Monitoring of different metabolic states of Sacchoromyces cerevisiae via NIR spectroscopy at low biomass concentration

Abstract

Microorganisms, such as the yeast Saccharomyces cerevisiae, are used in fermentations in biotechnology to produce a wide variety of products - from beer to medicines. These industrial bioprocesses are subject to strict legal guidelines that require monitoring of the quality and quantity of the individual products. These controls are followed up with samples from the process. However, on-line sampling is a loss of biomass and/or product, there is a risk of contamination of the process, and the preparation of the samples is very time-consuming. On-line measurements via spectroscopic methods can provide a lot of data in real time without touching or disturbing the contents of the bioreactor.With the help of NIR data from fermentations from 2016 and 2019, analyses of the metabolism of the yeast were made at a biomass concentration of less than or equal 2.5g/L.The fermentations were exposed to dierent process conditions. On the one hand, standard fermentations were carried out in which the best possible conditions for the yeasts were set. In the following analyses, these served as target values for an ideally running process. Furthermore, various disturbance processes were recorded in which the environment of the yeast (the medium) was deviated into a sub-optimal range. Too much or too little base was added so that the pH affected the metabolism, or the oxygen supply was interrupted to force a change to the anaerobic metabolism. These disturbances should simulate realistic failures of individual parameters in the industry. The results obtained were processed using the R software and analysed using PCA. Off-line samples served as controls for the present metabolism.The results from the spectral data showed that a statement about the respective metabolism can be made. The standard fermentations can be distinguished from the disturbed fermentations as well as the disturbed processes among each other. A comparison with data recorded in 2016 gave similar results with regard to the physiological data. The spectral data from 2016 and 2019 gave different results, but there was insufficient data to give an exact reason for the origin of this deviation.If the results of the 2019 data are reproducible through further fermentations, a model for the detection of deviations can be created. Furthermore, it can be determined - by increased time sampling - how early a difference from the optimum can be recognized. Depending on the result of the trend over time, an earlier intervention would be possible if there was a signicant deviation in the suboptimal running process.