1 Introduction
Lentiviral vectors (LVV’s) are one of the fastest growing vectors under development within the cell and gene therapy industry (McCarron, Donnelley, McIntyre, & Parsons, 2016). They offer many advantages including the ability to transduce dividing and non-dividing cells, offer high transduction efficiency, and low immunogenicity (Kotterman, Chalberg, & Schaffer, 2015). These favourable characteristics make them a promising tool for use in the treatment of various genetic and acquired diseases such as beta-thalassemia (Cavazzana-Calvo et al., 2010), Parkinson’s disease (Palfi et al., 2014) as well as oncology (Levine, 2015). With the increasing use of LVV in translational research and clinical programs, robust and scalable production processes are of critical importance in view of implementing these novel therapies for routine use (Merten, Hebben, & Bovolenta, 2016). However, current LVV batch or fed-batch upstream processes are limited by the inability to achieve high cell densities with the HEK293T cell line, resulting in poor yields (Manceur et al., 2017). Optimisation of cell culture conditions is therefore necessary to improve upstream efficiency.
Initial optimisation of suspension cell culture is traditionally carried out at small scale (i.e. microlitre to millilitre scale) under conditions that aim to mimic the large-scale bioreactor environment (Kumar, Wittmann, & Heinzle, 2004), and where possible, at high-throughput. However, adopting such an approach for LVV production employing conventional scales (e.g. miniature stirred tank reactors (STRs) and shake flasks) is both costly and impractical. Microwell based systems offer a suitable alternative to obtain key process design data early and cost-effectively (Martina Micheletti & Lye, 2006).
Previous studies have demonstrated the use of 24-standard round well (24-SRW) microtiter plates for the suspension cultivation of mammalian cell lines such as hybridoma (Barrett, Wu, Zhang, Levy, & Lye, 2010; M. Micheletti et al., 2006) and Chinese Hamster Ovary (CHO) (Chaturvedi, Sun, O’Brien, Liu, & Brooks, 2014; Mora et al., 2018; Silk et al., 2010) for monoclonal antibody production. However, examples of their use for LVV production are few in literature. Guy et al. , 2013 describe cultivation of HEK293 cells in 24-SRW plates to establish operating conditions for a 0.5 L wave bioreactor culture and demonstrated that the 24-SRW is an effective scale-down model (Guy, McCloskey, Lye, Mitrophanous, & Mukhopadhyay, 2013). HEK293T stable cell line cultivation in 24-deep square well (24-DSW) plates has yet to be demonstrated in literature. Compared to the 24-SRW plate, the 24-DSW format is advantageous as it offers 3-6 times greater working volume. Studies have also shown that square shaped wells have a better mixing and oxygen transfer profile compared to round well plates (Wouter A. Duetz & Witholt, 2004). To adopt the microwell system as a scale-down tool for LVV production, cell culture performance such as growth kinetics and LVV productivity must be matched with bench to pilot scale bioreactor processes. If not, there is no guarantee that the optimised process conditions achieved using microwells will be maintained upon scale-up.
Scale translation in cell culture is notoriously challenging due to the number and complexity of the variables affecting the physical and biological process. Criteria such as gas-liquid mass transfer coefficient (kLa), power input per unit volume (P/V) and mixing time have been used in chemical processes and adopted in biochemical engineering for bioreactor scale translation for different cultures (Alvin W. Nienow, 2006). Matched kLa is often used for scaling cell cultures with high oxygen demand (Ferreira-Torres et al., 2005). The oxygen demand for mammalian cells (e.g. HEK293) is relatively low compared to bacterial cell lines (e.g. E. coli) (Alvin W. Nienow, 2006), therefore matched kLa was not considered for the HEK293T stable producer cell line cultures. The use of matched P/V in this study was limited by the lack of accurate methods of determining P/V in microwells. Current methods only provide estimates of the power input using computational fluid dynamics (CFD) predications of the mean energy dissipation (Zhang, Lamping, Pickering, Lye, & Shamlou, 2008). As there are no direct methods to compare against, the CFD predictions are yet to be validated as accurate models of the actual P/V. In comparison, there are now more accurate and semi-automated methods of measuring mixing time such as the dual indicator system for mixing time (DISMT) technique. This method also enables mixing time to be characterised across reactors with different scales and geometries. Furthermore, Sani and Baganz, 2016 demonstrated matched mixing time as a suitable criterion for scaling between 0.5 L micro-bioreactor (MBR) and 5 L STR for a GS-CHO cell line (Sani & Baganz, 2016). However, scale translation for smaller scale vessels with different geometries is yet to be established. Therefore, in this work, we evaluate mixing time as a basis for HEK293T stable LVV producer cell line culture scale-translation between 24-DSW microtiter plates and bench-scale 2 L STR. First, we characterise mixing times in the 2 L STR and 24-DSW plates using the DISMT technique. Thereafter, growth kinetics, LVV productivity and cell metabolism of the HEK293T stable producer cell line are compared between the microwell and STR cultures at matched mixing time.