Introduction

Hemophilia A is a bleeding disorder caused by a deficiency in the coagulation factor VIII (FVIII) that affects one in 5,000–10,000 males (Hoyer 1994; Soucie et al. 1998). Treatment for hemophilia A patients consists of replacement therapy using human plasma-derived FVIII (pdFVIII) or recombinant FVIII (rFVIII), which is available to a small percentage of patients worldwide (20–30 %) (Stonebracker et al. 2010).

Recombinant products have the advantage of unlimited availability and safety. Despite the number of the recombinant products available, the cost of such products is still unavailable and/or unaffordable. This problem may be addressed by more efficient and cost-effective production methods (Lynch et al. 1993). In this area, one of the most significant challenges is the establishment of optimized cell lines with high expression levels (Mei et al. 2006).

Currently, commercially-available rFVIII products are produced in murine cells (CHO and BHK) using conventional plasmid vectors (Spencer et al. 2011). The gene delivery mediated by non-viral vectors is less efficient than viral vectors and the transgene expression has a short duration (Nishikawa and Huang 2001). Additionally, recombinant proteins produced in non-human cells can present different patterns of glycosylation, which create a product more likely to cause immune reactions in patients (Spencer et al. 2011). Human cell lines have been an alternative since these expression systems are expected to produce recombinant proteins with post-translational modifications that better resemble their natural counterpart, reducing potential immunogenic reactions (Durocher and Butler 2009).

The SK-HEP human cell line is under consideration for the production of FVIII due to its high rFVIII expression levels when compared with other human and murine cells (Herlitschka et al. 1998; Picanco et al. 2007). However, the rFVIII produced by SK-HEP cells has not been functionally tested in vivo and the potential of these cells to be used as host cells for industrial-scale recombinant protein production has not been shown either.

In this study, we have produced human recombinant B-domain deleted FVIII (rBDDFVIII) in the human SK-HEP-1 cell line using a lentiviral vector with the cytomegalovirus (CMV) promoter, which was described as effective for rFVIII expression (Russo-Carbolante et al. 2011). We also performed an initial characterization of this production platform and evaluated its scale-up potential. The commercial production of rFVIII in Brazil might improve the treatment of the patients who currently receive infusions of purified pdFVIII.

Materials and methods

rBDDFVII-producing cell line modified with lentiviral vector

The rBDDFVIII-producing cell line was generated from the human liver adenocarcinoma cell line SK-HEP-1 (DSMZ ACC 141) and a replication-defective self-inactivating HIV-1-derived vector containing the coagulation factor FVIII gene [cPPT-C(FVIIIDB)IGWS] (Picanco et al. 2007). rBDDFVIII SK-HEP cells, here referred to as rSK-HEP, was cultured in Dulbecco’s modified eagle medium (DMEM) supplemented with 100 IU/ml penicillin, 0.1 mg/ml streptomycin and 10 % v/v fetal bovine serum (FBS), at 37 ºC in a 5 % v/v CO2 atmosphere.

Animals

Hemophilia A mice (B6;129S4-F8 tm1kaz/J) were obtained from Jackson Laboratory (Bar Harbor, ME, USA). Adult mice aged 8–12 weeks were used for the tail-clipping bleeding and pharmacokinetic studies. All animal manipulations were done in accordance with institutional guidelines under approved protocols (082/2008).

rBDDFVIII production experiments: static and suspension cultures

To analyze rBDDFVIII production in static culture, rSK-HEP cells were seeded at 80,000 cells/cm2 in 150 mm culture plates. After reaching 70 % confluence, the supernatant was harvested every 24 h and replaced with fresh medium supplemented with 10 % (v/v) FBS. The supernatants were centrifuged (600g) to remove cell debris and stored at −80 °C until further analyses. Controls were not supplemented with FBS.

The suspension culture for rBDDFVIII production was performed in Cytodex-3 (GE Healthcare) and Cultispher-G (Percell Biolytica) microcarriers cultured in 125 ml spinner flasks (Techne) following manufacturer’s instructions. Briefly, 1 × 105 cells/ml rSK-HEP cells, previously expanded in culture plates, were mixed with appropriate microcarrier concentration (3 g l−1 for Cytodex-3 and 1 g l−1 for Cultispher-G), previously hydrated, sterilized by autoclaving, and equilibrated in pre-warmed (37 °C) 10 % FBS-containing medium.

The culture was performed with only 1/3 of the final volume and with intermittent stirring (2 min stirring at 30–40 rpm every 30 min) for the first 3 h of culture. After cell attachment, the medium was added to its final volume (50 ml for Cytodex-3 and 75 ml for Cultispher-G) and the stirring was kept at 40 rpm. Medium was replaced according to the requirements of each individual experiment; as soon as the pH of the medium decreased to approx. 7.1, 50 % of its working volume it was replaced by pre-warmed fresh medium. Samples of 2 ml were taken daily for monitoring the cell density and viability, as well as the rBDDFVIII concentration.

Western blot of rBDDFVIII

Western immunoblotting of rBDDFVIII secreted by rSK-HEP cell line was performed using a 400×-reduced culture medium. A total of 0.81 mg proteins/ml was separated by 8 % SDS/PAGE and transferred to a nitrocellulose membrane. Initially the nitrocellulose membrane was incubated with blocking solution to eliminate nonspecific marker. The membrane was incubated with MAk Anti-Human FVIII-LCh (Green Mountain Antibodies, Burlington VT USA) diluted in blocking solution (1:1,000) as primary antibody. Subsequently, the membrane was incubated with anti-mouse IgG biotinylated secondary antibody diluted in blocking solution (1:1,000). The avidin conjugate (ImmunoPure Avidin; HRP conjugate) diluted in PBS (1:3,000) was incubated with the membrane and it was used to reveal the TMB peroxidase substrate kit. Commercial rBDDFVIII (Refacto) was used as standard.

rBDDFVIII in vitro stability

The in vitro stability of rBDDFVIII present in the culture supernatant was evaluated for 24 h at 37 °C and compared with a commercial plasma-derived FVIII (pdFVIII) (Hemofil M, Baxter). The biological activity of rBDDFVIII and pdFVIII was analyzed in the presence of FVIII-deficient plasma (FVIIIDP) (AMAX) in order to mimic the biological matrix, and also in its absence. To better compare the results, the residual biological activity was normalized with initial values that were 0.7; 0.76; 0.72 and 1.18 IU/ml for pdFVIII, pdFVIII + FVIIIDP, rBDDFVIII and rBDDFVIII + FVIIIDP, respectively.

Flow cytometry

rSK-HEP cells were immunophenotypically characterized by flow cytometry with vWF (DAKO A0082) or human albumin (Sigma A-0433). Cells were incubated with the antibodies according to the manufacturer’s instructions. Non-specific IgG of the corresponding class worked as negative control. Cell suspensions were analyzed on FACSort flow cytometer (Becton–Dickinson, Mountain View, CA, USA) using CellQuest software.

Immunocytochemistry assay

The non-recombinant SK-HEP-1 and rSK-HEP cells were fixed in 4 % (v/v) paraformaldehyde washed with PBS and incubated with 0.1 M glycine. Cells were permeabilized with 0.2 % Triton X-100 and then blocked with 1 % BSA and 2 % goat serum in PBS. Initially, the rSK-HEP cells were incubated with primary antibodies monoclonal mouse anti-FVIII human (QED 10104) and polyclonal rabbit anti-vWF human (DAKO A0082). Anti-mouse AlexaFluor 594 (A-21207, Molecular Probes, Invitrogen, USA) and anti-rabbit AlexaFluor 660 (A-21073, Molecular Probes, Invitrogen, USA) were used as secondary antibodies. The nucleus was stained with DAPI (VYSIS, USA). Since rSK-HEP cell was modified with lentiviral vector that carried GFP and rBDDFVIII, to avoid the superposition with the GFP fluorescence, AlexaFluor 660 was used to stain vWF in rSK-HEP and SK-HEP-1 cells and green color was chosen in confocal software as pseudocolor to indicate the vWF presence. Cells were visualized using a confocal laser scanning microscope (CLSM).

rBDDFVIII biological activity profile in hemophilia A mice

To analyze the rBDDFVIII biological activity profile, mice received 50 IU/kg of culture supernatant containing rBDDFVIII (n = 3 mice/time) or pdFVIII (n = 2 mice/time) in the tail vein. The blood samples were collected at 0.5, 2 and 6 h post-infusion by retroorbital bleeding and added at a 10:1 (v/v) ratio to acid citrate dextrose (ACD). Plasma was separated by centrifugation (9,300×g) and samples were stored at −80 °C until analysis. The biological activity profiles were determined by the average values of the FVIII biological activity at each time point.

Severe tail-bleeding model

Mice were anesthetized with ketamine 10 % (100 mg/kg) and xylazine 2 % (10 mg/kg) intraperitoneally. Bleeding was induced by cutting the tail in a region with 3 mm diameter. 10 min prior to injury, the animals received culture supernatant containing rBDDFVIII (50 IU/kg) (n = 5 mice), pdFVIII (50 IU/kg) (n = 5 mice), or no treatment (n = 5 mice) and were observed for 30 h.

Cell counts and viability

Viability and density of suspended cells were measured by the trypan blue dye exclusion method. Adhered cell concentration in Cytodex-3 culture was estimated by counting the stained nuclei with crystal violet in haemocytometer. Cell density was estimated in Cultispher-G culture after microcarrier’s digestion with dispase solution for 10 min at 37 °C.

Measurement of FVIII activity

The biological activity of in vitro and in vivo FVIII molecules (rBDDFVIII and pdFVIII) was determined by chromogenic assay (Coamatic FVIII, DiaPharma Group) following manufacturer’s instructions.

Statistical analysis

The results were expressed as mean values and standard deviation. The two-way ANOVA was used to compare differences between the indicated groups in the in vitro stability and biological activity profile in vivo. The log-rank test was used to compare the survival curves after tail-clipping challenge and p < 0.05 was considered significant.

Results

Characterization of in vitro rBDDFVIII production and stability profile

The in vitro protein stability and the kinetic expression are important for evaluating the production potential of the expression system. For an initial characterization of the kinetic expression, the rSK-HEP cells were cultured in static flasks and the biological activity assessed every 24 h, after medium exchange. Figure 1 shows that rSK-HEP cells were able to produce rBDDFVIII levels about 2.6 IU/ml in culture medium containing 10 % FBS and 1.6 IU/ml in absence of FBS supplementation. After 72 h static culture, the median cumulative production of rBDDFVIII with 10 % FBS was 82.5 ± 17.7 IU whereas in the serum-free condition it was 45.3 ± 13.1 IU (Fig. 1a). As expected, the rBDDFVIII production with 10 % FBS was 1.8-times higher than the production seen in the serum-free (SF) condition. On a cell basis, the mean rBDDFVIII production during static cultures was 2.1 IU/106 cells/24 h for serum-containing medium and 1.5 IU/106 cells/24 h for serum free medium.

Fig. 1
figure 1

In vitro rBDDFVIII production and stability. a Production of rBDDFVIII by static culture in culture medium containing 10 % (v/v) FBS and in the absence of FBS supplementation (n = 2). b In vitro stability at 37 °C of rBDDFVIII and pdFVIII molecules and also incubated in FVIII-deficient plasma (FVIIIDP) (n = 2). SF serum free medium, FBS fetal bovine serum

Full size image

The stability profile of rBDDFVIII was assessed over 24 h at 37 °C and then compared to the stability profile of a commercial pdFVIII. Both rBDDFVIII and pdFVIII were also incubated in FVIII-deficient plasma in order to mimic the biological matrix. As shown in Fig. 1b, after 24 h the rBDDFVIII maintained about 17 % of its initial biological activity whereas the pdFVIII did not present any biological activity. When FVIII samples were incubated with FVIII-deficient plasma, the remaining FVIII activity was 71 and 50 % for rBDDFVIII and pdFVIII, respectively. Note that the decay profile of the biological activity was lower in rBDDFVIII, regardless of the presence and absence of FVIII-deficient plasma.

rSK-HEP cells express von Willebrand factor

The rBDDFVIII biological activity profile at 37 °C, as previously shown, indicates the presence of a factor in the rSK-HEP cell culture that might protect rBDDFVIII from degradation. Therefore, we investigated whether these factors were the human albumin or the von Willebrand factor (vWF), FVIII’s most important stabilizers. Indeed, about 10 % of the rSK-HEP cells express vWF but not albumin, as shown in Fig. 2a. We also detected intracellular vWF co-localized with rBDDFVIII (Fig. 2b). The non-recombinant SK-HEP-1 cell line also expressed vWF (Online Resource 1). Western blot analysis demonstrated that rSK-HEP secreted FVIII protein that was correctly cleaved into a light chain (80 kDa) with only a small fraction of uncleaved rBDDFVIII protein (170 kDa) (lane 3, Fig. 2c) and this pattern was identical to the one from commercial recombinant Factor VIII (ReFacto) (lane 2, Fig. 2c). rBDDVIII was not detected in concentrated supernatants from non recombinant SK-HEP cells (lane 4, Fig. 2c).

Fig. 2
figure 2

In vitro characterization of the rSK-HEP cell line. a vWF and albumin expression by flow cytometry. b vWF and FVIII expression by confocal microscopy in the rSK-HEP cell line. c Western blot with anti-human FVIII-LCh. Std: molecular weight marker; Lane 2 Refacto 1000 I.E. Lane 3 rSK-HEP supernatant. Lane 4 non recombinant SK-HEP supernatant. In lane 2 and 3, the band above 150 kDa indicates the non-processed molecule and the band above 75 kDa indicates the light chain of rBDDFVIII with 80 kDa

Full size image

rBDDFVIII is functionally active in hemophilia A mice

The hemophilia mouse model can be used to evaluate the efficacy of FVIII molecules because it does not express FVIII. In order to investigate the rBDDFVIII activity profile in vivo, 50 IU/kg of rBDDFVIII or of pdFVIII was infused in mice. This dosage level was chosen to give an initial activity of 1 IU/ml or 100 % FVIII/ml. Figure 3a shows that after 0.5, 2, and 6 h, the mice that received rBDDFVIII presented an average FVIII activity of 0.66 ± 0.22; 0.54 ± 0.08; and 0.26 ± 0.04 IU/ml, respectively. The mice that received pdFVIII at the same time as above presented 0.42 ± 0.12, 0.29 ± 0.13, and 0 IU/ml of FVIII activity in the plasma. [Note that mice that received rBDDFVIII presented a higher half-life, around 2.6 h, than those receiving the commercial pdFVIII that presented a half-life of 0.6 h.]

Fig. 3
figure 3

In vivo characterization of rBDDFVIII potential. a Biological activity profile of rBDDFVIII (n = 3) and pdFVIII (n = 2) infused in hemophilia A mice. b Survival curve of hemophilia A mice submitted to treatment with rBDDFVIII (n = 5), pdFVIII (n = 5), or without treatment (n = 5) after challenge by tail clipping. Significance level: *p < 0.05; **p < 0.01; ***p < 0.001

Full size image

To show that our rBDDFVIII is able to correct a bleeding event in hemophilia A mice, we applied 50 IU/kg body weight of rBDDFVIII, pdFVIII, or PBS (as negative control) and then induced tail bleeding. Both the rBDDFVIII group and the pdFVIII group were successful in stopping the bleeding. Moreover, all untreated hemophilia A mice died within 30 h (Fig. 3b). The mice that received rBDDFVIII and pdFVIII were followed for 30 days and showed no complications during this period.

rSK-HEP cells grew and produced active rBDDFVIII in a scalable culture system

After verifying the in vitro and in vivo functionality of rBDDFVIII, we evaluated the scale-up potential of these cells by culturing rSK-HEP cells suspended on microcarriers in spinner flasks. The rBDDFVIII-producing cells were able to grow on suspended microcarriers for over 300 h, reaching a maximum of 1.74 × 106 cells ml−1 for Cultispher-G and 3.17 × 106 cells ml−1 for Cytodex-3 (Fig. 4a, b).

Fig. 4
figure 4

Characterization of cell growth in spinner flasks using Cultispher-G and Cytodex-3 microcarriers. a Growth of the adherent and suspension cells on Cultispher-G microcarriers and b on Cytodex-3 microcarriers. Values shown are the averages (mean ± SD) of the duplicated samples from one representative experiment

Full size image

The rBDDFVIII production profile was similar in both microcarrier cultures, as shown in Fig. 5a. The average production was 1.4 ± 0.56 IU/10e6 cells/24 h in the Cultispher-G culture and 0.42 ± 0.19 IU/106 cells/24 h in Cytodex-3. The higher specific production in Cultispher-G culture is probably due to lower adherent cell concentrations. The total cumulative production with Cytodex-3 microcarrier was 300 IU (334 h in suspension culture) and 480 IU (310 h in suspension culture) with Cultispher-G (Fig. 5b). Even after 200 h in suspension culture with Cultispher-G and Cytodex-3, rSK-HEP cells produced rBDDFVIII levels above 1.0 and 0.4 IU/106 cells/24 h, respectively.

Fig. 5
figure 5

Production of rBDDFVIII in suspension culture on Cultispher-G and Cytodex-3 microcarriers. a Biological activity and b Cumulative production. Values shown are the averages (mean ± SD) of the duplicated samples from one representative experiment

Full size image

Discussion

We have shown that rSK-HEP cells genetically modified with a lentiviral vector containing a CMV promoter produced high levels of rBDDFVIII that showed to be functional in vivo. We also showed that rSK-HEP cells grew well in a scalable system. Hepatic cell lines can produce rFVIII in high concentrations (Chen et al. 1999; Herlitschka et al. 1998; Picanco et al. 2007; Picanco-Castro et al. 2008). Indeed, the levels of rBDDFVIII produced by rSK-HEP cells in this study (1.5–2.1 IU/106 cells/24 h) were higher than those usually reported for rFVIII expression in murine cells (Dooriss et al. 2009; Herlitschka et al. 1998; Shi et al. 2003; Spencer et al. 2011) and similar to that obtained by Herlitschka et al. (1998), who found a rFVIII expression of 0.1–3.5 IU/106 cells/24 h in SK-HEP cells. This may be explained since SK-HEP cells share characteristics with hepatic cells in regard to FVIII mRNA transcription and translation (Herlitschka et al. 1998). Additionally, vWF expression in SK-HEP cells (Heffelfinger et al. 1992) may contribute to the expression levels reported here and by others (Herlitschka et al. 1998; Picanco et al. 2007). We also observed the vWF co-localized with rBDDFVIII.

It is possible that the increased stability of rBDDFVIII, as shown in Fig. 1b, is associated with the cellular presence and co-localization of the vWF. However, we have no information on the presence of the vWF in the supernatant and cannot rule out that factors other than the vWF are associated with our findings. We hypothesized that vWF may be one of the factors responsible for the rBDDFVIII stability and therefore the purification and vWF addition in the formulation of the rBDDFVIII may increase the molecule stability even more.

The rBDDFVIII had functional activity and stability after infusion in hemophilia A mice. We used pdFVIII as control and found similar activity profiles between rBDDFVIII and pdFVIII, but rBDDFVIII showed a lower biological activity decrease. The estimated half-life of rBDDFVIII in hemophilic mice was 2.6 h whereas the pdFVIII half-life was 0.6 h. A half-life of 4.1 h was found by Mordenti et al. (1996), when recombinant FVIII (Kogenate, Bayer) was infused in Balb/C mice (Mordenti et al. 1996). The in vivo functionally of rBDDFVIII was also demonstrated by mouse tail-clipping challenge. We observed that our molecule was able to correct the bleeding event.

To evaluate the scale-up potential of the rSK-HEP cells we performed the cell culture in a system that is also used at an industrial scale, microcarrier culture technology. We observed that rSK-HEP cells grew in the suspended microcarriers, reaching cell densities about 1.7 × 106 cells ml−1 for Cultispher-G microcarrier and 3.2 × 10 6 cells ml−1 for Cytodex-3 microcarrier, and remained viable for more than 300 h. We also showed that this culture system was able to produce 300–500 IU of active rBDDFVIII molecules. Even with prolonged culture periods, rBDDFVIII production remained stable.

The microcarrier culture may not be the best choice for a commercial scale production due mostly to microcarrier costs. Nevertheless, it has been shown that the high cost may be compensated by a greater production, if compared to protein production in suspension-adapted cells (Tharmalingam et al. 2008).

The Brazilian Health Ministry spends about US$ 100 million per year on blood-derived FVIII products imported from other countries. Production of rFVIII in this country may provide lower costs and greater benefits to Brazilian society as a whole and allow patients to have access to a more efficient and safer treatment for hemophilia A. The production system described in this study can be optimized to develop a new platform for commercial rFVIII production.

Conclusions

We demonstrated that rSK-HEP cells produced a high concentration of active rBDDFVIII in static and scalable culture. Moreover, the rBDDFVIII produced was functionally active in hemophilia A mice. We are motivated to work on further adaptation of rSK-HEP cells to serum-free suspension cultures in order to compare the rBDDFVIII productivity between the systems.