Hybrid Approach to State Estimation for Bioprocess Control.

An improved state estimation technique for bioprocess control applications is proposed where a hybrid version of the Unscented Kalman Filter (UKF) is employed. The underlying dynamic system model is formulated as a conventional system of ordinary differential equations based on the mass balances of the state variables biomass, substrate, and product, while the observation model, describing the less established relationship between the state variables and the measurement quantities, is formulated in a data driven way. The latter is formulated by means of a support vector regression (SVR) model. The UKF is applied to a recombinant therapeutic protein production process using Escherichia coli bacteria. Additionally, the state vector was extended by the specific biomass growth rate µ in order to allow for the estimation of this key variable which is crucial for the implementation of innovative control algorithms in recombinant therapeutic protein production processes. The state estimates depict a sufficiently low noise level which goes perfectly with different advanced bioprocess control applications.

Bioreactor control improves bioprocess performance.

The performance of bioreactors is not only determined by productivity but also by process quality, which is mainly determined by variances in the process variables. As fluctuations in these quantities directly affect the variability in the product properties, combatting distortions is the main task of practical quality assurance. The straightforward way of reducing this variability is keeping the product formation process tightly under control. Purpose of this keynote is to show that there is enough evidence in literature showing that the performance of the fermentation processes can significantly be improved by feedback control. Most of the currently used open loop control procedures can be replaced by relatively simple feedback techniques. It is shown by practical examples that such a retrofitting does not require significant changes in the well-established equipment. Feedback techniques are best in assuring high reproducibility of the industrial cultivation processes and thus in assuring the quality of their products. Many developments in supervising and controlling industrial fermentations can directly be taken over in manufacturing processes. Even simple feedback controllers can efficiently improve the product quality. It's the time now that manufacturers follow the developments in most other industries and improve process quality by automatic feedback control.

Data-based optimization of protein production processes.

While data-based modeling is possible in various ways, data-based optimization has not been previously described. Here we present such an optimization technique. It is based on dynamic programming principles and uses data directly from exploratory experiments where the influence of the adjustable variables u were tested at various values. Instead of formulating the performance index J as a function of time t within a cultivation process it is formulated as a function of the biomass x. The advantage of this representation is that in most biochemical production processes J(x) only depends of the vector u of the adjustable variables. This given, mathematical programming techniques allow determining the desired optimal paths u(opt)(x) from the x-derivatives of J(x). The resulting u(opt)(x) can easily be transformed back to the u(t) profiles that can then be used in an improved fermentation run. The optimization technique can easily be explained graphically. With numerical experiments the feasibility of the method is demonstrated. Then, two optimization runs for recombinant protein formations in E. coli are discussed and experimental validation results are presented.

Increasing batch-to-batch reproducibility of CHO-cell cultures using a model predictive control approach.

By means of a model predictive control strategy it was possible to ensure a high batch-to-batch reproducibility in animal cell (CHO-cell) suspensions cultured for a recombinant therapeutic protein (EPO) production. The general control objective was derived by identifying an optimal specific growth rate taking productivity, protein quality and process controllability into account. This goal was approached indirectly by controlling the oxygen mass consumed by the cells which is related to specific biomass growth rate and cell concentration profile by manipulating the glutamine feed rate. Process knowledge represented by a classical model was incorporated into the model predictive control algorithm. The controller was employed in several cultivation experiments. During these cultivations, the model parameters were adapted after each sampling event to cope with changes in the process' dynamics. The ability to predict the state variables, particularly for the oxygen consumption, led to only moderate changes in the desired optimal operational trajectories. Hence, nearly identical oxygen consumption profiles, cell and protein titers as well as sialylation patterns were obtained for all cultivation runs.

Insights into large-scale cell-culture reactors: I. Liquid mixing and oxygen supply.

In the pharmaceutical industry, it is state of the art to produce recombinant proteins and antibodies with animal-cell cultures using bioreactors with volumes of up to 20 m(3) . Recent guidelines and position papers for the industry by the US FDA and the European Medicines Agency stress the necessity of mechanistic insights into large-scale bioreactors. A detailed mechanistic view of their practically relevant subsystems is required as well as their mutual interactions, i.e., mixing or homogenization of the culture broth and sufficient mass and heat transfer. In large-scale bioreactors for animal-cell cultures, different agitation systems are employed. Here, we discuss details of the flows induced in stirred tank reactors relevant for animal-cell cultures. In addition, solutions of the governing fluid dynamic equations obtained with the so-called computational fluid dynamics are presented. Experimental data obtained with improved measurement techniques are shown. The results are compared to previous studies and it is found that they support current hypotheses or models. Progress in improving insights requires continuous interactions between more accurate measurements and physical models. The paper aims at promoting the basic mechanistic understanding of transport phenomena that are crucial for large-scale animal-cell culture reactors.

Insights into large-scale cell-culture reactors: II. Gas-phase mixing and CO2 stripping.

Most discussions about stirred tank bioreactors for cell cultures focus on liquid-phase motions and neglect the importance of the gas phase for mixing, power input and especially CO(2) stripping. Particularly in large production reactors, CO(2) removal from the culture is known to be a major problem. Here, we show that stripping is mainly affected by the change of the gas composition during the movement of the gas phase through the bioreactor from the sparger system towards the headspace. A mathematical model for CO(2)-stripping and O(2)-mass transfer is presented taking gas-residence times into account. The gas phase is not moving through the reactor in form of a plug flow as often assumed. The model is validated by measurement data. Further measurement results are presented that show how the gas is partly recirculated by the impellers, thus increasing the gas-residence time. The gas-residence times can be measured easily with stimulus-response techniques. The results offer further insights on the gas-residence time distributions in stirred tank reactors.

Simplified off-gas analyses in animal cell cultures for process monitoring and control purposes.

Batch-to-batch reproducibility of animal cell cultures can significantly be enhanced using process control procedures. Most informative signals for advanced process control can be derived from the volume fractions of oxygen and carbon dioxide in the vent line of the reactors. Here we employed simple low-cost sensors, previously not considered for off-gas analysis at a laboratory-scale cell cultures, and compared them with a simultaneously used quadrupole mass spectrometer, i.e., the standard equipment. A decisive advantage is that the sensors did not need any calibration and are easy to use. We show that monitoring and advanced control of cell cultures can significantly be simplified using the devices tested here and that the same batch-to-batch reproducibility can be obtained with much less effort than before.

Simple control of fed-batch processes for recombinant protein production with E. coli.

A very simple but effective process control technique is proposed that leads to a high batch-to-batch reproducibility with respect to biomass concentration as well as the specific biomass growth rate profiles in E. coli fermentations performed during recombinant protein production. It makes use of the well-established temperature controllers in currently used fermenters, but takes its information from the difference between the controlled culture temperature T (cult) and the temperature T (coolin) of the coolant fed to the fermenter's cooling jacket as adjusted by the fermenter temperature controller. For process control purposes this measured difference is corrected regarding stirrer influences and cumulated before it is used as a new process control variable. As a spin-off of this control, it becomes possible to estimate online the oxygen mass transfer rates and the corresponding k(L)a values during the real cultivation process.

Selective expression of the soluble product fraction in Escherichia coli cultures employed in recombinant protein production processes.

Recombinant proteins produced in Escherichia coli hosts may appear within the cells' cytoplasm in form of insoluble inclusion bodies (IB's) and/or as dissolved functional protein molecules. If no efficient refolding procedure is available, one is interested in obtaining as much product as possible in its soluble form. Here, we present a process engineering approach to maximizing the soluble target protein fraction. For that purpose, a dynamic process model was developed. Its essential kinetic component, the specific soluble product formation rate, if represented as a function of the specific growth rate and the culture temperature, depicts a clear maximum. Based on the dynamic model, optimal specific growth rate and temperature profiles for the fed-batch fermentation were determined. In the course of the study reported, the mass of desired soluble protein was increased by about 25%. At the same time, the formation of inclusion bodies was essentially avoided. As the optimal cultivation procedure is rather susceptible to distortions, control measures are necessary to guarantee that the real process can be kept on its desired path. This was possible with robust closed loop control. Experimental process validation revealed that, in this way, high dissolved product fractions could be obtained at an excellent batch-to-batch reproducibility.

Control of cultivation processes for recombinant protein production: a review.

The current state-of-the-art in control of cultivation processes for recombinant protein production is examined including the quantitative knowledge that can be activated for this purpose and the measurement techniques that can be employed for control at industrial manufacturing sites.

Product formation kinetics in genetically modified E. coli bacteria: inclusion body formation.

A data-driven model is presented that can serve two important purposes. First, the specific growth rate and the specific product formation rate are determined as a function of time and thus the dependency of the specific product formation rate from the specific biomass growth rate. The results appear in form of trained artificial neural networks from which concrete values can easily be computed. The second purpose is using these results for online estimation of current values for the most important state variables of the fermentation process. One only needs online data of the total carbon dioxide production rate (tCPR) produced and an initial value x of the biomass, i.e., the size of the inoculum, for model evaluation. Hence, given the inoculum size and online values of tCPR, the model can directly be employed as a softsensor for the actual value of the biomass, the product mass as well as the specific biomass growth rate and the specific product formation rate. In this paper the method is applied to fermentation experiments on the laboratory scale with an E. coli strain producing a recombinant protein that appears in form of inclusion bodies within the cells' cytoplasm.

Periplasmic production of native human proinsulin as a fusion to E. coli ecotin.

Native proinsulin belongs to the class of the difficult-to-express proteins in Escherichia coli. Problems mainly arise due to its small size, a high proteolytic decay, and the necessity to form a native disulfide pattern. In the present study, human proinsulin was produced in the periplasm of E. coli as a fusion to ecotin, which is a small periplasmic protein of 16 kDa encoded by the host, containing one disulfide bond. The fusion protein was secreted to the periplasm and native proinsulin was determined by ELISA. Cultivation parameters were studied in parallel batch mode fermentations using E. coli BL21(DE3)Gold as a host. After improvement of fed-batch high density fermentation conditions, 153 mg fusion protein corresponding to 51.5mg native proinsulin was obtained per L. Proteins were extracted from the periplasm by osmotic shock treatment. The fusion protein was purified in one step by ecotin affinity chromatography on immobilized trypsinogen. After thrombin cleavage of the fusion protein, the products were separated by Ni-NTA chromatography. Proinsulin was quantified by ELISA and characterized by mass spectrometry. To evaluate the influence of periplasmic proteases, the amount of ecotin-proinsulin was determined in E. coli BL21(DE3)Gold and in a periplasmic protease deficient strain, E. coli SF120.

Improving the batch-to-batch reproducibility of microbial cultures during recombinant protein production by regulation of the total carbon dioxide production.

Batch-to-batch reproducibility of fermentation processes performed during the manufacturing processes of biologics can be increased by operating the cultures at feed rate profiles that are robust against typically arising disturbances. Remaining randomly appearing deviations from the desired path should be suppressed automatically by manipulating the feed rate. With respect to the cells' physiology it is best guiding the cultivations along an optimal profile of the specific biomass growth rate mu(t). However, there are two problems that speak for further investigations: Upon severe disturbances that may happen during the fermentation, the biomass concentration X may significantly deviate from its desired value, then a fixed mu-profile leads to a diminished batch-to-batch reproducibility. Second, the specific growth rate cannot easily be estimated online to a favourably high accuracy, hence it is difficult to determine the deviations in mu from the desired profile. The alternative discussed here solves both problems by keeping the process at the corresponding total cumulative carbon dioxide production-profile: it is robust against distortions in X and the controlled variable can accurately be measured online during cultivations of all relevant sizes. As compared to the fermentation practice currently used in industry, the experimental results, presented at the example of a recombinant protein production with Escherichia coli cells, show that CPR-based corrections lead to a considerably improved batch-to-batch reproducibility.

Improving the batch-to-batch reproducibility in microbial cultures during recombinant protein production by guiding the process along a predefined total biomass profile.

In industry Escherichia coli is the preferred host system for the heterologous biosynthesis of therapeutic proteins that do not need posttranslational modifications. In this report, the development of a robust high-cell-density fed-batch procedure for the efficient production of a therapeutic hormone is described. The strategy is to guide the process along a predefined profile of the total biomass that was derived from a given specific growth rate profile. This profile might have been built upon experience or derived from numerical process optimization. A surprisingly simple adaptive procedure correcting for deviations from the desired path was developed. In this way the batch-to-batch reproducibility can be drastically improved as compared to the process control strategies typically applied in industry. This applies not only to the biomass but, as the results clearly show, to the product titer also.

Open-loop control of the biomass concentration within the growth phase of recombinant protein production processes.

Recombinant protein production processes are typically divided into two phases. In the first one, pure cell propagation takes place, while in the second one product formation is switched on within the cells by adding an inducer. In the initial biomass formation phase, the cell density is rather low and, hence, the measurement quantities that could be used to determine the process' state depict small values and are rather severely distorted by measurement noise. Because of these measurement problems, the fermentation cannot be reliably controlled by feedback control during this first production phase; instead, the process must be controlled in an open-loop fashion. The consequence, worked out in this paper, is to design substrate feed rate profiles for the growth phase in such a way that they are robust with respect to the main disturbances observed in practice. The robustness of the biomass formation is shown to be primarily dependent on the specific growth rate adjusted in the first hours. High batch-to-batch reproducibility can be obtained with exponential feeding profiles F(t) corresponding to specific growth rates micro(set) well below the maximal specific growth rate micro(max) of the organism. The reduction in the growth rate needed to obtain a robust process behavior depends on the inaccuracies in the initial biomass concentrations. Quantitative feed rate profiles were obtained by numerical simulation and these results were validated experimentally by means of a series of cultivation runs, where a recombinant pharmaceutical protein was produced. All experimental data confirmed the assumptions made in the robust process design study.

Estimation of biomass concentrations in fermentation processes for recombinant protein production.

Online biomass estimation for bioprocess supervision and control purposes is addressed. As the biomass concentration cannot be measured online during the production to sufficient accuracy, indirect measurement techniques are required. Here we compare several possibilities for the concrete case of recombinant protein production with genetically modified Escherichia coli bacteria and perform a ranking. At normal process operation, the best estimates can be obtained with artificial neural networks (ANNs). When they cannot be employed, statistical correlation techniques can be used such as multivariate regression techniques. Simple model-based techniques, e.g., those based on the Luedeking/Piret-type are not as accurate as the ANN approach; however, they are very robust. Techniques based on principal component analysis can be used to recognize abnormal cultivation behavior. For the cases investigated, a complete ranking list of the methods is given in terms of the root-mean-square error of the estimates. All techniques examined are in line with the recommendations expressed in the process analytical technology (PAT)-initiative of the FDA.