experiments and (2) the interpretation of the data from scale-down experiments.
These two branches of application are presented in the following sections, with
respect to the state of the art, as reported in current literature and future perspectives.
Model-Based Design of Scale-Down Experiments
The concept of scale-down experiments developed in the past 3 decades usually
involves creating one type of stress (e.g., dissolved oxygen limitation, excess
substrate, excess metabolite concentrations, and perturbations in pH) in the scaledown simulator. However, considering an actual larger-scale bioreactor, concentration gradients arise from mixing effects in a 3-dimensional space, combined with the
uptake of substrates and release of metabolites by cells. Moreover, the type of
gradients are always coupled and may co-exist (e.g., pH-oxygen-substrate gradients)
in the larger bioreactor. Therefore, at best, the scale-down simulators are only a gross
estimate of the actual environments in large bioreactors. Additionally, the fraction of
cells exposed to a given gradient in a large-scale bioreactor has been variable in the
definition of the scale-down model. In multi-compartment scale-down simulators,
this has been in the range of 10%–30%, whereas in pulse-based single-compartment
simulators, the total population is subjected to the stresses, without population
subgroups. Notwithstanding these challenges, important physiological responses
have been reported by researchers using these physical approximations of larger
bioreactors. In this light, experimental set-ups for scale-down studies can improve
greatly when they are combined with the ideas of digitalization in the industry
4.0 era.
The scale-down model, defined on the blueprint of the environmental heterogeneity of the larger scale, will contain process specifications such as the gradient
profiles, zone definitions with defined boundaries, residence time distributions,
magnitudes of gradients, frequencies and other experimental inputs that should be
implemented in the simulator. With such a scale-down model in hand, the scaledown experiment can be designed on a computer, and results of the design sent to
intelligent equipment (pumps, pipettes, agitators, liquid handlers, etc.) to control a
small-scale bioreactor to mimic the blueprint of the larger-scale bioreactor. That is,
mixed-gradient zones, containing excess substrate with acidic pH, or limited oxygen
with high hydrodynamic stress, or any desired combination of gradients can be
created with the model, and the size of the zone and its frequency/duration can be
varied randomly to more closely resemble actual gradient dynamics in the larger
scale. A step in this direction is the work of Anane et al. [29] who used a mechanistic
model of E. coli [101], in combination with a mechanistic description of the gradient
profiles of a multi-compartment scale-down bioreactor [1] to calculate glucose
pulses in high-throughput scale-down experiments. The outputs of the two models
were integrated into the operation scheme of the high-throughput system to reproduce heterogenous glucose conditions in minibioreactors. The calculated glucose
pulses were commensurate with the physiology of the strain, as the pulse sizes were
derived from uptake capacities and physiological limits of the strain, as well as
18
P. Neubauer et al.
These two branches of application are presented in the following sections, with
respect to the state of the art, as reported in current literature and future perspectives.
Model-Based Design of Scale-Down Experiments
The concept of scale-down experiments developed in the past 3 decades usually
involves creating one type of stress (e.g., dissolved oxygen limitation, excess
substrate, excess metabolite concentrations, and perturbations in pH) in the scaledown simulator. However, considering an actual larger-scale bioreactor, concentration gradients arise from mixing effects in a 3-dimensional space, combined with the
uptake of substrates and release of metabolites by cells. Moreover, the type of
gradients are always coupled and may co-exist (e.g., pH-oxygen-substrate gradients)
in the larger bioreactor. Therefore, at best, the scale-down simulators are only a gross
estimate of the actual environments in large bioreactors. Additionally, the fraction of
cells exposed to a given gradient in a large-scale bioreactor has been variable in the
definition of the scale-down model. In multi-compartment scale-down simulators,
this has been in the range of 10%–30%, whereas in pulse-based single-compartment
simulators, the total population is subjected to the stresses, without population
subgroups. Notwithstanding these challenges, important physiological responses
have been reported by researchers using these physical approximations of larger
bioreactors. In this light, experimental set-ups for scale-down studies can improve
greatly when they are combined with the ideas of digitalization in the industry
4.0 era.
The scale-down model, defined on the blueprint of the environmental heterogeneity of the larger scale, will contain process specifications such as the gradient
profiles, zone definitions with defined boundaries, residence time distributions,
magnitudes of gradients, frequencies and other experimental inputs that should be
implemented in the simulator. With such a scale-down model in hand, the scaledown experiment can be designed on a computer, and results of the design sent to
intelligent equipment (pumps, pipettes, agitators, liquid handlers, etc.) to control a
small-scale bioreactor to mimic the blueprint of the larger-scale bioreactor. That is,
mixed-gradient zones, containing excess substrate with acidic pH, or limited oxygen
with high hydrodynamic stress, or any desired combination of gradients can be
created with the model, and the size of the zone and its frequency/duration can be
varied randomly to more closely resemble actual gradient dynamics in the larger
scale. A step in this direction is the work of Anane et al. [29] who used a mechanistic
model of E. coli [101], in combination with a mechanistic description of the gradient
profiles of a multi-compartment scale-down bioreactor [1] to calculate glucose
pulses in high-throughput scale-down experiments. The outputs of the two models
were integrated into the operation scheme of the high-throughput system to reproduce heterogenous glucose conditions in minibioreactors. The calculated glucose
pulses were commensurate with the physiology of the strain, as the pulse sizes were
derived from uptake capacities and physiological limits of the strain, as well as
18
P. Neubauer et al.