T5-09 Prediction of Bacillus weihenstephanensis Acid Resistance Using Gene Expression Quantification as Molecular Biomarkers

Tuesday, July 30, 2013: 11:00 AM
213BC (Charlotte Convention Center)
Noemie Desriac, ADRIA Development, Quimper, France
Louis Coroller, LUBEM-UMT 08.3 PHYSI'Opt, Quimper, France
Daniele Sohier, ADRIA, Quimper, France
Florence Postollec, ADRIA Development, Quimper, France
Introduction: The physiological state of vegetative cells has a great impact on bacterial resistance. The Omics data allow understanding of  microbiological behavior to environmental change due to the process, or storage conditions. The next step is to integrate these data in a quantitative approach.

Purpose: In this study, an integrative approach was followed to select molecular biomarkers of bacterial resistance to further predict the acid inactivation of Bacillus weihenstephanensis.

Methods: Combining gene expression and survival ability, potential biomarkers of acid resistance of B. weihenstephanensis were identified. RT-qPCR gene expression of 31 genes was quantified (i) during exposure to sublethal conditions and correlated to a subsequent acid-resistance (3D values i.e. the time necessary to lose 99.9% of the bacterial population) and (ii) throughout bacterial inactivation and correlated to an instant first bacterial decrease, i.e. the time necessary to lose 90% of the bacterial population at time t. An individual gene selection was made, and a selection of gene set was also made using a Partial Least Square (PLS) regression. The main advantage of this method is to take gene expression interactions into account.

Results: In sublethal conditions, 4 genes exhibited a linear correlation between their expression and subsequent bacterial resistance. They could be selected as direct biomarkers. While 9 genes, named long-acting biomarkers, showed an up-regulation during short adaptation time and were correlated to an increased acid-resistance over time. It highlighted the importance of non-linear correlation particularly when focusing at transcriptional level. In lethal conditions a set of 8 genes were selected to predict acid-resistance. Using this PLS model, the bacterial resistance of two independent samples was predicted at 2.1 h and 3.3 h whereas the resistances were observed at 2.8 h and 3.7 h, respectively. Further investigations are running to validate the mathematical model in process like conditions.

Significance: This study underlines the possibility to integrate the bacterial physiology state, using molecular biomarkers, into bacterial behavior modeling and thereby further improve microbial risk assessment.