Comparison of a Production Process in a Membrane-Aerated StirredTank and up to 1000-L Airlift Bioreactors Using BHK-21 Cells andChemically Defined Protein-Free Medium

Friedemann Hesse,† Maria Ebel, Nadine Konisch, Reinhard Sterlinski,Wolfgang Kessler, and Roland Wagner*

German Research Centre for Biotechnology (GBF), Mascheroder Weg 1, D-38124 Braunschweig, Germany

The applicability of a protein-free medium for the production of recombinant humaninterleukin-2 with baby hamster kidney cells in airlift bioreactors was investigated.For this purpose, a BHK-21 cell line, adapted to grow and produce in protein-freeSMIF7 medium without forming spheroids in membrane-aerated bubble-free biore-actors, was used as the producer cell line. First, cultivation of the cells was establishedat a 20-L scale using an internal loop airlift bioreactor system. During the culturingprocess the medium formulation was optimized according to the specific requirementsassociated with cultivation of mammalian cells under protein-free conditions in abubble-aerated system. The effects of the addition of an antifoam agent on growth,viability, productivity, metabolic rates, and release of lactate dehydrogenase wereinvestigated. Although it was possible to establish cultivation and production at a20-L scale without the use of antifoaming substances, the addition of 0.002% silicon-oil-based antifoaming reagent improved the cultivation system by completely prevent-ing foam formation. This reduced the release of lactate dehydrogenase activity to thelevel found in bubble-free aerated stirred tank membrane bioreactors and led to areduction in generation doubling times by about 5 h (17%). Using the optimizedmedium formulation, cells were cultivated at a 1000-L scale, resulting in a cultureperformance comparable to the 20-L airlift bioreactor. For comparison, cultivationswith protein-containing SMIF7 medium were carried out at 20- and 1000-L scales.The application of protein supplements did not lead to a significant improvement inthe cultivation conditions. The results were also compared with experiments performedin a bubble-free aerated stirred tank membrane bioreactor to evaluate the influenceof bubbles on the investigated culture parameters. The data implied a higher metabolicactivity of the cells in airlift bioreactors with a 150% higher glucose consumption rate.The results of this study clearly demonstrate the applicability of a protein-freechemically defined medium for the production of recombinant proteins with BHK cellsin airlift bioreactors.

Introduction

Culture media used for the production of recombinantproteins with mammalian cells even today often containundefined and protein-rich supplements such as serum.Major problems arise concerning the animal origin ofthese supplements leading to a high contamination risk,reduced product safety, and difficult licensing procedures(Hesse and Wagner, 2000).

To provide an alternative, a number of protein-free cellculture media have been developed recently, and theirapplicability for the production of biopharmaceuticals wasproven for several process conditions (Mizrahi and Lazar,1991; Kessler, 1994). Ryll et al. (1990) for example,showed that the production of recombinant human in-terleukin-2 (IL-2) was possible under protein-free condi-

tions in hollow fiber and stirred tank bioreactors. Zanget al. (1995) established cultivation of the CHO-SSF3 cellline using the protein-free FMX-8 medium and demon-strated the applicability of this cultivation system for theproduction of urokinase-type plasminogen activator (uPA)and a humanized immunoglobulin G κ light chain (IgGLC) in compact loop and stirred tank bioreactors. Cruzet al. (1998) successfully used the protein-free SMIF6medium for the production of a fusion protein with arecombinant BHK-21 cell line. They also showed thatadaptation to protein-free medium can be performed ina single step. The cell specific productivity was 200%higher compared to the control using DMEM mediumsupplemented with 10% FCS. Recently, Chen et al. (2000)developed the chemically defined protein-free mediumDF6S for the cultivation of the cell line CHO-2DS toproduce human prothrombin in suspension batch cultureusing a bubble-free membrane aeration bioreactor anddemonstrated growth and production specific efficiencycomparable to the conventional serum-containing me-dium. Serum-free and protein-free media have been

* To whom correspondence should be addressed. Phone:+49-(0)531-6181-104. Fax: +49-(0)531-6181-488. Email:wagner.roland@gbf.de.

† Present address: Institute of Applied Microbiology, Universityof Agricultural Sciences, A-1190 Vienna, Muthgasse 18, Austria.

833Biotechnol. Prog. 2003, 19, 833−843

10.1021/bp0257630 CCC: $25.00 © 2003 American Chemical Society and American Institute of Chemical EngineersPublished on Web 04/19/2003

shown to support the growth of different cell lines(Murakami, 1990; Mizrahi and Lazar, 1991). Unlikeserum-containing media that may be utilized for a broadrange of cell types and culture conditions, protein-freemedia are generally highly specific (Jayme, 1991). There-fore, we developed a cell-line-specific protein-free mediumline combined with a reliable adaptation protocol(Scharfenberg and Wagner, 1995). Adaptation to a protein-free medium is generally accompanied by the formationof spheroids, which desintegrate during subsequentprocess adaptation to suspensions supported by specificmedium components (Cruz et al., 1998; Hesse andWagner, 2000).

Airlift bioreactors are used for the production ofrecombinant proteins with mammalian cell cultures(Birch and Arathoon, 1990; Pickett, 1987). The processesdescribed in the literature using airlift reactors for thispurpose either depend on serum-containing or protein-containing cell culture media (Fountoulakis et al., 1995;Russo et al., 1997; Cosgrove et al., 1995) or use adherentcell clones attached to various types of microcarriers(Gray et al., 1992; Crowley et al., 1991). However, theapplicability of protein-free media for the production ofrecombinant proteins in suspension cultures with mam-malian cells has not been reported for airlift bioreactorsystems so far. In this study, we describe the establish-ment and optimization of the cultivation of a BHK cellline to produce recombinant human interleukin-2 at 20-and 1000-L scales in an airlift bioreactor system underprotein-free conditions.

Materials and MethodsCell Line. The interleukin-2-producing recombinant

cell line BHK-21pSVIL2 was used for all experiments.It was derived from the cell line BHK-21C13-B bytransfection with an expression plasmid based on theplasmid pBR322 (Conradt et al., 1986; Conradt et al.,1989). It expresses human interleukin-2 constitutivelyunder the control of the SV40 promoter (Conradt et al.,1989). The cell line was obtained from the geneticengineering department of the German Research Centrefor Biotechnology (GBF). BHK-21pSVIL2 has been usedfor several years as a model cell line in various processescomparing different bioreactor configurations and cultureconditions (Wagner and Lehmann, 1988; Ryll et al., 1990;Kratje and Wagner, 1992; Kratje et al., 1994). After a6-month period of cultivation under continuous agitatingconditions and exposure to SMIF7 medium (contains 0.9mmol L-1 CaCl2), cells completely adapted to suspensionconditions as indicated by a substantial reduction of theaggregates to 2-3 cells, and 90% of the cells remainedas single cells in suspension (Figure 1).

Bioreactors and Culture Conditions. Three differ-ent bioreactors were used in this study to performcultivations at a benchtop (2.5 L), pilot (20 L) andproduction scale (1000 L), respectively.

Cultivation in MSR2.5. A membrane stirrer bio-reactor designed at GBF (Lehmann et al., 1987; Wagnerand Lehmann, 1988; Ryll et al., 1990) with 2.5-L workingvolume (MSR2.5, total volume 2.8 L) was used for allbench-scale cultivations. This bioreactor was equippedwith a bubble-free aeration system using a polypropylenemembrane tubing (S6, AKZO Nobel, Wuppertal, Ger-many). The process gas was supplied by a gas-mixingstation (MKS, Munich, Germany) as a mixture of air,oxygen, nitrogen, and carbon dioxide. The flow rate wasadjusted to 200 mL min-1. The carbon dioxide content ofthe mixture was maintained at 12.5% to adjust culturepH below 7.4. The proportions of the other gases werevaried to maintain a constant oxygen pressure in theculture. The dissolved oxygen concentration was con-trolled at 40% air saturation for all experiments usingan autoclavable oxygen probe (Ingold, Steinbach, Ger-many). All cultivations were operated in batch mode.Samples were taken daily to monitor the cultivationparameters cell concentration, viability, product concen-tration, glucose concentration, L-glutamine concentration,lactate concentration, and lactate dehydrogenase (LDH)activity.

Cultivation in ALR30. The 30-L airlift fermenter(ALR30) with 20-L working volume used for all experi-ments at a pilot scale was a modified prototype built byBraun-Diessel Biotech (Melsungen, Germany). The biore-actor was equipped with two autoclavable oxygen probes(Ingold, Steinbach, Germany), a pH electrode (Ingold),and a Pt100 thermoelectric probe (Braun-Diessel Bio-tech). The process was controlled by a digital control unit(DCU1, Braun-Diessel Biotech), which operated theBIOSTAT ED supply unit and a gas mixing station (bothBraun-Diessel Biotech). Gas mixtures of variable con-tents of air, oxygen, and nitrogen were used to aeratethe cultures. Carbon dioxide was added at intervals tomaintain culture pH below 7.4. The flow rate wasadjusted to 0.4 L min-1 for batch cultures and variedbetween 0.4 and 2.0 L min-1 in continuously perfusedcultures. In perfusion experiments a Prostak (Open-Channel-Module PSGVAG101, equipped with hydrophilicPVDF microporous (0.22 µm) Durapore membranes)cross-flow filtration unit (Millipore, Eschborn, Germany)with 2.49-m2 filtration area (3 × 0.83 m2) was used ascell retention device. A tangential flow rate of 300-400L h-1 was applied for that purpose. Continuously per-fused cultures were performed to provide sufficientamounts of cells for the inoculation of the 1.5-m3 airliftbioreactor. The dissolved oxygen concentration was con-trolled at 40% air saturation for all experiments. Cultiva-tion parameters were determined from samples takenonce a day.

Cultivation in ALR1500. The 1.5-m3 airlift reactor(ALR1500) used for the cultivations at a production scalewas designed by Yonsel (1991) in collaboration with anindustrial partner (Diessel, Hildesheim, Germany) andfirst described by Yonsel and Deckwer (1991). Thissystem was used before in an investigation on the largescale production of Tetrahymena thermophila (Hellenb-roich et al., 1999). The bioreactor was equipped with fourautoclavable oxygen probes (Ingold) and two pH elec-trodes (Ingold). The control unit was also designed byDiessel and was equipped with two Simatik S 5 processcontrol units (Siemens AG, Munich, Germany). A gasmixture consisting of variable proportions of air, oxygen,

Figure 1. Single cell suspension of BHK-21pSVIL2 cultivatedin ALR30 under protein-free medium conditions using SMIF7.The cells did not aggregate to clusters or spheroids under eitherlow (5 × 105 mL-1; a), medium (5 × 106 mL-1 ; b), or high(1 × 107 mL-1 ; c) cell concentrations.

834 Biotechnol. Prog., 2003, Vol. 19, No. 3

and nitrogen was used to aerate the cultures. Carbondioxide was added at intervals to maintain the culturepH below 7.4. The flow rate was adjusted to 24 L min-1,and the dissolved oxygen concentration was maintainedat 40% air saturation. All experiments were performedin batch mode. Samples were taken daily to monitor theprogress of the cultivations.

Cell Culture Media. Protein-Free SMIF7 Media.The protein-free SMIF7 medium (GIBCO Life Technolo-gies, Karlsruhe, Germany, custom-made) was based onSMIF6 (Scharfenberg and Wagner, 1995) and optimizedfor the cultivation of baby hamster kidney (BHK) celllines. It was used as a basal medium for all experiments.To provide shear-force protecting properties, SMIF7 wassupplemented with 0.1% Pluronic F-68 (Sigma) in allexperiments performed in airlift bioreactors and in someof the experiments performed in MSR2.5 (SMIF7p). Toprevent foam formation SMIF7p was additionally supple-mented with 0.002% of the antifoaming substance Q7-2587 (Dow Corning, Midland, USA) in some of thecultivations as stated below (SMIF7pa) (Table 2).

Protein-Containing SMIF7 Medium. The protein-containing SMIF7 medium (SMIF7prot) was obtained bysupplementation of SMIF7 with 1.0 g L-1 albumin(Albumax I, GIBCO Life Technologies), 10.0 mg L-1

insulin (Sigma), 10.0 mg L-1 transferrin (Biotest, Frank-furt/Main, Germany), 0.1% Pluronic F-68 (Sigma) and0.002% Q7-2587.

Analytical Methods. Cell Counting. Dead cells werestained according to the trypan blue exclusion dyemethod (Roche Diagnostics, Basel, Switzerland) andcounted using a Neubauer-type hemocytometer. To guar-antee a minimum margin of error, total cell concentrationwas determined using a Casy Counter (Scharfe, Reut-lingen, Germany) after lysis of cells in 90% lysis buffer(citric acid, 0.1 mol L-1, N-acetyl-trimethylammonium-bromide, 27.5 mmol L-1) by counting the released nuclei.

Glucose and Lactate. Glucose and lactate concentra-tions were determined in cell-free culture supernatantsobtained by centrifugation of cell culture samples(13,000 × g, 3 min) using a YSI 2700 Select glucose andlactate analyzer (Yellow Springs Instruments, Ohio).

Interleukin-2 (IL-2) Assay. Interleukin-2 activitiesof cell-free culture supernatants, obtained either bycentrifuging (250 × g, 3 min) sterile samples (for experi-ments with MSR2.5) or by sterile filtration with lowprotein-binding filters (Millex-GV, Millipore, Eschborn,Germany) (for all experiments with airlift bioreactors),were determined by the [3H]thymidine incorporationassay using the IL-2-dependent murine CTLL-2 cell line(ATCC, TIB-214) as described (Gillis et al., 1978). Alaboratory IL-2 preparation was used as internal stan-

dard. The specific activity of the purified IL-2 amountedto 107 U mg-1 (Conradt et al., 1985).

Amino Acid Analysis. Quantitative determination offree amino acids in cell-free culture supernatants wasperformed according to Larsen and West (1981). Briefly,after treatment with perchloric acid to precipitate theproteins, samples were mixed with o-phthaldialdehydein the presence of mercaptoethanol. The isoindole deriva-tives formed were separated by reversed phase HPLC(column, AminoOPA, 3 µm, Grom Herrenberg, Germany)using a System 450 MT2 (Kontron Instruments, Ham-burg, Germany) and analyzed with a SFM 25 fluores-cence detector (Kontron Instruments). Norvalin was usedas an internal standard for quantification.

Analysis of Lactate Dehydrogenase (LDH) Activ-ity. LDH activity in cell-free culture supernatants wasdetermined according to the method of Pesce et al. (1964).Briefly, 100-500 µL of sample was mixed with 30 mmolL-1 sodium phosphate buffer (pH 7.4) containing 1 mmolL-1 pyruvate and 0.27 mmol L-1 NADH (BoehringerMannheim, Germany) in a total volume of 1 mL. Thelinear reduction of the absorbance at 340 nm wasmeasured for 1 min using a kinetic photometer (UltrospecK, LKB, Freiburg, Germany) to determine LDH activityin the sample. To determine the total intracellular LDHactivity, 2.9 × 106 cells at a viability of 98.1% weredisrupted for 30 min on ice using 5 mL of extractionbuffer (30 mmol L-1 phosphate buffer pH 7.2, 1%TRITON X-100).

Mathematical Analysis. Mitotic rate, generationdoubling time, and volumetric as well as cell specific rateswere calculated according following equations using thephase of exponential growth:

where CSi,X is the cell specific production and consumptionrate, c1 is the concentration of the component in thereactor at time t1 [mol mL-1], c2 is the concentration ofthe component in the reactor at time t2 [mol mL-1], t1and t2 are Time 1 or 2, respectively [h], and X, X1, andX2 are cell concentrations in the reactor system [mL-1].

ResultsEstablishment of Cultivation on 20-L Scale. A

model process using a BHK-21 cell line for the productionof recombinant human interleukin-2 in an airlift biore-actor system was established, applying a batch cultiva-

Table 1. Technical Data of Airlift Bioreactors ALR30 andALR1500 Used in This Study

reactor

ALR30 ALR1500a

reactor volume, VR [m3] 0.03 1.50maximum working volume, VW [m3] 0.02 1.20inner diameter, D [mm] 127 500height, H [mm] 1524 6000liquid height, HL [mm] 1270 5000HL/D 10 10inner loop diameter, DI [mm] 85 355head diameter, DH [mm] 234 800DH/D 1.8 1.6

a In the ALR1500 cultivations were performed with a workingvolume of 1.0 m3 instead of the maximum working volume of 1.2m3.

Table 2. Overview of Different Medium CompositionsUsed in This Study

medium

SMIF7 basalprotein-free

medium

0.1%Pluronic

F-680.002%Q7-2587

proteinsupplementsa

SMIF7 ×SMIF7p × ×SMIF7pa × × ×SMIF7prot × × × ×

a The protein supplements comprised 1.0 g L-1 albumin, 10 mgL-1 transferrin, and 10 mg L-1 insulin.

mitotic rate [h-1]: v )(ln X2 - ln X1)

(t2 - t1) ln 2

generation doubling time [h]: td ) ln 2v

cell specific rate [mol h-1]: CSi,X )(c2 - c1)

∫t1

t2X dt

Biotechnol. Prog., 2003, Vol. 19, No. 3 835

tion strategy. The BHK-21pSVIL2 cell line was usedbefore in several investigations on the production ofrecombinant human interleukin-2 using different processconditions, including the application of protein-free media(Lucki-Lange and Wagner, 1991; Kratje and Wagner,1992; Kratje et al., 1994). Prior to the experimentsdescribed here, the producer cell line used was adaptedto protein-free SMIF7 medium for optimal growth andproductivity in single cell suspension cultures withoutforming spheroids, as shown in Figure 1, using a 2.5-Lmembrane-aerated bioreactor system (MSR2.5, Lehmannet al., 1987; Wagner and Lehmann, 1988). Figure 2 showsrepresentative data on selected cultivation parameters.Cells grew with an average generation doubling time of31.7 ( 2.5 h up to a maximum cell concentration of 1.3-1.6 × 106 mL-1. Viability was always higher than 87%.At day 6-7, L-glutamine was depleted and, therefore, cellgrowth reduced. An increasing amount of dying cells inthis phase of the culture was also indicated by anincrease in LDH release, which could not be observed bythe trypan blue exclusion method, expressed as viability(see Figure 2b). This indicated that LDH was releasedby lysed cells. During the exponential growth phase anoverall cell specific productivity of 3.9 ( 0.7 U 10-5 h-1

could be obtained.To enable cultivation of the producer cell line in the

bubble-aerated airlift bioreactor, we modified the culturemedium by supplementing SMIF7 medium with 0.1%Pluronic F-68 to provide sufficient shear force protection(SMIF7p). Applying this modification, cultivation of theBHK-21pSVIL2 cell line was possible in the ALR 30.Figure 3 shows a representative cultivation course com-prising all determined cultivation parameters. The aver-age generation doubling time was 30.9 ( 2.6 h, the samerange as observed in the MSR2.5 bioreactor. Viability wasalways higher than 86% until day 3.5-4.5, when L-glutamine was depleted. Nutrient depletion occurredabout 2.5 days earlier in airlift-cultures compared tocultures in MSR2.5. Moreover, the cell specific glucose

uptake rate was about 2 times higher, and the cellspecific lactate secretion rate was about 2.5 times higherin the ALR30 (see Figure 3). According to the forcesassociated with bubble disruption (Cherry and Hulle,1992), shear stress is much higher in sparged than innonsparged cultivation systems (Oh et al., 1989; 1992).Nevertheless, the LDH release observed in the ALR30cultures, being 14.2 ( 6.0 nkat L-1 h-1, was just slightlyhigher than in MSR2.5 (12.3 ( 1.0 nkat L-1 h-1, seeFigure 3). Cell specific productivities obtained in ALR30were about 2 times higher compared to those in MSR2.5.

A foam layer of about 5-10 cm thickness alwaysformed on the surface of the culture broth using SMIF7pmedium in ALR30. To prevent foam formation entirely,we further supplemented the SMIF7p medium with thesilicone oil antifoaming substance Q7-2587 (SMIF7pa)(Figure 4). In a series of experiments the amount neededto suppress foam formation was determined to be 0.002%(data not shown). This medium (SMIF7 + 0.1% PluronicF-68 + 0.002% Q7-2587) was used for all further experi-ments with protein-free medium. The application ofSMIF7pa led to a significant improvement in cultivationconditions. First, the generation doubling time of the cellswas reduced by about 5 h (17%) to 25.4 ( 0.9 h. Cellspecific glucose uptake increased to about 421.8 ( 87.9fmol h-1, while cell specific lactate secretion was in-creased to 587.3 ( 102.2 fmol h-1. LDH release wasreduced to the level found in MSR2.5 cultures. Theaverage cell specific productivity observed with thissystem was 5.8 ( 1.6 U 10-5 h-1 in the range betweenthat observed in MSR2.5 cultures using SMIF 7 mediumand in ALR30 cultures using SMIF7p medium. No foamformation was observed when 0.002% Q7-2587 wasapplied.

SMIF7pa was also used for cultivations in MSR2.5 forcomparison. The average generation doubling time ob-tained in these cultures was 32.0 ( 2.6 h, comparable tothat observed in MSR2.5 using SMIF7. Cell specificglucose uptake (245.7 ( 44.1 fmol h-1) and lactate

Figure 2. Representative cultivation course in the MSR2.5 bioreactor using SMIF7 medium: (a) Change in glucose (4), lactate (3),and L-glutamine (0) concentration compared to the change in cell concentration (b). (b) Change in viability (2), LDH activity in theculture supernatant (1), and IL-2 concentration ()) compared to the change in cell concentration (b). Seven fermentations wereperformed using this configuration.

836 Biotechnol. Prog., 2003, Vol. 19, No. 3

secretion (375.2 ( 48.9 fmol h-1) rates were lower thanthose found in culture supernatants of ALR30 but higherthan in MSR2.5 when using SMIF7 medium. This mightbe explained by the 200% higher cell specific productivityobserved in MSR2.5 when SMIF7pa medium was used.

Scale-Up to 1000-L Scale. SMIF7pa medium, im-proved for cultivation in airlift reactors, was used for thecultivation at a 1000-L scale. To obtain the cell massrequired for inoculation of the 1.5-m3 airlift bioreactor

(ALR1500), we established perfusion cultures in ALR30.This was achieved by the use of a Prostak cross filtrationunit as a cell retention device. The glucose concentrationin the culture was used as key parameter to control theperfusion rate, which was adjusted to guarantee astationary concentration of about 1 g L-1. To achieve this,the perfusion rate was varied between 0 and 4 reactorvolumes per day. Figure 5 shows a representative courseof these cultures. Maximum cell concentrations of (0.6-

Figure 3. Representative cultivation course in the ALR30 bioreactor using SMIF7p medium. (a) Change in glucose (4), lactate (3),and L-glutamine (0) concentration compared to the change in cell concentration (b). (b) Change in viability (2), LDH activity in theculture supernatant (1), and IL-2 concentration ()) compared to the change in cell concentration (b). Five fermentations were performedusing this configuration.

Figure 4. Representative cultivation course in the ALR30 bioreactor using SMIF7pa medium. )a) Change in glucose (4), lactate(3), and L-glutamine (0) concentration compared to the change in cell concentration (b). (b) Change in viability (2), LDH activity inthe culture supernatant (1), and IL-2 concentration ()) compared to the change in cell concentration (b). Five fermentations wereperformed using this configuration.

Biotechnol. Prog., 2003, Vol. 19, No. 3 837

1.1) × 107 mL-1 were obtained. The average generationdoubling time was 27.3 ( 1.6 h. Viability was alwayshigher than 82%. At the end of these cultures, when cellconcentrations exceeded 3 × 106 mL-1, the released LDHactivity considerably increased up to about 6 µkat L-1.Although we always observed very high viabilities, thisindicates a significant number of cells dying duringcultivation. On the basis of the cell specific LDH activityof BHK-pSVIL2, which has been determined as 10.3 fkat(10.3 × 10-15 kat), the accumulated dead cell populationwas calculated to 5.8 × 108 L-1, representing only 19%of the total cells formed during the culture period. Whencell concentrations of 8 × 106 to 2 × 107 mL-1 werereached, the culture broth was transferred to the ALR1500 system using a stainless steel transfer vessel (TB25,Naue GmbH, Weiterstadt, Germany).

We were able to apply the same cultivation protocol tocultivations in ALR1500 as already employed in ALR 30.Figure 6 shows a representative cultivation course in-cluding all determined parameters. Cultures in ALR1500were started with cell concentrations of 2.0 × 105 mL-1.Cells grew exponentially up to maximum cell concentra-tions of about 1.6 × 106 mL-1, which were reached at theend of the growth phase on day 4. Viability was alwayshigher than 91% until day 4, when glucose and L-glutamine were depleted. The average generation dou-bling time was 25.2 ( 1.9 h, and glucose uptake, lactatesecretion, and L-glutamine uptake rates showed averagevalues of 382.3 ( 25.0, 631.4 ( 21.2, and 64.5 ( 8.2 fmolh-1, respectively, during the exponential growth phase.These parameters differed by about 11% compared tothose obtained with ALR30. However, LDH release at6.1 ( 1.0 nkat L-1 h-1 was significantly lower in ALR1500.

Comparison to Protein-Containing Medium. Todemonstrate the performance of the protein-free SMIFmedium in comparison to protein-containing cell culture

media, cultivations with SMIF7prot medium were per-formed in all bioreactors used. SMIF7prot mediumconsisted of SMIF7 medium supplemented with 1.0 g L-1

albumin, 10.0 mg L-1 insulin, and 10.0 mg L-1 transfer-rin. For the experiments performed in this study,SMIF7prot medium was further supplemented with 0.1%Pluronic F-68 and 0.002% Q7-2587.

In the MSR2.5 system the use of SMIF7prot mediumled to a 6.1-h reduction in the average generationdoubling time, compared to the use of SMIF7pa medium.Cell specific glucose uptake (222.0 ( 25.6 fmol h-1) andlactate secretion (453.5 ( 38.7 fmol h-1) rates were onlyslightly reduced, while the cell specific productivity (6.4( 3.3 U 10-5 h-1) was significantly reduced (about 30%).The L-glutamine uptake rate was not significantlyaffected. The released LDH activity observed here (5.0( 1.8 nkat L-1 h-1) was the lowest of all configurationsinvestigated.

Application of SMIF7prot medium in the ALR30system surprisingly led to a 3.06-h increase in generationdoubling time. Compared to cultures performed withSMIF7pa medium, glucose and L-glutamine uptake aswell as lactate secretion rates were reduced between 19%and 35%. Cell specific productivities were nearly identicalfor ALR30-SMIF7pa and ALR30-SMIF7prot, at 5.8 (1.6 and 5.9 ( 1.7 U 10-5 h-1, respectively. LDH releasewas reduced by 22.9% when SMIF7prot medium wasapplied.

At the 1000-L scale, only one single cultivation usingSMIF7prot medium was performed. In this experiment,we observed the lowest generation doubling time (23.2h) of all experiments. Moreover, the released LDHactivity was relatively low (6.1 nkat L-1 h-1) and in thesame range as under SMIF7pa medium conditions.Interestingly, we observed relatively high cell specificglucose (557.5 fmol h-1) and L-glutamine (98.8 fmol h-1)uptake and lactate secretion (753.0 fmol h-1) rates. These

Figure 5. Representative cultivation course in the ALR30 bioreactor in perfusion mode using SMIF7pa medium. (a) Change inglucose (4), lactate (3), L-glutamine (0) concentration, and perfusion rate (s) compared to the change in cell concentration (b). (b)Change in viability (2), LDH activity in the culture supernatant (1), and IL-2 concentration ()) compared to the change in cellconcentration (b). Four fermentations were performed using this configuration.

838 Biotechnol. Prog., 2003, Vol. 19, No. 3

metabolic rates represented the highest values that weobserved during the whole study. A cell specific produc-tivity of 4.3 U 10-5 h-1 was measured, indicating thesame range as observed in the two experiments applyingSMIF7pa medium.

Discussion

The main objective of this study was to demonstratethe applicability of a chemically defined, completelyprotein-free medium for the cultivation of a BHK-21producer cell line on an industrial scale using a bubble-aerated airlift bioreactor system. We also wanted to showthe feasibility of medium optimization and scale-upprocedures during process development when applyingprotein-free medium conditions. For this purpose weinvestigated the establishment of the well-optimizedprocess for the production of interleukin-2 (IL-2) using aBHK-21 host cell line (Ryll et al., 1990). The process wasestablished at a 20-L pilot and 1000-L production scale,applying the protein-free SMIF7 medium as the basalculture medium. Additionally, we compared the processperformance using protein-free medium conditions toconventional standard environmental conditions using aprotein-containing medium based on the same basalformulation.

At the beginning of this study, cultivation of theproducer cell line was established under protein-freecultivation conditions at a benchtop scale in a 2.5-Lmembrane-aerated stirred tank bioreactor (MSR2.5).Therefore, the first task to establish this process was totransfer the cultivation process from the bubble-free,membrane-aerated system to a bubble-aerated bioreactorsystem. In general, bubble-aerated systems expose thecells to higher shear stress than bubble-free nonspargedsystems (Oh et al., 1989; 1992). The protein-free SMIF7medium has been successfully used for cultivation of theBHK-21 cell line in MSR2.5 (Lucki-Lange and Wagner,1991). However, it was originally designed to be used in

bubble-free aerated cultivation systems such as spinnerflasks or membrane-aerated bioreactors that were as-sociated with low shear stress. Therefore, SMIF7 doesnot contain any shear-protecting substances. To providesufficient shear force protection we modified the culturemedium by adding the surfactant Pluronic F-68 at aconcentration of 0.1% (SMIF7p). Pluronic F-68 wasshown to have significant cell-protecting properties (Zhanget al., 1992a; Wu et al., 1995) and is widely used for thispurpose (Papoutsakis, 1991). By applying this modifica-tion we were able to establish cultivation in ALR30 underprotein-free conditions. Growth characteristics of BHK-21pSVIL2 using SMIF7p in ALR30 were comparable tothose in MSR2.5 cultures using SMIF7, as indicated bynearly identical generation doubling times of about 30 hin both systems. This is significantly lower than thegeneration doubling time of 50 h reported by Merten etal. (1994), who used the same recombinant human IL-2-producing BHK-21 cell line cultured in serum-free, butinsulin-containing MDSS2 medium in a 1-L continuouslyperfused stirred tank Biolafitte ICC bioreactor with aworking volume of 1.6 L. These data can be compared tothe results obtained earlier with the same cell line indifferent bioreactor and medium configurations showinga variation of growth characteristics depending on thebioreactor system and the medium formulation. Thegeneration doubling time was reported to be 72 h inMSR2.5 using a 1:1 mixture of DMEM and Ham’s F12protein-free medium (Ryll et al., 1990). An even highergeneration doubling time of 115.1 h was observed forBHK-21pSVIL2 in a fluidized bed bioreactor systemusing porous borosilicate glass beads as immobilizationmaterial for cell entrapment, and the same protein-freemedium (Kratje and Wagner, 1992; Kratje et al., 1994).However, 15.6% more LDH was released by the cells inthe ALR30-SMIF7p system. Similar observations werereported by Al-Rubeai et al. (1993), who suggested thathigh hydrodynamic stress, as generated by heavy agita-

Figure 6. Representative cultivation course in the ALR1500 bioreactor using SMIF7pa medium. (a) Change in glucose (4), lactate(3), and L-glutamine (0) concentration compared to the change in cell concentration (b). (b) Change in viability (2), LDH activity inthe culture supernatant (1), and IL-2 concentration ()) compared to the change in cell concentration (b). Two fermentations wereperformed using this configuration.

Biotechnol. Prog., 2003, Vol. 19, No. 3 839

tion or bubble bursting, may lead to the violent and rapiddestruction of the cells just after the onset of cell death.Nevertheless, a 92.3% higher cell specific productivitywas achieved with the ALR30-SMIF7p configuration.There was one major problem associated with the use ofSMIF7p in ALR30. In all experiments performed withthis combination we observed the formation of a foamlayer of about 5-10 cm thickness on the surface of theculture broth. As described by Zhang et al. (1992b), thepresence of a foam layer in mammalian cell cultures leadsto a significant loss of cells by concentrating them in thefoam layer as a result of adsorption onto rising bubbles.Therefore, we decided to further supplement the SMIF7pmedium with a silicone-oil-based emulsion (Q7-2587)used in industrial and medical applications, to suppressfoam formation. The resulting optimized protein-freemedium formulation (SMIF7pa) then consisted of SMIF7,0.1% Pluronic F-68, and 0.002% Q7-2587. Compared tothe primary culture system (MSR2.5-SMIF7) the ap-plication of SMIF7pa in ALR30 led to a significantimprovement of the culture performance indicated bycritical cell growth parameters (Figure 7). The generationdoubling time was reduced by 5.5 h and the cell specificproductivity was about 47.1% higher compared to

MSR2.5-SMIF7, while the vitality of the cells was notinfluenced (indicated by the same LDH release in bothsystems). Nevertheless, we always observed a higherenergy demand characterized by higher glucose andL-glutamine uptake rates as well as higher lactatesecretion rates in airlift bioreactors compared to themembrane-aerated MSR2.5 system. This might be dueto the higher shear stress associated with airlift biore-actors. This observation is confirmed by data reportedin Frangos et al. (1988), who found increased metabolicrates in response to high mechanical stress.

In comparison to the ALR30-SMIF7p configuration weobserved a slight reduction of the released LDH activityto the level found with the MSR2.5-SMIF7 system.Obviously, the application of the antifoaming supplementQ7-2587 did not lead to a further improvement in cellprotection but to a significant reduction of the generationdoubling time of about 5 h, most probably because of theprevention of a foam layer, which is also associated withthe reduced loss of cells.

We also performed experiments applying the SMIF7pamedium in the MSR2.5 system to investigate the influ-ence of the used shear force protecting supplements oncultivation in a membrane bioreactor. Although the

Figure 7. Cultivation parameters obtained during medium optimization and scale-up. Bioreactor-medium configurations: (A)MSR2.5-SMIF7; (B) MSR2.5-SMIF7pa; (C) ALR30-SMIF7p; (D) ALR30-SMIF7pa; (E) ALR1500-SMIF7pa. Investigated cultivationparameters: (a) generation doubling time, (b) LDH release, (c) cell specific IL-2 productivity, (d) glucose uptake rate, (e) lactatesecretion rate, (f) L-glutamine uptake rate.

840 Biotechnol. Prog., 2003, Vol. 19, No. 3

generation doubling time was not affected, LDH releasewas reduced by 46.9%, indicating that even in a biore-actor system associated with rather low shear stressthese supplements can provide a significant improvementin the cellular physiology leading to higher vitalities ofthe cells. Surprisingly, the application of SMIF7paresulted in a 65.1% higher glucose uptake, a 91.6% higherlactate secretion, and a 22.1% increased L-glutamineuptake rate. This higher energy demand might be dueto the 2.2-times higher cell specific productivity comparedto the MSR2.5-SMIF7 configuration. These data there-fore also revealed an improvement in culture perfor-mance when applying the SMIF7pa medium in theMSR2.5 system.

Because of identical geometrical data, such as the ratioof liquid height to inner diameter or the ratio of headdiameter to inner loop diameter (Table 1), cultivationparameters in ALR30 were completely adopted to the1000-L scale (ALR1500). This facilitated the scale-upprocedure because the cultivation protocol used forALR30 based on SMIF7pa medium could be directlyapplied to cultivation of the BHK-21 producer cell linein ALR1500. During scale-up, most of the determined

cultivation parameters did not change significantly,indicating a comparable performance of the investigatedcultivation systems. For example, the values of themetabolic rates as well as the generation doubling timeand the cell specific productivity either were identical tothose obtained with the ALR30-SMIF7pa configurationor only differed by up to 11%. However, LDH release wassignificantly lower in ALR1500, indicating a lower shearstress in this bioreactor. A similar effect was reportedby Handa-Corrigan et al. (1989) in bubble column reac-tors with increasing column height. They suggested thatthis effect might be due to the fact that in a higher reactorthe cells are less often exposed to the “destructive” zonewhere bubble disruption occurs.

To demonstrate the applicability of the protein-freeSMIF7 medium for an application of industrial relevance,we also compared the growth-promoting and productiv-ity-increasing performance to that obtained using a“conventional” cell culture medium. For this purpose weperformed experiments using the protein-containingSMIF7prot medium in all three bioreactor systemsinvestigated. SMIF7prot medium was also based onSMIF7 medium but additionally contained albumin,

Figure 8. Comparison of cultivation parameters obtained with the protein-free medium SMIF7pa and those obtained with theprotein-containing medium SMIF7prot. Bioreactor-medium configurations: (A) MSR2.5-SMIF7pa; (B) MSR2.5-SMIF7prot; (C)ALR30-SMIF7pa; (D) ALR30-SMIF7prot; (E) ALR1500-SMIF7pa; (F) ALR1500-SMIF7prot. Investigated cultivation param-eters: (a) generation doubling time, (b) LDH release, (c) cell specific IL-2 productivity, (d) glucose uptake rate, (e) lactate secretionrate, (f) L-glutamine uptake rate.

Biotechnol. Prog., 2003, Vol. 19, No. 3 841

transferrin, and insulin as protein supplements. Ad-ditionally, Pluronic F-68 and Q7-2587 were added in thesame amounts as in SMIF7pa to provide sufficient shearforce protection and to suppress foam formation in theairlift bioreactors. This also ensured comparability withthe results obtained with the optimized protein-freemedium. In MSR2.5, the use of SMIF7prot medium ledto a slight improvement in cell growth, indicated by a6.1-h reduction in generation doubling time compared tocultures with SMIF7pa, and a very low LDH release of5 nkat L-1 h-1 (Figure 8b). Cell specific productivity,however, was slightly reduced (25.4%; Figure 8c) whilethe metabolic rates were almost the same as observedwith SMIF7pa medium (Figure 8d-f). In the ALR 30system the use of SMIF7prot medium caused a reductionin glucose uptake (26.2%), lactate secretion (22.8%), andL-glutamine uptake (16.1%) rates, as well as a reductionin LDH release of 18.6 % compared to cultures withSMIF7pa. However, cell specific productivity was notaffected while generation doubling time was even in-creased by 3 h when SMIF7prot medium was used(Figure 8a). At the 1000-L scale, cell specific productivity,LDH release (Figure 8b), and generation doubling time(Figure 8a) were not significantly different whenSMIF7prot medium was used compared to cultures withSMIF7pa, while the metabolic rates were higher (Figure8d-f). However, as only one experiment with SMIF7protmedium in ALR1500 was performed, these differencesshould not be overvalued. The main objective of this studywas to show whether the cultivation parameters were ina range comparable to those observed with SMIF7pamedium. Taken together, the results obtained withprotein-containing medium were nearly identical to thoseobtained with protein-free medium. There was also nosignificant improvement in cellular physiology under thedifferent cultivation conditions when protein supple-ments were added.

ConclusionsThe application of the protein-free SMIF7 medium for

the cultivation of BHK-21 cells is feasible using bubble-aerated airlift bioreactors up to a 1000-L scale whenPluronic F-68 is added to protect the cells from hydro-dynamic stress and a silicone oil supplement is used toprevent foam formation. These supplements also improvecultivation conditions when applied in membrane-aeratedstirred tank bioreactors. Addition of protein supplementssuch as albumin, transferrin, or insulin to the mediumdoes not further improve cultivation conditions.

AcknowledgmentThis work was part of the demonstration project

“Application of Protein-free Medium for Industrial Pro-cesses with Mammalian Cell Cultures Producing Bio-pharmaceuticals”, funded by the European Commision,grant BIO4-CT97-2140.

References and NotesAl-Rubeai, M.; Emery, A. N.; Chalder, S.; Goldman, M. H. A

flow cytometric study of hydrodynamic damage to mam-malian cells. J. Biotechnol. 1993, 31, 161-177.

Birch, J. R.; Arathoon, R. Suspension culture of mammaliancells. Bioprocess Technol. 1990, 10, 251-270.

Chen, Z.; Iding, K.; Lutkemeyer, D.; Lehmann, J. A low-costchemically defined protein free medium for a recombinantCHO cell line producing prothrombin. Biotechnol. Lett. 2000,22, 837-841.

Cherry, R. S.; Hulle, C. T. Cell death in the thin films of burstingbubbles. Biotechnol. Prog. 1992, 8, 11-18.

Conradt, H. S.; Geyer, R.; Hoppe, J.; Grotjahn, L.; Plessing, A.;Mohr, H. Structures of the major carbohydrates of naturalhuman interleukin-2. Eur. J. Biochem. 1985, 153, 255-261.

Conradt, H. S.; Ausmeier, M.; Dittmar, K. E. J.; Hauser, H.;Lindenmaier, W. Secretion of glycosylated human interleu-kin-2 by recombinant mammalian cell lines. Carbohydr. Res.1986, 149, 443-450.

Conradt, H. S.; Nimtz, M.; Dittmar, K. E. J.; Lindenmeier, W.;Hoppe, J.; Hauser, H. Expression of glycosylated humaninterleukin-2 in recombinant mammalian cells. J. Biol. Chem.1989, 264, 17368-17373.

Cosgrove, L.; Lovrecz, G. O.; Verkuylen, A.; Cavaleri, L.; Black,L. A.; Bentley, J. D.; Howlett, G. J.; Gray, P. P.; Ward, C. W.;McKern, N. M. Purification and properties of insulin receptorectodomain from large-scale mammalian cell culture. ProteinExpression Purif. 1995, 6, 789-798.

Crowley, J.; Marsden, W. L.; Gray, P. P. Effect of dilution rateon the metabolism and product formation of a recombinantmammalian cell line growing in a chemostat with internalrecycle of cells. In Animal Cell Culture and Production ofBiologicals; Sasaki, R., Ikura, H., Eds.; Kluwer Academic:Dordrecht, 1991; pp 275-281.

Cruz, H. J.; Moreira, J. L.; Stacey, G.; Dias, E. M.; Hayes, K.;Looby, D.; Griffiths, B.; Carrondo, M. J. T. Adaptation of BHKcells producing a recombinant protein to serum-free mediaand protein-free medium. Cytotechnology 1998, 26, 59-64.

Fountoulakis, M.; Mesa, C.; Schnid, G.; Gentz, R.; Manneberg,M.; Zulauf, M.; Dembic, Z.; Garotta, G. Interferon γ receptorextracellular domain expressed as IgG fusion protein inChinese Hamster Ovary cells. J. Biol. Chem. 1995, 270,3958-3964.

Frangos, J. A.; McIntire, L. V.; Eskin, S. G. Shear stress inducedstimulation of mammalian cell metabolism. Biotechnol. Bioeng.1988, 32, 1053-1060.

Gillis, S.; Ferm, M. M.; Ou, W.; Smith, K. A. T-cell growthfactor: parameters of production and a quantitative microas-say for activity. J. Immunol. 1978, 129, 2027-2032.

Gray, P. P.; Jirasripongpun, K.; Gebert, C. Maximizing mam-malian cell expression of heterologous proteins in continuousflow bioreactors. In Harnessing Biotechnology for the 21stCentury; Bose, A., Ladisch, M. R., Eds.; American ChemicalSociety: Washington, DC, 1992; pp 197-201.

Handa-Corrigan, A.; Emery, A. N.; Spier, R. E. Effect of gas-liquid interfaces on the growth of suspended mammaliancells: mechanisms of cell damage by bubbles. Enzyme Microb.Technol. 1989, 11, 230-235.

Hellenbroich, D.; Valley, U.; Ryll, T.; Wagner, R.; Tekkanat, N.;Kessler, W.; Ross, A.; Deckwer, W.-D. Cultivation of Tetrahy-mena thermophila in a 1.5 m3 airlift bioreactor. Appl. Micro-biol. Biotechnol. 1999, 51, 447-455.

Hesse, F.; Wagner, R. Developments and improvements in themanufacturing of human therapeutics with mammalian cellcultures. Trends Biotechnol. 2000, 18, 173-180.

Jayme, D. W. Nutrient optimization for high-density biologicalproduction applications. Cytotechnology 1991, 5, 15-30.

Kessler, N. Cellular culture in media withour serum. Biofutur1994, 133, 3-18.

Kratje, R. B.; Wagner, R. Evaluation of the production ofrecombinant human interleukin-2 in a fluidized bed biore-actor. Biotechnol. Bioeng. 1992, 39, 233-242.

Kratje, R. B.; Reimann, A.; Hammer, J.; Wagner, R. Cultivationof recombinant Baby Hamster Kidney cells in a fluidized bedbioreactor system with porous borosilicate glass. Biotechnol.Prog. 1994, 10, 410-420.

Larsen, B. R.; West, F. G. A method for quantitative amino acidanalysis using precolumn o-phthaldialdehyde derivation andhigh performance liquid chromatography. J. Chromatogr. Sci.1981, 19, 259-265.

Lehmann J.; Piehl, G. W.; Schulz, R. Bubble-free cell cultureaeration with porous moving membranes. Dev. Biol. Stand.1987, 66, 227-240.

842 Biotechnol. Prog., 2003, Vol. 19, No. 3

Lucki-Lange, M.; Wagner, R. Conditions for the production ofrecombinant IL-2 in stirred suspension culture using aprotein-free medium. In Production of Biologicals fromAnimal Cells in Culture; Spier, R. E., Griffiths, J. B., Meigner,B., Eds.; Butterworth-Heinemann: Oxford, 1991; pp 180-186.

Merten, O. W.; Kierulff, J. V.; Castignolles, N.; Perrin, P.Evaluation of the new serum-free medium (MDSS2) for theproduction of different biologicals: use of various cell lines.Cytotechnology 1994, 14, 47-59.

Mizrahi, A.; Lazar, A. Media for cultivation of animal cells: anoverview. In Animal Cell Culture and Production of Biologi-cals; Sasaki, R., Ikura, H., Eds.; Kluwer Academic: Dor-drecht, 1991; pp 159-180.

Murakami, H. What should be focused in the study of cellculture technology for production of bioactive proteins? Cy-totechnology 1990, 3, 3-7.

Oh, S. K. W.; Nienow, A. W.; Al-Rubeai, M.; Emery, A. N. Theeffects of agitation intensity with and without continuoussparging on the growth and antibody production of hybridomacells. J. Biotechnol. 1989, 12, 45-62.

Oh, S. K. W.; Nienow, A. W.; Al-Rubeai, M.; Emery, A. N.Further studies of the culture of mouse hybridomas in anagitated bioreactor with and without continuous sparging. J.Biotechnol. 1992, 22, 245-270.

Papoutsakis, E. T. Media additives for protecting freely sus-pended cells against agitation and aeration damage. TrendsBiotechnol. 1991, 9, 316-324.

Pesce, A.; McKay, R. H.; Stolzenbach, F.; Cahn, R. D.; Kaplan,N. O. The comparative enzymology of lactic dehydrogenases.J. Biol. Chem. 1964, 239, 1753-1761.

Pickett, A. M. Large scale production of rDNA pharmaceuti-cals: experience and prospects. World Biotech Rep. 1987, 1,153-162.

Russo, G.; Gast, A.; Schlaeger, E.-J.; Angiolillo, A.; Pietropaolo,C. Stable expression and purification of a secreted humanrecombinant prethrombin-2 and its activation to thrombin.Protein Expression Purif. 1997, 10, 214-225.

Ryll, T.; Lucki-Lange, M.; Jager, V.; Wagner, R. Production ofrecombinant human interleukin-2 with BHK cells in a hollowfibre and a stirred tank reactor with protein-free medium. J.Biotechnol. 1990, 14, 377-392.

Scharfenberg, K.; Wagner, R. A reliable strategy for theachievement of cell lines growing in protein-free medium. InAnimal Cell Technology: Developments towards the 21stCentury; Beuvery, E. C., Griffiths, J. B., Zeijlemaker, W. P.,Eds.; Kluwer Academic: Dordrecht, 1995; pp 619-623.

Wagner, R.; Lehmann, J. The growth and productivity ofrecombinant animal cells in a bubble-free aeration system.Trends Biotechnol. 1988, 6, 101-104.

Wu, J.; Daugulis, A. J.; Faulkner, P.; Goosen, M. F. A. Protectiveeffects of polymer additives on animal cells exposed to rapidlyfalling liquid films. Biotechnol. Prog. 1995, 11, 127-132.

Yonsel, S. Entwicklung und Erprobung eines Turm-Bioreaktorsim Pilotmaâstab. In Fortschrittberichte VDI; VDI-VerlagGmbH: Dusseldorf, 1991; Vol. 3, No. 266, pp 1-131.

Yonsel, S.; Deckwer, W.-D. Hochinstrumentierter Airlift-/Bla-sensaulen-Reaktor im Pilotmaâstab. BioEngineering 1991, 7,24-30.

Zang, M.; Trautmann, H.; Gandor, C.; Messi, F.; Asselbergs, A.;Leist, C.; Fiechter, A.; Reiser, J. Production of recombinantproteins in Chinese Hamster Ovary cells using a protein-freecell culture medium. Bio/Technology 1995, 13, 389-392.

Zhang, Z.; Al-Rubeai, M.; Thomas, C. R. Effect of Pluronic F-68on the mechanical properties of mammalian cells. EnzymeMicrob. Technol. 1992a, 14, 980-983.

Zhang, S.; Handa-Corrigan, A.; Spier, R. E. Foaming and mediasurfactant effects on the cultivation of animal cells in stirredand sparged bioreactors. J. Biotechnol. 1992b, 25, 289-306.

Accepted for publication March 24, 2003.

BP0257630

Biotechnol. Prog., 2003, Vol. 19, No. 3 843