Establishing a fully closed bioprocess to generate autologous TCR-T cells
To streamline our TCR-T autologous bioprocess and lower COGM, we identified the costliest materials and evaluated if they were necessary for generating a high yield of transgene positive T cells. Published T cell bioprocesses enrich T cells from the leukopak prior to activation5, 9,10, however the CD4 and CD8 magnetic beads and single-use kit used for selection make up a significant portion of the raw material costs of manufacturing. We demonstrated that eliminating the T cell enrichment step resulted in comparable T cell purity following expansion to T cells enriched from a CliniMACS Plus at day8 (Fig 1B & Supplemental Fig. 1D) . Unexpectedly, these leuko-apheresed cells also had higher transduction efficiency compared to T cells enriched from CD4+ and CD8+ positive selection independent of whether the transgene was a CAR, GFP or TCR when tested with three different donors in small-scale studies (Fig. 1A) .
Many established bioprocesses involve open steps where cells are moved between culture vessels for T cell activation and transduction, mainly due to the use of plate-bound activators and transduction enhancers. Some processes begin expansion in static vessels or bags before transferring to a rocking bioreactor(Eyles et al., 2019; Highfill & Stroncek, 2019). We hypothesized that these separated unit operations and associated open steps adversely affect process performance by subjecting T cells to an unfavorable environment such as deprivation of heterogeneous cell contact, cytokines, and insufficient/timely medium exchange due to lack of perfusion(Chan & Shlomchik, 2000; Deola et al., 2008). We envisioned integrating these steps into one single bioreactor to simplify the workflow and aid in the efficiency of cell engineering and expansion. To enable this, a soluble activator was used to eliminate the need to pre-coat bags. Through optimization of process parameters, we demonstrated that adding ImmunoCult CD3/28/2 to leuko-apheresed cells in a semi-static bioreactor with rocking speed and angle at 2 rpm x 2° is effective in enabling T cell activation (Supplemental Fig. 1A) . Cost of lentiviral vector limits the number of cells that can be transduced, and transductions with a multiplicity of infection (MOI) of 5 or higher must be performed in a gas permeable bag due to the low volume necessary for 1-2e8 cells at an optimal cell density. We found a MOI of 1 was efficient for transduction of multiple TCRs(Supplemental Fig. 1B ). Due to the low amount of lentivirus needed for efficient transduction, we concluded that it is feasible to transduce roughly 7e8 cells in the bioreactor on day1. Together, these findings enabled an integrated workflow for TCR-T cell production by incorporating cell activation, transduction, and expansion in a single bioreactor immediately after the leukopak wash step using the Sepax Pro (Fig. 2 ).
This fully closed bioprocess yields an average of 10e9 T cells from ~1e9 leukopak cells as starting material within 7-8 days, and a peak of 20-30e9 cells can be reached in 10 days(Fig. 1C) . Viability dips midway through the process due to non-T cell death, but rebounds to >97% by day6 of expansion (Fig. 1D & Supplemental Fig. 1C) . Variation is observed in growth curve and in-process viability due to donor-to-donor variability, however, end-of-process viability is consistently above 90% (Fig. 1D) . Semi-continuous perfusion ensures timely replenishment of nutrients, removal of undesired metabolites and cell debris and maintains glucose levels above 2g/L and lactate levels below 2 g/L, respectively (Fig. 1E & 1F) . Cells harvested at the end of process (EOP) showed TCR expression of CD8+ T cells detected by MHC dextramer, ranging from 30% to 70% with mean expression at 55%(Fig. 1G) . Mean TCR expression of CD3+ T cells was 30%, however, due to MHC class I tetramer being only able to detect TCR expression on CD8+ cells (Fig. 1G) . Staining with a Vbeta antibody Vβ13.2 targeting the TCR β chain confirmed that CD4+ cells are transduced, but CD4+ TCR expression is undetectable by dextramer staining (Supplemental Fig. 1C) . 97% purity of CD3+ T cells was achieved at EOP regardless of CD3+ percentage of leukopak starting material (20-80%) (Fig. 1H) . Harvested TCR-T cells showed potent and specific cytotoxicity against MAGE-B2 peptide loaded T2-Luc cells at E:T ratio of 1 after 24 hours during in vitro functional assays (Fig. 1I) .