Novel chemometric strategy based on the application of artificial neural networks to crossed mixture design for the improvement of recombinant protein production in continuous culture

CAROLINE DIDIER,; FORNO G; MARINA ETCHEVERRIGARAY,; RICARDO KRATJIE,; GOICOECHEA, HÉCTOR C

ANALYTICA CHIMICA ACTA

Elsevier

Lugar: Amsterdam; Año: 2009 vol. 650 p. 167 - 174

0003-2670

The optimal blends of six compounds that should be present in culture media used in recombinant proteinproduction were determined by means of artificial neural networks (ANN) coupled with crossedmixture experimental design. This combination constitutes a novel approach to develop a medium forcultivating genetically engineered mammalian cells. The compounds were collected in two mixtures ofthree elements each, and the experimental spacewas determined by a crossed mixture design. Empiricaldata from51 experimental units were used in a multiresponse analysis to train artificial neural networkswhich satisfy different requirements, in order to define two newculturemedia (Medium 1 andMedium 2)to be used in a continuous biopharmaceutical production process. Thesemediawere tested in a bioreactorto produce a recombinant protein in CHO cells. Remarkably, for both predicted media all responses satisfiedthe predefined goals pursued during the analysis, except in the case of the specific growth rate ()observed forMedium 1. ANN analysis proved to be a suitable methodology to be used when dealing withcomplex experimental designs, as frequently occurs in the optimization of production processes in thebiotechnology area. The presentwork is a newexample of the use of ANN for the resolution of a complex,real life system, successfully employed in the context of a biopharmaceutical production process