Harvesting recombinant microbial cells using crossflow filtration

Harvesting recombinant microbial cells using crossflow filtration F. James Bailey, R. Thomas Warf* and Robert Z. Maigetter M e r c k S h a r p & D o h m e R e s e a r c h Laboratories, * M e r c k P h a r m a c e u t i c a l M a n u f a c t u r i n g Division, W e s t Point, P e n n s y l v a n i a

A contained, crossflowfiltration (CFF) membrane system is described for harvesting Saccharomyces cerevisiae and Escherichia coil cells. This system is portable and can be cleaned and sanitized in place. Low- and high-cell density (LCD, HCD) fermentations o f recombinant cells in 10- to 200-1 volumes were used as the starting material. LCD fermentations, up to 8.3 g 1-: dry weight (dcw) o f S. cerevisiae, with volumes o f 10 to 200 l were harvested and diafiltered in 0.5 and 1.5 h, respectively. HCD 200-1 fermentations o f S. cerevisiae (47-63 g l 1 dcw) were harvested and diafiltered in appproximately 2 h. E. coli fermentations, LCD and HCD (up to 16.2 g 1-I dcw), o f 200-1 volumes were harvested and diafiltered in 2.3 h while employing 14 and 75 f t 2 o f membrane area, respectively. Using hollow f b e r orflat sheet membranes from different sources, cell harvesting times were less than 2.5 h. These studies demonstrate that CFF is an efficient method for harvesting and diafiltering recombinant S. cerevisiae and E. coli cells from fermentation broth.

Keywords:Cell harvesting; crossflowfiltration;recombinant;Saccharomyces; E. coli; fermentation

Introduction The separation of cells from fermentation broths is a critical step in the processing of biotechnologicallyderived products. Traditional separation methods include centrifugation and rotary vacuum filtration. L2 However, such techniques are not satisfactory when aseptic processing and aerosol containment 3 are required. The emergence of recombinant DNA technology and its evolution from the laboratory bench to the manufacturing facility has resulted in a shift away from traditional methods of cell separation. 4 Since the 1970s, crossflow microfiltration (CFF) has emerged as an important tool for microbial cell harvesting) The theory and application of CFF have been described. 6-8 Briefly, in CFF for cell harvesting, the cell suspension is constantly circulated across the membrane surface, thereby providing a sweeping action which helps keep the membrane pores from plugging. The filtrate passes through the membrane into the extracapillary space. The net result is a concurrent concentration of the cells and elimination of soluble media

Address reprint requests to Mr. Bailey at the Merck Sharp and Dohme Research Laboratories, West Point, Pennsylvania19486 Received 17 April 1989; revised 25 August 1989 © 1990 Butterworth-Heinemann

components. Whereas the filtration rate is dependent upon the retentate velocity, the driving force through the membrane is determined by pressure. The transmembrane pressure (ATMP) is defined as: A T M P : (Pi 2 P ° ) - P f

where Pi, Po, and Pfare the membrane inlet, outlet, and filtrate pressures, respectively. F i g u r e 1 is a schematic representation of a CFF process. Although pressure profiles will vary for each particular harvesting condition, the ranges reported here are Pi (10-33 psi), Po (0-5 psi), and Pr (0-1 psi). CFF has also proven useful for protein separation, 8 enzyme purification,9 and removal of cell debris. 10Advantages of CFF are efficiency, the ability to sanitize the harvesting equipment, diafiltration of soluble media components, and containment of aerosols. The need for greater containment of recombinant organisms and aseptic processing has focused our cell separation efforts toward CFF. Giorgio 11outlines physical containment requirements as issued by the National Institutes of Health as guidelines 12regulating large-scale recombinant experiments. Many of our fermentation broths provide the starting material that is used to prepare products for clinical testing; therefore, avoiding contamination is essential.

Enzyme Microb. Technol., 1990, vol. 12, September

647

Papers Pi

Po

Heat Exchanger

Feed =lk • o o o O•oO •o o 0 o 0 o 0 ~ 0oO• =lk Concentrate Inlet V O ° • o • o a O ~ W o • ~ • o • • O ,,r (orRetentate)

Oo-O-;-7-

O tlo

Membrane Filtrate

TampQ~netentate (Cens) PSI

or Permeate

PSi

Figure 1 Schematic representation of crossflow filtration process. Pi, Po, and Pf are the retentate inlet, outlet, and filtrate pressure, respectively. Used by permission13

vlltrate Flow Meter

LoadP~~~Membrane Narvest

This paper describes CFF cell harvesting for the recovery of cells from low- and high-cell density (LCD, HCD) fermentations of recombinant Saccharomyces cerevisiae and Escherichia coil grown in volumes of 10, 12.5, and 200 1. We present CFF harvesting runs that are representative of many that have been performed in our laboratory. Membranes and harvesting equipment from different manufacturers were tested and found satisfactory for harvesting cells. CFF is an efficient, contained method for harvesting microbial cells.

Materials and methods

Fermentation All fermentations were performed with recombinant S. cerevisiae and E. coli strains grown in complex media in 10- to 200-1 volumes. Hollow fiber and flat sheet membranes from different manufacturers were used ( Table 1 ). Cell densities were determined by measuring absorbance at 660 nm (A660)or by dry cell weight (dcw) expressed in g I.-]. Dry cell weight was determined by filtration of culture fluids followed by drying of the cells for 5 min in a microwave oven.

Diafiltration Removal of soluble components is desirable for ease of further downstream processing. This can be accomplished easily by diafiltering the cell suspension with buffer. Diafiltrations were performed using cold (4°C), 0.1 M phosphate buffered saline, pH 7.2, following an initial 4X concentration as described by Tutunjian. ]3 Final cellular concentrations, approximately 50% cell

Table 1

Membranes used for cell harvesting

Membrane manufacturer

Membrane type

P• re size (/~m)

Area (ft 2 )

Membrane type

Am•con Am•con Am•con A.G. Technology Mill•pore Romicon Romicon

H5MP01-43 H15MP01-43 H26MP01-43 CFP-2-E-55R Durapore PM500 PM500

0.1 0.1 0.1 0.2 0.45 0.05 0.05

5 15 26 23 10 5 25

Hollow fiber Hollow fiber Hollow fiber Hollow fiber Flat sheet Hollow fiber Hollow fiber

648

Figure 2 Process diagram for CFF harvesting of 200-1 fermentation broth

suspensions, were 10X or 20X those of the initial fermentation volume. The process volume is defined as the sum of the fermentation volume plus the volume of diafiltration buffer used.

Sanitization The CFF equipment was sanitized with 250 parts per million (ppm) sodium hypochlorite for 18 h at 22°C. After 18 h, sodium hypochlorite was drained and the CFF equipment was rinsed with three system volumes of sterile, distilled water. Hypochlorite determinations were made with a Chemetrics CL-1.6 Hypochlorite test kit, Calverton, VA. To minimize corrosion to stainless steel components, equipment manufacturers do not recommend concentrations greater than 250 ppm.

Harvesting system Figure 2 is a diagram of the CFF cell harvesting system that was used to harvest 200-1 HCD fermentations. The stainless steel process tank has 2-inch sanitary inlet and outlet connections. All piping in the retentate loop was 2-inch sanitary construction. The valves have a teflon diaphragm (ITT Engineered Valves, Lancaster, PA). Temperature and pressure gauges (Anderson Instrument Co., Fultonville, NY) were of sanitary construction. Venting points were equipped with sterile 0.22-/~m Sealkleen filters (Pall Corporation, East Hills, NY) to prevent aerosol release. Membrane reuse is a feature that makes CFF attractive. ]4 Membranes that are to be reused should be cleaned after use and stored according to the manufacturer's instructions.

Results

Harvesting low-cell density fermentations o f recombinant S. c e r e v i s i a e LCD fermentations of recombinant yeasts (up to 8.3 g 1-] dcw) grown in 10- and 200-1 volumes were performed in complex media and harvested by CFF (Table 2). Harvesting operations were performed at 22°C and ATMP was maintained at 9-12 psi. A 10-1 LCD fermentation was harvested in 0.5 h. Since this was a reason-

E n z y m e M i c r o b . T e c h n o l . , 1990, v o l . 12, S e p t e m b e r

Crossflow filtration: F. J. Bailey et al. Table 2

Harvesting low-cell density (LCD) fermentations of recombinant S. cerevisiae Fermentation harvest parameters

Filtration run

Filtration system

Membrane type

1

Amicon DC-10 Amicon DC-30 Amicon DC-30

Amicon 0.1 # m Amicon 0.1 /~m Amicon 0.1 /~m

2 3

Area (ft 2)

Process time (h)

Filtrate rate (I m 2 h 1)

15

0.5

65

N.D. b

--

--

200

250

1.5

120

15-20

0

0

200

250

1.3

80

20-21

0

0

Absorbance (A660)

DCW (g I 1)

Broth vol (I)

5

12.0

8.3

10

15

10.8

7.3

26

14.4

N.D. ~

Process vol (I)

Pressures (psi) Pi Po

Pf

Harvests were performed at 22°C a N.D., not determined b N.D., not determined, although typical values w o u l d be Pi = 10-15 psi, Po = 0 psi, Pf = 0 psi

able time for small-scale operation, the system was scaled up to harvest 200 1 of fermentation broth. Using either a 15 or 26 ft 2 Amicon 0.1-/~m hollow fiber cartridge, total processing time for concentrating 200 1 of fermentation broth and diafiltering with 50 I of buffer was less than 1.5 h.

Harvesting high-cell density fermentations of recombinant S. c e r e v i s i a e The desire for increased volumetric productivity led us to perform HCD fermentations (47-63 g 1-1 dcw) of recombinant S. cerevisiae. The data for cell harvesting of HCD fermentations are shown in Table 3. Initially, a 10-1 HCD fermentation was harvested in a manner similar to that for the 10-1 LCD fermentations. Total processing time for a 10-1 HCD fermentation was 0.5 h. Since the dissolved solids content of the HCD fermentation media was much greater than that of the LCD media, this dictated an increase in diafiltration

Table 3

buffer volume to adequately wash the cells. Thus, for the 200-1 HCD fermentations, the volume of diafiltration buffer was increased from 0.25X the initial fermentation volume for the batch fermentations to approximately 1.5X for the HCD fermentations. The first attempt at harvesting a 200-1 HCD fermentation having a dcw of 60.4 g 1-~ met with failure (Run 5, Table3). After 5 h of filtration, only 1601 of broth was filtered, and the process was terminated. Inadequate cooling capacity and the extended processing time resuited in a significant temperature increase (38°C) of the retentate stream. To help circumvent this problem, a heat exchanger was incorporated into the harvest system (Figure 2). In order to harvest a 200-1 HCD fermentation within a reasonable time (approximately 2 h), we recognized that increases in both membrane surface area and pump capacity were required. Therefore, subsequent HCD S. cerevisiae harvests employed the large-scale equipment described in Figure 2. Harvest runs 6 and 7 (Table 3) were performed in a

Harvesting high-cell density (HCD) fermentations of recombinant S. cerevisiae Fermentation harvest parameters

Filtration run

Filtration system

Membrane type

4

Amicon DC-10 Amicon DC-30 Amicon DC-90 Amicon DC-90 Millipore ProStak

Amicon 0.1 /~m A.G. Tech 0.2/~m Amicon 0.1 /~m A.G. Tech 0.2/~m Millipore 0.45/~m

5 6 7 8

Area (ft 2)

Absorbance (A660)

DCW Broth (g I 1) vol (I)

Process vol (I)

Process time (h)

Filtrate rate ([ m -2 h -1)

Pressures (psi) Pi Po Pf

0.5 b

108

N.D. d

--

__

5

88

51.7

10

25

23

92

60.4

200

160

5.0 a,b

15

18-23

2-5

0

45(3 x 15)

110

63.3

200

500

1.8 c

66

20-25

0

0

69(3 x 23)

96

53.1

200

480

2.1 c

36

10-20

0

0

30(3 x 10)

88

47.6

200

500

1.9 b

94

26-33

2-4

0-1

a Filtration terminated after 5 h b Harvested at 22°C c Harvested at 4°C d N.D., not determined although typical values are shown in (b) Table 2

Enzyme Microb. Technol., 1990, vol. 12, September

649

Papers 80

~'~ E

I

I

I

I

I

I

of broth. These cartridges were selected based upon preliminary test results where the Romicon 0.05-/~m membrane exhibited slightly faster filtration rates than did the Amicon 0.1-/~m membrane. E. coli harvesting runs 11 and 12 are shown in Table 4. Run 11 is an LCD fermentation that was harvested and concentrated 20-fold to approximately a 50% cell suspension in under 2.5 h using 14 f t 2 membrane area. Run 12 is an H C D fermentation that was harvested and concentrated 10-fold to approximately a 50% cell suspension in less than 2.5 h using 75 ft z membrane area. The filtration kinetics of this H C D fermentation (run 12) are plotted in Figure 4. Harvesting was performed at room temperature, since studies had shown that the polypeptide product produced by the cells was not labile at temperatures of 18°C-25°C. Before processing, the broth was cooled at 25°C. After setting the initial Pi, the ATMP was held constant at 9 psi throughout the filtration. The filtration rate profile is typical where the flux rate decreases and the cell concentration increases, v'15

I

60

40 L.=

u_ 20

Diafiltration

= [

0 14

b.-

~

8

g

6

rr

4 2 0 15

v

Discussion

10

5 0.0

I 0.2

I 0.4

I 0.6

[ 0.8

I 1.0

I 1.2

1.4

I 1.6

1.8

Time (HRS)

Figure

3

Kinetics of harvesting a high-cell density recombinant

S. cerevisiae f e r m e n t a t i o n

cold (4°C) room to minimize heat build-up so as to protect a potentially labile product. Tenfold concentrations of the initial fermentation volume were achieved following diafiltration. The filtration kinetics for the harvest of an H C D S. cerevisiae fermentation (run 6) are shown in Figure 3. The fermentation broth was cooled to 12°C prior to harvesting, and the diafiltration buffer was held at 4°C. During harvesting, the filtrate rate in 1m -2 h -I increased with increasing ATMP. Throughout the first 0.4 h of harvesting, the Pi was slowly increased from 0 to 25 psi and a constant 10 psi ATMP was then maintained. The filtrate rate slowly increased from 43 to 76 1 m -2 h -1 with the onset of diafiltration and a decrease of the retentate temperature.

Harvesting o f recombinant E. coli fermentations C F F was applied to harvesting fermentations (up to 16 g 1 ~dcw) of recombinant E. coli ( Table 4). All filtrations were performed at 22°C. Amicon and Romicon hollow fiber cartridges were used to harvest I0-1 and 12.5-1 fermentations of recombinant E. coli. Cartridges from both manufacturers proved acceptable and comparable; cells were concentrated 20-fold to approximately a 50% cell suspension in 0.5 h. This process was scaled up using Romicon PM500 cartridges to harvest 200 l 650

The scale-up for S. cerevisiae L C D fermentations from 10 1 to 200 I was accomplished by an increase in membrane area and retentate pumping capacity. The Amicon DC-30 filtration system with 1½-inch sanitary piping will a c c o m m o d a t e up to 26 ft 2 hollow-fiber cartridge area. Although the processing time with 26 ft 2 was slightly faster than that for 15 ft z, one would have expected a much faster processing time with greater membrane area. H o w e v e r , given the pumping capacity of the DC-30, 15 ft 2 membrane area was adequate to harvest the LCD yeast fermentation in run 2. Due to the pressure drop along the length of the 26 ft 2 hollow-fiber cartridge, this cartridge did not provide a significant reduction in processing time compared to the 15 ft z cartridge. The harvesting of a 10-1 H C D Saccharomyces fermentation (run 4) was accomplished in 0.5 h, even though the fermentation had a cell concentration o f 51.7 g I J dcw. During our first attempt at harvesting a 200-1 HCD S. cerevisiae fermentation (run 5), there was inadequate pumping capacity and insufficient membrane area. These limitations were o v e r c o m e by using the harvesting system shown in Figure 2. Several conclusions may be drawn from the results of experiments utilizing either 45 ft 2 (3 x 15 ft 2 Amicon 0.1-/~m cartridges) or 69 f t 2 (3 X 23 f t 2 A.G. Technology 0.2-/~m cartridges). The increasing filtration rate c o r r e s p o n d e d to the removal of media components through diafiltration. Processing at temperatures below the antifoam cloud point ~6 was manifested in reduced m e m b r a n e fouling. Hollow-fiber cartridges proved acceptable for harvesting and diafiltering 200-1 H C D fermentations with total process volumes of 500 I in approximately 2 h. While we had demonstrated the usefulness of hollowfiber cartridges, we also successfully demonstrated that the Millipore ProStak Pilot Plant could harvest a H C D

Enzyme Microb. Technol., 1990, vol. 12, September

Crossflow filtration: F. J. Bailey et al. Harvesting of low (LCD) and high-cell density (HCD) recombinant E. coil fermentations

Table 4

Filtration run

Filtration system

Membrane type

9

Amicon DC-10 Amicon DC-10 Amicon DC-30 Romicon Pilot 50

Amicon 0.1 /zm Romicon PM500 Romicon PM500 Romicon pm500

10 11 12

Area (ft 2)

Fermentation harvest parameters Absorbance DCW Broth Process (A660) (g I-v) vol (I) vol (I)

5

4.3

5.7

5

5.0

N.D. c

14

1.5

N.D. c

200

16.2

200

75(3x25)

60.8

Process time (h)

Filtrate rate (I m 2 h 1)

Pressures (psi) Pi Po Pf

10

15

0.5 ~

65

N.D. d

12.5

12.5

0.5 a

54

N.D. d

250

2.3 a

84

20

0

0

370

2.3 b

23

20

0

1

Harvests p e r f o r m e d at 22°C a LCD f e r m e n t a t i o n b HCD f e r m e n t a t i o n c N.D., not d e t e r m i n e d d N.D., not determined, a l t h o u g h typical values are s h o w n in (b) Table 2

recombinant S. cerevisiae fermentation (run 8) in under 2 h using 30 ft 2 membrane area. Due to the fiat sheet configuration of the Millipore membranes, the ATMP was 14-18 psi. Operating pressures for flat sheets can be as high as 60-80 psi for P~, while the P~ limit for the Amicon and A. G. Technology hollow-fiber cartridges

40

-

~

1

I

I

E 30 v

~ 2o ~_ lO

~1~

Diafiltrati

25 ,

~- 24 E ~

22

g 2o

\

rr 18

15

10 J

Q. 1-<3 0 0.0

I

I

I

I

0.5

1.0

1.5

2.0

2.5

Time (HRS) Figure 4 Kinetics of harvesting a high-cell density recombinant E. coil fermentation

is 25 psi. It is likely that the 0.45-/xm pore size of the Millipore membrane and the higher operating pressure attainable in the ProStak were the factors that contributed to a process time equivalent to that for the hollowfiber cartridges. The filtration rate in 1 m -2 h -1 is frequently the primary measure of cell harvesting performance. However, 1 m 2 h i values can be misleading, especially when excess membrane area is used. For example, filtration runs 6 and 7 ( Table 3) are comparable in terms of cell mass and time of harvesting. The filtration rate for run 7 is much less than that for run 6:36 versus 66 1 m 2 h--l. This difference results from excess membrane area employed in run 7. While there may have been an excess of membrane area for harvesting the recombinant S. cerevisiae fermentations, the total process time is critical. When harvesting potential labile biological materials, increased process time may reduce product yields. It is also desirable to get the fermentation and harvesting equipment cleaned and ready for storage within one working day. Amicon 0.1-#m and Romicon PM500 cartridges were used successfully to harvest 10-1 and 12.5-1 fermentations of recombinant E. coli in 30 rain. In addition, a 200-1 LCD recombinant E. coil fermentation was harvested in 2.3 h with a 14 ft 2 Romicon PM500 cartridge. To harvest a H C D 200-1 recombinant E. coil fermentation, 75 ft 2 (3 × 25 ft 2 Romicon PM500 cartridges) membrane area was used, which reduced the harvesting time to under 2.5 h. Similar to the conclusions of Cheryan,17 we found that a smaller m e m b r a n e pore is less prone to plugging by the bacteria, thus resulting in a greater flux rate. Previous experiences with harvesting recombinant E. coil fermentations indicated that these bacteria harvest more slowly than do yeast. The filtration kinetics for this H C D filtration are typical for a bacterial filtration where the filtrate rate declines as the cell concentration increases. In our laboratory, C F F has proved to be an accept-

E n z y m e M i c r o b . T e c h n o l . , 1990, v o l . 12, S e p t e m b e r

651

Papers 2 3

able and reproducbile method for harvesting microbial cell fermentations. As demonstrated, CFF was used to harvest LCD and HCD fermentations of recombinant S. cerevisiae and E. coli. For 200-1 fermentations, total processing times were under 2.5 h with process volumes of 500 1. Potential product degradation was minimized due to short processing times. Additional benefits are those of sanitary processing and aerosol containment, especially when harvesting recombinant organisms. The capability to scale up CFF for the harvesting of microbial cells most certainly will increase as products of biotechnology progress from the research laboratory to the production facility.

10

Acknowledgements

11 12

We gratefully acknowledge F. Kovach, our predecessor in CFF; C. E. Carty, C. A. Schulman, W. K. Herber, and R. W. Ellis for critical reading of the manuscript; and G. Albanesius for careful preparation of the manuscript.

13 14

4 5 6 7 8 9

15 16

References 1

652

Wang, D. 1. C. and Sinskey, A. J. in Adv. Appl. Microbiol. 1970, 12, 121

17

Enzyme Microb. Technol., 1990, vol. 12, September

Bowden, C. Chem. Engineer 1985, 50 Hanisch, W. H., Fuhrman, S., Harano, D., and Pemberton, M. in Proceedings Aust. Biotechnol. Con[: 1982, p. 153 Tutunjian, R. S. Bio/Technology 1985, 3, 615 Hanisch, W. in Membrane Separation in Biotechnology McGregor, W. C., ed.) Marcel Dekker, New York, 1986, p. 61 Henry, J. D. in Recent Developments in Separation Science (Li, N. N., ed.) CRC Press, 1972, Vol. 2, p. 205 Henry, J. D. and Allred, R. C. Devel. Ind. Microbiol. 1972, 13, 117 Porter, M. C. in Handbook o f Separation Techniques McGrawHill, 1979, 2-3 Le, M. S., Spark, L. B. and Ward, P. S. J. Membr. Sci. 1984, 21, 219 Quirk, A. V. and Woodrow, J. R. Biotechnol. Left. 1983, 5, 277 Giorgio, R. J. and Wu, J. J. Trends Biotechnol. 1986, 4, 60-65 Guidelines for Research Involving Recombinant DNA Molecules, National Institutes of Health (Federal Register, Vol. 49, No. 227t. Friday, November 23, 1984, 46266-46291 Tutunjian, R. S. Devel. Ind. Microbiol. 1984, 25, 415 Lasky, M. and Grant, D. Am. Biotechnol. Lab. 1985, Nov./ Dec., 16-21 Le, M. S. and Atkinson, T. Process Biochem. 1985, 20, 26 Kroner, K. H., Schutte, H., Husted, H. and Kula, M. R. in 3rd European Congress on Biotechnology. Verlag Chemie, 1984, I11-549 Cheryan, M. in Uhrafiltration Handbook Technomic Publishing, Lancaster, PA, 1986, p. 279