Bioreactors are defined as containers used to provide a controlled environment to transfer nutrients and oxygen to cell cultures in adequate concentrations that allow the cell to maintain its primary and secondary metabolic activity. Because plant cells, as well as other micro-organisms, are more sensitive and less stable than chemical compounds, bioreactor designs must be robust enough to provide a greater degree of control over process disturbances and contamination and achieve high productivities, high quality products, and cost effectiveness. The bioreactor design and its optimal operation depend on the determination of the operating conditions giving rise to the required product formation, minimizing the cost of the process
[74]. The most common bioreactor configurations utilized for commercial and large-scale production consist in stirred tank bioreactor (STB), wave stirred bioreactor (WSB), air-lift bioreactor (ALB), and bubble column (BC). The selection of the bioreactor configuration is frequently established by its optimal performance in terms of metabolic activity and kinetics of cell cultures, economic costs, and its flexible operation regarding maintenance of cultures by controlling operational conditions, such as temperature, pH, aseptic, mixing, aeration, and scalability.
Table 3 shows some characteristics, advantages, and disadvantages of these types of bioreactors.
Engineering Aspects in the Plant Cell Suspension Culture
Engineers designing or optimizing bioreactor technologies must both consider the effect of operating conditions on the complex interaction between transport phenomena, thermodynamics, growth kinetics, metabolic activity, and maintenance of plant cell cultures and, based on it, propose methodologies to transfer information observed in flask cultures to larger bioreactor scales. Some operational conditions are critical because they can cause a decrease in biomass, a low PDSM production, or a loss of cell viability. Table 4 shows some CSC that have been successfully scaled from flask cultures to large-scale bioreactors.
Table 4. Comparison of operating conditions used for SCC in flask and bioreactor to produce PDSM. The list of examples makes no claim to be complete.
The scaling up of CSC carried out in a flask culture demands the use of bioreactor engineering to characterize the impact of operating conditions on growth kinetics, cell deactivation, and transport phenomena and, hence, on the metabolic activity and production rates of PDSM. To this end, in what follows, main aspects to be considered during the scaling up of CSC, from the screening of plant cells to the industrial-scale bioreactor design, are mentioned and analyzed.
The screening of a set of plant cells is considered as the first stage during the scaling up of CSC
[48][49]. Screening takes place in shake flasks. In these laboratory bioreactors, hydrodynamic and transport phenomena negatively impact on the growth kinetics, cell viability, metabolic activity, and production rates of PDSM. For instance, in these bioreactors, the production of PDSM involves two-phase systems (liquid culture phase and cell culture phase) neglecting the effect of operating conditions, including the impact of the oxygen concentration, on the microscopic and macroscopic performance of the shake flask. In this context, apparent results regarding cell growth kinetics, cell viability rates, and production rates of PDSM are observed. In these conditions, promising plant cells are identified and selected to be evaluated in larger bioreactor configurations, such as those presented in
Table 3.
The second step accounts for characterization of cell growth kinetics, cell viability rates, metabolic activity, and production rates of PDSM under controlled operating conditions in bench-scale bioreactors with similar configurations to those systems to be implemented at the commercial scale, i.e., bench-scale bioreactors accounting for three phases (liquid–gas–cells) (see
Table 3). Thus, during the analysis of bench-scale systems, the coupling of experimentation with mathematical modeling is essential for stating the basis for the scaling up of CSC
[89][90][91]. Herein, cell growth kinetics and production rates of PDSM are the main response variables to maximize during CSC. It is worth mentioning that their experimental and theoretical characterization makes possible the connection between the microscopic world of the metabolic cell activity and the macroscopic world of the bioreactor performance and, hence, the downstream processing. Besides, the experimental characterization of these cell mechanisms and their analysis using mathematical models lead to the construction of the engineering tool for the scaling up and optimization of the bioreactor configuration, allowing a better understanding of CSC during the production of PDSM. In particular, the use of bench-scale bioreactors allows for identifying and controlling those operating conditions where transport phenomena favor the kinetics of the CSC.
Based on the kinetics, since in CSC it is not possible to develop intrinsic kinetic models, there are two types of models that can be developed in bench-scale bioreactors: extrinsic ones, where transport phenomena are explicitly included during the modeling of the bioreactor; and apparent ones, where transport phenomena resistances impact during the experimentation but they are not considered during the modeling of the bench-scale bioreactors
[90][92][93][94][95][96]. Thus, to determine extrinsic kinetic models, it is recommended to carry out a regime analysis to identify and model those transport phenomena limiting the production of PDSM. Experiments make possible the development of the corresponding model, relating kinetics with macroscopic variables, namely the concentration of substrates and PDSM, cell growth, and cell viability involved during the operation of the bench-scale bioreactor. The kinetic model depends on the quality of the experimental data and it is only reliable for the range of operational conditions utilized during its development. When the kinetic model is based on metabolic steps of the reaction, the mathematical complexity increases but leads to a better physical representation of the CSC during the production of PDSM. Besides, the loss of cell viability caused by operational aspects, i.e., a toxic compound, cell shear stress, or cell sintering, is modeled by empirical expressions whose parameters involve physical meaning
[97], such as the generalized power law equation (GPLE)
[98][99][100]. Finally, the Monod model offers an adequate explanation for the reaction rates of growing cells, but it has no mechanistic basis
[101][102]. Moreover, the Monod model is only applicable when cells are in a metabolic equilibrium, namely when the composition of the macromolecules in the cell remains in a pseudo-steady state during the CSC.
Table 5 presents some kinetic models to describe cell growth rate. It is worth mentioning that, in transient experiments, when the concentration of a substrate or PDSM is brusquely modified, Monod kinetics are not suitable and the kinetic model must account for the cell metabolism
[97][103]. There are, in the literature, several models that have no mechanistic grounds but account for some biological features of the cell growth
[97][104]. These models offer an acceptable description of the cell growth and metabolic activity due to fluctuation in the concentration of substrates and products. In these models, cell mass is divided into compartments, and the rate of formation of each compartment has different stoichiometry and kinetics.
Table 5. Models used to describe kinetics and deactivation in whole cells
[96][99][100][101].
In bench-scale bioreactors, it is experimentally complicated to minimize transport resistances
[98][99][100][105]. In the fluid bulk, concentration, temperature, or radiative gradients can be present. Hydrodynamics impact on mass and heat transfer mechanisms from the gas phase to the liquid phase and from the liquid phase to the cell phase. Moreover, cell growth can impact on mass and heat transfer mechanisms. Although complicated, a proper kinetic analysis must account for the effect of fluid dynamics on transport phenomena and, hence, on cell growth, cell viability, and metabolic activity.
During the screening at the laboratory bioreactors or during the operation of the bench-scale bioreactor, the response surface methodology (RSM) is a potential tool to guide experimental designs. RSM leads to the following advantages
[106][107][108][109][110]:
- (1)
-
It defines an establishment of the relationship between responses (yield, cell viability, oxygen concentration, etc.) and control operating conditions (temperature, pressure, initial concentration, power input, agitation rate, etc.).
- (2)
-
It predicts the effect of control operating condition on responses.
- (3)
-
It gives inferences on the significance of the operating conditions on the performance of the reactor.
- (4)
-
It allows the determination of the operating window where the bioreactor meets its best performance.
On the above end, RSM couples experimental designs, and mathematical and statistical methods
[111][112]. Firstly, an experimental design is proposed; the evaluation of this experimental design constitutes the so-called response surface design (RSD). The suitability of the RSD depends on its orthogonally, ratability, and uniform precision
[112]. Secondly, the empirical model is then developed; it is approximated by a polynomial equation that accounts for elements that consist of powers and cross-product powers, constant coefficients referred to as parameters, and a random experimental error. Albeit empirical, first-degree and second-degree polynomial equations are usually used to fit observations and carry out the optimization. To this end, every model and its reliability depends on the RSD, i.e., first-order designs are used to fit observations with the first-degree models, and observations out of second-order designs are fitted with second-degree models
[111][112][113]. The most common first-order designs are 2k factorial, Plackett–Burman, and simplex designs, while the most common second-order designs are 3k factorial, central composite, and the Box–Behnken designs. Note that the choice of a proper RSD is essential since the quality of prediction, as measured by the size of the prediction variance, depends on it; thus, the lower the variance, the better the fit of the responses. On this basis, a single RSD is not able to satisfy all criteria, but it is considered as robust if it meets the assumptions related to the model and the error distribution
[111][112]. Finally, the assessing of the results uses both statistical tests, i.e., F-value, t-value, and confidence interval, and graphical tests, i.e., variance dispersion graphs, fraction of design space plots, and quantile plots. Graphical methods
[108][109] based on quantile dispersions have also been used to compare experimental designs for estimating variance components in an analysis of variance (ANOVA) situation. RSM can lead to the identification of the operational window where CSC presents its higher yields to PDSM, which, in turn, will be essential in the conceptual design and scaling up of the bioreactor configuration.
Because of the advent of computation in the last years, the bioreactor design not only depends on empirical, but also deterministic approaches, which allows the proper determination of hydraulics, fluid dynamics, mass transport, heat transfer, radiative transfer, and kinetics from different bioreactor configurations at various scales. This information is transferred to design and scale up the industrial bioreactor. The design of this reactor strongly depends on the development of a model coupling kinetics and transport phenomena at both the cell and bioreactor level, including the fluid and the gas phase. This is, however, a complex task, since it needs experiments and mathematical solutions that are not trivial. It is worth stressing that, during the construction of this model, fluid dynamics are yet the bottleneck during the scaling up of a bioreactor configuration because of their impact on transport phenomena, kinetics, and, hence, on the global production of PDSM.
Based on the above, a model accounting for kinetic, deactivation, and all transport mechanisms should be developed from the laboratory to the bench scale. This model should be constructed following a framework based on computational fluid dynamics (CFD). The model needs to be validated at the bench scale before using it to design the industrial bioreactor. The preliminary dimensions of the reactor need to be obtained from the utilization of the practical know-how reported in the literature or experimental and modeling results obtained at the bench scale. It will make the scaling up process more efficient and reliable. Developing a model for the use of CFD allows the consideration of fluid dynamics along with its effect on transport phenomena, which leads to obtaining operating conditions where mixing, hydrodynamics, and transport phenomena are improved without affecting the operating cost of the process. A criterion when designing the industrial-scale bioreactor is to achieve a compromise between operating expenses and yield of the PDSM. At the end of the scaling-up process, the experimentation and investment cost as that compared using an empirical or heuristic approach will be significantly minimized.
In addition to the aforementioned, the scaling up of CSC becomes more challenging when observing how operating conditions impact on the production of PDSM. Operating conditions influence in different scenarios and magnitudes the performance of cell cultures during the production of PDSM, from the supply of nutrients (oxygen, light, ionic strength, pH) to the implementation of mechanical and pneumatic work to keep the process operating in optimal conditions. In further sections, a discussion about the main operating variables in bioreactors and their effect on the performance of cell culture will be provided.