Broadley-James Corporation, Emerson Process Management, and the University of Texas at Austin are working together to examine and quantify the potential for faster optimization of batch operating points, process design, and cycle times. We're also looking for more reproducible and predictable batch endpoints.
The objective of this effort is to show that the impact of PAT can be maximized through the integration of dynamic simulation and multivariate analytics in a laboratory-optimized control system during product development.
Data from bench-top and pilot-plant cell culture runs are being used to create multivariate analytic and high-fidelity, first-principle cell culture models to prototype process changes. The tools that are being used could provide significant improvement in process development and process control by laying a foundation for real-time release capabilities as defined by the PAT guidelines (1). Potential benefits include a more automated, seamless, and effective commercialization process that should translate to faster time to market and early release of biological products.
Currently more than 400 biotechnology medicines are in development for more than 100 diseases (2). These products generally require overlapping and iterative stages for process development and commercialization: cell line selection and development, media optimization, process conditions optimization and verification, scale-up, project definition, and plant design.Photo 1:
Beta tests are in progress to explore the use of a new dynamic model and online data analytics in the development and scale-up of new products derived from mammalian cell lines. A key objective is to make results fully public to encourage extensive use and advancement of the concepts and methodology.
Although our tests involve bench-top and pilot-plant bioreactors, these tools are important for industrial bioreactors as well. Data collected during product and process development often may be the best source for initial model development, particularly because changes must be minimized in production runs. Using these tools in a laboratory provides an opportunity to evaluate their performance and establish an effective basis in a production system for real-time release.Laboratory Set-Up
The CHO cell line for these beta tests uses a glutamine synthetase (GS) expression system to produce an antibody. GS catalyses biosynthesis of glutamine from glutamate and ammonia. The GS expression system uses methionine sulphoxamine (MSX) in culture media to inhibit endogenous GS. These CHO cells are modified with the recombinant protein of interest and an exogenous GS selective marker (3, 4). No exogenous glutamine can be added to this system, or the selection pressure will break down and lead to a culture with a great deal of genetic drift for the recombinant protein being produced. Another side effect of this metabolic change is that the GS system consumes its own waste ammonia, so ammonia levels tend to be lower in these cultures. These cells also tend to produce less lactate. Consequently, this system requires less sodium carbonate to maintain pH, resulting in lower osmolality than other CHO selection systems. The lower ammonia and osmolality levels contribute to an increase in culture longevity, resulting in batch times of 22-30 days rather than the 10-12 days normally associated with CHO lines. Cells are maintained and batches are run in a proprietary, serum free media supplemented with glutamate before and during the run.
Our laboratory bioreactors are controlled by a laboratory-optimized industrial distributed control system (DCS) that can use embedded modeling, analytical, monitoring, tuning, and advanced control tools. For the beta tests, we embedded those tools in a new DCS station that was connected by OPC communication protocol to an existing laboratory DCS application station. Multiple 7-liter bench-top bioreactors are connected to the laboratory DCS (Photo 1) for process optimization and model development. A 100-liter jacketed pilot-plant size “single-use” bioreactor (Photo 2) was connected to the laboratory DCS for scale up of the process and the model.
A key part of our test set-up is the use of on-line and at-line analytical measurements. Each bioreactor has a near-infrared (NIR) probe and a dissolved carbon dioxide (DCO2) electrode as well as dissolved oxygen (DO) and pH electrodes. Our at-line analyzer (Photo 3) has an automated sample system that every four hours provides measurements of lactate, ammonia, glutamine, glutamate, cell count, cell viability, cell diameter, and osmolality. These measurements enable an extension of basic control (e.g., proportional and proportional-integral control) and the addition of advanced control (e.g., model predictive control) for cell culture. These feedback controllers inherently transfer variability from key process variables (e.g., DO and media formulations) to feeds (e.g., air, O2, glucose, and amino acids) and provide a standardized and direct method for adjustment of the process variables according to their set-points by a higher level of control or supervision (5).
Additionally a method has been developed to provide the fastest possible automated approach to a PID controller set-point (3). The extension of feedback control offers these advantages:
eliminates the need to develop schedules for automated dosing or feed profiling
makes design of experiments less complex
allows for fewer batches for design of experiments
provides more reproducible batches
speeds transitions to new process conditions
facilitates more effective data analytics by elimination of unmeasured disturbances
improves commercialization of optimization opportunities by making them faster and more definable.