ARTICLE

Comparative Metabolite Analysis to UnderstandLactate Metabolism Shift in Chinese Hamster OvaryCell Culture Process

Jun Luo,1 Natarajan Vijayasankaran,2 Jennifer Autsen,1 Rodell Santuray,1 Terry Hudson,1

Ashraf Amanullah,1 Feng Li1

1Oceanside Pharma Technical Development, Genentech, Inc., 1 Antibody Way, Oceanside,

California 92056; telephone: 760 231 2127; fax: 760 231 2465; e-mail: li.feng@gene.com2Late Stage Cell Culture, U.S. Biologics Pharma Technical Development, Genentech, Inc.,

South San Francisco, California

Received 23 March 2011; accepted 26 July 2011

Published online 30 September 2011 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/bit.23291

ABSTRACT: A metabolic shift from lactate production (LP)to net lactate consumption (LC) phenotype was observed incertain Chinese hamster ovary (CHO) cell lines during theimplementation of a new chemically defined medium(CDM) formulation for antibody production. In addition,this metabolic shift typically leads to process performanceimprovements in cell growth, productivity, process robust-ness, and scalability. In our previous studies, a correlationbetween a key media component, copper, and this lactatemetabolism shift was observed. To further investigate thisphenomenon, two complementary studies were conducted.In the first study, a single cell line was cultivated in twomedia that only differed in their copper concentrations, yetwere known to generate an LP or LC phenotype with that cellline. In the second study, two different cell lines, which wereknown to possess inherently different lactate metaboliccharacteristics, were cultivated in the same medium witha high level of copper; one cell line produced lactatethroughout the duration of the culture, and the otherconsumed lactate after an initial period of LP. Cell pelletand supernatant samples from both studies were collected atregular time intervals, and their metabolite profiles wereinvestigated. The primary finding from the metabolic anal-ysis was that the cells in LP conditions exhibited a lessefficient energy metabolism, with glucose primarily beingconverted into pyruvate, sorbitol, lactate, and other glyco-lytic intermediates. This decrease in energy efficiency may bedue to an inability of pyruvate and acetyl-CoA to progressinto the TCA cycle. The lack of progression into the TCAcycle or overflow metabolism in the LP phenotype resultedin the inadequate supply of ATP for the cells. As a conse-quence, the glycolysis pathway remained the major source ofATP, which in turn, resulted in continuous LP throughoutthe culture. In addition, the accumulation of free fatty acidswas observed; this was thought to be a result of phospholipidcatabolism that was being used to supplement the energyproduced through glycolysis in order tomeet the needs of LPcells. A thorough review of the metabolic profiles indicated

that the lactate metabolic shift could be related to theoxidative metabolic capacity of cells.

Biotechnol. Bioeng. 2012;109: 146–156.

� 2011 Wiley Periodicals, Inc.

KEYWORDS: CHO cell culture; fed-batch; lactate; metabo-lomics; copper; oxidative pathway

Introduction

Fed-batch cell culture processing of Chinese hamster ovary(CHO) cells is a widely used platform technology for thesynthesis of therapeutic glycosylated proteins. In a CHO cellculture process the effect of high concentrations of lactatecan be detrimental to cells due to the increased acidificationof culture medium, which ultimately results in highosmolality caused by the alkali addition necessary to controlthe culture pH (Li et al., 2010). Therefore, it is desirable tounderstand and control lactate metabolism in fed-batchprocesses. Different hypotheses have been proposedregarding regulation of lactate metabolism, such as nutrientlimitation or balance, transport limitation, metabolicblockage, and NAD/NADH balance (Glacken et al., 1986;Greiner et al., 1994). Corresponding approaches in cell linegeneration, media optimization, and process control havebeen developed, including maintaining low glucose concen-tration (Xie and Wang, 1994), coupling glucose feed tolactate level and pH (Gagnon et al., 2011), using alternatecarbon sources (Altamirano et al., 2004), optimizingoperating conditions (Moore et al., 1997), manipulatinglactate dehydrogenase (Chen et al., 2000), pyruvatecarboxylase (Kim and Lee, 2007), and pyruvate dehydroge-nase kinases (Zhou et al., 2011), and engineering anti-apoptotic genes into cell lines (Dorai et al., 2009). Recently,it has been reported that the culture medium compositionCorrespondence to: F. Li

146 Biotechnology and Bioengineering, Vol. 109, No. 1, January, 2012 � 2011 Wiley Periodicals, Inc.

(Ma et al., 2009) and copper concentration (Vijayasankaranet al., submitted; Qian et al., 2011) can affect lactateproduction (LP), and hence its shift from net production toconsumption in CHO cells.

Copper plays an important role in the function of anumber of enzymes, such as cytochrome C oxidase complexIV (Zeng et al., 2007), a key protein in the electron transportchain; and copper/zinc superoxide dismutase (Tainer et al.,1983), a protein involved in defense against oxidative stress.The effect of copper on the respiratory chain is welldocumented; combining that knowledge with the discoveryof an impact of copper on lactate metabolism, a linkagebetween oxidative capacity and lactate metabolism washypothesized. During industrial cell culture process devel-opment, selection of lower lactate producing cell lines anddevelopment of well-balanced culture medium are typicallythe preferable route for limiting lactate due to the ease ofimplementation at different development and productionstages. Therefore, finding a common mechanism that canhelp explain variation between cell lines and the impact ofmedia on LP will pave a practical path for controlling lactatein industrial cell culture process development.

In our experience, two common lactate phenotypes areobserved in fed-batch cultures. The desirable phenotype ischaracterized by limited lactate synthesis during the initialgrowth phase, ideally no more than 3g/L, followed by lactateconsumption (LC), usually after cessation of cell growth.The undesirable phenotype is characterized by a high level ofLP and accumulation throughout the entire run. Severalgenomics and proteomics studies (Korke et al., 2004;Pascoe et al., 2007, Seow et al., 2001) have been conductedto shed light on lactate related metabolite shifts. Themetabolite analysis approach complements and enhancesthe proteomics and genomics approaches because metabo-lites are the end products of cellular regulatory processes,and metabolite levels can be regarded as the ultimateresponse of biological systems to genetic or environmentalchanges (Fiehn, 2002).

In this study, two sets of complementary experiments(Case 1 and Case 2) were designed to investigate thepresence or absence of a shift into an LC phase. In Case 1, cellline A was grown in two different chemically-defined mediawith the only formulation difference being the copperconcentration. The cultivated cells in the media with a highcopper concentration demonstrated lactate synthesize andthen consumption (LC), and the media with a low copperconcentration demonstrated lactate synthesize but noconsumption (LP). In the Case 2, two different CHO lineswere used, with cell line B being LP and cell line C being LC.We undertook a comparative approach and hypothesizedthat there is a shared mechanism or pathway for the LP orLC traits regardless of the cell line or culture conditions.Furthermore, the effects of copper and cell line on lactatemetabolism shift were discussed. The correlation betweenthe lactate metabolic shift and the cells’ oxidativemetabolic capacity was proposed based on the metaboliteanalysis data.

Materials and Methods

Cell Culture

Cell lines A, B, and C are in-house developed CHO antibodyproducing cell lines derived from a dihydrofolate reductase-deficient (DHFR-) CHO DUK-XB11 host (Urlaub andChasin, 1980) by stable transfection of a plasmid encodinggenes for DHFR and humanized antibody heavy chain andlight chain. The cell lines were generated in separatetransfections with different antibody sequences (Shen et al.,2006). The frozen cells were thawed in seed train media. Twonon-selective inoculum passages in the absence of metho-trexate (MTX) were used to scale up the culture usingmedium with same copper concentration as the productionbasal media. Both the seed train medium and the productionbasal medium were chemically-defined proprietaryformulations. The copper concentration in productionmedium was 1� (low copper) and 20� (high copper)respectively. Copper concentrations were selected based onprevious component titration study (data not shown).Production culture was inoculated at 1.5� 106 cells/mL witha working volume of 1.5 L in 2-L stirred bioreactors(Applikon, Foster City, CA). The culture was controlledusing a DeltaV-based BioNet system (Broadley-James,Irvine, CA) to maintain the desired process parameters(pH at 7.0, temperature constant at 378C for day 0–3 andthen 358C for day 3 to the end of run, agitation rate at275 rpm, dissolved oxygen level at 30%). An in-houseproprietary chemically-defined feed medium (with 50 g/L ofglucose) was added to the bioreactor on day 3, 6, and 9 at10% of the initial working volume. When glucoseconcentration was below 3 g/L, bolus glucose was addedto the bioreactor to increase the glucose concentration by6 g/L. This level of glucose is not expected to be low enoughto affect glucose uptake rate (Miller et al., 1988). Viable cellcount (VCC) and viability were measured daily using aVi-Cell (Beckman Coulter, Fullerton, CA). Lactate andglucose were determined daily using a Nova Bioprofile 400(Nova Biomedical, Waltham, MA).

Sample Collection for Metabolomic Assay

Both cell pellet and supernatant samples were collected dailyfor metabolomic analysis. Three replicates from threeseparate cultures were submitted for select time pointsbased on lactate profiles, while one replicate was submittedfor other time points to limit the number of totalsamples. To collect cell samples, cell culture brothcontaining 2� 107 viable cells was removed from thebioreactor and immediately chilled in an ice bucket. Thecells were centrifuged at 1,000g for 3min in a refrigeratedcentrifuge (Bechman Coulter, Fullerton, CA) and thenrinsed twice using cold phosphate buffered saline (PBS).After two washes, the cell pellet was flash frozen in liquidnitrogen. A 2mL spent medium sample was collected forwhich cell broth was first centrifuged at 1,000g for 3min and

Luo et al.: Lactate Metabolism in CHO Cell Culture Process 147

Biotechnology and Bioengineering

then the supernatant was filtered through a 0.2mmpolyethersulfone filter. The cell pellets and the supernatantwere frozen at �808C and shipped in dry ice to MetabolonInc. (Durham, NC) for analysis.

Sample Analysis by Metabolon

Sample analysis was conducted at Metabolon Inc.Metabolon developed a platform that integrates thechemical analysis, metabolites identification and relativequantification, data reduction, and quality assurancecomponents of the process. The methodology is detailedelsewhere (Lawton et al., 2008). Briefly, samples wereinventoried and stored at�808C before assay. At the time ofanalysis, one aliquot of each sample was thawed and extractsprepared according to Metabolon’s standard protocol.Metabolon’s proprietary solvent was used to removeprotein, dislodge any small molecules bound to proteinor physically trapped in the precipitated protein matrix, andrecover a wide range of chemically diverse metabolites. Theextracted samples were then split into equal parts foranalysis on their GC/MS and LC/MS/MS platforms. Alsoincluded were several technical replicate samples createdfrom a homogeneous pool containing a small amount of allstudy samples (Client Matrix). The order in which thesamples from the different time points and cell lines wereprocessed was completely randomized. Compounds wereidentified by automatically matching the combination ofchromatographic retention and mass spectra signaturescompared to Metabolon’s reference library entries usingtheir proprietary software developed for creating libraryentries from known chemical entities and then automati-cally fitting those spectra to experimentally derived spectra.Raw data for each compound in each sample werenormalized to correct for variation resulting from instru-ment inter-day tuning differences. The relative normalizedintensity (RNI) for each metabolite was obtained fromMetabolon for further analysis. Welch’s Two Sample t-testswere used to analyze the data and to identify biochemicalsthat differed in the cells and media between the two testconditions.

Results and Discussion

Lactate Shifts in Case 1 (Effect of Copper) and Case 2(Effect of Cell Line)

Triplicate bioreactor cultures were performed for eachcondition to capture the process variation, and all readingswere plotted in average with standard deviation (Fig. 1). Thegrowth (VCC and viability), metabolites (lactate andglucose), and antibody titer profiles from Case 1 andCase 2 are shown in Figure 1. In Case 1 (Fig. 1A-1E), similargrowth profiles were observed until day 5, after which higherVCC was observed in the cultures grown in high coppermedium until the end of the run, and similar viability

profiles were observed throughout the run. In addition, thecells in low copper medium maintained LP until the end ofthe run, while the cells in high copper medium shifted to LCstarting on day 4. A higher glucose level and higher celldensity in high copper media suggested that cells in thatmedia have a lower specific glucose consumption rate.Higher productivity was obtained after day 6 in high coppermedium due to the improved growth and specificproductivity. In case 2 (Fig. 1F-1J), a significant growthdifference (more than 2-fold higher VCC in cell line C) andviability separation (starting from day 5 with lower viabilityobserved in cell line B) were observed. Also cell line Bmaintained LP throughout the culture, even though the netrate slowed after day 5 when the VCC peaked. Cell line Ctransitioned from LP to consumption on day 7; meanwhile,the cell density continued to increase until day 9, after whichthe VCC remained essentially constant. Significantly highertiter was observed in cell line C starting from day 5.

The initial lactate profiles observed in all conditions areconsistent with the fundamental idea of the cells’ reliance onglycolysis early in culture. In all four test conditions, cellsengaged heavily in cytosolic energy production via theconsumption of glucose during glycolysis, especially duringthe rapid exponential growth phase (first 3–4 days), basedon the correlation between cell growth and the rapid lactateincrease. Once the majority of cellular division and growthceased, the specific glucose consumption rate was alsoreduced. This metabolic ‘‘switch’’ is presumed to be drivenby a decreased dependence on glycolysis and an enhancedengagement in oxidative pathways for energy production(Rossignol et al., 2004).

Comparison of Metabolic Profile of LC versus LP

Triplicate samples from days 0, 3, 5, 8, and 10 of Case 1 andtriplicate samples from days 1, 3, 6, 9, and 12 of Case 2 weretested for statistical data analysis. Table I shows the totalnumber of named metabolites detected and the number ofsignificantly differing metabolites (P< 0.05) from bothcases when comparing the LC and LP conditions at thetriplicate time points by t-test. More than 200 metaboliteswere identified in cell pellet (intracellular) samples andaround 150 metabolites were identified in media (extracel-lular) samples. Because samples from Case 1 and Case 2 wereanalyzed at different times, not all the same metabolites weredetected in both cases. In Case 1, only up to 10% of the totalmetabolites were significantly different by the time of thelactate shift in high copper (day 5), and that differenceincreased to up to 25% by the end of the run. In this case theone component modification in media limited the varia-tions; whereas, in Case 2 with different cell lines thereappeared to be a greater extent of metabolic variance. By thetime of the lactate shift in cell line C (day 7), 50% of themetabolites showed significant differences from cell line B,and that increased to 68% by the end of the run. Comparedto Case 1, more of differences in Case 2 could be fromdifferent cell growth, metabolism, and environment

148 Biotechnology and Bioengineering, Vol. 109, No. 1, January, 2012

Figure 1. The growth and lactate profile from Case 1 (effect of copper concentration) using cell line A and Case 2 (effect of cell line) using cell line B and C. A: Case 1 viable

cell density (D); (B), Case 1 viability (�); (C), Case 1 lactate (&); (D), Case 1 glucose (^); (E), Case 1 antibody titer (*). (F), Case 2 viable cell density (D); (G), Case 2 viability (�); (H),Case 2 lactate (&); (I), Case 2 glucose (^); (J), Case 2 antibody titer (*). Dashed line for lactate continuous production with low copper (Case 1) and Line B (Case 2). Solid line for

lactate net consumption with high copper (Case1) and Line C (Case2).

Table I. Number of metabolites detected and significantly differing metabolites between LC and LP conditions in Case 1 and Case 2.

Named metabolites

Case 1: Effect of copper Case 2: Effect of cell line

Detected

No. of significantly differing metabolites

Detected

No. of significantly differing metabolites

Day 0 Day 3 Day 5 Day 8 Day 10 Day 1 Day 3 Day 6 Day 9 Day 12

Intracellular 245 6 24 19 32 60 219 50 72 112 139 149

Extracellular 147 3 6 1 13 35 154 26 28 43 89 99

Luo et al.: Lactate Metabolism in CHO Cell Culture Process 149

Biotechnology and Bioengineering

(osmolarity and pCO2) rather than being directly related tothe lactate shift. In addition, the number of significantlydiffering metabolites increases with time for both cases,which suggests that the cell phenotypes might already bedifferent before the lactate profiles diverged. Thus, thechanges before or concurring with the lactate shift couldcontribute to or magnify the impending metabolic diver-gence, while the changes after the shift could be theconsequence or cellular response to this divergence. Becauseof the comparable cell growth, metabolism, environment,and protein productivity in Case 1 prior to the lactate shift,data analysis was primarily focused on the significantchanges in metabolites from Case 1, with the results fromCase 2 used to supplement the findings. In case of noidentical metabolite being detected in Case 2, the metabolitein the same category was used.

Amino Acid Metabolism

In the Case 1 copper study, the cells in low copper mediumwith LP have higher levels of many, but not all intracellularamino acids when compared to cells in high coppermedium. Some exceptions include arginine, aspartate,cysteine, and histidine. At the same time, similar levels ofextracellular amino acids were typically observed whencompared to cells in high copper medium (exceptionsinclude cystine, aspartate, and asparagine). The five aminoacids with the most significant intracellular changes beforeday 5 (lactate shift) are shown in Figure 2A-2E (intracellu-lar) and Figure 2G-2K (extracellular). Aside from support-ing cell growth and protein production, the next majorfunction of amino acids is to supply the TCA cycle withmetabolic intermediates through catabolic reactions. Withsimilar growth, viability, and titer until day 5, this elevationin intracellular amino acids in the LP batches may be theresult of differences in catabolic efficiency between the LPand LC conditions. Thus, the cells in low copper mediummay not be catabolizing amino acids as efficiently.Figure 2F and L show that alanine experienced a similarmetabolic shift to lactate, except the shift started from day 6,which is consistent with the report by Ma et al. (2009).Similar and magnified trends were observed in Case 2 ofusing different cell lines. Cell line B with LP had higher levelsfor all amino acids in the cell pellets, and differences were upto 21-fold on day 12 (Data not shown). In supernatant, mostof the amino acids were several-fold higher in cell line B aswell, except for threonine, asparagine, and cysteine. Thus,data from both Cases 1 and 2 suggest that cells withcontinuous LP have a lower capacity to consume aminoacids.

Carbohydrate Metabolism

Sorbitol Pathway

In Case 1, elevated intracellular sorbitol and fructose (Fig. 3)were observed in cells with low copper media (LP). The

sorbitol pathway is an alternate pathway for glucose andultimately leads to fructose production (Brownlee, 2001).Fructose can re-enter glycolysis after conversion to fructose-1-P or fructose-6-P (Jeffery and Jornvall, 1983). Therefore,sorbitol is a probable indicator that the glucose entering thecells is not efficiently channeled into glycolysis in the lowcopper condition, compared to the high copper condition.Tomlinson and Gardiner (2008) have shown that whenglucose levels rise the enzyme that converts glucose tosorbitol, aldose reductase (AR), more efficiently capturesand converts glucose to sorbitol. Specifically, in our study,sharp increases of intracellular pyruvate and fructose on day4, day 7, and day 10 following batch feeds (which containglucose) on day 3, day 6, and day 9 were observed. Towardthe end of the run, intracellular sorbitol levels were low forcells in both media, which indicates the cells in low copperwere still able to consume the accumulated sorbitol. Inaddition, even though extracellular sorbitol increased inboth high and low copper conditions, no significantdifference between their levels was observed, which was inagreement with the similar viability profiles for low versushigh copper conditions (Fig. 1B). In Case 2, higherintracellular levels of sorbitol (up to 1780-fold increase)and fructose (up to 135-fold increase) were observed in cellline B compared to cell line C (Fig. 3). In addition, thesupernatants of cell line B had significantly elevated levels ofsorbitol and fructose (17- and 4.5-fold) (Fig. 3). Althoughsorbitol can be synthesized by cells to counteract increasedosmotic pressure (Neuhofer and Beck, 2005), it does notfreely diffuse through the cell membrane. Therefore, in Case2, the increased extracellular sorbitol in cell line B comesfrom the induced cell death or compromised membraneintegrity releasing the increased intracellular sorbitol, whichcorresponds to the viability separation starting from day 5(Fig. 1G).

Glycolytic Pathway

In Case 1, intracellular pyruvate, although highly variable, iselevated in the cells in low copper media during days 4–8(excluding an outlier at day 7) (Fig. 4). In Case 2, pyruvateshowed a steady intracellular increase after day 4 incontinuous lactate producing cell line B. Generated fromglucose through the glycolytic pathway, pyruvate is a keyintermediate in the network of metabolic pathways. It isconverted to lactate using the enzyme lactate dehydrogenase.The increase of pyruvate might be due to increased glycolysisor reduced TCA cycle catabolic capacity.

TCA Cycle Metabolism

In Case 1, the secretion of molecules like 2-hydroxybutyrate(AHB) and 3-methyl-2-oxovalerate (a branch chain aminoacid alpha-keto acid) in the low copper condition aresuggestive of less TCA cycle catabolic capacity (Fig. 5)because both are ultimately destined for the TCA cycle forcatabolism. In general, most metabolites (specifically, acetyl-

150 Biotechnology and Bioengineering, Vol. 109, No. 1, January, 2012

CoA) in the TCA cycle are at a higher level in the continuouslactate producing cell line B of Case 2 (Fig. 5). Acetyl-CoAcan originate from glycolysis, breakdown of lipids (beta fattyacid oxidation), and breakdown of amino acids. In addition,a blockage in the entry of acetyl-CoA into the TCA cycle

would also cause its accumulation. In addition to higherlevel of acetyl-CoA, the cell line B also had higher levels of itsprecursors: Succinate and oxaloacetate but lower levels of itsproducts: Citrate, cis-aconitate, isocitrate, and malate (datanot shown). These facts may indicate that less acetyl-CoA is

Figure 2. Intracellular amino acid profiles in Case 1 (A–F) media evaluation using medium with low copper (LP,&) and high copper (LC,~). Extracellular amino acid profiles

in Case 1 (G–L) media evaluation using medium with low copper (LP,&) and high copper (LC,~). The X-axis shows the different time points in days. The Y-axis shows the relative

normalized intensity (RNI) for that metabolite median scaled to 1.

Luo et al.: Lactate Metabolism in CHO Cell Culture Process 151

Biotechnology and Bioengineering

entering the TCA cycle in cell line B. The results from Case 1and Case 2 suggest a decreased TCA cycle capacity in lactateproducing conditions, which further indicates impairedmitochondrial function.

Lipid and Cholesterol Metabolism

In Case 1, an increase in both intracellular and extracellularfatty acids (e.g., laurate) is observed throughout the run(Fig. 6) in the low copper condition. In Case 2, an increase inmost of the fatty acids along with glycerol and glycerol-3-P(G3P) was observed in cell line B (Fig. 6) when compared tocell line C.

Free fatty acids are an important source of fuel for manytissues since they can yield relatively large quantities of ATP.The increased levels of fatty acids could come from elevatedtriglyceride or phospholipid breakdown, increased synthe-sis, or altered beta fatty acid oxidation or secretion. It hasbeen reported that under conditions of oxygen deficiency intissues there is a buildup of free fatty acids (Ford, 2002;

Neely et al., 1976). This buildup is due to increasedphospholipid breakdown and decreased beta fatty acidoxidation (Ford, 2002; Neely et al., 1976). Therefore, themetabolomic profile observed in the continuous lactateproducing conditions might be similar to hypoxia (lack ofoxygen), which could result in the breakdown of membranephospholipids. Once the free fatty acids are released fromthe triglycerides and the phospholipids, before entering theTCA cycle, they need to be broken down into acetyl-CoAthrough beta oxidation. As discussed before, acetyl-CoA wasincreased significantly in cell line B (Fig. 5). Thus, elevatedlevels of fatty acids are likely reflective of less mitochondrialoxidative catabolic capacity.

Summary of Metabolites

In summary, many intermediates (amino acid, pyruvate,fatty acids, etc.) in multiple pathways, including the TCAcycle and those leading up to mitochondrial processing, areincreased at various level in the continuous lactate

Figure 3. Intracellular and extracelluar sorbitol and fructose profiles in Case 1 (A, B, E, F) media evaluation using medium with low copper (LP, &) and high copper (LC,~)

versus in Case 2 (C, D, G, H) cell line evaluation using cell line B (LP, &) and cell line C (LC, D). The X-axis shows the different time points in days. The Y-axis shows the relative

normalized intensity (RNI) for that metabolite median scaled to 1.

152 Biotechnology and Bioengineering, Vol. 109, No. 1, January, 2012

producing cultures (Fig. 7). This provides a strongindication of impaired mitochondrial and energy metabo-lism in the LP cultures; thus, the observed variations inlactate metabolism could be related to the oxidativemetabolic capacity of cells.

The metabolomic data show significant differences wellbefore the lactate divergence. Although these differences are

all subtle, their compilation argues that the cells in thelactate net consumption conditions are primed for greateroxidative metabolism and engage in it to a larger extent fromthe beginning of the culture.

Effect of Copper (Case 1)

As an essential cofactor for mitochondrial function, copperdeficiency has been shown to cause respiratory impairment(Zeng et al., 2007). Gybina and Prohaska (2008) found thatthe activity of cerebellar mitochondrial cuproenzymecytochrome c oxidase is markedly lower in Cu deficientrat pups and is accompanied by higher lactate levels,indicating mitochondrial inhibition. Copper deficiency alsoshowed shifts in energy substrate utilization from carbohy-drate to fat in rat models (Hoogeveen et al., 1994), whichresulted in a decreased respiratory quotient. To maintain anadequate energy balance, the demand for fat as a substratefor oxidation may be increased in response to the reducedcarbohydrate oxidation resulting from the impairedcarbohydrate metabolism. In addition, data from a studyby Kelly et al. (1974) indicated that the energy metabolism ofthe myocardium of copper-deficient rats was augmentedthrough the increased activities of lactate dehydrogenase andseveral mitochondrial oxidative enzymes.

Since both cells in the low copper medium and cells underhypoxia displayed increased fatty acid levels, a linkagebetween copper deficiency and hypoxia is proposed. Under alow copper environment, the oxygen limitation does notcome from the environmental oxygen level. Instead, theinability to effectively use oxygen is due to a disabledoxidative phosphorylation enzyme (e.g., cytochrome c

Figure 5. Intracellular and extracelluar TCA cycle pathway metabolite profiles in Case 1 (A, B, and C) media evaluation using medium with low copper (LP, &) and high

copper (LC, ~) versus in Case 2 (D) cell line evaluation using cell line B (LP, &) and cell line C (LC, D).

Figure 4. Intracellular pyruvate profiles in Case 1 (A) media evaluation using

medium with low copper (LP,&) and high copper (LC,~) versus in Case 2 (B) cell line

evaluation using cell line B (LP, &) and cell line C (LC, D).

Luo et al.: Lactate Metabolism in CHO Cell Culture Process 153

Biotechnology and Bioengineering

oxidase) from the copper deficiency, essentially histotoxichypoxia.

With high copper concentration, higher respiratorycapacity would also reduce the need for the cells to generateenergy through aerobic glycolysis and hence would limitlactate synthesis rates. Our results suggest that respiratoryimpairment could be a key mechanism by which copperimpacts cell phenotype with respect to lactate metabolism.Furthermore, similar results of altering lactate metabolismwith different concentrations of copper have been con-firmed with three different cell lines (unpublished data).

Effect of Cell Line (Case 2)

Isolating recombinant cell lines for large scale productionremains costly and time consuming; due to recombinant celllines generated from the same parent cell line can display abroad range of growth, productivity, and metaboliccharacteristics. Screening for desirable characteristics (i.e.,lactate net consumption) early in clone selection isadvantageous. In the proteomics study on the comparisonof lactate continuous producing cell line and the net LC cellline, Pascoe et al. (2007) pointed out that the changingproteins included glycolytic enzymes (i.e., enolase andpyruvate kinase), antioxidant enzyme (i.e., peroxiredoxin),and proteins related to antibody production, proteinprocessing (i.e. BiP/GRP 78, PDI), and cell structure.Korke et al. (2004) compared the cells from two differentsteady states of different glucose to lactate production molarratio (0.08 vs. 1.4) using gene profiling and proteomictechniques. Because only moderate changes in the geneexpression in terms of the number of genes and the extent oftheir differential expression level were observed, it suggeststhat the metabolic shift evolves from both changes at enzymelevels andmodulation of reaction rates through nutrient andmetabolic intermediate concentration. Therefore, identify-ing the potential metabolites associated with the lactate shiftmay indicate useful gene targets for metabolic engineering tofurther improve the cell line development. In the Case 2study, a significant amount of metabolites changed beforethe lactate shift in all pathways for cell line B and C. WithCase 2 study alone, it is impossible to narrow down oridentify the potential pathway or mechanism for the lactateshift. The complementary study is required. Only aftercombining Cases 1 and 2, the importance of energymetabolism or mitochondria oxidative capacity to thelactate shift is clearly seen.

A previous study established that mitochondrial mem-brane potential is correlated to the extracellular glucoseconcentration, and thus to the effective specific glucoseuptake rate (Borth et al., 1993). Hinterkorner et al. (2007)used Rhodamine 123, a lipophilic cationic dye for cellsorting to improve the energy metabolism of existingproduction cell lines. From the cell line with high LP, asubclone sorted for reduced mitochondrial membranepotential was found to have a lower specific lactateformation rate compared to the parental cell line in batchcultures. Therefore, prescreening the parental cells orselecting the clone with favorable mitochondrial membranepotential or simply with more mitochondria provides apotential viable solution to control the lactate metabolism.

Lactate Metabolism and Oxidative Capacity

Valenti et al. (2002) reported evidence of the existence ofmitochondrial L-lactate carriers in 2002. Recently, thediscovery of lactate carrier-mediated transport processesand mitochondrial L-lactate dehydrogenase (mL-LDH) inmammalian cells suggests that mitochondria play an

Figure 7. Balance between glycolysis and mitochondria oxidative capacity.

Figure 6. Intracellular fatty acid metabolite profiles in Case 1 (A) media evalua-

tion using medium with low copper (LP, &) and high copper (LC, ~) versus in Case 2

(B) cell line evaluation using cell line B (LP, &) and cell line C (LC, D).

154 Biotechnology and Bioengineering, Vol. 109, No. 1, January, 2012

important role in lactate metabolism (Passarella et al., 2008).Briefly, lactate produced from glucose through glycoslysis inthe cytosol can enter mitochondria via a proton compen-sated symport, and it is then oxidized inside themitochondria via mL-LDH to pyruvate (Fig. 7). In addition,mitochondria are extremely susceptible to oxidative damagebecause 2–4% of the oxygen consumed by mitochondria isconverted to superoxide anions by the electron transportchain and also because mitochondria have limited protec-tion from oxidative stress with its naked DNA. When theelectron transport chain is disrupted, mitochondria wouldbe the first cellular components affected, and in consequencethe cell can no longer aerobically produce ATP for energy.Rossignol et al. (2004) demonstrated that the type of energysubstrate can change the mitochondrial composition,structure, and function regarding its ability to use glycolysisversus oxidative phosphorylation in HeLa cells. In our study,it appears that the high oxidative capacity of LC cells madelactate a preferred substrate after cell growth cessation, evenin the presence of excess glucose.

CONCLUSIONS

In summary, two sets of complementary experiments wereconducted to elucidate the mechanism of a lactate shiftobserved in bioreactor cultures; the studies evaluatedmetabolic changes resulting from the use of different cellculture media and cell lines. In both cases, all cultivatedmammalian cells engaged heavily in cytosolic energyproduction via the consumption of glucose throughglycolysis, especially during the rapid exponential growthphase. After that, for low copper (Case 1) or cell line B (Case2), continuous LP was observed, whereas for high copper(case 1) or cell line C (Case 2), a lactate shift fromproduction to net consumption was exhibited. Based on theamount and extent of metabolite changes, Case 1 was chosenas the primary focus for this analysis. The changes observedin Case 1, such as increased intracellular amino acids,pyruvate, and fatty acids, were seen in Case 2 to an evengreater extent.

Major differences between lactate metabolism profileswere identified, indicating a more efficient energy metabo-lism in the lactate net consumption conditions throughimproved conversion of metabolites into TCA cycle. Theeffect of copper concentration on lactate metabolismprovides an important practical approach to controllingthe lactate profile in industrial cell culture. The impact ofrobust oxidative pathways in CHO cell culture processes wasdiscussed based on experiments results and literatures. Tofurther verify that LC is unequivocally related to oxidativemetabolic capacity, the reverse experiments, such as treatinglactate net consumption conditions with various agentsknown to either shut down or impair oxidative metabolism(e.g., rotenone, Li et al., 2003), may provide valuable insight.At the same time, genomic/target probing or proteomicapproaches to study the effects of copper concentration onlactate metabolism could be done to bolster the conclusion.

Overall, an improved understanding of the underlying causefor the LP versus consumption behavior could yield gains inboth productivity and robustness of industrial cell cultureprocesses.

The authors are grateful for the valuable contributions and technical

discussions by Robert Hamilton,Masaru Shiratori, Inn Yuk, and Robb

Shawley. They also acknowledge Kurt Boudonck and Kirk Beebe from

Metabolon, Inc. for the technical support and collaboration.

References

Altamirano C, Paredes C, Illanes A, Cairo JJ, Godia F. 2004. Strategies for

fed-batch cultivation of t-PA producing CHO cells: Substitution of

glucose and glutamine and rational design of culture medium.

J Biotechnol 110(2):171–179.

Borth N, Kral G, Katinger H. 1993. Rhodamine 123 fluorescence of

immortal hybridoma cell lines as a function of glucose concentration.

Cytometry 14(1):70–73.

Brownlee M. 2001. Biochemistry and molecular cell biology of diabetic

complications. Nature 414:813–820.

Chen K, Liu Q, Xie L, Sharp PA, Wang DIC. 2000. Engineering of a

mammalian cell line for reduction of lactate formation and high

monoclonal antibody production. Biotechnol Bioeng 72(1):55–61.

Dorai H, Kyung YS, Ellis D, Kinney C, Lin C, Jan D, Moore G, Betenbaugh

MJ. 2009. Expression of anti-apoptosis genes alters lactate metabolism

of Chinese hamster ovary cells in culture. Biotechnol Bioeng 103(3):

592–608.

Fiehn O. 2002. Metabolomics – the link between genotypes and phenotypes.

Plant Mol Biol 48:155–171.

Ford DA. 2002. Alterations in myocardial lipid metabolism during

myocardial ischemia and reperfusion. Prog Lipid Res 41(1):6–26.

Gagnon M, Hiller G, Luan Y-T, Kittredge A, DeFelice J, Drapeau D. 2011.

High-end pH-controlled delivery of glucose effectively suppresses

lactate accumulation in CHO fed-batch cultures. Biotechnol Bioeng

108(6):1328–1337.

Glacken MW, Fleischaker RJ, Sinskey AJ. 1986. Reduction of waste product

excretion via nutrient control: Possible strategies for maximizing

product and cell yields on serum in cultures of mammalian cells.

Biotechnol Bioeng 28(9):1376–1389.

Greiner EF, Guppy M, Brand K. 1994. Glucose is essential for proliferation

and the glycolytic enzyme induction that provokes a transition to

glycolytic energy production. J Biol Chem 269(50):31484–31490.

Gybina AA, Prohaska JR. 2008. Copper deficiency results in amp-activated

protein kinase activation and acetylcoa carboxylase phosphorylation in

rat cerebellum. Brain Res 14(1204):69–76.

Hinterkorner G, Brugger G, Muller D, Hesse F, Kunert R, Katinger H, Borth

N. 2007. Improvement of the energy metabolism of recombinant cho

cells by cell sorting for reduced mitochondrial membrane potential.

J Biotechnol 129(4):651–657.

Hoogeveen RCAM, Reaves SK, Reid PM, Reid BL, Lei KY. 1994. Copper

deficiency shifts energy substrate utilization from carbohydrate to fat

and reduces fat mass in rats. J Nutr 124(9):1660–1666.

Jeffery J, Jornvall H. 1983. Enzyme relationships in a sorbitol pathway that

bypasses glycolysis and pentose phosphates in glucose metabolism.

Proc Natl Acad Sci USA 80(4):901–905.

Kelly WA, Kesterson JW, Carlton WW. 1974. Myocardial lesions in the

offspring of female rats fed a copper dificient diet. Exp Mol Pathol

20(1):40–56.

Kim SH, Lee GM. 2007. Functional expression of human pyruvate carbox-

ylase for reduced lactic acid formation of Chinese hamster ovary cells

(DG44). Appl Microbiol Biotechnol 76:659–665.

Korke R, Gatti Mde L, Lau A, Lim J, Seow T, Chung M, Hu W. 2004. Large

scale gene expression profiling of metabolic shift of mammalian cells in

culture. J Biotechnol 107(1):1–17.

Lawton K, Berger A, Mitchell M, Milgram K, Evans A, Guo L, Hanson R,

Kalhan S, Ryals J, MilburnM. 2008. Analysis of the adult human plasma

metabolome. Pharmacogenomics 9(4):383–397.

Luo et al.: Lactate Metabolism in CHO Cell Culture Process 155

Biotechnology and Bioengineering

Li N, Ragheb K, Lawler G, Sturgis J, Rajwa B, Melendez JA, Robinson JP.

2003. Mitochondrial complex I inhibitor rotenone induces apoptosis

through enhancing mitochondrial reactive oxygen species production.

J Biol Chem 278(10):8516–8525.

Li F, Vijayasankaran N, Shen AY, Kiss R, Amanullah A. 2010. Cell culture

processes for monoclonal antibody production. mAbs 2(5):466–

479.

Ma N, Ellet J, Okediadi C, Hermes P, McCormick E, Casnocha S. 2009.

A single nutrient feed supports both chemically defined NS0 and CHO

fed-batch processes: Improved productivity and lactate metabolism.

Biotechnol Prog 25(5):1353–1363.

Miller WM, Wilke CR, Blanch HW. 1988. Transient responses of hybrid-

oma cells to lactate and ammonia pulse and step changes in continuous

culture. Bioproc Biosyst Eng 3(3):113–122.

Moore A, Mercer J, Dutina G, Donahue CJ, Bauer KD, Mather JP,

Etcheverry T, Ryll T. 1997. Effects of temperature shift on cell cycle,

apoptosis and nucleotide pools in CHO cell batch cultures. Cytotech-

nology 23(1–3):47–54.

Neely J, Rovetto M, Whitmer J. 1976. Rate-limiting steps of carbohydrate

and fatty acid metabolism in ischemic hearts. Acta Med Scand Suppl

587:9–15.

Neuhofer W, Beck F-X. 2005. Response of renal medullary cells to osmotic

stress. Contrib Nephrol 148:21–34.

Pascoe DE, Arnott D, Papoutsakis ET, Miller WM, Andersen DC. 2007.

Proteome analysis of antibody-producing CHO cell lines with different

metabolic profiles. Biotechnol Bioeng 98(2):391–410.

Passarella S, Bari Ld, Valenti D, Pizzuto R, Paventi G, Atlante A. 2008.

Mitochondria and l-lactate metabolism. FEBS Lett 582(25–26):3569–

3576.

Qian Y, Khattak SF, Xing Z, He A, Kayne PS, Qian N-X, Pan S-H, Li ZJ.

2011. Cell culture and gene transcription effects of copper sulfate on

cho cells. Biotechnol Prog 27(4):1190–1194.

Rossignol R, Gilkerson R, Aggeler R, Yamagata K, Remington SJ, Capaldi

RA. 2004. Energy substrate modulates mitochondrial structure and

oxidative capacity in cancer cells. Cancer Res 64:985–993.

Seow TK, Korke R, Liang RCMY, Ong S-E, Ou K, Wong K, HuW-S, Chung

MCM. 2001. Proteomic investigation of metabolic shift in mammalian

cell culture. Biotechnol Prog 17(6):1137–1144.

Shen AY, Van de Goor J, Zheng L, Reyes AE, Krummen LA. 2006.

Recombinant DNA technology and cell line development. In:

Ozturk SS, Hu W-S, editors. Cell culture technology for pharmaceu-

tical and cell-based therapies. Boca Raton, FL: CRC Press Taylor &

Francis Group.

Tainer JA, Getzoff ED, Richardson JS, Richardson DC. 1983. Structure and

mechanism of copper, zinc superoxide dismutase. Nature 206(5940):

284–287.

Tomlinson DR, Gardiner NJ. 2008. Glucose neurotoxicity. Nat Rev

Neurosci 9(1):36–45.

Urlaub G, Chasin LA. 1980. Isolation of Chinese hamster cell mutants

deficient in dihydrofolate reductase activity. Proc Natl Acad Sci USA

77(7):4216–4220.

Valenti D, Bari Ld, Atlante A, Passarella S. 2002. L-lactate transport into rat

heart mitochondria and reconstruction of the l-lactate/pyruvate shut-

tle. Biochem J 364(Pt1):101–104.

Vijayasankaran N, Shawley R, Yang Y, Shiratori M, Jamison K, Varma

S, Caffaleatte C, Brooks D, Meier S, Zhang B, Gawlitzek M, Kiss R.

Effect of copper on metabolism and productivity of industrial chinese

hamster ovary cell cultures. Biotechnol Prog Submitted.

Xie L, Wang DIC. 1994. Fed-batch cultivation of animal cells using different

medium design concepts and feeding strategies. Biotechnol Bioeng

43(11):1175–1189.

Zeng H, Saari JT, Johnson WT. 2007. Copper deficiency decreases complex

IV but not complex I, II, III, or V in themitochondrial respiratory chain

in rat heart. J Nutr 137(1):14–18.

Zhou M, Crawford Y, Ng D, Tung J, Pynn AFJ, Meier A, Yuk IH,

Vijayasankaran N, Leach K, Joly J, Snedecor B, Shen A. 2011. Decreas-

ing lactate level and increasing antibody production in Chinese

Hamster Ovary cells (CHO) by reducing the expression of lactate

dehydrogenase and pyruvate dehydrogenase kinases. J Biotechnol

153:27–34.

156 Biotechnology and Bioengineering, Vol. 109, No. 1, January, 2012

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具