Protein Expression and Purification 26 (2002) 14–18 www.academicpress.com

An economical 20 litre bench-top fermenter Michael A. Thiel,a,1 Douglas J. Coster,a Christos Mavrangelos,b Heddy Zola,b and Keryn A. Williamsa,* a

Department of Ophthalmology, Flinders University of South Australia, Adelaide, Australia b Child Health Research Institute, Adelaide, Australia Received 4 December 2001, and in revised form 26 May 2002

Abstract We describe an economical 20 litre bench-top fermenter suitable for production of recombinant antibody fragments in bacterial expression systems. The bacterial culture contained within a polycarbonate carboy is mixed (400–600 rpm) and aerated (1 vessel vol./min) by a high-shear radial flow impeller mounted on a hollow stainless steel shaft, through which pressurised air is pumped. Air is dispersed as fine bubbles into the culture medium by the turbine impeller, without the need for a porous sparger. A stainless steel baffle stabilised by a gliding counterweight increases mixing. The components can easily be disassembled for cleaning and sterilisation. Temperature (range 20–37 °C) and pH (range 7.0–7.5) are controlled manually. Using the apparatus, it proved possible to achieve Escherichia coli cell culture densities equivalent to an optical density at 600 nm (OD600 ) of 30–32, compared with OD600 4–6 in shake flasks. A yield of 40 mg/litre/day of a recombinant antibody fragment was obtained with the fermenter, which was 15-fold more than the yield of 2.5 mg/litre/day achieved in shake flasks. The fermenter may be particularly suited for research purposes. Ó 2002 Elsevier Science (USA). All rights reserved.

The amount of a recombinant protein needed for animal experiments usually exceeds the levels that can be achieved in shake-flask cultures. Medium-to-large scale production requires the use of sophisticated bacterial fermenter systems. However, commercially available fermenters are expensive and may exceed the budget available to an academic institution. Here, we describe a low-cost 20 litre bench-top fermenter suitable for medium-scale expression of recombinant proteins such as antibody fragments in Escherichia coli. The major factors that limit E. coli culture densities are a lack of oxygen and the accumulation of metabolic byproducts [1,2]. Thus, important considerations for any bacterial fermenter are the aeration and mixing systems. For good aeration, air must be dispersed into very fine bubbles to maximise the available surface area. Effective mixing is required to maintain an even distribution of bacterial cells throughout the culture, which becomes more viscous at higher densities. The *

Corresponding author. Fax: +618-8277-0899. E-mail address: keryn.williams@lfinders.edu.au (K.A. Williams). 1 Present address: Department of Ophthalmology, University of Zurich, Zurich, Switzerland.

fermenter we describe proved satisfactory for production of several hundred milligrams of a recombinant antibody fragment.

Materials and methods Recombinant antibody construct A single chain fragment variable (scFv) antibody fragment with specificity for rat CD4 was engineered from a log-phase culture of the OX38 hybridoma cell line (European Collection of Animal Cell Cultures, Porton Downs, Wiltshire, UK), using the methods described by Krebber et al. [3]. The scFv gene was cloned into the pAK100 vector using the SfiI restriction enzyme (New England Biolabs, Beverly, MA, USA) and transformed into E. coli strain HB2151 for fragment selection by flow cytometry (vide infra). The scFv genes from clones expressing functional scFv were cloned into a modified pAK400 vector (pHB400) containing both an skp insert (encoding a periplasmic foldase) from the pHB100 vector and a Shine–Dalgarno sequence for high-yield scFv expression in E. coli [4,5]. A 6-histidine

1046-5928/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 1 0 4 6 - 5 9 2 8 ( 0 2 ) 0 0 5 2 7 - 2

M.A. Thiel et al. / Protein Expression and Purification 26 (2002) 14–18

15

tag engineered into the expression vectors permitted detection of fragments and subsequent purification over an immobilised metal–ion affinity column [6]. Vectors were the gift of Professor A. Pl€ uckthun (University of Zurich, Zurich, Switzerland). Bacterial shake-flask cultures E. coli HB2151 were grown in Terrific Broth (TB, containing 12 g/litre bacto- tryptone, 24 g/litre bactoyeast extract, 4 ml/litre glycerol, 17 mM KH2 PO4 , and 36 mM K2 HPO4 ) [7]. Bacto-tryptone and bacto-yeast extract were purchased from Oxoid (Basingstoke, Hampshire, UK). All media were supplemented with 25 lg/ml chloramphenicol (Sigma Chemical, St Louis, MO, USA), just prior to inoculation with E. coli. Cultures of 500 ml were grown in baffled 3 litre shake flasks, to allow enough headspace for aeration. Cultures were incubated on an orbital shaking platform (200–300 rpm) at 37 °C. Description of bench-top 20 litre fermenter The fermenter, which was constructed in the Department of Biomedical Engineering, Flinders University, Adelaide, Australia, is shown in Fig. 1. A 23 litre transparent polycarbonate carboy (Nalgene, Nalge Nunc International, Rochester, NY, USA) was used as the fermentation vessel. The bacterial culture was mixed and aerated using a centrally placed mixer (range 23–2300 rpm, catalogue numbers P-50002-35 (drive motor unit) and P-50002-07 (control unit), Cole– Parmer Instrument Company, Vernon Hills, IL, USA) together with a stainless steel, high-shear radial flow impeller (overall diameter 51 mm, catalogue number P04560-30, Cole–Parmer) mounted on a hollow stainless steel shaft (outer diameter 7.9 mm, internal diameter 7.4 mm, and length 750 mm) purchased from a local plumbing supplier. The shaft was connected to the mixer via the precision collet integral to the drive motor. Air was pumped through two parallel copper coils (each coil comprising 7 m tubing, internal diameter 5.5 mm) at a rate of 10 l/min (1 vessel vol./min). The coils were immersed in a water bath to allow temperature equilibration of the air. Air was then pumped via silicon tubing fitted with a 0.22 lm filter (Minisart, Sartorius, G€ ottingen, Germany) to a tube connected by a tightly fitting bearing to the hollow, rotating mixer shaft (Fig. 2A). The fermentation vessel was fitted with a 2.5 cm diameter side-port unit (Fig. 2B) containing a second air inlet, to improve aeration and mixing of the culture periphery. The whole side-port unit was removable to allow access for the addition and removal of culture medium. Air was dispersed into the culture medium as fine bubbles by the turbine impeller, without the need

Fig. 1. Design of the bench-top 20 litre fermenter. A 23 litre polycarbonate carboy; B mixer with through the engine shaft design; C hollow shaft for air flow; D air inlets; E turbine impeller; F stainless steel tube to remove samples; G removable thermometer probe inside a hollow stainless steel shaft; H removable stainless steel baffle with stabilisation weight and spring-fixation to the bottle neck; I copper coils for air– temperature equilibration; J flowmeter and pressure valve for air-flow control; K 0.22 lm air filter; L water bath heater; M water bath; and N stand for mixer engine. The culture is mixed by the central turbine impeller mounted on a hollow stainless steel shaft. The stainless steel baffle inside the bottle increases mixing. The baffle is stabilised by a spring inside the bottle neck and a gliding counterweight on the bottom of the flask. Air is pumped through the hollow mixer shaft and dispersed by the impeller. Additional air is supplied by a side-port air line. The side- port unit also contains a sampling port and a temperature probe.

for a porous sparger. Antifoam #289 (Sigma Chemical, St Louis, MO, USA) was added at 0.1 ml/litre culture volume, to prevent foaming of the medium. To further increase mixing of the culture, a removable baffle (Fig. 2C) was placed inside the bottle. The baffle was stabilised by a spring inside the bottle neck (Figs. 2C and D) and a gliding counterweight (Fig. 2C) on the bottom of the flask. Baffle, counterweight, and spring were constructed in-house from stainless steel. The side-port unit (Fig. 2B) also contained a sampling port and a metal tube, which acted as a conductor for the temperature probe (catalogue number 90052-00, Cole–Parmer). For temperature control, the fermentation vessel was placed in a water bath. With the exception of the mixer motor, the whole fermenter was designed to be autoclavable. A photograph of the rig in operation is shown in Fig. 3.

16

M.A. Thiel et al. / Protein Expression and Purification 26 (2002) 14–18

Fig. 2. Photographs of component parts of the fermenter. (A) The mixer drive unit showing the bearing between the air inlet tubing and the hollow, rotating mixer shaft; (B) the removable 2.5 cm diameter side-port unit, showing the second air inlet, the sampling port, and the conductor for the temperature probe; (C) the counterweighted baffle used to increase aeration of the mixed culture; (D) details of the spring holding the baffle inside the neck of the fermentation vessel.

Periplasmic extraction of recombinant protein Bacterial cells were harvested (10,000g, 20 min, 4 °C) and either frozen immediately to )70 °C or kept at 4 °C for scFv extraction within 12 h. ScFv fragments were extracted from the periplasm of the pelleted bacteria with a periplasmic extraction buffer containing 50 mM Na2 B4 O7 , 150 mM NaCl, and 1 mM ethylene diamine tetraacetic acid (EDTA) at pH 8.0, essentially as described by Pack et al. [8]. Two ml extraction buffer was added per gram wet cell pellet and the suspension was shaken vigorously for 60 min on ice before centrifugation at 10,000g at 4 °C for 30 min. The supernatant containing the periplasmic extract was incubated at 4 °C for 1 h with 10 mg/litre DNAse I (Boehringer–Mannheim, Indianapolis, IN, USA) and then passed through a 0.22 lm filter to remove cell detritus. To remove EDTA, the periplasmic extract underwent a buffer exchange against sodium borate buffer (50 mM Na2 B4 O7 , 1 M NaCl, pH 8.0) in a Amicon DL10 ultrafiltration system (Amicon Corporation, Danvers, MA, USA) using a 10 kDa cut-off cartridge. During this step, the

Fig. 3. Photograph of the rig during a fermentation run. The fermentation vessel sits in a water bath for temperature control.

periplasmic extract was also concentrated 5-fold. The extract was supplemented with imidazole to a final concentration of 20 mM to prevent microbial growth and stored at 4 °C. Flow cytometry Binding activity of OX38 scFv protein was measured by high-sensitivity immunofluorescence detected by flow cytometry, as previously described [9], using normal rat thymocytes as the CD4-positive target cells. In brief, 50 ll cell suspension at 2  107 cells/ml in Dulbecco’s A phosphate-buffered saline (PBS) containing 20 mM sodium azide (PBS-azide) was incubated with 50 ll periplasmic extract at 4 °C for 30 min. Cells were washed with PBS-azide and incubated in 50 ll of 1/250 dilution of anti-polyHIS antibody (Sigma Chemical, St Louis, MO, USA) in PBS-azide for 30 min at 4 °C. Cells were washed as before and resuspended in 50 ll of 1/100 dilution of biotinylated goat anti-mouse antibody (DAKO, Carpinteria, CA, USA) in PBS-azide for 30 min at 4 °C. After washing, cells were incubated with

M.A. Thiel et al. / Protein Expression and Purification 26 (2002) 14–18

17

50 ll of 1/50 dilution of streptavidin–phycoerythrin (Sigma Chemical, St Louis, MO, USA) in PBS-azide for 30 min at 4 °C. Finally, cells were fixed in 50 ll of 5% v/v formaldehyde and 10 mM glucose in PBS-azide, and analysed within 24 h in a FACScan flow cytometer (Becton–Dickinson, Mountain View, CA, USA).

Results and discussion Production of a recombinant protein in the 20 litre fermenter Shake-flask cultures of E. coli OX38 scFv/pHB400 were grown to an OD600 of approximately 3, at which point they were used to inoculate the 20 litre fermenter. Ten litres of TB culture medium supplemented with 0.5% glucose, 5 mM MgSO4 , and 25 lg/ml chloramphenicol was inoculated with 1 litre E. coli OX38scFv/ pHB400 shake- flask culture and allowed to grow at 20– 25 °C for 12 h (OD600 approximately 10–12) before another 10 litre of fresh TB medium was added. Once an OD600 of approximately 10 had been reached, the culture was induced with 0.25 mM isopropyl-b-thiogalactopyranoside (IPTG; Promega, Madison, WN, USA) and the temperature was maintained at 25 °C thereafter by adding ice to the water bath. Cultures grew to an OD600 of 22–24 over a total fermentation period of approximately 18–20 h (12 separate fermentations), at which point growth stopped. A representative growth curve is shown in Fig. 4. During fermentation, the pH of the culture was adjusted within the range 7.0–7.5 by addition of 5 M NaOH or HCl and temperature was controlled by addition of ice to the water bath. During the initial bacterial growth phase, the pH tended to drop below 7.0 followed by a sudden reversal, with a tendency to rising pH values thereafter until the end of the fermentation. The pH of a culture may increase when E. coli is forced to switch to consumption of endogenous protein as the energy source [10,11]. To investigate whether the observed rise in pH was caused by carbon depletion, a culture was established in which 1 ml glycerol/litre culture medium was added every hour once an OD600 of 12 had been reached [12]. The addition of glycerol completely abrogated the need for addition of HCl to maintain a neutral pH and resulted in prolonged growth of the culture with final OD600 values of 30–32 over 10 separate fermentation runs (Fig. 5), without the need for any pH adjustment during fermentation. Typical 20 litre fermentation processes resulted in 850–1000 g wet cell pellet (40–50 g wet cells/litre). Crude periplasmic extraction of optimised cultures revealed yields of soluble and correctly folded OX38 scFv fragments of approximately 40 mg/litre fermentation culture, as measured by antigen binding activity using flow

Fig. 4. Representative example of a growth curve for E. coli OX38 scFv/pHB400 using the 20 litre fermenter. The initial fermentation volume was 11 litre. A further 10 litre of culture medium was added after 12 h at an OD600 of 10 and the culture was induced with 0.25 mM IPTG after 14 h. The OD600 reached approximately 23 before growth stopped.

cytometry. In contrast, the maximal OD600 achievable in the shake-flask culture was 4–6 and maximal yields were approximately 2.5 mg/litre. Successful 20 litre fermentations were carried out with two other scFv fragments with completely different binding specificities (data not shown), indicating the general applicability of the equipment for this class of recombinant proteins. The large body of published literature on the engineering of recombinant antibody fragments contrasts with the small amount of accessible information on how such fragments can be produced in the research laboratory in yields of several hundred milligrams [6]. Production of recombinant proteins in this medium-scale range usually involves the use of commercial fermenters, the cost of which can easily exceed the financial resources of academic institutions. Our 20 litre bench-top fermenter cost approximately US $2000 and was constructed within a university workshop. Despite the requirement for manual control of parameters such as temperature and pH, it was possible to produce 40 mg active scFv per litre of fermentation culture per day. This was a 15-fold increase as compared with expression levels of the same fragment in shake-flask cultures. It compares reasonably well with the levels of various antibody fragments obtained in commercial high-density fermentation systems

18

M.A. Thiel et al. / Protein Expression and Purification 26 (2002) 14–18

References

Fig. 5. Effect of additional glycerol supplements. Representative example of a growth curve for E. coli OX38 scFv/pHB400 in the 20 litre fermenter. Glycerol was added (1 ml/litre culture fluid/hour) from hours 14 to 19 of fermentation and abrogated the need for manual adjustment of pH with HCl. An OD600 of 35 was achieved before fermentation was terminated.

for non- mammalian cells, in which yields ranging from 1 mg/litre to 3 g/litre have been reported [13–16].

Conclusion We describe an inexpensive 20 litre bench-top fermenter, in which it was possible to generate 800 mg of soluble, active recombinant scFv antibody fragment in a day.

Acknowledgments This work was supported by the National Health and Medical Research Council of Australia, the Ophthalmic Research Institute of Australia, the Flinders Medical Centre Foundation, the Swiss National Science Foundation, EMDO Foundation, and the Swiss Foundation for the Prevention of Blindness. We gratefully acknowledge the skills of Mr. L. Bahr, Department of Biomedical Engineering, Flinders University, who constructed the equipment described.

[1] G. Georgiou, P. Valax, Expression of correctly folded proteins in Escherichia coli, Curr. Opin. Biotechnol. 7 (1996) 190–197. [2] E. Bech Jensen, S. Carlson, Production of recombinant human growth hormone in Escherichia coli: expression of different precursors and physiological effects of glucose, acetate and salts, Biotech. Bioeng. 36 (1990) 1–11. [3] A. Krebber, S. Bornhauser, J. Burmester, A. Honegger, J. Willuda, H.R. Bosshard, A. Pl€ uckthun, Reliable cloning of functional antibody variable domains from hybridomas and spleen cell repertoires employing a reengineered phage display system, J. Immunol Methods 201 (1997) 35–55. [4] H. Bothmann, A. Pl€ uckthun, Selection for a periplasmic factor improving phage display and functional periplasmic expression, Nat. Biotechnol. 16 (1998) 376–380. [5] C. Mavrangelos, M. Thiel, P.J. Adamson, D.J. Millard, S. Nobbs, H. Zola, I.C. Nicholson, Increased yields and activity of soluble singlechain antibody fragments by combining high-level expression and the Skp periplasmic chaperonin, Protein Expr. Purif. 23 (2001) 289–295. [6] A. Pl€ uckthun, A. Krebber, C. Krebber, U. Horn, U. Kn€ upfer, R. Wenderoth, L. Nieba, K. Proba, D. Riesenberg, Producing antibodies in Escherichia coli: from PCR to fermentation, in: J. McCafferty, H.R. Hoogenboom, D.J. Chiswell (Eds.), Antibody Engineering: A Practical Approach, Oxford University Press, Oxford, 1996, pp. 203–252. [7] J. Sambrook, E. Fritsch, T. Maniatis, Bacterial media, antibiotics and bacterial strains, Appendix A, in: Molecular Cloning: A Laboratory Manual, second ed., Cold Spring Harbor Laboratory Press, Plainview, NY, 1989, pp. 1–13. [8] P. Pack, M. Kujau, V. Schroeckh, U. Knupfer, R. Wenderoth, D. Riesenberg, A. Pl€ uckthun, Improved bivalent miniantibodies, with identical avidity as whole antibodies, produced by high cell density fermentation of Escherichia coli, Biotechnol. 11 (1993) 1271–1277. [9] H. Zola, S.H. Neoh, B.X. Mantzioris, J. Webster, M.S. Loughnan, Detection by immunofluorescence of surface molecules present in low copy numbers. High sensitivity staining and calibration of flow cytometer, J. Immunol. Methods 135 (1990) 247–255. [10] S.Y. Lee, High cell-density culture of Escherichia coli, Trends Biotechnol. 14 (1996) 98–105. [11] R. Verma, E. Boleti, A.J. George, Antibody engineering: comparison of bacterial, yeast, insect and mammalian expression systems, J. Immunol. Methods 216 (1998) 165–181. [12] D.J. Korz, U. Rinas, K. Hellmuth, E.A. Sanders, W.D. Deckwer, Simple fed-batch technique for high cell density cultivation of Escherichia coli, J. Biotechnol. 39 (1995) 59–65. [13] J.F. Rippmann, M. Klein, C. Hoischen, B. Brocks, W.J. Rettig, J. Gumpert, K. Pfizenmaier, R. Mattes, D. Moosmayer, Procaryotic expression of single-chain variable-fragment (scFv) antibodies: secretion in L-form cells of Proteus mirabilis leads to active product and overcomes the limitations of periplasmic expression in Escherichia coli, Appl. Environ. Microbiol. 64 (1998) 4862–4869. [14] T. Kretzschmar, L. Aoustin, O. Zingel, M. Marangi, B. Vonach, H. Towbin, M. Geiser, High-level expression in insect cells and purification of secreted monomeric single-chain Fv antibodies, J. Immunol. Methods 195 (1996) 93–101. [15] A. Sotiriadis, T. Keshavarz, E. Keshavarz-Moore, Factors affecting the production of a single-chain antibody fragment by Aspergillus awamori in a stirred tank reactor, Biotechnol. Prog. 17 (2001) 618–623. [16] U. Horn, W. Strittmatter, A. Krebber, U. Kn€ upfer, M. Kujau, R. Wenderoth, I. M€ uller, S. Matzku, A. Pl€ uckthun, D. Riesenberg, High volumetric yields of functional dimeric miniantibodies in Escherichia coli, using an optimized expression vector and highcell-density fermentation under non-limited growth conditions, Appl. Microbiol. Biotechnol. 46 (1996) 524–532.