Modeling perfusion at small scale using ambr15TM

Reaching cell densities higher than 80 million with the minimum possible perfusion rates is a goal for an increasing proportion of processes developed by biopharmaceutical companies. With the goal of fulfilling the industry needs for better commercial and customized perfusion media, SAFC evaluated different small-scale perfusion models to achieve an efficient work flow that can accommodate perfusion systems. SAFC successfully uses an optimized work flow for the development of media and feeds for fed-batch cell culture that integrates high-throughput screening, statistical tools and bench-top bioreactor scale studies. In this model, 96-deep well plates are used for the initial high throughput screening, followed by further development in spin tubes or shake flasks. At this time, there is no commercially available cell separation device that can be used for scales of 30mL or lower. The application of the 96-deep well plate or spin tubes model for perfusion showed to have severe limitations, specifically when trying to optimize processes to extremely low cell specific perfusion rates (CSPR). In order to develop media that can sustain the desired high densities and productivity at the desired low CSPRs, we needed a representative model that provided enough throughput to apply our statistical analysis. With this goal, we evaluated an alternative small scale model using the automation and process control offered by the ambr15TM. In this work, we show how ambr15TM fits in the work flow for perfusion media development and its comparability to a chemostat bioreactor. Conclusions and future work • The use of batch TPPs or well plates allows for high throughput screening that can be used for the selection of media to be used for perfusion. However, this work shows that observations made on TPPs do not always transfer to a chemostat system. It is not known at this time if this observation is due to differences between a dynamic and a continuous process or due to process parameters such as pH and DO and this is currently being investigated. • Differences when ranking different media based on their specific growth and productivity between TPPs and ambr15TM were found. The ranking in specific productivity was more similar between TPPs and ambr15TM if the comparison is made on the growth phase of the chemostat. While the absolute values are not comparable (Fig. 3B) the ranking and behavior on the stability of productivity was comparable (Fig. 1C and 2B). This fact underlines the need for a model that can represent a steady state as opposed to a dynamic model. • The chemostat run in ambr15TM is a semi-chemostat. Because of the discontinuity of the model when compared to a fully continuous process, limitations on growth and potentially productivity can become visible at higher dilution rates than in bioreactors (Figure 3A and B). • Work published by Heltmann (2015) and Henry et. al (2008) showed comparability in specific rates between chemostat and perfusion at similar CSPR. In this work we showed that ambr15TM can be used as a tool to model a chemostat and evaluate different media that ultimately is going to be used in perfusion mode. We expect that this model is going to allow us to evaluate media with very different characteristics in order to perform media component optimization at a relatively high throughput. Figure 1. High seed density batch culture • Peak cell densities between 10-15x106vc/mL were observed. Culture longevity was inversely correlated to the peak cell density (Fig. 1A) • Cells stopped producing at the peak density with the exception of M4 (Fig. 1B). • Specific growth rates ranked as M3 > M4> M1 > M2 (data not shown) • Specific productivity ranked as M4 > M2 ≥ M1 > M3 (Fig. 1C) CSPR: cell specific perfusion rate (nL/cell*d) D: dilution rate (vvd) F: volume of medium exchanged (mL) IVCD: integral viable cell density (cell*d/mL) μ: growth rate (d-1) N: number of media exchanges per day P: IgG concentration (mg/L) qp: specific productivity (pcd) X: cell density (vc/mL) V: working volume (mL) Small scale models