Synergism Between Two High Throughput Platforms Efficiently Predicts Productivity

Rachel Legmann, Principal Scientist, SimCell Applications, Seahorse Bioscience

Small differences in manufacturing processes can affect the efficacy and safety of a therapeutic product. In efforts to rapidly bring high quality drugs to market, statistically designed experiments are a powerful strategy for effective cell culture clone screening and process development. Only through the use of miniaturized high throughput technologies can one realize the full benefits of multi-factorial experimentation for process development.

The SimCell is one such system that fulfills this need by providing a high throughput controlled culture environment equivalent to the bench scale reactors used today. It performs fed-batch protocols with pH, dissolved oxygen (DO), and glucose control across hundreds of micro-bioreactors. Cell density, metabolites and titer profiles permit rapid and more accurate analysis using multiple metrics. Such content results in more predictive experiments earlier in the development process. Very high throughput allows wider design spaces to be covered with greater statistical depth, providing better optimization and process robustness.

Unfortunately, many analytical techniques are not designed to meet the throughput requirements of the SimCell output as they are challenged by small harvest samples volumes and lack of a rapid, reliable and easy to use analysis. These challenges were addressed in this case study with the use of the ForteBio Octet QK platform to quantify monoclonal antibody during the production phase and at the end of the experiment. The results demonstrate that early selection of productivity levels is key in developing a robust process.

A mixed level, full factorial design was implemented in this DOE study to optimize and characterize production of a monoclonal antibody by a CHO cell line. DO and pH were varied at two levels, while feed rate and feed schedule strategy were varied across three levels. The impacts of these variations on cell growth and product yield were examined and compared. Each of the 36 unique experimental conditions were run with six replicates to yield a total of 216 micro-bioreactors. The temperature was set to 36.5°C for the duration of a 13-day experiment. DO and pH were set according to the DOE. Glucose was controlled to 2 g/L by the addition of glucose feed, if necessary, based on the off-line measurements starting on day 3 of culture. Feed rate and time of addition were performed according to the DOE. Samples were used to measure total cell density, viability and product titer on day 7, day 10 and at harvest on day 13.

After the completion of the factorial DOE in the micro-bioreactors (SimCell), a subset of 2 conditions were duplicated in conventional bench-scale bioreactors using 3L glass vessels (Applikon) operated at a 1L working volume. Two replicates for each condition were performed. MAb concentrations in crude samples from bench top bioreactors and micro-bioreactors were quantified using a plate-based, high throughput assay on an Octet QK instrument (ForteBio). Twenty microliters of a 1:10 diluted culture was transferred to a 96-well plate and further diluted, 1:10, with CD-CHO media. The plate was then transferred to the Octet instrument where MAb concentrations for each sample were measured. The total analysis time for a single 96-well plate was approximately 30 minutes.

The overall objective of this study was to examine the potential synergies between two high throughput platforms, SimCell and Octet QK, to predict titer outcome and compare the data to standard predictions made by bench-scale bioreactors.

In this study a DOE with 216 micro-bioreactor experiments was conducted in the span of less than a month. The response of an IgG producing CHO fed-batch process to 4 factors with respect to large scale production was extensively characterized. Statistical analysis of the results identified interactions and effects of four factors on product yield as determined by ForteBio for this defined process space.

Figure 2 shows the overall performance of the 36 unique experimental conditions in the factorial design and two conditions in bench top bioreactors using titer as the metric. Manipulation of process parameters is shown to have very significant impact on titer. It is clear that the titer results between the two systems have strong quantitative correlation between the predictions of the micro-bioreactor and the results of the bench-scale bioreactor models. A previous study showed good correlation (R² value of 0.97 with a slope of 1.02) between SimCell micro-bioreactors and bench-scale bioreactors for terminal product titer as measured by the Octet (Legmann et al, 2009).

The results demonstrate that combination of these two high throughput platforms acts synergistically as an effective and efficient tool to predict productivity. This combined platform permits a single experimenter to increase the scope of experimentation by a factor of twenty or more. These larger experiments allow the potential design space to be penetrated at greater statistical depth in less time. 

References

Legmann R, Schreyer HB, Combs RG, McCormick EL, Russo AP, Rodgers ST. 2009. A predictive high-throughput scale-down model of monoclonal antibody production in CHO cells. Biotechnol Bioeng 104(6):1107–1120.