Submitted to: Meeting Abstract
Publication Acceptance Date: January 18, 2000
Publication Date: N/A
Abstract only

Technical Abstract: The goal of the work was to overcome the selectivity limitations inherent with whole cell microbial biosensors for use in multi-analyte mixtures. Glucose and ethanol were chosen as model analytes because a number of fermentation processes involve mixtures of these substrates. In one phase of the study, analytical techniques were tested using a combination of an amperometric microbial sensor, based on Gluconobacter oxydans cells, and a glucose-specific enzyme sensor (Reshetilov et al. 1998). In a second phase, an amperometric microbial sensor, based on cells of Pichia methanolica, replaced the enzyme sensor. Data were processed using polynomial approximations and an artificial neural network. Nearly complete additivity for total glucose and ethanol concentrations of up to 0.6 mM was observed with the amperometric sensor based on whole cells of G. oxydans. Within the range of concentrations from 1.0 to 10.0 mM, greatest accuracy was achieved with dual microbial sensors coupled to the artificial neural network. This system provided estimates of glucose and ethanol with high coefficients of determination (R**2= 0.99 for both analytes). Results demonstrated the potential for selective analyses of complex mixtures using a system composed of whole cell microbial sensors. In addition, it was shown that an artificial neural network was useful for data analysis. Reshetilov et al. 1998. Biosensors & Bioelectronics 13: 787-793.