Optimizing Xuri W25 bioreactor parameters
To further shorten the end-to-end process time, we evaluated whether several bioreactor parameters, including seeding density, dissolved oxygen (DO) level and rocking agitation, could accelerate cell growth. The same donor apheresis was split into five tested bioreactor conditions listed in Table 1 .
Xuri#1, the standard condition where cells were seeded at lower density (3.5e6 cells/ml) with lower agitation (3 rpm x 3°), yielded < 4e9 engineered T cells at day9 while Xuri#2, Xuri#3, and Xuri#5 all significantly promoted cell growth to around 10e9 cells through increasing rocking agitation or gassing oxygen directly (DO feedback loop control) to promote a greater level of oxygen transfer(Fig. 3A) . Moreover, Xuri#2, Xuri#3, and Xuri#5 with increased rocking speed and angle demonstrated that enhancing agitation also improved in-process cell viability from day6 to day9 compared to Xuri#1 (Fig. 3B) . In addition, shortening the bioprocess from 11 days to nine days did not adversely affect T cell purity, as comparable % CD3 T cells were found for all conditions upon harvest(Fig. 3C) .
While enhanced oxygen mixing through increasing agitation or direct oxygen input promoted cell growth, those two approaches contrasted in their impact on transduction efficiency. Having direct and steady oxygen delivery through DO feedback loop control from the start (Xuri#3) compromised transduction efficiency and resulted in the lowest TCR surface expression (Fig. 3D) . In contrast, DO feedback loop control at a later stage when the bioreactor was fully scaled up (Xuri#5) or enhancing oxygen mixing through rocking agitation at 500 mL volume (Xuri#2) led to TCR levels similar to standard conditions (Xuri#1) (Fig. 3D) . In addition, a higher inoculation density at 7e6 cells/ml (Xuri#4) led to a ~15% increase in transduction efficiency compared with conditions at 3.4e6 cells/ml (Xuri#2) (Fig. 3D) . Correlating with higher TCR expression in cells from Xuri#4 and % Tcm for cells from Xuri#2, INF-ɣ release was also modestly enhanced compared with Xuri#1 when co-cultured with MAGE-B2 peptide loaded T2 cells for 24 hours at E:T ratio at 1 (Fig. 3E) . No distinct differences were found for % T2 cells lysis at 48 hours due to saturation of T2 cell lysis above 99% (Fig. 3F) .
T cell activation and expansion usually comes at the cost of differentiation into effector phenotype (CD45RO+, CCR7-, CD95+) from naïve (CD45RA+, CCR7-, CD95-), stem (CD45RA+, CCR7+, CD95+) and central memory (CD45RO+, CCR7+, CD95+) pool, a phenomenon associated with a shortened life span, T cell exhaustion and compromised therapeutic persistence in the patient(Henning, Roychoudhuri, & Restifo, 2018; Wherry, 2011). Cells from Xuri#2, Xuri#3, and Xuri#5, which had greater oxygen/mass mixing compared to Xuri#1, contained significantly higher % Tcm cells than cells from Xuri#1 and Xuri#4 when compared at yield of 10e9 cells (Fig. 3G) . It is also notable that while cells from #Xuri4 had the highest TCR surface expression (Fig. 3D) among all the conditions tested, cell growth and %Tcm was lower than its counterpart conditions tested. On the other hand, while cells from Xuri#3 showed higher % Tcm and enhanced cell growth, TCR surface expression of those cells was 3-fold less than cells from Xuri#2(Fig. 3D) . This suggests that while oxygen promotes T cell growth, it may negatively impact transduction when supplied in excess during T cell activation. Overall, higher inoculation density and greater rocking agitation adopted at 0.5 L culture volume enables a shorter end-to-end bioprocess time with optimal T cell phenotype and potency.