Pilot-scale process development and scale up for antifungal production

Abstract

A pilot-scale fermentation was developed for an antifungal compound produced by a filamentous fungus. Replacement of galactose with lactose (20-fold cost savings) and a threefold phosphate reduction (15 to 5 g/L) improved productivity 2.5-fold. Addition of supplements--glycine, cobalt chloride, and trace elements--resulted in a further twofold productivity increase, greater process robustness, and less foaming which reduced antifoam addition tenfold (30 to <3 mL/L). Mid-cycle lactose limitations were addressed by raising initial lactose levels (40 to 120 g/L) resulting in another twofold productivity increase. Overall, peak titers increased tenfold from 45 +/- 9 to 448 +/- 39 mg/L, and productivities improved from 3 to 25 mg/L day. Despite its high productivity, process scale up was challenged by high broth viscosity (5,000-6,000 cP at 16.8 s(-1)). Gassed power requirements at the 600 L scale (4.7 kW/1,000 L) exceeded available power at the 15,000 L scale (3.0 kW/1,000 L), and broth transfer to the downstream isolation facility was hindered. Mid-cycle broth dilution with up to five 10 vol% additions of 12 wt% lactose solution or whole medium-reduced viscosity three- to fivefold (1,000-1,500 cP at 16.8 s(-1)), gassed power within scale-up limits (2.5 kW/1,000 L), and peak titer by up to 45%. The process was scaled up to the 15,000 L working volume based on constant aeration rate (vvm) and peak impeller tip speed, raising superficial velocities at similar shear. This strategy maximized mass transfer rates at target gassed power per unit volume levels, and along with controlled broth viscosity, precluded multiple dilution additions. A final titer of 333 mg/L with one dilution addition was achieved, somewhat lower than expected, likely owing to inhibition from some unmeasured volatile compound (not believed to be carbon dioxide) during an extended period of high back-pressure in the early production phase.

Process performance of parallel bioreactors for batch cultivation of Streptomyces tendae

“…The impeller speeds of the milliliter-scale stirredtank bioreactors were set to ensure a similar power input compared to the liter-scale stirred-tank bioreactor, as it is often done for cultivations with filamentous microorganisms [13]. Shake flask cultivations were run with a small filling volume of 50 mL (one-tenth of the total volume of the shake flask) to ensure a higher oxygen transfer.…”

“…The reason for this change is the intertwined mycelial structure which is pulled apart and aligns if the shear rate is increased [11]. The exact knowledge of the apparent viscosity, and the average shear rate accordingly, at the small-scale is important for scale-up considerations as engineering variables like the oxygen transfer coefficient (k L a) and the volumetric power consumption are significantly affected by changes in viscosity [12,13]. Despite these facts, no comparative study of the performance of different small-scale bioreactor systems and their scale-up abilities is available so far for filamentous microorganisms.…”

Targeted purification development enabled by computational biophysical modeling

“…Second, the binding kinetics between the protein and the chromatographic ligand should allow for a shorter residence time, which will convert to a shorter loading time and higher recovery, also reducing the likelihood of any undesired proteolytic events. Finally, the percentage of recovered protein upon elution should be high enough to reduce costs and ensure robustness of the selected purification platform . When compared to empirical results, an in‐silico binding affinity (Δ G ) of protein–ligand complex in the range of −5 to −3 kcal/mol gives enough room for strong binding and good recovery upon empirical optimization for insulin purification.…”

Scale-up of micro-aerobic 1,3-propanediol production with Klebsiella pneumonia CGMCC 1.6366

“…Since agitation rate was scaled based on ITS (proportional to nD where n was the agitation speed and D was the impeller diameter), shear rate (proportional to n 2 D 2 for a turbulent peak Reynold's number of 10 4 ) likely remained similar [25]. Thus, it was expected that broth viscosity would remain similar.…”