Hybrid modeling as a QbD/PAT tool in process development: an industrial <Emphasis Type="Italic">E. coli</Emphasis> case study

Process understanding is emphasized in the process analytical technology initiative and the quality by design paradigm to be essential for manufacturing of biopharmaceutical products with consistent high quality. A typical approach to developing a process understanding is applying a combination of design of experiments with statistical data analysis. Hybrid semi-parametric modeling is investigated as an alternative method to pure statistical data analysis. The hybrid model framework provides flexibility to select model complexity based on available data and knowledge. Here, a parametric dynamic bioreactor model is integrated with a nonparametric artificial neural network that describes biomass and product formation rates as function of varied fed-batch fermentation conditions for high cell density heterologous protein production with E. coli. Our model can accurately describe biomass growth and product formation across variations in induction temperature, pH and feed rates. The model indicates that while product expression rate is a function of early induction phase conditions, it is negatively impacted as productivity increases. This could correspond with physiological changes due to cytoplasmic product accumulation. Due to the dynamic nature of the model, rational process timing decisions can be made and the impact of temporal variations in process parameters on product formation and process performance can be assessed, which is central for process understanding.     

Process design for recombinant protein production based on the promoter,P<Subscript><Emphasis Type="Italic">malK</Emphasis></Subscript>

P_malK is induced through activation of MalT, by the formation of maltotriose and cyclic adenosine monophosphate (cAMP). The possibility to influence endogenous inducer levels is used to vary the production rates in specifically designed production protocols. Induction based on a batch process protocol on maltose gives low production rates, as the result of a lack of cAMP, which is shown to be of major importance to fully induce this promoter. Two mechanisms are thus used to influence the levels of maltotriose and/or cAMP formation: (1) catabolite derepression achieved from low glucose concentration and (2) catabolite derepression/inducer exclusion from diauxic growth on glucose/maltose. Fed-batch processes based on limited amounts of glucose result in product accumulation of up to 10% of the total protein. Depending on the feed of limiting glucose, different production profiles are developed. The initial increase in the production rate is due to maltotriose formation from endogenous glycogen degradation while, later in the process, production can be further supported by elevated levels of cAMP, provided the feed rate is sufficiently low. The introduction of maltose after a preceding fed-batch process on glucose can be efficiently used to produce maltotriose in combination with cAMP formation in the event of catabolite derepression. This leads to higher production rates and a further increase in product accumulation of up to 30% of the total protein. The diauxic growth phase resulting from the shift in carbon source can be shortened and even avoided by the design of the preceding feed-rate of glucose. It is postulated that proper design of the inoculum and initial phases of production can reduce basal levels of product formation. With this promoter, the production rate can be as high as 65 units mg^–1 h^–1 and the time to reach a maximal production rate can be designed to take up to 8 h. Furthermore, the duration of the production rate can be as long as 7 h.     

Cell engineering of <Emphasis Type="Italic">Escherichia coli</Emphasis> allows high cell density accumulation without fed-batch process control

A set of mutations in the phosphoenolpyruvate:carbohydrate phosphotransferase system (PTS) was used to create Escherichia coli strains with a reduced uptake rate of glucose. This allows a growth restriction, which is controlled on cellular rather than reactor level, which is typical of the fed-batch cultivation concept. Batch growth of the engineered strains resulted in cell accumulation profiles corresponding to a growth rate of 0.78, 0.38 and 0.25 h⁻¹, respectively. The performance of the mutants in batch cultivation was compared to fed-batch cultivation of the wild type cell using restricted glucose feed to arrive at the corresponding growth profiles. Results show that the acetate production, oxygen consumption and product formation were similar, when a recombinant product was induced from the lacUV5 promoter. Ten times more cells could be produced in batch cultivation using the mutants without the growth detrimental production of acetic acid. This allows high cell density production without the establishment of elaborate fed-batch control equipment. The technique is suggested as a versatile tool in high throughput multiparallel protein production but also for increasing the number of experiments performed during process development while keeping conditions similar to the large-scale fed-batch performance.     

Dynamic mathematical models of batch experiments and fed-batch cultures for cyclic adenosine monophosphate production by <Emphasis Type="Italic">Arthrobacter</Emphasis> A302

A glucose utilizing strain, Arthrobacter A302 was used for cyclic adenosine monophosphate (cAMP) production in batch modes. The non-structured model in a 5 l stirred tank bioreactor for understanding, controlling, and optimizing the fermentation process was proposed using the logistic equation for microbial growth, the Luedeking-Piret equation for product formation and Luedeking-Piret-like equation for substrate uptake, respectively. The production of cAMP was a mixed-growth-associated pattern. Based on model prediction, a comparison of calculated value using the parameters evaluated above with another experimental data in 30 l bioreactor was used to test the model. The results predicted from the model were in good agreement with the experimental observations in 30 l bioreactor, which demonstrated that the model might be useful for the development and optimization of production of cAMP in industrial scale. Based on estimated kinetic parameters, three different fed-batch modes, constant rate and intermittent (once and repeated), were adopted in order to obtain more cAMP accumulation. Furthermore, the final production of cAMP reached 11.24 g l⁻¹ after 72 h incubation using three stages feeding strategy. In particular, the cAMP productivity (0.156 g l⁻¹ h⁻¹) was successfully improved by 22.83, 11.43 and 9.86%, respectively, compared with the modes of the batch, constant rate fed-batch and intermittent fed-batch once.     

Dynamic optimization of hybridoma growth in a fed-batch bioreactor

This study addressed the problem of maximizing cell mass and monoclonal antibody production from a fed-batch hybridoma cell culture. We hypothesized that inaccuracies in the process model limited the mathematical optimization. On the basis of shaker flask data, we established a simple phenomenological model with cell mass and lactate production as the controlled variables. We then formulated an optimal control algorithm, which calculated the process-model mismatch at each sampling time, updated the model parameters, and re-optimized the substrate concentrations dynamically throughout the time course of the batch. Manipulated variables were feed rates of glucose and glutamine. Dynamic parameter adjustment was done using a fuzzy logic technique, while a heuristic random optimizer (HRO) optimized the feed rates. The parameters selected for updating were specific growth rate and the yield coefficient of lactate from glucose. These were chosen by a sensitivity analysis. The cell mass produced using dynamic optimization was compared to the cell mass produced for an unoptimized case, and for a one-time optimization at the beginning of the batch. Substantial improvements in reactor productivity resulted from dynamic re-optimization and parameter adjustment. We demonstrated first that a single offline optimization of substrate concentration at the start of the batch significantly increased the yield of cell mass by 27% over an unoptimized fermentation. Periodic optimization online increased yield of cell mass per batch by 44% over the single offline optimization. Concomitantly, the yield of monoclonal antibody increased by 31% over the off-line optimization case. For batch and fed-batch processes, this appears to be a suitable arrangement to account for inaccuracies in process models. This suggests that implementation of advanced yet inexpensive techniques can improve performance of fed-batch reactors employed in hybridoma cell culture.     

Morphologically structured model for antitumoral retamycin production during batch and fed-batch cultivations of Streptomyces olindensis

A morphologically structured model is proposed to describe trends in biomass growth, substrate consumption, and antitumoral retamycin production during batch and fed-batch cultivations of Streptomyces olindensis. Filamentous biomass is structured into three morphological compartments (apical, subapical, and hyphal), and the production of retamycin, a secondary metabolite, is assumed to take place in the subapical cell compartment. Model accounts for the effect of glucose as well as complex nitrogen source on both the biomass growth and retamycin production. Laboratory data from bench-scale batch and fed-batch fermentations were used to estimate some model parameters by nonlinear regression. The predictive capability of the model was then tested for additional fed-batch and continuous experiments not used in the previous fitting procedure. The model predictions show fair agreement to the experimental data. The proposed model can be useful for further studies on process optimization and control.     

An unstructured kinetic model of macromolecular metabolism in batch and fed-batch cultures of hybridoma cells producing monoclonal antibody

Growth profiles of the batch and fed-batch culture of hybridoma cells producing monoclonal antibody were simulated using an unstructured model. The model describes the production of cellular macromolecules and monoclonal antibody, the metabolism of glucose and glutamine with the production of lactate and ammonia, and the profiles of cell growth in batch and fed-batch culture. Equations describing the cells arrested in G1 phase [T.I. Linardos, N. Kalogerakis, L.A. Behie, Biotechnol. Bioeng. 40 (1992) 359–368; E. Suzuki, D.F. Ollis, Biotechnol. Bioeng. 34 (1989) 1398–1402] were included in this model to describe the increase of the specific antibody productivity in the near-zero specific growth rate, which was observed in the recent experiments in fed-batch cultures of this study and the semi-continuous culture of hybridoma cells [S. Reuveny, D. Velez, L. Miller, J.D. Macmillan, J. Immnol. Methods 86 (1986) 61–69]. This model predicted the increase of specific antibody production rate and the decline of the specific production rate of cellular macromolecules such as DNA, RNA, protein, and polysaccharide in the late exponential and decline phase of batch culture and at lower specific growth rates in the fed-batch culture.     

A rational approach to improving productivity in recombinant Pichia pastoris fermentation

A Mut(S) Pichia pastoris strain that had been genetically modified to produce and secrete sea raven antifreeze protein was used as a model system to demonstrate the implementation of a rational, model-based approach to improve process productivity. A set of glycerol/methanol mixed-feed continuous stirred-tank reactor (CSTR) experiments was performed at the 5-L scale to characterize the relationship between the specific growth rate and the cell yield on methanol, the specific methanol consumption rate, the specific recombinant protein formation rate, and the productivity based on secreted protein levels. The range of dilution rates studied was 0. 01 to 0.10 h(-1), and the residual methanol concentration was kept constant at approximately 2 g/L (below the inhibitory level). With the assumption that the cell yield on glycerol was constant, the cell yield on methanol increased from approximately 0.5 to 1.5 over the range studied. A maximum specific methanol consumption rate of 20 mg/g. h was achieved at a dilution rate of 0.06 h(-1). The specific product formation rate and the volumetric productivity based on product continued to increase over the range of dilution rates studied, and the maximum values were 0.06 mg/g. h and 1.7 mg/L. h, respectively. Therefore, no evidence of repression by glycerol was observed over this range, and operating at the highest dilution rate studied maximized productivity. Fed-batch mass balance equations, based on Monod-type kinetics and parameters derived from data collected during the CSTR work, were then used to predict cell growth and recombinant protein production and to develop an exponential feeding strategy using two carbon sources. Two exponential fed-batch fermentations were conducted according to the predicted feeding strategy at specific growth rates of 0.03 h(-1) and 0.07 h(-1) to verify the accuracy of the model. Cell growth was accurately predicted in both fed-batch runs; however, the model underestimated recombinant product concentration. The overall volumetric productivity of both runs was approximately 2.2 mg/L. h, representing a tenfold increase in the productivity compared with a heuristic feeding strategy.     

High-level production of a monoclonal antibody in murine myeloma cells by perfusion culture using a gravity settler

A perfusion system is described for the production of a human monoclonal antibody in non-secreting murine myeloma (NS0) cells that was previously shown to be difficult to produce at high levels using fed-batch culture. The perfusion system was based on the use of a commercially available cell settler as the separation device to separate the cells from the culture. Separation efficiency of the cell settler was above 98%. Based on the growth and glucose consumption rates, fresh media was added to the culture and the turnover rate for the bioreactor was set at a maximum of 1.5 times the bioreactor volume per day. The perfusion process resulted in twice the maximum viable cell densities and up to three times the total protein production in a 53-day run period when compared to the fed-batch process. In addition, charge heterogeneity of the antibody as measured by ion exchange chromatography was lower for material purified from the perfusion runs compared to fed-batch. Perfusion mode of culture using a commercially available gravity settler is therefore a viable alternative to fed-batch mode for high-level production of this monoclonal antibody in NS0 cells.     

Influence of culture modes on the microbial production of hyaluronic acid by <Emphasis Type="Italic">Streptococcus zooepidemicus</Emphasis>

The principal objective of this study was to assess the effects of culture modes including batch culture, pulse fed-batch culture, constant feeding rate fed-batch culture, and exponential fed-batch culture on the production of hyaluronic acid (HA) by Streptococcus zooepidemicus. Batch cultures had the highest levels of HA productivity, whereas fed-batch cultures were more favorable with regard to cell growth, and exponential fed-batch cultures evidenced the highest cell concentrations. A two-step culture model was proposed to enhance HA production: an exponential fed-batch culture was conducted prior to 8 h and then sucrose supplementation was applied for 8 h to start the batch fermentation of S. zooepidemicus. HA production and productivity were increased by 36 and 37% in the proposed two-step culture process as compared with that observed in the batch culture, respectively. The proposed two-step culture model can be applied in the production of secondary metabolites, and particularly of the exopolysaccharides.     

Use of fed-batch cultivation for achieving high cell densities for the pilot-scale production of a recombinant protein (phenylalanine dehydrogenase) in Escherichia coli

A fed-batch process for the high cell density cultivation of E. coli TG1 and the production of the recombinant protein phenylalanine dehydrogenase (PheDH) was developed. A model based on Monod kinetics with overflow metabolism and incorporating acetate utilization kinetics was used to generate simulations that describe cell growth, acetate production and reconsumption, and glucose consumption during fed-batch cultivation. Using these simulations a predetermined feeding profile was elaborated that would maintain carbon-limited growth at a growth rate below the critical growth rate for acetate formation (mu < mu(crit)). Two starvation periods are incorporated into the feed profile in order to induce acetate utilization. Cell concentrations of 53 g dry cell weight (DCW)/L were obtained with a final intracellular product concentration of recombinant protein corresponding to approximately 38% of the total cell protein. The yield of PheDH was 129 U/mL with a specific activity of 1.2 U/mg DCW and a maximum product formation rate of 0.41 U/mg DCW x h. The concentration of aectate was maintained below growth inhibitory levels until 3 h before the end of the fermentation when the concentration reached a maximum of 10.7 g/L due to IPTG induction of the recombinant protein.     

Kinetics of inclusion body production in batch and high cell density fed-batch culture of Escherichia coli expressing ovine growth hormone.

A process for maximizing the volumetric productivity of recombinant ovine growth hormone (r-oGH) expressed in Escherichia coli during high cell density fermentation process has been devised. Kinetics of r-oGH expression as inclusion bodies and its effect on specific growth rates of E. coli cells were monitored during batch fermentation process. It was observed that during r-oGH expression in E. coli, the specific growth rate of the culture became an intrinsic property of the cells which reduced in a programmed manner upon induction. Nutrient feeding during protein expression phase of the fed-batch process was designed according to the reduction in specific growth rate of the culture. By feeding yeast extract along with glucose during fed-batch operation, high cell growth with very little accumulation of acetic acid was observed. Use of yeast extract helped in maintaining high specific cellular protein yield which resulted in high volumetric productivity of r-oGH. In 16 h of fed-batch fermentation, 3.2 g l-1 of r-oGH were produced at a cell OD of 124. This is the highest concentration of r-oGH reported to date using E. coli expression system. The volumetric productivity of r-oGH was 0.2 g l-1 h-1, which is also the highest value reported for any therapeutic protein using IPTG inducible expression system in a single stage fed-batch process.     

Enhanced truncated-t-PA (CT-b) expression in high-cell-density fed-batch cultures of <Emphasis Type="Italic">Pichia pastoris</Emphasis> through optimization of a mixed feeding strategy by response surface methodology

Recently, Pichia pastoris has been the focal point of interest as an expression system for production of many recombinant proteins. The study and optimization of feeding strategy are of major importance to achieve maximum volumetric productivity in fed-batch cultivations. Among different feeding strategies used in P. pastoris fed-batch cultures, those trying to maintain a constant specific growth rate have usually resulted in superior productivities. The objective of the present study was to investigate and optimize the co-feeding of glycerol and methanol to attain maximum expression of t-PA in P. pastoris fed-batch cultures with constant specific growth rate. The experiments were designed by response surface methodology, considering the specific feeding rates of methanol and glycerol as independent variables. In each experiment, glycerol and methanol were fed according to a predetermined equation to maintain a constant specific growth rate. It was found that with glycerol feeding for higher specific growth rates, the inhibitory properties of glycerol are more pronounced, while the best expression level was achieved when the ratio of µ _set glycerol to that of methanol was around 1.67. In all specific growth rates tested, almost a similar ratio of the specific glycerol feeding rate to that of methanol led to the maximum protein production and activity. The statistical model predicted the optimal operating conditions for µ _set glycerol and that of methanol to be 0.05 and 0.03 h^?1, respectively. Applying the optimum strategy, maximum of 52 g/L biomass, 300 mg/L t-PA and 340,000 IU/mL enzyme activity were obtained.     

A new approach to define optimized range of medium composition for enhancement of hairy root production in fed-batch process

This work proposes a new methodology to identify the best medium concentrations for fed-batch production of hairy root using Datura innoxia as a model. Firstly, the role of each component on the growth rate is investigated separately. Then, an experimental design allows refining the optimization studying the interactions between the major species. The result analysis let to define concentration range optimized for fed-batch process. The work novelties lie in two aspects. Firstly, concentrations have been kept constant during each run. Thus, biomass uptakes do not affect the optimization and the growth rate is maintained constant during the exponential phase. Secondly, the effects of salts are generally studied. In this work, the influences of each ion are investigated in order to avoid bias due to the counter-ion effects. Compared to the classical B₅ medium, the optimized medium shows a significant improvement leading to more than 80% increase of final biomass production.     

Cell bank characterization and fermentation optimization for production of recombinant heavy chain C-terminal fragment of botulinum neurotoxin serotype E (rBoNTE(H(c)): antigen E) by Pichia pastoris

A process was developed for production of a candidate vaccine antigen, recombinant C-terminal heavy chain fragment of the botulinum neurotoxin serotype E, rBoNTE(H(c)) in Pichia pastoris. P. pastoris strain GS115 was transformed with the rBoNTE(H(c)) gene inserted into pHILD4 Escherichia coli-P. pastoris shuttle plasmid. The clone was characterized for genetic stability, copy number, and BoNTE(H(c)) sequence. Expression of rBoNTE(H(c)) from the Mut(+) HIS4 clone was confirmed in the shake-flask, prior to developing a fed-batch fermentation process at 5 and 19 L scale. The fermentation process consists of a glycerol growth phase in batch and fed-batch mode using a defined medium followed by a glycerol/methanol transition phase for adaptation to growth on methanol and a methanol induction phase resulting in the production of rBoNTE(H(c)). Specific growth rate, ratio of growth to induction phase, and time of induction were critical for optimal rBoNTE(H(c)) production and minimal proteolytic degradation. A computer-controlled exponential growth model was used for process automation and off-gas analysis was used for process monitoring. The optimized process had an induction time of 9 h on methanol and produced up to 3 mg of rBoNTE(H(c)) per gram wet cell mass as determined by HPLC and Western blot analysis.     

Introduction of the carbohydrate-activated promoter P(malK) for recombinant protein production

A production protocol for the use of the malK promoter was established. The protocol includes two phases: an initial fed-batch phase on glucose to reach a high cell density and a fed-batch phase on maltose for production of the desired recombinant protein. It is suggested that this cultivation scheme could be used for all promoters that are catabolite repressed by glucose and where growth and production need to be separated. The specific feature of this system is shown by its ability to control the rate of synthesis of the product protein, ss-galactosidase. In the production phase with a constant feed or an exponential feeding of 0.1 h(-1) it took 4 h longer to reach the maximum specific production rate than with the higher dilution rates of 0.25 h(-1) and 0.4 h(-1), respectively. In the above experiments a dilution rate of 0.3 h(-1) in the growth phase was used. The volumetric production of this system could furthermore be extended to 40 h. All protocol procedures so far tested resulted in the same maximum production rate, but reached in different lengths of time. It is argued that this system is particularly well suited for the production of proteins that have a complex structure and/or need to be produced in a soluble form or to be exported to the periplasm.     

Maximization of production of secreted recombinant proteins in Pichia pastoris fed-batch fermentation

Pontryagin's Maximum Principle has been applied for optimization of secreted proteins from Pichia pastoris fed-batch fermentation. The objective of this work is to maximize the total accumulated product per unit operation time under different given conditions and system constraints. To obtain optimal solutions, an automated curve-fitting software, Table Curve 2D, was employed to construct the necessary mathematical models and solve the complicated functions. In the solution processes, the end of the glycerol batch phase was defined as the initial state of the system, the end of the methanol fed-batch phase as the final state, the cell mass produced along with product accumulated as state variables, and the specific growth rate (mu) as the control variable. Initially, a relationship between the specific production rate (rho) and mu was established. Then, according to Pontryagin's Maximum Principle, the admissible range of mu and its trajectories for the optimal operations were determined. Four representative cases with different combinations of the operation time along with the initial and final states were evaluated. A close correlation was obtained between the predicted values of the model equation with the experimental results from the Pichia pastoris fed-batch fermentations producing secreted alpha-galactosidase. The approaches proposed here greatly simplify the computational processes and validate the optimization strategy as a generalized approach to maximize the yield from fed-batch fermentations.     

Use of chemostat data for modelling extracellular-inulinase production by Kluyveromyces marxianus in a high-cell-density fed-batch process

Production of extracellular inulinase by low-cell-density (2 kg dry weight·m⁻³) sucrose-limited chemostat cultures of Kluyveromyces marxianus obeyed saturated kinetics at dilution rates ranging from 0.02 to 0.5 h⁻¹. A non-structured Monod-type equation, describing the relation between specific growth rate and specific extracellular-inulinase production rate, was used to fit experimental data. THis equation was subsequently incorporated in a model for the production of biomass and extracellular inulinase in a high-cell-density (> 100 kg dry weight·m⁻³) fed-batchculture of K. marxianus grown on sucrose. The model adequately described biomass production in the fed-batch culture. However, the production of extracellular inulinase in the fed-batch process was slightly higher than predicted by the model. This observation may be related to differences in growth conditions between in the chemostat and fed-batch cultures.     

Production of pediocin SM-1 by Pediococcus pentosaceus Mees 1934 in fed-batch fermentation: Effects of sucrose concentration in a complex medium and process modeling

Production of pediocin SM-1 by Pediococcus pentosaceus Mees 1934 was investigated in semi-aerobic, pH-controlled, batch and fed-batch fermentations using a complex medium containing sucrose as the main source of carbon. The effects of sucrose concentration were studied in fed-batch fermentations in which a sucrose solution was added at stable feeding rates (5, 7, 9 and 10 g/l/h). The results showed that pediocin is produced as a product of the primary metabolism and its titer could be greatly improved by adjusting the sucrose feeding rate in fed-batch fermentation. The maximum titer of pediocin of 145 AU/ml was obtained in the fed-batch culture with 7 g/l/h feeding rate and that was 119% higher compared to the titer obtained in batch culture. Higher feeding rates (9 and 10 g/l/h) resulted in decreased pediocin yields while biomass levels appeared to be rather unaffected. The specific rate of pediocin formation was also sensitive to sucrose concentration levels. A mathematical model developed on the basis of well-known rate equations for batch and fed-batch cultures and growth associated production, described successfully cell growth, sucrose assimilation, lactate production and pediocin production in fed-batch culture.     

Increased heterologous protein production by Saccharomyces cerevisiae growing on ethanol as sole carbon source

Saccharomyces cerevisiae is a widely used host organism for the production of heterologous proteins, often cultivated in glucose-based fed-batch processes. This production system however has many factors limiting the productivity, mainly towards the end of the fermentation. For the optimised production of a Camelid antibody fragment this process was evaluated. In shake flask cultivations, it was found that ethanol has a strong effect on productivity increase and therefore glucose and ethanol fed-batch fermentations were compared. It appeared that specific heterologous protein production was up to five times higher in the ethanol cultivation and could be further optimised. Then the key characteristics of ethanol fed-batch fermentations such as growth rate and specific production were determined under ethanol limitation and accumulation and growth limiting conditions in the final phase of the process. It appeared that an optimal production process should have an ethanol accumulation throughout the feed phase of approximately 1% v/v in the broth and that production remains very efficient even in the last phase of the process. This productivity increase on ethanol versus glucose was also proven for several other Camelid antibody fragments some of which were heavily impaired in secretion on glucose, but very well produced on ethanol. This leads to the suggestion that the ethanol effect on improved heterologous protein production is linked to a stress response and folding and secretion efficiency.