Chromatographic Purification: Data Science, Chemistry, and Process Engineering Are Driving Bioprocess Innovation Forward

When people decry “the lack of innovation” in biomanufacturing, I often find myself scratching my head. Maybe we’re working from different angles on the concept — for some, it might mean only those advances that are truly disruptive and replace a previous paradigm completely. But what I see in the biopharmaceutical industry currently is an explosion of new ideas, fresh approaches to established technologies, and incremental improvements that all add up to paint a dynamic picture of innovation indeed.

“Creativity and problem-solving are very much alive and well in this part of the industry,” said David Wood (professor of chemical and biomolecular engineering at Ohio State University) in moderating a Recovery and Purification session at Biotech Week Boston in September 2022. “I think they have to be for us to evolve, develop new modalities, and address new patient challenges.”

With upstream production groups outputting an increasingly diverse set of product streams, downstream process engineers have been under mounting pressure to devise and implement new purification solutions to the downstream bottleneck problem. Many groups are turning to technology suppliers and even academic researchers for help, and the results have been impressive so far — with more advancements to come.

Data Science
The intersection of information technology and life science brings many benefits to the biopharmaceutical industry — some eagerly anticipated and others coming as welcome surprises. Whereas opportunities for next-generation screening and multiattribute analysis might have been expected, that’s not necessarily true for downstream process modeling and process analytics. But we at BPI have shared some interesting such discussions over the past few years (1–4). Our colleagues in Informa’s Taylor and Francis (T&F) division and the BPI Conference series producers have done the same.

Process Modeling: In a 2018 T&F article, collaborators from Pakistan’s COMSATS University Islamabad and Germany’s Max Planck Institute for Dynamics of Complex Technical Systems describe their work in developing a nonlinear general-rate model to simulate fixed-bed chromatographic columns packed with core beads instead of fully porous particles (5). Because of their thin, porous coatings and solid, impenetrable cores, such “core-shell adsorbents” have short pathways for intraparticle diffusion. Their narrow-peak elution makes them popular with analytical chemists, and Cytiva’s Capto Core media are one example used in bioprocessing. The proposed nonlinear model could be useful in predicting the optimal layer thicknesses for such particles in for a given process.

In a 2020 mAbs article, authors from Bristol Myers Squibb and the The State University of New York College of Environmental Science and Forestry describe another modeling approach for process optimization based on controlled kinetics of disulfide-bond reoxidation during monoclonal antibody (MAb) purification (6). Because disulfide bonds are important to protein folding and structural stability, their disruption can present a problem in downstream processing, leading to product-related impurities such as low–molecular-weight variants and even potential failure to meet product specifications.

“Although many mitigation strategies have been developed to prevent disulfide reduction,” the authors write, “to the best of our knowledge, reforming disulfide bonds from the reduced antibody in manufacturing has not previously been reported.” They explored a “rescue” strategy to repair broken disulfide bonds. Under optimal redox conditions, >90% of reduced MAbs could be reoxidized to form intact antibodies on protein A resin in an hour. The team developed a kinetic model based on elementary oxidative reactions to help optimize those conditions and predict product purity. Combining a deep understanding of interchain disulfide-bond reoxidation with the predictive model yielded a process that generated high-purity antibodies with substantial cost savings.

In a March 2022 talk at BPI West in San Diego — and a BPI article based on it later that year — authors from Sanofi and Cytiva reported on mechanistic modeling of hydrophobic-interaction chromatography (HIC) for protein purification (2). The team used DSPx modeling software to simulate a HIC step in downstream processing of a vaccine antigen candidate, creating a “digital twin” for rapidly performing hundreds to thousands of in silico chromatography experiments, thus saving both materials and time. Proof-of-concept application cases included a massive sampling study to provide design-space understanding, and a pilot scale-up run confirmed the prediction performance of the model.

“Current challenges of chromatography modeling,” the authors cautioned, “include the unavailability or difficulty of using mechanistic models to describe more complex bioprocesses and certain adsorbent modalities. In addition, the time-consuming initial phase of model development limits the usefulness of this approach during early development stages” (2). They recommended “hybrid modeling approaches to allow for flexibility in combining both mechanistic knowledge and process data to improve the speed and accuracy of chromatography modeling without requiring deep knowledge of mechanistic details.”

In a September 2022 talk at Biotech Week Boston, Samantha Wadsworth from Regeneron Pharmaceuticals did focus on the details, specifically regarding transient charge variants (TCVs) in cation-exchange (CEX) chromatography (7). She said that mechanistic modeling provides a means to accelerate process development “while revealing the ‘why’ behind separation behaviors.” Because gradient experiments are important to model calibration, developers need to investigate the root cause of split peaks that show up in CEX gradients. Although such formations often can be attributed to TCVs, their transient nature makes them difficult to study and account for in process models. By fitting a “dummy” monomer to related experimental data, Wadsworth’s team demonstrated a strategy to account for TCVs when generating mechanistic models. Related work continues on structural modeling to understand histidine–solvent exposure as a function of buffer conditions and potentially to control TCV formation by modulating buffer systems.

Process Analytics: Also at BPI West in 2022, Thiago Millen of the Jefferson Institute for Bioprocessing described an approach to applying osmolality measurements for ultrafiltration/diafiltration (UF/DF) monitoring before CEX (8). As a potential alternative to protein A affinity capture in MAb processing, CEX has demonstrated low selectivity for antibodies in clarified cell-culture media. Some process engineers have improved antibody binding to CEX columns through dilution, a concentration/diafiltration step, or pH adjustment before column loading. Thus, careful control of feed/buffer concentration and pH is required for optimizing performance of CEX resins in such applications.

Millen described a partnership with Advanced Instruments whereby his group studied the impact of monitoring osmolality at line during UF/DF of mammalian cell-culture material. Comparing that approach with conductivity monitoring helped the team build a model based on osmolality measurements. “We observed that using a CEX column to capture rituximab was feasible after slight adjustments.” After subsequent anion-exchange (AEX) polishing, final-product concentration and buffer exchange resulted in a rituximab formulation concentration of 20 mg/mL. Osmolality monitoring of UF/DF operations before and after purification showed good correlation with the conductivity data, suggesting that at-line osmolality monitoring could be used to control UF/DF process performance.

“Our findings indicate a high sensitivity of osmolality readings to the TFF and UF/DF operation conditions,” Millen concluded. This might be a more powerful parameter than conductivity to identify slight process deviations. The main challenge is to replace traditionally established conductivity monitoring with osmolality monitoring. “Introduction to new downstream processes (e.g., for viral vectors, which still lack a robust platform purification process) would be interesting. A high-throughput osmolality measurement system would be helpful for process development and process characterization studies.”

Another interesting process analytical technology (PAT) approach is under investigation by a BioPhorum working group, as reported by Daniel Some of Wyatt Technology at BPI West in March 2022 and by Ryan MacDougall of Lonza at BWB later that year (9). The project focuses on real-time multiangle light scattering (RT-MALS) as a tool for monitoring high–molecular-weight product variants or aggregates in downstream processing. Traditional MALS measures indirect process parameters such as temperature and pressure, with UV signals indicating general concentration. RT-MALS can measure direct product attributes such as molar mass (10³–10⁹ g/mol), aggregate size (10–250 nm), and particle concentration. That should make it possible to differentiate product from impurities and to build in process flexibility and control based on data outputs. In-line monitoring, real-time release, and aggregate monitoring are key goals for streamlining production, improving quality, and increasing yield and process consistency/robustness with decreased deviations due to earlier quality attribute understanding. However, accuracy of RT-MALS results depend on stable pH and conductivity values, which will require users to develop an algorithm that can define pooling criteria to enable comparison with UV-based results.

Chromatographic Chemistry
Great strides and important refinements also are emerging in the chemistry and materials of chromatographic purification. Many drug developers partner with technology suppliers and academic researchers to help them solve problems and find new approaches to separation and purification.

For example, Bristol Myers Squibb worked with 3M on developing adsorptive filters for two-step purification of biologics (10). Citing concerns over a substantial increase in both process- and product-related impurities from high-concentration cell cultures, the authors described a need in the biopharmaceutical industry for new process technologies “that offer higher productivity and improved economics without sacrificing process robustness.” To that end, the BMS group evaluated 3M’s Emphaze adsorptive hybrid filters (AHFs) with quaternary amine AEX and salt-tolerant biomimetic ST-AEX ligands for clearance of process-related host-cell proteins (HCPs), DNA, and viruses, as well as product-related soluble aggregates. The team studied interactions governing adsorptive removal of impurities during filtration by evaluating the effects of different filter types, feed streams, and process conditions on impurity removal.

BMS found that the ionic capacity of those filters significantly exceeded that of other adsorptive depth filters, providing for substantially improved reduction of soluble anionic impurities (e.g., DNA, HCPs, and a model virus). Adding a filter aid provided additional hydrophobic functionality that led to removal of aggregates. “Implementing AHF demonstrated improved process-related impurity removal and viral clearance after protein A chromatography and enabled a two-step purification process,” the authors concluded. “The consequences of enhanced process performance are far reaching because it allows the downstream polishing train to be restructured and simplified, and chromatographic purity standards to be met with a reduced number of chromatographic steps.”

Seeking to adapt the protein A affinity step itself, researchers at the KTH Royal Institute of Technology in Sweden worked to develop a calcium-dependent domain (Z_Ca) derived from protein A for a relatively gentle antibody purification method (11). The resulting Z_Ca-multimer purification matrix enables use of less acidic elution conditions than are required for conventional protein A affinity chromatography. That gives it utility for purifying sensitive antibodies and other crystallizable fraction (Fc)–containing proteins. The optimized multimeric variant demonstrated a dynamic binding capacity for IgG of 35 mg/mL resin. Recovery and HCP/DNA clearance proved to be comparable to those of commercial MabSelect resins.

The researchers investigated a number of elution conditions for antibody purification and obtained complete elution of all captured IgG2 and IgG4 antibodies at neutral pH. They concluded, “This optimized protein ligand and the proposed purification method offer a unique strategy for effective and mild purification of antibodies and Fc-fusion proteins that cannot be purified under conventional acidic elution conditions due to aggregation formation or loss of function.”

Suppliers have presented a number of posters at BPI conferences this year touting advancements in chromatographic chemistry.

At BPI West in late February 2023, for example, representatives from Bio-Rad Laboratories highlighted the value of mixed-mode resins for purification of adenoassociated viruses (12), an AEX method downstream processing of single-stranded oligonucleotides (13), and prepacked columns with ceramic hydroxyapatite (CHT) media facilitating process scalabilty (14). Thermo Fisher posters focused on affinity chromatography (15, 16), as did a Tosoh Bioscience entry (17).

In the latter, Jukka Kervinen described an Fc-receptor–based analytical method that can predict antibody-dependent cellular cytotoxicity (ADCC) based on fucose interference (17). Glycan structures attached to conserved Asn-297 influence the efficacy, safety, and stability of antibodies, making characterization of their N-linked glycosylation patterns an important part of MAb quality control. “In particular,” Kervinen writes, “the core fucose is known to decrease ADCC and thus potency.” The analytical chromatography technique presented separates MAbs based on their affinity to the Fc receptor IIIa ligand, and results are consistent with those of a cell-based receptor assay. That provides a faster and simpler alternative for assessing MAb structural consistency in manufacturing.

The Thermo Fisher Scientific posters both featured CaptureSelect media with affinity ligands synthesized from the distinctive heavy-chain antibodies found in Camelidae species (15, 16), which lack the light chain and CH1 domain found in other antibodies. Existing as a single polypeptide chain lends those antibodies stability. They are expressed by yeast cells, so the chromatography resin is free of animal-sourced products, which allows it to be used in both clinical and commercial manufacturing. “Advances in biotherapeutics are generating an increasing range of complex molecules that present unique and often complex purification challenges,” writes Cole Dickson (16). “By taking advantage of antibody-based selectivity, camelid heavy-chain antibody fragments (VHHs) have proven to be a reliable immunoaffinity chromatography solution in the downstream process of biologics.”

Ankita Singh (Bio-Rad Laboratories) highlighted the scalability of prepacked Foresight Pro CHT columns, emphasizing that the packing meets performance criteria before and after shipping. Purification studies demonstrate efficient comparable purification on prepacked columns with inner diameters (IDs) of 8 mm to 33 cm. Erik Verona showed how Bio-Rad addresses the unique challenges of oligonucleotides obtained through solid-phase synthesis (13). They come with incomplete or erroneous sequences that must be removed. A high degree of negative charge makes them separable with AEX chromatography. Experiments with a 5-mL column provided ~90% yields and >97% purity of full-length oligos. In a related poster, Matt Emig showed how Bio-Rad’s AEX and mixed-mode media could be combined to purify adenoassociated viruses, with the former used for capture and the latter for polishing (12).

AEX chromatography is a key unit operation for reduction of process-related contaminants following protein A affinity MAb capture. As bioreactor processes have become more productive through increasing cell densities, HCP concentrations have increased. Particular attention must be placed on removing HCPs that catalyze product or excipient degradation and destabilize bulk drug substance. As purification processes are simplified, clearance of such proteins becomes more challenging, forcing process engineers to develop specific strategies to address them.

At BPI Europe, Ankur Solanki of 3M Ireland also featured AEX, in this case focusing on HCP removal (18). He compared single-use Polisher ST membrane adsorbers with different ligand types across a number of MAb feeds, including proteins that are known to facilitate product degradation and destabilization.

Thomas Valorose (Astrea Bioseparations) touted HCPure mixed-mode chromatography for streamlining the polishing process by combining AEX and HIC binding methods, thus reducing two traditional steps to one (19). He demonstrated the tunability of the resulting purification platform for a variety of conditions and its ability to purify problematic feed streams. Also from Astrea, Gráinne Dunlevy highlighted HCP removal from lentivirus production streams (20).

Mark Hicks of Purolite introduced his company’s new protein A resin designed for high-pH elution of “atypical IgG constructs” (21). Many such products are prone to aggregation, are expressed at high titers, and come along with numerous variants. Hicks showed the value of Praesto Jetted A50 HipH media’s ability to elute at a broad pH range without compromising yield, purity, or elution volume.

Greta Hulting of Cytiva focused on improving lentivirus polishing with Capto Core 700 resin (22). Purification of such enveloped RNA viruses is difficult because of their sensitivity to pH, salt, temperature, shear forces, and other factors. “Efforts must be made to minimize conditions that negatively affect a good physical and infectious titer recovery,” Hulting writes. AEX is often the choice but typically comes with recovery losses. Following optimized capture from clarified feeds with weak-AEX Capto DEAE resin, using pH 7.0 improved infectious recovery of a Capto Core 700 polishing step without changing the physical recovery. Taking residence time down to 0.7 minutes was possible without negative effects. Total DNA and proteins in the resulting material were below the limit of quantitation (LoQ) for all conditions.

Julia Bartmann of YMC described another refinement of IEX chemistry that applied to oligonucleotides, proteins, and MAbs alike (23). Hydrophilic polymethacrylate-based IEX resins work at high flow rates and generate low back pressures. Bartmann highlighted the potential for improving resolution with relatively high eluent temperatures, which could increase overall process productivity. That combined with the potential for large-volume sample loading (thanks to relatively high dynamic binding capacities) should make for economical and efficient purification processes.

Process Engineering
New therapeutic modalities are challenging old ways of thinking in purification, and thanks to technological advances such as those above, some of the most creative problem-solving is happening in the realm of process engineering. The SUNY–BMS collaboration mentioned above is a good example (6). Others that have appeared in T&F journals over the past few years include

• an investigation into the effects of salt additives in protein L affinity chromatography for purification of tandem single-chain variable fragment (scFv) bispecific antibodies (24) from the Bioprocessing Technology Institute at Singapore’s Agency for Science, Technology and Research (A*STAR)
• a report from Germany’s University of Applied Sciences and University of Giessen on CEX purification of oncolytic measles virus for a cancer vaccine (25)
• a BMS case study in MAb process intensification (26).

With its potential in biomanufacturing for increasing productivity, improving sustainability, and reducing costs, it’s no wonder that process intensification has gained widespread interest. BMS enhanced seed-culture cell densities and production-bioreactor inoculation density, which increased expression titers from fed-batch culture by eightfold. The downstream process needed multiple changes to accommodate those improvements: new high-capacity resins for both the protein A and AEX steps and a change from bind–elute to flow-through mode for the CEX step. Multicolumn chromatography was implemented for protein A capture, and integrated AEX–CEX polishing steps enabled the process to run semicontinuously with reduced resin requirements, buffer consumption, and process time. “The hybrid-intensified process described here is easy to implement in manufacturing and lays a good foundation to develop a fully continuous manufacturing with even higher productivity in the future,” the authors concluded (23).

The remainder of this insert adds to the growing volume of literature demonstrating that downstream process innovation is not only alive and well, but kicking into high gear. BPI’s managing editor discusses the ins and outs of mixed-mode chromatography with the University of Virginia’s Nick Vecchiarello. A contract manufacturer and technology supplier report on their collaborative work toward a purification platform for adenoassociated viruses. And BPI advertiser Tosoh Bioscience details scFv technology that may be compatible with the A*STAR work mentioned above. We look forward to your continued contributions.

References
1 Maiti S, Spetsieris K. Advanced Data-Driven Modeling for Biopharmaceutical Purification Processes. BioProcess Int. 19(9) 2021: 44–51; https://bioprocessintl.com/manufacturing/process-monitoring-and-controls/advanced-data-driven-mvda-modeling-for-biopharmaceutical-purification-processes.

2 Li AS, et al. Mechanistic Modeling for a Hydrophobic-Interaction Chromatography Process: Use in Vaccine Antigen Purification. BioProcess Int. 20(9) 2022: S10–S17; https://bioprocessintl.com/manufacturing/vaccines/advancing-in-silico-tools-for-vaccine-development-and-process-modeling.

3 Rosado PJ, Merheb B, Toro A. Real-Time, Data-Driven, and Predictive Modeling: Accelerating Digital Transformation in Drug Substance Commercial Manufacturing. BioProcess Int. 21(1–2) 2023: 38–45; https://bioprocessintl.com/analytical/pat/real-time-data-driven-and-predictive-modeling-accelerating-digital-transformation-in-drug-substance-commercial-manufacturing.

4 Kim N, et al. Predictive Algorithm Modeling for Early Assessments in Downstream Processing: Using Direct Transition and Moment Analysis To Assess Chromatography Column Integrity at Production Scale. BioProcess Int. 21(3) 2023: 24–31; https://bioprocessintl.com/downstream-processing/chromatography/predictive-algorithm-modeling-for-early-assessments-in-downstream-processing-using-direct-transition-and-moment-analysis-to-assess-chromatography-column-integrity-at-production-scale.

5 Akram N, Qamar S, Seidel-Morgenstern A. Nonlinear Model of Liquid Chromatography Considering Finite Rates of Adsorption-Desorption Kinetics and Core-Shell Adsorbents. J. Liquid Chromatog. Rel. Technol. 41(17–18) 2018: 964–972; https://doi.org/10.1080/10826076.2018.1519832.

6 Tan Z, et al. On-Column Disulfide Bond Formation of Monoclonal Antibodies During Protein A Chromatography Eliminates Low Molecular Weight Species and Rescues Reduced Antibodies. mAbs 12(1) 2020: 1829333; https://doi.org/10.1080/19420862.2020.1829333.

7 Wadsworth S, Vartak A, Reilly J. Transient Charge Variants During Cation-Exchange Chromatography Gradients and Their Implications for In Silico Modeling. Biotech Week Boston: Boston, MA, 28 September 2022: https://streamly.video/video/transient-charge-variants-during-cation-exchange-chromatography-gradients-and-their-implications-for-in-silico-modeling.

8 Millen T. Osmolality as a Tool for UF/DF Monitoring Before Cation Exchange Chromatography Capture of Monoclonal Antibodies. BPI West: San Diego, CA, 16 March 2022.

9 Is It Possible To Do Real-Time Analysis of Aggregates in Downstream Protein Solutions? BioPhorum News 4 May 2022; https://www.biophorum.com/is-it-possible-to-do-real-time-analysis-of-aggregates-in-downstream-protein-solutions.

10 Singh N, et al. Development of Adsorptive Hybrid Filters to Enable Two-Step Purification of Biologics. mAbs 9(2) 2017: 350–364; https://doi.org/10.1080/19420862.2016.1267091.

11 Scheffel J, et al. Optimization of a Calcium-Dependent Protein A-Derived Domain for Mild Antibody Purification. mAbs 11(8) 2019: 1492–1501; https://doi.org/10.1080/19420862.2019.1662690.

12 Verona E. Purification of Single-Stranded Oligonucleotides Using Anion Exchange Chromatography. Poster at BPI West: San Diego, CA, February 2023.

13 Singh A. Scalable Purification Using GMP Ready CHT Prepacked Process-Scale Columns. Poster at BPI West: San Diego, CA, February 2023.

14 Emig M. Mixed-Mode Chromatography for AAV8 Scalable Purification. Poster at BPI West: San Diego, CA, February 2023.

15 Eke J. Addressing Purification Challenges in Bioprocessing: CaptureSelect FcXP Affinity Resin and POROS CaptureSelect FcXP Affinity Resin. Poster at BPI West: San Diego, CA, February 2023.

16 Dickson C. Advantage of Antibody-Based Selectivity in the Purification of Biologics. Poster at BPI West: San Diego, CA, February 2023.

17 Kervinen J. A Simple Fc Receptor-Based Affinity Chromatography Method that Predicts ADCC Activity Based on Fucose Interference. Poster at BPI West: San Diego, CA, February 2023.

18 Solanki A. Comparative Investigations of Single-Use AEX Ligand Types for Removal of Common and Problematic Host Cell Proteins in Biopharmaceutical Purification. Poster at BPI Europe: Amsterdam, The Netherlands, May 2023.

19 Valorose T. Advances in Mixed-Mode Chromatography: HCPure Host Cell Protein Clearance Resin Case Studies. Poster at BPI Europe: Amsterdam, The Netherlands, May 2023.

20 Dunlevy G. A Novel Purification Technology to Increase Processing Efficiency, Purity, and Recovery of Lentiviral Particles for Viral Vector Development. Poster at BPI Europe: Amsterdam, The Netherlands, May 2023.

21 Hicks M. New Protein A Resin for Elution at High pH. Poster at BPI Europe: Amsterdam, The Netherlands, May 2023.

22 Hulting G. How to Improve Lentivirus Polishing Recovery Using Capto Core 700 Resin. Poster at BPI Europe: Amsterdam, The Netherlands, May 2023.

23 Bartmann J. High-Performance Resins for Downstream Processing of Oligonucleotides, Proteins, and Monoclonal Antibodies. Poster at BPI Europe: Amsterdam, The Netherlands, May 2023.

24 Chen SW, et al. Investigation of the Effect of Salt Additives in Protein L Affinity Chromatography for the Purification of Tandem Single-Chain Variable Fragment Bispecific Antibodies. mAbs 12(1) 2020: 1718440; https://doi.org/10.1080/19420862.2020.1718440.

25 Eckhardt D, et al. Purification of Oncolytic Measles Virus By Cation-Exchange Chromatography Using Resin-Based Stationary Phases. Sep. Sci. Technol. 57(6) 2022: 886–896; https://doi.org/10.1080/01496395.2021.1955267.

26 Xu J, et al. Biomanufacturing Evolution from Conventional to Intensified Processes for Productivity Improvement: A Case Study. mAbs 12(1) 2020: 1770669; https://doi.org/10.1080/19420862.2020.1770669.

Cheryl Scott is cofounder and senior technical editor of BioProcess International (part of Informa Connect Life Sciences); 1-212-600-3429; [email protected].