Fructo-oligosaccharides (FOS) are prebiotic low-calorie sweeteners that are synthesized by the transfer of fructose units from sucrose by enzymes known as fructosyltransferases. If these enzymes generate β-(2,6) glycosidic bonds, the resulting oligosaccharides belong to the neoseries (neoFOS). Here, we characterized the properties of three different fructosyltransferases using a design of experiments approach based on response surface methodology with a D-optimal design. The reaction time, pH, temperature, and substrate concentration were used as parameters to predict three responses: The total enzyme activity, the concentration of neoFOS and the neoFOS yield relative to the initial concentration of sucrose. We also conducted immobilization studies to establish a cascade reaction for neoFOS production with two different fructosyltransferases, achieving a total FOS yield of 47.02 ± 3.02%. The resulting FOS mixture included 53.07 ± 1.66 mM neonystose (neo-GF3) and 20.8 ± 1.91 mM neo-GF4.

The enzymatic production of prebiotic fructo-oligosaccharides (FOS) from sucrose involves fructosyltransferases (FFTs) and invertases, both of which catalyze forward (transferase) and reverse (hydrolysis) reactions. FOS yields can therefore be increased by favoring the forward reaction. We investigated process conditions that favored transferase activity in the yeast strain Kluyveromyces lactis GG799, which expresses a native invertase and a heterologous FFT from Aspergillus terreus. To maximize transferase activity while minimizing native invertase activity in a scaled-up process, we evaluated two reactor systems in terms of oxygen input capacity in relation to the cell dry weight. In the 0.5-L reactor, we found that galactose was superior to lactose for the induction of the LAC4 promoter, and we optimized the induction time and induction to carbon source ratio using a response surface model. Based on the critical parameter of oxygen supply, we scaled up the process to 7 L using geometric similarity and a higher oxygen transport rate, which boosted the transferase activity by 159%. To favor the forward reaction even more, we deleted the native invertase gene by CRISPR/Cas9 genome editing and compared the ΔInv mutant to the original production strain in batch and fed-batch reactions. In fed-batch mode, we found that the ΔInv mutant increased the transferase activity by a further 66.9%. The enhanced mutant strain therefore provides the basis for a highly efficient and scalable fed-batch process for the production of FOS.

The hydrodynamic characterization of bioreactors is an integral element in the industrial development of cell and gene therapy platforms where mild process conditions with low power inputs are desired. In this technology, modular single‐use equipment is frequently used for which accurate techniques to measure or predict scaling parameters, such as power input, are of high interest. In this work, computational fluid dynamics as well as two experimental approaches are used to determine the power input in a small‐scale bioreactor and validate the employed techniques. Particular attention is given to the implications arising from low‐speed operation with low power input conditions.