March 15, 2008 (Vol. 28, No. 6)

Bo Xu
Jens Gram

Method Can Improve Success Rate of Fermentation Process Development & Transfer

Astandardized approach for using process metabolic monitoring for enhancing microbial production of recombinant proteins is described in this tutorial by CMC Biopharmaceuticals. Implementation is based on readily available equipment such as off-gas composition analyzers used in new ways as a means of quickly applying the basic principles of process analytical technology (PAT) to bioprocesses.

Process Understanding

One of the simplest and best known of all expression systems, Escherichia coli, exhibits many different physiological patterns under varying growth and production conditions. It is difficult to monitor the consumption of carbon sources directly due to lack of reliable online glucose sensors. Monitoring exhaust-gas composition, however, is a simple means to increase process understanding during process development and optimization, thus greatly aiding in scale-up.

The basic use of fermentor exhaust-gas composition monitoring is to acquire metabolic data that reveals the physiological state of the cell population. Cell populations exhibit two related yet distinctly different metabolism profiles—anabolic and catabolic (Figure 1).


Figure 1

Anabolism is used by a cell to replenish new building blocks for cellular repair and growth, while catabolism is used to replenish cell energy and reduce power. To best understand a new process, it is first desirable to determine the total utilization of carbon source(s) and oxygen as well as how a cell distributes the carbon source for different metabolic routes. This information can, in turn, be used to derive control strategies with high success rates for process development.


Table

Intracellular consumption of carbon sources can be deduced from real-time oxygen measurements as follows:

C6H12O6 (glucose)+6O2 (oxygen) ? 6 CO2 (carbon dioxide)+ 6H2O (water)

With continuous oxygen-uptake rate (OUR) data and frequent glucose off-line measurements, a whole picture of the metabolic status of a cell growing and expressing a target protein in a bioprocess can be acquired.

Interpretation of OUR is given in the Table. With a high specific feed rate using glucose only, around 40% of the sugar is used by cell catabolism, while the ratio of catabolic activity to anabolic activity decreases during low specific feed rates. When a feed contains glucose and complex components, a rather constant ratio (60%) of catabolic to anabolic activity is observed regardless of the carbon-limitation condition being relaxed or stringent.

It should be noted that the ratio of catabolic to anabolic activity is dependent on other factors besides source fed and carbon limitation. A change in the process temperature or pH, or the addition of product inducer may also alter this ratio. Thus, online registration of oxygen-consumption demand provides indicative information to allow a better understanding of the process under development. It also provides an opportunity for retrospective data comparison, which is necessary for process optimization and scale-up.

Automatic Feed Start

The availability of automatic process control systems for monitoring metabolic changes can be used to increase the quality of development data as well as to develop high-yielding production runs. An example given here is the control of feed start during fed-batch, high cell-density microbial fermentation.

Traditionally, while implementing a fed-batch process, initiation of feed is done by visually observing the rise in the dissolved oxygen tension (DOT) signal combined with a steep decrease in stirrer speed (if DOT-stirrer control is applied), a process that can be labor intensive. The use of reliable oxygen-consumption rates to trigger feed allows well-characterized and controlled process operations as well as determination of the state of the organism’s metabolic profile.

As depicted in Figure 2, the signal used to automatically turn on a feed pump is a decrease in the monitored OUR, which is equivalent to a decrease in stirrer speed or an increase in DOT, called DOT spiking. The magnitude of OUR reduction is given by the simple expression in Equation 1, which must pass self checking as per the requirements given in Equations 2 and 3 in order to exclude noise and short-lived signal oscillations.


Figure 2

Equation 1 Signal pick-up Signal = OUR value<(OUR max value* 0.5)
Equation 2 Signal Self-Checking Time r = (Timer+1)*Signal
Equation 3 Signal Self-Checking Signal Flag = (Timer> = (120/CalcCycle– 1))

(Note: The terminology is based on the Sartorius Stedim programmable control-and-supervision SCADA system used for generating this data.)

Thus, a reduction by 50% of OUR (Equation 1) meets the feed auto-start criterion after being subject to a self-check for validity.

Figure 3 shows an example of using the described strategy to automatically link the batch-growth phase and the following fed-batch phase. The auto-feed start has proven to work seamlessly, allowing for flexible study plans, multiple parallel studies, and automatic process control that is scale independent. It can be integrated into a manufacturing-scale protocol used as an assisting tool to help supervise processes.


Figure 3

Process Scale-Up

Solid scale-up data is required to ensure smooth scale transfer leading to first-time-right (FTR) initiation of a process to large-scale operation. A practical and easy-to-use variable is the measured OUR and CER (carbon dioxide evolution rate) that are used to follow the outcome of a process under transfer. OUR and CER are dimensionless parameters and as such, can be evaluated before and after process scale-up to judge whether or not a process has been successfully transferred (Figure 4).

OUR and CER values can be used to fine-tune the configuration of large-scale fermentors so as to accommodate a process’ oxygen-transfer demand. While a standard stirred tank may be able to deliver between 200–250 mmol/L/h of oxygen, the actual applicable values may be lower than this value. Awareness of this limitation allows a development scientist to set boundary conditions for benchtop systems.


Figure 4

When oxygen enrichment is used to supplement the usual feed of sterile, heated and filtered atmospheric air, higher oxygen-consumption rates can be realized. Since the evolution of metabolic heat is proportional to the oxygen consumption, OUR can provide a good indication of heat generation, thus allowing calculation of the required cooling capacity of the target fermenter. Capacity must be large enough to remove all surplus heat generated from metabolism, agitation, and aeration in order to maintain a constant temperature.

Today, most biopharmaceutical fermentation manufacturing facilities are equipped with a fermenter exhaust-gas analyzer. Data provided by these devices can reflect cellular metabolic activities very well. Creative application of this data gives a good understanding of the metabolic state of growing/producing organisms, thereby enhancing process understanding, process optimization, and process transfer to large-scale manufacturing bioreactors.

Bo Xu, Ph.D. ([email protected]), is principal scientist, and Jens Gram, Ph.D., is director at CMC Biopharmaceuticals. Web: www.cmcbio.com.

Previous articleGenentech’s Option to Pursue Exelixis’ Oncology Candidate Triggers $3M Fee
Next articleConnecting Curious Cancer Clues