Saccharomyces cerevisiae has long been grown in fed-batch cultivation by the brewing and baking industry, where it was selected principally for its vigorous growth. It has also been used in the biopharmaceutical industry for the production of protein therapeutics, such as recombinant insulin.
In this tutorial we describe a simple high-intensity fed-batch fermentation process for the production of pharmaceutical-grade proteins in S.cerevisiae that has been developed using automatic feed control. The system has been successfully validated at the manufacturing scale at Novozymes Delta’s (biopharmaceuticals.novozymes.com) Nottingham facility, operating to a final working volume of 8,000 L.
Novozymes Delta undertook extensive strain development, molecular biology, and process development to offer S.cerevisiae as an alternative system for the commercial production of an extensive range of proteins representing many classes of biotherapeutics. The system confers the advantages of stable, high-level protein expression using animal-component free materials. It also delivers a highly competitive cost of goods typical of a microbial fermentation-based process.
This technology was commercially validated by the production of Recombumin®, a commercially available recombinant human albumin authorized for manufacturing use in human therapeutics. The FDA and the EMEA approved the use of Recombumin in the production of childhood vaccines for measles, mumps, and rubella for M-M-R® II (Merck & Co.and M-MRVAXPRO™ (Sanofi Pasteur MSD).
The system is optimized for the production of recombinant proteins where glycosylation does not naturally occur or can be designed out without impacting product efficacy. There is no induction phase, so one can avoid the use of methanol, characteristic of Pichia yeast-based processes. The constitutive expression system has not had any problems with proteolysis or other forms of degradation that makes a separate induction phase advantageous.
The laboratory-developed system of automatic feed control has worked well on scale up from 10–8,000 L. The production bioreactor, however, was designed to use overpressure to achieve maximum oxygen transfer capacity. Hence, a modification to the dissolved oxygen tension (DOT) control strategy was required.
DOT is normally controlled to 20% of air saturation. Satisfactory aerobic growth of S.cerevisiae, however, can be obtained over a wide range of dissolved oxygen concentration. Higher DOT values are not harmful, although some attention is required in setting up the DOT controller.
The technical difficulty is that DOT can change very rapidly. At high cell density, the rate of oxygen uptake can reduce DOT from 100% air saturation to zero in a few seconds if oxygen is not constantly supplied. The DOT probe itself, however, responds relatively slowly, typically requiring 30–60 seconds to equilibrate. This means that an oversensitive controller results in hunting with rapid and extreme oscillations in DOT and stirrer speed.
The solution is to set the controller to make relatively modest adjustments in stirrer speed with a sufficient delay between successive actions to allow for the probe response.
The DOT control was set up as a cascade system, first controlling stirrer speed by increasing stepwise and then, smoothly to give 20% air saturation, until its maximum is reached. This is followed by increases in airflow and overpressure, keeping the stirrer at its maximum speed to provide better mixing and foam control. With each step change there is an immediate rise in DOT, which then declines gradually as the culture grows, until 20% air saturation is reached, triggering the next step increase.
S. cerevisiae as a Process Organism
To produce recombinant proteins in high yields and of optimum quality, S.cerevisiae has been subjected to a series of genetic manipulations. The success of the design of the disintegration vector, designed as whole 2-µm vectors in a cir° background (Figure 1), dispels the common view that yeast episomal plasmids are too unstable for industrial use, as no plasmid loss occurs during the production process. The stability of this system was demonstrated in semicontinuous operation over a time scale of months.
The process is robust and only two physiological consequences of the genetic changes resulting from many accumulative genetic changes have been observed.
The first phenomenon is a reduction in the critical growth, µcrit, which is the highest rate at which growth is fully aerobic without production of ethanol or acetate. Values above µcrit will result in the build-up of unwanted by-products. Although a lower µcrit value theoretically results in a reduction in bioreactor productivity, this is of little economic significance, since at large scale, factors such as mass and heat transfer limit the maximum growth rate. The change in µcrit can be resolved by lowering the parameter used in the control algorithm that determines the effective growth rate in the process.
The second phenomenon is a tendency of the organism to produce acetate under conditions where there is a slight excess in nutrient supply. Ethanol production is readily detected by a rise in Respiratory Quotient (RQ) determined by exit gas analysis. Hence, the control algorithm is designed to adjust the feed rate automatically. Acetic acid cannot be detected by a change in RQ but can be identified by changes in conductivity. Novozymes Delta used this principle to develop a subroutine in the automatic control procedure to adjust the feed rate appropriately.
If acetate is present, there is a second substrate available for oxidation in addition to the sucrose in the feed. The result is that metabolism is not strictly limited by the feed rate. This can be detected by a limitation check. Under conditions of carbon limitation by feed supply, a reduction in feed rate causes an almost immediate reduction in catabolism and respiratory rate.
The first indication of this is a rise in DOT or fall in stirrer speed if it is controlling DOT followed by a decline in oxygen uptake and CO2 evolution determined by exit gas analysis. The procedure here is to make a periodic reduction, once every few hours, in growth rate by about 20% for a period of 20 minutes. If the expected changes indicating reduced respiratory rate are observed, the culture was in limitation before the reduction and the previous feed rate can be restored.
If the respiration rate remains unchanged, this indicates the presence of acetate, whose increased consumption compensates for the reduction in feed rate. Since the feed rate cut has already caused an increase in acetate consumption, no further action is usually required. A subsequent limitation check a few hours later will show that the process is now under control. If necessary, repeated cuts can be imposed.
Automated Process Control
The operation of the automated process control system in Figure 2 shows the time course of the feed rate from an early stage in the fermentation and the control of DOT throughout the process in a well-controlled fermentation. The feed rate increases exponentially until approximately the last 12 hours, when it was held constant as maximum oxygen transfer capacity was approached. The periodic limitation checks are apparent as spikes projecting downward from the line. Additionally, in most cases, a corresponding increase in DOT or reduction in stirrer speed can also be seen. In no case was the culture found to be out of carbon limitation.
The initially high DOT corresponding to the low starting biomass declined progressively as increasing biomass and feed rate resulted in increasing oxygen uptake. The control strategy using a cascade of increasing stirrer speed, airflow rate, and overpressure successfully maintained a DOT above 20% air saturation throughout the fermentation ensuring there was always sufficient dissolved oxygen for aerobic growth.
A robust and reliable process for the industrial-scale production of pharmaceutical-grade recombinant proteins in S. cerevisiae was achieved by overcoming the issues encountered during scale up. Evidence of this process’ success is the commercial-scale manufacture of Recombumin and transfer of the technology to partners taking production of therapeutic proteins to the tonnage scale.
Commercial-scale manufacture successfully validated the disintegration vector system, which provides excellent genetic stability. The automatic procedures developed in the laboratory for the control of the fed-batch process function at plant scale.
The in-house technical capabilities and systems developed in this process can be used for commercial production of an extensive range of proteins, representing many classes of biotherapeutics.