Bioprocess Control Research Articles

Additive Uncertainty Consideration for Nonlinear and Multivariable Bioprocess Control

Biological processes are becoming more frequent nowadays due to the wide variety of products obtained from them and their possibility of making environmentally friendly some processes while high standard products are obtained. Nevertheless, controlling them has many difficulties due to their complex dynamic (multivariable and highly nonlinear systems) subject to modeling uncertainties and external disturbances presence. In this paper, two possibilities of improvement for a previously presented technique are proposed. In the first one, an approach based on the error estimation using Newtons backward interpolation is included in the design equations to decrease the uncertainties effect; while in the second one, some tracking error integrators are added in the control action calculation. Alternatives are applied in a bioethanol system, tested under different conditions and compared to show the improvements.

Coagulation factor IX analysis in bioreactor cell culture supernatant predicts quality of the purified product

Coagulation factor IX (FIX) is a complex post-translationally modified human serum glycoprotein and high-value biopharmaceutical. The quality of recombinant FIX (rFIX), especially complete γ-carboxylation, is critical for rFIX clinical efficacy. Bioreactor operating conditions can impact rFIX production and post-translational modifications (PTMs). With the goal of optimizing rFIX production, we developed a suite of Data Independent Acquisition Mass Spectrometry (DIA-MS) proteomics methods and used these to investigate rFIX yield, γ-carboxylation, other PTMs, and host cell proteins during bioreactor culture and after purification. We detail the dynamics of site-specific PTM occupancy and structure on rFIX during production, which correlated with the efficiency of purification and the quality of the purified product. We identified new PTMs in rFIX near the GLA domain which could impact rFIX GLA-dependent purification and function. Our workflows are applicable to other biologics and expression systems, and should aid in the optimization and quality control of upstream and downstream bioprocesses.

Visible/Short-wave near-infrared hyperspectral analysis of lipid concentration and fatty acid unsaturation of Scenedesmus obliquus in situ

Microalgae lipid and its degree of fatty acid unsaturation (DU) have the ability to undergo programmatic changes in response to different environmental and culture conditions. A real-time and reliable approach to monitor the change patterns of lipid-accumulation and DU is crucial for microalgal culture and bioprocess control. In this study, a fast visual and nondestructive method based on visible/short-wave near-infrared hyperspectral imaging (V/SNIR HSI) at 400–1000 nm region was developed to investigate the dynamic changes and content distribution of lipids and DU of Scenedesmus obliquus treated by nitrogen-low stress (0.3 g/L NaNO3, NL) and nitrogen-free stress (0 g/L NaNO3, NF). After collecting the average spectra from the region of interests (ROIs), the competitive adaptive reweighted sampling (CARS) algorithm and the successive projection algorithm (SPA) were applied to pick effective wavebands from the spectral data for the establishment of the lipid concentration and DU estimation models including partial least squares regression (PLSR) and multiple linear regression (MLR). The prediction results show the CARS-PLSR (Rp = 0.9762, RMSEp = 0.0143) and CARS-MLR (Rp = 0.9477, RMSEp = 0.5718) performed the best for lipids and DU of S. obliquus, respectively. Based on the optimal models, the spatial distribution of lipid abundance and DU in liquid suspension were visualized, which could be helpful in exploring the dynamic process of lipid synthesis and DU of S. obliquus. The overall results show that V/SNIR HSI is a promising technique for fast and noninvasively monitoring the lipid-accumulation and its characteristics i.e. DU in situ, and is expected to be applied for efficient and high-quality lipid production in microalgae industries.

A genome-scale metabolic network model and machine learning predict amino acid concentrations in Chinese Hamster Ovary cell cultures.

The control of nutrient availability is critical to large-scale manufacturing of biotherapeutics. However, the quantification of proteinogenic amino acids is time-consuming and thus is difficult to implement for real-time in situ bioprocess control. Genome-scale metabolic models describe the metabolic conversion from media nutrients to proliferation and recombinant protein production, and therefore are a promising platform for in silico monitoring and prediction of amino acid concentrations. This potential has not been realized due to unresolved challenges: (1) the models assume an optimal and highly efficient metabolism, and therefore tend to underestimate amino acid consumption, and (2) the models assume a steady state, and therefore have a short forecast range. We address these challenges by integrating machine learning with the metabolic models. Through this we demonstrate accurate and time-course dependent prediction of individual amino acid concentration in culture medium throughout the production process. Thus, these models can be deployed to control nutrient feeding to avoid premature nutrient depletion or provide early predictions of failed bioreactor runs.

Modelling Cell Metabolism: A Review on Constraint-Based Steady-State and Kinetic Approaches

Studying cell metabolism serves a plethora of objectives such as the enhancement of bioprocess performance, and advancement in the understanding of cell biology, of drug target discovery, and in metabolic therapy. Remarkable successes in these fields emerged from heuristics approaches, for instance, with the introduction of effective strategies for genetic modifications, drug developments and optimization of bioprocess management. However, heuristics approaches have showed significant shortcomings, such as to describe regulation of metabolic pathways and to extrapolate experimental conditions. In the specific case of bioprocess management, such shortcomings limit their capacity to increase product quality, while maintaining desirable productivity and reproducibility levels. For instance, since heuristics approaches are not capable of prediction of the cellular functions under varying experimental conditions, they may lead to sub-optimal processes. Also, such approaches used for bioprocess control often fail in regulating a process under unexpected variations of external conditions. Therefore, methodologies inspired by the systematic mathematical formulation of cell metabolism have been used to address such drawbacks and achieve robust reproducible results. Mathematical modelling approaches are effective for both the characterization of the cell physiology, and the estimation of metabolic pathways utilization, thus allowing to characterize a cell population metabolic behavior. In this article, we present a review on methodology used and promising mathematical modelling approaches, focusing primarily to investigate metabolic events and regulation. Proceeding from a topological representation of the metabolic networks, we first present the metabolic modelling approaches that investigate cell metabolism at steady state, complying to the constraints imposed by mass conservation law and thermodynamics of reactions reversibility. Constraint-based models (CBMs) are reviewed highlighting the set of assumed optimality functions for reaction pathways. We explore models simulating cell growth dynamics, by expanding flux balance models developed at steady state. Then, discussing a change of metabolic modelling paradigm, we describe dynamic kinetic models that are based on the mathematical representation of the mechanistic description of nonlinear enzyme activities. In such approaches metabolic pathway regulations are considered explicitly as a function of the activity of other components of metabolic networks and possibly far from the metabolic steady state. We have also assessed the significance of metabolic model parameterization in kinetic models, summarizing a standard parameter estimation procedure frequently employed in kinetic metabolic modelling literature. Finally, some optimization practices used for the parameter estimation are reviewed.

Application of Statistical Modeling and Hypothesis Testing to Reinforce Model Validation Concepts in Bioprocess Control Laboratory

Application of Statistical Modeling and Hypothesis Testing to Reinforce Model Validation Concepts in Bioprocess Control Laboratory

Antonio A. Alonso. In memoriam. Contributions to modelling and control of food and bioprocesses

Antonio A. Alonso. In memoriam. Contributions to modelling and control of food and bioprocesses

Selection of a Minimal Suboptimal Set of EFMs for Dynamic Metabolic Modelling

Mathematical modelling supports the understanding of basic biological mechanisms and is the basis for bioprocess simulation, prediction, control and optimization. Dynamic macroscopic models can be derived from the concept of elementary flux modes (EFMs), which provide a comprehensive representation of all possible pathways through a metabolic network. As the number of EFMs drastically increases with the size of the metabolic network, a procedure to reduce the number of EFMs and select the most informative ones is required. For this purpose, this study proposes a methodology to select a minimal suboptimal set of elementary flux modes allowing the development of reduced macroscopic models. The algorithm is divided into two steps. First, the concept behind the cosine-similarity algorithm is extended for a large number of EFMs to cut the initial set by removing all the collinear modes. Next, the algorithm is used to extract only the most informative modes from the reduced set by means of a series of optimization problems. The algorithm performance is illustrated with datasets from hybridoma cultures in batch and perfusion modes, and two metabolic networks with different levels of description.

Online deep neural network-based feedback control of a Lutein bioprocess

An online adaptive deep neural network (DNN) scheme has been introduced for the tracking control of a nonlinear bioprocess with uncertain internal dynamics. First, a detailed controllability analysis is conducted for the Lutein bioprocess to represent the bioprocess as a nonlinear system in affine form. Next, a controller consisting of a DNN-based function approximator is designed for the nonlinear Lutein production bioprocess. It is demonstrated that closed-loop tracking control of a bioprocess for a desired yield profile is possible only with two inputs. The set point trajectory to yield maximum Lutein production is shown by the proposed online adaptive deep NN controller. The proposed controller exhibits self-learning capability under closed loop condition, due to the online learning phase. In other words, no explicit offline learning phase is required and online learning is preferred due to lack of a priori training data for approximating complex nonlinear functions. Simulation results are provided to confirm the performance of the proposed approach.

Real-time bioprocess monitoring using a mid-infrared fibre-optic sensor

This work proposes a simple, easy-to-fabricate and cost-efficient sensor based on mid-infrared (MIR) spectroscopy for in-situ real-time bioprocess monitoring. A process involving the production of sophorolipid biosurfactants by way of fermentation was studied with the aim of monitoring varying substrate and product concentrations over time and also distinguishing between the main two structurally different types of biosurfactant produced: lactonic and acidic sophorolipids. First, measurements were carried out using ATR-FTIR spectroscopy. This was followed by comparative measurements using a fabricated U-bent fibre-optic sensor designed for in-situ monitoring. It was demonstrated for the first time that varying concentrations of sophorolipids in a fermentation broth could successfully be monitored over time using an in-situ MIR fibre-optic sensor. Moreover, acidic and lactonic sophorolipids could simultaneously be identified at varying concentrations. The utilisation of such a sensor in bioprocesses can provide a wealth of real-time data, allowing for improved control of feeding rates and better process characterisation, providing significant opportunities for bioprocess control, resulting in increased production and decreased overall costs.

Parameter and state estimation of backers yeast cultivation with a gas sensor array and unscented Kalman filter.

Real‐time information about the concentrations of substrates and biomass is the key to accurate monitoring and control of bioprocess. However, on‐line measurement of these variables is a challenging task and new measurement systems are still required. An alternative are software sensors, which can be used for state and parameter estimation in bioprocesses. The software sensors predict the state of the process by using mathematical models as well as data from measured variables. The Kalman filter is a type of such sensors.In this paper, we have used the Unscented Kalman Filter (UKF) which is a nonlinear extension of the Kalman filter for on‐line estimation of biomass, glucose and ethanol concentration as well as for estimating the growth rate parameters in S. cerevisiae batch cultivation, based on infrequent ethanol measurements. The UKF algorithm was validated on three different cultivations with variability of the substrate concentrations and the estimated values were compared to the off‐line values.The results obtained showed that the UKF algorithm provides satisfactory results with respect to estimation of concentrations of substrates and biomass as well as the growth rate parameters during the batch cultivation.

Towards smart biomanufacturing: a perspective on recent developments in industrial measurement and monitoring technologies for bio-based production processes

The biomanufacturing industry has now the opportunity to upgrade its production processes to be in harmony with the latest industrial revolution. Technology creates capabilities that enable smart manufacturing while still complying with unfolding regulations. However, many biomanufacturing companies, especially in the biopharma sector, still have a long way to go to fully benefit from smart manufacturing as they first need to transition their current operations to an information-driven future. One of the most significant obstacles towards the implementation of smart biomanufacturing is the collection of large sets of relevant data. Therefore, in this work, we both summarize the advances that have been made to date with regards to the monitoring and control of bioprocesses, and highlight some of the key technologies that have the potential to contribute to gathering big data. Empowering the current biomanufacturing industry to transition to Industry 4.0 operations allows for improved productivity through information-driven automation, not only by developing infrastructure, but also by introducing more advanced monitoring and control strategies.

Optimizing the biosynthesis of oxygenated and acetylated Taxol precursors in Saccharomyces cerevisiae using advanced bioprocessing strategies.

Taxadien-5α-hydroxylase and taxadien-5α-ol O-acetyltransferase catalyze the oxidation of taxadiene to taxadien-5α-ol and subsequent acetylation to taxadien-5α-yl-acetate in the biosynthesis of the blockbuster anticancer drug, paclitaxel (Taxol®). Despite decades of research, the promiscuous and multispecific CYP725A4 enzyme remains a major bottleneck in microbial biosynthetic pathway development. In this study, an interdisciplinary approach was applied for the construction and optimization of the early pathway in Saccharomyces cerevisiae, across a range of bioreactor scales. High-throughput microscale optimization enhanced total oxygenated taxane titer to 39.0 ± 5.7 mg/L and total taxane product titers were comparable at micro and minibioreactor scale at 95.4 ± 18.0 and 98.9 mg/L, respectively. The introduction of pH control successfully mitigated a reduction of oxygenated taxane production, enhancing the potential taxadien-5α-ol isomer titer to 19.2 mg/L, comparable with the 23.8 ± 3.7 mg/L achieved at microscale. A combination of bioprocess optimization and increased gas chromatography-mass spectrometry resolution at 1 L bioreactor scale facilitated taxadien-5α-yl-acetate detection with a final titer of 3.7 mg/L. Total oxygenated taxane titers were improved 2.7-fold at this scale to 78 mg/L, the highest reported titer in yeast. Critical parameters affecting the productivity of the engineered strain were identified across a range of scales, providing a foundation for the development of robust integrated bioprocess control systems.

Model-Free Extremum Seeking Control of Bioprocesses: A Review with a Worked Example

Uncertainty is a common feature of biological systems, and model-free extremum-seeking control has proved a relevant approach to avoid the typical problems related to model-based optimization, e.g., time- and resource-consuming derivation and identification of dynamic models, and lack of robustness of optimal control. In this article, a review of the past and current trends in model-free extremum seeking is proposed with an emphasis on finding optimal operating conditions of bioprocesses. This review is illustrated with a simple simulation case study which allows a comparative evaluation of a few selected methods. Finally, some experimental case studies are discussed. As usual, practice lags behind theory, but recent developments confirm the applicability of the approach at the laboratory scale and are encouraging a transfer to industrial scale.

From Microbial Communities to Distributed Computing Systems.

A distributed biological system can be defined as a system whose components are located in different subpopulations, which communicate and coordinate their actions through interpopulation messages and interactions. We see that distributed systems are pervasive in nature, performing computation across all scales, from microbial communities to a flock of birds. We often observe that information processing within communities exhibits a complexity far greater than any single organism. Synthetic biology is an area of research which aims to design and build synthetic biological machines from biological parts to perform a defined function, in a manner similar to the engineering disciplines. However, the field has reached a bottleneck in the complexity of the genetic networks that we can implement using monocultures, facing constraints from metabolic burden and genetic interference. This makes building distributed biological systems an attractive prospect for synthetic biology that would alleviate these constraints and allow us to expand the applications of our systems into areas including complex biosensing and diagnostic tools, bioprocess control and the monitoring of industrial processes. In this review we will discuss the fundamental limitations we face when engineering functionality with a monoculture, and the key areas where distributed systems can provide an advantage. We cite evidence from natural systems that support arguments in favor of distributed systems to overcome the limitations of monocultures. Following this we conduct a comprehensive overview of the synthetic communities that have been built to date, and the components that have been used. The potential computational capabilities of communities are discussed, along with some of the applications that these will be useful for. We discuss some of the challenges with building co-cultures, including the problem of competitive exclusion and maintenance of desired community composition. Finally, we assess computational frameworks currently available to aide in the design of microbial communities and identify areas where we lack the necessary tools.

Special Issue: Bioprocess Monitoring and Control

Bioprocesses can be found in different areas such as the production of food, feed, energy, chemicals, and pharmaceuticals [...]

Development of process analytical technology to monitor the cell secretome during differentiation of pluripotent stem cells

Background & Aim Induced pluripotent stem cells (iPSC) offers a renewable source for reliably and robustly generating off-the-shelf cell products for regenerative medicine. A number of differentiation protocols have been developed and PSC-derived cells have been used in a variety of animal models to treat disease and progressed through to clinical trials in Asia and more recently the United States. However, the manufacturing processes for expansion and differentiation of iPSC cells remain reliant on poorly scalable technologies with limited in process monitoring despite typically > 14 days of culture and manipulation required; increasing the risks and costs of batch manufacturing failure. The industrialisation of pluripotent stem cell-derived cell therapies requires the development of process analytical technologies (PAT) for the monitoring and control of large-scale cell differentiation bioprocesses. The extracellular microenvironment provides fertile hunting ground to determine the impact of processing conditions on process outcome with the myriad of molecules consumed and produced by cells during differentiation alluding to cell state and potential cell fate. PAT that targets the spent media enables control of critical processing parameters to maintain product critical quality attributes without requiring cell-based analysis to be conducted. Methods, Results & Conclusion Using two exemplar differentiation processes, going from induced pluripotent stem cells to, either, definitive endoderm or cardiomyocytes, we have characterised the cell secretome with respect to metabolites and secreted proteins, and performed multi-parametric analysis to determine potential biomarkers for monitoring future large-scale differentiation processes using cost- and time-effective PAT alternatives.

Chemical and Biological Sensor Capsules for Real-Time Measurement of Cell Properties in Bioreactors

Introduction Real-time bioprocess monitoring and control is needed for the scalable production and deployment of cancer cell therapies at reasonable cost. For example, CAR-T cell therapy shows promise as an effective treatment for cancer with high (70-90%) remission rates.1 However, the cost of these treatments is too expensive for their mass adoption.2 One reason for the steep price is the difficulty scaling production while ensuring delivery of an efficacious therapy to the patient. One factor preventing scalability is the lack of effective process analytical technologies. As a potential solution to this issue, we have developed a fully integrated, wireless, 3D-printed sensor capsule to be used for multiplexed sensing of critical quality attributes (CQAs), namely pH, glucose, lactate, and IL-2 concentrations within cell bioreactors. CQAs are used not only as metrics for cell viability, but also as determinants of a cell’s ability to deliver efficacious treatment. Unlike current process monitoring technology, capsule sensors are buoyant and can be propelled using impellers within the bioreactor. This allows for measurements uniformly inside the bioreactor. To our knowledge, this is the first time that 3D-printed, wireless sensor capsules have been developed for use in cell bioreactors for cancer cell therapy. Methods The capsule consists of electrochemical sensors as well as data acquisition and wireless transmission electronics integrated into a capsule made of a biocompatible polymer (Figure 1). The sensors for glucose, lactate, pH, and IL-2 along with Ag on-chip pseudo-reference electrode (pseudo-RE) were fabricated using standard lithographic techniques on silicon substrates. The reference electrode consisted of a thin-film of Ag that was later coated with polyvinyl butyral (PVB) to improve stability.3 IL-2 and pH sensing was achieved using gold electrodes. A pH-sensitive Al2O3 or HfO2 layer is deposited through atomic layer deposition on the pH sensor to achieve pH-sensitivity. Glucose and lactate sensing was achieved amperometrically using enzymes, glucose or lactase oxidase respectively, immobilized on a Pt electrode biased to 0.7 V. Detection could also be achieved potentiometrically by using the pH sensor to detect changes in pH caused by conversion of the enzyme product, H2O2, to H+.Figure 1 shows the capsule and sensor design. The capsule was 3D printed using a Stratasys Connex-350 and was designed to be fully insulated from solution except for an opening on the top revealing the sensors. The sensor chip is installed within the cap of the capsule which can be screwed off, but with wired connection to the electronics housed in the body of the capsule. This allows for simple replacement of the sensor chip while retaining the capsule components for reuse. Results and Conclusions Before testing within the capsule, the sensors were tested using microfluidics. Figure 2 shows the performance of the Ag/PVB pseudo-RE. The pseudo-RE shows a stable potential in different pH buffer solutions and similar sensitivity results when compared to a commercial reference electrode for pH sensing. By having a stable microfabricated pseudoRE, the sensing components will not limit miniaturization of the capsule. Figure 3 shows the response of the pH, lactate, and glucose sensors at different concentrations. The pH sensor shows high sensitivity (~53 mV/pH) over the range of pH 3 to 7. Glucose and lactate sensing was tested in 1X PBS. The glucose sensor shows a linear response to glucose concentrations from 1-6 mM with saturation at ~8 mM validating the use of the sensor within cell culture media.4 The lactate sensor shows sensitivity in the concentration range of 1-5 mM with saturation occurring around ~6 mM. The division of T cells requires aerobic glycolysis where glucose is converted to lactate. However, not all glucose is expected to be converted. The lactate concentration expected to be produced is ~1 mM.5 Selectivity towards IL-2 is achieved using anti-IL2 antibodies. Figure 4a shows the functionalization scheme for detection of IL-2. A thiol self-assembled monolayer (SAM) is attached to the gold electrode and then modified for amine conjugation using NHS/EDC. Anti-IL2 can then bind to the SAM and unbound sites are blocked with ethanolamine. Figure 4b reveals antibody attachment occurs optimally at pH 5. Surface plasmon resonance shows successful attachment of 10 ug/mL of IL-2 to the functionalized gold sensor. The next step is to demonstrate IL-2 sensing potentiometrically. The capsule’s wireless sensing capability was validated using pH sensing in a beaker. The pH sensing results in Figure 5b were obtained wirelessly from the capsule and were nearly identical to the results measured using our microfluidic system and Keithley-4200 lab instrument (Fig. 5a). This work has successfully demonstrated wireless sensing using a 3D-printed sensor capsule and functionality of electrochemical sensors for multiple CQAs for cell growth in bioreactors. These sensor capsules can enable scalability of the cell manufacturing process while ensuring delivery of an efficacious treatment. Future work will focus on capsule sensor validation in cell growth media and monitoring of cell growth in bioreactors.

Development of Novel Bioreactor Control Systems Based on Smart Sensors and Actuators.

Bioreactors of various forms have been widely used in environmental protection, healthcare, industrial biotechnology, and space exploration. Robust demand in the field stimulated the development of novel designs of bioreactor geometries and process control strategies and the evolution of the physical structure of the control system. After the introduction of digital computers to bioreactor process control, a hierarchical structure control system (HSCS) for bioreactors has become the dominant physical structure, having high efficiency and robustness. However, inherent drawbacks of the HSCS for bioreactors have produced a need for a more consolidated solution of the control system. With the fast progress in sensors, machinery, and information technology, the development of a flat organizational control system (FOCS) for bioreactors based on parallel distributed smart sensors and actuators may provide a more concise solution for process control in bioreactors. Here, we review the evolution of the physical structure of bioreactor control systems and discuss the properties of the novel FOCS for bioreactors and related smart sensors and actuators and their application circumstances, with the hope of further improving the efficiency, robustness, and economics of bioprocess control.

Adaptive predictive control of bioprocesses with constraint-based modeling and estimation

Control of biotechnological processes is currently recipe-based with insufficient ability to handle possible uncertainties, which results in suboptimal production processes. To address this problem, model-based optimization and control approaches can be implemented to derive optimal control strategies. However, for reliable performance of model-based control, it is crucial to use flexible and adaptive control strategies which address biological variability while compensating for uncertainties. In this work, we present an approach for adaptive control of a bioprocess based on dynamic flux balance models. A previously developed bilevel approach for bioprocess optimization is implemented inside a model predictive control (MPC) routine. To account for model uncertainties, a moving horizon estimation algorithm is combined with the MPC in order to estimate uncertain parameters of the underlying model online for different metabolic modes. We apply this method to maximize the productivity of a target metabolite under microaerobic conditions by adapting the degree of oxygen-limitation online.