Boosted structured additive regression for Escherichia coli fed-batch fermentation modeling.

The quality of biopharmaceuticals and patients' safety are of highest priority and there are tremendous efforts to replace empirical production process designs by knowledge-based approaches. Main challenge in this context is that real-time access to process variables related to product quality and quantity is severely limited. To date comprehensive on- and offline monitoring platforms are used to generate process data sets that allow for development of mechanistic and/or data driven models for real-time prediction of these important quantities. Ultimate goal is to implement model based feed-back control loops that facilitate online control of product quality. In this contribution, we explore structured additive regression (STAR) models in combination with boosting as a variable selection tool for modeling the cell dry mass, product concentration, and optical density on the basis of online available process variables and two-dimensional fluorescence spectroscopic data. STAR models are powerful extensions of linear models allowing for inclusion of smooth effects or interactions between predictors. Boosting constructs the final model in a stepwise manner and provides a variable importance measure via predictor selection frequencies. Our results show that the cell dry mass can be modeled with a relative error of about ±3%, the optical density with ±6%, the soluble protein with ±16%, and the insoluble product with an accuracy of ±12%. Biotechnol. Bioeng. 2017;114: 321-334. © 2016 Wiley Periodicals, Inc.

The potential of random forest and neural networks for biomass and recombinant protein modeling in Escherichia coli fed-batch fermentations.

Product quality assurance strategies in production of biopharmaceuticals currently undergo a transformation from empirical "quality by testing" to rational, knowledge-based "quality by design" approaches. The major challenges in this context are the fragmentary understanding of bioprocesses and the severely limited real-time access to process variables related to product quality and quantity. Data driven modeling of process variables in combination with model predictive process control concepts represent a potential solution to these problems. The selection of statistical techniques best qualified for bioprocess data analysis and modeling is a key criterion. In this work a series of recombinant Escherichia coli fed-batch production processes with varying cultivation conditions employing a comprehensive on- and offline process monitoring platform was conducted. The applicability of two machine learning methods, random forest and neural networks, for the prediction of cell dry mass and recombinant protein based on online available process parameters and two-dimensional multi-wavelength fluorescence spectroscopy is investigated. Models solely based on routinely measured process variables give a satisfying prediction accuracy of about ± 4% for the cell dry mass, while additional spectroscopic information allows for an estimation of the protein concentration within ± 12%. The results clearly argue for a combined approach: neural networks as modeling technique and random forest as variable selection tool.

Evaluation of three industrial Escherichia coli strains in fed-batch cultivations during high-level SOD protein production.

BACKGROUND: In the biopharmaceutical industry, Escherichia coli (E. coli) strains are among the most frequently used bacterial hosts for producing recombinant proteins because they allow a simple process set-up and they are Food and Drug Administration (FDA)-approved for human applications. Widespread use of E. coli in biotechnology has led to the development of many different strains, and selecting an ideal host to produce a specific protein of interest is an important step in developing a production process. E. coli B and K-12 strains are frequently employed in large-scale production processes, and therefore are of particular interest. We previously evaluated the individual cultivation characteristics of E. coli BL21 and the K-12 hosts RV308 and HMS174. To our knowledge, there has not yet been a detailed comparison of the individual performances of these production strains in terms of recombinant protein production and system stability. The present study directly compared the T7-based expression hosts E. coli BL21(DE3), RV308(DE3), and HMS174(DE3), focusing on evaluating the specific attributes of these strains in relation to high-level protein production of the model protein recombinant human superoxide dismutase (SOD). The experimental setup was an exponential carbon-limited fed-batch cultivation with minimal media and single-pulse induction. RESULTS: The host strain BL21(DE3) produced the highest amounts of specific protein, followed by HMS174(DE3) and RV308(DE3). The expression system HMS174(DE3) exhibited system stability by retaining the expression vector over the entire process time; however, it entirely stopped growing shortly after induction. In contrast, BL21(DE3) and RV308(DE3) encountered plasmid loss but maintained growth. RV308(DE3) exhibited the lowest ppGpp concentration, which is correlated with the metabolic stress level and lowest degradation of soluble protein fraction compared to both other strains. CONCLUSIONS: Overall, this study provides novel data regarding the individual strain properties and production capabilities, which will enable targeted strain selection for producing a specific protein of interest. This information can be used to accelerate future process design and implementation.

Implementation of proton transfer reaction-mass spectrometry (PTR-MS) for advanced bioprocess monitoring.

We report on the implementation of proton transfer reaction-mass spectrometry (PTR-MS) technology for on-line monitoring of volatile organic compounds (VOCs) in the off-gas of bioreactors. The main part of the work was focused on the development of an interface between the bioreactor and an analyzer suitable for continuous sampling of VOCs emanating from the bioprocess. The permanently heated sampling line with an inert surface avoids condensation and interaction of volatiles during transfer to the PTR-MS. The interface is equipped with a sterile sinter filter unit directly connected to the bioreactor headspace, a condensate trap, and a series of valves allowing for dilution of the headspace gas, in-process calibration, and multiport operation. To assess the aptitude of the entire system, a case study was conducted comprising three identical cultivations with a recombinant E. coli strain, and the volatiles produced in the course of the experiments were monitored with the PTR-MS. The high reproducibility of the measurements proved that the established sampling interface allows for reproducible transfer of volatiles from the headspace to the PTR-MS analyzer. The set of volatile compounds monitored comprises metabolites of different pathways with diverse functions in cell physiology but also volatiles from the process matrix. The trends of individual compounds showed diverse patterns. The recorded signal levels covered a dynamic range of more than five orders of magnitude. It was possible to assign specific volatile compounds to distinctive events in the bioprocess. The presented results clearly show that PTR-MS was successfully implemented as a powerful bioprocess-monitoring tool and that access to volatiles emitted by the cells opens promising perspectives in terms of advanced process control.