HIGH DENSITY MAMMALIAN CELL GROWTH IN LEIBOVITZ ...medium with fructose or galactose, and a similar...

J. Cell Set. 78, 173-189 (1985) 173Printed in Great Britain © The Company of Biologists Limited 1985

HIGH DENSITY MAMMALIAN CELL GROWTH INLEIBOVITZ BICARBONATE-FREE MEDIUM: EFFECTSOF FRUCTOSE AND GALACTOSE ON CULTUREBIOCHEMISTRY

D. BARNGROVER*, J. THOMASf AND W. G. THILLYDepartment of Applied Biological Sciences, Room El8-666, Massachusetts Institute ofTechnology, Cambridge, Massachusetts 02139, U.SA.

SUMMARY

The most commonly used buffering system for mammalian cell cultures is a bicarbonate/CO2system, which requires CO2 regulators and incubators to supply a constant level of CO2. As areplacement, Leibovitz developed a bicarbonate-free medium, L15, with relatively high levels ofcertain amino acids in the free base form. We found that a modified form of L1S, containing 10 mM-fructose instead of galactose, supported high density growth of Vero and MDCK cells, with main-tenance of a stable pH and lactate/pyruvate ratio. We report here investigations of Vero and MDCKcell growth and culture biochemistry at different concentrations of the two carbohydrates. Theinitial fructose concentration in the medium affected the eventual pH of the medium, the rate ofproduction of lactic acid and ammonia, and the fructose utilization rate. The initial galactoseconcentration affected the growth rate but did not affect eventual culture pH, the rates of lactate andammonia production, or the rate of its own utilization. Thus, Leibovitz' formula, modified tocontain lOmM-fructose, appears to yield satisfactory stability of culture pH and the lactate/pyruvate ratio. At all concentrations of galactose tested, the lactate/pyruvate ratio drifted out of thephysiological range.

INTRODUCTION

The most common method in use today for buffering mammalian cell cultures isthe bicarbonate/CO2 system, which is based on the following equilibrium:

CO2 + H2O < - > HCO3" + H+

The use of bicarbonate is an attempt to mimic the buffering system of blood, butit has at least two major drawbacks. (1) The p/Q of bicarbonate is 6*1, which is farremoved from the desired pH range of cell culture media (7-0-7-4). (2) Althoughbicarbonate is cheap, supplying a constant level of CO2 to cell cultures is definitelynot, as it requires expensive CO2 incubators or fermentors.

•Present address: Integrated Genetics, 31 New York Ave, Framingham, MA 01701, U.S.A.fPresent address: Genentech, Inc., 460 Point San Bruno Blvd, South San Francisco, California

94080, U.S.A.

Key words: bicarbonate-free Leibovitz, media, mammalian, cell culture, Vero cells, MDCK cells,fructose.

174 D. Barngrover, J. Thomas and W. G. Thilly

In high density cell culture, the production of CO2 during metabolism complicatesthe control of CO2 levels in the cell environment.

A simpler solution to this buffering problem was developed by Leibovitz (1963).His L15 medium and a subsequent modification (Waymouth, 1981) were designed foruse in cultures in free exchange with room air and were based on the following

6-4 -

Days

Fig. 1. Cell number (A) and pH of the medium (B) in microcarrier cultures of Vero cells.Cultures were grown in 100 ml total volume with Sg microcarrier/1 and the indicatedmedium with 10% horse serum. Cultures with DMEM or HiGEM were maintained inan atmosphere of 10 % CO2 :90 % air, while cultures with L15 were maintained in 100 %air. ( • ) DMEM-glucose; (•) HiGEM-fructose; (A) LIS-galactose; (T) L15-fructose.

High density mammalian cell growth media 175

principles. (1) The use of high levels of the free base forms of amino acids (arginine,cysteine and histidine) to set the desired pH. (2) The use of galactose instead ofglucose as a carbohydrate source, which results in a lower production of lactic acid andhence less need for buffering (Eagle, Barban, Levy & Schulze, 1958). (3) The use ofpyruvate and alanine to generate the required bicarbonate for synthesis reactions.

Earlier research from this laboratory has shown that the use of fructose instead of

7-6

7-3

o-7-O

6-7

6-4

r — " ' •

-

•

i 1 1—

i i i i

B

-

-

1 1 1 1

Days

Fig. 2. Cell number (A) and pH of the medium (B) in microcarrier cultures of MDCKcells. Conditions (and symbols) were the same as in Fig. 1.


glucose in cell culture media buffered with bicarbonate/CCh results in a lower lacticacid production and a more stable pH and lactate/pyruvate ratio (Imamura, Crespi,Thilly & Brunengraber, 1982). Therefore, it seemed reasonable to examine Leibovitzmedium with the original galactose-containing formula or with fructose as carbo-hydrate source.

x

g

5 -

2 -

7-6

7-3

Q.7-0

6-7

1 1 1

f̂c A—A-—

-

i i

.A.

A i

- - .

-

i i i

10 12 14Days

Fig. 3. Cell number (A) and pH of the medium (B) in microcarrier cultures of Vero cells.Initial conditions (and symbols) were as in Fig. 1. After day 6, the medium was replacedwith medium containing 2 % horse serum, and with fresh medium every 3 days thereafter(indicated by arrowheads).


Another consideration with regard to analysing the effects of type and concentra-tion of carbohydrate arises from the nature of energy sources in mammalian cellculture. Reitzer, Wice & Kennell (1979) have reported that for cultured HeLa cells,glutamine is the major energy source, not sugar. From the distribution of label from[U-14C]glutamine, Reitzer et al. concluded that a majority of the glutamine wascompletely oxidized, though a significant fraction of the label (13 %) did accumulatein lactate. They also concluded that when 2mM-fructose was used, 90% of it wasmetabolized via the oxidative pathway of the pentose phosphate cycle, with very littlegoing through glycolysis. The utilization of glucose and glutamine is reciprocallyregulated in human diploid fibroblasts (Zielke et al. 1978); that is, high levels of eachreduce the utilization of the other. We were interested in determining if fructose andgalactose had the same effect in Vero cells. One of the major byproducts of glutaminemetabolism is ammonia, and earlier studies have shown that the accumulation ofammonia matches the decrease in glutamine (Imamura et al. 1982). We thereforemeasured ammonia production, as well as lactate, pyruvate and fructose levels in L15microcarrier cultures with different kinds and concentrations of carbohydrates.

MATERIALS AND METHODS

Cells

Madin Darby canine kidney cells (MDCK) and Vero African green monkey cells were obtainedfrom Flow Laboratories, Inc. (McLean, VA). Stocks were maintained in 490cm2 roller bottles in

Lactate

Carbohydrate —(Fructose,Glucose,Galactose)

NADH,H+-

-•Pyruvate

Aspartate

Ammonia

Glutamine

Fig. 4. Outline of amino acid and carbohydrate utilization for energy production.


50

20

10

5

2

1 — ' -i

Vero

-

L l _

1 1 1 1 1

A

«———•"• ""~y~ •

-

-

1 1 1 1 _ 1

Days

Fig. 5. Cell number (A) and pH of the medium (B) in microcarrier cultures of Vero cellsin L15 plus 10% horse serum with the indicated initial concentrations of fructose. Cul-tures (100 ml total volume) were maintained in free exchange with room air (# ) 5 mM, ( • )10mM, (A) 15mM, ( • ) 20mM-fructose.

100 ml HiGEM (Flow Laboratories) with 10% horse serum (GIBCO, Grand Island, NY). MDCKcells were used between passages 64 and 66 and Vero between 139 and 145. All cells were routinelytested for mycoplasma by DNA fluorescence testing (Russell, Newman & Williamson, 1975) (CellCulture Centre, Massachusetts Institute of Technology) and found to be negative.


Fig. 6. Lactate levels in the medium of the Vero microcarrier cultures shown in Fig. 5.Symbols as in Fig. 5 for Figs 6-10.

MediaDulbecco's modified Eagle's medium (DMEM), HiGEM (a modification of DMEM containing

fructose instead of glucose), L15 and L15 without galactose were obtained as liquid media fromFlow Laboratories. Media were supplemented with 4 mM-glutamine (Flow), 10 % horse serum andother supplements as indicated in the separate experiments.

Microcarrier culturesMicrocarrier cultures were grown in 100 ml cultures in 250 ml siliconized spinner bottles (Wilbur

Scientific, Boston, MA) at 37 °C. Cultures with DMEM or HiGEM were kept in incubators contain-ing 10% CO2 :90% air, while those containing L15 were maintained in 100% air.

Cultures were initiated by combining washed microcarriers (Flow Laboratories) and cells tryp-sinized from confluent roller bottles as previously described (Levine, Wong, Wang & Thilly, 1977).All cultures contained 5 g (dry weight) microcarrier per litre of medium; the microcarriers were keptin suspension by stirring with a Teflon-coated suspended stir bar at 60rev./min.

Growth and metabolite determinationsCell growth was determined in daily 1 ml samples, using the method of enumeration of nuclei

previously described (Levine et al. 1977). The pH was measured in separate 2 ml samples obtainedat the same time. These samples were then deproteinized with 2 ml of 6 % perchloric acid and storedfrozen ( -20°C) . After thawing and neutralization with 2M-Na2HCO3, samples were taken forenzymic determination of fructose (Bernt & Bergmeyer, 1974), ammonia (Kun & Kearney, 1974),and lactate and pyruvate (Sigma Technical Bulletin, 1980).

RESULTS AND DISCUSSION

Cell growth and pH changes in Vero and MDCK culturesFigs 1B and 2B show the pH of the Vero and MDCK cell cultures as they increase


in cell number (shown in Fig. 1A and 2A) from an initial seeding density of 3 X 105

to 2 X 106 to 3 X 106 cells/ml in stirred batch culture. The mean values for simul-taneous duplicate cultures are shown. No significant difference in net cell growth orrate of growth occurred when the bicarbonate/CO2 buffering system (DMEM withglucose or HiGEM with fructose) was replaced with Leibovitz bicarbonate-free

2 25

Fig. 7. The ratio of lactate levels to pyruvate levels in the medium of the Vero microcarriercultures shown in Fig. 5.


formula. This shows that neither cell line had a bicarbonate requirement for growthnot fulfilled by its own metabolism.

As expected, pH dropped rapidly when cell number increased in glucose-containing medium (Imamura et al. 1982), as seen for Vero cells in Fig. 1A. The pHremained essentially constant (7-2-7-4) in the media containing fructose or galactose,with or without bicarbonate/CCh-based buffering during the exponential growthphase. A general increase in pH followed growth cessation for Vero cells in Leibovitzmedium with fructose or galactose, and a similar increase was seen with MDCK cellsgrown in Leibovitz medium with fructose. This may be a matter of some concern,since Vero cells are unusually sensitive to pH above 7-4. In our experience, the cellsdetach from microcarriers above this pH.

Since maintenance of Vero cells as non-growing confluent monolayers is necessaryfor production of a number of viruses (e.g. Rubella), we also examined the effect ofbuffering system and carbohydrate source in maintenance culture (Fig. 3). We foundthat 50 % medium replacement every 3 days, beginning on day 6 (post-confluence),led to desirable maintenance behaviour for bicarbonate/CO2-buffered HiGEM andfor bicarbonate-free Leibovitz medium with galactose or fructose. Again, a generalincrease in pH was seen with the bicarbonate-free Leibovitz formulation, but this wasnot associated with any loss in microcarrier-attached cells.

U.

10 -

5 •

0 1

Fig. 8. Fructose levels in the medium of the Vero microcarrier cultures shown in Fig. 5.


Fig. 9. The rate of fructose utilization in the Vero microcarrier cultures shown in Fig. 5.

Since there was no significant difference in growth when the bicarbonate/CO2system was replaced with the bicarbonate-free formula, further studies concentratedon the latter.

Culture biochemistry

Fig. 4 summarizes our view of the metabolic possibilities for the experimentsdescribed. Basically, our view is derived from Reitzeref al. (1979), who demonstratedthe role of glutamine as an energy source, and from Zielke et al. (1978), who notedan interrelationship between carbohydrate and glutamine utilization. Starting fromthese considerations, we monitored the following quantities: ammonia concentrationas an indicator of amino acid (primarily glutamine) catabolism, pyruvate and lactateconcentrations as indicators of carbohydrate catabolism, the pyruvate/lactate ratio asan estimate of cellular redox potential, and free sugar concentrations as a directestimate of their utilization.


6 -

Fig. 10. Ammonia levels in the medium of the Vero microcarrier cultures shown in Fig. 5.

Growth of Vero cells in bicarbonate-free medium with fructose

In our first set of experiments, Vero cells were grown as stirred batch cultures inbicarbonate-free Leibovitz medium with 5, 10, 15 or 20mM-fructose. Cell numberand culture pH were measured daily (Fig. 5A,B). Cell growth to confluence wassimilar at all fructose concentrations, but microscopic examination of the culturesrevealed some abnormal morphology of cells on microcarriers post-confluence at20mM-fructose. Cultures started with 15 and 20 mia-fructose demonstrated drops inpH below 7-0 by confluence, while a range of 7-1-7-3 was maintained through 6 daysin cultures started with 10 mM. At 5 mM-fructose, a general increase in pH was notedpost-confluence. The pH changes were coincident with changes in lactic acid con-centration in the cultures (Fig. 6). Analysis of the lactate/pyruvate ratios (Fig. 7)showed that the 10 mM-f ructose medium resulted in cultures that remained in therange of physiological redox potentials for the duration of our observations. However,15 and 20 miw-f ructose concentrations resulted in unphysiologically high values.Strangely, the 5 mM-fructose cultures showed an initial increase to high lactate/pyruvate ratios and then declined to levels lower than the physiological range.

Fructose concentrations declined most rapidly at high initial fructose concentra-tions (Fig. 8). The cellular rate of fructose utilization was initially high for the 20 mM-fructose-containing medium but declined to very low values after cell confluence and


50

20

n1O

x 10

Cel

ls/m

l

2

1 •

VeroL15-galactose

-

-

• i

A

-

1 1 1 1 1

Days

Fig. 11. Cell number (A) and pH of the medium (B) in microcarrier cultures of Vero cellsin L15 with the indicated initial levels of galactose. Cultures (100 ml total volume) weremaintained in free exchange with room air. ( T ) 20mM, (A) 15 mM, ( • ) lOmia,5 mM-galactose.

after the fructose concentration in the culture was reduced to about 17 mM. The ratesof fructose utilization were about the same when cultures were started at 10 or 15 mM-fructose. These rates also became slower after confluence to a low rate by day 6.Utilization began slowly in 5 mM-fructose-containing cultures, rose to a maximum ata time somewhat later than cessation of cell growth, and then declined (Fig. 9).


3

1 -

Fig. 12. Lactate levels in the medium of the Vero microcarrier cultures shown in Fig. 11.Symbols as in Fig. 11 for Figs 12—14.

Fig. 10 shows that ammonia production was rapid during cell proliferation, prob-ably indicating rapid glutamine utilization. The highest ammonia levels were attainedat the lowest fructose concentrations, which may reflect carbohydrate sparing ofglutamine, as suggested by Zielke et al. (1978).

Overall, a picture emerges of the desirability of Leibovitz bicarbonate-free formulawith fructose at 10 mM as a medium for growing Vero cells in batch cultures in whichstable, physiological pH and redox potential are important.

Growth of Vero cells in LI5 with galactose

We next examined the same parameters for Vero growth in batch microcarrierculture using the original Leibovitz formula with galactose as the carbohydrate sourcein varying initial concentrations. Fig. 1 1A and B reveal that, while all tested galactoseconcentrations resulted in stable pH (7-0-7-3), higher galactose concentrations(IS mM and 20 mM) resulted in significant diminution of cell growth rate and final celldensity.

Lactate concentration showed an initial increase in galactose-containing cultures(Fig. 12) similar to that seen with fructose (Fig. 6). However, whereas lactate con-centration remained constant or increased in fructose-containing media, it showed a

186 D. Barngrvver, J. Thomas and W. G. Thilly

marked decrease from a maximum at day 2 at all concentrations of galactose tested.Lactate/pyruvate ratios (Fig. 13) fluctuated widely for all galactose concentrations.The ratio rose well above the physiological range at 5, 10 and 15 mM initial galactoseconcentrations, but at 20 mM-galactose the ratio remained within the physiologicalrange.

Ammonia concentration (Fig. 14) rose continuously in media containing 5 and10 mM-galactose but appeared to reach a maximum post-confluence with 15 and20 mM-galactose.

Our initial rationale for using fructose or galactose in cell culture medium was tominimize decreases in pH by minimizing lactate production. Since Vero and MDCKcells are sensitive to alkaline pH values (above 7-4), pH control in this direction is alsoimportant. We suggested that the major determinants of pH change were the lactateand ammonia produced by the cells. To test this hypothesis, we converted the changesin pH measured to acid equivalents produced (using data from a pH titration of L15with 10% horse serum) and did a multiple regression analysis of the cumulativechanges in acid equivalents (dependent variable) on the cumulative changes in lactateand ammonia (independent variables). For the fructose-containing cultures, theregression equation was:

Fig. 13. The ratio of lactate levels to pyruvate levels in the Vero microcarrier culturesshown in Fig. 11.


Fig. 14. Ammonia levels in the medium of the Vero microcarrier cultures shown inFig.11.

y = 1-0 + 0-53 (lactate production) - 0-11 (ammonia production).

Analysis of variance showed that the overall regression equation was significant(P < 0 • 01) and accounted for 80 % of the variability in production of acid equivalents.Individually, the coefficient for lactate production was significantly different from 0(P < 0-01), while the coefficient for ammonia production was not. While lactate levelsare a major determinant of pH changes, only 1 mole of acid equivalents was producedfor every 2 moles of lactate. This indicates that the production or consumption ofsome other components must be involved in determining the pH of the medium, bydamping the effects of lactic acid.

For the galactose-containing cultures, the regression equation was:

y = - 0-64 + 0-37 (lactate) - 0-01 (ammonia).

Though this regression was significant (P< 0-01), the regression accounted for only30 % of the variability in the acid equivalent data. Individually, when the effect of theother was subtracted, neither coefficient was significantly different from zero. Lacticacid production did not affect pH in the galactose-containing cultures as it did in thefructose-containing cultures, indicating that other unknown components determine


the pH of the medium. Possibilities include CO2 production (as the bicarbonate ion),acidic metabolities, or other amino acids.

In summary, high density microcarrier cultures of Vero and MDCK cells can begrown and maintained in bicarbonate-free media. If CO2 is a required nutrient, assuggested by the work of Itagaki & Kimura (1974) and Geyer & Chang (1958), thenthe cells must be producing sufficient CO2 to fulfil their requirements. In the fructose-containing cultures, ammonia production is reciprocally regulated by the level offructose, and lactic acid production can be used to predict changes in medium pH.Thus, controlling lactic acid production will control decreases in the pH. However,regression analysis showed that some other components are also involved in determin-ing changes in medium pH. In galactose-containing cultures, the level of galactose atconcentrations giving optimal cell growth does not appear to regulate ammoniaproduction, and lactic acid production is not a good predictor of changes in the pHof the medium. These results indicate that the overall metabolic dynamics of fructoseand galactose in Vero cells are significantly different, and that there are additionalcomponents affecting changes in the medium pH that remain to be identified.

These experiments have shown that pH can be regulated in a bicarbonate-freemedium, such as L15, by the level and type of carbohydrate in the medium. Theapproach of Leibovitz eliminates the need for expensive CO2 regulation or the use ofother organic buffering compounds. Using the L15 medium formula, fructose is thecarbohydrate of choice for Vero and MDCK cells, since a reasonably stable pH andlactate/pyruvate ratio can be maintained in the high density microcarrier cultures.

This research was supported by gifts from Flow General Inc.

REFERENCES

BERNT, E. & BERGMEYER, H. U. (1974). D-fructose. \n Methods of Enzymatic Analysis (ed. H. U.Bergmeyer), p. 1304. Deerfield Beach, FL: Verlag Chemie International.

EAGLE, H., BARBAN, S., LEVY, M. & SCHULZE, H. O. (1958). The utilization of carbohydrates byhuman cell cultures..?, biol. Chem. 233, SS1-558.

GEYER, R. P. & CHANG, R. S. (1958). Bicarbonate as an essential for human cells in vitro. ArchsBiochem. Biophys. 73, 500-506.

IMAMURA, T., CRESPI, C. L., THILLY, W. G. & BRUNENGRABER, H. (1982). Fructose as acarbohydrate source yields stable pH and redox potential in microcarrier cell culture. Analyt.Biochem. 124, 353-358.

ITAGAKI, A. & KIMURA, G. (1974). Tes and hepes buffers in mammalian cell cultures and viralstudies: problems of carbon dioxide requirement. Expl Cell Res. 83, 351-361.

KUN, E. & KEARNEY, E. B. (1974). Ammonia. In Methods of Enzymatic Analysis (ed. H. U.Bergmeyer), p. 1802. Deerfield Beach, FL: Verlag Chemie International.

LEIBOVITZ, A. (1963). The growth and maintenance of tissue-cell cultures in free gas exchange withthe atmosphere. Am. J. Hyg. 78, 173-180.

LEVINE, D. W., WONG, J. S., WANG, D. I. C. & THILLY, W. G. (1977). Microcarrier cell culture:New methods for research-scale application. Somat. Cell Genet. 3, 149-155.

REITZER, L. J., WICE, B. M. & KENNELL, D. (1979). Evidence that glutamine, not sugar, is themajor energy source for cultured HeLa cells. J. biol. Chem. 254, 2669-2676.

RUSSELL, W. C , NEWMAN, S. & WILLIAMSON, D. H. (1975). A simple cytochemical techniquefor demonstration of DNA in cells infected with mycoplasmas. Nature, Land. 253, 461—462.

Sigma Technical Bulletin no. 726-UV and no. 826-UV (1980). St Louis, MO: Sigma Chemical Co.


WAYMOUTH, C. (1981). Major ions, buffer systems, pH osmolality and water quality. In TheGrowth Requirements of Vertebrate Cells inVitro (ed. C. Waymouth, R. G. Ham&P. J. Chappie),p. 106. Cambridge University Press.

ZIELKE, H. R., OZAND, P. T., TILDON, J. T., SEVDALIAN, D. A. & CORNBLATH, M. (1978).

Reciprocal regulation of glucose and glutamine utilization by cultured human diploid fibroblasts.J. cell. Physiol. 95, 41-48.

(Received 22 February 1985 -Accepted 6 May 1985)

HIGH DENSITY MAMMALIAN CELL GROWTH IN LEIBOVITZ ...medium with fructose or galactose, and a similar...

Documents

Transcript of HIGH DENSITY MAMMALIAN CELL GROWTH IN LEIBOVITZ ...medium with fructose or galactose, and a similar...