Comprehensive Comparison of Baculoviral and Plasmid Gene Delivery in Mammalian Cells.

(1) Recombinant protein production in mammalian cells is either based on transient transfection processes, often inefficient and underlying high batch-to-batch variability, or on laborious generation of stable cell lines. Alternatively, BacMam, a transduction process using the baculovirus, can be employed. (2) Six transfecting agents were compared to baculovirus transduction in terms of transient and stable protein expression characteristics of the model protein ACE2-eGFP using HEK293-6E, CHO-K1, and Vero cell lines. Furthermore, process optimization such as expression enhancement using sodium butyrate and TSA or baculovirus purification was assessed. (3) Baculovirus transduction efficiency was superior to all transfection agents for all cell lines. Transduced protein expression was moderate, but an 18-fold expression increase was achieved using the enhancer sodium butyrate. Ultracentrifugation of baculovirus from a 3.5 L bioreactor significantly improved the transduction efficiency and protein expression. Stable cell lines were obtained with each baculovirus transduction, yet stable cell line generation after transfection was highly unreliable. (4) This study demonstrated the superiority of the BacMam platform to standard transfections. The baculovirus efficiently transduced an array of cell lines both transiently and stably and achieved the highest efficiency for all tested cell lines. The feasibility of the scale-up of baculovirus production was demonstrated and the possibility of baculovirus purification was successfully explored.

Explainable deep learning enhances robust and reliable real-time monitoring of a chromatographic protein A capture step.

The application of model-based real-time monitoring in biopharmaceutical production is a major step toward quality-by-design and the fundament for model predictive control. Data-driven models have proven to be a viable option to model bioprocesses. In the high stakes setting of biopharmaceutical manufacturing it is essential to ensure high model accuracy, robustness, and reliability. That is only possible when (i) the data used for modeling is of high quality and sufficient size, (ii) state-of-the-art modeling algorithms are employed, and (iii) the input-output mapping of the model has been characterized. In this study, we evaluate the accuracy of multiple data-driven models in predicting the monoclonal antibody (mAb) concentration, double stranded DNA concentration, host cell protein concentration, and high molecular weight impurity content during elution from a protein A chromatography capture step. The models achieved high-quality predictions with a normalized root mean squared error of <4% for the mAb concentration and of ≈10% for the other process variables. Furthermore, we demonstrate how permutation/occlusion-based methods can be used to gain an understanding of dependencies learned by one of the most complex data-driven models, convolutional neural network ensembles. We observed that the models generally exhibited dependencies on correlations that agreed with first principles knowledge, thereby bolstering confidence in model reliability. Finally, we present a workflow to assess the model behavior in case of systematic measurement errors that may result from sensor fouling or failure. This study represents a major step toward improved viability of data-driven models in biopharmaceutical manufacturing.

Sensors and chemometrics in downstream processing.

The biopharmaceutical industry is still running in batch mode, mostly because it is highly regulated. In the past, sensors were not readily available and in-process control was mainly executed offline. The most important product parameters are quantity, purity, and potency, in addition to adventitious agents and bioburden. New concepts using disposable single-use technologies and integrated bioprocessing for manufacturing will dominate the future of bioprocessing. To ensure the quality of pharmaceuticals, initiatives such as Process Analytical Technologies, Quality by Design, and Continuous Integrated Manufacturing have been established. The aim is that these initiatives, together with technology development, will pave the way for process automation and autonomous bioprocessing without any human intervention. Then, real-time release would be realized, leading to a highly predictive and robust biomanufacturing system. The steps toward such automated and autonomous bioprocessing are reviewed in the context of monitoring and control. It is possible to integrate real-time monitoring gradually, and it should be considered from a soft sensor perspective. This concept has already been successfully implemented in other industries and requires relatively simple model training and the use of established statistical tools, such as multivariate statistics or neural networks. This review describes a scenario for integrating soft sensors and predictive chemometrics into modern process control. This is exemplified by selective downstream processing steps, such as chromatography and membrane filtration, the most common unit operations for separation of biopharmaceuticals.

Mass transfer of proteins in chromatographic media: Comparison of pure and crude feed solutions.

Elucidation of intraparticle mass transfer mechanisms in protein chromatography is essential for process design. This study investigates the differences of adsorption and diffusion parameters of basic human fibroblast factor 2 (hFGF2) in a simple (purified) and a complex (clarified homogenate) feed solution on the grafted agarose-based strong cation exchanger Capto S. Microscopic investigations using confocal laser scanning microscopy revealed slower intraparticle diffusion of hFGF2 in the clarified homogenate compared to purified hFGF2. Diffusive adsorption fronts indicated a strong contribution of solid diffusion to the overall mass transfer flux. Protein adsorption methods such as batch uptake and shallow bed as well as breakthrough curve experiments confirmed a 40-fold reduction of the mass transfer flux for hFGF2 in the homogenate compared to pure hFGF2. The slower mass transfer was induced by components of the clarified homogenate. Essentially, the increased dynamic viscosity caused by a higher concentration of dsDNA and membrane lipids in the clarified homogenate contributed to this decrease in mass transfer. Moreover, binding capacity for hFGF2 was much lower in the clarified homogenate and substantially decreased the adsorbed phase driving force for mass transfer.

Characterization of hydrodynamics and volumetric power input in microtiter plates for the scale-up of downstream operations.

Parameter estimation for scale-up of downstream operations from microtiter plates (MTPs) is mostly done empirically because engineering correlations between microplates and stirred tank reactors (STRs) are not yet available. It is challenging to change the operation mode from shaken MTPs to large-scale STRs. For the scale-up of STRs, volumetric power input is well-established although it is unclear whether this parameter can be used to transfer the operations from MTPs. We determine the volumetric power input in MTPs via the temperature increase caused by the motion of the liquid. The hydrodynamics in MTPs are studied with computational fluid dynamics (CFD). Mixing is investigated in 96-, 24-, and 6-well MTPs to cover different geometries, filling volumes, shaking diameters, and shaking frequencies. All CFD simulations are validated by experimental results, which now allows prediction of the volumetric power input and hydrodynamics at various conditions in MTPs without the need for further experiments. We provide a map of the power input achievable in MTPs. Based on this map, from knowing about large-scale conditions, adequate microscale conditions can be adjusted for process development. This enables the direct scale-up of downstream unit operations from MTPs to STRs.

Technology transfer of a monitoring system to predict product concentration and purity of biopharmaceuticals in real-time during chromatographic separation.

Technological developments require the transfer to their location of application to make use of them. We describe the transfer of a real-time monitoring system for lab-scale preparative chromatography to two new sites where it will be used and developed further. Equivalent equipment was used. The capture of a biopharmaceutical model protein, human fibroblast growth factor 2 (FGF-2) was used to evaluate the system transfer. Predictive models for five quality attributes based on partial least squares regression were transferred. Six out of seven online sensors (UV/VIS, pH, conductivity, IR, RI, and MALS) showed comparable signals between the sites while one sensor (fluorescence) showed different signal profiles. A direct transfer of the models for real-time monitoring was not possible, mainly due to differences in sensor signals. Adaptation of the models was necessary. Then, among five prediction models, the prediction errors of the test run at the new sites were on average twice as high as at the training site (model-wise 0.9-5.7 times). Additionally, new prediction models for different products were trained at each new site. These allowed monitoring the critical quality attributes of two new biopharmaceutical products during their purification processes with mean relative deviations between 1% and 33%.

Scale up of a chromatographic capture step for a clarified bacterial homogenate - Influence of mass transport limitation and competitive adsorption of impurities.

A model-based approach for scaling up chromatographic capture step was developed. The purification of human basic fibroblast growth factor protein 2 (FGF2) from an E. coli homogenate on a cation exchange resin was selected as a case study. Non-ideal effects accompanying the capture operation were examined, including: reduction in the protein diffusivity in the presence of the homogenate, competitive adsorption between FGF2 and undefined impurities, and flow behavior in external column volumes. The viscosity of the homogenate was measured as a function of dilution degree and shear stress, and its contribution to the diffusivity reduction was quantified. A dynamic model was formulated which accounted for underlying kinetic and thermodynamic dependencies. The model parameters were determined for a lab scale system using a small 2-mL column. The model was successfully used to scale up the capture operation from the lab scale column to a preparative bench scale column of about 1 L volume.

Real-time monitoring and model-based prediction of purity and quantity during a chromatographic capture of fibroblast growth factor 2.

Process analytical technology combines understanding and control of the process with real-time monitoring of critical quality and performance attributes. The goal is to ensure the quality of the final product. Currently, chromatographic processes in biopharmaceutical production are predominantly monitored with UV/Vis absorbance and a direct correlation with purity and quantity is limited. In this study, a chromatographic workstation was equipped with additional online sensors, such as multi-angle light scattering, refractive index, attenuated total reflection Fourier-transform infrared, and fluorescence spectroscopy. Models to predict quantity, host cell proteins (HCP), and double-stranded DNA (dsDNA) content simultaneously were developed and exemplified by a cation exchange capture step for fibroblast growth factor 2 expressed in Escherichia coliOnline data and corresponding offline data for product quantity and co-eluting impurities, such as dsDNA and HCP, were analyzed using boosted structured additive regression. Different sensor combinations were used to achieve the best prediction performance for each quality attribute. Quantity can be adequately predicted by applying a small predictor set of the typical chromatographic workstation sensor signals with a test error of 0.85 mg/ml (range in training data: 0.1-28 mg/ml). For HCP and dsDNA additional fluorescence and/or attenuated total reflection Fourier-transform infrared spectral information was important to achieve prediction errors of 200 (2-6579 ppm) and 340 ppm (8-3773 ppm), respectively.

Prediction of the Quantity and Purity of an Antibody Capture Process in Real Time.

Regulatory recommendations for quality by design instead of quality by testing raise increasing interest in new sensor technologies. An online monitoring system for downstream processes is developed, which is based on an array of online detectors. Besides standard detectors (UV, pH, and conductivity), our chromatographic workstation is equipped with a fluorescence and a mid-infrared spectrometer, a light scattering, and a refractive index detector. The combination of these sensors enables the prediction of specific protein concentration and various purity attributes, such as high molecular weight impurities, DNA and host cell protein content during the elution phase of a chromatographic antibody capture process. Prediction models solely based on online signals are set up providing real-time predictions. No mechanistic models or information about the chromatographic runs is used. These predictions allow online pooling decisions replacing time- and labor-intensive laboratory measurements. Different process variations, such as changes in the column load or elution buffer, are introduced to test the predictive power of the models. Extrapolation of the models worked well when the column load is changed, whereas model adjustment is necessary when the elution conditions are changed considerably.

A two-step process for capture and purification of human basic fibroblast growth factor from E. coli homogenate: Yield versus endotoxin clearance.

A two-step purification process for human basic fibroblast growth factor (FGF-2) from clarified E. coli homogenate has been developed in which the impurity level after the second step is below the limit of quantification. Endotoxin content is cleared to 0.02 EU/µg FGF-2 and the overall yield is 67%. The performance of the cation exchanger Carboxymethyl-Sepharose Fast Flow (CM-SFF) was compared to the affinity resin Heparin-SFF regarding the impurity profile and product quality in the elution peak. The CM-SFF eluate was further purified using hydrophobic interaction resin Toyopearl-Hexyl-650C. The relative amounts of target product, host cell proteins (HCPs), dsDNA, endotoxin, monomer content, and high molecular weight impurities differed along the elution peak depending on the applied method. The bioactive monomer (>99%) was obtained with a yield of 48% for CM-SFF and 68% for Heparin-SFF. A half-load reduction in CM-SFF increased the yield up to 67% without deterioration of the impurity content. Assuming a dose of 400 µg FGF-2, endotoxin was reduced to 188 EU/dose, dsDNA <10 ng/dose, and HCP <2 ppm/dose using the cation exchanger. In the pooled eluate fractions, dsDNA was removed 4-fold (291 ng/mL) and endotoxin 14-fold (0.47 EU/µg FGF-2) more efficiently by CM-SFF than by affinity chromatography. In contrast, HCP clearance was 3-fold (13 ppm) more efficient with Heparin-SFF than CM-SFF. In contrast to process monitoring by UV280nm or SDS-PAGE, this characterization is the basis for a Process Analytical Technology attempt when correlated with online monitored signals, as it enables knowledge-based pooling according to defined quality criteria.

Prediction tool for loading, isocratic elution, gradient elution and scaling up of ion exchange chromatography of proteins.

An efficient mathematical tool for the design and scaling up of protein chromatography is suggested, in which the model parameters can be determined quickly over a wide operating space without large material investments. The design method is based on mathematical modelling of column dynamics and moment analysis. The accuracy of the dynamic models that are most frequently used for simulations of chromatographic processes is analyzed, and possible errors that can be generated using the moment analysis are indicated. The so-called transport dispersive model was eventually employed for the process simulations. The model was modified to account for the protein dispersion in void volumes of chromatographic systems. The manner of the model calibration was suggested, which was based on a few chromatographic runs and verified over a wide space of the operating parameters, including composition and flow rate of the mobile phase, column dimensions, residence time, and mass loading. The model system for the study was ion-exchange chromatography. The analysis was performed based on the elution profiles of basic fibroblast growth factor 2 and lysozyme, on two different IEX media.

Influence of cavitation and high shear stress on HSA aggregation behavior.

Neither the influence of high shear rates nor the impact of cavitation on protein aggregation is fully understood. The effect of cavitation bubble collapse-derived hydroxyl radicals on the aggregation behavior of human serum albumin (HSA) was investigated. Radicals were generated by pumping through a micro-orifice, ultra-sonication, or chemically by Fenton's reaction. The amount of radicals produced by the two mechanical methods (0.12 and 11.25 nmol/(L min)) was not enough to change the protein integrity. In contrast, Fenton's reaction resulted in 382 nmol/(L min) of radicals, inducing protein aggregation. However, the micro-orifice promoted the formation of soluble dimeric HSA aggregates. A validated computational fluid dynamic model of the orifice revealed a maximum and average shear rate on the order of 108 s-1 and 1.2 × 106 s-1, respectively. Although these values are among the highest ever reported in the literature, dimer formation did not occur when we used the same flow rate but suppressed cavitation. Therefore, aggregation is most likely caused by the increased surface area due to cavitation-mediated bubble growth, not by hydroxyl radical release or shear stress as often reported.

Impact of Cavitation, High Shear Stress and Air/Liquid Interfaces on Protein Aggregation.

The reported impact of shear stress on protein aggregation has been contradictory. At high shear rates, the occurrence of cavitation or entrapment of air is reasonable and their effects possibly misattributed to shear stress. Nine different proteins (α-lactalbumin, two antibodies, fibroblast growth factor 2, granulocyte colony stimulating factor [GCSF], green fluorescence protein [GFP], hemoglobin, human serum albumin, and lysozyme) are tested for their aggregation behavior on vapor/liquid interfaces generated by cavitation and compared it to the isolated effects of high shear stress and air/liquid interfaces generated by foaming. Cavitation induced the aggregation of GCSF by +68.9%, hemoglobin +4%, and human serum albumin +2.9%, compared to a control, whereas the other proteins do not aggregate. The protein aggregation behaviors of the different proteins at air/liquid interfaces are similar to cavitation, but the effect is more pronounced. Air-liquid interface induced the aggregation of GCSF by +94.5%, hemoglobin +35.5%, and human serum albumin (HSA) +31.1%. The results indicate that the sensitivity of a certain protein toward cavitation is very similar to air/liquid-induced aggregation. Hence, hydroxyl radicals cannot be seen as the driving force for protein aggregation when cavitation occurs. Further, high shear rates of up to 108 s-1 do not affect any of the tested proteins. Therefore, also within this study generated extremely high isolated shear rates cannot be considered to harm structural integrity when processing proteins.

Integrated process development-a robust, rapid method for inclusion body harvesting and processing at the microscale level.

Escherichia coli stores large amounts of highly pure product within inclusion bodies (IBs). To take advantage of this beneficial feature, after cell disintegration, the first step to optimal product recovery is efficient IB preparation. This step is also important in evaluating upstream optimization and process development, due to the potential impact of bioprocessing conditions on product quality and on the nanoscale properties of IBs. Proper IB preparation is often neglected, due to laboratory-scale methods requiring large amounts of materials and labor. Miniaturization and parallelization can accelerate analyses of individual processing steps and provide a deeper understanding of up- and downstream processing interdependencies. Consequently, reproducible, predictive microscale methods are in demand. In the present study, we complemented a recently established high-throughput cell disruption method with a microscale method for preparing purified IBs. This preparation provided results comparable to laboratory-scale IB processing, regarding impurity depletion, and product loss. Furthermore, with this method, we performed a "design of experiments" study to demonstrate the influence of fermentation conditions on the performance of subsequent downstream steps and product quality. We showed that this approach provided a 300-fold reduction in material consumption for each fermentation condition and a 24-fold reduction in processing time for 24 samples.

Microscale disruption of microorganisms for parallelized process development.

Escherichia coli, Saccharomyces cerevisiae, and Pichia pastoris are the standard platforms for biopharmaceutical production with 40% of all between 2010 to 2014 approved protein drugs produced in those microbial hosts. Typically, products overexpressed E. coli and S. cerevisiae remain in the cytosol or are secreted into the periplasm. Consequently, efficient cell disruption is essential for high product recovery during microbial production. Process development platforms at microscale are essential to shorten time to market. While high-pressure homogenization is the industry standard for cell disruption at large scale this method is not practicable for experiments in microscale. This review describes microscale methods for cell disruption at scales as low as 200 µL. Strategies for automation, parallelization and miniaturization, as well as comparability of the results at this scale to high pressure homogenization are considered as those criteria decide which methods are most suited for scale down. Those aspects are discussed in detail for protein overexpression in E. coli and yeast but also the relevance for alternative products and host such as microalgae are taken into account. The authors conclude that bead milling is the best comparable microscale method to large scale high-pressure homogenization and therefore the most suitable technique for automated process development of microbial hosts with the exception of pDNA production.

Trend analysis of performance parameters of pre-packed columns for protein chromatography over a time span of ten years.

Pre-packed small scale chromatography columns are increasingly used for process development, for determination of design space in bioprocess development, and for post-licence process verifications. The packing quality of 30,000 pre-packed columns delivered to customers over a period 10 years has been analyzed by advanced statistical tools. First, the data were extracted and checked for inconsistencies, and then were tabulated and made ready for statistical processing using the programming language Perl (https://www.perl.org/) and the statistical computing environment R (https://www.r-project.org/). Reduced HETP and asymmetry were plotted over time to obtain a trend of packing quality over 10 years. The obtained data were used as a visualized coefficient of variation analysis (VCVA), a process that has often been applied in other industries such as semiconductor manufacturing. A typical fluctuation of reduced HETP was seen. A Tsunami effect in manufacturing, the effect of propagation of manufacturing deviations leading to out-of-specification products, was not observed with these pre-packed columns. Principal component analysis (PCA) showed that all packing materials cluster. Our data analysis showed that the current commercially available chromatography media used for biopharmaceutical manufacturing can be reproducibly and uniformly packed in polymer-based chromatography columns, which are designed for ready-to-use purposes. Although the number of packed columns has quadrupled over one decade the packing quality has remained stable.

Mixing at the microscale: Power input in shaken microtiter plates.

Power input and local energy dissipation are crucial parameters for the engineering characterization of mixing and fluid dynamics at the microscale. Since hydrodynamic stress is solely dependent on the maximum power input, we adapted the clay/polymer method to obtain flock destruction kinetics in six-, 24-, and 96-well microtiter plates on orbital shakers. We also determined the specific power input using calorimetry and found that the power input is at the same order of magnitude for the six- and 96-well plates and the laboratory-scale stirred tank reactor, with 40 to 90 W/m3 (Re' = 180 to 440), 40 to 140 W/m3 (Re' = 320 to 640), and 30 to 50 W/m3 (Re = 4000 to 8500), respectively. All of these values are significantly below 450 to 2100 W/m3 determined for the pilot-scale reactor. The hydrodynamic stress differs significantly between the different formats of MTPs, as the 96-well plates showed very low shear stress on the shaker with a shaking amplitude of 3 mm. Thus, the transfer of mixing conditions from the microtiter plate to small-scale and pilot-scale reactors must be undertaken with care. Our findings, especially the power input determined by the calorimetric method, show that the hydrodynamic conditions in laboratory- and pilot-scale reactors cannot be reached.

Prediction of inclusion body solubilization from shaken to stirred reactors.

Inclusion bodies (IBs) were solubilized in a µ-scale system using shaking microtiter plates or a stirred tank reactor in a laboratory setting. Characteristic dimensionless numbers for mixing, the Phase number Ph and Reynolds number Re did not correlate with the kinetics and equilibrium of protein solubilization. The solubilization kinetics was independent of the mixing system, stirring or shaking rate, shaking diameter, and energy input. Good agreement was observed between the solubilization kinetics and yield on the µ-scale and laboratory setting. We show that the IB solubilization process is controlled predominantly by pore diffusion. Thus, for the process it is sufficient to keep the IBs homogeneously suspended, and additional power input will not improve the process. The high-throughput system developed on the µ-scale can predict solubilization in stirred reactors up to a factor of 500 and can therefore be used to determine optimal solubilization conditions on laboratory and industrial scale.

Matrix-assisted refolding of autoprotease fusion proteins on an ion exchange column: a kinetic investigation.

Matrix-assisted refolding is an excellent technique for performing refolding of recombinant proteins at high concentration because aggregation during refolding is partially suppressed. The autoprotease N(pro) and its engineered mutant EDDIE can be efficiently refolded on cation-exchangers. In the current work, denatured fusion proteins were loaded at different column saturations (5 and 50 mg mL(-1) gel), and refolding and self-cleavage were initiated during elution. The contact time of the protein with the matrix significantly influenced the refolding rate and yield. On POROS 50 HS, the refolding rate was comparable to a batch refolding process, but yield was substantially higher; at a protein concentration of 1.55 mg mL(-1), an almost complete conversion was observed. With Capto S, the rate of self-cleavage increased by a factor of 20 while yield was slightly reduced. Processing the autoprotease fusion protein on Capto S at a high protein loading of 50 mg mL(-1) gel and short contact time (0.5h) yielded the highest productivity.

Refolding of Npro fusion proteins.

The autoprotease Npro significantly enhances expression of fused peptides and proteins and drives the formation of inclusion bodies during protein expression. Upon refolding, the autoprotease becomes active and cleaves itself specifically at its own C-terminus releasing the target protein with its authentic N-terminus. Npro wild-type and its mutant EDDIE, respectively, were fused N-terminally to the model proteins green fluorescent protein, staphylococcus Protein A domain D, inhibitory peptide of senescence-evasion-factor, and the short 16 amino acid peptide pep6His. In comparison with the Npro wild-type, the tailored mutant EDDIE displayed an increased rate constant for refolding and cleavage from 1.3 x 10(-4) s(-1) to 3.5 x 10(-4) s(-1), and allowed a 15-fold higher protein concentration of 1.1 mg/mL when studying pep6His as a fusion partner. For green fluorescent protein, the rate constant was increased from 2.4 x 10(-5) s(-1) to 1.1 x 10(-4) s(-1) when fused to EDDIE. When fused to small target peptides, refolding and cleavage yields were independent of initial protein concentration, even at high concentrations of 3.9 mg/mL, although cleavage rates were strongly influenced by the fusion partner. This behavior differed from conventional 1st order refolding kinetics, where yield strongly depends on initial protein concentration due to an aggregation reaction of higher order. Refolding and cleavage of EDDIE fusion proteins follow a monomolecular reaction for the autoproteolytic cleavage over a wide concentration range. At high protein concentrations, deviations from the model assumptions were observed and thus smaller rate constants were required to approximate the data.