JOURNAL OF BIOSCIENCE AND BIOENGINEERING Vol. 103, No. 1, 50-59. 2007 DOI: 10.1263/jbb.103.50

02007, The Society for Biotechnology, Japan

Cholesterol Delivery to NSO Cells: Challenges and Solutions in Disposable Linear Low-Density Polyethylene-Based Bioreactors

Jessica Okonkowski,' Uma Balasubramanian,' Craig Seamans,' Serena Fries,' Jinyou Zhang,' Peter Salmon,' David Robinson,' and Michel Chartrain'*

Merck Research Laboratories, Bioprocess R&D, PO Box 2000, RY80Y-105, Rahway NJ 07065, USA'

Received 23 May 2006/Accepted 13 October 2006

We report the successful cultivation of cholesterol dependent NSO cells in linear low-density polyethylene (LLDPE) Wave Bioreactors when employing a low ratio of cyclodextrin to choles-terol additive mixture. While cultivation of NSO cells in Wave Bioreactors was successful when us-ing a culture medium supplemented with fetal bovine serum (FBS), cultivation with the same cul-ture medium supplemented with cholesterol-lipid concentrate (CLC), which contains lipids and synthetic cholesterol coupled with the carrier methyl-(1-cyclodextrin (m(ICD), proved to be prob-lematic. However, it was possible to cultivate NSO cells in the medium supplemented with CLC when using conventional cultivation vessels such as disposable polycarbonate shake-flasks and glass bioreactors. A series of experiments investigating the effect of the physical conditions in Wave Bioreactors (e.g., rocking rate/angle, gas delivery mode) ruled out their likely influence, while the exposure of the cells to small squares of Wave Bioreactor film resulted in a lack of growth as in the Wave Bioreactor, suggesting an interaction between the cells, the CLC, and the LLDPE contact surface. Further experiments with both cholesterol-independent and cholesterol-dependent NSO cells established that the concurrent presence of m(ICD in the culture medium and the LLDPE film was sufficient to inhibit growth for both cell types. By reducing the excess mlCD added to the culture medium, it was possible to successfully cultivate cholesterol-dependent NSO cells in Wave Bioreactors using a cholesterol-mDCD complex as the sole source of exogenous cho-lesterol. We propose that the mechanism of growth inhibition involves the extraction of choles-terol from cell membranes by the excess mICD in the medium, followed with the irreversible ad-sorption or entrapment of the cholesterol-mpCD complexes to the LLDPE surface of the Wave Bioreactor. Controlling and mitigating these negative interactions enabled the routine utilization of disposable bioreactors for the cultivation of cholesterol-dependent NSO cell lines in conjunction with an animal component-free cultivation medium.

[Key words: disposable bioreactor, NSO cell line, cholesterol, cyclodextrin, animal component - free medium]

The introduction of therapeutic monoclonal antibodies (mAbs) has profoundly impacted the therapies of several cancer types and immune disorders, translating into great benefits to patients (1-6). As of today, there are eighteen therapeutic mAbs currently registered for marketing in the US (7, 8). The production of mAbs at the industrial scale is complex and presents many technical challenges. Although many cells lines can be potentially used, the upstream pro-duction of most currently commercialized therapeutic mAbs is achieved in mammalian cells of mouse (NSO cell line) or Chinese hamster ovary (CHO cell line) origin that are culti-vated in large conventional bioreactors (9-11).

The cell line NSO originates from mouse plasmacytoma cells that have undergone several steps of cloning and selec-tion to yield immortalized non-inununoglobulin G (IgG) se-

* Corresponding author. e-mail: chartrain@merck.com phone: +1-732-594-4945 fax: +1-732-594-4400

creting B cells (12). These cells have been extensively used for the production of mAbs with fusion technology and more recently molecular biology techniques, and are currently used as the production platform for several therapeutic mAbs currently on the market (13, 14). Unlike most mammalian cell lines that down-regulate de novo cholesterol synthesis in the presence of an exogenous source of cholesterol and rapidly up regulate the pathway when cultivated in a me-dium lacking an exogenous source of cholesterol, NSO cells seldom grow in the absence of exogenous cholesterol, and are therefore routinely cultivated in the presence of choles-terol (15-18). Exogenous cholesterol can, for example, be satisfied by albumin-bound cholesterol present in the serum added to the cultivation medium (17-19). Biochemical stud-ies, based on the feeding of intermediates in the synthesis of cholesterol, have established that the NSO cholesterol require-ment was linked to a difficulty in up-regulating a 3-keto-steroid reductase responsible for the demethylation of lanos-

50

VOL. 103, 2007

CULTIVATION OF NSO CELLS IN DISPOSABLE BIOREACTORS 51

terol to C-29 sterol intermediates (20-22). Most recently, Seth et al. studied differential gene and protein expression levels in both cholesterol-dependent and independent cells (23). Using this information, they further demonstrated that cholesterol dependence was the result of an epigenetic gene silencing caused by methylation upstream of the region cod-ing for the the I 713-hydroxysteroid dehydrogenase type 7 that catalyzes the conversion of lanosterol to lathosterol (24). They further confirmed these findings by restoring choles-terol independence to NSO cells through cloning and overex-pressing the 17p-hydroxysteroid dehydrogenase type 7 (25).

Although NSO cells have been routinely cultivated using serum as the source of cholesterol, recent events linked to abnormal prions which cause bovine spongiform encephali-tis (BSE), have generated an incentive for the avoidance of animal-sourced ingredients in the manufacture of biopharma-ceuticals such as therapeutic mAbs (26). Consequently, the development of non-animal sourced cholesterol carriers such as methyl-(3-cyclodextrin (mi3CD) complexed with synthetic cholesterol, have met with success in supporting the culti-vation of NSO cells in serum-free media, and their use is be-coming standard (27, 28). Briefly, mf3CDs are ring-like cyclic oligosaccharides structures that present hydrophilic external faces and a hydrophobic interior cavity (29). These proper-ties lend mpCD and other cyclodextrin molecules the ability to encapsulate and effectively solubilize hydrophobic mole-cules in aqueous solutions. This unique property, together with a lack of toxicity, has been exploited for the delivery of water insoluble drugs, as well as the delivery of cholesterol to in vitro cultivated mammalian cells.

Recently, the introduction of disposable bioreactors has triggered interest in the mammalian cell cultivation commu-nity. These bioreactors are sterile plastic bags ranging from 2 1 to 10001 that are filled from 10% to 50% of their nomi-nal volume with cultivation medium. The head space can be filled or continuously gassed with the desired gas mixture, and mixing and gas transfer are achieved through rocking of the bags on a platform (30). The advantages of these dispos-able bioreactors include lower cost, ease of use, flexibility, and efficient aseptic transfers via tubing welding. Although these bioreactors are limited in their available volume and cannot yet replace very large bioreactors, they have the po-tential to be used for smaller scale laboratory cultivations or for inocula toward the inoculation of large conventional bioreactors (31-33).

While we routinely and successfully cultivated NSO cells in disposable bioreactors using a serum-containing medium, the cultivation of NSO cells in these bioreactors proved to be far more challenging with a chemically-defined medium con-taining cholesterol-lipid concentrate (CLC), a lipid mixture containing a cholesterol-mf3CD complex. Similar and con-current observations were made by others (34), who specu-lated that cholesterol was entrapped into ink-bottle like pores that formed at the surface of the linear low-density polyethyl-ene (LLDPE) film Supported by the use of cholesterol-inde-pendent NSO cells whose ability to grow in Wave Bioreactors (Wave Biotech LLC, Somerset, NJ, USA) can be lost when mPCDs are added to the cultivation medium, our data re-ported here place mpCD at the center of this growth inhibi-tion mechanism.

We propose that excess empty m(3CD present in the culti-vation medium extracts membrane-bound cholesterol. This is followed with the cholesterol-mPCD complexes irrevers-ibly binding to the LLDPE film of the Wave Bioreactor. The combination of these two mechanisms results in a choles-terol depleted, non-growth permissive environment. This model is consistent with known interactions of cyclodex-tins with biological membranes (35-38), and can be recon-ciled with an entrapment model (34). Based on the assump-tion that the growth inhibition results from excess empty m3CD in the cultivation medium, we have shown that careful control of the amount of mPCD used to deliver cholesterol can lead to the successful cultivation of cholesterol-depen-dent NSO cells in the presence of an LLDPE surface. This approach allows the successful cultivation of cholesterol-dependent NSO cells in LLDPE-based Wave Bioreactors using a cultivation medium free of animal-sourced compo-nents.

MATERIALS AND METHODS

Chemicals All media formulations were proprietary and pre-pared in-house (26). Dialyzed fetal bovine serum (FBS) was pur-chased from HyClone (Logan, UT, USA) and heat-inactivated at 56°C for 30 min. EX-CYTE (animal sourced lipoprotein supple-ment) was purchased from Serologicals Corporation (Norcross, GA, USA). The 250 x Cholesterol-lipid Concentrate (CLC, a non-animal sourced cholesterol-lipid concentrate, containing cholesterol bound with m3CD) was purchased from Invitrogen Corporation (Carlsbad, CA, USA). SyntheChol and mi3CD were purchased from Sigma (St. Louis, MO, USA).

Cell line The GS-NSO cell line secreting a recombinant mono-clonal antibody was obtained from Lonza Biologics (Slough, UK). This glutamine synthetase (GS) deficient mouse cell line has been developed to be used in conjunction with a plasmid carrying both genes for the GS and the recombinant protein of interest. Cultiva-tion in a glutamine-free medium allows for efficient post-transfec-tion selection of cells carrying genes for GS and the protein of in-terest (12).

Cultivation vessels Cultivation vessels were either Coming (Coming, NY, USA) vented disposable polycarbonate shake-flasks (125 ml to 500 ml) or 2 / Wave Bioreactors with contact layer com-posed of linear low-density polyethylene and extemal gas-imper-meable layer composed of LLDPE/ethyl vinyl acetate (EVA) co-polymer from Wave Biotech (Somerset, NJ, USA).

Cutivation medium and solutions All media were prepared using a non-animal sourced, low-protein Merck Proprietary Medium as the basal medium (26). In these experiments, additional com-ponents were added to the basal medium prior to sterile filtration through a 0.22 pm cellulose acetate low protein-binding filter unit (Millipore, Billerica, MA, USA).

Preparation of cholesterol and mpCD stock solutions and complexes A 10 g// synthetic cholesterol stock solution was prepared by dissolving SyntheChol powder in 100% ethanol at 37°C with subsequent filtration using a Steri-flip 0.22 p.m filter unit (Millipore), followed by storage at 4°C. A 35 g// mpCD stock solution was prepared by dissolving the mpCD powder in distilled water at room temperature with subsequent filtration using a Steri-flip 0.22 pm filter unit, followed by storage at 4°C.

The following procedure was used to prepare approximately 100 ml of the cholesterol-m(3CD complex solution at a 1:70 weight ratio of cholesterol to ml3CD (1:21 molar ratio). A volume of 4.8 ml of the 10 g// synthetic cholesterol stock solution was added to 96 ml of the 35 gll mIICD stock solution at room temperature. Cloudiness

2 3

4 5

Cultivation time (d)

2 3

Cultivation time (d)

0

5

52 OKONKOWSKI ET AL. J. Blosa. BIOENG.,

of the solution initially occurred, but with continued mixing be-came clear as the mI3CD complexed with the insoluble cholesterol.

Cultivation conditions Shake-flasks All shake-flasks were filled at a 12% working-

volume ratio (e.g., 30 ml culture in 250 ml flask) and were incu-bated in a 37°C humidified incubator containing 5% CO 2 . All shake-flasks were cultivated at an agitation rate of 75 rpm.

Seed culture Cells to be used as an inoculum source were maintained in shake-flasks throughout the experiments by diluting the culture with fresh pre-warmed medium to 1.5 x 10 3 viable cells per ml (vc/ml) or 2.0 x 10 3 vc/ml every 3 to 4 d, respectively, to maintain the cells in continuous exponential growth. Average cel-lular doubling time in shake-flasks for most media formulations was approximately 24 h, while maximum cell densities ranged from 1.0 x 106 ye/in] to 1.4 x 10 6 vc/ml. These seed cultures served as the inocula for all Wave Bioreactor cultivations.

Wave bioreactors Experiments were performed in 21 Wave Bioreactors manufactured with LLDPE using a 1 I working vol-ume. A sterile glass bottle with a weldable tubing assembly, asepti-cally filled with the required volume of fresh pre-warmed medium and seed culture to make up the 1 I working volume, was sterile-welded to the Wave Bioreactor addition port. The contents of the bottle were pumped in the Wave Bioreactor using a peristaltic pump at a rate of approximately 3 ml/s. Immediately following addition of the medium and cells, the Wave Bioreactor was inflated with sterile air from house supply or from a gas cylinder, and was sealed off with the provided clamps. The Wave Bioreactor was placed in a 37°C incubator and was rocked at 17 rpm with an 8° rocking angle unless otherwise indicated. Inoculation viable-cell densities were 2.0 x103 vc/ml unless otherwise indicated.

Shake-flask controls The Wave Bioreactor control shake-flasks were prepared for all experiments by removing a 30 ml ali-quot of Wave Bioreactor culture approximately 15 min post-inocu-lation with a sterile 30 ml syringe connected to a piece of weld-able sterile tubing. Each aliquot was transferred inside a biosafety cabinet to a sterile 250 ml Corning disposable vented shake-flask, which was incubated at 37°C and 75 rpm with 5% CO, for the du-ration of the experiment.

Wave Bioreactor coupon preparation The total surface area of the moderately inflated Wave Bioreactor film exposed to the cell culture was measured (approximately 1540 cm 2) and the ratio of exposed film to culture volume was calculated. This ratio was used to determine the area of film required for a considerably smaller shake-flask cell culture volume. To prepare the coupons, a sterile Wave Bioreactor was removed from its packaging inside a biosafety cabinet. Using sterilized forceps and scissors, 2 x 2 cm squares were cut from the Wave Bioreactor film and transferred to a sterile shake-flask. A total of 12 such squares were required for a 30 ml cell culture working volume inside a shake-flask.

Sample analysis Samples were aseptically removed from the Wave Bioreactors and from the control shake-flasks approximately every 24 h until the culture was passaged or until the viability of the cells decreased below 50%, at which point the culture was ter-minated. All samples were analyzed for pH and dissolved 0 2/CO2

concentration on a blood gas analyzer (Radiometer A/S, Bronshoj, Denmark), viable-cell concentration and viability on a Cedex auto-mated cell counter (Innovatis A4 Bielefeld, Germany), and me-tabolites (glucose, lactate and ammonia) using a Nova Bioprofile 100 (Nova Biomedical, Waltham, MA, USA).

Adaptation of NSO cells to cholesterol -independence Cells previously cultivated with CLC-containing medium were adapted to cholesterol-independence by diluting out the CLC over several direct passages using inoculation densities of 2.0 x 10 5 vc/ml and cultivating under normal conditions. For the first two to three pas-sages, the growth rate substantially declined, yet recovered by the fourth passage. After six passages, the cells were considered cho-

lesterol-independent and were ready for experimental use.

RESULTS AND DISCUSSION

Inconsistent cell growth in Wave Bioreactors In an initial experiment to explore alternate cholesterol sources to serum, the growth of a GS-NSO cell line secreting a re-combinant monoclonal antibody was evaluated in Wave Bioreactors using FBS, EX-CYTE, or CLC. Prior to inocu-lation of a Wave Bioreactor, the cells were cultivated for at least three passages in flasks with culture medium contain-ing the respective source of exogenous cholesterol to be in-vestigated. The Wave Bioreactors and the control shake-flasks were placed under their respective incubation condi-tions described in the Materials and Methods section.

Data presented in Fig. IA and 1B show that when culti-vated in Wave Bioreactors in medium supplemented with either FBS or EX-CYTE, peak cell concentrations of 1 x 10 6

FIG. 1. Cholesterol-dependent NSO cell growth (A) and viability (B) profiles when cultivated in Wave Bioreactors and shake-flasks in a medium supplemented with various cholesterol sources. Wave Bioreactor (solid circles) and control shake-flasks (empty circles) containing the cultivation medium supplemented with CLC. Wave Bioreactor (solid diamonds) and control shake-flasks (empty diamonds) containing the cultivation medium supplemented with FBS. Wave Bioreactor (solid triangles) and control shake-flasks (empty triangles) containing the cultivation medium supplemented with EX-CYTE.

6 7

16+6

2 3 4 5

Cultivation time (d)

16.2

Vol.. 103, 2007

CULTIVATION OF NSO CELLS IN DISPOSABLE BIOREACTORS 53

100

50

80

70

60

so

5 40

30

20

10

0

4 5

Cultivation time (d)

FIG 2. Growth (A) and viability (B) profiles of cholesterol-de-pendent NSO cells originating from Wave Bioreactors and transferred to shake-flasks at various time-points post-inoculation. Symbols: con-trol Wave Bioreactors (solid circles); culture removed immediately post-inoculation of the Wave Bioreactors (solid diamonds); culture removed 4 h post-inoculation of the Wave Bioreactors (solid trian-gles); culture removed 7.5 h post-inoculation of the Wave Bioreactors (solid squares); culture removed 22 h post-inoculation of the Wave Bioreactors (solid hexagons).

vc/ml and 0.7 x 106 vc/ml were achieved, respectively. In both cases, about 85-90% of the cells were viable after three days of cultivation. In contrast, the cells experienced rapid death (0% viability within 48 h post-inoculation) in the me-

dium supplemented with CLC. However, cell growth and viability in the control shake-flasks were acceptable with all three cholesterol supplements. This experiment was re-peated, and a similar outcome was observed. While these data show that, in most cases, the cells are capable of grow-ing in Wave Bioreactors, they also show that, although ac-ceptable growth can be achieved in shake-flasks when using CLC as the cholesterol carrier, growth is inhibited in the Wave Bioreactor. Taken together, these observations suggest the existence of a negative interaction between the Wave Bioreactor, the CLC-containing medium, and the cells.

In a following experiment, cells obtained from aliquots drawn from the inoculated Wave Bioreactors approximately 15 min post-inoculation and cultivated in the control shake-flasks grew well, suggesting that this negative interaction does not occur immediately upon filling and inoculation of the Wave Bioreactor, but takes place over a more extended period of time. To further understand this phenomenon, in-oculated medium aliquots obtained from a Wave Bioreactor at time points that spanned from immediately post-inocula-tion to 22 h post-inoculation were transferred and cultivated in shake-flasks. Data presented in Fig. 2A and 2B show that, when cells and medium were obtained from the Wave Bioreactor at inoculation, acceptable growth was achieved upon transfer to the shake-flask. Cells that were obtained 4 h post-inoculation, once transferred to the shake-flask along with the cultivation medium, experienced a difficult growth. Neither the maximum cell concentration (0.6 x 10 6 vc/ml ver-sus 1.0 x 106 vc/ml) nor the initial growth rate (p= —0.55 d-' versus y=0.60 d-') achieved in the shake-flasks were com-parable to those obtained with cells obtained within 15 min post inoculation. Figure 2B shows that viability reached a low point of about 25% after 2 d of cultivation. This dip in viability was followed with a slow recovery phase that led to a peak viability of about 70% after 7 d of cultivation. The transfer to shake-flasks of cells and medium that had been exposed 7.5 h and longer to the Wave Bioreactor environ-ment resulted in a rapid and complete loss of the cell viabil-ity within 3 d of cultivation. Although not immediate, the ex-posure of the cells to a Wave Bioreactor in the presence of a CLC-containing medium, irreversibly leads to a complete loss of viability.

Effect of cultivation conditions The first hypothesis investigated was that when cultivated in the presence of CLC, the cells were more sensitive to the cultivation con-ditions, especially the wave-like motion and the method of

TABLE 1. Growth of NSO cells in Wave Bioreactor under various cultivation conditions in a medium containing CLC

Condition Standard value Changed value Positive growth rate

first 48 h in Wave Bioreactor

Positive growth rate first 48 h

in control shake-flask

Gas delivery 5% CO, initial fill 0.1% CO, initial fill No Yes Gas delivery 5% CO, initial fill 5% CO, continuous flow No Yes Gas delivery 5% CO, initial fill Variable CO, continuous flow No Yes

(maintain pH 7.15) Rocking rate 17 rpm Static No Yes Rocking rate 17 rpm 4 rpm No Yes Rocking rate 17 rpm 9 rpm No Yes Rocking rate 17 rpm 25 rpm No Yes Rocking angle 8° 4° No Yes Rocking rate and angle 17 rpm, 8° 4 rpm, 4° No Yes

0 2 3

Cultivation time (d)

4 5

4 0

*

C

ims 100

so

ea

70

0-1 60

53 30

>40

30

20

10

102 . 0 4 5 2 3

Cultivation time (4)

2 3

Cultivation time (d)

9 o

0

-0

D

54 OKONKOWSKI ET AL.

gassing, used in the Wave Bioreactors. First, the gassing method and CO, concentration, which both directly affect culture pH, were evaluated. In the preliminary evaluation work described above, the inoculated 21 Wave Bioreactor was inflated to capacity with 5% CO, and all ports were clamped for the duration of the cultivation cycle. Such incu-bation conditions often result in a substantial reduction in pH after 3 or 4 d of cultivation. Three alternative gassing methods were evaluated: initial gassing with 0.1% CO, fol-lowed by sealing of the bag, continuous gassing with 5% CO,, and variable continuous gassing with CO, concentra-tion adjusted daily to maintain pH at 7.1.

J. BIOSCI.

Despite the fact that acceptable growth was achieved in the control shake-flasks incubated under standard conditions, the cells in all Wave Bioreactors experienced rapid death as observed in the past (Table 1). The stripping of CO 2 that oc-curred with the continuous gassing conditions led to higher pH values (pH 7.2-7.3) when compared to cultures that were only gassed initially (pH 7.0-7.1). It is, however, unlikely that these pH changes explain the complete cell death ob-served here, since variations of pH in this range are ob-served in successful cultivations as well. A compounding effect of higher pH values with the unknown mechanism of cell death, however, may be possible.

-

E

2 3

4

5

Cultivation time (d)

100

90

eo

70

F £50 g 40

30

20

10 F

2 3

4

Cultivation time (d)

FIG 3. Growth and viability profiles of cholesterol-independent NSO cells cultivated in shake-flasks using medium supplemented with various sources of cholesterol in the presence of Wave Bioreactor LLDPE film coupons. (A, B) Medium supplemented with CLC (solid circles). Solid lines denote the presence of coupons, dotted lines denotes the absence of coupons (control). (C, D) Medium supplemented with mi3CD (solid squares). Solid lines denote the presence of coupons, dotted lines denotes the absence of coupons (control). (E, F) Medium supplemented with cholesterol (solid triangles). Solid lines denote the presence of coupons, dotted lines denotes the absence of coupons (control).

0

CULTIVATION OF NSO CELLS IN DISPOSABLE BIOREACTORS 55 VoL. 103, 2007

In a second set of experiments, the effects of rocking rate and angle were examined to determine if cell shear or exces-sive oxidation of CLC components due to the unique rock-ing motion of the Wave Bioreactor played a role in this neg-ative interaction. Wave Bioreactors were cultivated under various rocking rates (all with an 8° rocking angle). The data presented in Table 1 show that growth was inhibited at all rocking rates investigated, including static conditions. Similar effects were observed when varying the rocking angle, and when concurrently varying the rocking angle and rocking rate. Taken together, these data suggest that exces-sive cell shear or aeration is unlikely to be the cause of the negative interaction.

Effect of chemical environment Since the effect of the physical environment alone can likely be eliminated, the only difference between conditions achieving growth and lack of growth in the Wave Bioreactors is the source of cho-lesterol used. Rapid cell death was observed when using CLC, a proprietary complex mixture of mI3CD, lipids, and cholesterol. To decouple the physical environment, especial-ly the wave-like motion, from the chemical environment, a study using coupons (cut up sections of Wave Bioreactor LLDPE film) was performed in shake-flasks. The shake-flasks containing coupons were incubated alongside control flasks that did not contain coupons.

Results in Fig. 3A and 3B show that when cholesterol-in-dependent NSO cells (see the Materials and Methods section for origin of cholesterol-independent cell line) were culti-vated in the presence of the coupons and the cholesterol-m 13CD complex, rapid cell death occurred. Conversely, a positive growth rate was achieved in the control shake-flasks. This data set points to possible interactions between the cell contact surface of the Wave Bioreactor and the culti-vation medium containing the m[CD-cholesterol complex.

To further explore these potential interactions, choles-terol-independent NSO cells were cultivated in shake-flasks in the presence of mpCD alone or cholesterol alone, with and without coupons Results presented in Fig. 3C and 3D show that a high reduction in cell viability followed by a total lack of growth was observed with NSO cholesterol-in-dependent cells in the presence of mPCD and coupons. This negative effect was however not as strong as that observed in the presence of CLC (Fig. 3A, B), as some viability and growth, albeit low, were observed under these cultivation conditions. In contrast, when cultivated in the presence of the coupons and cholesterol, the cells grew well (Fig. 3E, F). For all cases, cells cultivated in the control flasks not con-taining the coupons grew as expected.

These data indicate that the presence of coupons in the shake-flasks cannot be accountable by itself for the cell death since growth was achieved in the presence of coupons and cholesterol. When considered together, these data point to a negative interaction between the mpCD and the LLDPE sur-face of the Wave Bioreactor, resulting in unfavorable condi-tions for the growth of both cholesterol-dependent and cho-lesterol-independent NSO cells.

Confirmation of cyclodextrin-LLDPE negative inter-action in Wave Bioreactors To further confirm the hy-pothesis outlined above, we evaluated the cultivation of cholesterol-independent NSO cells in 21 Wave Bioreactors,

2 3

4

Cultivation time (d)

too

90

80

70

60

50

40

30

20

10

0

0 2 3 4 5

Cultivation time (d)

FIG. 4. Growth (A) and viability profiles (B) of cholesterol-in-dependent NSO cells cultivated in Wave Bioreactors using medium supplement with various sources of cholesterol. Medium supple-mented with mi3CD/cholesterol (solid hexagons). Solid line denote Wave Bioreactor, dotted line denote shake-flask control. Medium sup-plemented with cholesterol (solid triangles). Solid line denote Wave Bioreactor, dotted line denote shake-flask control. Medium supple-mented with mfICD (solid squares). Solid line denote Wave Bioreactor, dotted line denote shake-flask control. Un-supplemented medium (solid diamonds). Solid line denote Wave Bioreactor, dotted line de-note shake-flask control.

which were filled with medium containing m(3CD, choles-terol, cholesterol-mfCD complex, or devoid of any addi-tion. Data presented in Fig. 4A and 4B show that cells culti-vated in a Wave Bioreactor in a medium containing milCD alone or the cholesterol-mfiCD complex were not able to grow. However, for both cases, expected growth was achieved in the control flasks. The data also show that cells cultivated in the absence of any additives or in the presence of choles-terol alone grew well in Wave Bioreactors and in shake-flask controls.

The data presented in Figs. 2 to 4 support the hypothesis that the probable cause for the NSO cell death observed in the Wave Bioreactors and the coupon-containing shake-flasks is the presence of mfICD in the medium. Interaction between the LLDPE surface of the Wave Bioreactor and the cyclodextrin resulted in an environment not amenable to cell

15+4 -0 3 4

Cultivation time (d)

6 7 1e+4 0

2 3 4

Cultivation time (d)

56 OKONKOWSKI ET AL. J. BIOSCI. BIOENG.,

15+4 2 3 4 5

Cultivation time (d) 6

0

2 3 4 5

7

Cultivation time (d)

e

le+6

C O

C 0

1e+5

C

`1)

FIG. 5. Growth of cholesterol-dependent NSO cells cultivated in Wave Bioreactors using the medium supplemented with cholesterol/m(3CD at various ratios. Solid lines represent the Wave Bioreactor culture, while dotted lines represent the control shake-flask. (A) Cholesterol (3 mg//) in a 1:21 molar ratio with inflCD (solid circles). (B) Cholesterol (1.2 mg//) in a 1:0.6 molar ratio with m(3CD (solid diamonds). (C) Cholesterol (1.8 mg//) in a I :0.6 molar ratio with mi3CD (solid triangles). (D) Cholesterol (2.4 mg//) in a 1:0.6 molar ratio with ml3CD (solid squares).

growth. Since cyclodextrins are known to extract choles-terol from biological membranes (29, 35-38), it can be speculated that empty mflCD present in large excess de-pletes the NSO cells of their membrane-bound cholesterol. This phenomenon must occur in all types of cultivation ves-sels, but the reverse reaction must occur as well, resulting in an equilibrium that is conducive to the growth of choles-terol-dependent NSO cells. However, in the presence of the LLDPE film surface of the Wave Bioreactor, the choles-terol-4CD complexes may be irreversibly entrapped or immobilized, thus pulling this equilibrium toward the deple-tion of cholesterol from the cultivation medium. The mech-anism of binding, or entrapment, of the complex could be due to strong electrostatic interactions or to entrapment into ink bottle-like pores of the LLDPE film (34). The sequester-ing of the cholesterol-mpCD complex to the LLPDE film results in the creation of a cholesterol sink that translates into a non-growth permissive environment. This disruption to the membrane of the cells is so dramatic that loss of viability is experienced within a few hours post exposure to the LLDPE/m13CD environment, and even upon transfer to a growth permissive environment such as a polycarbonate shake-flask, the cells continue to lose viability. Figure 6 summarizes the proposed interactions outlined here.

Although the cultivation of NSO cells in Wave Bioreactors

can be achieved by using cholesterol-independent cells, the adaptation of new NSO transfectants producing a recombi-nant protein of interest to cholesterol independence may not be consistently achievable. Driven by the avoidance of ani-mal-sourced ingredients, understanding and potentially con-trolling the effect of mr3CD on the cultivation of cholesterol-dependent NSO cells in LLDPE-based bioreactors may there-fore present significant operational value.

Cholesterol to mflCD ratio reduction Since it is speculated that mflCD is the root cause for the growth in-hibition of NSO in LLDPE-based bioreactors, lowering the concentration of empty mi3CD by manipulating the ratio of cholesterol to mi3CD may help to reduce the negative inter-action and may help to restore growth of the cholesterol-de-pendant NSO cells in a Wave Bioreactor. Using a method previously presented (Balasubramanian et al., poster at the 229th ACS National Meeting, San Diego, CA, USA, 13 to 17 March, 2005), media solutions containing various ratios of cholesterol to mi3CD were prepared and evaluated for their potential to support the growth of cholesterol-depen-dent NSO cells in the presence of LLDPE film.

The data presented in Table 2 summarize the findings of a screening experiment using shake-flasks filled with medium containing 2 or 3 mg// of cholesterol mixed with mI3CD (con-centration refers to the amount of cholesterol added per liter

Vol.. 103, 2007 CULTIVATION OF NSO CELLS IN DISPOSABLE BIOREACTORS 57

TABLE 2. Growth of NSO cells with cholesterol-m(3CD complexes at various ratios and cholesterol concentrations in shake-flasks in the presence of LLDPE coupons

Medium Weight ratio of cholesterol to

m3CD in medium

Molar ratio of cholesterol to

m13CD in medium

Presence of Wave Bioreactor

film coupons

Positive growth rate first 48 h

in shake-flask

MPM + CLC n/a n/a Yes No MPM + 2 mg// chol-m0CD 1:70 1:21 Yes No MPM + 2 mg// chol-m8CD 1:45 1:13 Yes No MPM + 2 mg// chol-m3CD 1:25 1:7.3 Yes No MPM + 3 mg// chol-m(3CD 1:2 1:0.6 Yes Yes MPM + 3 mg// chol-m3CD 1:2 1:0.6 No Yes

of medium) at various ratios in the presence of LLDPE cou-pons. The data show that molar ratios of cholesterol to mf3CD ranging from 1:21 to 1:7.3 led to rapid cell death, presum-ably due to an excess of mf3CD. A mixture of 3 mg/I of cho-lesterol with mr3CD at a molar ratio of 1:0.6 did, however, adequately support growth similar to that achieved in the control shake-flask.

To further demonstrate the effective use of the lower 1:0.6 molar ratio of cholesterol to mf3CD, as well as to determine the optimal final concentration of cholesterol-mllCD com-plex, cells were cultivated in Wave Bioreactors with three cholesterol concentrations at a cholesterol to mf3CD ratio of 1:0.6. In addition, a control Wave Bioreactor with the origi-nal high molar ratio of cholesterol to ml3CD (1:21) was also inoculated with cholesterol-dependent NSO cells. Figure 5 shows that, as expected, a high 1:21 cholesterol to mf3CD ratio resulted in rapid cell death in the Wave Bioreactors. However, the lower cholesterol to mf3CD molar ratio of 1:0.6, when added at cholesterol concentrations ranging from 1.2 mg// to 2.4 mg//, supported growth of the choles-terol-dependent NSO cells in the Wave Bioreactors. A 1:0.6

FIG. 6. Proposed interactions between the cholesterol-m(3CD com-plex; NSO cells, and LLDPE film surface of a Wave Bioreactor. The interaction of the LLDPE wave plastic surface, the cholesterol/m(3CD, and the NSO cells results in an environment impermissible to cell growth. The empty milCD is capable of removing the membrane-bound cholesterol (solid triangle) with the cholesterol/m3CD irrevers-ibly binding or being entrapped with/to the LLDPE plastic film. The net result of these interactions is the total depletion of cholesterol from both the cultivation medium and the membrane of the cells.

molar ratio of cholesterol to mf3CD corresponds to an ex-cess of cholesterol, thus further supporting the hypothesis that excess mf3CD is the root cause for this apparent toxic-ity. The data also show that higher concentrations of choles-terol (2.4 mg/I) resulted in better growth, with a peak cell density 1.6 x 10' vc/ml, a value that compares favorably with the CLC- and FBS-containing media peak cell densities of 1.2 x 106 vc/ml achieved in shake-flasks. This scale-up ex-periment demonstrates that once mechanistically understood, the negative interaction of mf3CD and LLDPE surface of Wave Bioreactors can be controlled to support the success-ful cultivation of cholesterol-dependent NSO cells in these disposable cultivation vessels.

Conclusion While the cultivation of cholesterol-de-pendent NSO cells in LLDPE-based Wave Bioreactors is simple to achieve when using serum as a source of choles-terol, achieving similar growth in the presence of the syn-thetic cholesterol carrier mf3CD proved to be problematic. Rapid cell death was observed when cultivating NSO cells in a Wave Bioreactor using CLC, a lipid mixture that con-tains mf3CD as the cholesterol carrier.

After ruling out the influence of the Wave Bioreactor physical environment and evaluating the growth of choles-terol-independent NSO cells, we were able to demonstrate that the co-presence of mf3CD and the LLDPE film surface of the Wave Bioreactor is sufficient to create an environ-ment unsuitable to the growth of the cells. We speculate that since mf3CD is in large excess in the CLC formulation and has a very high affmity for cholesterol, the mf3CD mole-cules remove membrane-bound cholesterol from the cells. Although a similar extraction of cholesterol from the mem-branes of NSO cells by m /3CD likely occurs in all types of cultivation vessels, re-delivery of the cholesterol to the mem-branes must also occur at a similar or greater rate, therefore resulting in a growth permissive environment. We further speculate that the cholesterol-mf3CD complexes present a high affinity for the LLDPE Wave Bioreactor film, and are irreversibly bound.

This problem could be mitigated via several options. First, the use of a cholesterol-independent clone, thus eliminating the need for mf3CD, could alleviate this problem, although obtaining such a clone is reported to be unpredictable (18). Second, the use of a cell line genetically engineered to over-express the enzyme needed for cholesterol synthesis, Hsd 17b7, has been recently proposed as an alternative to using cholesterol-dependent cells (25). Third, the use of other plastic-based film surfaces (34), is a good practical solution, however the use of LLDPE presents advantages in terms of

58 OKONKOWSKI ET AL. J. BIOSCI. BIOENG.,

mouse myeloma cells and its application to the isolation of Ns-1 hybridomas. Anal. Biochem., 130, 445-453 (1983). Keen, M. and Steward, T.: Adaptation of cholesterol-re-quiring NSO mouse myeloma cells to high density growth in a fully defined protein-free and cholesterol-free culture medium. Cytotechnology, 17, 203-211 (1995). Birch, J., Boraston, R., Metcalfe, H., Brown, M., Bebbington, C., and Field, It: Selecting and designing cell lines for improved physiological characteristics. Cytotechnology, 15, 11-16(1994). Chen, J.-K., Okamoto, J., Sato, J., Sato, G., and McClure, El.: Biochemical characterization of the cholesterol-depen-dent growth of the NS-1 mouse myeloma cell line. Exp. Cell Res., 163, 117-126 (1986). Chen, J.-K. and Li, L.: Sterol depletion reduces receptor-mediated low-density lipoprotein binding in NS-1 mouse my-eloma cells. Exp. Cell Res., 117, 76-85 (1987). Sato, J., Cao, H-T., Kayada, Y., Cabot, M., Sato, G., Okamoto, T., and Welsh, C.: Effects of proximate choles-terol precursors and steroid hormones on mouse myeloma growth in serum-free medium. In Vitro Cell. Dev. Biol., 24, 1223-1228 (1988). Seth, G., Philp, R., Denoya, C., McGrath, K., Stutzman-Engwall, K., Yap, M., and Hu, W-S.: Large-scale gene ex-pression analysis of cholesterol dependence in NSO cells. Biotechnol. Bioeng., 90, 552-567 (2005). Seth, G., Ozturk, M., and Hu, W-S.: Reverting cholesterol auxotrophy of NSO cells by altering epigenetic gene silenc-ing. Biotechnol. Bioeng., 93, 820-827 (2006). Seth, G., Mclvor, S., and Hu, W-S.: 17-13-Hydroxysteroid dehydrogenase type 7 (Hsdl 767) reverts cholesterol auxotro-phy in NSO cells. J. Biotechnol., 121, 241-252 (2006). Zhang, J. and Robinson, D.: Development of animal-free, protein-free and chemically-defined media for NSO cell cul-ture. Cytotechnology, 48, 59-74 (2005). Walowitz, J., Fike, IL, and Jayme, D.: Efficient lipid deliv-ery to hybridoma culture by use of cyclodextrin in a novel granulated dry-form medium technology. Biotechnol. Prog., 19, 64-68 (2003). Gorfien, S., Paul, B., Walowitz, J., Keem, IL, Biddle, W., and James, D.: Growth of NSO cells in protein-free, chemi-cally defined medium. Biotechnol. Prog., 16, 682-687 (2000). Christian, A., Haynes, P., Philips, M., and Rothblat, G.: Use of cyclodextrins for manipulating cellular cholesterol con-tent. J. Lipid Res., 38, 2264-2272 (1997). Singh, V.: Disposable bioreactor for cell culture using wave-induced agitation. Cytotechnology, 30, 140-158 (1999). Hardy, J. and Priester, P.: Considerations for use of dispos-able technology in contract manufacturing. Bioprocess Int., 2(suppL 4), 32-55 (2004). Fries, S., Glazomitsky, K., Woods, A., Forrest, G., Hsu, A., Olewinski, It, Robinson, D., and Chartrain, M.: Evalua-tion of disposable bioreactors: rapid production of recombi-nant proteins by several animal cell lines. Bioprocess Int., 3(suppl. 6), 36-44 (2005). Pierce, L. and Shabram, P.: Scalability of a disposable bio-reactor form 25L-500L run in perfusion mode with a CHO-based cell line: a tech review. Bioprocessing, 2(suppl. 4), 51-56 (2004). Kadarusman, J., Bhatia, IL, McLaughlin, J., and Lin, W.: Growing cholesterol-dependent NSO myeloma cell line in the wave bioreactor system: overcoming cholesterol-polymer in-teraction by using pretreated polymer or inert fluorinated eth-ylene propylene. Biotechnol. Prog., 21, 1341-1346 (2005). Senin, I., Hoppner-Heitmanns, D., Polkovnikova, 0., Churumova, V., Tildiomirova, N., Philippov, P., and Koch, K-W.: Recoverin and rhodopsin kinase activity in detergent-resistant membrane rafts rod outer segments. J. Biol. Chem.,

ease of manufacture of the bags and low cost. We believe that for the short term, however, reducing the amount of

nifiCD added to the cultivation medium by controlling the 18.

cholesterol to mr3CD ratio may prove to be a widely appli- cable and effective method. For our case, this solution was

proven to successfully support the cultivation of cholesterol- 19. dependent NSO cells in LLDPE-based Wave Bioreactors in a medium free of animal components. Such an approach should be applicable to all cholesterol-requiring de novo

transfectants, and should therefore enable the implementa- 20.

tion of streamlined processes based on the use of disposable bioreactors in conjunction with animal-component free cul- tivation media. 21.

REFERENCES

22.

1. Andreakos, E., Taylor, P., and Feldmann, M.: Monoclonal antibodies in immune and inflammatory diseases. Curr. Opin. Biotechnol., 13, 615-620 (2002).

2. Hinoda, Y., Sasaki, S., Ishida, T., and Imai, K.: Mono- 23. clonal antibodies as effective therapeutic agents for solid tu-mors. Cancer Sci., 95, 621-625 (2004).

3. Harris, IL, Shire, S., and Winter, C.: Commercial manu-

facturing scale formulation and analytical characterization of 24. therapeutic recombinant antibodies. Drug Dev. Res., 61, 137-154 (2004).

4. Carter, P.: Improving the efficacy of antibody-based cancer 25. therapies. Nat. Rev., 1, 118-129 (2001).

5. Roskos, L., Davis, C., and Schwab, G.: The clinical phar-

macology of therapeutic monoclonal antibodies. Drug Dev. 26. Res., 1, 108-120 (2004).

6. Souriau, C. and Hudson, P.: Recombinant antibodies for

cancer diagnosis and therapy. Expert Opin. Biol. Then, 3, 27. 305-318 (2003).

7. Reichert, J. and Pavlou, A.: Monoclonal antibody market. Nat. Rev., 3, 383-384 (2004).

8. Reichert, J., Rosensweig, C., Faden, L., and Dewitz, M.: 28. Monoclonal antibody successes in the clinic. Nat. Biotechnol., 23, 1073-1078 (2005).

9. Andersen, D. and Reilly, D.: Production technologies for 29. monoclonal antibodies and their fragments. Curr. Opin. Biotechnol., 15, 456-462 (2004).

10. Wurm, F.: Production of recombinant protein therapeutics in 30. cultivated mammalian cells. Nat. Biotechnol., 22, 1393-1398 (2004). 31.

11. Werner, IL, Noe, W., Kopp, K., and Schluter, M.: Appro-priate mammalian expression systems for biopharmaceuticals. Drug Res., 48, 870-880 (1998). 32.

12. Barnes, L., Bentley, C., and Dickson, A.: Advances in ani-mal cell recombinant protein production: GS-NSO expression system. Cytotechnology, 32, 109-123 (2000).

13. Whitford, W.: NSO serum-free culture and applications. Bioprocess Int., 1(12), 36-47 (2003). 33.

14. Chu, L. and Robinson, D.: Industrial choices for protein production by large-scale cell culture. Curr. Opin. Biotechnol., 12, 180-187 (2001).

15. Sato, J.: Cholesterol requirement of NS-1 mouse myeloma 34. cells for growth in serum-free medium. Mol. Biol. Med., 2, 121-134 (1984).

16. Sato, J., Kawamoto, T., and Okamoto, T.: Cholesterol re-quirement of P3-X63-Ag8 and X63-Ag8.653 mouse myeloma

cells for growth in vitro. J. Exp. Med., 165, 1761-1766 35.

(1987). 17. Kawamoto, T., Sato, J., Le, A., McClure, D., and Sato, G.:

Development of a serum-free medium for growth of NS-1

Vol,. 103, 2007

CULTIVATION OF NS0 CELLS IN DISPOSABLE BIOREACTORS 59

47, 48647-48653 (2004). 36. Bartel, S., Diehl, H., and Ojeda, F.: Methyl-beta-cyclo-

dextrins and liposomes as water-soluble carriers for choles-terol incorporation into membranes and its evaluation by a microenzymatic fluorescence assay and membrane fluidity-sensitive dyes. Anal. Biochem., 258, 277-284 (1998).

37. Larbi, A., Douziech, N., Khalil, A., Dupuis, G., Gherairi,

S., Guerard, K-P., and Fulop, T., Jr.: Effects of methyl-beta-cyclodextrin on T lymphocytes lipid rafts with aging. Exp. Gerontol., 39, 551-558 (2004).

38. Launikonis, B. and Stephenson, G.: Effects of membrane cholesterol manipulation on excitation-contraction coupling in skeletal muscle of toad. J. Physiol., 534 Pt. 1, 71-85 (2001).