An artificial intelligence approach to Bacillus amyloliquefaciens CCMI 1051 cultures : application to the production of anti-fungal compounds

Abstract

Article history: Received 2 February 2010 Received in revised form 15 July 2010 Accepted 19 July 2010 Available online 27 July 2010 Keywords: Bacillus amiloliquefaciens Spore formation Anti-fungal activity Neural networks 1. Introduction Biopesticides based on natural endophytic bacteria to control plant diseases are a promising ecological alternative to chemical treatments. Bacillus species produce a wide variety of metabolites with interesting biological activities, among them iturinic lipopep- tides antibiotics (Bottone and Peluso, 2003; Cho et al., 2003; Moyne et al., 2001). The antimicrobial activity exhibited by Bacillus sp. is dependent on the culture medium composition, and different nitro- gen sources can result in the production of different antibiotics (Besson et al., 1987; Chevanet et al., 1986; Davis et al., 1999; Volpon et al., 2000). Aspartic acid is the preferred nitrogen source for the production of iturinic compounds by Bacillus subtilis (Besson et al., 1987) and Bacillus amyloliquefaciens (Caldeira et al., 2006, 2007, 2008). With increasing culture time, the nutrient content changes and adverse environmental conditions appear. Thus, incubation time is another factor influencing antibiotic production (Caldeira et al., 2008; Feio et al., 2004; Moyne et al., 2001), as the response to adverse environmental conditions can lead to activation of differ- ent mechanisms for the production of antibiotics giving a compet- itive advantage to the producer microorganism (Dieckmann et al., 2001). The link between antibiotic production and Bacillus sporulation is not fully understood. During production of lipopeptides in sub- ⇑ Corresponding author. Tel.: +351 266 745 315; fax: +351 266 745 303. E-mail address: hvicente@uevora.pt (H. Vicente). 0960-8524/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biortech.2010.07.080 abstract The combined effect of incubation time (IT) and aspartic acid concentration (AA) on the predicted bio- mass concentration (BC), Bacillus sporulation (BS) and anti-fungal activity of compounds (AFA) produced by Bacillus amyloliquefaciens CCMI 1051, was studied using Artificial Neural Networks (ANNs). The values predicted by ANN were in good agreement with experimental results, and were better than those obtained when using Response Surface Methodology. The database used to train and validate ANNs con- tains experimental data of B. amyloliquefaciens cultures (AFA, BS and BC) with different incubation times (1–9 days) using aspartic acid (3–42 mM) as nitrogen source. After the training and validation stages, the 2–7-6–3 neural network results showed that maximum AFA can be achieved with 19.5 mM AA on day 9; however, maximum AFA can also be obtained with an incubation time as short as 6 days with 36.6 mM AA. Furthermore, the model results showed two distinct behaviors for AFA, depending on IT.