Modeling production of antifungal compounds and their role in biocontrol product inhibitory activity.

Partial least squares (PLS) regression modeling was used to relate the antifungal activity of Bacillus subtilis solid-state fermentation extracts to the individual high-performance liquid chromatography (HPLC) peaks from those extracts. A model was developed that predicted bioassay inhibition based on the extract HPLC profile (R(2) = 0.99). Concentrations of the members of the antifungal lipopeptide families iturin A and fengycin were found to correlate positively with extract inhibition, but a peak with unidentified chemical composition (designated as peak 48) showed the strongest correlation with extract inhibition. HPLC data were used to construct models for the production of iturin A, fengycin, and peak 48 as a function of the substrate moisture content, incubator temperature, and aeration rate in the solid-state bioreactors. Maximum production of all compounds occurred at the highest moisture content (1.7 g/g dry basis) and lowest incubator temperature (19 degrees C) tested. Optimal aeration rates for the production of the two known lipopeptides and peak 48 were 0.1 and 1.5 L/min, respectively.