Flow Cytometric fingerprinting microbial quality

01 January 2014 → 31 December 2017
Regional and community funding: IWT/VLAIO
Research disciplines
  • Engineering and technology
    • Environmental microorganism biotechnology
flow cytometric fingerprinting microbial communities
Project description

Bacteria are ubiquitous on earth and typically form complex microbial communities of coexisting genotypes and phenotypes. These communities are important for our modern society as many different industrial applications are facilitated by bacteria. These applications range from the production of fine chemicals, pharmaceuticals, and food to wastewater treatment, bioremediation, bioaugmentation, and bioleaching. In the future, the importance of bioprocesses is likely to increase due to the technological, economical, and ecological benefits of microbiologically-driven techniques. Microbial communities are the result of the complex interactions among bacteria and between bacteria and their environment. They are therefore constantly in flux and react sensitively on changed conditions. As a consequence, microbial communities are also good indicators of how processes operate. For example, water quality is monitored based on the absence or presence of certain indicator organisms. To improve bioprocesses or to monitor microbial communities for quality purposes, techniques for community characterization are necessary. In this doctoral research we explored how flow cytometry could be used for this purpose. Existing fingerprinting methods were improved and the sensitivity of this technique was tested for the taxonomic and phenotypic characterization of microbial communities. Since flow cytometry relies on staining and since staining can decrease the resolution and speed of the method, Raman spectroscopy, a label-free alternative, was also investigated for community characterization. Bacteria are very small and fluorescent dyes staining a specific feature of microbial cells are necessary to detect them. The type of dye dramatically impacts the cytometric fingerprints as only the stained features are assessed. In Chapter 2 different multicolor staining protocols were compared to increase the amount of physiological information per cell and consequently to improve the fingerprints. We found that not only the number of dyes are important but also the stability of the dyes. Results showed that SG and SGPI are the most stable dyes and therefore favored for cytometric fingerprinting. In Chapter 3, the cytometric fingerprinting toolbox was improved with a similarity based approach. To test the sensitivity of this unsupervised approach, 29 Lactobacillus species and strains were compared. Results showed that 27 out of 29 species and strains could be discriminated with SG or SGPI staining. The sensitivity of the method was found to be between 1% and 5% of change in the fingerprints, depending on the overlap between the bacteria fingerprints. The method is also reproducible but a standardized growth protocol is necessary. To illustrate the possible impact of the different growth stages on the fingerprints, a comparison of the fingerprints of a batch culture were compared. The results showed that the fingerprints did change sufficiently for the fingerprinting algorithm. To further investigate the impact of the microbial growth stage on the cytometric fingerprints, an E. coli batch fermentation was monitored over time (Chapter 4). Results showed that the fingerprints changed due to a phenotypic switch caused by substrate depletion. This switch occurred before conventional methods such as exhaust-gas analysis could detect the substrate depletion. As a consequence, we conclude that flow cytometric fingerprinting can be used as a monitoring tool for bioprocess operations. A contrast analysis shows that the sensitivity of the method is related to the sensitivity of the nucleic acid dyes to detect small changes in the nucleic acid composition. Because labeling bacteria with fluorescent dyes reduces the amount of physiological information comprising the fingerprints, an alternative label-free method such as Raman spectroscopy was evaluated (Chapter 5). Several data processing pipelines were compared and results showed that phenotypic differences could be detected by both supervised and unsupervised methods with high accuracy. Moreover, we developed and tested a new algorithm which quantifies bacteria species in a community based on the average spectrum of the community. Quantification with this algorithm is possible but different results were found for different bacteria mixes. To compare the sensitivity and the complementarity of flow cytometry and Raman spectroscopy, an experiment with microcosms was set up (Chapter 6). When two bacteria species were cocultured, they both changed their phenotypes though in different ways. This could be established with both flow cytometry and Raman spectroscopy but the combination of the two methods confirmed the observations. This also illustrates that the two methods are complementary. Microbiology is an important aspect of water quality. To reduce health risks, bacteria in drinking water are mitigated by disinfection but it is impossible to remove all bacteria. Therefore a good monitoring of the microbial quality, especially in drinking water distribution systems is necessary. In Chapter 7, we compared the planktonic microbial communities with the microbial communities in the biofilm which are putatively the most important fraction of bacteria in DWDS. Results showed that both the type of materials in contact with the water and the origin and treatment of the water influenced the cytometric fingerprints and hence the microbial community. No relationship could be established between the bulk and biofilm fingerprints. Furthermore, we tested if an Enterobacter amnigenus, a drinking water contaminant, could colonize biofilms and subsequently contaminate drinking water. Results showed that the Enterobacter could do this, regardless of the type of material or of a pre existing biofilm. When Enterobacter was incubated with sterilized water, the bacteria grew better, probably because of necrotrophic growth or because of a lack of competition. In Chapter 8, we tested the sensitivity of cytometric community fingerprinting by adding elements which could be used by the aquatic microbial community to grow. We showed that the fingerprints are mainly sensitive to (re)growth of the microbial community. Finally, we installed an online flow cytometer in a full-scale water treatment plant to illustrate the potential of the method as early-warning system and as water quality monitoring tool. Water after different filtration steps was monitored and we showed that after every filtration unit (UF, RO), bacteria could be detected. The cell density was related to the type of filter, the age, and fluctuated in function of the operation of the treatment plant. To conclude, in this doctoral research we showed that flow cytometric fingerprinting can be used for community characterization and that it correlates with genotypic and phenotypic changes in the microbial community. The speed and automation of the method makes it an ideal candidate for industrial monitoring of bioreactors or water distribution systems. Raman spectroscopy is an alternative approach which is potentially more powerful as it is can easily discriminate among genotypes and phenotypes. Yet, Raman spectroscopy cannot be automated and more research is necessary to adapt it for industrial applications. However, the possibility to both characterize the species composition and the phenotypic composition of a community in less than a few hours makes it a promising tool for research.