phosphate feeding improves high-cell-concentration ns0 myeloma culture performance for monoclonal...

Phosphate Feeding ImprovesHigh-Cell-Concentration NS0 MyelomaCulture Performance for MonoclonalAntibody Production

Vivian M. deZengotita,1,2 William M. Miller,2 John G. Aunins,1

Weichang Zhou1

1Fermentation and Cell Culture, Bioprocess R&D, Merck ResearchLaboratories, Merck & Co., Inc., West Point, Pennsylvania, USA2Chemical Engineering Department, Northwestern University,Evanston, Illinois, USA

Received 2 November 1999; accepted 23 April 2000

Abstract: Phosphorus depletion was identified in high-cell-concentration fed-batch NS0 myeloma cell culturesproducing a humanized monoclonal antibody (MAb). Inthese cultures, the maximum viable and total cell con-centration was generally ca. 5 × 109 and 7 × 109 cells/L,respectively, without phosphate feeding. Depletion of es-sential amino acids, such as lysine, was initially thoughtto cause the onset of cell death. However, further im-provement of cell growth was not achieved by feeding astoichiometrically balanced amino acid solution, whicheliminated depletion of amino acids. Even though ahigher cell viability was maintained for a longer period,no increase in total cell concentration was observed. Af-terwards, phosphorus was found to be depleted in thesecultures. By also feeding a phosphate solution to elimi-nate phosphorus depletion, the cell growth phase wasprolonged significantly, resulting in a total cell concen-tration of ca. 17 × 109 cells/L, which is much greater thanca. 7 × 109 cells/L without phosphate feeding. The maxi-mum viable cell concentration reached about 10 × 109

cells/L, twice as high as that without phosphate feeding.Apoptosis was also delayed and suppressed with phos-phate feeding. A nonapoptotic viable cell population of6.5 × 109 cells/L, as compared with 3 × 109 cells/L withoutphosphate feeding, was obtained and successfully main-tained for about 70 h. These results are consistent withthe knowledge that phosphorus is an essential part ofmany cell components, including phospholipids, DNA,and RNA. As a result of phosphate feeding, a muchhigher integral of viable cell concentration over time wasachieved, resulting in a correspondingly higher MAb titerof ca. 1.3 g/L. It was also noted that phosphate feedingdelayed the cell metabolism shift from lactate productionto lactate consumption typically observed in recombi-nant NS0 cultures. The results highlight the importanceof phosphate feeding in high-cell-concentration NS0 cul-tures. © 2000 John Wiley & Sons, Inc. Biotechnol Bioeng 69:

566–576, 2000.

Keywords: NS0 cells; fed-batch culture; glutamine syn-thetase; phosphorus; lactate consumption; cell metabo-lism; apoptosis; monoclonal antibody

INTRODUCTION

Mammalian cell culture has been used increasingly for theproduction of therapeutic and diagnostic proteins. Since the1980s, many therapeutic proteins expressed by mammaliancells have been marketed with total annual sales in 1997 ofat least 4 billion U.S. dollars (Thayer, 1998). Among thecommercialized protein products, there are at least 12monoclonal antibodies (MAbs). High MAb dosages (e.g., of2 to 15 mg/kg of body weight) are often required for thera-peutic efficacy. Thus, lowering the cost of manufacturingMAbs is not only highly desirable, but also necessary tomeet commercialization challenges.

Recent years have witnessed intensive bioprocessing re-search to enhance productivity (Bedard et al., 1997; Bibilaand Robinson, 1995; Bibila et al., 1994; Bushell et al., 1994;DiStefano et al., 1996; Honda et al., 1998; Lenas et al.,1997; Robinson et al., 1994, 1995; Ryu and Lee, 1999; Xieand Wang, 1996; Zhou et al., 1995, 1997a). Notable is therelentless pursuit of either a perfusion or fed-batch processthat results in an increased cell concentration and extendedculture lifetime. Efficient perfusion and fed-batch culturestrategies have been developed in parallel with cell line andculture medium optimization (Bebbington et al., 1992;Gould et al., 1992; Itoh et al., 1995; Sato et al., 1988; Singhet al., 1997; Simpson et al., 1998; Xie and Wang, 1996;Zhou et al., 1995). Fed-batch culture offers various advan-tages over other culture operations such as batch, perfusion,and continuous. It is easy to implement, and allows a highproduct concentration and high yield of product on medium.However, its performance depends strongly upon the effec-tiveness of the feeding strategy.

In general, feeding strategy development has involved theuse of stoichiometric analysis to design concentrated nutri-ent solutions (without inorganic components) that are fed

Correspondence to:W. Zhou, P.O. Box 4, WP17-201, 770 SumneytownPike, West Point, PA 19486. Telephone: 215-652-7431; fax: 215-993-4884; e-mail: [email protected]

© 2000 John Wiley & Sons, Inc.

based on process parameters such as cell concentration,oxygen uptake rate (OUR), and glucose concentration. Forexample, in a hybridoma fed-batch culture, an optimizedsalt-free nutrient concentrate was fed based on on-line mea-surements of OUR in order to maintain the glucose andglutamine concentrations at low, but not limiting, levels (Huet al., 1998; Zhou et al., 1995, 1997b). Even though a highviable cell concentration (six-fold greater than in batch cul-ture) was maintained for a long period of time with theseimprovements, the maximum total and viable cell concen-trations were not increased further. This suggests that othernutrients not included in the feeding solution were limiting.These limitations seem to have been eliminated in a hybrid-oma fed-batch culture that employed a stoichiometric feed-ing strategy in which dialyzed serum, phosphate, and tracesalts were also fed (Xie and Wang, 1996). The viable (17 ×109 cells/L) and total (50 × 109 cells/L) cell concentrationswere increased by about twofold over a previous effort thatdid not include these additions. However, cell viability wasbelow 50% for most of the culture (100 to 549 hours). Indesigning the feeding strategy, the total phosphate require-ment per cell was assumed to be twice the phosphate re-quirement for DNA and RNA synthesis (0.084 mmol/109

cells) (Xie and Wang, 1996).The work presented here is a continuation of previous

efforts to develop a highly productive fed-batch culture us-ing an amplified glutamine synthetase (GS)-expressing NS0cell line for the production of MAb against CD18 (Bibila etal., 1994; Bibila and Robinson, 1995; DiStefano et al.,1996; Robinson et al., 1994, 1995; Zhou et al., 1997a).Initial optimization efforts focused on using salt-free com-plete medium concentrates (Bibila et al., 1994) or concen-trated solutions of key nutrients (Robinson et al., 1994,1995) and a simple feeding rate strategy. Subsequently,apoptosis was found to be the main mechanism of cell deathin these NS0 fed-batch cultures. DiStefano et al. (1996)showed that nutrient feeding delayed and reduced the oc-currence of apoptosis in NS0 cultures. Some groups havetransfected cell lines with anti-apoptotic genes such as Bcl-2in an effort to suppress the occurrence of apoptosis in cul-ture (Simpson et al., 1998). Unfortunately, overexpressionof particular anti-apoptotic genes is not effective in everycell type. For example, overexpression of Bcl-2 did notprolong the culture lifetime of an NS0 myeloma cell line(Murray et al., 1996). Regardless, overexpression of anti-apoptotic genes offers protection, but does not eliminate celldeath completely. Therefore, eliminating the stress duringculture that induces apoptosis may be a more practical so-lution.

Realizing the importance of avoiding depletion of essen-tial nutrients, a feedback nutrient control mechanism wasimplemented based on the integral of viable cell concentra-tion over time (IVC), also known as the viable cell index(Zhou et al., 1997a). A maximum viable cell concentrationof about 6 × 109 cells/L and a final IVC of 1.6 × 1012

cells? h/L were achieved. In combination with the high pro-ductivity of the clone, an unprecedented high MAb concen-

tration of >2.7 g/L was obtained. However, the maximumtotal cell concentration was limited to ca. 9 × 109 cells/L. Itshould be noted that no phosphate was fed. So far, regard-less of the feeding strategy used, the nutrients usually fedare glucose, amino acids, and other critical nutrients such asinsulin, cholesterol, and vitamins. Phosphate and other saltsare often not included in order to avoid increasing osmolal-ity in the culture.

Although carbon, hydrogen, nitrogen, and oxygen aremore abundant, phosphorus is also essential for life(Mathews and van Holde, 1990). Phosphorus is present innucleotides and important coenzymes, such as nicotinamideadenine dinucleotide (NAD), derived from nucleotides(Bailey and Ollis, 1986). Phosphorus also plays an essentialrole in biological energy transfer because adenosine triphos-phate (ATP) is the major form of chemical energy currencyin all cells. Phospholipids constitute a significant fraction ofthe cell membrane lipids (Mathews and van Holde, 1990);therefore, phosphorus availability can limit the total cellnumber that can be achieved during culture. In an earlystudy for formulation of a chemically defined culture me-dium for mouse fibroblast cell clone 929-L, the phosphorusrequirement for synthesis of cell mass was determined to be0.2 mmol/109 cells (Higuchi, 1970). This may explain whyfeeding strategy optimization efforts that did not includephosphate feeding extended the culture lifetime, but did notincrease the total cell concentration.

Our objective was to further characterize cellular nutrientrequirements, cell metabolism, and the mechanism of celldeath in NS0 fed-batch culture. New feeding solutions weredesigned based on stoichiometric nutritional analysis,whereas on-line measurement of OUR was used to controlthe nutrient feeding rate. The importance of phosphate feed-ing for further improvement of high-cell-concentration cul-ture performance is highlighted. Also, the role that phos-phate may play in the characteristic metabolic shift in re-combinant NS0 fed-batch cultures from lactate productionto lactate consumption is discussed.

MATERIALS AND METHODS

Cell Line and Culture Conditions

An amplified NS0 cell line that expresses GS and a human-ized MAb against CD18 (hIB4) was used (Robinson et al.,1995). The cell line, culture medium, feeding solutions, andculture conditions have been described previously (Zhou etal., 1997a). All cultures were inoculated using cells from1-L spinner flasks and performed in a 2-L Biostat (B. BraunBiotech, Inc., Allentown, PA). The initial fed-batch culturesemployed the same concentrated amino acid feeding solu-tion used previously (Zhou et al., 1997a). This concentratewas reformulated for later cultures based on analysis ofcellular nutrient requirements. In addition, a phosphate so-lution (20 g/L NaH2PO4) was fed to study the effects ofphosphate feeding on cell growth, cell metabolism, celldeath, and MAb productivity.

DEZENGOTITA ET AL.: PHOSPHATE FEEDING IMPROVES NS0 CULTURE PERFORMANCE 567

Off-line Analysis

Samples were taken twice a day for off-line analysis. Cul-ture broth was immediately injected into a blood gas ana-lyzer (Ciba Corning) to measure pH, dissolved oxygen(DO), and dissolved CO2. Cell concentration and viabilitywere determined using the trypan blue dye exclusionmethod with a hemacytometer.

A fluorescence assay with a dye solution containing 100mg/mL acridine orange and 100mg/mL ethidium bromide(Molecular Probes, Eugene, OR) in phosphate-buffered sa-line (PBS) was used to identify apoptotic cell populations(Mercille and Massie, 1994). Eight microliters of the dyesolution was added to 60mL of cell suspension. A minimumof 200 cells were counted using a hemacytometer. Fourdifferent populations were identified: (i) V-NA, viablenonapoptotic; (ii) V-A, viable cells undergoing apoptosis;(iii) NV-A, nonviable cells that are apoptotic; (iv) N, ne-crotic cells. In general, good agreement was obtained be-tween the viable populations determined using the trypanblue and fluorescence assays (data not shown).

Culture broth was centrifuged and assayed immediatelyfor glucose, lactate, ammonia, and phosphate on a KodakBiolyzer (Eastman Kodak Co., Rochester, NY). For phos-phate measurement, the analyzer employs two chemical re-actions and colorimetric detection. First, phosphate reactswith ammonium molybdate at pH 4.2 to form ammoniumphosphomolybdate, which reacts withp-methylamino-phenol sulfate in the second reaction to form heteropoly-molybdate blue. The final product is then detected at awavelength of 660 nm. Phosphate concentration is reportedas elemental phosphorus concentration. Osmolality wasmeasured using a vapor pressure osmometer (Wescor Inc.,Logan, UT). Supernatant was frozen and stored for aminoacid and MAb analysis. Amino acid concentrations weremeasured using an orthophthalalehyde/9-fluorenylmethyl-chloroformate (OPA/FMOC) derivatization method with anHP 1090 HPLC system (Hewlett Packard, Boise, ID). MAbconcentration was determined with affinity chromatographyusing high purified hIB4 MAb as standard. A Poros ProteinA column (PerSeptive Biosystems, Framingham, MA) wasused to specifically bind the MAb. Samples were loadedwith a phosphate buffer and eluted by decreasing the pH ofthe mobile phase from 7.2 to 2. The detection wavelengthwas 280 nm.

Feeding Strategies

Three feeding strategies were used: Initial feeding strategy(FS-I)—a concentrated amino acid solution was fed off-linewithout phosphate feeding; New amino acid solution feed-ing strategy (FS-N)—an optimized amino acid solution wasfed off-line without phosphate feeding; and Phosphate feed-ing strategy (FS-P)—phosphate was fed off-line and theoptimized amino acid solution was fed on-line based onOUR, which was measured dynamically once every hourusing a procedure described previously (Zhou and Hu,

1994; Zhou et al., 1997a). In addition, concentrated solu-tions of glucose, nucleosides, bovine insulin, bovine trans-ferrin, bovine low-density lipoprotein (Excyte VLE), bovineserum albumin (Albumax), ethanolamine,b-mercaptoetha-nol, Pluronic-F68, and methionine sulfoximine (MSX; tomaintain selection pressure) were combined at each time-point and fed off-line based on experimentally determinedcell yields. Nutrient feeding was initiated at about 50 h in allfed-batch cultures.

During fed-batch cultures FS-I and FS-N, feeding wasperformed based on the forward calculation of IVC assum-ing a constant specific growth rate, specific uptake rates,and cell yields over time. The feeding volume of a particularnutrient solution was calculated using the following equa-tion:

Vi =qiVc

Ci*

tn

tn+1xv dt (1)

where Vi is the volume of the concentrated solution ofirequired at timetn, qi is the specific consumption rate ofnutrienti, Ci is the concentration ofi in the feeding solution,Vc is the reactor volume at timen, andxv is the viable cellconcentration. A constant specific growth rate of 0.02 h−1

was used in the cell growth phase to project the viable cellconcentration for timetn+1 when the next feeding wouldtake place:

(xv)n+1 4 (xv)n exp(0.02 × (tn+1 − tn)) (2)

During the stationary and death phases, the same viable cellconcentration was assumed at timetn and tn+1.

When using the feeding strategy FS-P, the stoichiomet-rically balanced amino acid solution was fed on-line basedon OUR measurements. There is a direct correlation be-tween viable cell concentration and OUR during the growthphase of NS0 fed-batch cultures (Zhou et al., 1997). Also, adirect correlation was observed between the specific OURand specific amino acid consumption rates. Amino acidswere fed on-line every hour that OUR measurements tookplace according to the following equation:

VAA feed ~n! =Vc × OUR × Dt

aCfeed(3)

whereVAA feed (n) is the volume of amino acid solution fedat timetn, Vc is the reactor volume at timetn, Dt is 1 h, andCfeed is the concentration of a representative amino acid inthe amino acid solution. Leucine was selected as the repre-sentative amino acid, even though others could have beenselected because the amino acid solution is stoichiometri-cally balanced.a is the stoichiometric feeding coefficient(mole oxygen per micromole leucine). The reactor volumewas updated on-line after each feeding. All other feedingsolutions including phosphate were combined and fed off-line. The sample size was adjusted to compensate for theoff-line additions. Therefore, the reactor volume was onlyaltered by on-line feeding of the amino acid solution.

568 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 69, NO. 5, SEPTEMBER 5, 2000

Estimation of Specific Rates

The specific cell growth rate (m); the specific consumptionrates of oxygen, glucose, and the amino acids; the specificlactate production or consumption rate; the specific ammo-nia production rate; and the specific MAb production ratewere calculated using the following mass balance equations:

@SV#t = SoVo + *0

tFinSfeeddt − *

0

t~FoutS!tdt − *

0

tqs~xvV!tdt

(4)

*0

tFinSfeeddt = Sfeed(Vfeed and

*0

t~FoutS!tdt = (~SVsample! (5)

SoVo + Sfeed(Vfeed− (~SVsample! − @SV#t = qs*0

t~xvV!tdt

(6)

whereS is the concentration of viable cells, substrate, orproduct;V is the reactor volume;Sfeed is the concentrationof S in the feeding solution;Fin andFout are volumetric flowrates into and out of the reactor, respectively;Vfeed is thefeeding volume;Vsample is the volume removed from theculture; andqs is the specific growth, consumption, or pro-duction rate of the correspondingS. The rates were deter-mined by plotting the left-hand side of Eq. (6) vs. the inte-gral of (xV V) over time. These plots were fitted with astraight line with a linear regression coefficient of about 1.The slope of the line corresponds to the average specificrate. In some cases, the specific rates changed throughoutthe culture and several lines were fitted. Specific oxygenuptake rate was determined by dividing the online OURvalue by the viable cell concentration at each timepoint. Theapoptotic cell death rate was determined as in DiStefano etal. (1996). Standard error propagation was used in the errorcalculation for the metabolic rates (Glantz, 1981).

RESULTS

Because feeding was done based on the cell concentration(and OUR in FS-P) observed at each timepoint, the concen-trations of cells and metabolites are presented without ad-justing for dilution due to nutrient feeding. Nutrient feedingand volume removal are accounted for in the equations usedto calculate specific growth rate and specific metabolic ratesfor direct comparison among cultures. It should be notedthat the improvement in cell concentration achieved withFS-P would be even greater if dilution had been taken intoaccount.

NS0 Fed-Batch Culture withFeeding Strategy FS-I

Several NS0 fed-batch cultures were performed using feed-ing strategy FS-I. The amino acid feeding solution was de-signed based on amino acid uptake rates in NS0 batch cul-ture (Robinson et al., 1994). These NS0 fed-batch culturesusing FS-I did not yield cell growth and death results dif-

ferent from those obtained previously (Zhou et al., 1997a).In a typical FS-I culture, a total of 0.35 L was fed to aninitial volume of 1.5 L. Cell death occurred after a maxi-mum viable cell concentration of ca. 5 × 109 cells/L wasreached at about 100 h (data not shown). At the onset of celldeath, lactate became consumed, and ammonia and gluta-mine accumulated (data not shown). As shown in Figure 1,phosphorus (which was not initially measured) and severalamino acids were depleted during the cultivation: phospho-rus (95 h), lysine (95 h), methionine (112 h), cystine (133h), and glycine (133 h). The fact that these amino acidsbecame limiting, while others accumulated, indicated thatthe amino acid feeding solution was not stoichiometricallybalanced.

The specific uptake rates for amino acids in FS-I cultureare shown in Table I. Based on these amino acid uptakerates, IVC values that could be supported by the amino acidconcentrations in the old amino acid (AA) feeding solutionwere calculated. It can be seen that the concentrations of thefour limiting amino acids could support only a relativelylow IVC. Feeding this AA solution more frequently to keepthese four amino acids (highlighted in bold in Table I) frombecoming limiting would result in the accumulation of otheramino acids that are present at concentrations that can sup-port higher IVC. To prevent this, a stoichiometrically bal-anced amino acid solution (the new AA solution) was de-veloped based on the FS-I uptake rates, so that all aminoacids in the new AA solution would support the same IVC.This solution was used in feeding strategies FS-N and FS-Pto eliminate amino acid limitation as a cause of cell death inNS0 cultures, while avoiding excessive feeding.

NS0 Fed-Batch Culture withFeeding Strategy FS-N

Cell Growth, Cell Death, and MAb Productivity

In fed-batch culture with feeding strategy FS-N, which usedthe new stoichiometrically balanced amino acid solution, a

Figure 1. Concentration profiles for phosphorus (h) and the amino acidsthat became limited in a representative NS0 fed-batch culture with feedingstrategy FS-I: cystine (j); glycine (s); lysine (l); and methionine (n).


total of 0.66 L was fed to an initial volume of 1.5 L. Aminoacid concentrations were not lower than their initial valuesnor higher than 3 mM (data not shown). Cells grew expo-nentially to a viable cell concentration of 5.5 × 109 cells/Lat ca. 90 h with viability of >90% (Fig. 2), while phosphoruswas rapidly consumed during this time period and becameexhausted at ca. 90 h. This maximum viable cell concen-tration is similar to that obtained with the old AA solution.Even though no further cell growth was observed, likely dueto phosphate limitation, an average viable cell concentration

of 5 × 109 cells/L (trypan blue method) was maintained forabout 55 h.

Apoptosis was confirmed to be the main mechanism ofcell death in all of the NS0 cultures. In the FS-N culture, theviable cell population undergoing apoptosis increased atabout 100 h (Fig. 3). The apoptotic nonviable populationincreased 24 h after apoptosis was detected. The apoptoticcell death rate was 0.011 h−1 between 89 and 500 h. Eventhough the viable cell population was maintained for about

Table I. Comparison of the old amino acid solution used in feeding strategy FS-I and the stoichiometrically balanced amino acid solution used in feedingstrategies FS-N and FS-P.a

Amino acids

Specific AA consumptionrates in FS-I cultures(10−9 mmol/cell ? h)

Old AA solution(mmol/L)

Theoretical IVC withold AA solution(1010 cells? h/L)

New AA solution(mmol/L)

ARG 4.1 ± 0.3 119 2.9 278ASN 10.2 ± 0.1 1136 11.2 681ASP 2.5 ± 0.9 75 3.0 170CYS* 3.1± 0.6 64 2.0 210GLU 14.0 ± 1.1 444 3.2 936GLY * 0.74± 0.00 0 0.0 50HIS 1.1 ± 0.2 36 3.4 72ILE 8.0 ± 0.2 248 3.1 533LEU 8.7 ± 0.5 248 2.8 586LYS* 4.2± 0.2 55 1.3 279MET * 1.71± 0.04 34 2.0 115PHE 2.1 ± 0.1 61 3.0 138PRO 3.7 ± 1.4 65 1.8 246SER 5.1 ± 0.3 190 3.7 342THR 3.6 ± 0.5 168 4.6 243TRP 0.73 ± 0.01 20 2.7 49TYR 2.1 ± 0.3 67 3.1 144VAL 7.4 ± 0.2 256 3.5 495

aThe specific amino acid consumption rates are presented as the average from two fed-batch cultures ± SD. The amino acids in bold with an asteriskbecame limiting during the FS-I cultures. The new amino acid solution could theoretically support an IVC of 6.7 × 1010 cells? h/L.

Figure 2. Viable cell concentration (d), total cell concentration (s), andMAb concentration (m) profiles in the NS0 fed-batch culture with feedingstrategy FS-N with a stoichiometrically balanced amino acid solution andwithout phosphate feeding. Phosphorus (h) became limited in this cultureat about 100 h. The concentrations are presented without adjusting fordilution effects. This culture was divided into four phases: N1 (0 to 89 h);N2 (89 to 145 h); N3 (145 to 235 h); and N4 (235 to 575 h).

Figure 3. Distribution of different cell populations in the NS0 fed-batchculture with feeding strategy FS-N. The number of cells undergoingapoptosis increased when phosphorus became limited. Symbols: viablenon-apoptotic (V-NA,m); viable apoptotic (V-A,n); non-viable apoptotic(NV-A, s); and necrotic (N,d).


55 h (Fig. 2), a constant viable non-apoptotic populationwas not observed (Fig. 3).

The final MAb concentration was 0.5 g/L (Fig. 2) with afinal IVC of 1.1 × 1012 cells? h/L. The specific MAb pro-duction rate (qMAb) remained constant at 4.8 (±0.02) × 10−13

g/cell ? h.

Cell Metabolism

To facilitate presentation of the different metabolic shifts,the FS-N culture was divided into four phases according tothe viable cell concentration profile (Fig. 2): exponentialgrowth (N1: 0 to 89 h); approximately constant viable cellconcentration (N2: 89 to 145 h); decline in viable cell con-centration (N3: 145 to 235 h); and extensive cell death (N4:235 to 575 h).

Selected metabolic parameters are shown in Figure 4.Glucose was maintained above 3 mM throughout the culture(data not shown). Three shifts in lactate metabolism were

observed (Fig. 4A). Lactate reached a concentration of 5mM in phase N1 where it was produced with a specific rate(qlac) of 1.6 × 10−11 mmol/cell? h. It then became consumedin phase N2 at a rate of 1.3 (±0.1) × 10−11 mmol/cell? h. Inphase N3, there was no significant net production or con-sumption of lactate. In phase N4, lactate was producedagain, especially after 350 h. The specific glucose consump-tion rate (qglc) decreased by 48%, from 2.9 (±0.9) × 10−11

mmol/cell? h in phase N1 to 1.5 (±0.1) × 10−11 mmol/cell ? h in phase N2. The next significant shift inqglc was anincrease when lactate was again produced in phase N4 (datanot shown). The specific oxygen consumption rate (qO2

)decreased during the first two phases (Fig. 4A). When lac-tate consumption stopped, an increasing trend inqO2

wasobserved. It is unclear why an increase inqglc andqO2

oc-curred during N4. It may be due to the accumulation ordepletion of other components that were not measured. Theratio of qO2

to qglc remained >6 throughout the culture afterphase N1. This suggests that most of the glucose consumedwas completely oxidized for energy production. Glutamineand ammonia did not accumulate significantly in phase N1(Fig. 4B), indicating a balance between glutamine synthesisand consumption by the cells. Both glutamine and ammoniastarted to accumulate when lactate was consumed. Gluta-mine stopped accumulating when lactate was again pro-duced in phase N4 (216 h). At 400 h, ammonia and gluta-mine reached concentrations of 6.1 and 1.4 mM, respec-tively. Alanine was produced in the first two phases. Therewas little change in alanine concentration in phase N3, butalanine was produced again in phase N4.

NS0 Fed-Batch Culture withFeeding Strategy FS-P

When phosphorus was measured retrospectively in FS-I andFS-N culture samples, it was found that phosphorus consis-tently became depleted when there was a sharp decrease inviable non-apoptotic cell concentration (Figs. 1 to 3). Themaximum cell concentration was likely limited by the avail-able phosphorus amount of ca. 25 mg/L in the basal medium(Iscove’s modified Dulbecco’s medium). In fact, theamount of phosphorus consumed per cell was found to beessentially the same in all three FS-I and FS-N cultures. Byplotting phosphorus consumed vs. total cell concentration,the phosphorus requirement was determined at slightly lessthan 5 mg/109 cells on average (Fig. 5). As a result, aphosphate solution was also fed to keep the phosphorusconcentration around the initial level of the culture medium(ca. 25 mg/L) throughout fed-batch culture using feedingstrategy FS-P (Fig. 6). A total of 1.6 L was fed in thisculture with an initial volume of 1.2 L.

Cell Growth, Cell Death, and MAb Productivity

By eliminating phosphorus limitation, the cell growth phasewas prolonged significantly and a viable cell concentrationof ca. 10 × 109 cells/L was obtained at 180 h with viability

Figure 4. Metabolic profiles in the NS0 fed-batch culture with feedingstrategy FS-N without phosphate feeding. (A) Lactate concentration (j)and specific oxygen consumption rate (n). (B) Alanine (h), glutamine(s), and ammonia (d) concentrations.


>85% (Fig. 6). The apparent specific cell growth rate (0.022h−1) was similar with and without phosphate feeding. Anaverage of 9 × 109 viable cells/L (trypan blue method) wasmaintained for about 70 h. The total cell concentrationreached was about 17 × 109 cells/L, which was much highercompared with ca. 7 × 109 cells/L in FS-N without phos-phate feeding. As for the FS-N culture, a direct correlationbetween the total cell concentration and the amount of phos-phorus consumed was obtained (Fig. 7). The phosphorusrequirement was not affected by phosphate feeding, and wasdetermined to be 4.9 mg or 0.16 ± 0.02 mmol per 109 cells.

Figure 8 shows the distribution of different cell popula-tions. The apoptotic cell death rate after exponential growth(178 to 500 h) was 0.011 h−1, similar to that without phos-phate feeding. However, a cell growth rate of 0.005 h−1 was

calculated over the same period for the nonapoptotic popu-lation in contrast to 0.001 h−1 in the FS-N culture. As aresult, a non-apoptotic viable cell concentration of 6.5 × 109

cells/L was maintained for approximately 100 h (130 to 230h), even though the apoptotic non-viable population steadilyincreased after 150 h (Fig. 8). This was a significant im-provement over the FS-N culture where the non-apoptoticviable cell concentration steadily decreased.

A final MAb concentration of ca. 1.3 g/L was obtained(Fig. 6) with a final IVC of 2.9 × 1012 cells? h/L. A shift inqMAb was observed. TheqMAb was 4.01 (±0.02) × 10−13

g/cell ? h until 180 h, when the maximum viable cell con-centration was reached. Thereafter,qMAb increased slightlyand remained at 5.5 (±0.05) × 10−13 g/cell ? h for the rest ofthe culture. TheseqMAb values, as well as those obtained inthe FS-I and FS-N cultures, are low relative to the specific

Figure 5. The cell yield from phosphorus was determined in NS0 fed-batch cultures with feeding strategies that did not include phosphate feed-ing. FS-I [(d) 4.9 mg per 109 cells; (n) 4.9 mg per 109 cells] and FS-N[(l) 4.7 mg per 109 cells]. On average, slightly less than 5 mg (0.16 mmol)of phosphorus was consumed per 109 cells.

Figure 6. Viable cell concentration (d), total cell concentration (s),MAb concentration (m), and phosphorus concentration (h) profiles in theNS0 fed-batch culture with feeding strategy FS-P with a stoichiometricallybalanced amino acid solution and phosphate feeding. The concentrationsare presented without adjusting for dilution effects. This culture was di-vided into four phases: P1 (0 to 178 h); P2 (178 to 300 h); P3 (300 to 370h); and P4 (370 to 826 h).

Figure 7. Profiles of the total cell density (m and d) and cumulativephosphorus consumption (n and s) in the NS0 fed-batch culture withfeeding strategy FS-N (triangles) and FS-P (circles).

Figure 8. Distribution of different cell populations in the NS0 fed-batchculture with feeding strategy FS-P. Symbols: viable non-apoptotic (V-NA,m); viable apoptotic (V-A,n); and non-viable apoptotic (NV-A,s); andnecrotic (N,d).


productivity of 1.3 to 1.7 × 10−12 g/cell ? h observed previ-ously in batch and fed-batch cultures with this cell line(Bibila et al., 1994; Robinson et al., 1995; Zhou et al.,1997a).

Cell Metabolism

The FS-P culture was also divided into four phases accord-ing to the viable cell concentration profile (Fig. 6): expo-nential growth (P1: 0 to 178 h); approximately constantviable cell concentration (P2: 178 to 300 h); decline inviable cell concentration (P3: 300 to 370 h); and extensivecell death (P4: 370 to 826 h).

Selected metabolic parameters are shown in Figure 9.Similar to the FS-N culture, glucose was maintained above3 mM (data not shown) and the same three shifts in lactatemetabolism were observed (Fig. 9A). Lactate was producedin phase P1 and then became consumed in phase P2. In

comparison with the FS-N culture, the onset of lactate con-sumption was delayed by 50 h.qlac shifted from productionat 2.8 (±0.1) × 10−11 mmol/cell? h in P1 (67% above that ofFS-N) to consumption at 5.4 (±0.2) × 10−11 mmol/cell? h inP2 (60% below that of FS-N). There was no significant netproduction or consumption of lactate in P3. Finally, lactatewas produced again in P4.qglc decreased during lactateconsumption by 38%, from 4.8 (±0.3) × 10−11 mmol/cell? hin P1 to 3.0 (±0.1) × 10−11 mmol/cell? h in P2. An increasein qglc was again observed when lactate was produced againin P4. In the first phase of both the FS-N and FS-P cultures,glucose was partly converted to lactate with a ratio of 0.6mol/mol, comparable to the ratio of 0.8 mol/mol observedpreviously in other cultures with this cell line (Zhou et al.,1997a). After an initial increase,qO2

decreased in phase P1and did not change when lactate was consumed. When lac-tate consumption ceased in P3, an increase inqO2

was ob-served. The ratio ofqO2

to qglc decreased from 5 in phase P1to 3 in phase P2 when lactate was consumed. This ratiodecrease was not observed in the FS-N culture. Alanine,glutamine, and ammonia concentrations are shown in Fig-ure 9B. The alanine concentration profile was similar to thatfor lactate, but with a delay of about 50 h before the onsetof alanine production or consumption. Glutamine and am-monia did not accumulate significantly throughout the FS-Pculture, with values of only 0.3 and 1.1 mM, respectively, at400 h (Fig. 9B).

DISCUSSION

During the course of development and optimization of nu-trient feeding strategies for NS0 fed-batch cultures, nutrientlimitation has been found to be the major cause of celldeath. Initially, four amino acids were found depleted infed-batch cultures using an amino acid solution developedbased on consumption rates in batch culture (FS-I). Using astoichiometrically balanced amino acid solution, amino aciddepletion was eliminated as a cause of cell death (FS-N).However, there was no significant improvement in total orviable cell concentration (Table II). The fact that cell growthwas not significantly improved suggests that some other

Figure 9. Metabolic profiles in the NS0 fed-batch culture with feedingstrategy FS-P with phosphate feeding. (A) Lactate concentration (j) andspecific oxygen consumption rate (n). (B) Alanine (h), glutamine (s),and ammonia (d) concentrations.

Table II. Comparison of the effect of the three different feeding strate-gies on cell concentration and specific MAb productivity in the NS0 fed-batch cultures.

Feeding strategy FS-I FS-N FS-P

Amino acid limitation Yes No NoPhosphorus limitation Yes Yes NoMaximum xv (109 cells/L) 5 6 10Maximum xt (109 cells/L) 6 7 17Duration of growth phase (h) 90 90 180Cell yield on phosphorus

(mg/109 cells)4.9 ± 0.2 4.7 ± 1.3 4.9 ± 0.4

Specific MAb productivity(10−13 g/cell ? h)

6.9 ± 0.3 4.8 ± 0.02 5.5 ± 0.05


nutrient(s) may have been limiting the cell growth. In fact,phosphorus was retrospectively found to be depleted whenthe maximum cell concentration was reached in the FS-Iand FS-N cultures.

Inclusion of phosphate in the feeding strategy (FS-P) re-sulted in a significant extension of the cell growth phase andculture lifetime. The maximum viable cell concentrationand the final IVC were 10 × 109 cells/L and 2.9 × 1012

cells? h/L, respectively, at 830 h, resulting in a final MAbconcentration of ca. 1.3 g/L. This was a significant improve-ment over 5 × 109 viable cells/L, a final IVC of 1.1 × 1012

cell ? h/L at 575 h, and a final MAb concentration of 0.5 g/Lwithout phosphate feeding in the FS-N culture (Table II), aswell as other previously reported NS0 fed-batch cultures(Bibila et al., 1994; Robinson et al., 1995; Zhou et al.,1997a). In the best NS0 fed-batch culture reported previ-ously (Zhou et al., 1997a), a lower maximum viable cellconcentration of 6 × 109 cells/L and a lower final IVC of 1.6× 1012 cells? h/L at 672 h were obtained, even though thefinal MAb concentration of 2.7 g/L was higher due to ahigher specific productivity. No explanation has been foundfor this lower productivity, which was observed in all of thecultures (batch and fed-batch) performed using the samecell source during the time period of the work presentedhere. Once this discrepancy was observed, the productivitydecrease was confirmed by assaying the samples from theprevious study (Zhou et al., 1997a) and from this study atthe same time using the highly purified standard. The pro-ductivity decrease may be due to long-term storage of cells,which may have had negative effects on cell line stability, asthe cells were derived from the same cell bank and thawed1 to 2 years apart.

The IVC was used as the basis for nutrient feeding be-cause it is relatively simple to use. It is especially useful foradjusting feeding rates for nutrients with a very low con-sumption rate. However, it is not suitable for very frequentadjustments. It may also result in fluctuating nutrient con-centrations. The on-line feeding strategy based on OURmeasurements can overcome these disadvantages. OUR hasbeen shown to be a sensitive parameter for monitoring themetabolic activity and state of a culture. Because there wasa direct correlation between the oxygen consumption rateand the amino acid consumption rates, online OUR mea-surements were used successfully to maintain concentra-tions of amino acids at about initial levels. Overfeeding,which may result in excessive and inhibitory high levels ofnutrients and significant increases in osmolality, and nutri-ent depletion were avoided by use of a combination of thetwo parameters. Because the feeding strategy FS-N, whicheliminated amino acid limitation as the cause of cell death,did not result in a significant improvement in viable cell ortotal cell concentration, it is clear that the improvements intotal cell and viable cell concentration observed with phos-phate feeding cannot be attributed to the use of an on-lineamino acid feeding strategy rather than inclusion of phos-phate feeding. Also, the same cell yield on phosphorus was

obtained regardless of the nutrient feeding strategy em-ployed in the NS0 fed-batch cultures (Table II).

Culture lifetime and productivity are a function of theviable non-apoptotic cell population. The purpose of devel-opment and optimization of feeding strategies is to maxi-mize and maintain this population by preventing the onsetof apoptotic death. While feeding of limiting nutrients, suchas amino acids, delayed and suppressed apoptosis inducedby starvation of these nutrients in hybridoma and NS0 cul-tures (DiStefano et al., 1996; Franeˆk, 1995; Franeˆk andSramkova, 1997), additional phosphate feeding further de-layed the onset of apoptosis in the FS-P culture.

Phosphorus is essential for energy transfer within cells, aswell as for the synthesis of nucleic acids and phospholipids.The amount of phosphorus available can limit the total cellnumber that can be achieved during culture. In this work, asimilar requirement of 0.16 mmol phosphorus/109 cells wasdetermined in all NS0 cultures using three feeding strate-gies. This value is similar to the 0.2 mmol/109 cells deter-mined for a mouse fibroblast cell clone 929-L (Higuchi,1970). It is consistent with a phosphorus requirement of 0.1mmol/109 cells in a previous NS0 fed-batch culture withoutphosphate feeding reported previously (Zhou et al., 1997a),which was estimated by dividing the available phosphorusby the total cell concentration achieved. Somewhat lowervalues have been calculated in the same way for a hybrid-oma fed-batch culture (0.068 mmol/109 cells; Zhou et al.,1995) and assumed for development of feeding solutions ina hybridoma fed-batch culture (0.084 mmol/109 cells; Xieand Wang, 1996).

A common observed metabolic shift in NS0 fed-batchcultures, where lactate becomes consumed (Bibila et al.,1994; Zhou et al., 1997a), was also observed in this work.This shift correlates with the end of the exponential growthphase (Figs. 4 and 9), and coincides with the onset of apop-totic cell death, and accumulation of ammonia and gluta-mine. Along with the low lactate yields from glucose ob-served in these fed-batch cultures, lactate consumption sug-gests a low energy contribution from glycolytic metabolism(Table III). This is consistent with the high ratios ofqO2

toqglc and qO2

to qlac observed. A similar observation wasmade for culture of glioma cells, where exogenous lactatewas found to be the major substrate for oxidative metabo-lism (Bouzier et al., 1998).

Phosphate limitation is known to affect gene expressionin and growth of bacteria (Osorio and Jerez, 1996; Seegerand Jerez, 1992; Spector, 1990). Phosphate limitation alsoaffects the metabolism of mammalian cells, such as eryth-rocytes from uremic patients (Bevington et al., 1995). Gly-colytic flux was inhibited by 20% when pH was loweredfrom 7.4 to 7.2 at a normal plasma phosphate concentrationof 1 mM. However, this inhibition was blocked when theintracellular phosphate concentration was increased throughin vitro phosphate addition. This is consistent with the ob-servations that lactate consumption was not only delayed,but also decreased in magnitude with phosphate feeding(Table III).


Several possible explanations exist for lactate consump-tion in NS0 fed-batch cultures. Lactate can be consumed toform pyruvate (Fig. 10). Pyruvate can be converted intoacetyl-CoA, which enters the tricarboxylic acid (TCA)cycle and becomes completely oxidized. For the completeoxidation of 1 mol of pyruvate, 3 mol of oxygen are re-quired. Low lactate yields from glucose, along with the highoxygen to glucose consumption ratio observed in these NS0fed-batch cultures (Table III), suggests that the flux throughthe oxidative pathway is significant. Pyruvate can also beconverted to lactate and to alanine through the transamina-tion pathway where glutamate is converted toa-ketogluta-rate. One mole of pyruvate is required to produce 1 mol oflactate or alanine. The inverse reactions where pyruvate isformed from lactate or alanine can also take place. Whenphosphorus becomes limited, there may be a decrease inATP availability within the cells. In the FS-N culture, thespecific glucose and oxygen uptake rates decreased to dif-ferent extents, whereas the oxygen to glucose consumptionratio increased after the first metabolic shift (Table III).Phosphorus limitation may inhibit the reaction in glycolysisthat converts glyceraldehyde-3-phosphate into 1,3-biphosphoglycerate where NAD+ and triose phosphate arerequired. When this reaction is inhibited, pyruvate becomeslimited and lactate consumption to produce pyruvate may be

induced. NAD+ required for this reaction is regeneratedthrough oxidative phosphorylation. Even though both oxy-gen consumption and pyruvate availability are decreased,the ratio between oxygen and glucose consumption is veryhigh at >6 mmol/mmol, indicating a primarily oxidative cellmetabolism.

Lactate consumption still occurred in the FS-P culturewhere phosphate was not depleted. However, lactate con-sumption was 40% of that observed in the FS-N culture.During lactate consumption,qglc decreased by 38% in theFS-P culture as opposed to the 50% decrease observed inthe FS-N culture, whereasqO2

decreased by 66% as opposedto 50%. This resulted in a decrease in the ratio ofqO2

to qglc

in the FS-P culture. It is of interest to note that alanine wasalso consumed in parallel to lactate consumption in the FS-Pculture, but not in the FS-N culture in which phosphate wasnot fed. Parallel lactate and alanine consumption may fur-ther suggest a limited pyruvate availability when cells weredying despite phosphate feeding. The fact that cells werestill dying and both lactate and alanine were consumed sug-gests that other factors may cause the cell death and me-tabolism shift in this phase. Cell death in the FS-P culturecould have resulted from the accumulation of other metabo-lites and/or nutrients, which are yet to be identified and maybe detrimental to NS0 cells at higher concentrations. Limi-tation of nutrients (e.g., trace metals) that were not fed couldalso have resulted in cell death. Neither ammonia nor lactatehave been built up to levels that could inhibit cell growth orresult in apoptotic cell death (Bibila et al., 1994; Mercilleand Massie, 1994). In addition, pH was maintained constantat 7.2, while no CO2 accumulation was observed throughoutthe culture. The culture osmolality, which ranged from 260to 300 mOsm/kg throughout the culture, did not reach in-hibitory levels of 400 mOsm/kg (Bibila et al., 1994).

In summary, phosphorus depletion was identified in high-cell-concentration fed-batch NS0 cultures producing a hu-manized MAb. The total cell concentration was limited bythe available phosphorus. Phosphate feeding not only sig-nificantly prolonged the cell growth phase and increased themaximum cell concentration, but also extended the culturelifetime. As a result, much higher IVC and MAb produc-tivity were achieved with phosphate feeding than without it.In addition, NS0 cell metabolism was also affected by phos-phate feeding, which delayed the shift from lactate produc-tion to lactate consumption. These results highlight the im-portance of phosphate feeding in high-cell-concentrationNS0 cultures.

Table III. Comparison of some metabolic rates in the NS0 fed-batch cultures with (FS-P) and without (FS-N) phosphate feeding.

Feedingstrategy

Culturephase

Specific glucoseconsumption rate

(10−13 mol/cell ? h)

Specific lactateproduction rate


Specific oxygenconsumption rate


Apparent yieldof lactate from glucose

(mol/mol)

Ratio of oxygento glucose uptake

(mol/mol)

Ratio of oxygento lactate uptake

(mol/mol)

FS-N N1 2.9 ± 0.9 1.6 2.0 0.56 6.9 −12.3N2 1.5 ± 0.1 −1.3 ± 0.1 1.4 — 9.5 +11.0

FS-P P1 4.8 ± 0.3 2.8 ± 0.1 2.5 0.58 5.3 −9.0P2 3.0 ± 0.1 −0.54 ± 0.02 0.84 — 2.8 +15.6

Figure 10. This metabolic pathway illustrates the possible metabolites(enclosed in boxes) that lead to pyruvate. Alanine was consumed during thelactate consumption phase in FS-P. Reactions where pyruvate is used arealso presented.


These experiments were performed by V.M.d.Z. during an in-dustrial internship at Merck Research Laboratories. Northwest-ern University was supported in part by a Predoctoral Biotech-nology Training Grant (NIH GM08449). The authors thankRoger Olewinski for performing the amino acid analyses.

References

Bailey JE, Ollis DF. 1986. Biochemical engineering fundamentals, 2ndedition. New York: McGraw-Hill.

Bebbington CR, Renner G, Thompson S, King D, Abrams D, YarrantonGT. 1992. High level expression of a recombinant antibody frommyeloma cells using a glutamine synthetase gene as an amplifiablemarker. Bio/Technology 10:169–175.

Bedard C, Perret S, Kamen AA. 1997. Fed-batch culture of Sf-9 cellssupports 3 × 107 cells per mL and improves baculovirus-expressedrecombinant protein yields. Biotechnol Lett 19:629–632.

Bevington A, Brough D, Baker FE, Hattersley J, Walls J. 1995. Metabolic-acidosis is a potent stimulus for cellular inorganic-phosphate genera-tion in uremia. Clin Sci 88:405–412.

Bibila TA, Ranucci CS, Glazomitsky K, Buckland BC, Aunins JG. 1994.Monoclonal antibody process development using medium concen-trates. Biotechnol Progr 10:87–96.

Bibila TA, Robinson DK. 1995. In pursuit of the optimal fed-batch processfor monoclonal antibody production. Biotechnol Progr 11:1–13.

Bouzier AK, Voisin P, Goodwin R, Canioni P, Merle M. 1998. Glucoseand lactate metabolism in C6 glioma cells: evidence for the preferen-tial utilization of lactate for cell oxidative metabolism. Devel Neurosci20:331–338.

Bushell ME, Bell SL, Scott MF, Spier RE, Wardell JN, Sanders PG. 1994.Enhancement of monoclonal antibody yield by hybridoma fed-batchculture, resulting in extended maintenance of viable cell population.Biotechnol Bioeng 44:1099–1106.

DiStefano DJ, Mark GE, Robinson DK. 1996. Feeding of nutrients delaysapoptotic death in fed-batch cultures of NS0 myeloma cells. Biotech-nol Lett 18:1067–1072.

Franek K. 1995. Starvation-induced programmed cell death of hybridomacells: prevention by amino acid mixtures. Biotechnol Bioeng 45:86–90.

Franek K, SramkovaK. 1997. Cell suicide in starving hybridoma culture:survival-signal effect of some amino acids. Cytotechnology 23:231–239.

Glantz SA. 1981. Primer of biostatistics. New York: McGraw-Hill.Gould SL, DiStefano DJ, Cuca G, Robinson DK, Silberklang M. 1992.

Major metabolic changes accompany transfection and selection forhigh level expression of recombinant genes. In Vitro Cell Dev Biol28:162A.

Higuchi K. 1970. An improved chemically defined culture medium forstrain L mouse cells based on growth responses to graded levels ofnutrients including iron and zinc ions. J Cell Physiol 75:65–72.

Honda H, Lenas P, Watanabe H, Kitade T, Kobayashi T. 1998. Humanantithrombin III variant production from recombinant BHK cells in afed-batch culture with on-line control of glucose and glutamine con-centrations. J Ferment Bioeng 85:532–535.

Hu WS, Zhou W, Europa LF. 1998. Controlling mammalian cell metabo-lism in bioreactors. J Microbiol Biotechnol 8:8–13.

Itoh Y, Ueda H, Suzuki E. 1995. Overexpression of bcl-2, apoptosis sup-pressing gene: prolonged viable culture period of hybridoma and en-hanced antibody production. Biotechnol Bioeng 48:118–122.

Lenas P, Kitade T, Watanabe H, Honda H, Kobayashi T. 1997. Adaptivefuzzy control of nutrients concentration in fed-batch culture of mam-malian cells. Cytotechnology 25:9–15.

Mathews CK, van Holde KE. 1990. Biochemistry. Redwood City, CA:Benjamin/Cummings.

Mercille S, Massie B. 1994. Induction of apoptosis in nutrient-deprivedcultures of hybridoma and myeloma cells. Biotechnol Bioeng 44:1140–1154.

Murray K, Ang CE, Gull K, Hickman JA, Dickinson AJ. 1996. NS0 my-eloma cell death: influence ofbcl-2 overexpression. Biotechnol Bio-eng 51:298–304.

Osorio G, Jerez CA. 1996. Adaptive response of the archaeonSulfolobusacidocaldariusBC65 to phosphate starvation. Microbiology 142:1531–1536.

Robinson DK, DiStefano DJ, Gould SL, Cuca G, Seamans TC, BenincasaD, Munshi S, Chan CP, Lee DK, Stafford-Hollis J, Hollis GF, Jain D,Ramasubramanyan K, Mark GE, Silberklang M. 1995. Production ofengineered antibodies in myeloma and hybridoma cells—enhance-ments in gene expression and process design. In: Wang H, Imanaka T,editors. Antibody engineering. American Chemical Society sympo-sium series 604. Washington, DC: ACS. p 1.

Robinson DK, Seamans TC, Gould SL, DiStefano DJ, Chan CP, Lee DK,Bibila TA, Glazomitsky K, Munshi S, Daugherty B, O’Neill PalladinoL, Stafford-Hollis J, Hollis GF, Silberklang M. 1994. Optimization ofa fed-batch process for production of a recombinant antibody. Ann NYAcad Sci 745:285–296.

Ryu JS, Lee GM. 1999. Application of hypoosmolar medium to fed-batchculture of hybridoma cells for improvement of culture longevity. Bio-technol Bioeng 62:120–133.

Sato DJ, Cao H-T, Kayada Y, Cabot MC, Sato HH, Okamoto T, Welsh CJ.1988. Effects of proximate cholesterol precursors and steroid hor-mones on mouse myeloma growth in serum-free medium. In Vitro CellDev Biol 24:1223–1228.

Seeger M, Jerez CA. 1992. Phosphate limitation affects global gene-expression inThiobacillus ferrooxidans.Geomicrobiol J 10:227–237.

Simpson NH, Singh RP, Perani A, Goldenzon C, Al-Rubeai M. 1998. Inhybridoma cultures, deprivation of any single amino acid leads toapoptotic death, which is suppressed by the expression of thebcl-2gene. Biotechnol Bioeng 59:90–98.

Singh RP, Simpson NH, Perani A, Goldenzon C, Al-Rubeai M. 1997.Essential amino acid deprivation of hybridoma cells results in theinduction of high levels of apoptosis, which is suppressed bybcl-2expression. Genet Eng Biotechnologist 17:121–124.

Spector MP. 1990. Gene-expression in response to multiple nutrient-starvation conditions inSalmonella typhimurium.FEMS MicrobiolEcol 74:175–183.

Thayer AM. 1998. Great expectations: biopharmaceutical industry expectssuccess from new drug therapies and a full product development pipe-line. Chem Eng News August 10, 1998, p 19–31.

Xie L, Wang DIC. 1996. High cell density and high monoclonal antibodyproduction through medium design and rational control in a bioreactor.Biotechnol Bioeng 51:725–729.

Zhou W, Chen C-C, Buckland B, Aunins J. 1997a. Fed-batch culture ofrecombinant NS0 myeloma cells with high monoclonal antibody pro-duction. Biotechnol Bioeng 55:783–792.

Zhou W, Hu W-S. 1994. On-line characterization of a hybridoma cellculture process. Biotechnol Bioeng 44:170–177.

Zhou W, Rehm J, Europa A, Hu W-S. 1997b. Alteration of mammalian cellmetabolism by dynamic nutrient feeding. Cytotechnology 24:99–108.

Zhou W, Rehm J, Hu W-S. 1995. High viable cell concentration fed-batchcultures of hybridoma cells through on-line nutrient feeding. Biotech-nol Bioeng 46:579–587.


phosphate feeding improves high-cell-concentration ns0 myeloma culture performance for monoclonal...

Documents