4.2.2 High-Throughput Execution of Scale-Down Experiments
in Parallel Cultivation Systems
High-throughput experiments in parallel cultivation platforms have become common in bioprocess development laboratories. In the past decade, there has been an
exponential increase in the adoption of these systems for early bioprocess development [45, 110, 111]. At the same time, due to Quality-by-Design (QbD) guidelines,
there has been an increasing demand to fully characterize bioprocesses at the
development phase, to forestall unforeseen consequences of the final process,
upon scale-up [112, 113]. This requirement demands that all conditions, including
actual large-scale process conditions are considered and tested in the early development phases of the process. Therefore, the question of whether cultivations in
minibioreactors are adaptable to mimic concentration gradients and the heterogeneous environments that exist in large-scale bioreactors has become very important,
and should be addressed.
A few studies conducted in high-throughput cultivation systems that consider the
heterogeneous conditions of larger bioreactors are reported in the literature. As
described above, Janakiraman et al. [107] matched the volumetric aeration rates
(vvm) between parallel Ambr15
® cultivations of CHO cells and a 15,000 L
production-scale bioreactor. They used this criterion to mimic the carbon dioxide
profile of the production bioreactor in the minibioreactor cultivations, which led to
similar productivity and product quality profiles in both the 15 ml bioreactors and the
15,000 L scale. In another study, Velez-Suberbie et al. [114] used the power per unit
volume (P/V) as a scale-down criterion to compare Ambr15 cultivations of E. coli
with 20 L bioreactor cultivations [114]. Perhaps the most comprehensive work in
this regard was reported by Anane et al. [29], who used model-calculated glucose
pulses to induce both dissolved oxygen and glucose gradient zones in 15 ml parallel
minibioreactors. A key aspect of their work was the use of robotic liquid handling
stations and mechanistic models in the operation of the scale-down set-up. These
smart equipment were interphased with the minibioreactors, such that model outputs
describing specific gradient conditions could be implemented in selected
minibioreactors by the robotic system. Their results in E. coli fermentation development showed significant accumulation of non-conventional amino acids in the
recombinant protein product, as well as accumulation of acetate in the scale-down
cultivations, when compared to cultivations under homogeneous conditions.
The results of scale-down cultivations as performed in high-throughput
minibioreactor systems so far show that it is possible to mimic large-scale environmental conditions in miniaturized bioreactors. Particularly, the physiological
responses of both E. coli and CHO cells to the induced heterogeneous conditions
in minibioreactors, as discussed above, is a proof of concept that gradient profiles
that are relevant in industrial-scale cultivations can be reproduced in milliliter scale
for scale-down studies. However, the adoption of enabling technological methods,
such as robotic liquid handling stations and mechanistic modeling is fundamental for
the successful operation of such minibioreactor facilities as scale-down platforms.
Potential of Integrating Model-Based Design of Experiments Approaches and. . .
21
in Parallel Cultivation Systems
High-throughput experiments in parallel cultivation platforms have become common in bioprocess development laboratories. In the past decade, there has been an
exponential increase in the adoption of these systems for early bioprocess development [45, 110, 111]. At the same time, due to Quality-by-Design (QbD) guidelines,
there has been an increasing demand to fully characterize bioprocesses at the
development phase, to forestall unforeseen consequences of the final process,
upon scale-up [112, 113]. This requirement demands that all conditions, including
actual large-scale process conditions are considered and tested in the early development phases of the process. Therefore, the question of whether cultivations in
minibioreactors are adaptable to mimic concentration gradients and the heterogeneous environments that exist in large-scale bioreactors has become very important,
and should be addressed.
A few studies conducted in high-throughput cultivation systems that consider the
heterogeneous conditions of larger bioreactors are reported in the literature. As
described above, Janakiraman et al. [107] matched the volumetric aeration rates
(vvm) between parallel Ambr15
® cultivations of CHO cells and a 15,000 L
production-scale bioreactor. They used this criterion to mimic the carbon dioxide
profile of the production bioreactor in the minibioreactor cultivations, which led to
similar productivity and product quality profiles in both the 15 ml bioreactors and the
15,000 L scale. In another study, Velez-Suberbie et al. [114] used the power per unit
volume (P/V) as a scale-down criterion to compare Ambr15 cultivations of E. coli
with 20 L bioreactor cultivations [114]. Perhaps the most comprehensive work in
this regard was reported by Anane et al. [29], who used model-calculated glucose
pulses to induce both dissolved oxygen and glucose gradient zones in 15 ml parallel
minibioreactors. A key aspect of their work was the use of robotic liquid handling
stations and mechanistic models in the operation of the scale-down set-up. These
smart equipment were interphased with the minibioreactors, such that model outputs
describing specific gradient conditions could be implemented in selected
minibioreactors by the robotic system. Their results in E. coli fermentation development showed significant accumulation of non-conventional amino acids in the
recombinant protein product, as well as accumulation of acetate in the scale-down
cultivations, when compared to cultivations under homogeneous conditions.
The results of scale-down cultivations as performed in high-throughput
minibioreactor systems so far show that it is possible to mimic large-scale environmental conditions in miniaturized bioreactors. Particularly, the physiological
responses of both E. coli and CHO cells to the induced heterogeneous conditions
in minibioreactors, as discussed above, is a proof of concept that gradient profiles
that are relevant in industrial-scale cultivations can be reproduced in milliliter scale
for scale-down studies. However, the adoption of enabling technological methods,
such as robotic liquid handling stations and mechanistic modeling is fundamental for
the successful operation of such minibioreactor facilities as scale-down platforms.
Potential of Integrating Model-Based Design of Experiments Approaches and. . .
21