Journal of Biotechnology, 15 (1990) 71-90

71

Elsevier BIOTEC 00411

Methods for increasing monoclonal antibody production in suspension and entrapped cell cultures" biochemical and flow cytometric analysis as a function of medium serum content Carole Heath *, Robert Dilwith 1 and Georges Belfort Bioseparations Research Center, Department of Chemical Engineering, Rensselaer Polytechnic Institute, Troy, N Y 12180-3590 and 1 Wadsworth Center for Research, New York State Department of Health, Empire State Plaza, Albany, N Y 12201, U.S.A.

(Received 1 November 1988; accepted 20 April 1989)

Summary The growth and antibody production of the SP2/0-derived hybridoma HB124 (ATCC) grown in media containing varying amounts of fetal bovine serum (FBS) were monitored using biochemical and flow cytometric methods. H y b r i d o m a s grown in 100 ml spinner flasks with RPMI-1640 containing varying amounts of serum demonstrated that cell growth, viability and I g G production show significant changes when serum content is decreased from 10.0 to 5.5 to 1.0 and 0.5%. A longer lag phase resulted when the lower serum content media were used. Cellular rates of glucose uptake showed a significant increase as serum levels were lowered. Similarly, exponential phase I g G production rates increased as the amount of serum was decreased, probably as a result of the decreased rate of exponential growth. Flow cytometric analysis showed a similar increase in cellular I g G content as medium serum levels declined. In contrast, the m a x i m u m I g G concentrations were found in flasks containing 1% FBS or above with the lowest concentration in the 0.5% FBS flask being due to the lower numbers of viable cells. Cells grown in microporous hollow fiber reactors were fed with medium containing serum which was decreased stepwise with time. Decreasing medium serum

Correspondence to: G. Belfort, Bioseparations Research Center, Department of Chemical Engineering,

Rensselaer Polytechnic Institute, Troy, NY 12180-3590, U.S.A. * Carole Heath is currently an Assistant Professor in the Thayer School of Engineering, Dartmouth College, Hanover, NH 03755, U.S.A. 0168-1656/90/$03.50 © 1990 Elsevier Science Publishers B.V. (Biomedical Division)

72 content stepwise from 10 to 2.5% resulted in increased antibody production. However, complete removal of serum from the medium resulted in a significant drop in antibody productivity. Cumulative antibody production was equivalent for cells grown entirely in medium containing 10% FBS and for those which experienced a drop to 2.5% FBS. To compare a defined serum-free medium preparation with medium containing 10% FBS, cells were again grown in batch suspension culture and analyzed. The growth rates were similar but there was a significant difference in IgG production rates. The serum-free culture exhibited both higher cellular production rates and higher IgG concentrations. These results indicate that decreasing medium serum content can adversely affect antibody yield because of lower cell viabilities, not because of lower production rates. Use of a defined serum-free medium, as done in this study, results in higher yields because of a higher IgG production rate as well as good cell growth and viability. Hybridoma; Monoclonal antibody; Serum; Flow cytometry; Suspension culture; Entrapped culture

Introduction

Hybridoma cells are typically grown in a basal medium (RPMI, DME, MCDB, F-12, or some combination of these) supplemented with up to 20% fetal bovine serum (FBS). The FBS helps to maintain continued cell growth and antibody production by supplying a complex mixture of proteins, growth factors and other macromolecules. Despite these advantages, the disadvantages of using serum are numerous. Besides being an expensive and sometimes unavailable component of the growth medium, serum is undefined and can vary in composition from one lot to another. Serum may also be a source of unwanted contaminants, such as mycoplasma and other viruses, although this problem has nearly been eliminated by improved processing techniques and testing. Perhaps most importantly, the presence of this heterogeneous supplement severely complicates the subsequent separation and recovery of the desired antibody from the culture medium, increasing the final product cost. The use of serum is particularly undesirable in the preparation of human therapeutic proteins since the presence of foreign antigens can stimulate an immune response, rendering a second treatment dangerous if not impossible. Serum proteins, approximately half of which is albumin, are the most difficult contaminants to remove from culture media (Scott et al., 1987). To circumvent these problems, researchers have been investigating the use of serum-free media for hybridomas (Chang et al., 1980; Murakami et al., 1982; Cleveland et al., 1983; Kawamoto et al., 1983, 1986; Steimer, 1984; Murakami, 1984). Numerous commercial preparations of serum-free and serum-reduced media have become available for use in recent years.

73 Cleveland et al. (1983) have shown that eight hybridoma lines successfully grown with serum-free media achieved levels of IgG production which were comparable to those from cells grown in serum-supplemented media. This behavior was maintained over several months of growth. Tharakan and Chau (1986a) also found similar antibody production rates in fed batch cultures of hybridomas with serum-free and serum-supplemented (10%) media over a period of one months. Takazawa et al. (1988) demonstrated comparable antibody productivities by mouse-human hybridomas from both serum-supplemented and defined serum-free media. Antibody productivity by hybridomas in entrapped systems has been shown previously to be unaffected by decreasing serum content, perhaps because of the high cell densities and the resulting local concentration of endogenous growth factors. Altshuler et al. (1986a) demonstrated that reducing serum content from 20 to 6.9% over a period of 16 d in a hollow fiber membrane bioreactor resulted in maintenance of high antibody yields by hybridomas. This work extends that investigation further by reducing serum levels from 10 to 0% in hollow fiber bioreactors and observing the effect on antibody yield. The thrust of this study is to compare the effects of decreasing medium serum content with use of a defined serum-free preparation. The observations used to make this comparison include hybridoma growth rate, cell viability, glucose uptake rate, antibody production and flow cytometric measurements of total cellular antibody content. The motivation for including the flow cytometric measurements arises from the interest in knowing the effect of changing medium serum content on the distribution of cellular properties, (i.e., antibody and DNA content) rather than simply the mean population value such as average antibody production per cell obtained from measuring medium concentrations. Previous analysis of antibody content in hybridomas by flow cytometry (Altshuler et al., 1986b) has shown that there exist two distinct populations of cells: those containing high IgG and those with low IgG content. Antibody content, as measured by flow cytometry, is believed to be directly related to the antibody production rate of the cell. Further investigations have indicated that both culture age and growth conditions may affect the distribution of IgG content in the two populations (Heath, 1988). To further explore the influence of medium components, ceils grown in cultures with different levels of serum were analyzed using flow cytometry.

Materials and Methods

Murine hybridoma cells (ATCC HB124) producing an IgG2a directed towards bovine insulin were grown at 37 °C in 5% CO 2 atmosphere. The medium used in the variation of serum content experiments consisted of RPMI-1640 (Gibco, Grand Island, NY) and 1.5 g l-1 sodium bicarbonate supplemented with 3.2 ml 50X MEM amino acids, 1.6 ml 100X non-essential amino acids and 1.6 ml 100X MEM vitamins per liter (Gibco, Grand Island, NY). Gentamicin sulfate (Tri Bio Laboratories, State College, PA) was used at a concentration of 25 gg ml -a. Fetal bovine serum (KC Biological, Lenexa, KS) was added as indicated. The defined serum-free

74 medium, hereafter referred to as RDplus, consisted of a 1 : 1 mixture of RPMI-1640 and DME, 2 mM L-glutamine, 15 mM Hepes, 10 mM ethanolamine, 10 mM sodium selenite, 5 mg 1-1 insulin, 5 mg 1-1 transferrin, 2.2 g 1-1 sodium bicarbonate and 110 g 1-1 sodium pyruvate. This medium formulation was kindly suggested by Dr. J. Denry Sato at the W. Alton Jones Cell Science Center in Lake Placid, NY. RDplus was supplemented with the same concentration of amino acids, vitamins and antibiotic as listed above for the serum-containing RPMI medium. Each medium was sterilized via membrane filtration (double layer 0.2 /zm membrane, Millipore Corp., Bedford, MA). The reactors used in this study were 100 and 250 ml spinner flasks (Bellco Biotechnology) and microporous (0.2 ~m) polysulfone hollow fiber modules (Model CFP-C-2-4, A / G Technology, Needham, MA). Samples were removed at indicated intervals for the appropriate assays. Total and viable cell counts were calculated as the average of six counts using 0.4% Trypan blue dye and a hemacytometer. Glucose concentrations were measured in triplicate using the enzymatic Trinder Assay (Sigma, St. Louis, MO). Antibody concentrations were determined in triplicate using a modified ELISA (Altshuler et al., 1986a). Flow cytometric analyses of cellular antibody and DNA distributions were obtained by the method of Altshuler et al. (1985), which consists of ethanol fixation of the sample (106 cells) followed by staining with excess. Meloy fluoresceinated goat anti-mouse IgG 1 or IgG 2 (GAM-IgG] was used for the negative control and GAM-IgG 2 was used as the positive stain) and propidium iodide. The negative control results in a small peak acounting for background and auto-fluorescence. Measurements were taken with an EPICS C flow cytometer (Coulter Electronics, Hialeah, FL). For consistency in the flow cytometric measurements, the instrument was calibrated daily with uniform fluorescent beads. The cell samples were excited at 488 nm from an argon laser. The forward light scatter and red fluorescence (DNA) signals were used as 'gates' to exclude doublets, cell fragments and nonviable cells from the antibody histograms. The green fluorescence peak (antibody) of the negative control for each sample was then aligned to a specific channel by adjusting the voltage of the photomultiplier tube. This was done to account for differences in background and auto-fluorescence among cells in different media and stages of growth so that positive histograms could be compared on the basis of peak location and shape. Using this technique we were able to compare relative amounts and distributions of IgG 2 in the different viable cell populations.

Experimental procedure Variation of serum content in batch suspension culture

Cells were grown in 100 ml spinner flasks with supplemented RPMI containing 10.0, 5.5, 1.0 and 0.5% FBS. Each flask was seeded on day 0 with 6 x 10 4 viable cells per ml (93.6% viable). Samples were taken daily over a five-day period to determine cell counts and viability. The supernatants were saved for analysis of

75 glucose and IgG content. Flow cytometric measurements of cellular IgG and D N A were conducted on days 0, 3 and 5. Variation of serum content in entrapped culture Three microporous hollow-fiber reactors were set up as shown in Fig. 1. On day 0, the extracapillary space (29.5 ml) of each module was seeded with 2.75 × 107 cells per ml. The medium from the 500 ml reservoir, changed every 4 d, was pumped continuously through the hollow fiber lumens. The medium for module 1 was maintained at 10% FBS throughout the 37-d run. The serum level in the medium of module 2 was changed from 10 to 2.5% FBS at day 16. With module 3, the serum levels were decreased from 10 to 7.5% FBS at day 4, to 5% at day 12, to 2.5% at day 20 and finally to 0% FBS at day 28. Reservoir samples were taken every 2 d to measure the antibody concentration by ELISA. Because the membranes were microporous, it was assumed that all or most of the antibody would be transported from the cell region to the medium region (reservoir). Thus samples from the extracapillary space were not collected. Flow cytometric measurements and cell counts were not determined due to the difficulty of obtaining representative samples from the hollow fiber extracapillary space. Serum-free medium in batch suspension culture Two 100 ml spinner flasks, one containing supplemented RPMI with 10% FBS and the other containing supplemented RDplus (serum-free), were used to culture hybridoma cells. The 10% serum flask was seeded at 5.5 × 104 viable cells per ml (95.2% viable) and the RDplus flask was seeded with 6.0 × 104 viable cells per ml

RESERVOIR INLE"=T

AIR I

F I-LTER

l>,=-¢'1 RESERVOIRDRAIN

MEDIARESERVOIR SILICONE TUBING

r~~SAMPLE PORTS, CELLS

--

ROLLER PUMP

L1

PREFILTER

--

®°.--

IJ

HOLLOW FIBER REACTOR

INCUBATOR 37=C

Fig. 1. Hollowfiber system for entrapped culture of hybridomaswith stepwisedecreaseof serum content.

76 (98.1% viable). The cells used to seed the two flasks were originally from one T-25 flask containing R P M I with 10% FBS. Approximately 10 d before the start of the experiment cells from the T-25 flask were divided into two T-25 flasks, one containing RPMI with 10% FBS and the other containing the serum-free medium. This was done to allow the hybridomas a short period to adjust to the RDplus. N o weaning was necessary since the cells adapt quickly to the serum-free medium. Samples were taken over a 5-d period for determination of cell count and viability, glucose concentration and IgG content. Flow cytometric analysis of cellular D N A (not shown) and IgG were also monitored on a daily basis.

Results

Variation of serum-content in batch suspension culture Different levels of serum in medium, in most cases, resulted in different values of the measured cellular parameters. Comparisons among the various batch cultures were made on the basis of behavior in the exponential growth phase not on a day-by-day basis since cells in the different medium conditions were not all in exponential growth at the same time. Decreasing serum content from 10 to 0.5% caused a decrease in the rate of exponential growth from 1.00 to 0.48 d-1 (see Table 1). The growth curves are indicated for each level of FBS in Fig. 2a. Cell viabilities (not shown) were lower for serum levels of 1.0 and 0.5% during the first 3 d of culture. Beyond this age, the viabilities in the 10 and 5.5% serum cultures became lower due to the faster growth of these cells resulting in earlier nutrient depletion and by-product accumulation than in the lower serum flasks. Glucose uptake rates during exponential growth (Table 1) showed a steady increase as serum levels declined. Most importantly, medium antibody concentrations (Fig. 2b) for the 0.5% FBS case was lower than those of the other three cases which were all approximately the same at day 5. However, as shown in Table 1, the IgG production rate per million cells during exponential growth increased as the serum level was lowered,

TABLE 1 Growth and antibody production in spinner flasks with serum content variation Parameter Growth rate of exponential phase (d-1) Glucose uptake rate of exponential phase (mg d -1) per 106 viable cells IgG production rate of exponential phase (/~g d- 1) per 106 viable cells IgG yield referred to medium (mg 1-1) IgG yield referred to serum (mg 1-1)

Medium serum content (%) 10 1.00 2.7-t-0.3 39.1 + 5.0 19.7 197

5.5 0.91 3.2+0.1 44.5 + 7.4 18.0 327

1.0

0.5 0.52

0.48

4.2+0.7

5.0+ 0.2

62.4+ 7.6 20.3 2 026

62.7 + 10.8 16.9 3 374

77

106

b

30

:o

E

• • •

10% 5.5% 1.0%

20 ~, 0.5% 105 C

d~

>

104

•

0

I

I

1

2

•

10%

•

5.5%

•

1.0%

x

0.5% I

3

0 0 0

10

t

4

Time (days)

5

0

0

1

2

3

4

5

Time (days)

in RPMI

Fig. 2. Decreasing serum content medium results in changes in hybridoma growth (a) and medium antibody concentrations for suspension culture

(b).

with the rates being equivalent at serum levels of 1.0 and 0.5%. Thus, the lower medium concentration of IgG was due to a lower viable cell count not to a lower production rate. Interestingly, the highest production rates corresponded to the lowest growth rates during exponential growth. An important economic consideration is indicated in the last row of Table 1 where the amount of antibody produced per liter of serum increases significantly with the decrease in the percentage of serum. The flow cytometric analysis of cellular IgG content shows that the percentage of cells beyond a marker, chosen to fall between the two IgG peaks of the 10% FBS sample on day 3, increased as the serum content decreased for both days 3 (Fig. 3) and 5 (Fig. 4) of culture growth. Because the control peaks for each sample were aligned to the same channel, comparisons can be made between different positive samples (i.e., done on different days, etc.). For this reason, the cursor location chosen on day 3 of the 10% FBS sample was used for all subsequent, samples, and provides the basis for indicating changing percentages of cells exhibiting high antibody content. Comparison of the two histograms (3- and 5-d) for each culture, shows that while the percentage of cells with high IgG content stayed approximately the same for the 10 and 5.5% serum levels, the percentage of high IgG content cells in the lower serum cultures increased significantly between the two analyses. This may be a simple reflection of the state of the cells. Even though the IgG analysis was obtained from live cells by means of the gating techniques, the cellular conditions, e.g. growing well or preparing to die, may be estimated by comparing the viabilities. Ceils in the 10% serum culture showed a tremendous drop in viability from days 3 to 5 (82 to 16%), while those in the 0.5% serum flask experienced a much smaller drop in viability (61 to 48%) between the 2 d. Interestingly, the flow cytometric analysis shows that specific antibody productivity is directly related both to the percentage of hybridomas with high IgG 2 content and to the cellular growth rate. Comparison among the cultures during the exponential growth phase shows that

78

Serum Content Variation

10.0% FBS

40.8~6

1

3 Days Post Seed

5.5% FBS

I I,OZ FBS

51.36 ~.

....

~~All~ .

.

.

.

.

.

.

.

.

O. 5% FBS

51.24

Log Green Fluorescence (IgG)

Fig. 3. Histograms of IgG content versus relative cell n u m b e r on day 3 for the four serum content variation suspension cultures as obtained by flow cytometry. Both peaks indicate positive fluorescence. The negative controls (no t shown) for all samples have been aligned so that the peak of each falls on the same channel.

79 Serum Content Variation 5 Days Post Seed

44.66

z

I. 0% FBS .M

58.4? :{

0.5% FBS

62.6Z

i[.......aliilisli

,,,

..........

NAME. J ~ "~ L-~!~ CH/FIG

Log Green Fluorescence (IgS)

Fig. 4. Histograms of IgG content versus relative cell n u m b e r on day 5 for four serum content variation suspension cultures as obtained by flow cytometry. Both peaks indicate positive fluorescence. The negative controls (not shown) for all samples have been aligned so that the peak of each falls on the same channel.

80 30 • • E

Module1 Module2

20

v

-O >-

o

10

i

0

i

10

i

20

i

30

40

Time (days)

Fig. 5. Cumulative antibody yields in the hollow fiber modules as influenced by medium serum content.

specific a n t i b o d y p r o d u c t i v i t y increases as the percentage of cells i n the high I g G peak increases a n d as the cell growth rate decreases (see T a b l e 1 a n d Figs. 3 a n d 4).

Variation of serum content in entrapped culture It should be emphasized that the a n t i b o d y c o n c e n t r a t i o n s were calculated o n l y for the reservoir; a n t i b o d y i n the extracapillary space of the m e m b r a n e was n o t i n c l u d e d i n the m e a s u r e m e n t s which m a y have resulted i n calculated p r o d u c t i o n rates lower t h a n actual. Fig. 5 shows the a n t i b o d y yields for the three m o d u l e s over a 37-d period. As c a n b e seen i n the figure, the productivities of the 10% F B S a n d the 10 to 2.5% FBS cases were very similar. However, as i n d i c a t e d i n T a b l e 2, the 10 to 0% case showed rate increases as serum c o n t e n t was lowered u n t i l s e r u m was completely removed causing a sharp drop i n productivity. Similarly, decreasing m e d i u m serum c o n t e n t did n o t affect the overall yield except i n the case i n c l u d i n g complete serum r e m o v a l (see T a b l e 3). Since cell c o u n t s were n o t o b t a i n e d d u e to the i n h e r e n t difficulty of collecting a representative sample, the average p r o d u c t i o n rates listed i n T a b l e 3 were based o n extracapillary volume. W h e t h e r the d r o p i n productivity with 0% FBS is d u e to decreased p r o d u c t i o n rates per cell or d u e to

TABLE 2 Antibody production rates in the hollow fiber systems Medium serum content (%) 10 7.5 5.0 2.5 0 Overall

Averageproduction rate per reactor volume * (rag ml-1 d -1 ) Module 1

Module 2

Module 3

23.0 23.0

22.0 23.3 22.7

14.3 19.3 22.5 26.4 6.8 17.9

* Based on 29.5 ml extracapillary space.

81 TABLE 3 Growth and antibody production in hollow fiber systems with serum content variation Parameter

Module 1

2

3

IgG production rate per reactor volume (/~g ml- 1d - l ) Maximum IgG concentration (/~g m1-1) IgG yield referred to medium (nag 1-1) IgG yield referred to serum (nag 1-1)

23.0 10.0 5.6 55.8

22.7 8.2 5.6 95.8

17.9 6.5 4.1 93.0

Module 1 was maintained at 10% FBS throughout the run. Module 2 was run with 10% FBS until day 16 when the serum content was dropped to 2.5%. Module 3 was run with 10% FBS until day 4, 7.5% until day 12, 5% until day 20, 2.5% until day 28 and 0% for the remainder of the run.

lower numbers of viable cells is not known but the previous experiment in suspension culture would support the latter.

Defined serum-free medium in batch suspension culture The growth rates for cells cultured in RPMI with 10% serum and in the RDplus were very similar, as shown in Fig. 6a. See also Table 4. Glucose uptake rates during exponential growth were higher in the culture containing 10% serum probably because of the higher concentration of glutamine in RDplus (584 mg 1-1 versus 300 mg 1-1). Medium antibody concentrations, however, were significantly different (Fig. 6b). Because the cell numbers were nearly equal, the antibody production rate per million cells was much higher for the RDplus culture. Note that comparisons between experiments should be done carefully as evidenced by the differences between the 10% FBS spinner flask cultures shown in Tables 1 and 4. The culture age and growth history have significant effects on growth and antibody production. Comparisons can usually be made with a degree of confidence only within an

10 7

• •

E O. t0

50

Q

10% RDplus

--:E

106

O.

C O

o o -~

10 5

10 4

~

0

,

I

,

,

I

2 4 Time (days)

i

40

30

20 10 0 0

2 4 Time (days)

6

Fig. 6. Comparison of hybridoma growth (a) and medium antibody concentrations (b) for cells grown with RPMI/10% FBS and RDplus in suspension culture.

82 TABLE 4 Growth and antibody production in spinner flasks with RPMI/10% FCS and RDplus Parameter

RPMI/10% FCS

Growth rate of exponential phase (d-1) Glucose uptake rate of exponential phase (mg d -1) per 10 6 viable cells IgG production rate of exponential phase (/~g d -1) per 10 6 viable cells IgG yield referred to medium (nag l-a) IgG yield referred to serum (nag 1-1 )

RDplus

0.99

1.~

3.5±1.0

2.1±0.2

24.5±4.8 4.8 48.0

49.8±6.0 21.7

experiment because each of the flasks for a particular experiment was seeded with cells from one culture vessel. Another important observation from Fig. 6b is that antibody concentrations began to decrease after day 3 or 4, indicating product degradation. This is especially apparent in the RDplus culture. Surprisingly, flow cytometric analysis of cells from the two cultures shows that internal antibody concentrations were approximately equal at day 3 (Fig. 7a) and showed significant differences in the histogram profile only at day 5 (Fig. 7b). Thus, it appears that both production and secretion rates must be higher for the RDplus culture, otherwise a greater accumulation of internal antibody would be observed. One

RPMI/IO%

Day 3

/""

'1 =z

FBS

RPMI/IO% FBS -

-

61.31%

if!" LogGreen Fluorescence(lgG)

-

I~y 5

-

60.34%

- - i t ,

1 i

Log Green Fluo~s~ce (IgG)

Fig. 7. Histograms showing the distribution of cellular antibody for cells grown in RPMI/10% FBS and in RDplus a t (a) day 3 and (b) day 5 of the batch culture.

83 (c)

(a)

16weeks

12 weeks

ca. 112 generations

29*/o

ca. 149 generations

71%

(d)

(b)

18 weeks

13.5 weeks ca. 126 generations

ca. 168 generations

54.5?.

L

....

147.

d a . t I..A Lll.a .

Log Green Fluorescence (IgC)

Log Green Fluorescence (IgG)

Fig. 8. Histograms of the IgG content (log of green fluorescence) versus relative cell number for hybridomas of different ages. Times indicate weeks since revival from liquid nitrogen storage. Notice the gradual decrease in the percentage of cells with higher IgG content as culture age increases. All cells used in flow cytometric analysis for this figure are in exponential growth.

surprising result is that medium antibody concentration started to decrease in the RPMI/10% FBS culture after day 3 yet from Fig. 6b one can see that cell viability is still high. We have no explanation for this observation other than to suggest that perhaps protease release by dying cells may not be the only mechanism of antibody catabolism. Comparison of Fig. 7 with Figs. 3 and 4 reveals an interesting phenomenon, that is, hybridomas at different ages (i.e., number of generations) demonstrate distinct

100 80 (9 tE

60 40

O_

20 0

I

10

12

i

I

14 16 Time (weeks)

I

18

,

20

Fig. 9. Linear decay of the percentage of hybridomas exhibiting high cellular antibody content as shown by flow cytometry.

84

histograms of IgG content. The hybridomas used in the serum content variation study had been cultured for 16 weeks (approximately 72 generations) after being revived from liquid nitrogen storage. In contrast, the hybridomas used in the RPMI and FCS versus RDplus study were only 8 weeks old (about 36 generations). We have seen this type of behavior with the same hybridoma previously as shown in Fig. 8. For a particular batch of cells, the decrease in the number of cells with high IgG content follows the linear decay with time as shown in Fig. 9. We have also observed a dual peak IgG distribution with a second hybridoma (NYSDH-T4), indicating that this is not a cell-line specific phenomenon (Altshuler et al., 1986b).

Discussion

Decreasing medium serum content, whether by a simple reduction in serum level or the use of a defined serum-free preparation (RDplus), has been shown in this study to result in higher rates of antibody productivity per cell for hybridoma HB124. The reason for this is not known but has also been shown by Tharakan et al. (1986b), Samoilovich et al. (1987) and Glassy et al. (1988). In fact, Tharakan and Chau (1986) suggest that serum may actually inhibit antibody secretion although a mechanism for this action was not presented. I t is also apparent from this study that a simple reduction in medium serum level will result in a slower growth rate and an initial drop in cell viability. With serum levels from 10 to 1.0%, the drops in viable cell count are offset by higher specific antibody production rates resulting in a steady yield of protein. However, when the serum level is decreased to 0.5%, the increase in IgG production rate cannot offset the lower viable cell count, resulting in a lower yield of antibody. Use of a defined serum-free medium avoids this lower yield without necessitating the use of serum. Cells counts and viabilities were maintained at values nearly indistinguishable from those obtained using medium containing 10% FBS. The antibody yield was thus greatly increased because of the much higher rate of productivity per cell. These results strongly advocate the use of a defined serum-free medium for culturing HB124 and other hybridoma cells. An important added benefit is the absence of serum proteins (other than those added in small known quantities to the medium) making purification of the product far less complicated and expensive. Decreasing serum content in RPMI medium resulted in both a decrease of the growth rate and an increase in the specific antibody production rate. Glacken et al. (1988) have shown for another hybridoma using initial rate data that antibody production is not a function of medium serum content. This leads us to believe that the specific antibody production rate is dependent in an inverse fashion on the exponential growth rate of hybridomas grown with RPMI medium in batch culture. The same inverse relationship has also been shown by others (Reuveny et al., 1986; Birch et al., 1987; Dean et al., 1988; Miller et al., 1988). In continuous culture, Ray et al. (1989) found that antibody productivity attained a maximum at less than the maximum growth rate, but that the productivity then decreased as growth rate increased. However, Low et al. (1987) demonstrated that the antibody production

85 rate in continuous culture reached a maximum below the maximum growth rate and that this production rate was maintained as growth rate increased. A model by Suzuki et al. (1988) indicates that hybridomas arrested in the G1 phase of the cell cycle (as a result of lowered growth rate) have higher specific antibody production rates than those in other cell cycle phases, suggesting a logical basis for the inverse relation between production and growth rates. With less energy funneled into cell growth and division, it is likely that more energy is available for antibody production. This study has also demonstrated that it is important, especially when using a defined serum-free medium, such as RDplus, to terminate batch runs before the onset of product degradation or alternatively to chemically inhibit protease action (by specific enzyme inhibitor or pH control). Schiaeger et al. (1987) have shown that in the supernatants of serum-reduced or serum-free hybridoma culture significant proteolytic digestion of antibodies occurs when the pH drops below 4.5. As seen in the batch cultures of RPMI/FCS and RDplus, antibody breakdown begins to occur after day 3 or 4 corresponding to the end of the exponential growth phase. At this point in a batch culture, cell death ensues resulting in cell lysis and release of proteases into the medium. According to Karl et al. (1988), cell death is not a necessary requirement for proteolytic degradation of antibody since an acid proteinase is apparently secreted from the hybridomas into the culture medium. Cell death will certainly result in larger quantities being released into the medium. As indicated, the degradation effect is more significant for the RDplus culture most likely because of the absence of serum proteins to dilute the degradation of antibody. This product loss can be avoided by terminating the culture before the start of cell death, using continuous culture with process control a n d / o r collecting the product on a periodic or continuous basis. It should be mentioned that not all hybridomas, or other mammalian cells, will thrive and increase productivity with the same medium. There is a plethora of literature on the requirements of various cell lines in defined serum-free media (see McKeehan et al., 1977; Chang et al., 1980; Murakami et al., 1982; Darfler and Insel, 1984; Iscove, 1984; Murakami, 1984; Sato et al., 1984; Steimer, 1984; Waymouth, 1984; Kawamoto et al., 1986) and the possibilities have certainly not been exhausted. Comparison of the batch and entrapped systems on the basis of their antibody yields is not appropriate in this study because the hybridomas used in each system were not the same culture age or from the same batch. For a fair comparison, cells from a particular culture should be grown simultaneously in the batch and entrapped systems. Only then should product yields be compared. This point is aptly demonstrated by comparing the production rates of IgG for the exponential phase for the two 10% FBS cultures in Tables 1 and 4. Table 1 indicates a production rate of 39.1 mg IgG d -1 per 106 cells while Table 4 shows a value of 24.5 for the same hybridoma cell line but from a different culture (and different age) of cells. For the same reason, and because representative cell counts are difficult to obtain in the membrane modules, specific antibody production rates also cannot be compared between the batch and entrapped systems. The production rates obtained during

86

this experiment were lower than expected and could have been caused by low cell viabilities or low specific productivities. Potentially, however, the hollow fiber bioreactors should result in higher antibody yields than from batch systems as long as high cell viability can be maintained. The existence of mass transfer limitations in membrane systems leading to substrate and waste concentration gradients makes the goal of sustained high cell viability a difficult one. Low cell viability results in protease release and the potential for subsequent product degradation. Entrapped systems may be more likely to foster lower cellular growth rates due to the high cell density which would lead one to expect increased specific antibody productivity as has been shown to occur with the batch experiments in this study (see Table 1). The flow cytometry results indicated that the distribution of antibody content per cell varies both on a short-term (within an experiment) and on a long-term (over many generations) basis. Figs. 3, 4 and 7 illustrate the short-term variations while Fig. 8 demonstrates changes occurring over a longer time period. The younger populations, i.e., those cultures which have undergone fewer divisions since revival from liquid nitrogen storage, generally show a larger fraction of cells with higher antibody content. The experiment examining the effect of serum level in medium on hybridoma growth and antibody production clearly demonstrated that specific antibody productivity was directly correlated both to the percentage of cells in the high IgG 2 peak (from flow cytometric analysis) and to the cellular growth rate. Within an experiment, higher productivity was seen in hybridoma cultures which showed higher IgG 2 content (demonstrating the utility of flow cytometric analysis) and lower growth rates suggesting that cultures be maintained intentionally at lower growth rates. However, these correlations do not always hold between different batches of the same hybridoma cell line. More experiments are needed before any conclusions can be drawn. The possibility exists that cells containing less antibody may also grow faster than those with higher antibody content since the percentage of the former population steadily increases with the culture age. This could be supported in theory by a shortened doubling time as a consequence of lower energy and substrate levels shuttled into antibody production. On the other hand, the increase in the percentage of lower IgG-containing cells may be due to a time- or age-related phenomenon of chromosome mutation of loss. Murine hybridomas, although generally more stable than their human counterparts, are known to experience chromosomal instabilities. The question remains as to whether the frequency of such occurrences is high enough to result in the observed distributions. Regardless of the cause, antibody productivity in continuous cultures is likely to change with time unless a method is found to prevent this behavior. Continuous culture with adequate process control and periodic reseeding of the culture are currently recommended. Conclusion

Studies of growth and productivity in media with various levels of serum content demonstrated that either serum-reduced or defined serum-free media can be used

87

for hybridoma cultivation without loss in antibody yield while simultaneously reducing the projected costs and complications of the purification process. RDplus actually resulted in higher yields than any of the combinations of serum in medium. A disadvantage of using RDplus in batch culture, as shown in this study, is that product degradation occurs quite rapidly and severely when cell death causes release of proteases into the culture medium. Maintaining high cell viability would alleviate this problem and thus comparison of serum-containing and defined serum-free media in continuous culture with process control is the next obvious step. Another possibility, based on the observation that the higher antibody production in serumcontaining medium occurred at suboptimal growth rates, is to alter the defined serum-free medium by removing or adding one or more components to slow the growth rate (while maintaining high viability) to see if antibody production can be further enhanced. Analysis of the cultures in the serum study by flow cytometry demonstrated that the percentage of cells in the high IgG 2 peak is directly correlated to the specific production rate of antibody. Cells with high IgG 2 content exhibited higher specific antibody production rates as well as lower growth rates. This suggests that cultures be intentionally maintained at lower growth rates and that antibody productivity be monitored either by flow cytometry or by ELISA on a periodic basis to observe the decrease in antibody production rate with time and thus dictate the optimum interval for reactor reseeding. Experiments to selectively collect cells with higher productivity using flow cytometric sorting techniques are currently under way and if successful should further improve reactor productivity.

Acknowledgements The authors would like to thank Dr. J. Denry Sato of the W. Alton Jones Cell Science Center in Lake Placid, NY for suggesting the formulation of the RDplus and Dr. Arye Gollan of A / G Technology, Needham, MA for donating the hollowfiber modules. The hollow fiber experiments were conducted in the Chemical Engineering Department at RPI by Dr. Gordon Altshuler, now at 3M Corp., St. Paul, MN and by Judith Sowek, now at Bristol-Myers, Wallingford, CT, U.S.A.

References Altshuler, G.L., Dilwith, R., Sowek, J. and Belfort, G. (1985) Dynamics of cellular property distributions in hybridoma batch cultures. Presented at the ACS 190th Annual Mtg., Chicago, IL. Altshuler, G.L., Dziewulski, D.M., Sowek, J. and Belfort, G. (1986a) Continuous hybridoma growth and monoclonal antibody production in hollow fiber reactors-separators. Biotectmol. Bioeng. 28, 646-658. Altshuler, G.L., Dilwith, R., Sowek, J. and Belfort, G. (1986b) Hybridoma analysis at the cellular level. Biotechnol. Bioeng. Syrnp. No. 17, 725-736. Birch, J.R., Thompson, P.W., Boraston, R., Oliver, S. and Lambert, K. (1987) The large-scale production of monoclonal antibodies in airlift fermentors, In: Webb, C. and Mavituna, F. (Eds.), Plant and Animal Ceils, Ellis Horwood Ltd., Chichester, England, p. 168.

88 Chang, T.H., Steplewski, Z. and Koprowski, H. (1980) Production of monoclonal antibodies in serum free medium. J. Immunol. Methods 39, 369-375. Cleveland, W.L., Wood, I. and Erlanger, B.F. (1983) Routine large-scale production of monoclonal antibodies in a protein-free culture medium. J. Immunol. Methods 56, 221-234. Darfler, F.J. and Insel, P.A. (1984) Growth of lymphoid cells in serum-free medium. In: Barnes, D.W., Sirbasku, D.A. and Sato, G.H. (Eds.), Methods for Serum-Free Culture of Neuronal and Lymphoid Cells, Alan R. Liss, New York, pp. 187-196. Dean, Jr., R.C., Karkare, S.B., Ray, N.G., Rundstadler Jr., P.W. and Venkatasubramanian, K. (1988) Large-scale culture of hybridoma and mammalian cells in fluidized bed bioreactors. In: Moo-Young, M. (Ed.), Bioreactor Immobilized Enzymes and Cells, Elsevier Applied Science, Essex, England, p. 125. Glacken, M.W., Adema, E. and Sinskey, A.J. (1988) Mathematical descriptions of hybridoma culture kinetics: I. Initial metabolic rates. Biotechnol. Bioeng. 32, 491-506. Glassy, M.C., Tharakan, J.P. and Chan, P.C. (1988) Serum-free media in hybridoma culture and moneclonal antibody production. Biotechnol. Bioeng. 32, 1015-1028. Heath, C.A. (1988) Engineering aspects of improved antibody production by hybridomas. Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY, U.S.A. Iscove, N.N. (1984) Culture of lymphocytes and hemopoietic cells in serum-free medium. In: Barnes, D.W., Sirbasku, D.A. and Sato, G.H. (Eds.), Methods for Serum-Free Culture of Neuronal and Lymphoid Cells, Alan R. Liss, New York, pp. 187-196. Karl, D.W., Bohn, M.A. and Flickinger, M.C. (1988) Cleavage of murine IgG2a by an acid proteinase released by hybridoma cells. Presented at the 196th ACS National Mtg., Los Angeles. Kawamoto, T., Sato, J.D., Anh, L., McClure, D.B. and Sato, G.H. (1983) Development of a serum-free medium for growth of NS-1 mouse myeloma cells and its application to the isolation of NS-1 hybridomas. Anal. Biochem. 130, 445-453. Kawamoto, T., Sato, J.D., McClure, D.B. and Sato, G.H. (1986) Serum-free medium for the growth of NS-1 mouse myeloma cells and the isolation of NS-1 hybridomas. Methods Enzymol. 121,266-277. Low, K.S., Harbour, C. and Barford, J.P. (1987) A study of hybridoma cell growth and antibody production kinetics in continuous culture. Biotechnol. Tech. 1, 239-244. McKeehan, W.L., McKeehan, K.A., Hammond, S.L. and Ham, R.G. (1977) Improved medium for clonal growth of human diploid fibroblasts at low concentrations of serum protein. In Vitro 13, 399-416. Miller, W.M., Blanch, H.W. and Wilke, C.R. (1988) A kinetic analysis of hybridoma growth and metabolism in batch and continuous suspension culture: effect of nutrient concentration, dilution rate, and pH. Biotechnol. Bioeng. 32, 947-965. Murakami, H. (1984) Serum-free cultivation of plasmacytomas and hybridomas. In: Barnes, D.W., Sirbasku, D.A. and Sato, G.H. (Eds.), Methods for Serum-Free Culture of Neuronal and Lymphoid Cells. Alan R. Liss, New York, pp. 197-205. Murakami, H., Masui, H., Sato, G.H., Sueoka, N., Chow, T.P. and Kano-Sueoka, T. (1982) Growth of hybridoma cells in serum-free medium: ethanolamine is an essential component. Prec. Natl. Acad. Sci. USA 79, 1158-1162. Ray, N.G., Karkare, S.B. and Runstadler, Jr., P.W. (1989) Cultivation of hybridoma cells in continuous cultures: kinetics of growth and product formation. Biotechnol. Bioeng. 33, 724-730. Reuveny, S., Velez, D., Miller, L. and MacMillan, J.D. (1986) Comparison of cell propagation methods for their effect on monoclonal antibody yield in fermentors. J. Immunol. Methods 86, 61-69. Samoilovich, S.R., Dugan, C.B. and Macario, A.J.L. (1987) Hybridoma technology: new developments of practical interest. J. Immunol. Methods 101, 153-170. Sato, J.D., Kawamoto, T., McClure, D.B. and Sato, G.H. (1984) Cholesterol requirement of NS-1 mouse myeloma cells for growth in serum-free medium. Mol. Biol. Med. 2, 121-134. Schlaeger, E.J., Eggimann, B. and Gast, A. (1987) Proteolytic activity in the culture supernatants of mouse hybridoma cells. Dev. Biol. Stand. 66, 403-408. Scott, R.W., Duffy, S.A., Moellering, B.J. and Prior, C. (1987) Purification of monoclonal antibodies from large-scale mammalian cell culture perfusion systems. Biotechnoi. Prog. 3, 49-56. Steimer, K.S. (1984) Serum-free growth of SP2/0-AG-14 hybridomas. In: Barnes, D.W., Sirbasku, D.A., and Sato, G.H. (Eds.), Methods for Serum-Free Culture of Neuronal and Lymphoid Cells, Alan R. Liss, New York, pp. 237-249.

89 Suzuki, E., Sayles, G.D. and Ollis, D.F. (1988) Cell cycle model for antibody production kinetics. Presented at the AIChE Annual Mtg., Washington, DC, U.S.A. Takazawa, Y., Tokashiki, M., Murakami, H., Yamada, K. and Omura, H. (1988) High-density culture of mouse-human hybridoma in serum-free defined medium. Biotechnol. Bioeng. 31, 168-172. Tharakan, J.P. and Chau, P.C. (1986a) Serum-free batch production of IgM. Biotechnol. Lett. 8, 457-462. Tharakan, J.P. and Chau, P.C. (1986b) IgG production kinetics in serum-free media. Biotechnol. Lett. 8, 529-534. Tharakan, J.P., Lucas, A. and Chau, P.C. (1986) Hybridoma growth and antibody secretion in serumsupplemented and low protein serum-free media. J. Immunol. Methods, 94, 225. Waymouth, C. (1984) Preparation and use of serum-free culture media. In: Barnes, D.W., Sirbasku, D.A. and Sato, G.H. (Eds.), Methods for Preparation of Media, Supplements and Substrates for Serum-Free Animal Cell Culture, Alan R. Liss, New York, pp. 23-68.