Rapid screening for metabolite overproduction in fermentor broths, using pyrolysis mass spectrometry with multivariate calibration and artificial neural networks

Abstract

Binary mixtures of model systems consisting of the antibiotic ampicillin with either Escherichia coli or Staphylococcus aureus were subjected to pyrolysis mass spectrometry (PyMS). To deconvolute the pyrolysis mass spectra, so as to obtain quantitative information on the concentration of ampicillin in the mixtures, partial least squares regression (PLS), principal components regression (PCR), and fully interconnected feedforward artificial neural networks (ANNs) were studied. In the latter case, the weights were modified using the standard backpropagation algorithm, and the nodes used a sigmoidal squashing function. It was found that each of the methods could be used to provide calibration models which gave excellent predictions for the concentrations of ampicillin in samples on which they had not been trained. Furthermore, ANNs trained to predict the amount of ampicillin in E. coli were able to generalise so as to predict the concentration of ampicillin in a S. aureus background, illustrating the robustness of ANNs to rather substantial variations in the biological background. The PyMS of the complex mixture of ampicillin in bacteria could not be expressed simply in terms of additive combinations of the spectra describing the pure components of the mixtures and their relative concentrations. Intermolecular reactions took place in the pyrolysate, leading to a lack of superposition of the spectral components and to a dependence of the normalized mass spectrum on sample size. Samples from fermentations of a single organism in a complex production medium were also analyzed quantitatively for a drug of commercial interest. The drug could also be quantified in a variety of mutant-producing strains cultivated in the same medium. The combination of PyMS and ANNs constitutes a novel, rapid, and convenient method for exploitation in strain improvement screening programs. 0 1994John Wiley & Sons, Inc.

Monitoring of complex industrial bioprocesses for metabolite concentrations using modern spectroscopies and machine learning: Application to gibberellic acid production

“…Some of the masses selected (e.g., m/z 55, 58, 67, 71, 85, 86, and 114), including m/z 75, were negatively correlated with GA3 titer, and whereas it would be valid to model on something (a substrate) that is disappearing in a manner proportionate to the analyte of interest, this approach may be more hazardous because the disappearance of a substrate does not guarantee its appearance in a product. Moreover, PyMS has signi®cant disadvantages in that: (1) the in vacuo thermal degradation step means that essentially all information on the structure or identity of the molecules producing the pyrolysate is lost; and (2) molecular reactions in the melt or pyrolysate±pyrolysate interactions in the gas phase can yield new molecular species (Goodacre et al, 1994b). Therefore, it is not sensible to use this destructive technique to attempt to elucidate precise structural information when the target analyte is a very complex, high-molecularweight molecule.…”

“…With recent developments in analytical instrumentation, these requirements are being ful®lled by spectroscopic methods, and the most common are pyrolysis mass spectrometry (PyMS) (Goodacre et al, 1994b;McGovem et al, 1999), Fourier transform infrared spectroscopy (FT-IR) (Mattu et al, 1997;McGovern et al, 1999;Winson et al, 1997) and dispersive Raman microscopy Shaw et al, 1999a). PyMS, FT-IR, and Raman spectroscopies are physicochemical methods that measure predominantly the bond strengths of molecules (PyMS) and the vibrations of bonds within functional groups (FT-IR and Raman) (Ferraro and Nakamoto, 1994;Griths and de Haseth, 1986;Meuzelaar et al, 1982;Schrader, 1995).…”

Quantitative Analysis of Multivariate Data Using Artificial Neural Networks: A Tutorial Review and Applications to the Deconvolution of Pyrolysis Mass Spectra

“…Thus, changes in these single ions can not be used accurately to estimate the percentage of cells of S. aureus mixed with E. coli. In other, less favourable cases, there may be interactions between constituents of the pyrolysate, which would change the mass spectra in a non-linear fashion (54,94,101). The next stage was to look at the relationship between the pyrolysis mass spectra of the binary bacterial mixtures using principal components analysis (PCA).…”

“…Thus, we have been able to follow the production of indole in a number of strains of E. coli grown on media incorporating various amounts of tryptophan (45), to estimate the amount of casamino acids in mixtures with glycogen (47), to quantify the (bio)chemical constituents of complex biochemical binary mixtures of proteins and nucleic acids in glycogen, and to measure the concentrations of tertiary mixtures of bacterial cells (52). More recently, within biotechnology, we have used PyMS and ANNs for the quantitative analysis of recombinant cytochrome bs expression in E. coli (49), and for effecting the rapid screening of the high-level production of desired substances in fermentor broths (54).…”

Validation using sensitivity and target transform factor analyses of neural network models for classifying bacteria from mass spectra

“…(J Am Soc Mass Spectrom 2002, 13, 10 -21) © 2002 American Society for Mass Spectrometry M ass spectra of microorganisms are complex and often require pattern recognition to overcome variations that arise from the measurement and biological factors. Artificial neural networks (ANNs) have been successfully applied to these spectra to characterize and identify microorganisms such as bacteria [1][2][3]. In general, ANNs have been shown to be a rapid and accurate method for classification and discrimination of various microorganisms using pyrolysis mass spectrometry (Py-MS) data.…”