Search

A fast, automated and accurate millidevice for determination of the apparent solubility of proteins and impurities and different industrially relevant precipitating agents.

The biopharmaceutical industry is still running in batch mode, mostly because it is a highly regulated industry sector. In the past, sensors were not readily available and in-process control was mainly executed off-line. The most important product parameters are quantity, purity and potency besides adventitious agents and bioburden. There is increasing economic pressure on time-to-market and also on the environmental sustainability of biopharmaceutical manufacturing. New concepts for manufacturing using disposable single-use technologies and integrated bioprocessing will dominate the future of bioprocessing. In order to ensure the quality of pharmaceuticals initiatives such as Process Analytical Technologies, Quality by Design and Continuous Integrated Manufacturing have been established. The vision must be that these initiatives together with technology development pave the way for process automation and autonomous bioprocessing without any human intervention. Then a 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 context of monitoring and control. Starting from statistical treatment of single and multiple sensors, establishing soft sensors with predictive chemometrics and hybrid models. A scenario is described how to integrate soft sensors and predictive chemometrics into modern process control. This will be exemplified by selective downstream processing steps such as chromatography and membrane filtration, the most common unit operations for separation of biopharmaceuticals.

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%., The authors show the challenges to transfer real time monitoring systems based on highly sensitive sensors and predictive models from the process science lab to two sites. Even though the exact same sensor set‐up was used at all three sites, the main challenge was the different sensitivity of sensors and data deviations due to non‐standardized process protocols. Partially, such deviations were compensated by data pre‐processing. Adaptions of the models considering those challenges enabled monitoring of in‐house processes with good precision.

Fusion protein technologies to facilitate soluble expression, detection, or subsequent affinity purification in Escherichia coli are widely used but may also be associated with negative consequences. Although commonly employed solubility tags have a positive influence on titers, their large molecular mass inherently results in stochiometric losses of product yield. Furthermore, the introduction of affinity tags, especially the polyhistidine tag, has been associated with undesirable changes in expression levels. Fusion tags are also known to influence the functionality of the protein of interest due to conformational changes. Therefore, particularly for biopharmaceutical applications, the removal of the fusion tag is a requirement to ensure the safety and efficacy of the therapeutic protein. The design of suitable fusion tags enabling the efficient manufacturing of the recombinant protein remains a challenge. Here, we evaluated several N-terminal fusion tag combinations and their influence on product titer and cell growth to find an ideal design for a generic fusion tag. For enhancing soluble expression, a negatively charged peptide tag derived from the T7 bacteriophage was combined with affinity tags and a caspase-2 cleavage site applicable for CASPase-based fusiON (CASPON) platform technology. The effects of each combinatorial tag element were investigated in an integrated manner using human fibroblast growth factor 2 as a model protein in fed-batch lab-scale bioreactor cultivations. To confirm the generic applicability for manufacturing, seven additional pharmaceutically relevant proteins were produced using the best performing tag of this study, named CASPON-tag, and tag removal was demonstrated.

BACKGROUND: Downstream processing of soluble recombinant proteins from Escherichia coli is complicated by the need to access the intracellular product by cell disruption and to separate the target protein from impurities, particularly host cell protein, DNA, endotoxins and lipids. We previously demonstrated the ability of the E. coli X‐press strain to leak high amounts of product to the culture medium without sacrificing viability. In this case study, we assessed the economic and ecological benefit of this strain for downstream processing in direct comparison to the industrial standard E. coli BL21(DE3). Staphylococcal Protein A was used as a model protein. We performed recombinant protein production, primary recovery and capture by anion exchange chromatography at laboratory scale, and used the data obtained for estimating costs and resource consumption by economic modeling. RESULTS: After primary recovery, the X‐press process resulted in a 1.5‐fold higher product purity, a 150‐fold lower DNA, 3.5‐fold lower endotoxin and 3.4‐fold lower lipid load compared to BL21(DE3). Consequently, anion exchanger binding capacity was increased 2.7‐fold and purity and concentration of the eluate also was increased. Extracellular protein production with X‐press resulted in a 25% reduction of costs and a 36% reduction of both water usage and water‐related CO₂ emissions compared to intracellular production with BL21(DE3). CONCLUSIONS: This case study performed with stapyhlococcal Protein A demonstrated the potential of E. coli X‐press to reduce costs for downstream processing and improve the environmental footprint by simplified primary recovery, lower impurity load and consequently higher chromatographic efficiency. © 2021 The Authors. Journal of Chemical Technology & Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

BACKGROUND: Recombinant proteins produced for use as biopharmaceuticals need to harbor their native N‐terminus. A drawback in expression of recombinant proteins as fusion proteins with an affinity fusion‐tag is that enzymatic or chemical processing is required to trim the artificial tag and release the true protein of interest. In many cases, however, this processing step generates an incorrect N‐terminus. RESULTS: Human fibroblast growth factor 2 (FGF2) was expressed as a fusion protein in Escherichia coli fed‐batch cultivations. The protein of interest (POI) carried an N‐terminal affinity fusion‐tag which enabled purification via affinity chromatography. After enzymatic removal of the affinity fusion‐tag with a circularly permuted human caspase‐2 (cpCasp2), the POI was further purified using subtractive affinity chromatography. Mass spectrometric analysis confirmed the authentic N‐terminus of the POI. The generated POI was highly pure with 42 ppm host cell protein, 3.7 μg mL⁻¹ dsDNA and ~ 1000 EU mL⁻¹ endotoxin. Only a small number of E. coli host cell proteins were co‐purified with the POI. Because of the high specificity of the novel protease cpCasp2, no off‐target cleavage could be observed. CONCLUSION: Our findings demonstrate that cpCasp2 can be used for the production of native proteins using a fusion‐protein process. This represents a first case study at large laboratory scale for the production of an industrially relevant protein. This technology constitutes the basis of a highly scalable cpCasp2‐based platform fusion protein process (CASPON technology) purification platform. © 2021 The Authors. Journal of Chemical Technology & Biotechnology published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Fusion protein technologies improve the expression and purification of recombinant proteins, but the removal of the tags involved requires specific proteases. The circularly permuted caspase-2 (cpCasp2) with its specific cleavage site, efficiently generates the untagged protein. While cleavage with cpCasp2 is possible before all 20 proteinogenic amino acids, cleavage before valine, leucine, isoleucine, aspartate and glutamate suffers from slow, and before proline extremely slow, turnover. To make the platform fusion protein process even more general such that any protein with an authentic N-terminus can be produced with high efficiency, the bacterial selection system PROFICS (PRotease Optimization via Fusion-Inhibited Carbamoyltransferase-based Selection) was used to evolve cpCasp2 into a variant with a catalytic turnover two orders of magnitude higher and the ability to cleave before any amino acid. The high specificity and the stability of the original circularly permuted protease was fully retained in this mutant, while the high manufacturability was mostly retained, albeit with decreased soluble titer. Four point-mutations are responsible for this change in activity, two of which are located in or near the binding pocket of the active site. This variant was named CASPON enzyme and is a major component of the CASPase-based fusiON (CASPON) platform technology. Applicability for the production of recombinant proteins was demonstrated by enzymatic removal of the CASPON tag from five model proteins. The CASPON tag enables high soluble expressions, affinity purification and good accessibility for cleavage. The five industry-relevant proteins of interest were FGF2, TNF, GH, GCSF and PTH.

Perfusion bioreactors are commonly used for the continuous production of monoclonal antibodies (mAb). One potential benefit of continuous bioprocessing is the ability to operate under steady-state conditions for an extended process time. However, the process performance is often limited by the feedback control of feed, harvest, and bleed flow rates. If the future behavior of a bioprocess can be adequately described, predictive control can reduce set point deviations and thereby maximize process stability. In this study, we investigated the predictive control of biomass in a perfusion bioreactor integrated to a non-chromatographic capture step, in a series of Monte-Carlo simulations. A simple algorithm was developed to estimate the current and predict the future viable cell concentrations (VCC) of the bioprocess. This feature enabled the single prediction controller (SPC) to compensate for process variations that would normally be transported to adjacent units in integrated continuous bioprocesses (ICB). Use of this SPC strategy significantly reduced biomass, product concentration, and harvest flow variability and stabilized the operation over long periods of time compared to simulations using feedback control strategies. Additionally, we demonstrated the possibility of maximizing product yields simply by adjusting perfusion control strategies. This method could be used to prevent savings in total product losses of 4.5–10% over 30 days of protein production.

Pre-packed columns have been increasingly used in process development and biomanufacturing thanks to their ease of use and consistency. Traditionally, packing quality is predicted through rate models, which require extensive calibration efforts through independent experiments to determine relevant mass transfer and kinetic rate constants. Here we propose machine learning as a complementary predictive tool for column performance. A machine learning algorithm, extreme gradient boosting, was applied to a large data set of packing quality (plate height and asymmetry) for pre-packed columns as a function of quantitative parameters (column length, column diameter, and particle size) and qualitative attributes (backbone and functional mode). The machine learning model offered excellent predictive capabilities for the plate height and the asymmetry (90 and 93%, respectively), with packing quality strongly influenced by backbone (∼70% relative importance) and functional mode (∼15% relative importance), well above all other quantitative column parameters. The results highlight the ability of machine learning to provide reliable predictions of column performance from simple, generic parameters, including strategic qualitative parameters such as backbone and functionality, usually excluded from quantitative considerations. Our results will guide further efforts in column optimization, for example, by focusing on improvements of backbone and functional mode to obtain optimized packings.

The baculovirus expression vector system is a very powerful tool to produce virus‐like particles and gene‐therapy vectors, but the removal of coexpressed baculovirus has been a major barrier for wider industrial use. We used chimeric human immunodeficiency virus‐1 (HIV‐1) gag influenza‐hemagglutin virus‐like particles produced in Tnms42 insect cells using the baculovirus insect cell expression vector system as model virus‐like particles. A fast and simple purification method for these virus‐like particles with direct capture and purification within one chromatography step was developed. The insect cell culture supernatant was treated with endonuclease and filtered, before it was directly loaded onto a polymer‐grafted anion exchanger and eluted by a linear salt gradient. A 4.3 log clearance of baculovirus from virus‐like particles was achieved. The absence of the baculovirus capsid protein (vp39) in the product fraction was additionally shown by high performance liquid chromatography‐mass spectrometry. When considering a vaccination dose of 109 particles, 4200 doses can be purified per L pretreated supernatant, meeting the requirements for vaccines with

The capture of recombinant antibodies from cell culture broth is the first critical step of downstream processing. We were able to develop a precipitation-based method for the capture and purification of monoclonal antibodies based on divalent cations, namely ZnCl2. Traditional precipitation processes have to deal with high dilution factors especially for resolubilization and higher viscosity due to the use of PEG as precipitation or co-precipitation agent. By the use of the crosslinking nature of divalent cations without the use of PEG, we kept viscosity from the supernatant and resolubilization dilution factors very low. This is especially beneficial for the solid–liquid separation for the harvest and wash of the precipitate in continuous mode. For this harvest and wash, we used tangential flow filtration that benefits a lot from low viscosity solutions, which minimizes the membrane fouling. With this precipitation based on ZnCl2, we were able to implement a very lean and efficient process. We demonstrated precipitation studies with three different antibodies, Adalimumab, Trastuzumab, and Denosumab, and a continuous capture case study using tangential flow filtration for precipitate recovery. In this study, we achieved yields of 70%.

A narrow residence time distribution (RTD) is highly desirable for continuous processes where a strict incubation time must be ensured, such as continuous virus inactivation. A narrow RTD also results in faster startup and shut down phases and limits the broadening of potential disturbances in continuous processes. A packed bed reactor with non-porous inert beads was developed to achieve narrow RTDs. The performance was defined as the ratio between the onset of the cumulative RTD and the median residence time (tx%/t50%). Laboratory-scale packed columns were used to study the influence of the column parameters on the RTD. A larger column with a void volume of 0.65 L and a length of 89 cm, packed with beads in a size range of 125 to 250 μm, achieved t0.5%/t50% >0.93 across flow rates from 0.1 to 9.8 mL/min. The RTD was significantly narrower than the RTDs of other reactor designs, such as the Coiled Flow Inverter and Jig in a Box. The pressure drop remained under 3 kPa for all tested flow rates. Fluorescent nanoparticles (30 and 200 nm) were used to mimic viruses. These two sizes showed less than 2% difference in terms of t1%/t50% and t0.01%/t50% scores. These results indicated that viruses travelled through the column at rates independent of size. This proposal of packed beds as incubation chambers for continuous virus inactivation is simple, scalable, and can be realized as single-use devices. Due to the low pressure drop, the system can be easily integrated into a fully continuous process.

Continuous virus inactivation (VI) has received little attention in the efforts to realize fully continuous biomanufacturing in the future. Implementation of continuous VI must assure a specific minimum incubation time, typically 60 min. To guarantee the minimum incubation time, we implemented a packed bed continuous viral inactivation reactor (CVIR) with narrow residence time distribution (RTD) for low pH incubation. We show that the RTD does not broaden significantly over a wide range of linear flow velocities—which highlights the flexibility and robustness of the design. Prolonged exposure to acidic pH has no impact on bed stability, assuring constant RTD throughout long term operation. The suitability of the packed bed CVIR for low pH inactivation is shown with two industry‐standard model viruses, that is xenotropic murine leukemia virus and pseudorabies virus. Controls at neutral pH showed no system‐induced VI. At low pH, significant VI is observed, even after only 15 min. Based on the low pH inactivation kinetics, the continuous process is equivalent to traditional batch operation. This study establishes a concept for continuous low pH inactivation and, together with previous reports, highlights the versatility of the packed bed reactor for continuous VI, regardless of the inactivation method., Virus inactivation is essential for product safety however its implementation in continuous mode remains a challenge while the biopharma industry aims at continuous biomanufacturing. This work describes the full viral inactivation of two model viruses by low pH using a packed‐bed reactor with narrow residence time distribution as incubation reactor.This novel approach enables an truly continuous operation while being as effective as the current batch approach.

The CASPON enzyme became an interesting enzyme for fusion protein processing because it generates an authentic N-terminus. However, the high cysteine content of the CASPON enzyme may induce aggregation via disulfide-bond formation, which can reduce enzymatic activity and be considered a critical quality attribute. Different multimerization states of the CASPON enzyme were isolated by preparative size exclusion chromatography and analyzed with respect to multimerization propensity and enzymatic activity. The impact of co-solutes on multimerization was studied in solution and in adsorbed state. Furthermore, protein–protein interactions in the presence of different co-solutes were measured by self-interaction chromatography and were then correlated to the multimerization propensity. The dimer was the most stable and active species with 50% higher enzymatic activity than the tetramer. Multimerization was mainly governed by a cysteine-mediated pathway, as indicated by DTT-induced reduction of most caspase multimers. In the presence of ammonium sulfate, attractive protein–protein interactions were consistent with those observed for higher multimerization when the cysteine-mediated pathway was followed. Multimerization was also observed under attractive conditions on a chromatographic stationary phase. These findings corroborate common rules to perform protein purification with low residence time to avoid disulfide bond formation and conformational change of the protein upon adsorption.

Purification of very large and complex, enveloped viruses, such as measles virus is very challenging, it must be performed in a closed system because the final product cannot be sterile filtered and often loss of virus titer and poor product purity has been observed. We developed a purification process where the clarified and endonuclease treated culture supernatant is loaded on a restricted access chromatography medium where small impurities are bound and the virus is collected in the flow-through, which is then concentrated, and buffer exchanged by ultra/diafiltration. Up to 98.5% of host cell proteins could be captured by direct loading of clarified and endonuclease treated cell culture supernatant. Reproducible process performance and scalability of the chromatography step were demonstrated from small to pilot scale, including loading volumes from 50 mL up to 9 L. A 10-fold virus concentration was achieved by the ultrafiltration using a 100 kDa flat-sheet membrane. The order of individual process steps had a large impact on the virus infectivity and total process yields. The developed process maintained virus infectivity and is twice as fast as the traditional process train, where concentration is performed before loading on the chromatography column. Capturing impurities by the restricted access medium makes it a platform purification process with a high flexibility, which can be easily and quickly adapted to other vectors based on the measles virus vector platform.

The trend in the biopharmaceutical industry is changing from batch process to continuous process. For continuous biomanufacturing, traceability of the material is required by regulatory authorities. The recent ICH draft guideline Q13 on continuous manufacturing of drug substances and drug products requests an "understanding of process dynamics as a function of input material attributes (e.g., potency, material flow properties), process conditions (e.g., mass flow rates) … One common approach is characterization of residence time distribution (RTD) for the individual unit operations and integrated system." Thus, it is necessary to trace material through individual continuous unit operations and the integrated process. The RTD of a process is obtained experimentally by injecting a pulse of an inert tracer into the inlet and measuring the broadening of the injected pulse in the outlet. We investigated the RTD of three-column periodic counter-current chromatography (PCC) using staphylococcal protein A affinity chromatography, with a focus on how the material distributes over subsequent cycles. A fluorescent-labeled antibody was used as the inert tracer under high salt concentration. The tracer was injected once in each run but at different points of the loading phase. We then analyzed the outlet of the column. In the elution phase, regardless of the point of injection, we observed an even distribution of the tracer. In the loading phase, a constant exchange between the antibody in the solid phase and the liquid phase was observed, meaning that sending the outlet of one chromatography column into another column to improve resin utilization causes higher residence time in the system for some portion of the material.

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.

Proteases serve as important tools in biotechnology and as valuable drugs or drug targets. Efficient protein engineering methods to study and modulate protease properties are thus of great interest for a plethora of applications. We established PROFICS (PRotease Optimization via Fusion-Inhibited Carbamoyltransferase-based Selection), a bacterial selection system, which enables the optimization of proteases for biotechnology, therapeutics or diagnosis in a simple overnight process. During the PROFICS process, proteases are selected for their ability to specifically cut a tag from a reporter enzyme and leave a native N-terminus. Precise and efficient cleavage after the recognition sequence reverses the phenotype of an Escherichia coli knockout strain deficient in an essential enzyme of pyrimidine synthesis. A toolbox was generated to select for proteases with different preferences for P1′ residues (the residue immediately following the cleavage site). The functionality of PROFICS is demonstrated with viral proteases and human caspase-2. PROFICS improved caspase-2 activity up to 25-fold after only one round of mutation and selection. Additionally, we found a significantly improved tolerance for all P1′ residues caused by a mutation in a substrate interaction site. We showed that this improved activity enables cells containing the new variant to outgrow cells containing all other mutants, facilitating its straightforward selection. Apart from optimizing enzymatic activity and P1′ tolerance, PROFICS can be used to reprogram specificities, erase off-target activity, optimize expression via tags/codon usage, or even to screen for potential drug-resistance-conferring mutations in therapeutic targets such as viral proteases in an unbiased manner.

The aim of this study was to semi-automate process analytics for the quantification of common impurities in downstream processing such as host cell DNA, host cell proteins and endotoxins using a commercial liquid handling station. By semi-automation, the work load to fully analyze the elution peak of a purification run was reduced by at least 2.41 h. The relative standard deviation of results among different operators over a time span of up to 6 months was at the best reduced by half, e.g. from 13.7 to 7.1% in dsDNA analysis. Automation did not improve the reproducibility of results produced by one operator but released time for data evaluation and interpretation or planning of experiments. Overall, semi-automation of process analytics reduced operator-specific influence on test results. Such robust and reproducible analytics is fundamental to establish process analytical technology and get downstream processing ready for Quality by Design approaches.

Protein and in particular antibody precipitation by PEG is a cost‐effective alternative for the first capture step. The 3D structure of precipitates has a large impact on the process parameters for the recovery and dissolution, but current technologies for determination of precipitate structures are either very time consuming (cryo‐TEM) or only generate an average fractal dimension (light scattering). We developed a light microscopy based reconstruction of 3D structures of individual particles with a resolution of 0.1–0.2 µm and used this method to characterize particle populations generated by batch as well as continuous precipitation in different shear stress environments. The resulting precipitate structures show a broad distribution in terms of fractal dimension. While the average fractal dimension is significantly different for batch and continuous precipitation, the distribution is broad and samples overlap significantly. The precipitate flocs were monofractal from micro‐ to nanoscale showing a random but consistent nature of precipitate formation. We showed that the fractal dimension and 3D reconstruction is a valuable tool for characterization of protein precipitate processes. The current switch from batch to continuous manufacturing has to take the 3D structure and population of different protein precipitates into account in their design, engineering, and scale up.

International audience; Polymer-grafted chromatography media, especially ion exchangers, are high performance materials for protein purification. However, due to the pore size limitation, conventional chromatography beads are usually not considered for the downstream processing of large biomolecules such as virus-like particles (VLPs). Contrariwise, since the outer surface of the chromatography beads provides satisfactory binding capacity for VLPs and impurities of smaller size can bind inside of the beads, conventional porous beads should be considered for VLP capture and purification. We used HIV-1 gag VLPs with a diameter of 100-200 nm as a model to demonstrate that polymer-grafted anion exchangers are suitable for the purification of bionanoparticles. The equilibrium binding capacity was 1 × 1013 part/mL resin. Moderate salt concentration up to 100 mM NaCl did not affect binding, allowing direct loading of cell culture supernatant onto the column for purification. Dynamic binding capacity at 10% breakthrough, when loading cell culture supernatant, was approximately 6 × 1011 part/mL column; only 1-log lower than for monoliths. Endonuclease treatment of the cell culture supernatant did not increase the dynamic binding capacity, suggesting that dsDNA does not compete for the binding sites of VLPs. Nevertheless, due to simultaneous elution of particles and dsDNA, endonuclease treatment is required to reduce dsDNA contamination in the product. Proteomic analysis revealed that HIV-1 gag VLPs contain different host cell proteins in their cargo. This cargo is composed of conserved proteins and other proteins that vary from one particle population to another, as well as from batch to batch. This process allowed the separation of different particle populations. HIV-1 gag VLPs were directly captured and purified from cell culture supernatant with a total particle recovery in the elution of about 35%. Columns packed with beads can be scaled to practically any dimension and therefore a tailored design of the process is possible.

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.

Pre-packed chromatography columns are routinely used in downstream process development and scale-down studies. In recent years they have also been widely adopted for large scale, cGMP manufacturing of biopharmaceuticals. Despite columns being qualified at their point of manufacture before release for sale, the suitability of pre-packed chromatography columns for protein separations at different scales has not yet been demonstrated. In this study, we demonstrated that the performance results obtained with small scale columns (0.5 cm diameter × 5 cm length, 1 mL column volume) are scalable to production sized columns (60 cm diameter × 20 cm length, 57 L column volume). The columns were characterized with acetone and blue dextran pulses to determine the packing density and packed bed consistency. Chromatography performance was evaluated with breakthrough curves including capacity measurements and with separation of a ternary protein mixture (lysozyme, cytochrome C and RNase A) with a step gradient. The equilibrium binding capacity and dynamic binding capacity were equivalent for all columns. The step gradient separation of the ternary protein mixture displayed similar peak profiles when normalized in respect to column volume and the eluted protein pools had the same purities for all scales. Scalable performance of pre-packed columns is demonstrated but as with conventionally packed columns the influence of extra column volume and system configurations, especially buffer mixing, must be taken into account when comparing separations at different scales.

Hydrophobic interaction chromatography is a very popular chromatography method for purification of proteins and plasmids in all scales from analytical to industrial manufacturing. Despite this frequent use, the complex interaction mechanism and the thermodynamic aspects of adsorption in hydrophobic interaction chromatography are still not well understood. Calorimetric methods such as isothermal titration calorimetry and flow calorimetry can help to gain a deeper understanding of the adsorption strength, the influence of salt type and temperature. They can be used to study conformational changes of proteins, which are often associated with the adsorption in hydrophobic interaction chromatography. This review offers a detailed introduction into the thermodynamic fundamentals of adsorption in hydrophobic interaction chromatography with a special focus on the potential applications of isothermal titration calorimetry and flow calorimetry for studying specific problems and relationships of the adsorption behavior of proteins and its various influencing factors. Models for characterizing conformational changes upon adsorption are presented together with methods for assessing this problem for different proteins and stationary phases. All of this knowledge can contribute greatly to forming a sound basis for method development, process optimization and finding modelling strategies in hydrophobic interaction chromatography.

In preparative and industrial chromatography, the current viewpoint is that the dynamic binding capacity governs the process economy, and increased dynamic binding capacity and column utilization are achieved at the expense of productivity. The dynamic binding capacity in chromatography increases with residence time until it reaches a plateau, whereas productivity has an optimum. Therefore, the loading step of a chromatographic process is a balancing act between productivity, column utilization, and buffer consumption. This work presents an online optimization approach for capture chromatography that employs a residence time gradient during the loading step to improve the traditional trade-off between productivity and resin utilization. The approach uses the extended Kalman filter as a soft sensor for product concentration in the system and a model predictive controller to accomplish online optimization using the pore diffusion model as a simple mechanistic model. When a soft sensor for the product is placed before and after the column, the model predictive controller can forecast the optimal condition to maximize productivity and resin utilization. The controller can also account for varying feed concentrations. This study examined the robustness as the feed concentration varied within a range of 50%. The online optimization was demonstrated with two model systems: purification of a monoclonal antibody by protein A affinity and lysozyme by cation-exchange chromatography. Using the presented optimization strategy with a controller saves up to 43% of the buffer and increases the productivity together with resin utilization in a similar range as a multi-column continuous counter-current loading process.

Multi-column counter-current chromatography is an advanced technology used for continuous capture processes to improve process productivity, resin capacity utilization and product consistency. However, process development is difficult due to process complexity. In this work, some general and convenient guidances for three-column periodic counter-current chromatography (3C-PCC) were developed. Boundaries and distributions of operating windows of 3C-PCC processes were clarified by model-based predictions. Interactive effects of feed concentration (c

Continuous chromatography is a promising technology for downstream processing of biopharmaceuticals. The operation of continuous processes is significantly different to batch-mode chromatography and needs comprehensive evaluation. In this work, the performances of four Protein A affinity resins were studied systematically for twin-column continuous capture processes. A model-based approach was used to evaluate the process performance (productivity and capacity utilization) under varying operation conditions, and the objective was to reveal the crucial resin properties for continuous capture. The trade-off between productivity and capacity utilization was found, and it is necessary to select appropriate resins for different feedstock and operation conditions. The capacity utilization heavily depends on mass transfer, and steep breakthrough curves are favorable for high capacity utilization. The productivity is determined by both equilibrium binding capacity and mass transfer, and the balance of feed amount and feed time is critical. Moreover, the influence of binding capacity and mass transfer on process productivity and parameter sensitivity with two important resin properties (equilibrium binding capacity qmax and effective pore diffusion coefficient De) were assessed by the model, and suitable resin parameter ranges for twin-column continuous capture were determined. The model-based approach is an effective and useful tool to evaluate the complex performance of different resins and guide the design of next-generation resins for continuous processes.

Chemically defined media are reconstituted batchwise and stored in hold tanks until use. To avoid large hold tanks and batchwise production of media, we developed continuous on‐demand reconstitutions directly from solids consisting of a hopper and a screw conveyor capable of feeding dry powdered media with the required precision ±5% at low dosing rates of 0.171 g min−1. A commercially available dry powdered cell culture medium was continuously fed over a duration of 12 h into a mixer which was connected to a UV‐cell for monitoring and the media were compared to a batchwise production. A comparable amino acid, carbohydrate, and osmolality profile to a batchwise reconstitution could be obtained. Cell cultivation showed comparable performance of batch and continuous reconstitution for two CHO cell lines producing the antibodies adalimumab and trastuzumab on a small and benchtop scale. In‐depth analysis of the produced antibodies showed the same glycosylation pattern, other posttranslational profiles such as methionine oxidation and deamidation compared to batchwise reconstitution. Therefore, we conclude a continuous reconstitution of the medium results in the same quality of the product. A continuous on‐demand media reconstitution will impact the supply chain and significantly reduce the floor space necessary for preparation and storage., In this study, we demonstrated a radical strategy for a continuous on‐demand reconstitution of a chemically defined media, which provides a solution for shrinking auxiliary buffer and media tanks necessary in continuous biomanufacturing. By using this strategy, we showed comparable performance and product quality of a CHO cell line expressing a monoclonal antibody.

The nucleocapsid protein (NP) of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is critical for several steps of the viral life cycle, and is abundantly expressed during infection, making it an ideal diagnostic target protein. This protein has a strong tendency to dimerization and interaction with nucleic acids. A native hydrophobic interaction chromatography hyphenated to multi-angle light scattering detection (HIC-MALS) method was established for in-process control, in particular, to monitor product fragmentation and multimerization throughout the purification process. High titers of the nucleocapsid protein were expressed in E. coli with a CASPON tag, using a growth-decoupled protein expression system. Purification was accomplished by nuclease treatment of the cell homogenate and a sequence of chromatographic steps. 730 mg purified NP per liter of fermentation could be produced by the optimized process, corresponding to a yield of 77%. The HIC-MALS method was used to demonstrate that the NP product can be produced with a purity of 95%. The molecular mass of the main NP fraction is consistent with dimerized protein as was verified by a complementary native size-exclusion separation (SEC)-MALS analysis. Peptide mapping mass spectrometry and host cell specific enzyme-linked immunosorbent assay confirmed the high product purity, and the presence of a minor endogenous chaperone explained the residual impurities. The HIC-MALS method enables to monitor the purity of the product and simultaneously access its molecular mass.

Deploying two salts in hydrophobic interaction chromatography can significantly increase dynamic binding capacities. Nevertheless, the mechanistic understanding of this phenomenon is lacking. Here, we investigate whether surface tension or ionic strength govern dynamic binding capacities of the chromatographic resin Toyopearl Butyl-650 M in dual salt systems. Small-angle X-ray scattering was employed to analyze the model proteins and the protein-resin adduct in the respective dual salt systems. The dual salt systems incorporate sodium citrate and a secondary sodium salt (acetate, sulfate, or phosphate). As model proteins, we used lysozyme, GFP, and a monoclonal antibody (adalimumab). Moreover, for the protein-resin adduct, we determined the model parameters of a self-avoiding random walk model fitted into the pair density distribution function of the SAXS data. Ionic strength is more predictive for dynamic binding capacities in HIC dual salt systems than surface tension. However, dynamic binding capacities still differ by up to 30 % between the investigated dual salt systems. The proteins exhibit extensive protein-protein interactions in the studied dual salt HIC buffers. We found a correlation of protein-protein interactions with the well-known Hofmeister series. For systems with elevated protein-protein interactions, adsorption isotherms deviate from Langmuirian behavior. This highlights the importance of lateral protein-protein interactions in protein adsorption, where monomolecular protein layers are usually assumed. SAXS analysis of the protein-resin adduct indicates an inverse correlation of the binding capacity and the excluded volume parameter. This is indicative of the deposition of proteins in the cavities of the stationary phase. We hypothesize that increasing protein-protein interactions allow the formation of attractive clusters and multilayers in the cavities, respectively.