(721e) Real Time and Inline Monitoring of Intracellular Lipid Accumulation Using Industrial Process Microscopes

Microbial lipids become increasingly favorable for industrial production due to their potential to replace petroleum as a main source of fuels and chemicals as well as to provide alternative feedstock for valuable polyunsaturated fatty acids.

The present work describes the development and methodology to apply a microscopy probe for the measurement of the sizes, shapes, colors and concentrations of microorganisms and its intracellular accumulation of lipids. The in-situ measurement of these parameters is a challenge, since the monitoring instrumentation has to be applicable under various process conditions and in different process phases. High particle concentrations, different particles in multiphase systems or changing particle features and velocities are challenging for an inline measurement method.

In a first study, morphological changes and lipid accumulation in fed-batch cultivations of a heterotrophic algae Crypthecodinium cohnii were monitored. Intracellular lipid compartments are identified and correlated to the cell size and finally to the accumulation of the polyunsaturated fatty acid docosahexaenoic acid (DHA). In a second study yeast fermentations of Yarrowia lipolytica with intracellular lipid production are investigated by the same inline microscopy.

Used method

The inline microscopy is bringing the focal plane of the microscopic objective into the sample, with an optical resolution of 0,055 µm/pix. This allows object detection and semantic segmentation of yeast and its intracellular accumulated lipids. The acquired images are analyzed with an artificial neural network. The employed MASK-RCNN is a deep convolutional network. This network learns to segment images in an end to end setting [1]. Automated image analysis classifies different cells types and identifies intracellular accumulated lipids under various growth conditions.

Results

In this study, changes in the cell size distribution of the heterotrophic microalgae C. cohnii were tracked with an offline holographic microscopy and the inline photo-optical microscope. On the basis of the cell size and broadness of the size distribution, the applied inline photo-optical measurements enabled to distinguish between cells in the growth phase with little or no intracellular lipid droplet accumulation and production phase with lipid droplet accumulation. Under conditions of low growth and high fatty acid accumulation, the cell sizes and its distributions were changing accordingly. The correlation between the predicted inline measurements based on the Sauter mean diameter and the measured intracellular DHA content with GC-FID was confirmed. The results obtained by digital holographic and the side scatter values from flow cytometry, which was performed as golden standard method for single cell measurement, were in good agreement with the inline microscopy, which enables this method to photo-optically measure DHA content in real-time.

Yeast is one of the most important species in biotechnology. In this study Baker’s yeast S. cerevisiae and its morphological changes during fermentation is monitored inline. The morphology of yeast cells is altering during maturation, depending on the growth rate and cultivation conditions. Inline microscopy was used to monitor such morphological changes of individual cells directly in the cell suspension. With automated image analysis it was possible to analyze budding and non-budding cells in parallel based on a trained artificial neural network. Deviations between automated and manual counting were considerably low.

The homogeneity among the population during the growth phase increased and at growth retardation, the portion of smaller cells increased due to a reduced bud formation. The maturation state of the yeast cells was determined by the budding index, which Is defined as the ratio between the number of budding cells and the total number of cells. Inline monitoring showed a linear correlation between the budding index (BI) and the growth rate during the batch cultivation. A real-time differentiation of growth activity across all process stages of several batch cultivations in complex media became feasible [2].

References

[1] Kaiming He, Georgia Gkioxari, Piotr Dollár, Ross B. Girshick, 2017, Computer Science, IEEE International Conference on Computer Vision (ICCV)

[2] Marbà-Ardébol, A.-M., Emmerich, J., Muthig, M., Neubauer, P. and Junne, S., Microbial Cell Factories, 2018, 17(1): 73.