(766e) Raman Spectroscopy to Monitor the Phenotypic Responses of Bacteria to Environmental Stresses

The chemical composition of a bacterial cell is determined by a complex function involving the genotype of the organism and its environmental conditions.  One reason for studying cellular chemical composition is to determine the physiological response to a toxic compound.  Indeed, this knowledge will help guide future metabolic engineering approaches to making highly resistant cells.  Producing bacteria that can withstand high concentrations of alcohols, alkane biofuels, or commodity chemicals would be advantageous.  However, physically determining the chemical composition of a cell routinely involves mass spectrometry and is labor and time-intensive.  Furthermore, chemical composition of a bacteria cell can change continuously, demonstrating the need for real-time measurements.  Here, we demonstrate a new technology combining Raman spectroscopy and multivariate statistics to effectively monitor cell chemical composition in real-time.  To demonstrate our novel approach, we exposed Escherichia coli K12 to various growth inhibitory levels of 1-butanol, isobutanol, 2-butanol, and 1,4-butanediol.  We performed mass spectrometry-based analyses of membrane fatty acids, chromatographic analysis of amino acids, fluorescence anisotropy determination of membrane fluidity, and routine analyses of protein, nucleotide, and carbohydrate content.  These measurements were all in agreement with specific Raman signatures obtained in near real-time and illuminated by comparative peak analysis and discriminant analysis.  Furthermore, results of the Raman study were used to identify differences in the mechanisms of toxicity among these 4-carbon alcohols to E. coli.  This approach has very broad applications for studying phenotype response and ultimately deriving metabolic engineerng strategies to address specific mechanisms of toxicity.  In addition, we have made our Raman Diagnostic Analysis (RDA) tool, containing all relevant multivariate statistical models, publicly available as a stand-alone application for the analysis of Raman data.