This project aimed to investigate aspects pertinent to the development o f a scaleable process for the production o f recombinant antibody fragments from microorganisms. A detailed search o f recent literature resulted in no reports to the present date o f complete processes for recombinant antibody fragment production processes and no universal expression and fermentation strategies were obvious. The advances made by Plucktbun and collègues (Skerra and Plucktbun, 1988; Plucktbun and Skerra, 1989; Plucktbun, 1991; Plucktbun and Pfitzinger, 1991; Skerra et aL, 1991) and other research groups (Better, et aL, 1988) by engineering expression o f a number o f functional periplasmic recombinant antibody fragments in Escherichia coli are perhaps the most significant molecular biological contributions to process development.
From an engineering viewpoint the entire process from host organism, vector construction and its expression characteristics, to the final product (in a suitable form for its intended application) must be considered for complete and optimal process design. The starting point for investigations in this study were strains o f Escherichia coli which could express Fv and scFv fragments o f the D1.3 monoclonal antibody and secrete them into the periplasmic space where accumulation o f functional antibody fragments occurred. Subsequent leakage o f these proteins to the extracellular culture medium was however observed during fermentation, the extent o f which was initially unpredictable. Investigations reported in Chapter 3 were concerned with a fed-batch process for production o f scFv fragments. In a 14 L fermenter, controlled exponential feeding to a low specific growth rate resulted in a process which routinely achieved E. coli biomass to cell densities o f 50 g/L dry cell weight in a reproducible manner. Induction o f this fermentation was performed by addition o f IPTG to activate expression o f scFv which was under the control o f the tac promoter. Feeding o f yeast extract solution at a uniform flow rate throughout induction was initiated simultaneously with IPTG addition. Increasing concentration o f the yeast extract feed affected both titre and location o f scFv fragments after 12 h o f induction, resulting in a 40-fold increase in scFv activity to titres o f 200 mg/L (40 mg scFv/g dry biomass) with almost 80% o f this located within the periplasm.
Monitoring fermentations to enable informed process decisions to be made is fundamental in both research and commercial fermentations. Data concerned with dissolved oxygen, pH and temperature which are measured by probes mounted in situ in the fermenter liquid, and respiration where mass spectrometry is utilised to monitor fermenter exit gas, are measured in real-time (Fig 8.1.1). Adjustments to the process may be made in response to these measurements where required, for example increase in stirrer speed and airflow when dissolved oxygen is reduced. Product monitoring in
real-time may be more problematic, especially in the case o f antibodies where activity is traditionally measured by ELISA after completion o f fermentation. Recent advances in biosensor technology to measure macromolecular interactions offer a promising alternative. An optical biosensor adapted for measurement o f D 1.3 antibody fragment binding to the target antigen was investigated as an alternative to ELISA for measurement o f Fv activity as a batch fermentation proceeded (Chapter 4). Although, exact concentrations o f active Fv could not be confidently determined, activity profiles generated by the biosensor over the course o f fermentation with respect to relative location o f the antibody fragment were reflected in ELISA analysis. The ability to determine exact concentrations o f antibody fragments in real process time is eagerly awaited.
Decreases in extracellular concentrations o f antibody fragments were observed in both batch and fed-batch protocols upon prolonged fermentation. Three possibilities were considered for this degradation. Firstly the presence o f proteases which could degrade both Fv and scFv fragments, secondly instability at fermentation temperature and thirdly exposure o f antibody proteins to a high shear environment combined with the presence o f air-liquid interfaces. The latter was investigated in bench scale reactors at controlled temperatures with affinity purified scFv spiked into PB SA buffer and fermentation medium (Chapter 5). ScFv fragments having increased thermal stability compared with Fv fragments (Glockshuber, et aL, 1990) were selected for this investigation. A severe decrease in activity which approximated a first order decay model was observed for scFv spiked into PB SA buffer, however fermentation broth filtrate from a non-antibody fragment producing E. coli strain had a protective effect and decay was not observed. The presence o f the antifoam agent PPG, and also the yeast extract content o f the fermentation broth were considered to be contributary factors to scFv protection. The reasons for degradation o f Fv and scFv still therefore remain unclear and investigations into the action o f proteases on these antibody fragments is required to determine their effects.
Having investigated fermentation and the ability to control the final location o f antibody fragments with respect to the cell it is necessary to consider product recovery. Preliminary investigations into the options for downstream processing were unable to elucidate a complete and reproducible recovery scheme (Chapter 6). To take advantage
o f the increased periplasmic concentrations o f active antibody fragments in a relatively pure form specific release o f antibody fragments from this location must be achieved. O f the specific methods reported in the literature none were able to effect complete release without affecting antibody fragment activity. Lysozyme treatment would have been considered a suitable candidate except for its interaction with these antibody fragments as the antigen. Interaction o f scFv with the bacteriophage lysozyme T4 was measured using the optical biosensor and no interaction was observed (Fig 8.1.2). This
Figure 8.1.1 - Time scale for analysis o f fermentation parameters and samples
Data concerning E. coli fermentations performed in these studies was collected via a number o f routes. Samples removed from the fermenter were required to undergo a number o f treatments prior to assay for the product, either by biosensor analysis or ELISA. Information concerning the product, unlike other fermenter data, was therefore not collected in real-time.
R eal-Tim e Fermentation Exit G a s a n a ly sis - M a ss S p e c Fermenter P robes D issolvsd O2, pH,
Temperature OUR, CER, RQ
Fermenter Sam p le Within 10 min o f sampling O D 6 6 0 n m 4- H om ogen ise ♦ OD 660n
A sep tic Serial
Dilution Centrifuge Centrifuge
Within 1 h H om ogenate Supem atant Cell Pellet P late onto Agar Supem atant sampling O sm otic S h o ck Protein A s s a y ▼ B io s e n s o r B io s e n s o r A n a iy s is A n a ly sis B io s e n s o A n a ly s is
P lasm id Stability Dry Cell ELISA W eight