3.1.1 Growth kinetics and sugar consumption
Transgenic rice cell growth profiles and sugar levels of the cyclical semicontinuous bioreactor run is shown in Figure 1a. As expected, the biomass concentration in each cycle increased during the growth phase and somewhat dropped during the induction phase due to the lack of exogenous carbon source as also seen in our previous semicontinuous rice cell cultures (Corbin et al., 2016; Trexler et al., 2005). The growth periods in cycles 1-4 were 26, 21, 10 and 9 days, respectively (Table S1). Whereas the lag phase was not known in cycle 1 and not observed in cycles 3 and 4, the rice cell culture in cycle 2 likely experienced ~4-day lag phase (semi-log plot; data not shown) because rice cells probably took some time to acclimatize in the presence of a large amount of fresh NB+S medium (the working volume was increased from 13 L, the final working volume of cycle 1, to 32.5-L, the initial working volume of cycle 2) (Table 1). In addition, the agitation rate in cycle 2 was increased to 150 rpm compared to 100 rpm in cycle 1; as a result, rice cells likely took time to adjust themselves to the new environment. Note that the culture sample at day 0 of cultivation was not taken to avoid any risks of contamination and to test the sterility of inoculation process.
The maximum specific growth rates (µmax) of cycles 1-4 were comparable at 0.10 ± 0.02, 0.07 ± 0.01, 0.10 ± 0.04 and 0.08 ± 0.03 day-1, respectively, corresponding to the doubling times (τD) of 6.7 ± 1.1, 9.8 ± 1.8, 6.8 ± 2.4 and 9.3 ± 3.8 day, respectively (Table S1) . The µmax values in the semicontinuous run were somewhat lower than our previous report for lab-scale semicontinuous bioreactor operation (Corbin et al., 2016) where µmax of 0.15 ± 0.01 and 0.22 ± 0.01 day-1 and τD of 4.7 ± 0.34 and 3.2 ± 0.12 day were found in the first and second cycle, respectively, using the same transgenic rice cell line and NB+S medium. The discrepancy between the two studies is likely due to different hydrodynamic stress since baffles were equipped in this study but not in the lab-scale study. Also it could be due to different oxygen transfer, especially near the bioreactor wall, or from different Di/Dt ratios since kLa was not maintained between the two scales.
The growth phases in cycles 3 and 4 were shorter than cycles 1 and 2 (Figure 1a) due to the use of half-strength sucrose (Table 1) and their higher initial biomass concentrations (Table S1). Sucrose gradually decreased due to its conversion to glucose and fructose by rice cell wall invertases and reached 0 g/L at day 24 of cycle 1, day 20 of cycle 2, day 10 of cycle 3, and day 8 of cycle 4 (Table S1) following by glucose depletion a few days later (Figure 1a). The shorter period of the growth phase is desirable not only to improve productivity but also to minimize energy and materials supply to the bioreactor. It is pertinent to mention here that the initial sucrose concentration in cycle 2 was ~20 g/L rather than 30 g/L since the newly added medium was diluted by the remaining culture, while NB+0.5xS used in cycles 3-4 was prepared by taking this dilution into account.