Cytotechnology 5: 129-139, 1991. 9 1991 Kluwer Academic Publishers. Printed in the Netherlands.

Long term and large-scale cultivation of human hepatoma Hep G2 cells in hollow fiber bioreactor Cultivation o f human hepatoma Hep G2 in hollow fiber bioreactor

Jiuan J. Liu 1, Bor-Shiun Chen 1, Te-Feng Tsai 1, Yun-Ju Wu 1, Victor F. Pang 2, Amy Hsieh 1, Jih-Han Hsieh 1 and Tong H. Chang 1

1Development Center for Biotechnology, 81 Chang Hsing Street, Taipei, Taiwan, ROC; 2pig Research Institute, Taiwan, P.O. Box 23, Chunan, Miaoli, Taiwan 35099, ROC Received 12 March 1990; accepted in revised form 23 August 1990

Key words: Hep G2, hollow fiber bioreactor

Abstract Long-term and large scale cultivation of an anchorage-dependent cell line using an industrial scale hollow fiber perfusion bioreactor is described. Hep G2 cells (a human hepatoma cell line) were cultivated in an Acusyst-P| (Endotronic) with a total fiber surface area of 7.2 m 2 (6 • 1.2 m 2) to produce Hep G2 crude conditioned medium (CCM). Pretreatment of the cellulose acetate hollow fibers with collagen enhances the attachment of the anchorage-dependent cells. We have succeeded in growing the Hep G2 cells in an antibiotics- and serum-free IMDM medium, supplemented with 50 ~tg/ml of Hep G2 CCM protein at inoculation. The Hep G2 cells replicate and secrete CCM protein in quantities comparable to those produced in DMEM containing 10% fetal calf serum (FCS). The highest CCM protein productivity during the 80-day cultivation was 1.1 g/day with a total of 30 g of protein accumulated. Hep G2 CCM (20--40 btg protein/ml) was comparable to or even better than 10% FCS in supporting the growth of Molt-4 (a human T leukemia cell line) and FO (a mouse myeloma cell line) cells in vitro. The availability of this large amount of Hep G2 CCM will aid the further purification and characterization of growth factor(s) which could be used as serum substituents.

Introduction Mammalian cell culture is becoming increasingly important in the productions of natural biomolecules and recombinant products for research and therapy. The growing demand for these products has prompted the development of long term mammalian cell culture to large scale and long term operation. Concentrating mammalian cells by immobilization procedures not only increases

the cell density but also the unit productivity, and therefore smaller bioreactors are required (Van Brunt, 1986; Merten, 1987). Reported methods of animal cell immobilization include microcarriers (Van Wezel, 1967), microporous matrices (Young & Dean, 1987), microencapsulation (Goosen et al. 1985), gel entrapment (Schreirer et al. 1982) and hollow fiber et al. 1981). The hollow fiber bioreactor is one of the most attractive approaches. The successful use of hollow fiber

130 module for the cultivation of mammalian cells was first reported by Knazek et al. in 1972, was then reported for the culturing of mammalian, plant, growing bacterial and yeast ceils (Belfort, 1989). Hollow fiber systems offer many advantages, including the support of large numbers of ceils in a small volume, the concentration and isolation of cellular products before harvest, and the increase of per-cell productivity by providing a protective and nutrient-rich environment (Knight, 1989). Both anchorage-dependent and suspension culture cell lines can be grown in these bioreactors, depending on the types of hollow fiber used. Suspension cultures grow well in cellulose acetate fibers and anchorage-dependent cells grow well in polypropylene (Van Brunt, 1986). We have successfully cultivated hybridoma cell line over extended periods along with the production of monoclonal antibody using cellulose acetate fibers in a large scale Acusyst-P| (Endotronics) hollow fiber bioreactor (Cheng et al., in press). We herein describe the successful cultivation of an anchorage-dependent cell (Hep G2) along with the production of growth factors using antibiotics- and serum-free medium in the same system.

Materials and methods Cells The human hepatoma cell line, Hep G2, was obtained from Dr. B. Knowles. Cells were initially grown in Dulbecco's modified Eagle's medium (DMEM) (Nissui) supplemented with 20 mM L-glutamine, 10 mM Na-pyruvate, 10 mM non-essential amino acid and 10% FCS. When slightly subconfluent, the cells were washed and maintained in the above medium without FCS. Hep G2 CCM was collected twice a week. Cell debris were removed by centrifugation and the CCM was pooled and concentrated by ultrafiltration using an Amicon or Pellicon PM-10 membrane. The protein concentration in the CCM was determined by the Bio-Rad assay (Bio-Rad La-

boratories, Richmond, CA) using bovine serum albumin as the standard protein.

Media To study the effect of medium compositions and FCS on the growth and product secretion of Hep G2 cells, 3 • 105 Hep G2 ceils were plated into a T-25 flask in the following media: DMEM, Iscove's modified Dulbecco's medium (IMDM) without or with various supplements (Ham's F12, pyruvate, glutamate, non-essential amino acid) in the presence (10%, 1%, 0.3%, 0.1%) or absence of FCS. Nonadherent cells were removed after 24 hrs. One hundred and ten hrs after inoculation, serum free medium was replaced. Cultures were terminated at 225 hrs after inoculation, the percentage of cell confiuency was recorded, the conditioned medium was harvested and total protein was determined. The effects of collagen a n d Hep G2 CCM on the attachment and growth of Hep G2 cells in serum free IMDM were studied. T-25 flasks were incubated with 5 ml collagen (10 ~tg/ml) at 37~ for 30 min. After washing the flasks several times with PBS, 2 • 104/cm 2 Hep G2 cells in serum free IMDM were inoculated. The nonadherent cells were removed 24 hrs later. The cultures were replaced with either serum free IMDM medium or IMDM supplemented with various amounts of Hep G2 CCM. The adherent Hep G2 cells were removed from the flask by trypsinization and counted in a hemocytometer every 24 hrs since inoculation for a week. The plating efficiency was the percentage of Hep G2 ceils attached to plates after 24 hrs as compared to the initial seeding density. The doubling times were calculated from the growth curves of individual batch static cultures.

Hollow fiber bioreactor An industrial scale Acusyst-P| (Endotronics, Coon Rapids, MN) hollow fiber bioreactor was used in this study. The Acusyst-P| instrument is designed to automate the production of mamma-

131 Solonoid Valve C02. >"'~" . . . . . . . "],. . . . . . . . . . . . -~---< air

~"-"~'-" IC Vent pH sample

t,

IC

IC flush

.~

pH Probe ....

] Bellow Pump 600 mllmin

"j

Air

Compressor

Check Valve

-~

135, ~

[ -"

1~

Chamber / ' " )a

I,. I .

I! 11

.~ Filter

e-._ ~a~

0.2/I m Filter

r'-~

i I I Integration

over flowl . . . . -_-_-_-_-]I Chamber

. . . . . . . . . . . . . . . . . . . . . . . . .

_.~

,o j

?

i Filter o.~176

xC hush medium reservoir & glucose supplement

/~ Gas Exchanger (polypropylene)

v.o. Probe

?=~

%

~_

harvest

I -o DO. sample } & metabolite

Fig. 1. Schematic diagram of Acusyst-P~ (Endotronics) system. Only one cartridge with one fiber is shown for simplicity sake.

lian cell secreted products. The system components are directed by the Acusoft-P software, monitor and feedback control pH, DO, temperature, glucose and lactate to user-defined set points. A schematic diagram of Acusyst-P| is shown in Fig. 1. The bioreactor consists of 6 hollow fiber cell culture cartridges with a total surface area of 7.2 m 2. Fiber membrane is made of Cuprophan (Cellulose Acetate) with a molecular weight cut off 6,000-10,000. The bioreactor includes 2 independent circulation systems, namely the integration circuit (IC) (the fluid path through the lumen of the bioreactor) and the expansion circuit (EC) (the fluid path through the extracapillary space (ECS)). The semi-permeable nature of the membrane allows nutrients and dissolved gasses to diffuse into the ECS, and waste products from the ECS into the luminal flow, while the cells and high molecular weight products are retained in the ECS. The software recognizes both the growth and production phases. The growth phase is defined to be the period with relatively higher pH environment, low cell density, low nutrient

requirements, and minimal need for additional medium pumping or waste removal. The production phase is the period with relatively lower pH environment, with high cell density, high nutrient requirements and maximal need for additional medium pumping or waste removal. Cell growth is monitored by the rate of glucose utilization and lactate production. The glucose and lactate setpoints, that are the minimal glucose and maximal lactate concentrations (mg/100 ml) desired in the flowpath, are determined as the time when the production per cell number per hour is the greatest. The growth and production phase pH setpoints are taken respectively as the points when the growth rate of the cells is the greatest and when the cells have entered into their secondary lag phase before starting into the death phase. These parameters were used to automate control of the pH, glucose and lactate level to the optimal cell growth condition. The Acusyst-P| system is located in a 10 thousand grade clean room. The hollow fiber culture cartridges, gas-exchanger and bellow pump were obtained in sterile condition from the

132 manufactures. Leaving out these components, the unit was constructed in parts and steam sterilized (121~ 55 rain) and were subsequently put together inside a laminar flow cabinet. The IC pre-cartridge samples were taken and the glucose and lactate concentrations were determined by assay kits purchased from Sigma (16UV, and 726-UV). The EC product samples were taken from the sample port near the harvest line. The total protein concentrations in the harvest were determined by Bio-Rad.

Morphologic examination

The cartridge was initially fixed with 10% neutral buffered formalin followed by filled with a 2% agarose solution. The cartridge was then trimmed longitudinally and sagittally, embedded in paraffin, sectioned at 5 ~tm, stained with hematoxylin and eosin (H&E), and examined by light microscopy.

Results and discussion Hep G2 CCM functional assay

Hep G2 CCM as serum substituent

The human T leukemia cell line Molt-4 was obtained from Cell Bank (Veterans General Hospital, Taipei, Taiwan, ROC) and the mouse myeloma cell line, FO from ATCC. Cells were routinely maintained in RPMI-1640 medium with 10% FCS. For the proliferation assay, 2 • 104 viable Molt-4 and FO cells at either lag (5 • 105/ml) or stationary (1 • 106/ml) phase were washed 5 times with serum free RPMI-1640, and were cultured for 4 days at 37~ in the 96-well microplate in the presence of various amounts o f Hep G2 CCM (32 to 0.5 ~tg total protein/well). Control cultures included cells incubated in RPMI 1640 with 10% and without FCS. The cultures were then labeled with 1 ~tCi of 3Hthymidine (6.7 Ci/m mole) (New England Nuclear, Boston, MA) for 18 hrs at 37~ The cells were recovered by filtration on glass fiber filters, which then were dried and the radioactivities were measured in a liquid scintillation counter in the presence of 2 ml Omni-Szintisol (Merck, Darmstadt). The A ct/min represents the mean difference of triplicate determinations in ct/min between cultures containing CCM or 10% FCS vs cultures without FCS. Enhancement ratio was calculated by dividing the A ct/min of cultures containing CCM by the A ct/min of cultures containing 10% FCS.

Serum is traditionally added to cell cultures to provide hormones, growth factors, binding and transport proteins, and other supplementary nutrients. However, the sources of serum are limited and their compositions are subjected to variation and seldom well defined. Thus, development of serum-flee and/or serum substituents is important. In addition, cultivation of mammalian cells under defined serum-flee conditions reduces the risk of virus and mycoplasm contamination, simplifies the purification of protein products and possibly reduces the cultivation cost. Knowles et al. (1984) isolated a human hepatoma-derived cell line Hep G2 and showed that 17 of the major human plasma proteins (o~-fetoprotein, albumin, transferrin, etc.) are synthesized and secreted by these cells. Hep G2 cells can synthesize and secret endothelial cell growth factor (ECGF) (Bowen-Pope, 1984) and platelet-derived growth factor-like substance (Mckeehan et al. 1986). Transcript of insulin-like growth factor IAI and transforming growth factor o~/B have also been detected in Hep G2 cells ( S u e t al. 1989). These findings strongly suggest that Hep G2 CCM may be a good candidate for serum substituent. We herein describe in this paper the long-term and large scale cultivation of Hep G2 ceils and production of CCM in antibiotics- and serum-free medium using industrial scaled hollow fiber bioreactor Acusyst-P| (Endotronics) System.

133 Table 1. Effects of medium compositions, FCS on the growth (percent confluency) and product secretion (total protein) of Hep G2 cells

Total protein (~tg/m!)

Confluency (%) [FCS] in medium

0%

IMDM 0 DMEM 0 DMEM (Supplemented) a 0 IMDM + F12 b 30 DMEM + F12 b 0 DMEM (Supplemented)a + F12 b 0

0.1%

0.3%

1%

10%

0%

0.1%

0.3%

1%

10%

70 30 5 90 0 0

100 65 10 100 10 5

100 90 40 100 90 45

100 100 100 100 100 100

0.0 0.0 0.0 30.8 0.0 0.0

51.2 26.0 17.6 60.4 0.0 0.0

68.4 37.6 18.4 70.8 28.4 29.6

66.0 56.8 25.2 85.2 62.4 38.4

94.0 72.0 54.4 ND ND ND

aDMEM supplemented with 20 mM L-glutamine, 10 mM Na-pyruvate and 10 mM non-essential amino acid. bmedium contained a quarter of Ham's F12 medium.

Development of serum-free medium for Hep G2 cultivation To develop serum-free medium for optimal cultivation of Hep G2 cells, we have cultured Hep G2 cells in various basal media in the presence or absence of various supplements. The effects of medium compositions and FCS on the growth of and products secreted by Hep G2 cells were compared. The results (Table 1) indicated that both the IMDM (0.3% FCS) and the DMEM (10% FCS) resulted in 100% confluency and 68-72 gg/ml protein production. Similar but a little better responses were found in cultures containing IMDM and Ham's F12 (25%) in the presence of 0.3% FCS. In the absence of FCS, none of the media supported the growth of Hep G2 cells. Anchorage-dependent mammalian cells must adhere to some sort of substratum in order to proliferate and differentiate and this adhesion is

regulated by some specific serum protein(s). Collagen is often employed as a substrate for cultured cells (Kleinman et al., 1981). Conditioned medium of Hep G2 cells stimulated the growth of another human hepatoma cell line HA22T cells (which grew poorly in the absence of FCS) and primary rat hepatocytes in vitro (Dr. C-K Chou, personal communication). We have studied the effects of collagen coating of T-flask and Hep G2 CCM on the attachment and growth of Hep G2 cells in a complete serum-free medium. As expected, in serum-free IMDM medium, the attachment of Hep G2 cells were enhanced when Tflask with 10 gg/ml of collagen coating was employed. The percent plating efficiency of Hep G2 cells in collagen coated flask was comparable to that found in cultures containing 0.3% FCS (Table 2). While in the absence of FCS, collagen coating alone did not support the growth of Hep G2 cells, but in the presence of 50 gg/ml of Hep G2 CCM protein, the doubling time of Hep G2

Table 2. Effects of collagen coating of T tasks and medium compositions on the attachment (plating efficiency) and growth (doubling time) of Hep G2 cells Collagen coating

Medium

Plating efficiency (%)

Doubling time (hrs)

+ + + +

IMDM IMDM + 0.3% FCS IMDM IMDM + 50 gg/ml CCM a IMDM + 10 ].tg/ml CCM a IMDM + 5 I.tg/ml CCM a

55.0 80.0 77.5 77.5 77.5 77.5

75.2 32.7 50.4 34.1 39.6 45.6

aHep G2 crude conditioned medium [protein].

134 a 80

- 64 Viable Cell Number 70"

58

60

52

50

46

A ,o

o x

~E

J~ E -,z Z

940

4O

34

.~

-28

~-

9

3O _e .Q r

20 - 22 10-

-16 0

10 30

60

90

120

b

Time

150

180

210

240

(hr)

560

7.6 ......... e .......

490

420

........... B. ........... ..... --.m m....

~

............

~

"El

E

.R

[Glucose]

~-

It actatr~

9

#t

7.4 7.2

g

350

~

280

6.8

==

210

6.6

7.0 9,~.-

..... "to.

o 2

140

6.4

i.i.o 6.2

70

6.0 30

60

90

120 Time

150

180

210

240

(hr)

Fig. 2. Batch culture data for Hep G2 cells. a. Total numbers of viable cells (Q) and total protein concentrations in Hep G2 conditioned medium. (e) b. The pH, glucose and lactate concentrations in Hep G2 conditioned medium.

cells was comparable to that of the cultured medium containing 0.3% FCS. Thus, coating flask with collagen enhances attachment, and Hep G2 CCM supports growth. In the combination of both, Hep G2 cells can grow in serum free IMDM medium.

Large-scale cultivation in hollowfiber bioreactor An industrial

scale hollow fiber bioreactor,

Acusyst-P| (Endotronics) was used to bring Hep G2 cell culture process to large scale production levels. To determine the desired pH, glucose and lactate setpoints for optimal growth of Hep G2 cells, batch static cultures were set up. Hep G2 cells were inoculated into collagen coated T-flask and after twenty-four hrs the nonadherent cells were washed away and IMDM containing 50 ~tg/ml Hep G2 CCM protein was added which then was replaced by fresh IMDM medium 72 hrs later. Batch cell growth, product production, and

135 of fibers and between various cartridges, the medium was circulated 5 times and then stopped. The integration circuit (IC) circulation was started approximately 2 hrs after inoculation using serum-free IMDM medium. Samples for pH, glucose, lactate were taken from the IC pre-cartridge sample port and were assayed every day. Medium pump rates were automatically controlled to the set points by the Acusoft-P software. Cycling between IC and EC was started after the 7th day of the inoculation to minimize gradients occurring in EC and also to permit equal distribution of nutrients and removal of waste products across the whole Cartridge. The system was switched to production phase at the 57th day after the inoculation. Conditioned medium was harvested after the 7th day of the inoculation from the EC sample

metabolite formation of the Hep G2 cells shown in Fig. 2a and Fig. 2b were comparable to those found in cultures containing 1% FCS (Data not shown). The set points of pH, glucose and lactate in growth and production phase were 7.43, 3.35 g/l, 0.95 g/1 and 7.2, 1.77 g/I, 1.64 g/l, respective-

ly. The hollow fibers (Cuprophan) were pretreated with collagen (10 I.tg/ml) for 30 min at 37~ followed by rinsing with distilled water and subsequently with sterile IMDM medium before the inoculation. A total of approximately 8 x 108 viable Hep G2 cells in 300 ml IMDM supplemented with about 40 mg Hep G2 CCM protein were inoculated into the extracapillary space of six hollow fiber cell culture cartridges. To promote an even distribution of cells on the bundle

a

7.50

3000.

.........i~....... GlucoseUptakeRate 2700 LactateProductionRate OxygenUptakeRate 2400

~ ;~ ~\' r/~ q~e--~'[

6.75 6.00

A~.....

X

"~" t'- ~ 2100

5.25

,, E18oo

/

..5o .~

ik

=,',-

~3 ~~a)~11200900 ~U

~ / ~ i

$

~

2.25:1 rl{)~::~x,_ o

600

1.50

3 0 0 -I

0

~

. ~

8

16

~t

24

|- 0 . 7 q

32

40

48

56

64

72

80

Elapsed Days Fig. 3a. Cultivation of Hep G2 cells in Acusyst-P| systemfor production of Hep G2 conditionedmedium. Glucoseuptake rate (A),

lactate production rate (I) and oxygenuptake rate (I).

136

Fig. 3b. Cultivationof Hep G2 cells in Acusyst-P| systemfor productionof Hep G2 conditionedmedium.Growthof Hep G2 cells on

the extemalsurfaceof a hollowfiber (arrow heads)afterbeingculturedfor 90 days.The cells are closely packedand someof themare degeneratingand necrotic (large arrows) x 1250. Inset: Rosetteformation(small arrows) • 800. H&E. port and continued for 73 days. After the 80th day, a semi-batch perfusion system with medium change of every 5-7 days was replaced. One of the cartridges was removed at 90th after inoculation, fixed and processed for morphological examination. The time course changes of glucose uptake rate, lactate production rate and oxygen uptake rate (See Appendix for calculations) shown in Fig. 3a indicates that the maximal growth of Hep G2 cells occurred around 48 days after inoculation with the maximal glucose uptake rate and lactate production rate at 2800 mg/hr and 1350 mg/hr, respectively. All the six cartridges turned to opaque and were covered with cells. Histologically, after being cultured for 90 days, the cells were closely packed to form dense, solid cell masses of various sizes, ranging from 10 to 30 or

more layers, in the spaces among the fibers (Fig. 3b). In some areas the cells were arranged in a palisading manner around centrally located clear spaces as rosettes (Fig. 3b inset). Cellular degeneration and necrosis, characterized by increased cytoplasmic eosinophilia and vacuolization, as well as pyknosis and karyorrhexis, were frequently observed. These changes were more often seen in the central region of those larger cell masses, although they were also present in the areas close to the fiber wall. The total number of viable Hep G2 cells after 80 day of incubation in 6 cartridges was estimated to be 1.1 • 1011 by glucose uptake rate. The protein production rate was influenced by the harvest pump rate (Fig. 3c), the product concentration (Fig. 3d) and the pH (Fig. 3e). In particular, a negative correlation between protein concentration and production rate suggested that

137 e

C

120

7r,0

.... ~..-

.... 'O""

Pr~lelflI~0~1. aide

t

Hmeid PumpR m

1200

110

1200

-.-

P~oitln Plod. Rill

7 55

pH

100 1000

?

90 "C

E

75O

1000

i

A

7 45

PJOO

800

7 4r)

I

Ui 6OO

r

!

=j

o

50"

i

735

" 600

~-

7 30

F40 ; ~-

a. 40o

I. 0 ;;

o4o0

. - 30

7 25

is

720

-20

200

2OO 715

10 Y

9

0

I 9 10

I 20

"

I 9 30

I

40

9

I 9 I SO 60

9

l 70

0

0

9

7 10

80

10

40

50

Elapsed

Days

r

70

RO

f

d 4

3000

30

Deya

Elapsed

3300

20

.... 0 " "

....O "

1200

1200

IPtotllnl

4~

Ploleln I~,od Rile

i 11 -10

PVolllnP~od RIll

ilonnll~O

ConvIt

2700 1000

9 1000

O

t

On

, ,,

2400 2100 800

800

070 o~ 1800 E

w

tr II

~ ' 1500 I o 1200

6OO ~o

9/00

n

600

oc, g

0 tl,.

200

> c

?(l(I I) q

3O0 9

o-q 0

10

20

30

40

El~nsad

50

60

70

AO

q 9 10

I 9 213

Days

!

9

qf)

= 9 4Q

12-Inp q ~ d

f_~... T.~T---~--~rl r'f)

Dayq

Fig. 3c-f. Cultivation of Hep G2 ceils in Acusyst-P| system for production of Hep (32 conditioned medium. c. protein production rate (rag/day) (- O -) and harvest pump rate (ml/hr) (- + -) d. protein production rate (rag/day) (- O -) and pH (- + -). e. protein concentration (mg/1) (- + -) and protein production rate (rag/day) (- O -) f. protein production rate (rag/day) (- O -) and conversion ratio (- + -)

i

Q ;)

7fl Pfl

138 production rate was under autocrine feedback regulation until the maximal growth of Hep G2 cells was obtained. The maximal productivity was 1.1 g/day and a total of 30 g of Hep G2 CCM protein was produced. The cells utilize glucose as carbon source for growth with the formation of lactate. The ratio of moles of lactate produced to moles of glucose utilized (conversion ratio) was 0.85 at the peak of cell growth and 0.25 at the initial culture period (Fig. 3f).

Hep G2 CCM functional assay The effects of Hep G2 CCM and FCS (10%) on the growth of Molt-4 and FO ceils were studied. The results (Fig. 4) indicated that both cell types grew comparably or even better in medium containing Hep G2 CCM with protein concentrations of 20-40 gg/ml than in medium containing 10% FCS. Thus, Hep G2 CCM may be used as a serum substituent. In conclusion, using antibiotics- and serumfree medium, a industrial scale hollow fiber bioreactor was successfully used for the large scale and long term culture of the adherent cell Hep

2?"

24 "

9

a

MOII.4 , L

Moll 4 , S 21" ....... t .......

0

FO , k

18"

0

n-"

FO S

15"

u I-

12"

W

$

10

15

20

25

30

35

Hop G2 C C M Protein ( la g/well)

Fig. 4. Dose responsive enhancement in growth o f Hep G2 conditioned medium. Molt-4 (-), FO ( - - - ) , lag phase (O) stationary phase (O).

G2. Hep G2 CCM is better than 10% FCS in supporting the growth of Molt-4 and FO cells in vitro. Further purification and characterization of growth factor(s) in Hep G2 CCM are underway.

Acknowledgement The work was supported by grant (DCB-038P301-78-05) from the Ministry of Economic Affairs (MOEA) of Taiwan, the Republic of China. The authors would like to thank Ms. ShyangYung Sun and Mr. Chi-Wai Kwang for their technical support and Dr. Chen-Kung Chou and Dr. Stephen Yue for reviewing of the article.

Appendix The definitions of glucose uptake rate (GUR), lactate production rate (LPR), oxygen uptake rate (OUR) and protein production rate (PPR) applied in computer software for the process control of Acusyst-P | G U R = [G x - F t G - E(G x - FtG)]/(1 - E) LPR = Ft(L - LoE)/(1 - E) PPR = Ft(P - POE)/(1 - E) If no fluids have been pumped into the flow path since the last sample, G U R = V(G o - G)/T LPR = V(L - Lo)/T where, F t = total flowrate = F m + Fg + F s + F b F m = flowrate o f m e d i u m being added to IC Fg = flowrate of glucose factor being added to IC F s = flowrate o f serum factor being added to EC F b = flowrate of base factor being added to IC G x = (FmGm) + (FgGg) + (FsGs) G m = concentration of glucose in medium G s = concentration of glucose in serum factor Gg = concentration of glucose in glucose factor E = e x p ( - FtT/V) T = time in hours since last sample V = system volume in ml = 1146 + 190 x 6 = 2286 (ml) G = current glucose concentration G O = previous glucose concentration L = lactate concentration of current sample L o = lactate concentration of previous sample P = protein concentration of current sample Po = protein concentration of previous sample O U R = (140 - D.O.eurrent) x flowrate x C = (140 - D.O.current) x 36 L/hr x 0.0013 m M / L / m m H g = (140 - D.O.current) x 0.0468 m M / h r / m m H g

139 where, flowrate = 0.1 L/minute/cartridge x 6 cartridge x 60 minutes/ hr = 36 L/hr 140 mmHg is saturation of medium (assumed) C is basically a solubility (saturation) constant and different for different medium because of the different salt concentration in medium formula Average C for media is about 0.0013 mM/L/mmHg.

References 1. Belfort G (1989) Membranes and bioreactors: a technical challenge in biotechnology. Biotechno. Bioeng. 33: 10471066. 2. Bowen-Pope DF, Vogel A and Ross R (1984) Production of platelet-derived growth factor-like molecules and reduced expression of platelet-derived growth factor receptors accompany transformation by a wide spectrum of agents. Proc. Natl. Acad. Sci. USA 81: 2396-2400. 3. Chen BC, Chen GC, Hsieh JH, Meng MH, Tsai TF, Liu JJ, Huang JH and Chang TH. Large-scale production of monoclonal antibodies using hollow fiber bioreactor. Chinese J. Micro. Immunol., in press. 4. Goosen MFA, O'Shea GM, Gharapetian HM, Chous S and Sun AM (1985) Optimization of microencapsulation parameters: Semipermeable microcapsules as a bioartificial pancreas. Biotechnol. Bioeng. 27: 146-150. 5. Kleinman HK, Klehe RJ and Martin GR (1981) Role of collagenous matrices in the adhesion and growth of cells. J. Cell Biol. 88: 473-485. 6. Knazek RA, Guillino PM, Kohler PO and Dedrick RL (1972) Cell culture on artificial capillaries: an approach to tissue growth in vitro. Science 178: 65-67.

7. Knight P (1989) Hollow fiber bioreactors for mammalian cell culture. Bio/Technology 7: 459--461. 8. Knowles BB, Howe CC and Aden DP (1984) Human hepatocellular carcinoma cell lines secrete the major plasma proteins and hepatitis B surface antigen. Science 209: 497--499. 9. Ku K, Kuo MJ, Delente J, Wildi BS and Feder J (1981) Development of a hollow-fiber system for large-scale culture of mammalian cells. Bioteehnol. Bioeng. 23: 79-95. 10. Mckeehan WL, Sakagami Y, Hoshi H and Mckeehan KA (1986) Two apparent human endothelial cell growth factors from human hepatoma cells are tumor-associated proteinase inhibitors. J. Biol. Chem. 261: 5378-5383. 11. Merten O-W (1987) Concentrating mammalian cells I. Large-scale animal cell culture. TIBTECH 5: 230--237. 12. Schreirer W, Nilsson K, Merten O-W, Katinger WD and Mosbach K (1982) Entrapment of animal cells for the production of biomolecules such as monoclonal antibodies. Dev. Biol. Stand. 55: 155-161. 13. Su T-S, Liu W-Y, Han S-H, Jansen M, Yang-Fen TL, Peng F-K and Chou C-K (1989) Transcripts of the insulin-like growth factors I and II in human hepatoma. Cancer Res. 49: 1773-1777. 14. Van Brunt J (1986) Immobilized mammalian cells: the gentle way to productivity. Bio/Technology 4: 505-509. 15. Van Wezel AL (1967) Growth of cell strains and primary cells on microcarriers in homogeneous culture. Nature 216: 64-65. 16. Young MW and Dean RC (1987) Optimization of mammalian-cell bioreactors. Bio/teehnology 5: 835-839.

Address for offprints: T.H. Chang, Development Center for Biotechnology, 81 Chang Hsing Street, Taipei, Taiwan, ROC