Discussion
Autologous cell therapy would benefit from a cost-effective, streamlined manufacturing process that yields cell numbers relevant for solid-tumor indications with high transgene expression and the desired T cell phenotype. The raw materials and the complexities of unit operations employed by typical bioprocesses, particularly the costly antibodies for magnetic cell selection and high number of touchpoints requiring skilled labor, drive up COGM(ten Ham et al., 2020). The closed autologous T cell bioprocess described here demonstrates that eliminating T cell enrichment by starting with washed leuko-apheresis material not only reduced COGM without compromising EOP CD3+ T cell purity, but also increased transduction efficiency. Expansion conditions including high seeding cell density (7e6 cells/mL), volume, rocking agitation, and low glycolytic environment (inhibiting glycolysis using 2-DG at 5mM) in the bioreactor also positively regulated TCR expression and T cell doubling time.
T cell manufacturing processes involving open steps require cells to be manually transferred between culture wares or fed in a biosafety cabinet within a Grade B cleanroom(Dietz, Padley, & Gastineau, 2007), increasing the risk of cross-contamination or microbial contamination and negatively affecting the efficiency and robustness of T cell manufacturing due to engineering control constraints. The only commercially available all-in-one T cell processing unit, the CliniMACS Prodigy (Miltenyi Biotec), is also limited by its chamber size, scale-out ability and fixed operation programs. Presented here is a fully closed T cell bioprocessing system utilizing single-use disposable transfer bags for raw material transfer and a single bioreactor for flexible and efficient T cell activation, transduction, and expansion. This approach not only reduces the end-to-end processing time and enhances key critical quality attributes of engineered T cells, it increases manufacturing efficiency by enabling processing of multiple lots in parallel within a Grade C cleanroom.
The critical quality attributes (CQA) of commercial autologous T cell therapy include identity, viability, purity, % transgene positive, potency, absence of impurities and sterility(Aijaz et al., 2018; Lipsitz, Timmins, & Zandstra, 2016). To develop and control an autologous bioprocessing system that meets these specifications while accommodating the inherent donor variabilities requires a delicately designed workflow guided by quality-by-design principles(Lipsitz et al., 2016). The integrated operation developed here contrasted with representative CAR-T cell bioprocess where T cells are enriched from leukopak, activated, and transduced in gas permeable bags, and transferred in a bioreactor for expansion, as depicted inSupplemental Fig. 2A . We have found that an integrated operation resulted in superior critical quality attributes of engineered T cells, including TCR-T cell yield, TCR expression, % memory subset, and cytotoxicity (Supplemental Fig. 2B-I ), during a study using a same donor leukopak. This enhancement of COA is possibly attributed to collective factors from soluble activators and non-T cell population, such as B cells and platelets, and bioreactor parameters(Bieback, Fernandez-Munoz, Pati, & Schafer, 2019; Canestrari, Steidinger, McSwain, Charlebois, & Dann, 2019; Chan & Shlomchik, 2000; Deola et al., 2008). Indeed, we found that activation plays a vital role in transduction efficiency and that ImmunoCult CD3/28/2 (StemCell Technologies) dramatically increased TCR expression compared to TransAct (Miltenyi Biotech) at the same MOI (Supplemental Fig. 1F) .
Stimulation drives differentiation of naïve T cell lymphocytes to effector cells and subjects them to activation-mediated apoptosis(Gett et al., 2003). The fitness of engineered T cells correlates with their therapeutic efficacy and long-term persistence post-transplantation(Gett et al., 2003). It also depends on the cellular characteristics that the manufacturing bioprocess cultivates, mainly through the cellular signals that activation antigen and cytokines trigger and the metabolic and mechanobiological environment of the bioreactor(Franco, Jaccard, Romero, Yu, & Ho, 2020; Rushdi et al., 2020; Scharping et al., 2021; Schluns & Lefrançois, 2003). Consistent with findings that mTOR-driven anabolic growth drives T cell differentiation from a naïve and memory-like state to effector mode after activation(Huang, Long, Zhou, Chapman, & Chi, 2020), we found that limiting glucose, the major carbon source of glycolysis, by using the glucose analog 2-DG effectively attenuated the differentiation of naïve T cells into Tem and preserved Tcm during expansion. 2-DG does not affect glycolysis as a single event but may result in a holistic starvation signal that leads to an overall lowered energetic profile(Sukumar et al., 2013). The slight increase in TCR expression that accompanied the enhanced memory phenotype may be a result of redirection of energetic metabolites from mitochondrial oxidation to biosynthesis. After antigen activation, both endogenous and exogenous IL-2 is engaged and further drive the antigen response while regulating pro-survival molecules, such as B cell lymphoma-2 (Bcl-2), until metabolic homeostasis is reached(Gett et al., 2003; Schluns & Lefrançois, 2003). In turn, the strength of activation signals have a profound impact on T cell fitness related to persistence(Gett et al., 2003; Schluns & Lefrançois, 2003). Increased activation signals lead to sustained activation post-production and could elicit life-threatening CRS(Hay, 2018; Obstfeld et al., 2017). Here it is shown that IL-2 withdrawal from feed medium coupled with semi-continuous perfusion resulted in a gradual decline of the activation state of T cells to less than 5% without expanding the population of FoxP3+ Tregs (data not shown). The low activation state of harvested T cells correlated with reduced non-specific killing of T2 cells without affecting their specific killing of target cells. More importantly, the lowered activation signals via IL-2 withdrawal enhanced metabolic fitness of T cells indicated by less polarized mitochondria and mitochondrial mass when in effector mode(Sukumar et al., 2016). Taken together, these results suggest that fine tuning activation signals during T cell manufacturing can impact the phenotypic characteristics of TCR-T cells. These optimizations enabled a robust manufacturing bioprocess that generated a high yield of quality TCR-T cells with a memory phenotype that has been shown to correlate with increased clinical efficacy while also reducing COGM and eliminating many of the manual touchpoints necessary in typical autologous manufacturing processes.