapeutic polypeptide at least one sdAb directed against VWF D′D3 domain.

Claims

1. A chimeric polypeptide comprising a polypeptide and at least one sdAb directed against von Willebrand Factor (VFW) D′D3 domain, wherein the polypeptide encodes a clotting factor, and wherein said sdAb comprises: i) a CDR1 having a sequence set forth as SEQ ID NO: 1, a CDR2 having a sequence set forth as SEQ ID NO: 2 and a CDR3 having a sequence set forth as SEQ ID NO: 3; ii) a CDR1 having a sequence set forth as SEQ ID NO: 5, a CDR2 having a sequence set forth as SEQ ID NO: 6 and a CDR3 having a sequence set forth as SEQ ID NO: 7; and/or iii) a CDR1 having a sequence set forth as SEQ ID NO: 9, a CDR2 having a sequence set forth as SEQ ID NO: 10 and a CDR3 having a sequence set forth as SEQ ID NO: 11.

2. An isolated single-domain antibody (sdAb) directed against von Willebrand Factor (VWF) D′D3 domain, wherein said sdAb comprises a CDR1 having a sequence set forth as SEQ ID NO: 1, a CDR2 having a sequence set forth as SEQ ID NO: 2 and a CDR3 having a sequence set forth as SEQ ID NO: 3; wherein the sdAb is fused with a clotting factor.

3. The chimeric polypeptide according to claim 1, wherein said chimeric polypeptide has an increased affinity and/or a reduced dissociation rate constant for VWF compared to a wild-type polypeptide encoding the clotting factor.

4. The chimeric polypeptide according to claim 1, wherein the clotting factor selected from the group consisting of FVII, FVIII, protein C and protein S.

5. The chimeric polypeptide according to claim 1, wherein said at least one single-domain antibody is fused at the N terminal end, at the C terminal end, both at the N terminal end and at the C terminal end of the chimeric polypeptide or is inserted within the sequence of the chimeric polypeptide.

6. The chimeric polypeptide according to claim 1, comprising two, three, four, or five sdAb directed against VWF.

7. The chimeric polypeptide according to claim 1, wherein the chimeric polypeptide comprises two sdAb directed against VWF which: i) replace the C-terminal part of the B domain of factor VIII (FVIII-KB13-bv, KB-VWF-008 and/or KB-VWF-011); ii) are fused to the C-terminus of FVIII; iii) simultaneously replace the C-terminal part of the B domain of factor VIII and are fused to C-terminus of factor VIII; or iv) are inserted at the C-terminus of factor VII.

8. The chimeric polypeptide according to claim 1, wherein the polypeptide comprises at least one single-domain antibody directed against a first antigen and at least one further binding site directed against a second antigen.

9. A chimeric polypeptide/VWF complex comprising the chimeric polypeptide of claim 1 and a VWF polypeptide, wherein the VWF polypeptide has an extended half-life within the chimeric polypeptide/VWF complex.

10. A pharmaceutical composition comprising a chimeric polypeptide according to claim 1 or a chimeric polypeptide/VWF according to claim 9, complex comprising the chimeric polypeptide, and a pharmaceutically acceptable carrier.

Description

FIGURES

(1) FIG. 1: Real-time analysis of association and dissociation of VWF interactions with FVIII and sdAbs. Association and dissociation curves for the binding of VWF to immobilized sdAbs and the binding of FVIII to immobilized VWF are plotted in FIG. 1. For the analysis, we focused on the dissociation phase. Apparent dissociation constants were 2.0±1.1×10−5 s−1 (KB-VWF-008), 0.6±0.5×10−5 s−1 (KB-VWF-011), 1.3-3.5×10−5 s−1 (KB-VWF013) and 2.2-3.0×10−3 s−1 (FVIII).

(2) FIG. 2: Effect of sdAbs on VWF binding to Factor VIII. Binding of FVIII to immobilized VWF was determined in the absence or presence of sdAbs or Mab418. Plotted is the percentage FVIII binding relative to FVIII binding in the absence of antibodies. FVIII binding is unaffected by the presence of KB-VWF-008, -011 or -013.

(3) FIG. 3: Factor VIII-sdAb fusion protein binds to VWF. The ability to form a complex with VWF was tested via transient expression of WT-FVIII-SQ, FVIII-SQ/p.Y1680F or FVIII-KB013bv/p.Y1680F in hemophilic mice. Four days after gene transfer, VWF/FVIII complexes were determined, which are expressed as the percentage of complex relative to WT-FVIII-SQ. As expected, the presence of the p.Y1680F mutation abrogated binding of FVIII to VWF (FVIII-SQ/p.Y1680F). In contrast, the introduction of KB-VWF-013 restored and even improved binding to VWF despite the presence of the p.Y1680F mutation.

(4) FIG. 4: Expression and functional analysis of FVIII-KB013bv. Purified FVIII-KB013 and WT-FVIII-SQ were incubated in the absence or presence of thrombin. Western blot analysis was performed to determine the presence of FVIII fragments. FVIII-KB013bv migrates predominantly as a single-chain protein when incubated in the absence of thrombin (lane 1), whereas WT-FVIII-SQ predominantly migrates as a heterodimeric protein (lane 3). After thrombin incubation, both FVIII-KB013bv and WT-FVIII-SQ are present as a heterodimeric protein, consisting of the thrombin-cleaved light chain and the heavy-chain derived A1 and A2 domains (lanes 2 & 4).

(5) FIG. 5: in vivo survival of FVIII-KB-013bv. FVIII-KB013bv or WT-FVIII-SQ were given intravenously to FVIII-deficient mice. At indicate time-points, blood was collected and FVIII activity was determined. Residual activity relative to activity at 3 min after injection is plotted against time after injection. FVIII-KB013bv is removed from the circulation slower than is WT-FVIII-SQ.

(6) FIG. 6: Correction of hemostasis in hemophilic mice 24 h after injection of FVIII-KB013bv. FVIII-KB013bv or B-domainless FVIII (Xyntha) were given intravenously to FVIII-deficient mice and 24 h after injection the terminal tip of the tail was amputated in anesthetized mice. Blood loss was monitored for 30 min. The volume of shed blood was determined and is presented for each mouse. Mice treated with FVIII-KB013bv lost significantly less blood compared to mice treated with wild-type B-domainless FVIII.

(7) FIG. 7: Fusion of KB-VWF-013 to coagulation factor VII induces complex formation with VWF. The ability to form a complex with VWF was tested via transient expression of wild-type FVII and FVII-KB013-bv in wild-type C57B16 mice. Four days after gene transfer, VWF/FVIII complexes were determined, which are expressed as OD450 nm. As expected, no complex formation with VWF could be detected for wild-type FVII. In contrast, VWF-FVII complexes were detected in all mice expressing FVII-KB013-bv. Thus, the fusion of FVII to KB-VWF-013 induces the capacity of FVII to bind to VWF.

EXAMPLES

Example A: Protein Domain Structure of VWF

(8) Bio-informatic analysis of the cDNA and protein sequences of VWF has revealed that the protein architecture distinguishes different types of domain structures. Originally, this domain structure consisted of a signal peptide (SP), A-domains, B-domains, C-domains, D-domains and a CK-domain arranged in the order: SP-D1-D2-D′-D3-A1-A2-A3-D4-B1-B2-B3-C1-C2-CK (Verweij C L et al. (1986) EMBO Journal, vol. 5, pp. 1839-1847). More recently an updated domain organization has been proposed, in which the domains are arranged in the following order: SP-D1-D2-D′-D3-A1-A2-A3-D4-C1-C2-C3-C4-C5-C6-CK (Zhou Y F et al. (2012) Blood, vol 120, pp. 449-458). Since the boundaries of the different domains may be varying from one publication to another, we use in this application the boundaries as defined in FIG. 1 of Lenting P J et al. (2015) Blood, vol 125, pp. 2019-2028).

Example B: Binding of sdAb to VWF or Fragments Thereof

(9) sdAbs KB-VWF-008, -011 and -013 were immobilized (5 μg/ml) in 10 mM NaHCO3, 50 mM Na2CO3 (pH 9.5) in a volume of 50 μl in half-well microtiter plates (Greiner Bio-One, Les Ulis, France) for 16 h at 4° C. As a positive control, polyclonal rabbit anti-VWF antibodies (Dako, Glostrup, Danmark) were immobilized in a similar fashion. As a negative control, no antibodies were immobilized. After washing the wells three times with 75 μl/well using Tris-buffered saline (pH 7.6) supplemented with 0.1% Tween-20 (TBS-T), wells were blocked with 75 μl/well of TBS-T supplemented with 3% bovine serum albumin (BSA) for 30 min at 37° C. Wells were washed as described above, and subsequently the following VWF preparations (diluted in Tris-buffered saline (pH 7.6) supplemented with 3% BSA, all at 2 μg/ml, 50 μl per well, 2 hours at 37° C.) were added to each of the immobilized sdAbs and both types of control wells: purified recombinant human VWF (rhVWF), purified recombinant murine VWF (rmVWF), VWF fragment SpII (a proteolytic fragment of plasma-derived (pd)-VWF following incubation with S. aureus V8-protease, which encompasses residues 2129-2813 of VWF; Denis C et al. (1993) Arteriosclerosis Thrombosis, vol 13, pp. 398-406), VWF fragment SpIII (a proteolytic fragment of pd-VWF following incubation with S. aureus V8-protease, which encompasses residues 764-2128 of VWF; Kalafatis M et al. (1987) Blood, vol 70, pp. 1577-1583), D′D3-HPC4 fragment (human VWF residues 764-1247 fused to the amino acid sequence EDQVDPRLIDGK (SEQ ID NO: 15), representing a recognition site for antibody HPC4), A1-A2-A3-HPC4 fragment (human VWF residues 1260-1874 fused to the amino acid sequence EDQVDPRLIDGK), hD1-D2-HPC4 fragment (human VWF residues 23-762 fused to the amino acid sequence EDQVDPRLIDGK), mD1-D2-HPC4 fragment (murine VWF residues 23-762 fused to the amino acid sequence EDQVDPRLIDGK)

(10) Wells were then washed three times with 75 μl/well using TBS-T. Bound VWF preparations were probed with peroxidase-labeled polyclonal rabbit anti-VWF antibodies (Dako, Glostrup, Danmark; diluted 1/6000) for rhVWF, rmVWF, SpII and SPIII or with peroxidase-labeled monoclonal antibody HPC4 (diluted 1/1000) for D′D3-HPC4, A1A2A3-HPC4, hD1D2-HPC4 and mD1-D2-HPC4 for 2 hours at 37° C. with 50 μl per well. Wells were then washed three times with 75 μl/well using TBS-T. Residual peroxidase activity was detected by measuring peroxidase-mediated hydrolysis of 3,3′,5,5′-tetramethylbenzidine.

(11) Negative binding (−) was defined as optical density (OD) being ≤0.5, moderate positive binding (+) was defined as OD being >0.5 and <1.0, strongly positive binding (++) was defined as OD being ≥1.0. Based on these definitions, none of the VWF preparations displayed moderate or strongly positive binding to the negative control (Table 1). All VWF preparations with the exception of mD1-D2-HPC4 had moderate or strongly positive binding to the positive control (polyclonal anti-VWF antibodies). None of the sdAbs bound to SpII, A1A2A3-HPC4, hD1-D2-HPC4 or mD1-D2-HPC4. In contrast, KB-VWF-008, -011 and -013 had moderate or strongly positive binding to rhVWF, spill and D′D3-HPC4, suggesting that the epitope of these three sdAbs is located within VWF residues 764-1247. Furthermore, sdAb KB-VWF-013 was the only one of the three tested sdAbs that reacted positively with rmVWF, showing that this sdAb cross-reacts with murine VWF.

(12) TABLE-US-00014 TABLE 1 belonging to example B: Binding of sdAb to VWF and fragments thereof D’D3- A1A2A3- hD1D2- mD1D2 sdAb rhVWF rmVWF SpII SpIII HPC4 HPC4 HPC4 HPC4 008 + − − + ++ − − − 011 ++ − − + ++ − − − 013 ++ + − ++ ++ − − − Control ++ + ++ ++ ++ ++ + − rhVWF; recombinant humanVWF rmVWF; recombinant murine VWF; spII: a proteolytic fragment of plasma-derived (pd)-VWF following incubation with S. aureus V8-protease, which encompasses residues 2129-2813 of VWF; spIII: a proteolytic fragment of pd-VWF following incubation with S. aureus V-8 protese, which encompasses residues 764-2128 of VWF; D’D3-HPC4: human VWF residues 764-1247 fused amino acid sequence EDQVDPRLIDGK; A1-A2-A3-HPC4: human VWF residues 1260-1874 fused to the amino acid sequence EDQVDPRLIDGK; hD1-D2-HPC4: human VWF residues 23-762 fused to the amino acid sequence EDQVDPRLIDGK; mD1-D2-HPC4: murine VWF residues 23-762 fused to amine acid sequence EDQVDPRLIDGK; control; polyclonal rabbit-antihuman VWF antibodies (Dako). −Negative binding defined as OD being ≤ 0.5; +Moderate positive binding defined as OD being >0.5-<1.0; ++Strongly positive binding defined as being ≥1.0

Example C: Real-Time Analysis of Association and Dissociation of VWF Interactions with FVIII and sdAbs

(13) The interaction between VWF and sdAbs was analyzed via bio-layer interferometry using Octet-QK equipment (Fortébio, Meldo Park, Calif., USA). To this end, sdAbs KB-VWF-008, -011 and -013 were diluted in 0.1 M Mes (pH 5.0) to a concentration of 10 μg/ml for coupling to EDC/NHS-activated amine-reactive biosensors (Fortébio, Menlo Park, Calif., USA). Sensors were rehydrated in 0.2 ml 0.1 M MES, pH 5.0 for 300 sec. Sensors were then activated via incubation with 0.1 ml 0.2 M EDC/0.095 M NHS mixture for 300 sec and subsequently incubated with 0.1 ml sdAb-solution for 600 sec. Unoccupied amine-reactive sites were quenched by incubating with 1M ethanolamine for 180 sec, and sensors were allowed to reach stable baseline levels via incubation with phosphate-buffered saline supplemented with 0.1% Tween-20 (PBS-T) for 300 sec. sdAb-coated sensors were then transferred to wells containing various concentrations of purified plasma-derived VWF (2.5, 25 & 250 μg/ml in PBS-T for KB-VWF-008 and -011 versus 25 & 250 μg/ml for KB-VWF-013) and incubated for 600 sec in order to visualize association of VWF to immobilized sdAbs. Following this association phase, sensors were transferred to wells containing PBS-T and incubated for 900 sec, allowing dissociation of the VWF-sdAb complex.

(14) In another set of experiments, we determined the association and dissociation of factor VIII to immobilized recombinant human VWF via biolayer-interferometry analysis, also using Octet-QK equipment. Amine-reactive biosensors were used to immobilize recombinant VWF (50 μg/ml in 0.1 M MES, pH 5.0). After hydration of the sensors via a 600-sec incubation with 0.1 M MES pH 5.0, sensors were activated with 0.1 ml 0.2 M EDC/0.095 M NHS mixture for 420 sec and subsequently incubated with 0.1 ml VWF-solution for 420 sec. Unoccupied amine-reactive sites were quenched by incubating with 1M ethanolamine for 420 sec, and sensors were allowed to reach stable baseline levels via incubation with Hepes-buffer (20 mM Hepes, 0.11 M NaCl, 0.005% Tween-20, 5 mM CaCl2, pH 7.3) for 600 sec. VWF-coated sensors were then transferred to wells containing various concentrations of purified recombinant full-length factor VIII (Kogenate; diluted to 3.5 nM or 1.4 nM in Hepes-buffer) and incubated for 600 sec in order to visualize association of FVIII to immobilized VWF. Following this association phase, sensors were transferred to wells containing Hepes-buffer and incubated for 600 sec, allowing dissociation of the VWF-FVIII complex.

(15) Association and dissociation curves are plotted in FIG. 1. When analyzing the data for the interaction between sdAbs and VWF versus the interaction between VWF and FVIII, we focused on the dissociation phase for both types of interaction. The dissociation rate constant for the VWF-FVIII interaction was calculated using an equation for a single exponential decay, and the dissociation rate constants were calculated to be 2.2×10−3 s−1 and 3.0×10−3 s−1 for FVIII concentrations of 3.5 nM and 1.4 nM, respectively. These values are similar to those described in the literature (0.3-6.0×10−3 s−1; Sandberg et al (2012) Thromb Res vol 130, pp 808-817; Dimitrov et al (2012) Biochemistry vol 51, pp 4108-4116; Zollner et al (2014) Thromb Res vol 134, pp 125-131). The dissociation constants for the sdAbs were could not be calculated accurately using an equation for a single exponential decay, as the dissociation was too slow during the period that was monitored. We used therefore a linear regression approach to determine the slope of the dissociation curve, which represents an apparent dissociation rate constant that probably over-estimates the true dissociation rate constant (i.e. in reality dissociation is slower than represented by the apparent dissociation rate constant). For KB-VWF-008, the apparent dissociation rate constant was 2.0±1.1×10−5 s−1 (mean±standard deviation; n=3 concentrations). For KB-VWF-011, the apparent dissociation rate constant was 0.6±0.5×10−5 s−1 (mean±standard deviation; n=3 concentrations). For KB-VWF-013, the apparent dissociation rate constants was 1.3×10−5 s−1 and 3.5×10−5 s−1 (for 250 □g/ml and 25 □g/ml, respectively). Thus, for each of the three sdAbs, the apparent dissociation rates constants for the interaction with VWF are at least 15-300-fold slower compared to those dissociation rates constants reported in the literature for the FVIII-VWF interaction, and at least 100-fold slower compared to the dissociation rate constant calculated for the VWF-FVIII interaction analyzed in the same Octet-QK equipment.

Example D: Effect of sdAbs on VWF Binding to Factor VIII

(16) Polyclonal rabbit anti-VWF antibodies (Dako, Glostrup, Danmark) were immobilized onto microtiter wells at 5 μg/ml in 50 mM Na2CO3 (pH 9.5) overnight at 4° C. in a volume of 50 μl. After washing thrice with Tris-buffered saline supplemented with 0.1% Tween-20 (TBS-T), wells were saturated with 3% BSA in TBS-T. Then rVWF (0.03-1.0 μg/ml; 50 μl/well) was added to the wells and incubated overnight at 4° C. After washing in TBS-T, wells were incubated twice with 75 μl of 0.35 M CaCl2 for 10 min at 37° C., followed by 6 washes with TBTS-T (75 μl/well). Then rFVIII (Kogenate-FS, Bayer Healthcare) diluted to a concentration of 1.5 U/ml was added in the presence or absence of 20 μg/ml of sdAb KB-VWF-008, -11 or -013 in a total volume of 50 μl. As a control, FVIII was added in the presence of the murine monoclonal anti-VWF antibody Mab418, which blocks binding of FVIII to VWF (Takahashi Y et al. (1987) Blood vol 70, pp 1679-1682). After 2 h at 37° C. and 3 washes with TBS-T (75 μl/well), bound FVIII was probed using peroxidase-labeled polyclonal sheep-anti-FVIII antibodies (Stago BNL, Leiden, the Netherlands) and detected by measuring peroxidase-mediated hydrolysis of 3,3′,5,5′-tetramethylbenzidine. For each VWF concentration, FVIII binding in the presence of sdAb or Mab418 was calculated relative to FVIII binding in the absence of sdAb or Mab418, and expressed in percentage binding (FIG. 2). Whereas the presence of Mab418 reduced binding of FVIII to VWF by 72±5% (mean±standard deviation; n=6; p<0.001 compared to control), the presence of each of the sdAbs left FVIII binding similar to that in the absence of any antibody (p>0.05 when tested using one-way ANOVA with multiple comparisons). This shows that sdAbs KB-VWF-008, -011 and -013 do not interfere with the binding of FVIII to VWF.

Example E: Factor VIII-sdAb Fusion Protein Binds to VWF

(17) cDNA constructs encoding wild-type B-domainless FVIII (WT-FVIII-SQ), B-domainless FVIII containing a Tyr to Phe replacement at position 1680 (FVIII-SQ/p.Y1680F) and FVIII-KB013bv containing a Tyr to Phe replacement at position 1680 (FVIII-KB013bv/p.Y1680F) were cloned into the pLIVE-plasmid (Mirus Bio, Madison, Wis., USA). Tyrosine at position 1680 is sulfated in WT-FVIII-SQ, a requirement for the binding to von Willebrand factor (VWF) and mutation of p.Tyr1680 to Phe is associated with a loss of VWF binding (Leyte A et al. (1991) J Biol Chem vol 266, pp 740-746). Plasmids (100 □g/mouse) were injected into factor VIII-deficient mice via hydrodynamic gene transfer: plasmids are diluted in 0.9% saline with the volume corresponding to 10% of the animal's bodyweight (i.e. 2 ml for a 20-gram mouse). The solution is injected in the tail vein within 5 seconds. Four days after gene transfer, blood was collected via retro-orbital puncture from isoflurane-anesthetized mice and plasma was prepared by centrifugation (1500 g for 20 min at 22° C.). Plasma was then used to measure VWF-FVIII complexes that were formed in the plasma of the mice. Complexes were determined as follows: microtiter wells were coated with polyclonal rabbit anti-VWF antibodies (5 μg/ml) as described in example D. After washing thrice with Tris-buffered saline supplemented with 0.1% Tween-20 (TBS-T), wells were saturated with 3% BSA in TBS-T. Then murine plasma samples (diluted 10-fold in TBS-T) were added to the wells and incubated 2 hours at 37° C. After 3 washes with TBS-T (75 μl/well), bound FVIII was probed using peroxidase-labeled polyclonal sheep-anti-FVIII antibodies (Stago BNL, Leiden, the Netherlands) and detected by measuring peroxidase-mediated hydrolysis of 3,3′,5,5′-tetramethylbenzidine. The amount of VWF-complex for mutants FVIII-SQ/p.Y1680F and FVIII-KB013bv/p.Y1680F was related to that of WT-FVIII-SQ, which was arbitrarily set as 100%. As anticipated, complex formation with VWF was strongly reduced for mutant FVIII-SQ/p.Y1680F (8% compared to 100% for WT-FVIII-SQ; see FIG. 3). In contrast, binding was increased 2.4 fold (238%) for variant FVIII-KB013bv/p.Y1680F, which contains the VWF-binding sdAbs. Since the p.Y1680F mutation abrogates natural VWF binding, these data show that while incorporated in the factor VIII protein, sdAb KB-VWF-013 is able to rescue binding to VWF. Thus, in the context of the fusion protein, sdAb KB-VWF-013 contributes to VWF binding.

Example F: Expression and Functional Analysis of FVIII-KB013bv

(18) Baby Hamster Kidney (BHK)-cells were transfected with cDNA encoding FVIII-KB013bv cloned in pcDNA3.1/Hygro and stable cell lines were obtained via selection with hygromycin. One clone was selected for the production of FVIII-KB013bv. FVIII-KB013bv was purified from the culture medium via affinity chromatography using VIIISelect-matrix as instructed by the manufacturer (GE Healthcare, Vélizy-Villacoublay, France). Purified FVIII-KB013bv was tested for activity and antigen. Five top-fractions were selected and chromogenic two-stage activity (Biophen FVIII:C; Hyphen Biomed, Neuville-sur-Oise, France) and factor VIII antigen levels (Girma J P et al (1998) Haemophilia vol 4 pp 98-103) were determined. Average activity was found to be 188±42 U/ml (mean±SD; n=5 consecutive elution fractions) and antigen was calculated to be 176±28 U/ml. Average activity/antigen ratio was 1.1±0.3, showing that FVIII-KB013bv displays full activity in the chromogenic two-stage activity assay.

(19) In a second analysis, FVIII-KB013bv and WT-FVIII-SQ were incubated with in the absence or presence of thrombin (10 nM) for 30 min at room temperature. Subsequently samples were analyzed via Western blotting using polyclonal sheep anti-FVIII antibodies. For samples incubated in the absence of thrombin, WT-FVIII-SQ is predominantly present in a cleaved form, consisting of a 90-kDa heavy chain and an 80-kDa light chain while some uncleaved material was also present (Lane 3 in FIG. 4). In contrast, for FVIII-KB013bv >90% of the preparation was present as a single-chain protein, appearing as a doublet (Lane 1 in FIG. 4). Of note, the size of the uncleaved FVIII-013bv is slightly larger than that of WT-FVIII-SQ, due to the insertion of two copies of sdAb KB-VWF-013 between the FVIII heavy and light chain (Lanes 1 & 3 in FIG. 4). In contrast, following incubation with thrombin, WT-FVIII-SQ and FVIII-013bv displayed a similar pattern for thrombin-activated FVIII, with a 70-kDa light chain and the separate A1 and A2 domains (Lanes 2 & 4 in FIG. 4). This analysis indicates that following thrombin activation, the inserted sdAb KB-VWF-013bv is removed from the protein, giving rise to the natural heterotrimeric FVIIIa protein.

Example G: In Vivo Survival of FVIII-KB-013bv

(20) Purified WT-FVIII-SQ or FVIII-kb013bv (both produced in BHK-M cells and purified using VIIISelect-affinity chromatography) were given intravenously (250-500 U/kg) to FVIII-deficient mice. At different time-points after injection (3 min, 30 min, 1 h, 2 h, 8 h and 24 h for WT-FVIII-SQ and 3 min, 4 h, 9 h, 21 h, 29 h and 48 h for FVIII-KB013bv) blood samples were obtained via retro-orbital puncture from isoflurane-anesthetized mice and plasma was prepared by centrifugation (1500 g for 20 min at 22° C.). Residual FVIII activity was measured using a chromogenic two-stage assay as instructed by the manufacturer (Biophen FVIII:C; Hyphen Biomed, Neuville-sur-Oise, France). Residual FVIII activity relative to activity at 3 min after injection was plotted against the time after injection (FIG. 5). This approach revealed that activity for FVIII-KB013bv remained higher WT-FVIII-SQ at later time-points. For instance, relative residual FVIII activity for WT-FVIII at 24 h was 0.72±0.23% (n=3), whereas for FVIII-KB013bv the relative residual activity at 29 h was more than 3-fold higher (2.62±0.25%; n=3; p=0.0007 in student t-test). When data where analyzed using an equation describing a single exponential decay (Graph Prism 5 for Mac OSX, GraphPad Software, La Jolla, Calif., USA), the half-life calculated for WT-FVIII-SQ was 1.1 h (95% confidence interval 0.9-1.5 h). For FVIII-KB013bv the half-life was calculated to be 2.1 h (95% confidence interval 1.7-2.9 h; p=0.0032 compared to WT-FVIII-SQ), 2-fold longer than the half-life for WT-FVIII-SQ. These results show that the presence of two copies of sdAb KB-VWF-013 has a significant beneficial effect on the survival of FVIII.

Example H: Correction of Hemostasis in Hemophilic Mice 24 h after Injection of FVIII-KB013bv

(21) 8-12 week old hemophilic mice were given WT-FVIII-SQ (Xyntha) or FVIII-KB013bv at a dose of 500 U/kg via intravenous tail injection. Twenty-four hours after injection, the terminal 3 mm of the tail-tip was amputated from ketamine/xylazine-anesthetized mice. The amputated tail was immersed immediately after transaction in a 50 ml tube full of warm physiological saline. Blood was collected for 30 min at 37° C. After 30 min, the mixture of blood and physiological saline was centrifuged at 1500 g. The red blood cells pellet was then lysed in H2O and the amount of hemoglobin was obtained by reading the absorbance at 416 nm. The volume of blood lost in each sample was calculated from a standard curve, which is obtained by lysing defined volumes (20 μl, 40 μl, 60 μl, 80 μl and 100 μl) of mouse blood in H2O to extract hemoglobin as described above. Blood loss for each mouse is presented in FIG. 6. For mice injected with FVIII-KB013bv, average blood loss was calculated to be 13±3 μl (mean±standard deviation; n=3 mice). For mice that received WT-FVIII-SQ, average blood loss was 194±146 μl (mean±standard deviation; n=5 mice), which is significantly more compared the mice injected with FVIII-KB013bv (p=0.0043 as determined using the Mann-Whitney test). Thus, FVIII-KB013bv displays hemostatic activity for a longer period of time than does WT-FVIII-SQ.

Example I: Use of FVIII-KB013bv as a Therapeutic Protein to Reduce the Formation of Allo-Antibodies

(22) Although VWF and FVIII circulate in plasma as a complex, there is a striking difference in the extent by which allo-antibodies develop following therapeutic application of these proteins. Development of allo-antibodies to VWF in response to replacement therapy is estimated to involve 5-10% of the patients with severe von Willebrand disease (James et al (2013) Blood vol 122, pp 636-640). In contrast, inhibitory allo-antibodies arise in up to 27% of previously untreated haemophilia A patients (Iorio et al. (2010) JTH vol 8, pp 1256-1265).

(23) The underlying reason for this difference in antibody development rate is unknown. Recently, it has been shown by Sorvillo and colleagues (Haematologica 2016 in press; doi:10.3324/haematol.2015.137067) that VWF remains associated at the surface of antigen-presenting cells without being endocytosed. In contrast, FVIII that was bound to this VWF is actually taken up by these cells and processed for incorporation into MHC-class II molecules, thereby allowing presentation to CD4+ T-cells. The notion that FVIII but not VWF enters into antigen-presenting cells could explain the antibody development is increased upon FVIII replacement therapy compared to VWF replacement therapy. A method that prevents dissociation of FVIII at the surface of the antigen presenting cell, and thereby uptake of FVIII by the antigen presenting cell would thus be a means to reduce the formation of allo-antibodies upon FVIII replacement therapy. One way to reduce dissociation of FVIII from VWF is by incorporating sdAbs against VWF in the FVIII protein, and an example hereof is FVIII-KB013bv of the present invention. FVIII-KB013bv could therefore be used as a therapeutic protein that is less immunogenic compared to FVIII that displays normal association-dissociation kinetics.

Example J: Fusion of KB-VWF-013 to Coagulation Factor VII Induces Complex Formation with VWF

(24) To determine whether sdAbs recognizing VWF can mediate binding of other proteins than FVIII to VWF, a cDNA was constructed encoding the sequence of human coagulation factor VII (FVII) fused to two copies of KB-VWF-013. Sequences encoding FVII and KB-VWF-013 were separated by a linker-sequence encoding a thrombin-cleavage site. The full sequence of this cDNA and corresponding protein is referred to as FVII-KB13-bv. FVII-KB-13-bv and WT-FVII were cloned into the pLIVE-plasmid (Mirus Bio, Madison, Wis., USA). Plasmids (100 μg/mouse) were injected into wild-type C57B16-mice via hydrodynamic gene transfer: plasmids are diluted in 0.9% saline with the volume corresponding to 10% of the animal's bodyweight (i.e. 2 ml for a 20-gram mouse). The solution is injected in the tail vein within 5 seconds. Four days after gene transfer, blood was collected via retro-orbital puncture from isoflurane-anesthetized mice and plasma was prepared by centrifugation (1500 g for 20 min at 22° C.). Plasma was then used to measure complexes between VWF and FVII or FVII-KB 13-bv that were formed in the plasma of the mice. Complexes were determined as follows: microtiter wells were coated with polyclonal sheep anti-human FVII antibodies (Affinity Biologicals, Ancaster ON, Canada) at a concentration of 2.5 μg/ml in 50 μl carbonate-buffer (0.07 M NaHCO3, 0.03 M Na2HCO3, pH 9.6) overnight at 4° C. Wells were washed thrice with Tris-buffered saline supplemented with 0.1% Tween-20 (TBS-T), then saturated with 5% BSA, 1% polyvinylpyrrolidone (PVP) in TBS-T for 2 hours at 37° C. and again washed 5 times with TBS-T. Then murine plasma samples (diluted 10-fold in 50 μl TBS-T containing 1% BSA) were added to the wells and incubated 2 hours at 37° C. After 5 washes with TBS-T (75 μl/well), bound FVII or FVII-KB13-bv was probed using peroxidase-labeled polyclonal rabbit anti-VWF antibodies (Dako) and detected by measuring peroxidase-mediated hydrolysis of 3,3′,5,5′-tetramethylbenzidine. Whereas for mice expressing FVII no signal above the background could be detected (OD450 nm=−0.038±0.033; mean±standard deviation; n=4 mice), suggesting the absence of complexes between VWF and FVII. In contrast, a clear signal was observed for plasma from each mouse expressing FVII-KB13-bv (OD450 nm=0.684±0.554; n=4; p=0.029 analyzed using Mann-Whitney test). This demonstrates that the fusion of FVII to sdAb KB-VWF-013 induces the protein to associate to circulating VWF.

REFERENCES

(25) Throughout this application, various references describe the state of the art to which this invention pertains. The disclosures of these references are hereby incorporated by reference into the present disclosure. 1. Mei B, Pan C, Jiang H, et al. Rational design of a fully active, long-acting PEGylated factor VIII for hemophilia A treatment. Blood 2010; 116(2):270-279. 2. Dumont J A, Liu T, Low S C, et al. Prolonged activity of a recombinant factor VIII-Fc fusion protein in hemophilia A mice and dogs. Blood 2012; 119(13):3024-3030. 3. Yee A, Gildersleeve R D, Gu S, et al. 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.