Model-based analysis of biocatalytic processes and performance of microbioreactors with integrated optical sensors

Highlights

• Mechanistic models developed to describe bioprocesses inside microbioreactors.

• Models allow fast identification of reaction mechanism, kinetics and limitations.

• Fluid flow and enzyme adsorption affects response of optical sensor inside μBR.

• Extra catalase and hydrogen peroxide in μBR disturb local oxygen concentrations.

• Framework created screening bioprocesses, sensor response and μBR designs.

Abstract

Design and development of scale-down approaches, such as microbioreactor (μBR) technologies with integrated sensors, are an adequate solution for rapid, high-throughput and cost-effective screening of valuable reactions and/or production strains, with considerably reduced use of reagents and generation of waste. A significant challenge in the successful and widespread application of μBRs in biotechnology remains the lack of appropriate software and automated data interpretation of μBR experiments. Here, it is demonstrated how mathematical models can be usedas helpful tools, not only to exploit the capabilities of microfluidic platforms, but also to reveal the critical experimental conditions when monitoring cascade enzymatic reactions. A simplified mechanistic model was developed to describe the enzymatic reaction of glucose oxidase and glucose in the presence of catalase inside a commercial microfluidic platform with integrated oxygen sensor spots. The proposed model allowed an easy and rapid identification of the reaction mechanism, kinetics and limiting factors. The effect of fluid flow and enzyme adsorption inside the microfluidic chip on the optical sensor response and overall monitoring capabilities of the presented platform was evaluated via computational fluid dynamics (CFD) simulations. Remarkably, the model predictions were independently confirmed for μL- and mL- scale experiments. It is expected that the mechanistic models will significantly contribute to the further promotion of μBRs in biocatalysis research and that the overall study will create a framework for screening and evaluation of critical system parameters, including sensor response, operating conditions, experimental and microbioreactor designs.

Introduction

A great variety of industrial products is produced via bio-based manufacturing processes. Before industrial application, screening, evaluation and further optimization of potentially interesting bio-based processes and/or microbial strains are the most resource-consuming steps. Therefore, various scale-down approaches have gained popularity, resulting in development and implementation of small scale reactors such as microbioreactors (μBRs) for the identification and characterization of bio-based reactions of interest prior to scale-up [[1], [2], [3]]. Moreover, extensive research and availability of sensing technologies that can be integrated inside the microfluidic platforms [[4], [5], [6]] open up the possibility for parallel operation [7] and continuous monitoring of various analytes [8] addressing the current needs of biotechnology [[9], [10], [11], [12]]. On-line monitoring of key system variables (dissolved O₂, pH, etc.) can provide a better understanding and be translated into improved control and optimization of the processes themselves [13]. Although optical O₂ sensors are extensively used in various macroscale applications [14], their integration into microfluidic devices faces challenges (e.g. biofouling, matrix stability) [6,15]. However, the compactness and easy handling of the optical sensors together with their non-invasive and non-destructive (no analyte consumption) quantitative response makes them attractive for in situ monitoring. In order to increase the successful application of μBRs, robust sensing technologies, appropriate software and automated data interpretation tools must be developed. This should significantly improve the exploitation of the performance, flexibility and capabilities of microfluidic platforms and the bioprocess itself by delivering information-rich experiments, as well as extracting as much information as possible from experimental data.

Modeling cascade enzymatic reactions is problematic from both experimental and mathematical points of view. Their kinetics and overall performance is strongly dependent on the scale, format and operation of the reactor. Therefore, reliable multi-analytical techniques are applied to identify and continuously monitor crucial system parameters required at the initial model development step for bio-based processes. Although glucose oxidase (GOx) is a well-studied and commonly used commercial enzyme, the reaction mechanism and kinetics vary considerably under different experimental conditions [[16], [17], [18]] and depending on the strain producing the enzyme [19]. Moreover, the use of deuterated glucose together with computer simulations [20] and an O₂ electrode [21] for an accurate reaction monitoring resulted in additional reactions necessary to complete the glucose-GOx mechanism.

The majority of enzymatic reactions have been studied and validated in large scale and batch operation mode where ideal mixing could be obtained. Lack of available experimental data and understanding behind continuous flow operation of μBRs, where mixing is mainly governed by diffusion and system heterogeneity, makes modeling cascade enzymatic reactions in microfluidic platforms difficult. Moreover, the complexity of bio-based processes requires sufficient relevant experimental data to guarantee reliability, accuracy, robustness and overall quality of the process model. Here, the oxidation reaction of glucose in the presence of GOx and catalase (Cat) was chosen as the example system for monitoring bio-processes inside a microfluidic platform (Fig. 1). The microreactor chip geometric characteristics, setup assembly and operation principles were previously described in [5] (referred as microreactor 2 type). The operation of the O₂ sensors integrated along the meander channel was based on the detection of molecular O₂ by a sensitive dye entrapped in a polymeric layer of the sensor spot.

In order to identify the reaction mechanism occurring inside the μBR, various reaction schemes were evaluated using a mechanistic model. The final model presented here is the result of a model structure selection method and solves non-linear ordinary differential equations (ODEs) which combine the biocatalytic kinetic term and the reactor performance. Several methods for kinetic model parameter estimation [[22], [23], [24]] were used as to guide for (bio)reaction model development and simulation results were compared with experimental data to validate the model accuracy. On-line reaction monitoring was possible by quantifying the O₂ production during catalyzed decomposition of H₂O₂ (Fig. 1) via integrated optical O₂ sensor spots along the meander channel. The off-line measurements of gluconic acid at the outlet provided by high-performance liquid chromatography (HPLC) were used as a set of reference data in the model structure selection step. The response of the integrated sensors in the proposed reaction conditions was validated by disturbing the dissolved O₂ concentration profile inside the microfluidic chip in the presence of excess amounts of catalase and H₂O₂. Moreover, the enzyme adsorption on the surface of the μBR (i.e. channel walls and sensor spots) was confirmed by means of computational fluid dynamics (CFD).