Studies on monoclonal antibody production by a hybridoma cell line (C1E3) immobilised in a fixed bed, porosphere culture system

Journal of Biotechnology, 15 (1990) 129-146

129

Elsevier BIOTEC 00533

Studies on monoclonal antibody production by a hybridoma cell line (C1E3) immobilised in a fixed bed, porosphere culture system A.J. Racher, D. Looby and J.B. Griffiths Division of Biologics, PHLS Centre for Applied Microbiology and Research, Porton Down, Salisbury, U.K.

(Received 12 January 1990; accepted 22 January 1990)

Summary The aim of this study was to investigate the potential of fixed beds of macroporous glass spheres as a production process for animal cell products. The growth, metabolism and monoclonal antibody expression of a mouse-mouse hybridoma cell line was investigated in order to both test the potential of and to optimise the system. After the initial growth phase, the culture went into a steady-state phase brought on by glutamine limitation. An event occurred after 120-160 h of steadystate operation which destabilised the culture, causing a decline in productivity, after which the culture recovered. This event was analysed in detail to determine its cause, and whether a major switch in metabolic function had occurred. The parameter which correlated most closely to antibody production rate was oxygen, but as this was kept constant in the void medium of the bed it has to be concluded that oxygen diffusion into the spheres was the regulatory factor. A comparison of the fixed bed and a flask culture identified interesting differences in glucose metabolism between the two systems. The data gave strong indications as to how the productivity of the fixed bed system can be further improved. This includes optimisation of the glutamine concentration and modifying the porous structure of the spheres to improve diffusion characteristics. Antibody production; Hybridoma; Immobilised culture; Physiology

Correspondence to: A.J. Racher, Div. of Biologics, PHLS Centre for Applied Microbiologyand Research,

Porton Down, Salisbury, Wiltshire SP4 0JG, U.K. 0168-1656/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)

130 Introduction

Immobilisation allows more efficient perfusion of animal cells and thus the potential for increasing unit cell density 20-50-fold from conventional batch densities. It also confers more stability to the culture system than free-suspension and provides the opportunity for longer-term culture periods. Most immobilised culture systems are very limited in volumetric scale-up potential and thus any system which is suitable for both high cell density and high volume application is of interest. Fixed beds of solid glass spheres have been widely used for growing anchorage dependent cells in both laboratory and industrial scale processes (Whiteside et al., 1979; Looby and Griffiths, 1987; Brown et al., 1988). However, glass sphere systems have certain limitations. They are not suitable for immobilising suspension cells due to excessive cell washout, and they are essentially low process intensity systems. These limitations have been overcome by substituting porous glass spheres (porospheres) for solid ones. Unit cell density increases significantly, by up to 20-fold for anchorage-dependent cells, and suspension cells can also be grown to similar densities without marked cell washout (Looby and Griffiths, 1988a,b) Siran porospheres are a very suitable matrix for immobilising suspension cells since no complicated immobilisation technique is required and the matrix is non-toxic to the cells. Due to the large pore size (60-300 #m) and open structure of interconnecting pores, cells are physically entrapped by filtration of medium containing cells. Up to 60% of the sphere is void volume. The cells growing in the internal structure of the spheres are protected from surface shear and can thus form dense aggregates. The combination of physical entrapment of cells and the loose adherence of cells in the sphere allows changes of medium without excessive loss of cells. The object of this study was to investigate the production of monoclonal antibody (mAb) and the physiology of hybridoma cells grown in a fixed bed, porosphere culture system. Cells grown in flask culture were also studied to improve our interpretation of the data from the fixed bed system. Flask culture has the advantage that cell densities can be readily monitored.

Materials and Methods

Cell line and medium The mouse-mouse hybridoma cell line CIE3, which secretes an IgG-class mAb against Toxoplasma gondii (Wright and Balfor, 1983), was obtained from Dr. S. Clark (PHLS CAMR). Cells were tested for mycoplasma infection and found to be negative by the Hoechst stain test. This cell line exhibits a pattern of both growth and non-growth associated mAb production (data not shown). The medium used was RPMI 1640 with 5% heat-inactivated FCS (56°C for 45 min) (Imperial Laboratories, Andover, U.K.). Glucose was included at 2 g 1-1.

131

Cell culture For both culture systems, cell seed was taken from mid-exponential phase cultures (4-5 x 105 cells per ml). In the flask culture (designated F2), 100 ml medium in a 175 cm 2 T-flask was inoculated to an initial density of about 105 viable cells per ml. Growth was allowed to continue at 37 o C, and samples were taken daily. At the points shown on the figures, the cells were removed from the conditioned medium by centrifugation and then resuspended in an equal volume of fresh medium, and growth allowed to continue. The carriers used in the fixed bed culture system were 5 mm diam., Siran porospheres (Schott Glaswerke, Mainz, F.R.G.). The design and operation of the fixed bed culture system have been previously described (Looby and Griffiths, 1988a,b). Briefly, the system consisted of a 1 1 (sphere bed volume) packed bed bioreactor for cell growth and a media reservoir vessel containing 10 1 medium. The cell seed was resuspended in sufficient medium to fill the void volume of the bed and introduced into the bottom of the bed of dried beads. Dry spheres absorb cells more fully into the internal structure of the spheres than do damp ones (Katinger, personal communication). The inoculation density was 2.0 x 106 viable cells per ml fixed bed volume. The bed was drained and refilled twice to promote even distribution of the inoculum. Medium perfusion was started immediately to give a linear flow rate of 2 cm min-1 increasing to 20 cm rain-1 with growth. In this study the system was operated in a repeated feed-and-harvest mode with replacement of medium in the reservoir (10 1) approximately every 24 h, after the first 72 h. The actual times of medium replacement are indicated in the figures. A list of the symbols used is given in Table 1.

TABLE 1 List of symbols Symbol

Unit

Definition

PATe Q Qo~y Q'~y. Y Y'

mmol d - 1 mmol 1-a d - l

Total ATP produced by the culture per day Volumetric uptake rate Total oxygen consumed by the culture per day Oxygen consumption rate pre-medium change Estimated volumetric m.Ab production rate Apparent yield of product from substrate (equals the ratio of product formation and substrate utifisation rates)

mrnol d - 1 mmol d- I

/~g m1-1 d-1 tool tool- 1

Subscripts ATP Gin lac

ATP Glucose Glutamine Lactate

NH4 +

Ammonium

glc

Superscripts ox

ATP produced by oxidative phosphorylation

132

Assays Cell numbers were determined in an Improved Neubauer chamber and the viability by trypan blue exclusion. Ammonium was assayed enzymatically using glutamate dehydrogenase (Bergmeyer and Beutler, 1984). Glucose was determined using a Beckman Glucose Analyser 2 (Beckman Instruments Inc., CA, U.S.A.). Glutamine was determined using glutamine synthetase as described by Mecke (1984). Lactate was assayed enzymatically using a commercial kit (Sigma Chemical Co., cat. no. 826-A). Protein was determined by the dye-binding assay of Bradford (1976). Cells were recovered from the porospheres by washing with PBS. The cells were lysed with ice-cold PBS + 0.5% v / v Triton X-100 (30 min incubation on ice). The washed spheres were treated in the same way to lyse any remaining cells. The lysates were pooled and cell debris removed by centrifugation. Lactate dehydrogenase (EC 1.1.1.27: LDH) activity was assayed spectrophotometrically at 30°C (Vassault, 1983) in either cell-free medium or cell lysates. One unit (U) of LDH activity was defined as 1/~mol NADH consumed per minute. Oxygen was monitored using polarographic oxygen electrodes (Ingold Ltd.) placed in the media inlet and outlet. The pH was monitored using a pH electrode (Russell Ltd.). The concentration of mAb secreted into the culture supernatants was assessed using the ELISA assay described by Clark et al. (1988), except that C1E3 antibody was used instead of the OKT3 one.

Calculations The oxygen consumption rates were calculated according to the method described by Lydersen (1987). The total ATP produced per day by the culture was estimated from the total lactate produced and oxygen consumed per day by the method of Miller et al. (1988a). A value of 2.0 was used for the P / O ratio while the glycolytic correction factor fglycolysis had the value of 1.0 (Miller et al., 1988a). Sample means were compared by an ANOVA test. Linear and stepwise regression analyses were done using the Statgraphics statistical package (Statistical Graphics Corp.).

Results

Definition of optimum medium change regime The fixed bed culture system was operated with either a variable percentage (50, 75 and 100%) (designated I1) or 100% (designated I2) change of the medium in the reservoir. When the bioreactor bed was drained at the end of each experiment (358 h for I1 and 382 h for I2), a greater proportion of cells recovered from culture I2 were viable (63%) compared to I1 (46%). The total cell density for I2 was also higher (2.82 X 107 and 3.92 × 107 cells per ml bed for I1 and I2, respectively). The cumulative mAb yield at 358 h for culture I2 was 183.4/xg m1-1 but only 113.4/~g

133 m1-1 for 11 (for the flask culture F2, the cumulative mAb yield was 72.3/tg ml-1). The difference between I1 and I2 is greater than can be accounted for by differences in cell numbers at the end of the experiment. These results clearly show that 100% replacement of the medium in the reservoir gives higher yields of both mAb and cells, with a greater proportion of viable cells, compared to replacement of a variable and lower proportion of the medium. Consequently, further studies in both the flask and fixed bed cultures used a regime of 100% replacement of the medium in the reservoir. Culture growth Culture F2. The graph of viable cell density for culture F2 (Fig. la) shows that the culture was in the exponential growth phase between 24-48 h, in the deceleration phase between 48-94 h, in the stationary phase between 94-290 h, and in the decline phase between 290-360 h. In the stationary growth phase, although the percent viability of the culture exhibited a downward trend, the viable cell density remained relatively constant (Fig. la). This indicates that replacement growth was occurring. This is supported by the observation of positive specific growth rates in the stationary growth phase (Fig. la). In culture F2, the volumetric glucose uptake rate (Fig. lb) increased during the deceleration growth phase to a relatively constant value. The increase in the volumetric glucose uptake rate parallelled the increase in viable cell density. When the viable cell density was relatively constant (between 94-290 h), the volumetric glucose uptake rate was also constant. The decline in the viable cell density at 290 h was accompanied by a decrease in the volumetric glucose uptake rate. The rate of release of LDH activity by culture F2 (Fig. lb) tended to zero in the exponential growth phase, but started to increase upon entry to the deceleration growth phase. In the stationary phase, the rate of release continued to rise, but at a lessening rate. In this period, the rate of release generally increased as the percent viability of the culture declined. The rate of LDH release increased markedly at the end of the stationary growth phase and entry to the decline phase, and was coincident with the decrease in volumetric glucose uptake rate. Culture 12. The glucose uptake profile of culture 12 (Fig. lc) shows that the uptake rate increased rapidly between 40-80 h, after which it slowed but continued to increase at a steady rate for the rest of the culture. The oxygen uptake profile of culture I2 (Fig. lc) shows a steady rise in uptake rate to a plateau at 210-280 h, before declining and then increasing again at the end of the culture. The profile of LDH release by culture 12 (Fig. lc) shows that the rate tended to zero during the first 80 h of the culture. Between 80-190 h, the rate of release of activity increased markedly, but the rate of increase gradually declined. Subsequently the rate decreased between 190-240 h, before increasing to a peak at 290 h and again decreasing.

0

~_.

~ ~-.

=~,~

~8~5

O0 ¢3

=~'zi o g'~=

,~ 0

~

..,

--I

o

Oxygen

o

L°\

" E~,O

LDH

~

uptake

rate

b~

-o

-

-

~_

d "1

rate

;>0

/

_

rate o

uptake

mmol

Glucose

i

/o "~

i°,

. . . . . . . ~ t . c, // 0

oi)o/

~

,,,

,~ko ~x\

release

°/

0

o(

o

o \'~, ",o "~,~,,

'\ \

'~',

o

I

O

./.~"

c, .\\ ¢,

I

,}'"'"/~

\\

\% D

O

mmol 1-1 d "1

~O

F

U m1-1 d -1

i

0 /O

%

ff

o

O

o.

/

.

~"

j,

~ ~

x

~

\1@

~

.

:

•

j

•

•

~

,

¢

10 5 m1-1

~, ~"

<~"~"

~.

~.

• !

d -1

s..~

.A • 5J~ ~

rate

viability

cells

Percent

.

•

..••

t

..

growth

..~_ ~ - ~

Viable

/

'4 ~;3k..~

Specific

.

_

-

-

-

-

-

-

135 TABLE 2 Comparison of cell viability for different positions in the bioreactor bed in culture 12 Position

Total cell density 106 cells per ml bed

LDH activity per protein per cell mU rag-110-6

Top Middle Bottom

27.2 34.3 30.3

5.8 5.2 5.3

Samples of spheres were taken from different points within the bed of culture 12 at the end of the experiment. The total cell density is the sum of cells recovered from PBS washes of the spheres plus the cell density inferred from a nuclei count done on the spheres.

Scale-up of fixed bed The effect of position within the bed upon cell viability was examined. The assumption behind the use of L D H as a marker for loss of culture viability is that the enzyme is only released upon damage of intact (i.e. viable) cells (Racher et al., 1990). Therefore, if a mixture of viable and non-viable cells is lysed, the greater the n u m b e r of viable cells, the more L D H activity that will be recovered. Samples of spheres were taken from the top, middle and bottom of the bed. The data (Table 2) indicate that the total cell counts at the three positions within the bed were similar. A N O V A analysis shows that the L D H activity recovered from the spheres at each point sampled was not significantly different.

Culture metabolism The initial and post-medium change glucose concentrations were always greater than 12 mmol 1-1 (Figs. 2b and 3c). In neither culture studied did the glucose concentration fall below 6 - 7 m m o l 1-1. No significant linear relationship was found between the glucose uptake rate and the post-medium change glucose concentration in the stationary phase of either culture. For culture F2, the stationary phase was taken as 73-263 h, and between 124-310 h for culture 12. In this period of culture F2, no significant linear relationship was found between the post-medium change glucose concentration and the specific growth rate. The m a x i m u m lactate concentrations measured were generally in the range 6-111 mmol 1-1 for both the culture systems (Figs. 2b and 3c). In the stationary phase of the cultures F2 and 12, no significant linear relationship was found between the post-medium change glucose concentration and the lactate production rate. The m a x i m u m a m m o n i u m concentrations measured did not exceed 3.5 mmol 1-1 (Figs. 2a and 3a). It has been found for the hybridoma C1E3 that a m m o n i u m only adversely affects cell growth and viability at concentrations of added a m m o n i u m greater than about 5 mmol 1-1 (data not shown). This suggests that in this study a m m o n i u m levels did not become limiting for cell growth or viability. The glutamine concentrations (Figs. 2a and 3a) at 0 h were about 4 m m o l 1-1. Before the first medium change, the glutamine concentration fell to 0-0.5 mmol 1-1 in both cultures. In the stationary phase of culture F2, the glutamine concentrations post-medium change were greater than 3 mmol 1-1 and decreased to below 1 m m o l

•

~

=

~ .

B"

~-.

~

[]

~ ~. ~-

,-,)~

~

~.~

~.~"

i;!

~"

~u

B

o

0

i o

d~

I fo

tool tool "1

G l u t a m i n e uptake rate m m o l I "1 d -1

Y'NH4+,GIn

............ ""..

6

i ¢~

Ab production rate p g m l " l d -1 0

Y'lac, glc

~'~

~-"

l

,.~,~ 0.

mmol I"1

mol mol "1

.~

concentrations

Glucose and l a c t a t e

0

Glutamine c o n c e n t r a t i o n

=

~.

mmol I"1

6

A m m o n i u m concentration m m o l l "1

o~

137 Glutamine uptake rate

Ammonium and glutamine

mmol I"1 d "1

concentrations

'~

'j.

mmol 1-1

~

o'-

@i i

~

(>

e,

j

•

.

o

~E ~

~..~,

oo

a.

0D

Y'NH4+,GIn

c/~

mol tool -1

~ e,'{

3.~ Mab

production

Glucose and lactate

rate

jjg m1-1 d -1

concentrations

.

t~

oo ATP °x

o Oxygen

production

~ uptake

. rate

~ rate

.

mmol

1-1

.

m m o l d -1

"~ ~ = ,~

~' " Y'lac, glc

tool tool -1

~,

~

.~

mmol d -1

L~

138

1-1 pre-medium change. During the stationary phase of culture 12, the glutamine concentration was generally in the range 0.25-0.75-1 mmol 1-1 pre-medium change, declining from a post-medium change level in excess of 2.5 mmol 1-1. A significant ( P < 0.01) positive, linear relationship was found in the stationary phase of the cultures between the glutamine uptake rate and the post-medium change glutamine concentration. From the regression equations for each culture, it was calculated that glutamine uptake ceased between 0.1-0.7 mmol 1-1. The metabolism of the other amino acids was examined between 150-280 h. These data show that between 170-200 h, there was a decrease in the concentration of most of the amino acids, although for the rest of the period examined, most were being produced by the culture (not shown). We could find no obvious relationship between these changes and variation in the other culture parameters.

Monoclonal antibody production Comparison of the mAb production data (Figs. 2c and 3d) for the two cultures reveals marked differences. Although the production rate increased during the exponential growth phase of both cultures (the assignment of different periods of culture 12 to the various phases of the growth cycle is discussed in the Discussion), differences were observed in the deceleration-stationary phases. In culture F2, the production rate peaked at the end of the exponential growth phase and entered a period of cyclical increase and decrease. In contrast, the mAb production rate of culture 12 continued to increase during the deceleration-stationary growth phase. After reaching a maximum, the rate declined before increasing to a level comparable to the peak value. As differences in mAb production were observed between the two cultures in the stationary growth phase, the relationship between mAb production and the various metabolic parameters in this period was examined further. For culture F2, the most significant linear relationship found was with a function of the glutamine uptake rate (Fig. 2c) and the apparent yield coefficient YNH4+,GIn ( P < 0.001) (Fig. 2c, Table 3). This model shows that there was a negative relationship between the mAb production rate and the glutamine uptake rate plus Y~H4+,CIn" Curves of observed and predicted mAb production rates in the stationary phase of culture F2 are presented in Fig. 4b. These show that the model predicts the observed production rate at different stages of the culture very well. In the stationary phase of culture 12, the most significant linear relationships found were with P~-p (Fig. 3d), with the oxygen uptake rate pre-medium change

TABLE 3 Best fit model for m A b production in the stationary growth phase by culture F2 Model

P

r

y = 22.0564-(5.5789 Y~H4+,GIn)--(4.0764 QGIn)

0.0006

0:9745

The model was calculated by stepwise regression of mAb production rates between 72.5-263.5 h on the metabolic data for this period. P probability that null hypothesis(differencebetween slope and zero is not real) is correct: r correlation coefficientof observed on predicted rate: other symbols as in Table 1.

139

30

-o

'7

a

20

10

,g

-

' :

,o

~o

'

3 6'O

~',

12

o

2

Time

h

F i g 4. Comparison o f observed m A b production rates with the rate curves predicted from the different

models o f m A b production. Culture I2 (a); Culture F2 Co). The times when the medium was changed are indicated by the vertical lines. Symbols: - observed; . . . . . . M o d e l I; . . . . . . . M o d e l II; . . . . . . Model III; . . . . . Model IV.

(Fig. 3d), or a combination of these parameters. The models were significant at P < 0.10 but not at P < 0.05. The best fit models are shown in Table 4. Models Ia and b are equivalent since P ~ e is the product of the oxygen uptake rate per 24 h and a constant. Both models give the same predicted values for the m A b production rate. TABLE 4 Models for m A b production in the stationary growth phase by culture I2 Model Ia Ib II III IV v

y y y y y y

= = = = = =

(0.1252 P ~ ) - 4 . 3 4 7 9 (0.5008 Q'xy)-4.3489 (0.6977 Q~xy*)-4-4358 (0.7712 Q ' x y . ) + ( o . 0 6 1 1 P ~ v ) - ( 2 . 8 1 2 9 QGm)-12.7329 (0.4961 Qoxy* )+(0.0891 Qo~)-14.3157 21.0342-(0.9086 Qon)-(0.0915 Y~H4+,GI,)

P

r

0.0742 0.0742 0.0743 0.1261 0.0946 0.9612

0.5865 0.5865 0.5864 0.7417 0.7024 0.0959

The models were obtained by the stepwise regression of m A b production rates in the period 124-310 h on the metabolic data for this period. P probability that null hypothesis (difference between slope and zero is not real) is correct: r correlation coefficient of observed on predicted rate: other symbols as in Table 1.

140 The best fit models obtained by stepwise regression are I and II. The sample correlation coefficient r is a measure of the degree of association between two variables. Examination of the values of r for each of the models in Table 4 shows that models III and IV are better than I and II. Visual examination of curves of m A b production rate with time suggests that models I and III show the best agreement with the observed data (Fig. 4a), although neither model gives as good agreement as the model for culture F2 (Fig. 4b). A model containing both the glutamine uptake rate and Y NHa+.GIn ' was found not to be significant ( P = 0.96). D a t a for oxygen uptake (and hence ATP production) were not available for culture F2, as it was an un-gassed, flask culture. Consequently, the relationship between oxygen uptake (and A T P production) to m A b production could not be examined.

Discussion Conventionally glucose and oxygen uptake rates are used as markers of cell growth in a culture when direct measurements are not possible. However, these can be contradictory, as for example in Fig. lc, which shows the glucose utilisation rate beginning to decline at 80 h and the oxygen rate at 180 h. Release of L D H , which is a marker for the loss of culture viability (Racher et al., 1990), has proved to be a useful additional measurement to understand the growth kinetics of a culture. Its value was demonstrated in the flask culture (Fig. la,b), where cell counts could be done, and the correlation between high cell viability and low L D H release, and loss of viability with increasing rates of L D H release was established. By analogy with the interpretation of the relationship between the profiles of culture growth and viability with the release of L D H activity for culture F2, it would appear that culture 12 entered the deceleration-stationary growth phase at about 80 h. The profile of L D H release for culture 12 suggests that the culture remained in this phase until about 190 h. At which point marked fluctuations in the culture viability began. The profile of the release of L D H activity suggests that the culture entered stationary phase at about 80 h. Between 80-190 h, the glucose and oxygen uptake profiles suggest that the viable cell density was increasing at a constant rate. In this period the rate of release of L D H activity also increased at a constant rate. Therefore it would appear that the loss of viable cells was balanced by replacement growth to maintain the culture biomass in a steady state (although not a true stationary state), as concluded from the cell growth and viability data for culture F2. Between 200-240 h, the steady state was upset. The interpretation of the relationship between changes in glucose, oxygen and L D H rates and changes in culture biomass and viability is unclear. The glucose uptake data suggest that the culture continued to grow, but the oxygen data suggest a loss of viable cells. These apparently contradictory results may in fact result from changes in the metabolism of one or both compounds, accentuating the fact that both glucose and oxygen

141 uptake are only indirect measurements of growth. The fluctuations in the L D H release rate suggest the rate of loss of culture viability decreased, then increased before it decreased again. This does not agree with the interpretation of either the glucose or oxygen rates, although this lack of agreement does support the idea of a change in glucose and oxygen metabolism. The only conclusion which can be made with any certainty is that the culture was no longer in a steady state. These observations emphasise the point that caution must be exercised when interpreting culture growth from one variable independently of others. These results also need analysis in order to determine what event(s) occurred at 200-240 h to destabilise the culture. Comparison of the glucose concentrations reported in this study (Figs. 2b and 3c) with published studies on the relationship between cell growth and glucose Concentration (Morell and Froesch, 1973; Low and Harbor, 1985; Miller et al., 1988b) suggests that none of the initial and post-medium change glucose concentrations were limiting for growth. Additionally, as 5 mmol 1-1 glucose can support a significant amount of growth (Miller et al., 1988b) then even the lowest pre-medium change concentrations measured would not appear to be growth limiting. Lactate concentrations up to 28 mmol 1-1 have been reported to have no adverse effect on the growth of hybridomas (Reuveny et al., 1986). This suggests that the lactate concentrations observed in this study (never in excess of 7 mmol 1-1) (Figs. 2b and 3c) did not inhibit growth. The average values of Yl~c,~c in the two culture systems for the period of each experiment were markedly different - approximately 1.5 and 0.7 for cultures F2 and 12, respectively (Figs. 2b and 3c). This suggests that there were marked differences in the fate of glucose in the two systems. These results can be interpreted as indicating that in the flask culture glucose was preferentially channelled through the glycolytic pathway to lactate, and that in culture 12, glucose was preferentially channelled into the pentose phosphate pathway, the main alternative pathway of glucose metabolism. Glucose therefore did not appear to be a limiting factor or to cause the loss of the steady state. Glutamine, however, decreased in concentration to below 0.5 mmol 1- ~ prior to the first medium change and this was coincident with a greater than 70% decrease in the specific growth rate (Figs. l a and 2a). At this stage of the culture, none of the other metabolites studied was apparently at a limiting concentration. Therefore, it suggests that exhaustion of the glutamine was a major factor contributing to the slowing of growth and the passage from the exponential growth phase into stationary phase in both cultures. Glutamine must be supplied at 3-1.6 mmol 1-1 in the presence of excess glucose, to prevent marked alteration in the final cell density or reduction in the length of stationary phase (Miller et al., 1988b). On the basis of the relationship between the glutamine uptake rate and the post-medium change glutamine concentration, it would appear that in culture F2 the glutamine concentration fell to a level at which the transport system could no longer function, with the result that it became limiting for growth (see below). However, during this period, the specific growth rate showed a significant ( P < 0.05) negative linear relationship to the post-medium change glutamine concentration. The relationship

142 and interpretation of these observations is unclear to us. It suggests that other factors are also modulating culture growth in this period. In the stationary phase of cultures F2 and 12, a significant negative linear relationship ( P < 0.01 and P < 0.05, respectively) was found between the apparent yield coefficient YNH4+.Gln and the glutamine uptake rate. This suggests that at low glutamine uptake rates, more glutamine was being deamidated - possibly as a result of glutamine metabolism for ATP production - than at higher rates. In the period 160-240 h of culture 12, the ATP production rate exhibited a steady increase (Fig. 3d). But, concomitantly, the post-medium glutamine concentration and therefore glutamine uptake rate decreased (Figs. 3a and b) i.e. the availability of the main carbon source for ATP generation (Donnelly and Scheffler, 1976; Reitzer et al., 1979) decreased. We postulate that as the available glutamine decreases, a greater proportion of the glutamine absorbed by the cells is channelled into the TCA cycle with its concomitant deamidation/deamination. Depending upon the pathway of glutamine oxidation, between 6 and 21 mol ATP per mol glutamine can be formed (Glacken et al., 1986). In the stationary phase of culture 12, the only significant relationship found by linear regression analysis between the estimated values of P2~p and the various metabolic parameters examined was with the daily oxygen uptake rate in this period ( P < 0.001). This relationship was expected as the estimated values of P2~-p are the products of the daily oxygen uptake rate and a constant. This finding excludes the possibility that the variation in P2-~p was the result of fluctuation in the glutamine uptake rate. It suggests that the pathway of glutamine oxidation (and hence YAmGtn) was modulated by some unknown factor, resulting in a greater proportion of the glutamine taken into the cell being oxidised in the TCA cycle, with the concomitant production of ATP. Although the apparent yield coefficient YATP.Gtnmay vary, it does not necessarily follow that YNH4+,~I~ also varies, because of the multiplicity of pathways of glutamine oxidation. An explanation which cannot be discounted is that glucose was channelled into the TCA cycle for ATP production by oxidative phosphoryiation. However, the literature suggests that this is unlikely. Reports for a variety of cell lines suggest that greater than 80% of the glucose consumed is converted to lactate, with less than 5% entering the TCA cycle (Morell and Froesch, 1973; Lavietes et al., 1974; Donnelly and Scheffler, 1976;Reitzer et al., 1979; Lavietes and Coleman, 1980). The work of Zielke et al. (1978) suggests that at low glutamine concentrations glucose is channelled into the pentose phosphate pathway and not the TCA cycle. This indicates that glucose was not being used for the generation of ATP by oxidative phosphorylation, and supports the idea of more glutamine being channelled to ATP generation at low glutamine concentrations. The closest agreement found between the growth phases of the cultures and the various culture parameters was with mAb production. Thus this was analysed in detail to determine which physiological event, if any, was modulating mAb production. In the stationary phase of culture F2, the best model of mAb production shows that the volumetric production rate exhibits a negative relationship to the volumetric glutamine uptake rate and the apparent yield coefficient YNHa+.Gln"During this

143 period the glutamine uptake rate showed a positive relationship to the post-medium change glutamine concentration. From this relationship it was calculated that glutamine uptake ceases at 0.1-0.7 mmol 1-1. Additionally, Yr~H4+.rln exhibited a negative relationship to both the post-medium change glutamine concentration and to its uptake rate. So to maximise mAb production, it will be necessary to optimise the post-medium change glutamine concentration, and hence YNH4÷.CIn and the glutamine uptake rate. Whichever model is considered to be best for culture 12, it is clear that the most important factor affecting mAb production is the oxygen uptake rate - the generation of ATP by oxidative phosphorylation is critically dependent upon oxygen! In the immobilised culture 12, the oxygen level at the top of the bed was generally maintained at about 10% air saturation. The analysis of the data in Table 2 suggests that there was no noticeable gradient in cell density or viability from the bottom to the top of the bed. This indicates that concentration gradients of waste metabolites and nutrients are not adversely affecting reactor performance at a bed height of 20 cm and linear flow velocity of 20 cm min-1. Therefore it is unlikely that oxygen became limiting across the bed. However, the models for 12 suggest that mAb production is dependent upon the oxygen uptake rate. Therefore it is possible that oxygen uptake into the porosphere becomes limiting. Consequently the oxygen transfer characteristics of the porosphere support will need to be optimised in order to improve mAb production. In contrast to F2, the glutamine uptake rate did not make a significant contribution to the models of mAb production by culture 12 even though the glutamine concentrations, both pre- and post-medium change, and its uptake rate were in comparable ranges for both cultures. Comparison of the models for mAb production by cultures F2 and 12 shows that there were fundamental differences in the metabolism of the stationary phase cultures. In this period, the medium change regime was the same for both cultures. So the differences probably result from the cells being grown in different culture systems. In comparison with fluidised microcarriers, it is concluded that porospheres in fixed beds have relatively poor oxygen transfer characteristics. The relatively large diameter of the porospheres (5 mm) results in a longer diffusion path. Additionally, when suspended in a fluidised bed, the movement of each carrier relative to the flow of medium creates a nett pressure differential across the bead, which enhances the mass transfer i n t o / o u t of the sphere (Vournakis and Runstadler, 1989). The better oxygen transfer properties of microcarriers in fluidised beds suggests that higher mAb yields may be possible in such systems compared to fixed bed cultures.

Acknowledgements Siran porospheres were supplied by Mr. M. Radke, Schott Glaswerke, Mainz, F.R.G. We would like to thank Mrs. D. Fellows for expert technical assistance. This work was supported by the Biotechnology Group, Department of Trade and Industry.

144

References Bergmeyer, H.U. and Beutler, H-O. (1984) Ammonia. In: Bergmeyer, H.U., Bergmeyer, J. and Grassl, M. (Eds.), Methods of Enzymatic Analysis. VIII. Metabohtes 3: Lipids, Amino Acids and Related Compounds. Third Edition, Verlag Chemie, Weinheim, pp. 454-461. Bradford, M.M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilising the principle of protein-dye binding. Anal. Biochem. 72, 248-254. Brown, P.C., Figueroa, C., Costello, M.A.C., Oakley, R. and Maciukas, S.M. (1988) Protein production from mammalian cells grown on glass beads. In: Spier, R.E. and Griffiths, J.B. (Eds.), Animal Cell Technology 3, pp. 251-262. Clark, S.A., Looby, D. and Griffiths, J.B. (1988) The adaptation of hybridoma cell lines to grow and secrete monoclonal antibody in serum-free/defined medium. In: Spier, R.E., Griffiths, J.B., Stepherme, J. and Crooy, P.J. (Eds.), Advances in Animal Cell Biology and Technology for Bioprocesses, Butterworths, Guildford, pp. 291-297. Donnelly, M. and Scheffler, I.E. (1976) Energy metabolism in respiration-deficient and wild type Chinese hamster fibroblasts in culture. J. Cell Physiol. 89, 39-52. Glacken, M.W., Fleischaker, R.J. and Sinskey, A.J. (1986) Reduction of waste product excretion via nutrient control: possible strategies for maximising product and cell yields on serum in cultures of mammalian cells. Biotechnol. Bioeng. 28, 1376-1389. Lavietes, B.B. and Coleman, P.S. (1980) The role of lipid metabolism in neoplastic differentiation. J. Theor. Biol. 85, 523-542. Lavietes, B.B., Regen, D.H. and Demopoulos, H.B. (1974) Glutamate oxidation in 6C3HED lymphoma: effects of L-asparaginase on sensitive and resistant lines. Proc. Natl. Acad. Sci. USA 71, 3993-3997. Looby, D. and Griffiths, J.B. (1987) Optimisation of #ass-sphere immobilised bed cultures. In: Spier, R.E. and Griffiths, J.B. (Eds.), Modern Approaches to Animal Cell Technology, Butterworths, Guildford, pp. 342-352. Looby, D. and Griffiths, J.B. (1988a) Immobilisation of animal cells in fixed and fluidised porous glass sphere reactors. In: Spier, R.E., Griffiths, J.B., Stephenne J. and Crooy, P.J. (Eds.), Advances in Animal Cell Biology and Technology for Bioprocesses, Butterworths, Guildford, pp. 336-344. Looby, D. and Griffiths, J.B. (1988b) Fixed bed porous glass sphere (porosphere) bioreactors for animal cells. Cytotechnology 1, 339-346. Low, K. and Harbor, C. (1985) Growth kinetics of hybridoma cells: (2) the effects of varying energy source concentrations. Dev. Biol. Standard. 60, 73-79. Lydersen, B.J. (1987) Perfusion cell culture system based on ceramic matrices. In: Lydersen, B.J. (Ed.), Large Scale Animal Culture, Hanser Publishers, Munich, pp. 169-192. Mecke, D. (1984) L-Glutamine: colorimetric method with glutamine synthetase. In: Bergmeyer, H.U., Bergmeyer, J. and Grassl, M. (Eds.), Methods of Enzymatic Analysis. VIII. Metabolites 3: Lipids, Amino Acids and Related Compounds. Third Edition, Verlag Chemie, Weinheim, pp. 364-369. Miller, W.M., Wilke, C.R. and Blanch, H.W. (1988a) Transient responses of hybridoma metabolism to changes in the oxygen supply rate in continuous culture. Bioprocess Eng. 3, 108-111. Miller, W.M., Wilke, C.R. and Blanch, H.W. (1988b) A kinetic analysis of hybridoma growth and metabolism in batch and continuous suspension culture: effects of nutrient concentration, dilution rate, and pH. Biotechnol. Bioeng. 32, 947-965. MoreU, B. and Froesch, E.R. (1973) Fibroblasts as an experimental tool in metabolic and hormone studies. I. Growth and glucose metabolism of fibroblasts in culture. Eur. J. Clin. Invest. 3, 112-118. Racher, A.J., Looby, D. and Griffiths, J.B. (1990) Use of lactate dehydrogenase to assess changes in culture viability. Cytotechnology, in press. Reitzer, L.J., Wice, B.M. and Kennell, D. (1979) Evidence that glutamine, not sugar, is the major energy source for cultured HeLa cells. J. Biol. Chem. 254, 2669-2676. Reuveny, S., Velez, D., MacMillan, J.D. and Miller, L. (1986) Factors affecting cell growth and monoclonal antibody production in stirred reactors. J. Immunol. Methods 86, 53-59. Vassault, A. (1983) Lactate dehydrogenase: uv-method with pyruvate and NADH. In: Bergmeyer, H.U., Bergmeyer, J. and Grassl, M. (Eds.), Methods of Enzymatic Analysis. III. Enzymes 1: Oxidoreductases, Transferases. Third Edition, Verlag Chemie, Weinheim, pp. 118-126.

145 Vournalds, J.N. and Runstadler, P.W. (1989) Microenvironment: the key to improved cell culture products. Bio/Technology 7, 143-145. Whiteside, J.P., Whiting, B.R. and Spier, R.E. (1979) Development of a methodology for the production of foot-and-mouth disease virus from BHK21 C13 monolayer cells grown in a 100 1 (20 m2) glass sphere propagator. Dev. Biol. Standard. 42, 113-119. Wright, J.P. and Balfor, A.H. (1983) Monoclonal antibodies to Toxoplasrna gondii. Parasitology 87, LXVI. Zielke, H.R., Ozand, P.T., Tildon, J.T., Sevdalion, D.A. and Cornblath, M. (1978) Reciprocal regulation of glucose and glutamine utilisation by cultured human diploid fibroblasts. J. Cell Physiol. 95, 41-48.