Protein Expression and Purification 115 (2015) 165–175

Contents lists available at ScienceDirect

Protein Expression and Purification

journal homepage: www.elsevier .com/ locate /yprep

Development, upscaling and validation of the purification process forhuman-cl rhFVIII (Nuwiq�), a new generation recombinant factor VIIIproduced in a human cell-line

http://dx.doi.org/10.1016/j.pep.2015.08.0231046-5928/� 2015 The Authors. Published by Elsevier Inc.This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

⇑ Corresponding author.E-mail address: stefan.winge@octapharma.se (S. Winge).

Stefan Winge a,⇑, Louise Yderland a, Christoph Kannicht b, Pim Hermans c, Simon Adema c,Torben Schmidt d, Gustav Gilljam a, Martin Linhult a, Maya Tiemeyer e, Larisa Belyanskaya f, Olaf Walter f

aOctapharma AB, Elersvägen 40, SE-112 75 Stockholm, SwedenbOctapharma Molecular Biochemistry, Walther-Nernst-Str. 3, 12489 Berlin, GermanycBAC BV – Thermo Fisher Scientific, 28 Huizerstraatweg, 1411 GP Naarden, The NetherlandsdOctapharma GmbH, Altenhöferallee 3, D-60438 Frankfurt, GermanyeOctapharma GmbH, Im Neuenheimer Feld 590, 69120 Heidelberg, GermanyfOctapharma AG, Seidenstrasse 2, 8853 Lachen, Switzerland

a r t i c l e i n f o

Article history:Received 28 May 2015and in revised form 17 August 2015Accepted 20 August 2015Available online 28 August 2015

Keywords:Haemophilia ARecombinant factor VIIIHuman-cl rhFVIIINuwiq�

Purification processPathogen safety

a b s t r a c t

Introduction: Human-cl rhFVIII (Nuwiq�), a new generation recombinant factor VIII (rFVIII), is the firstrFVIII produced in a human cell-line approved by the European Medicines Agency.Aims: To describe the development, upscaling and process validation for industrial-scale human-clrhFVIII purification.Methods and results: The purification process involves one centrifugation, two filtration, five chromatog-raphy columns and two dedicated pathogen clearance steps (solvent/detergent treatment and 20 nmnanofiltration). The key purification step uses an affinity resin (VIIISelect) with high specificity forFVIII, removing essentially all host-cell proteins with >80% product recovery. The production-scalemulti-step purification process efficiently removes process- and product-related impurities and resultsin a high-purity rhFVIII product, with an overall yield of �50%. Specific activity of the final productwas >9000 IU/mg, and the ratio between active FVIII and total FVIII protein present was >0.9. The entireproduction process is free of animal-derived products. Leaching of potential harmful compounds fromchromatography resins and all pathogens tested were below the limit of quantification in the finalproduct.Conclusions: Human-cl rhFVIII can be produced at 500 L bioreactor scale, maintaining high purity andrecoveries. The innovative purification process ensures a high-purity and high-quality human-clrhFVIII product with a high pathogen safety margin.

� 2015 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license(http://creativecommons.org/licenses/by/4.0/).

1. Introduction

Haemophilia A is an X-linked hereditary blood coagulation dis-order caused by a deficiency in coagulation factor VIII (FVIII) [1,2].Patients with haemophilia A are predisposed to recurrent bleeds,particularly into joints and soft tissues culminating in debilitatingarthropathy and long-term morbidity [1,3]. Haemophilia A can beeffectively treated with recombinant (rFVIII) or plasma-derivedFVIII (pdFVIII) [1,2]. The development of inhibitors to FVIII is themost serious complication in the treatment of haemophilia A[4–8]. Inhibitors are estimated to develop in up to 38% of previously

untreated patients [9] and 1% of previously treated patients (PTPs)[10].

Endogenous FVIII undergoes extensive post-translational modi-fications (PTMs), particularly sulphation and glycosylation, whichare necessary for full functional activity of FVIII [11–14]. Thedegree of sulphation at specific residues, notably Tyr1680, has alsobeen shown to determine the binding affinity of FVIII for its in vivomolecular chaperone, von Willebrand factor (VWF) [13,15]. TheTyr1680 sulphation site is in close proximity to a peptide,V1670-D1687, within the acidic a3-domain of FVIII (see Fig. 1)which recent work by Chiu et al., using hydrogen–deuteriumexchange mass spectrometry, has shown to be involved in FVIII-VWF binding [16]. The close proximity of the sulphation site at

Fig. 1. Schematic representation of the domain structure of human pdFVIII (without the B domain as per Nuwiq�), showing sites of post-translational modification (PTM).Functional FVIII protein comprises of 5 major domains: A1, A2, A3, C1 and C2. The heavy and light chains are co-ordinated by a divalent metal ion. Within the acidic a3domain, Chiu et al. [16] identified a putative FVIII-VWF binding region, V1670-D1687. Figure adapted with kind permission from Kannicht et al. [15].

166 S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175

Tyr1680 supports the observed impact of this sulphation site onthe affinity of FVIII-VWF binding.

Human-cl rhFVIII (Nuwiq�, Octapharma AG, Switzerland) is anew generation human rFVIII, without chemical modification orfusion with any other protein/fragment, which is produced in ahuman cell-line (HEK293F). As a result, human-cl rhFVIII hashuman PTMs, is fully sulphated, and contains only human glycans[15,17,18]. The comparable sulphation profile of human-cl rhFVIIIto human pdFVIII is thought to confer a higher binding affinityfor VWF than other rFVIII products. Consequently this mayenhance the stability of human-cl rhFVIII and reduce the immuno-genic potential created from unbound, circulating FVIII [15,18].Furthermore, human-cl rhFVIII is devoid of antigenic non-humanepitopes present in rFVIII products derived from hamster cell-lines [15]. The combination of full sulphation and the absence ofantigenic carbohydrate epitopes has the potential to minimisethe intrinsic immunogenicity of infused rFVIII [15].

Clinical trials involving 135 PTPs with haemophilia A have con-firmed the efficacy of human-cl rhFVIII in the prevention and treat-ment of bleeds with no inhibitor development or treatment-related serious adverse events reported [19]. Human-cl rhFVIII isthe first rFVIII produced in a human cell-line approved by the Euro-pean Medicines Agency for the prevention and treatment of bleed-ing in patients with haemophilia A.

The human cell-line for producing human-cl rhFVIII, the PTMsand functional characteristics of the purified product have beenreported previously [15,17,18]. The aim of this article is to reportthe development, upscaling and process validation for industrial-scale purification of human-cl rhFVIII. Several publications havereported on the purification of rFVIII [20–25]. However, these pub-lications are mainly based on non-human cells, have no or limitedinformation in regard to the development process, upscaling toproduction scale, process variability (validation) and processyields. Recently there has also been increased focus on leachablesfrom process aids, equipment and chromatography resins usedduring the process, especially for the manufacturing of biopharma-ceutical drugs that are intended for lifelong use such as FVIII. To

our knowledge, for the first time, theoretical and practical aspectsof removing less-defined impurities such as leachables is presentedrelated to the safety of a rFVIII product.

2. Materials and methods

2.1. Overall purification process

A purification process was developed to optimise the purity andpathogen safety of human-cl rhFVIII (Fig. 2), including one cellremoval step (centrifugation/filtration), two filtration and fivechromatography column steps. The chromatography resins and fil-ters used were Capto MMCTM (step 2), SP Sepharose FF� (step 3),VIIISelect (step 6), Q Sepharose FF� (step 8), Superdex 200 p.g.TM

(step 9, all GE Healthcare Life Sciences, Sweden), Bio20 STAX (step1, PALL), Sartobind� Q (step 4, Sartorius Stedim, Denmark) and Pla-nova 20 NTM (step 7, N.V. Asahi Kasei Bioprocess, Belgium) [26–29].Molar (M) is equally used to mol/kg throughout this work, due toproduction related data input.

2.2. Cell cultivation

A proprietary derivative of the human embryonic kidney cellline HEK293, named 293F – adapted to grow in serum-free culturemedia, was used to express the protein [17,30–32]. Revival of a fro-zen cell bank vial with HEK293F cells was performed for each newproduction campaign for production of Nuwiq�. Cells were furtherpropagated and expanded with in-house proprietary medium inshaker flasks followed by further propagation in a 20 L bioreactor.When sufficient active cell material of the required quality hadbeen achieved, a series of sequential batches consisting of themajor part of the cell suspension were transferred to a productionbioreactor. The production bioreactor (100 L) was operated as aperfuse batch process using a membrane-based cell retention sys-tem. The perfusion rate was adjusted daily after each cell densitymeasurement to maintain a cell specific perfusion rate until the

Fig. 2. Overview of the human-cl rhFVIII purification process.

S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175 167

product was collected in a batch harvest (starting material for thepurification Fig. 2, step 1). All bioreactors used were baffled, con-tinuously controlled and maintained in regard of pressure, temper-ature, pH, oxygen and carbon dioxide according to cultivationstandard procedures. The cells were agitated (3–30W/m3) duringthe process [32].

2.3. Harvesting (Fig. 2, step 1)

A batch harvest process was used, whereby sodium chloride(NaCl) was added to the cell suspension to a final concentrationof 0.3 M for detachment of FVIII from cells. [26]. Cells wereremoved by centrifugation and filtration before the harvested

168 S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175

material was applied directly to the capture step without any dilu-tion or other adjustments.

2.4. Capture purification (Fig. 2, step 2)

Primary product purification and concentration (capture) wasperformed on a 2 L multimodal cation exchange chromatographycolumn (bed height 15 cm and diameter 14 cm) in which150–200 kg cell free harvest was applied after equilibration witha 0.3 M NaCl solution at pH 7 with a flowrate of 70 L/h. The columnwas further washed by a stepwise procedure with a total amountof 50 column volumes, lowering the pH to 6.5 by using differentNaCl concentrations (0.3 M, 1 M, 0.1 M and 0.3 M). The last washalso included a combination of arginine (0.4 M) and ethylene glycol(10%). FVIII was eluted and collected from the column by increas-ing the arginine concentration to 0.8 M. Subsequently the columnwas regenerated with a high salt solution combined with a non-ionic detergent (0.5 M NaCl + 1% Triton X-100), low (1 M HAc)and high (1 M NaOH) pH and finally stored in a weak (0.05 M)sodium hydroxide solution before start of next production batch.

2.5. Cation exchange purification (Fig. 2, step 3)

The cation exchange purification was performed on a 2 L col-umn (bed height 12 cm and column diameter 14 cm) column inwhich the eluate from purification step 2 had been diluted tentimes to a conductivity equal to approximately 0.1 M NaCl,pH 6.5, before applying to the column, which previously had beenequilibrated with a 0.1 M NaCl, pH 6.5 buffer at a flow of 55 L/h.The column was further washed by a stepwise procedure with atotal amount of 50 column volumes by increasing the NaCl concen-tration to 0.15 M. FVIII was eluted and collected from the columnby further increasing NaCl concentration to 0.3 M. Subsequentlythe column was regenerated with a high salt solution combinedwith a non-ionic detergent (0.5 M NaCl + 1% Triton X-100), low(0.1 M HAc) and high (1 M NaOH) pH and finally stored in a weak(0.05 M) sodium hydroxide solution before start of next productionbatch.

2.6. DNA removal (Fig. 2, step 4)

3–4 L of the FVIII containing fraction from purification step 3was processed over a 2500 cm2 anion exchange membrane equili-brated with the same solution used for eluting FVIII in step 3 at amaximum pressure of 3 bar. FVIII was recovered in the flowthrough fraction.

2.7. Pathogen inactivation (Fig. 2, step 5)

2–5 batches (10–25 kg) from step 4 were pooled before adding1% Triton X-100 and 0.3% Tri(n-butyl)phosphate (TnBP) and stir-ring at 22 �C for 30–60 min.

2.8. Affinity purification (Fig. 2, step 6)

Affinity purification was performed on a 1.5 L column (bedheight 10 cm and diameter 14 cm) in which the pathogen activatedsolution from purification step 6 was applied to the column previ-ously equilibrated with a 0.3 M NaCl, pH 6.5 buffer at a flow of22 L/h. The column was further washed at a flowrate of 55 L/h bya stepwise procedure with a total amount of 25 column volumesby increasing NaCl concentration from 0.3 M to 1 M. FVIII waseluted and collected from the column by adding 50% ethylenegly-col. Subsequently the column was regenerated with a high saltsolution (2 M NaCl) and a low pH solution (0.15 M phosphoric

acid + 0.17 M HAc + 2.2% (v/v) benzyl alcohol) and finally storedin a ethanol solution (20%) before start of next production batch.

2.9. Nanofiltration (Fig. 2, step 7)

3–5 L of the FVIII containing fraction from purification step 6was diluted 10 times resulting in a NaCl concentration of 0.1 Mat pH 7.5. The solution was processed over a 1 m2 nanofilter equi-librated with a 0.1 M NaCl, pH 7.5 buffer solution at a pressure of0.6–0.8 bar. FVIII was recovered in the flow-through fraction.

2.10. Anion exchange purification (Fig. 2, step 8)

Anion exchange purification was performed on a 0.5 L column(bed height 7 cm and column diameter 10 cm) in which the filtratefrom purification step 7 was applied to the column, previouslyequilibrated with a 0.1 M NaCl, pH 7.5 buffer solution at a flowof 13 L/h. The column was further washed by a stepwise procedurewith a total amount of 15 column volumes by increasing the NaClconcentration from 0.1 M to 0.3 M. FVIII was eluted and collectedfrom the column by further increase of the NaCl concentration to0.4 M. Subsequently the column was regenerated with a high saltsolution (2 M NaCl) and high pH (1 M NaOH) and finally storedin a weak (0.01 M) sodium hydroxide solution before start of nextproduction batch.

2.11. Size exclusion purification (Fig. 2, step 9)

Size exclusion purification was performed on a 5 L column (bedheight 60–70 cm, column diameter 10 cm) in which 200–400 mL ofthe FVIII solution from purification step 8 was applied to the col-umn, previously equilibrated with the formulation solution ([42],a native buffered sugar/amino acid/non-ionic detergent solutionat pH 7) at a flow rate of 65 mL/min. The monomeric FVIII fractionwas collected according to the retention time as measured byabsorbance at 280 nm. Subsequently the column was regeneratedwith a high pH solution (1 M NaOH) and finally stored in a weak(0.01 M) sodium hydroxide solution before start of next productionbatch.

2.12. Assessment of pathogen safety

The following pathogens were selected for testing of both safe-guarding steps (solvent/detergent [S/D] inactivation and 20 nmnanofiltration) in scaled-down models mimicking the productionscale: human immunodeficiency virus type 1 (HIV-1), pseudora-bies virus (PRV), and bovine viral diarrhoea virus (BVDV). In addi-tion, hepatitis A virus (HAV), porcine parvovirus (PPV) and prionprotein (Strain 263 K hamster-adapted scrapie; PrPSc) were inves-tigated to specifically test the nanofiltration step. All pathogensafety studies were performed in duplicate. Cytotoxicity and inter-ference tests were performed to determine necessary dilutions toavoid matrix effects. A validated assay system for endpoint titra-tion and large-volume plating was used. PrPSc was investigatedusing Western Blot analysis. The obtained logarithmic reductionfactors were determined using the method of Spearman/Kaerberand, where applicable, the Poisson distribution at 95% upper confi-dence limits (p = 0.05).

2.13. Analytical methods

2.13.1. FVIII activity (FVIII:C)FVIII biologic activity was measured with a chromogenic assay

(COATEST SP FVIII kit, 82 4086 63, Chromogenix/InstrumentationLaboratory, US), based on a two-stage photometric method thatmeasures the biological activity of FVIII as a cofactor [33–35].

S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175 169

The method complies with the requirements in the EuropeanPharmacopoeia.

2.13.2. FVIII antigen (FVIII:Ag)The amount of FVIII antigen content was measured with an

ELISA kit (ASSERACHROM� VIII:Ag, enzyme immunoassay for FVIII,kit, Diagnostica Stago, France), as further described [36] withreplacement of the provided kit buffer with Tris–NaCl buffer +1%bovine serum albumin for sample dilutions.

2.13.3. Quantification of DNADNA was quantified by a real time quantitative PCR (qPCR)

assay based on SYBR Green 1 chemistry [37] with some addedimprovements. During each PCR cycle a 115 base pair fragmentfrom the ALU sequence families was amplified by the primers,ALU115-F and ALU115-R. The ALU sequence family are short inter-spersed repeated DNA elements distributed throughout primategenomes, but the assay only amplify DNA from human origin.The procedure allows for high throughput analysis of residualHEK293F DNA in cell free tissue culture media and its downstreampurification processes.

2.13.4. Total protein (Bradford)Protein concentration was determined by Bradford assay [38].

2.13.5. Host-cell proteinDetermination of host cell proteins (HCP) deriving from HEK

293 cells was performed using an ELISA method [39].

2.13.6. Specific activityFVIII purity (or also called specific activity) for a sample, was

calculated from the value achieved from the FVIII:C analysis (bio-logical activity, IU/mL) and divided by the value achieved fromthe analysis of total protein concentration (mg/mL).

2.13.7. Sodium dodecyl sulphate polyacrylamide gel electrophoresis(SDS–PAGE) silver stained

Analysis was carried out under reducing conditions to deter-mine the molecular weight distribution in the sample.

2.13.8. FVIII Western BlotAfter SDS–PAGE, proteins were transferred by electrophoresis

to a nitrocellulose membrane and polyclonal sheep antibodiesdirected to FVIII were added, followed by horseradishperoxidase-conjugated rabbit anti-sheep secondary antibody. Thechemiluminescent substrate was added and FVIII polypeptidebands were detected using a charge-coupled device camera(ChemiDocTM XRS, BioRad).

2.13.9. Size exclusion chromatography (SEC)FVIII monomers, aggregates and fragments in samples were

measured using a HPLC system (Waters 2695) with a size exclusionchromatography (SEC) analytical column (Superdex 200, 10/300GL, GE Healthcare) processed under native buffer conditions(25 mM HEPES, 0.5 M NaCl, 0.3 M arginine, 50 mM CaCl2, 0.02%Polysorbate 80, pH 7.5, temperature 25 �C) at a flowrate of0.25 mL/min and a sample injection volume of 200 lL. A fluores-cence detector (280–340 nm) was used to detect separated peaks.

2.14. Binding kinetics

The kinetic parameters of the VIIISelect affinity ligand for bind-ing to rhFVIII and to the rhFVIII light chain (rhFVIII-LC) were deter-mined by surface plasmon resonance (SPR) on a BiacoreTM 3000system (GE Healthcare). The VIIISelect ligand and a control ligandselective for human chorionic gonadotropin (hCG) (BAC B.V.) were

immobilised as biotin conjugates onto separate flow cells of astreptavidin sensor chip (GE Healthcare). The amount of immo-bilised VIIISelect and control ligand (reference flow cell) for affinitymeasurements of rhFVIII (MW 170 kDa,) corresponded to 153 and150 resonance units (RU), respectively. For binding analyses withrhFVIII-LC (MW 80 kDa), these values were 303 and 749 RU,respectively. Each analyte was allowed to bind for 3 min at a flowrate of 30 ll/min followed by a dissociation step of 15 min at 30ll/min. Flow cells were regenerated for 1 min at 30 ll/min using0.1 M glycine pH 2 and subsequently equilibrated with bindingbuffer prior to the next measurement. Binding curves were gener-ated in duplicate for each sample at 5, 25, 50, 150 and 250 nM. ForrhFVIII-LC, a sample of 500 nM was also included. Binding analyseswere performed in 3 buffers: HBS-P (10 mM HEPES, 0.15 M NaCl,0.005% surfactant P20 pH 7.4); HBS-EP (10 mM HEPES, 0.15 MNaCl, 3 mM EDTA, 0.005% surfactant P20 pH 7.4); and HBS-CP(10 mM HEPES, 0.15 M NaCl, 10 mM CaCl2, 0.005% surfactant P20pH 7.4). The rate constants of association (kass) and dissociation(kdiss) were calculated using the BIA evaluation software applyinga standard 1:1 Langmuir local fit. Equilibrium dissociation con-stants (KD) were calculated as kass/kdiss.

2.15. Analysis of leachables from chromatography resins

Resin leakage for the five chromatography column purificationsteps was evaluated in the equilibration buffer after passingthrough the column prior to product loading, and in the product-containing eluate fraction of the respective step. Leachable com-pounds originating from the resin were measured to assess the sta-bility of each resin. In addition, the resin leachate componentswere analysed in the final product solution. All method validationsand analyses of leachable compounds were performed by BIASeparations (Ljubljana, Slovenia). The acceptance criteria for eachleachable compound were set according to the limits of quantifica-tion of the respective analytical method.

2.15.1. Capto MMC resinAllyl glycidyl ether was analysed using a gas chromatography

column (DB-FFAP, 30 m � 0.53 mm, 1 lm film) under constanthelium pressure carrier gas of 7.4 Psi with a temperature of100 �C injecting 2 lL sample at 220 �C, detecting with aFID-detector at 250 �C and using a makeup nitrogen gas with25 mL/min, an airflow of 450 mL/min and a hydrogen flow of40 mL/min. Sample was prepared by mixing 1:1:2 with ethanoland methylene chloride, before the sample was centrifuged3500 rpm for 3 min in a 15 mL tube and the bottom (methylenechloride phase) was further used for injection in the gas chro-matograph. N-benzoyl-DL-homocysteine was analysed using aHPLC column (YMC Hydrosphere C18, 150 � 4.6 mm; 3 lm) at25 �C with a flowrate of 1 mL/min and a mobile phase of 0.05%TFA and methanol 45/55 (V/V), injecting 50 lL sample anddetecting at 250 nm SP.

2.15.2. Sepharose FF resinAllyl glycidyl ether was analysed using gas chromatography,

according to the same procedure as described for the Capto MMCresin above.

2.15.3. VIIISelect resin6-Aminocaproic acid was analysed using a HPLC column

(Kromasil 100 C-18, 250 � 4.6 mm; 5 lm) at 30 �C with a flowrateof 1 mL/min and a mobile phase of 25% methanol, 0.55 g/L Hexane-1-sulphonic acid sodium salt, 10 g/L monobasic potassium phos-phate, pH 2.2. The sample was prepared by mixing 0.5 mL samplewith 1.0 mL mobile phase injecting 50 lL sample and detectingat 210 nm. The VIIISelect ligand was analysed using a commer-

170 S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175

cially available ELISA (VIIISelect ligand leakage ELISA, ThermoFischer/BAC 81-2860.01) based on quantitative sandwich enzymeimmunoassay technology.

2.15.4. Q Sepharose FF resinAllyl glycidyl ether was analysed using gas chromatography,

according to the same procedure as described for the Capto MMCresin above. Trimethylamine was analysed using a head spacegas chromatography column (WCOT fused silica, 60 � 0.32 mm,coating CP-Volamine) under constant helium pressure carrier gasof 27 Psi with a temperature of 45 �C injecting 1 mL sample at175 �C, detecting with a FID-detector at 250 �C and using a makeupnitrogen gas with 25 mL/min, an airflow of 450 mL/min and ahydrogen flow of 40 mL/min. Samples for injection was preparedby mixing 1:1:1 with DMSO and 2.5 M NaOH at 60 �C.

2.15.5. Superdex 200 p.g. resinAgarose and dextran were quantified by a standard total organic

carbon analysis using high temperature combustion at 1350 �C,followed by IR detection of combustion gases – the amount ofCO2 in combustion gases is direct proportional to the amount ofcarbon in sample. 1 ml of sample was pipetted into ceramic cru-cible. 20 lL of concentrated HCl was added and the sample wasdried at 45 �C for at least 24 h until complete dryness. Preparedsample was analysed using ELTRA Elemental Analyser (modelMetalyt 100 CS, Germany).

3. Results

3.1. Development of the human-cl rhFVIII purification process

A purification process was developed to optimise the purity andpathogen safety of human-cl rhFVIII (Fig. 2), which includes onecell removal step (centrifugation/filtration), two filtration and fivechromatography column steps. The harvesting procedure used abatch harvest of the cell suspension as the expressed FVIII adheresto the cells. The interaction between FVIII and the cells could bebroken by increasing the ion concentration (by addition of e.g.,NaCl), allowing for enhanced extraction of FVIII, minimising thecell destruction and release of intracellular impurities into the har-vest. By optimising the amount of salt added, a high FVIII recov-ery/concentration could be achieved (Supplementary material,Figure S1) with relatively low amounts of cell-related impuritiesand low protease activity. A further advantage of the salted batchharvest process, compared with the perfusion process used in gen-eral for rFVIII production [22], was the stability of FVIII due to inhi-bition and rapid reduction of proteases. The relatively smallvolume and the non-physiological environment facilitate rapidcapture column protease removal under reduced proteolytic activ-ity conditions.

Based on the advantageous harvest process with regards toproductivity and product stability, a multi-modal resin (CaptoTM

MMC) was implemented as the capture step (Supplementarymaterial, Figure S2). The main advantages of this resin were:(a) its ability to bind FVIII at relatively high salt concentrations,which minimises protease activity and avoids dilution that wouldotherwise prolong the process time; (b) that the column could bethoroughly washed before product elution, thereby removingimpurities, including proteases, host-cell proteins and DNA (upto 3 log reduction, Fig. 3a and b) early in the process; and (c) thatthe FVIII molecule could be eluted from the column under nativeconditions, resulting in a high product recovery (Fig. 3d). Afterproduct load, four steps of washing (Supplementary material,Table S1) are processed over the multi-modal chromatographycolumn before product elution. The first wash, Wash 1, mimicked

the conditions in the harvest solution and removed impuritieswith no interaction with the capture column. Wash 2, a high-salt wash step, removed any impurities binding to the resinthrough ionic interactions. Wash 3, a low-salt wash, reducedimpurities adhering to the resin through hydrophobic interac-tions. Wash 4, an ethylene glycol/low arginine wash, was chosenbecause it was shown to elute proteins other than FVIII from theresin through a combinatorial mode. Finally, FVIII was elutedfrom the column by increasing the arginine concentration to0.8 M. Arginine was used based on experimental data showinggood FVIII-eluting properties and its known protein anti-aggregation properties [40]. FVIII activity was found to be stablein the eluting buffer for at least 24 h, indicating no, or very low,protease activity. A subsequent cation-exchange chromatographystep further increases the purity of FVIII, mainly through removalof host-cell proteins, and ensures a suitable buffer environmentfor the subsequent DNA removal step. An SP Sepharose FFcolumn was found to be robust and reliable for this purpose.

Based on experimental data (Supplementary material, FiguresS3), a single-use anion-exchange membrane was used as a specificDNA removal step. Depending on DNA reduction of other steps inthe purification process, one or more specific single-use DNAremoval membranes could be introduced in the process. Owingto the excellent DNA removal properties (9 log) throughout theentire chosen purification process (Fig. 3b), one specific DNA-removal step was sufficient. Typically, 1.5–2.5 log reductions inDNA levels were achieved for the single-use DNA removal mem-brane with no product loss (Fig. 3d).

The S/D virus inactivation method was included as an extrasafety measure, even though the risk of virus contamination in arecombinant biopharmaceutical process is in principle negligible.In-process material was supplemented with 1% Triton X-100 and0.3% Tri-(n-butyl)-phosphate and incubated at room temperaturefor at least 30 min, according to state-of-the-art S/D virus inactiva-tion practice [41].

An affinity purification step was included, a key step in thepurification process. As the entire production process is free ofanimal-derived components, a prerequisite of the intended affinityligand was that it should be completely free of animal components.The affinity ligand chosen was based on a camelid-derived single-domain antibody fragment expressed in Saccharomyces cerevisiae.The affinity ligand resin (VIIISelect) was developed in collaborationbetween BAC, GE Healthcare and Octapharma. A prototype resinshowed promising purification properties (Fig. 4) and was furtherdeveloped to produce a resin with high specificity and stability.Small-scale cycling studies demonstrated that the VIIISelect col-umn could be used and regenerated at least 20 times without sig-nificant decrease in capacity, which was later verified inproduction scale (Fig. 5 and Supplementary material, Figure S7).Furthermore, no leakage of ligand from the resin could be detectedunder the selected process conditions (Supplementary material,Table S3).

The kinetic parameters for binding of VIIISelect ligand tointact rhFVIII and to its isolated light chain (rhFVIII-LC) weredetermined by SPR. No significant difference in association rates(kass) was observed between the two analytes in any of the bind-ing buffers tested (Table 1). When tested in HBS-P binding buffer,the free rhFVIII light chain showed an almost 20-fold higher dis-sociation rate (kdiss) compared with intact rhFVIII, indicating for-mation of a less stable complex with the VIIISelect ligand. Asimilar �20-fold increase in kdiss was observed for intact rhFVIIIwhen measured in buffer containing EDTA (HBS-EP). EDTA candeplete Ca2+ and Cu2+ ions from rhFVIII, which are essential forconserving the functional structure of rhFVIII. Furthermore, Cu2+

enhances the inter-chain affinity between the heavy and lightchains. Therefore, the observed increase in kdiss is likely due to

S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175 171

EDTA-induced formation of free rhFVIII light chains. The presenceof CaCl2 in the buffer (HBS-CP) did not affect the kineticparameters for rhFVIII binding.

For rhFVIII-LC, addition of EDTA slightly increased the kdisscompared with HBS-P, whereas addition of CaCl2 lowered thedissociation rate. Although Ca2+ ions can alter the conformationof the FVIII light chain, further studies are required to address thisas a possible cause.

The VIIISelect chromatography column step was shown toreduce process-related (Fig. 3b), product-related (Figs. 3c and 5b)and host-cell impurities (Figs. 3a and 5a), resulting in a pure con-centrated FVIII product fraction with more than 80% recovery(Fig. 3d).

The product-containing fraction from the VIIISelect column wasdiluted and passed through a Planova 20 NTM nanofilter (mean poresize: 19 nm). Downscaled virus validation studies (nanofiltermembrane surface area: 0.001 m2) confirmed that the nanofiltra-tion removed one of the smallest known viruses, PPV, by >4 log.In fact, none of the tested viruses or prions could be detected afterperforming the nanofiltration step under selected process condi-tions (Supplementary material, Table S2).

After nanofiltration, the product was concentrated and furtherpolished on a Q Sepharose FF anion-exchange chromatography col-umn that removed trace amounts of host-cell proteins (Supple-mentary materials, Figure S4 and Figure S5), DNA (Fig. 3b),process chemicals and FVIII degeneration products (Fig. 3c). Theproduct recovery was more than 80%.

The final step of human-cl rhFVIII purification is size exclusionchromatography (SEC) on a Superdex 200 p.g.TM resin (Supplemen-tary material, Figure S6). The main purpose of this step is further

Fig. 3. (a–d) Purification of FVIII as measured by specific activity (a), DNA content (b), FVThe number of batches used for each calculated value is indicated in the graphs. (a) Prospecific activity. (b) DNA content after each purification step in production batches, as mein production batches. (d) FVIII recovery over each purification step in production batch

polishing and preparation (formulation) of the product for thefreeze-drying step. SEC was preferred over ultrafiltration, anotherpotential technique for formulation, due to its combinatorial prop-erties of polishing, buffer exchange and removal of aggregates andfragments. The SEC purification technique fulfilled the highdemand for product stability, purity and quality, in order to opti-mise patient safety. The product recovery for this step was essen-tially 100% (Fig. 3d), and the end product fulfilled all set demandswith regards to purity (Figs. 3–6).

3.2. Upscaling and validation of the purification process

Upscaled and downscaled process models are critical for estab-lishment of a robust validated production process. The developedprocess was upscaled to a pilot plant scale to be able to producematerial for preclinical and clinical studies, and was thereafter fur-ther upscaled to a production plant whereby the process validationtakes place to verify purity, quality, safety and recovery under goodmanufacturing production conditions. Downscaled process modelsare critical to be able to perform virus clearance analyses and resincycling studies, which would be costly and inefficient on a largerscale. The manufacturing process and the equipment used werein principle comparable in the pilot and production plants. Productpurity, expressed as specific activity (FVIII IU/mg protein), DNAimpurities and process yields were highly comparable across thethree scales: laboratory, pilot and production (Supplementarymaterial, Figure S7a–c). Thirty-six production batches (twenty-eight in 100 L and eight in 500 L bioreactor scale) were tested withno demonstrable changes in product characteristics.

III:C/FVIII:Ag ratio (c) and FVIII yield (d) after each step of the purification process.tein impurities after each purification step in production batches, as measured byasured by pg DNA/1000 IU FVIII. (c) Ratio FVIII:C/FVIII:Ag after each purification stepes.

1 2 3 4 Fig. 4. Proof of concept purification of FVIII by chromatography over a prototypeVIIISelect affinity ligand resin, as assessed by silver-stained sodium dodecylsulphate polyacrylamide gel electrophoresis (SDS–PAGE). Lane 1: FVIII control;lane 2: material before the prototype affinity step; lane 3: flow-through fraction notbinding to the affinity ligand; lane 4: FVIII purified elution fraction.

172 S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175

Process-related (proteins and DNA) and product-related (ratioactive/inactive FVIII) impurities were effectively removed duringthe purification process (Fig. 3a–c). As reported previously, thepurified product was free of contamination by viruses and otherpathogens (Supplementary material, Table S2) [17]. The mean FVIIIyield for each purification step was more than 80%, with a yield forthe overall process of �50% (Fig. 3d). SDS–PAGE and Western Blotanalyses (Fig. 5a and b) demonstrated a progressive increase inpurity throughout the production process, with no degradationproducts observed in the product eluates after the first capture

a

Fig. 5. Silver stained sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDSsteps.

chromatography purification step. The host cell protein contentin the final product, as analysed with a specific host cell protein(HCP) assay, was <40 ng/1000 IU FVIII (data not shown), and thecontent of DNA in the final product (<10 pg DNA/1000 IU FVIII)was significantly lower than that required by regulatory demands(Fig. 3b). Aggregates and fragments, as measured by SEC, corre-sponded to <2% of the final product (Fig. 6).

3.3. Analyses of leachables from the purification columns

Leached resin was not detected in any of the samples in thepost-column equilibration buffer or in the product eluate fractions(Supplementary material, Table S3).

4. Discussion

Human-cl rhFVIII (Nuwiq�) is a new generation rFVIII product,without chemical modification or fusion with any other protein(or protein fragment), produced in a human cell-line. The produc-tion process for human-cl rhFVIII consists of separate cultivation,purification and pharmaceutical processes that utilise innovativetechniques, and completely avoids the use of human- or animal-derived additives [26–32,42,43]. The development of the humancell-line and purification scheme for human-cl rhFVIII has beenreported previously [17]. The highly pure product was shown tobe fully sulphated, glycosylated similarly to human pdFVIII, andto have high specific FVIII activity [15,17,18]. Studies in previouslytreated children and adults have shown that human-cl rhFVIII isefficacious in the prevention and treatment of bleeding episodes,and no patients have developed inhibitors or allergic reactions todate [19].

The definition of purification is to remove unwanted substancesfrom a final product. For a biopharmaceutical recombinant process,such as human-cl rhFVIII production, this includes removal of host-cell related impurities (e.g., proteins, DNA and potential viruses),

b

–PAGE) (a) and Western Blot (b) analyses of samples from each of the purification

Table 1Kinetic parameters determined by SPR of the VIIISelect ligand in binding to intactrhFVIII and isolated FVIII light chain under different conditions.

Analyte Binding buffer kass (�105 1/Ms) kdiss (�10�4 1/s) KD1 (nM)

rhFVIII HBS-P 3.34 ± 2.34 0.22 ± 0.04 0.07HBS-EP 4.74 ± 2.92 4.29 ± 1.03 0.91HBS-CP 1.81 ± 1.00 0.24 ± 0.08 0.13

rhFVIII-LC

HBS-P 1.71 ± 1.03 4.22 ± 0.55 2.47HBS-EP 1.38 ± 0.79 8.62 ± 1.36 6.25HBS-CP 1.04 ± 0.91 1.43 ± 0.32 1.38

kass, rate constant of association; KD, equilibrium dissociation constants; kdiss, rateconstant of dissociation; SPR, surface plasmon resonance.

1 KD calculated as kass/kdiss using average values of both parameters.

S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175 173

process additives (e.g., cell cultivation medium components, virusinactivation chemicals and purification step aids) and (potentially)leachables from equipment (e.g., plastic bags, tubing, filters) andother material used during the purification process (e.g., chro-matography resin ligands and support).

During development of a new purification process, there areseveral options when selecting the type and number of purificationsteps, depending on product properties and demand for purity inthe final product. It is important that the purification process isdesigned for maximum removal of known and unknown (e.g.,pathogenic) impurities. This is especially true for products suchas FVIII that are intended for lifelong treatment. The human-clrhFVIII purification process was developed with careful selectionof five chromatography column purification techniques (Fig. 2),each with a different action to ensure maximum overall reductionof impurities. The multi-modal cation chromatography columncapture step purifies using a combination of cationic, hydrogen,thiophilic and hydrophobic interactions. The cation exchange chro-matography column purification step interacts through negatively

Fig. 6. Aggregate and fragment size exclusion chromatography (SEC) analysis chromasubstance). The retention time reflects the molecular size, as detected with a fluorescenFVIII monomer peak of >99%.

charged groups. The affinity chromatography column step inter-acts with a ligand that has been specifically developed to bindintact FVIII. The anion exchange chromatography column stepinteracts through positively charged groups. The SEC column stepseparates according to molecular size. The first four column stepsbind the FVIII molecule. Before FVIII is eluted, the columns arewashed to waste with a total of 130 column volumes of differentwash buffers (Materials and methods 2.4, 2.5, 2.8 and 2.10), con-tributing to the safety measures for removing all types of potentialimpurities.

Chromatography resins can be a source of leachable as impuri-ties. Leachables for all chromatography column steps in the FVIIIpurification process were studied under actual process conditionsand no leaching from any resins was detectable. In addition, allplastic components (e.g., filters, tubing, bags and ampoules) usedin the production process have been evaluated in terms of leach-ables. This evaluation included review of materials, supplier infor-mation, risk analysis and, if deemed necessary, leaching studiesperformed under actual processing conditions. All plastic compo-nents used in the human-cl rhFVIII manufacturing process are cer-tified according to the requirements of the USP plastic class VI andare suitable for use in routine production of biopharmaceuticals.

Commercially available rFVIII products are purified based onchromatographic column purification, which include an affinitychromatography resin as a key step in the process [22]. An affinityresin developed based on a single domain antibody (now commer-cialised as VIIISelect) was used for purification of human-cl rhFVIII.The kinetic parameters for binding of VIIISelect ligand to intactrhFVIII and to its isolated light chain (rhFVIII-LC) were determinedby binding affinity studies (SPR). No significant difference in associ-ation rates (kass) was observed between the two analytes in any ofthe binding buffers tested (Table 1), indicating that the epitoperecognised by the VIIISelect ligand is equally accessible on the free

togram of a sample taken after the final human-cl rhFVIII purification step (drugt detector. The figure shows a typical chromatogram appearance for a human-cl rh

174 S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175

FVIII light chain as on intact FVIII molecules. Studies also showedthat the VIIISelect affinity ligand forms a more stable complex withintact rhFVIII molecules than with free rhFVIII-LC only. As shownduring thedevelopment time [29] and in theproductionprocess val-idation (Fig. 3c), rhFVIII-LC originating from the cultivation processwas removed during the VIIISelect purification step. A prerequisitefor the development of the affinity resin was that it should be com-pletely free of animal components, and this was achieved by usingyeast (S. cerevisiae) as the production system for the ligand. Thedeveloped resin was found to have excellent purification properties(Figs. 3a and 5a), with a product recovery ofmore than 80% (Fig. 3d).

The purification process resulted in a highly pure product, witha high overall yield of approximately 50%. This yield compares tothe 15% yield reported previously for another rFVIII product [25].This may be attributable to protection of human-cl rhFVIII fromproteolytic degradation during the cultivation/harvest step, thepropriety [26] non-physiological salt concentration used, and/orthe multi-modal capture chromatography step that includes sev-eral wash steps to remove potential proteases.

Host cell protein and DNA content were efficiently removedduring the purification process, as analysed by specific activity(units of FVIII/milligrams of protein, Fig. 3a), silver-stainedSDS–PAGE (Fig. 5a), specific HCP assay (data not shown) andq-PCR DNA analysis (Fig. 3b). In the final product, host cell proteincontent was <40 ng HCP/1000 IU FVIII and DNA content was <10 pgDNA/1000 IU FVIII, thereby fulfilling all demands for the removal ofthese main impurities from the cell cultivation process.

Non-biologically active FVIII molecules were removed duringthe later part of the purification process. No detectable FVIII degra-dation was observed (Fig. 5b) and the ratio of FVIII:C/FVIII:Ag wasclose to 1 (Fig. 3c) in the essentially monomeric (170 kD) final pro-duct (Fig. 6). This finding is potentially of great importance, as inac-tive product molecules might contribute to inhibitor developmentin patients [44].

The use of two dedicated pathogen safeguarding steps for inac-tivation (S/D) and removal (20 nm nanofiltration) of hypotheticallypresent pathogens is requested by health authorities for modernrecombinant production processes. The previously reported highpathogen safety, with regards to viruses and prions, of purifiedhuman-cl rhFVIII [17] was confirmed here (Supplementary mate-rial, Table S2).

5. Conclusion

Validation of the production process demonstrated that allrequirements for a robust and reproducible manufacturing processwere fulfilled, and that the production process is suitable for com-mercial production. Human-cl rhFVIII could be produced in biore-actors of up to 500 L and further purified resulting in high purityand high recoveries.

The quality demands for new biopharmaceutical products haveincreased over recent years as technology has developed and safetyrequirements have become more stringent. The human-cl rhFVIIIproduction process has utilised the most innovative technologyand combined a plethora of knowledge and experience to developa highly innovative and effective method for commercial-scale pro-duction of an essentially monomeric human-cl rhFVIII productwith high specific activity [26–28,30–32,42,43].

Conflicts of interest

S. Winge, L. Yderland, C. Kannicht, T. Schmidt, G. Gilljam, M. Lin-hult, M. Tiemeyer, L. Belyanskaya and O. Walter are employees ofOctapharma. P. Hermans and S. Adema are employees of BAC BV– Thermo Fisher Scientific.

Author contributions

S. Winge, L. Yderland, P. Hermans (VIIISelect affinity ligandstudies) and T. Schmidt (pathogen safety) wrote the first draft ofthe manuscript. All authors contributed to the interpretation ofthe data, manuscript revisions and approval of the submittedversion.

Intellectual property

For enquires of any proprietary rights mentioned in this article,please contact Octapharma AG, Legal Department, Lachen,Switzerland.

Acknowledgements

The authors would like to thank the following colleagues andformer employees (⁄) at Octapharma AB (Stockholm, Sweden) fortheir participation during the development and/or process valida-tion: Irene Agerkvist⁄, Carin Borgvall, Ulrika Ericsson⁄, Mats Jern-berg⁄, Elisabeth Lindner⁄, Cecilia Lindwall⁄, Kristina Martinell⁄,Helena Sandberg⁄, Christina Leo, Karin Stackerud, Erica Johansson,Ann-Charlotte Hinz, Ann-Christine Eskekärr, Lena Johansson,Tomas Åslund and Irena Knappik. The authors acknowledge thecollaboration and support of Laurens Sierkstra and Ingeborg vanGemeren at BAC BV – Thermo Fisher Scientific, Naarden, TheNetherlands, and Henrik Ihre and Henrik Neu from GE Healthcare,Custom Designed Media, Uppsala, Sweden. Editorial support wasprovided by Liz Bennett from nspm Ltd. (Meggen, Switzerland)with funding from Octapharma AG (Lachen, Switzerland).

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.pep.2015.08.023.

References

[1] A. Coppola, M. Di Capua, M.N. Di Minno, M. Di Palo, E. Marrone, P. Ierano, C.Arturo, A. Tufano, A.M. Cerbone, Treatment of hemophilia: a review of currentadvances and ongoing issues, J. Blood Med. 1 (2010) 183–195.

[2] M. Franchini, The modern treatment of haemophilia: a narrative review, BloodTransfus. 11 (2013) 178–182.

[3] K. Fischer, J.G. Bom, E.P. Mauser-Bunschoten, G. Roosendaal, H.M. Berg, Effectsof haemophilic arthropathy on health-related quality of life and socio-economic parameters, Haemophilia 11 (2005) 43–48.

[4] T.M. Brown, W.C. Lee, A.V. Joshi, C.L. Pashos, Health-related quality of life andproductivity impact in haemophilia patients with inhibitors, Haemophilia 15(2009) 911–917.

[5] S.C. Darby, D.M. Keeling, R.J. Spooner, S. Wan Kan, P.L. Giangrande, P.W. Collins,F.G. Hill, C.R. Hay, U.K.H.C.D. Organisation, The incidence of factor VIII andfactor IX inhibitors in the hemophilia population of the UK and their effect onsubsequent mortality, 1977–99, J. Thromb. Haemost. 2 (2004) 1047–1054.

[6] M. Morfini, S. Haya, G. Tagariello, H. Pollmann, M. Quintana, B. Siegmund, N.Stieltjes, G. Dolan, J. Tusell, European study on orthopaedic status ofhaemophilia patients with inhibitors, Haemophilia 13 (2007) 606–612.

[7] A. Gringeri, L.G. Mantovani, L. Scalone, P.M. Mannucci, C.S. Group, Cost of careand quality of life for patients with hemophilia complicated by inhibitors: theCOCIS Study Group, Blood 102 (2003) 2358–2363.

[8] J. Astermark, Prevention and prediction of inhibitor risk, Haemophilia 18(Suppl. 4) (2012) 38–42.

[9] S.C. Gouw, J.G. van der Bom, R. Ljung, C. Escuriola, A.R. Cid, S. Claeyssens-Donadel, C. van Geet, G. Kenet, A. Makipernaa, A.C. Molinari, W. Muntean, R.Kobelt, G. Rivard, E. Santagostino, A. Thomas, H.M. van den Berg, PedNet, R.S.Group, Factor VIII products and inhibitor development in severe hemophilia A,N. Engl. J. Med. 368 (2013) 231–239.

[10] M. Xi, M. Makris, M. Marcucci, E. Santagostino, P.M. Mannucci, A. Iorio,Inhibitor development in previously treated hemophilia A patients: asystematic review, meta-analysis, and meta-regression, J. Thromb. Haemost.11 (2013) 1655–1662.

[11] P.J. Lenting, O.D. Christophe, P. Gueguen, The disappearing act of factor VIII,Haemophilia 16 (2010) 6–15.

S. Winge et al. / Protein Expression and Purification 115 (2015) 165–175 175

[12] M.P. Kosloski, R.D. Miclea, S.V. Balu-Iyer, Role of glycosylation inconformational stability, activity, macromolecular interaction andimmunogenicity of recombinant human factor VIII, AAPS J. 11 (2009) 424–431.

[13] A. Leyte, H.B. Van Schijndel, C. Niehrs, W.B. Huttner, M.P. Verbeet, K. Mertens,J.A. van Mourik, Sulfation of Tyr1680 of human blood coagulation factor VIII isessential for the interaction of factor VIII with von Willebrand factor, J. Biol.Chem. 266 (1991) 740–746.

[14] S.A. Brooks, Appropriate glycosylation of recombinant proteins for human use:implications of choice of expression system, Mol. Biotechnol. 28 (2004) 241–255.

[15] C. Kannicht, M. Ramstrom, G. Kohla, M. Tiemeyer, E. Casademunt, O. Walter, H.Sandberg, Characterisation of the post-translational modifications of a novel,human cell line-derived recombinant human factor VIII, Thromb. Res. 131(2013) 78–88.

[16] P.L. Chiu, G.M. Bou-Assaf, E. Seth Chhabra, M.G. Chambers, R.T. Peters, J.D.Kulman, T. Walz, Mapping the interaction between factor VIII and vonWillebrand factor by electron microscopy and mass spectrometry, Blood(2015).

[17] E. Casademunt, K. Martinelle, M. Jernberg, S. Winge, M. Tiemeyer, L. Biesert, S.Knaub, O. Walter, C. Schroder, The first recombinant human coagulation factorVIII of human origin: human cell line and manufacturing characteristics, Eur. J.Haematol. 89 (2012) 165–176.

[18] H. Sandberg, C. Kannicht, P. Stenlund, M. Dadaian, U. Oswaldsson, C. Cordula,O. Walter, Functional characteristics of the novel, human-derived recombinantFVIII protein product, human-cl rhFVIII, Thromb. Res. 130 (2012) 808–817.

[19] L.A. Valentino, C. Negrier, G. Kohla, A. Tiede, R. Liesner, D. Hart, S. Knaub, Thefirst recombinant FVIII produced in human cells – an update on its clinicaldevelopment programme, Haemophilia 20 (Suppl. 1) (2014) 1–9.

[20] J. McCue, D. Osborne, J. Dumont, R. Peters, B. Mei, G.F. Pierce, K. Kobayashi, D.Euwart, Validation of the manufacturing process used to produce long-actingrecombinant factor IX Fc fusion protein, Haemophilia 20 (2014) e327–e335.

[21] B. Kelley, M. Jankowski, J. Booth, An improved manufacturing process forXyntha/ReFacto AF, Haemophilia 16 (2010) 717–725.

[22] B.G. Boedeker, Production processes of licensed recombinant factor VIIIpreparations, Semin. Thromb. Hemost. 27 (2001) 385–394.

[23] R. Jiang, T. Monroe, R. McRogers, P.J. Larson, Manufacturing challenges in thecommercial production of recombinant coagulation factor VIII, Haemophilia 8(Suppl. 2) (2002) 1–5.

[24] M. Griffith, H. Kingdon, S.L. Liu, W. Burkart, In-process controls andcharacterization of recombinate antihemophilic factor (recombinant), Ann.Hematol. 63 (1991) 166–171.

[25] L. Thim, B. Vandahl, J. Karlsson, N.K. Klausen, J. Pedersen, T.N. Krogh, M. Kjalke,J.M. Petersen, L.B. Johnsen, G. Bolt, P.L. Norby, T.D. Steenstrup, Purification andcharacterization of a new recombinant factor VIII (N8), Haemophilia 16 (2010)349–359.

[26] S. Winge, Method for improved isolation of recombinantly produced proteins.Patent application PCT/WO 2006103258-A1, published 2006.

[27] G. Gilljam, M. Jernberg, S. Winge, A. Neisser-Svae, A process for isolation andpurification of a target protein free of prion protein (prpsc). Patent applicationPCT/WO 2009024620-A2, published 2009.

[28] C. Borgvall, U. Ericsson, G. Gilljam, M. Jernberg, S. Winge, A process of purifyingcoagulation factor VIII. Patent application PCT/WO 2009156430-A1, published2009.

[29] S. Winge, E. Johansson, M. Dadaian, B. Fuchs, A process for manufacturingFactor VIII having an improved ratio of FVIII:C/FVIII:Ag. PCT/EP2015/051028,filed 2015.

[30] C. Hauser, A. Hoerster, C. Schroeder, M. Lehnerer, Production of recombinantblood clotting factors in human cell lines. Patent application PCT/WO2001070968-A2, published 2001.

[31] C. Schroeder, C. Wegmann, H. Ding, Serum-free stable transfection andproduction of recombinant human proteins in human cell lines. Patentapplication PCT/WO 2010020690-A1, published 2010.

[32] I. Agerkvist, P. Aizawa, A method of increasing the productivity of eucaryoticcells in the production of recombinant FVIII. Patent application PCT/WO2012156356-A1, published 2012.

[33] S. Rosen, Assay of factor VIII: C with a chromogenic substrate, Scand. J.Haematol. Suppl. 40 (1984) 139–145.

[34] G. Carlebjork, U. Oswaldsson, S. Rosen, A simple and accurate microplate assayfor the determination of factor VIII activity, Thromb. Res. 47 (1987) 5–14.

[35] A.R. Mire-Sluis, T. Gerrard, R.G. Das, A. Padilla, R. Thorpe, Biological assays:their role in the development and quality control of recombinant biologicalmedicinal products, Biologicals 24 (1996) 351–362.

[36] J.P. Girma, E. Fressinaud, A. Houllier, Y. Laurian, J. Amiral, D. Meyer, Assay offactor VIII antigen (VIII:CAg) in 294 haemophilia A patients by a newcommercial ELISA using monoclonal antibodies, Haemophilia 4 (1998) 98–103.

[37] N. Umetani, J. Kim, S. Hiramatsu, H.A. Reber, O.J. Hines, A.J. Bilchik, D.S. Hoon,Increased integrity of free circulating DNA in sera of patients with colorectal orperiampullary cancer: direct quantitative PCR for ALU repeats, Clin. Chem. 52(2006) 1062–1069.

[38] M.M. Bradford, A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dyebinding, Anal. Biochem. 72 (1976) 248–254.

[39] E. Engvall, P. Perlmann, Enzyme-linked immunosorbent assay (ELISA).Quantitative assay of immunoglobulin G, Immunochemistry 8 (1971) 871–874.

[40] D. Ejima, K. Tsumoto, T. Arakawa, Biotech applications of arginine, BioProcessInt. 3 (2005) 20–28.

[41] P.L. Roberts, Virus inactivation by solvent/detergent treatment using Triton X-100 in a high purity factor VIII, Biologicals 36 (2008) 330–335.

[42] B. Rippner, J. Österberg, U. Nilsson, E. Ivarsson, I. Agerkvist, New protectingcompositions for recombinantly produced factor VIII. Patent application PCT/WO 2010026186-A1, published 2010.

[43] H. Sandberg, P. Stenlund, C. Schröder, E. Casademunt, M. Tiemeyer,Recombinantly produced human factor VIII and IX. Patent application PCT/WO 2010020690-A1, published 2010.

[44] J.C. Lin, R.A. Ettinger, J.T. Schuman, A.H. Zhang, M. Wamiq-Adhami, P.C.Nguyen, S.M. Nakaya-Fletcher, K. Puranik, A.R. Thompson, K.P. Pratt, Six aminoacid residues in a 1200 A2 interface mediate binding of factor VIII to anIgG4kappa inhibitory antibody, PLoS ONE 10 (2015) e0116577.