Formulation and production of a blood‐free and chemically defined virus production media for VERO cells

Abstract Vaccines provide effective protection against many infectious diseases as well as therapeutics for select pathologies, such as cancer. Many viral vaccines require amplification of virus in cell cultures during manufacture. Traditionally, cell cultures, such as VERO, have been used for virus production in bovine serum‐containing culture media. However, due to concerns of potential adventitious agents present in fetal bovine serum (FBS), regulatory agencies suggest avoiding the use of bovine serum in vaccine production. Current serum‐free media suitable for VERO‐based virus production contains high concentrations of undefined plant hydrolysates. Although these media have been extensively used, the lack of chemical definition has the potential to adversely affect cell growth kinetics and subsequent virus production. As plant hydrolysates are made from plant raw materials, performance variations could be significant among different lots of production. We developed a chemically defined, serum‐free medium, OptiVERO, which was optimized specifically for VERO cells. VERO cell growth kinetics were demonstrated to be equivalent to EMEM‐10% FBS in this chemically defined medium while the plant hydrolysate‐containing medium demonstrated a slower doubling time in both two‐dimensional (2D) and 3D cultures. Virus production comparisons demonstrated that the chemically defined OptiVERO medium performed at least as good as the EMEM‐10%FBS and better than the plant hydrolysate‐containing media. We report the success in using recombinant proteins to replace undefined plant hydrolysates to formulate a chemically defined medium that can efficiently support VERO cell expansion and virus production.

have witnessed a major shift in the paradigm of viral vaccine manufacturing from more primitive manufacturing platforms, such as embryonated chicken eggs and primary and diploid cell culture systems, to more robust large-scale methods that utilize mammalian continuous cell line (CCL)-based systems (Hegde, 2015). VERO (African monkey kidney epithelial) cells have been utilized for the manufacturing of human vaccines for nearly 40 years and were at one point, the only CCL approved by the World Health Organization (Barrett, Mundt, Kristner, & Howard, 2009). As a result, VERO is currently the most widely utilized and best characterized CCL for vaccine production (Hegde, 2015).
VERO cells have a multitude of inherent benefits to serve as a viral production cell line, including the inability to express interferonα that can hinder viral replication, low doubling time, and relative ease of use (Desmyter, Melnick, & Rawls, 1968). VERO cells grow readily in Dulbecco's modified Eagle's medium (DMEM) or Eagle's minimum essential medium (EMEM) containing 10% fetal bovine serum (FBS; EMEM-10% FBS) and have been documented to produce a multitude of viruses for vaccine production at commercially viable levels (Barrett et al., 2009;Desmyter et al., 1968;Govorkova, Kodihali, Alymova, Fanget, & Webster, 1999). However, due to safety concerns, current guidance by regulatory bodies in addition to Good Manufacturing Practice compliant manufacturing of cell-based products (advanced therapy medicinal products) states that alternatives to FBS should be utilized due to poor reproducibility and safety concerns associated with FBS use (EMA, 2011;van der Valk et al., 2017). As a result, several serum-free media have been described for VERO-based virus production and are currently available to the market (Merten, Kierulff, Castignolles, & Perrin, 1994;Price, 1997Price, , 2002. These media have demonstrated excellent ability to support VERO cell growth and virus production (Price, 2002).
VERO cells can be readily adapted to serum-free conditions but remain dependent on the major protein components in serum for successful propagation. Serum albumin provides multiple functions in vitro, including antioxidant, macromolecule delivery, and free radical scavenger functions (Lee & Wu, 2015;Taverna, Maria, Mira, & Guidet, 2013). Serum transferrin delivers iron to cells via catalytic cycling (Steere et al., 2012). Cytokines deliver the necessary mitogenic signaling to maintain the proliferation rate and promote survival of VERO cells. Some commercially available serum-free media for VERO cells, such as VP-SFM, contain very low levels of protein (typically 10 ng/ml or less) that includes recombinant human epidermal growth factor (EGF) as the only full-length protein (Life Technologies, 1999).
The EGF cytokine appears to be important for VERO expansion (Life Technologies, 1999). Furthermore, iron chelates can be utilized as an efficient substitute for serum transferrin, and albumin can be replaced with plant-derived di-and tripeptides at a final concentration of 100-200 mg/L (Life Technologies, 1999). As a result, these plant hydrolysate-containing media are void of any animal components and are thus considered to be animal free.
The ability of these media to support VERO cell proliferation is mostly dependent on the presence of these plant-derived hydrolysates as withdrawal of this complex mixture results in cessation of cell proliferation (Mizrahi & Shahar, 1977). It is thought that protein hydrolysates act as a concentrated, balanced nutrient mixture that can at least in part replace serum (Jan, Jones, Emery, & al-Rubeai, 1994). However, current understanding of these complex mixtures cannot unequivocally provide evidence that beneficial activity resides in a small fraction of the mixture or whether the beneficial activity observed is mostly derived from interactions of individual components in the mixture (Franek, Hohenwarter, & Katinger, 2000). Indeed, attempts to separate peptide fractions via chromatography identified fractions with significantly variable growth-promoting activities in mammalian cell culture (Franek et al., 2000). Thus, media containing these undefined complex mixtures can complicate manufacturing efforts due to potential variable performance among different lots of the mixture made from poorly controlled raw material.
To circumvent the lack of chemical definition in virus production media, we have utilized recombinant human transferrin and recombinant human albumin to formulate a chemically defined serumfree media that is optimized specifically for production of virus in VERO cells in both two-dimensional (2D) and 3D culture formats.
The efficacy of this chemically defined media, known as OptiVERO, is demonstrated in both the 2D and 3D expansion as well as the production of three different viruses in comparison to both serum and plant hydrolysate-containing media. Overall, the data presented here provide proof of principle that poorly defined plant hydrolysates can be replaced with recombinant proteins to formulate a chemically defined, serum-free virus production media suitable for clinical application.  Burman-based designs, data were analyzed via linear regression and significance of means was determined using a t test. For central composite designs, polynomials to model the second-order response surface were generated and subsequently verified by the same experimental method described above.

| Two-dimensional multiple cell-passage expansion
Early passage VERO cells maintained in EMEM-10% FBS were extensively washed with DPBS to remove all traces of serum and subsequently removed from the growth surface using 0.05 ml/cm 2 TrypLE containing 1 mM EDTA. Equal volume of 0.25% soybean trypsin was added, and cells were collected with the addition of 0.2 ml/cm 2 DPBS. Cells were centrifuged at 200g for 5 min and resuspended in a minimal volume of prewarmed OptiVERO SFM (our chemically defined medium). Cells were seeded in T-75 flasks containing prewarmed EMEM-10% FBS, VP-SFM containing 4 mM L-glutamine, or OptiVERO SFM at an initial cell density of 10,000 cells/cm 2 . VERO cells were subcultured every 3-5 days until a total of 14 serial passages was attained. At each passage, population doublings were determined in each media formulation. Cumulative population doublings were determined and presented.
2.5 | Microcarrier-based (3D) cell expansion VERO cells were adapted to VP-SFM with 4 mM L-glutamine or the chemically defined media for at least three passages in T-flasks before 3D expansion on microcarriers. Briefly, cells were expanded in multiple T-150 flasks to achieve sufficient cells for microcarrier seeding. The day before seeding, untreated plastic microcarriers (Corning) in cell culture grade water (1,000 mg/ml; Sigma-Aldrich) and glass spinner flasks (Corning) were sterilized and allowed to cool overnight. On the day of cell inoculation, sterilized microcarriers were washed three times in EMEM to remove all traces of water.
Cold VP-SFM, EMEM-10% FBS, or our developed OptiVERO SFM was used on a final wash before seeding the microcarriers in the spinner flask containing respective medium. Microcarriers (4,200 mg) were added per spinner flask to generate a 10 cm 2 /ml microcarrier suspension for a total of 1,500 cm 2 per spinner flask. The spinner flasks were allowed to equilibrate at 37°C for a minimum of 3 hr before cell inoculation. VERO cells were trypsinized and collected from the T-150 flasks using standard methods as described above.
Cells were resuspended in 5 ml of each testing medium, passed through a 0.7 µm nylon cell strainer (Corning), and counted. Cells were adjusted to 1 × 10 6 viable cells/ml in each prewarmed growth medium and 15 ml of each cell suspension was added to the equilibrated spinner. Spinners were returned to the stir plate set at 40 rpm in a 37°C 5% CO 2 incubator for 48 hr. Starting at Day 2, spinners were monitored every day for cell growth, culture glucose, and pH.
Cultures with glucose levels below 2 g/L were adjusted back up to 2 g/L using 45% glucose (Sigma-Aldrich) and pH was restored to neutral with 7.5% sodium bicarbonate (Thermo Fisher Scientific). Cell growth was determined by treating the microcarriers in a lysis buffer solution containing 10 mM Sodium Citrate (Sigma-Aldrich) and 1% Triton X-100 (Sigma-Aldrich). Nuclei were harvested, stained (Viacount nuclear stain EMD Millipore), and counted on a pregated flow cytometer (Guava; EMD Millipore). Spinner runs were terminated when glucose consumption ceased (about 9 days). | 3279 the green fluorescent protein reporter gene, instead of the VP30 gene in the viral genome, was generated as previously described (Halfmann et al., 2008)

| Virus titration
DENV-2 and ZIKV titration to measure plaque forming unit was performed in 6-well plates of just confluent VERO cells as described previously (Huang et al., 2000). Briefly, virus samples were thawed on ice and subjected to 10-fold serial dilutions in basal DMEM media.
Growth media was removed from the confluent VERO monolayers and 100 µl of diluted virus was added per well. Inoculated cell plates were incubated at 37°C for 2 hr with gentle rocking of the plates every 15 min. Cells were overlayed with 4 ml overlay medium containing 0.8% SeaKem LE agarose. Plates were stored upside down at 37°C CO 2 incubator for 4-7 days (DENV-2 16681) or 4-5 days (ZIKV PRVABC 59) before adding a second overlay (2 ml/well) containing 4.8% of 1:300 neutral red solution (MP Biomedicals, Santa Ana, CA).
Plates were incubated for an additional 24-48 hr and virus plaques were counted for three consecutive days. For EbolaΔVP30 viruses, VERO-VP30 cells were plated in 12-well plates in MEM + 10% FBS and allowed to grow to confluence. On the day of infection, virus samples were thawed on ice and subjected to 10-fold dilutions in basal EMEM. Growth medium was removed from the confluent VERO-VP30 monolayer and 100 µl of diluted virus was added per well. Cells were incubated at 37°C for 60 min with gentle rocking of the plates every 15 min. Cells were washed twice with PBS, 1.5 ml of 1.25% methylcellulose media overlay [18] was added to each well, and cells were incubated at 37°C for 1 week. After fixation with 10% neutral formalin, plaques were visualized as previously described (Halfmann et al., 2008).

| Statistics
One-way analysis of variance with 95% confidence interval was utilized to determine the significance of differences in means between different media.

| Formulation of the animal component-free chemically defined virus production media
To formulate the chemically defined media, 37 different components (Figure 1a), including the recombinant albumin (Cellastim S) and recombinant transferrin (Optiferrin), were first titrated against VERO cells in basal DMEM/F12 to identify approximate concentration ranges that would be optimal for VERO cells. Plackett-Burman matrix designs were then utilized to screen novel formulations within the optimal concentration range of each component and linear regression analysis was performed. In total, two separate screens were performed in succession that were aimed at removing biologically inert or detrimental compounds using VERO cell growth as the readout (Figure 1a). A representative screen is shown in Figure 1b

| VERO cell expansion and morphology in OptiVERO
The serum-free OptiVERO medium was subjected to a 14-passage OptiVERO spinner were all significantly higher from 2 to 9 days after cell seeding (Figure 3).

| Virus production in VERO cells cultured with OptiVERO
Two different flaviviruses, DENV-2 16681 and ZIKV, and EbolaΔVP30 were used as proof of concept to determine the efficiency of OptiVERO in supporting virus production from VERO cells.
VERO cells cultured in OptiVERO were highly competent in the F I G U R E 1 Formulation of the VERO-optimized ACF media. Medium components (37 total) were first screened in basal DMEM/F12 to identify approximate optimal concentration ranges. Plackett-Burman matrix designs were then employed to determine optimal reagent mixtures in these identified concentration ranges using VERO cell proliferation in 48-well plates as the readout. (a) Linear regression analysis was performed on the 37 different components in Phase 1 using two separate Plackett-Burman experiments. Components 11 and 22 were identified as significant drivers of VERO cell proliferation with a linear regression slope of 1.85 and 2.04, respectively (p ≤ .01 and .001). Component 20 induced a strong yet insignificant induction of proliferation in VERO cells. Components 24-26 induced a significant inhibition of VERO cell proliferation as evidenced by highly negative slopes in linear regression analysis. (b) Representative data from Phase 1 Plackett-Burman matrix screening experiments. (c) For Phase 2, reagents from the 37 originally screened components were added and removed, and remaining component concentrations identified as significant in Phase 1 were optimized via steepest ascent/descent approaches. F I G U R E 3 Growth kinetics of VERO cells expanded in OptiVERO in 3D culture. VERO cells were adapted to each tested medium for at least three passages in T-flasks before microcarrier runs. Cells were seeded at an initial density of 10,000 cells/cm 2 at Day 0 in 150 ml spinners. At Day 2 postseeding, spinners were monitored every day for cell growth, culture glucose, and pH. When necessary, cultures were fed with 45% glucose and 7.5% sodium bicarbonate to maintain the culture glucose at 2 g/L and pH neutral. VERO cells cultured in OptiVERO readily attached to the microcarriers and demonstrated a significantly higher cell expansion than either EMEM-10% FBS (***) or VP-SFM ( ### ) as determined by one-way ANOVA. Data are a representative run of at least four independent experiments. 3D, three-dimensional; ANOVA, analysis of variance; EMEM, Eagle's minimum essential medium; FBS, fetal bovine serum significant issues with both a lack of reproducibility due to variation of composition as well as reliability of supply (van der Valk et al., 2017). It has been shown that different geographical and manufacturing lots of FBS can exhibit as large as 10-fold difference in known components, in addition to the absence or presence of undefined substances (Baker, 2016a(Baker, , 2016b. Growing demand for FBS across different areas of the biotech industry has put a severe strain on supply and resulted in a drastic price increase over the past decade. These conditions in combination with slack regulations are fertile grounds to encourage unethical practices and even blatant misconduct. Indeed, adulteration and fraud have been well noted in the serum industry. In 1994, 30,000 L of FBS were documented to be sourced from New Zealand; however, the country itself only reported a 15,000 L annual production (van der Valk et al., 2017). A second major adulteration case in 2013 was documented with the dilution of FBS with bovine albumin, water, and other growth-promoting substances by a major global supplier (Gstraunthaler, Lindl, & van der Valk, 2014). Investigations by the Food and Drug Administration concluded that ∼280,000 L of FBS were affected between 2008 and 2013 (Gstraunthaler et al., 2014). In addition, animal products could potentially contain undesirable or dangerous adventitious agents, which is a safety concern of using FBS in vaccine manufacturing.
Although the issues with FBS usage, especially in the context of manufacturing of the clinical product, are well known, the component proteins that FBS contains are critical drivers of VERO cell growth, survival, and ultimately virus production. Serum albumin, one of the most abundant proteins in plasma of vertebrates, has multiple roles in several different biological processes (Lee & Wu, 2015). Likewise, serum transferrin can reversibly bind Fe 3+ with nanomolar affinity and is one of the major vehicles for iron delivery to cells (Steere et al., 2012;Zhang et al., 2012). Thus, to have successful propagation of these cells in vitro, it is critical to incorporate proteins or compounds that can effectively substitute these bovine serum proteins.
Human cell propagation has utilized human serum-purified components to substitute for these bovine proteins and has fueled the xeno-free marketing revolution in serum-free media (Oikonomopoulous et al., 2015). However, use of protein derived from human serum does not mitigate the risk inherent in the variability of the serum source, limited supply availability, and risk of adventitious agent contamination. In addition, VERO cells are not of human origin Plant-derived di-and tripeptides offer another cost-effective and technically feasible solution to replace these bovine or human serumderived proteins. Serum was shown to be effectively reduced to ∼20% of the original incorporation level by use of vegetable proteins in BHK cells, though full serum withdrawal demonstrated a significant reduction in cell expansion capacity (Mizrahi & Shahar, 1977). Further, peptone (0.5-2% wt/vol) was noted to effectively propagate multiple cell types in multiple media formulations (Taylor & Parshad, 1977). However, these mixtures are extremely complex in their composition and the mechanistic contribution to produce the observed positive effects in mammalian cell culture media is poorly understood (Franek et al., 2000). Fractionation attempts via low-pressure liquid chromatography of both soy protein and wheat gluten demonstrated a significant spread of growth promotion and cell productivity using a murine hybridoma model (Taylor & Parshad, 1977). Furthermore, the inclusion of these complex ingredients in cell culture media can add unpredictability, such as the presence of unknown compounds that can potentially interfere with certain virus productivity.
Here, the protein components of FBS have been replaced with recombinant human serum albumin and recombinant human serum transferrin. These recombinant proteins were expressed in a nonmammalian expression platform that exhibits robust expression and exceptional scalability, and have been proven to be equivalent to their native counterparts both biochemically and functionally . Recombinant transferrin (Optiferrin) was shown to possess the correct sequence to serum-derived transferrin as determined by amino N-terminal sequencing and the correct molecular weight as determined by Matrix-Assisted Laser Desorption Ionization analysis (Steere et al., 2012;Zhang et al., 2012). Furthermore, the recombinant transferrin was able to both sequester and release iron as determined by in vitro iron-binding and exhibited equivalent growth-promoting activities of a murine hybridoma cell line as native human transferrin (Steere et al., 2012). Likewise, recombinant albumin (Cellastim S) has demonstrated equivalent biochemical and functional characteristics to the native version (Pennybaker & Alfano, 2018). Reverse-phase analytical chromatography (RP-UPLC) and liquid chromatography-mass spectrometry (LC-MS) demonstrated that the recombinant albumin contained the unmodified albumin species (with unmodified Cys-34) and modified albumin species (with cysteinylated Cys-34). These species were also present in human native albumin samples from commercial vendors (Pennybaker & Alfano, 2018). However, intact LC-MS analysis and RP-UPLC analysis demonstrated substantial variability in the amount of unmodified and modified species in the serum-derived albumins.
The ratio between unmodified and modified albumin species was shown to be highly conserved from lot to lot of recombinant albumin (Pennybaker & Alfano, 2018). In addition, the recombinant albumin was able to effectively expand primary human T cells that were comparable to native commercial preparations (Pennybaker & Alfano, 2018). The use of these recombinant proteins in cell culture virtually eliminates the risk of animal origin adventitious agent contamination, providing a significant advancement for the future development of the industry.
In conclusion, two highly scalable recombinant proteins, Cellastim S and Optiferrin, were utilized to replace the bovine components provided by serum to develop serum-free medium candidates for VERO cell culture. Formulations were subsequently designed around these two components in addition to recombinant human EGF, to provide adequate nutritional and mitogenic support specifically for VERO cell expansion. Further optimization work of both the protein and nonprotein components within the formulation produced a chemically defined blood component-free media, OptiVERO. This media supports rapid VERO proliferation in both 2D and 3D culture formats with a significantly lower doubling time than VP-SFM. Furthermore, this media supported robust viral replication in VERO cell cultures for multiple viruses. The chemically defined OptiVERO can enhance the safety and consistency of virus production during manufacturing leading to improved virus-based vaccines and therapeutics.