Products such as therapeutic proteins synthesised as recombinant proteins in microorganisms require that fermentation is optimised and can be easily integrated with subsequent product recovery steps if production costs are to be reduced. Investigations into a fed-batch fermentation process for the expression o f a recombinant single chain antibody fragment resulted in process improvement with respect to both time and titre. The original protocol comprised growth o f starter cultures in a complex medium. Transfer o f this inoculum to the fermenter containing a defined medium adapted for fed- batch fermentation resulted in a long lag phase. Medium downshifts by replacement o f one carbon source with another have previously been shown to have effects on protein synthesis (Johnsen, et al., 1977) in that polypeptide chain elongation is decreased. It was therefore not surprising that lag phase in the present fermentation could easily be minimised by a change o f inoculum medium so that starter cultures were prepared in identical medium to that in the fermenter.
Secretion o f scFv to the periplasmic space allows for production o f properly folded and active proteins at this cellular location, however subsequent leakage o f scFv fragments to the extracellular environment can occur. Control o f leakage to produce a single location for scFv fragments will determine the downstream process options for recovery.
The tac promoter is frequently used to control expression o f antibody fragments and other recombinant proteins from E. coli. Although induction normally takes place by addition o f IPTG to an optimal concentration in the culture medium, production o f scFv fragments was also observed prior to induction in this study.
It has been observed that antibody fragments expressed under the control o f the tac
promoter are released to the extracellular medium after several hours o f induction, even though signal sequences have directed the recombinant protein to the periplasm (Shibui and Nagahari, 1992; Takkinen, et al., 1991). Induction using IPTG also caused inhibition o f growth in an antibody fragment fermentations and this was shown to be due to the antibody product itself (Pack, et al., 1993). In this report, addition o f IPTG did not affect the periplasmic location, however, extracellular accumulation o f antibody fragments has occured with other promoter/inducer systems (Better and Horwitz, 1993). Release o f scFv to culture medium was prevented by use o f a complex medium feed at high concentration during the induction phase. In the present investigation the use o f high concentrations o f yeast extract feed resulted in concentrations o f approximately 200 mg/L fermentation broth o f active scFv in the periplasmic space o f E. coli cells. In comparison, concentrations o f scFv were 40-fold lower and largely extracellular when yeast extract feed concentration was 1 0-fold lower.
The roles o f yeast extract and other complex medium components have been investigated for their involvement in other E. coli fermentations for protein production. Investigations into glyceraldehyde 3-phosphate dehydrogenase (GAPDH) production indicated that inclusion o f both peptone and yeast extract in the medium allowed optimal enzyme production (Nancib, et al., 1991). Their results suggested that peptone stabilised the enzyme and yeast extract was found to assist high growth and effective acetate utilisation. Studies on the expression o f recombinant human insulin-like growth factor (IGF-1) from E. coli have found that supply o f an organic nitrogen source is essential to production, and no production occurs with inorganic nitrogen (Tsai, et al.,
1987). The same study also investigated the use o f dual feeds containing glucose, and organic nitrogen in the form o f yeast extract and bactotryptone. The highest concentrations o f IGF-1 were produced at an organic nitrogen feed rate o f 50 g/h. It was speculated that the yeast extract reduced proteolytic degradation o f the product as proteases would favour smaller peptides supplied in the feed. Unlike IGF-1, scFv fragments produced here are located in the periplasmic space, and are partitioned from
E. coli cytoplasmic proteases. However, scFv are subject to proteolysis from periplasmic proteases. Also in the present investigation thermoinduction is not utilised so proteases resulting from heat shock response will not be present.
For production o f recombinant |3-galactosidase in E. coli fermentation, supplementation with yeast extract resulted in increased biomass as well as increased productivity o f the enzyme (Li, et al., 1990). Two mechanisms for the involvement o f yeast extract were proposed in this study, namely provision o f components for polypeptide elongation and effects on regulation o f transcription and translation. In some studies yeast extract and other complex components have been replaced by individual amino acids and vitamins. This approach did not increase yields as much as yeast extract supplementation (Li, et al., 1990) and would be more costly for use on a process scale.
Stoichiometric mass balancing over the fed-batch fermentation (see Appendix 3) reveals that only 1.8 g o f the 86.4 g yeast extract supplied over 12 hours o f induction in Run 6
are required for product formation (assuming 100% conversion). It appears from our study that yeast extract has a role in prevention o f release o f periplasmic proteins from the cell. It cannot reverse the effects o f cell lysis in cultures where cell viability is reduced prior to yeast extract feeding such as Run 10 performed in the 42 L fermenter. It is possible that yeast extract strengthens cell walls o f E. coli. Gray, et al., (1972) observed that growth o f E. coli in complex medium resulted in cell walls which provided a greater resistance to disruption by high pressure homogenisation than when fermentation was performed in chemically defined medium.
The periplasm provides an environment where scFv is present as a concentrated solute, in a specific cell compartment, separated from the majority o f cellular proteases. Specific release o f scFv by techniques such as osmotic shock which will not rupture
inner cell membranes, will result in a feed stream suitable for purification. Although driving the fermentation towards extracelluar scFv production may appear a more attractive option for downstream processing as it avoids operations for cell disruption, our study has shown that titres are then low. Purification o f a large volume o f a dilute solution o f scFv may therefore not be straightforward. This study illustrates that both the concentration o f antibody fragment, and its location, can be influenced in ways which enhance the prospect o f effective large scale processing.