Establishment of dynamic microfluidic single-cell cultivation for microbial robustness analysis

Bioprocesses play a central role in advancing towards a biobased and sustainable economy. To effectively compete with established chemical processes, these bioprocesses must be developed with the final aim in mind, which includes the strategic design of microbial production hosts to operate reliably under production conditions. These conditions often involve gradients in process parameters that significantly influence productivity, population heterogeneity and yield. Microfluidic single-cell cultivation (MSCC) enables the cultivation of single cells and microcolonies over time in defined and dynamic environments. With live-cell imaging as analysis technique, growth, fluorescently tagged cellular functions, population heterogeneity, cell history and lineage can be observed. Yet applications of MSCC for bioprocess development are limited, although its ability to modify the cultivation environment rapidly and accurately would allow for the exploration of the effects of process parameter gradients on microbial production hosts. This thesis focussed on evaluating, if and how dynamic MSCC (dMSCC) could be used as scale-down tool and what unique knowledge can be gained from its application for bioprocess development. Initially, the conceptual capability of dMSCC to replicate bioprocess-relevant environmental dynamics was assessed. Microbial lifelines were identified as translational tool between different cultivation systems. Lifelines enable to take a single cell’s perspective in the bioreactor, which experiences e.g. by-product accumulation in the cultivation medium, but also existing process parameter gradients as a series of environmental changes in time. From a cell’s perspective, these changes have a timescale of mere seconds to hours, with varying occurrences, frequencies and amplitudes. For aerobic fermentations, fluctuations in carbon sources and dissolved oxygen are particularly significant. In dMSCC, any concentration of a soluble, non-gaseous process parameter, such as substrates, can be modulated within 1.6 s from a cell’s perspective, leveraging the microfluidic characteristics and principles of laminar flow and hydrodynamic focussing. Conceptually, dMSCC therefore covers both a sufficient timescale and magnitude for these soluble process parameters to be applied as scale-down device, yet fast gas dynamics remain challenging due to the gas permeability of the fabrication material polydimethylsiloxane. This thesis showcases further development and characterisation of the dMSCC principle to improve experimental throughput as well as knowledge on applicable environmental modulations. Robustness quantification of microbial strains was identified as an approach that could benefit from the application of dMSCC. Robustness is defined as a system’s ability to maintain its function in the face of perturbations. With the microbial production host being the heart of every bioprocess, the robustness of growth, production behaviour or yield of a microbe when faced with perturbations such as gradients, toxic substrate contaminants or cellular aging grants benefits for a bioprocess. With dMSCC it was possible to expand the existing concept for robustness quantification to assess the robustness of functions over time and within populations. To establish this methodology, the yeast *S. cerevisiae* CEN.PK 113-7D equipped with a fluorescent ATP biosensor was grown in glucose dynamics (starvation vs. excess) with varying frequencies. The results indicated that while fast dynamics (1.5 to 6 minutes) yielded the best robustness in growth, ATP levels were least robust to rapid changes every 1.5 minutes. In subsequent experiments, the robustness of several functions in the three yeast strains CEN.PK 113-7D, EthanolRed and PE2 against glucose (0.01 vs. 50 g/L) and pH (3.5 vs. 5) dynamics was compared. EthanolRed displayed the highest robustness over time and across the population for multiple functions when exposed to substrate oscillations. Dynamics in pH had less influence than glucose dynamics on the performance of robustness on all strains, both over time or within the population. In conclusion, the combination of robustness quantification and dMSCC has the potential to provide insights that can benefit bioprocess development: First, dynamics that alter robustness and production behaviour can be precisely identified, motivating bioprocess engineering to circumvent these dynamics. Furthermore, testing for robustness and performance of strains against inevitable large-scale dynamics could be included already during strain selection and development. Even with limitations in analytical techniques and the representation of gaseous dynamics, the findings of this thesis demonstrate that dMSCC is a promising scale-down tool, offering unique advantages through rapid, metabolism-independent environmental dynamics and single-cell resolution. Future research should focus on validating the transferability of findings between dMSCC and large-scale bioreactors, using the cell’s perspective to guide experimental design.