Premium
Artificial neural network‐genetic algorithm (ANN‐GA) based medium optimization for the production of human interferon gamma (hIFN‐γ) in Kluyveromyces lactis cell factory
Author(s) -
Unni Silpa,
Prabhu Ashish A.,
Pandey Rajat,
Hande Rohit,
Veeranki Venkata Dasu
Publication year - 2019
Publication title -
the canadian journal of chemical engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.404
H-Index - 67
eISSN - 1939-019X
pISSN - 0008-4034
DOI - 10.1002/cjce.23350
Subject(s) - akaike information criterion , artificial neural network , kluyveromyces lactis , box–behnken design , response surface methodology , genetic algorithm , microbiology and biotechnology , mathematics , chemistry , biological system , biology , chromatography , computer science , biochemistry , yeast , artificial intelligence , statistics , mathematical optimization , saccharomyces cerevisiae
In the current investigation, we have adapted response surface methodology (RSM) and artificial neural network‐genetic algorithm (ANN‐GA) based optimization to develop a defined medium for maximizing human interferon gamma production from recombinant Kluyveromyces lactis ( K. lactis ). In the initial screening studies, sorbitol and glycine emerged as a carbon and nitrogen source respectively having higher influence on hIFN‐γ production. Substrate inhibition studies were performed by varying the initial substrate concentration, and we found maximum hIFN‐γ concentration at 50 g L −1 of sorbitol. Inhibition kinetics studies were carried out using 3 and 4‐parametric models. Among the estimated models, the Moser model was observed as the best fitted model followed by the Luong model with R 2 values of 0.882 and 0.75, respectively. The model acceptability test was carried out using the extra sum of squares F‐test and Akaike information criterion (AIC). The Plackett‐Burman multifactorial design identified sorbitol, glycine, Na 2 HPO 4 , and MgSO 4 .7H 2 O as the parameters significantly influencing the hIFN‐γ production. Further, the Box‐Behnken design (BBD) followed by the artificial neural network coupled with genetic algorithm (ANN‐GA) was employed for the precise optimization of medium components. With ANN‐GA a maximum hIFN‐γ yield of 2.1 ±0.3 mg L −1 in shake flask level and 3.5 ±0.1 mg L −1 in reactor level was achieved. The findings of this study serve as a model for a process development strategy (bench scale to reactor scale) to achieve a high productivity of the desired protein from a microbial cell factory.