Enabling the full potential of a new therapeutic modality requires the development of a flexible and cost-effective manufacturing platform. A critical bottleneck in this process is the development of robust purification platform, usually reliant on sequential chromatography steps. Development of chromatographic steps is a laborious and costly task, as it is dependent of multiple iterations. Evaluating multiple parameters in a traditional methodology can lead to an exponential number of experiments. To overcome this, we propose the use of Bayesian optimisation to efficiently navigate the solution space. Using the development of mRNA affinity chromatography as a model, Bayesian optimization was used to enhance the dynamic binding capacity. This approach led to a 7.5-fold increase in capacity (1.8 mgRNA mL-1) relatively to the benchmark run in only 13 iterations. Additionally, model interpretability techniques were used to correlate predictions with the experimental results, while gaining process knowledge. Bayesian optimisation is a powerful and efficient tool for chromatography development, and in combination with model interpretability techniques, can have a real impact on process development using a QbD framework, and with potentially be used for automation and broader application in bioprocessing.

Traditional polyethersulfone (PES) filters, widely used for sterile, viral and ultrafil- tration often exhibit limited selectivity-permeability due to the heterogenous pore size distribution. Such limitations have sparked interest in developing novel isoporous mem- brane materials and fabrication techniques to overcome the selectivity-permeability upper bound. Among several promising candidates, block copolymer membranes pro- duced via the self-assembly and non-solvent induced phase separation (SNIPS) method offer distinct advantages, such as customisable pore size, narrow dispersity, high poros- ity and mechanical flexibility. However, achieving the desired structure formation in SNIPS remains a complex optimisation procedure, rendering this approach unsuitable for the rapid screening of new block copolymer candidates. This study explores a direct spin coating method to fabricate a poly(styrene)-block- poly(methyl methacrylate) (PS-b-PMMA) thin film composite membrane, integrating a block copolymer layer with rigid anodic aluminium oxide (AAO) support. The pro- cess involves depositing the polymer solution onto a water-filled AAO substrate via spin coating. Compared with the SNIPS method, this fabrication process greatly re- duces the complexity of optimisation to yield an isoporous membrane structure for filtration purposes. We present this approach as a straightforward and reliable plat- form method for the rapid screening and evaluation of block copolymer membranes in their initial development stages. When compared to commercial PES membranes with similar molecular weight cut-offs, these novel PS-b-PMMA thin film composite membranes showed similar transmission rates for Bovine Serum Albumin and a Mon- oclonal Antibody while providing a 9 fold enhancement in Thyroglobulin rejection. This indicates a superior performance in terms of cut-off precision. The membranes demonstrate potential for the removal of viruses and antibody aggregates in the down- stream processing of monoclonal antibody production, which could reduce the burden of chromatographic polishing steps. An advance which offers promise for improving efficiency and reducing costs in biopharmaceutical manufacturing.

mRNA is an emerging drug modality with high potential as both therapeutic and prophylactic products. The rising investment in this technology can be attributed to its versatility and rapid development of new drug candidates. Oligo deoxythymidine (oligo(dT)) ligand chemistries are of industrial relevance in chromatographic purification of mRNA, however, beaded oligo(dT) resins struggle with low dynamic binding capacity (DBC). To aid understanding of these beaded oligo(dT) resins, a structural analysis was conducted, followed by a characterisation of poly-adenine RNA binding behaviours. The aim establishing a structure-performance relationship to inform future resin design. To characterise structural properties of resins, Brunauer-Emmett-Teller analysis first determined the surface area of three resins, ranging from 13-18 m2.g-1. Mercury intrusion porosimetry characterised pore size distribution and scanning electron micrographs were produced of resin surfaces and interiors from bead cross-sections, observing modal pore diameters of 298-526 nm. To measure RNA binding performance, we tested binding of a 2100-10,000 nucleotide poly-adenine over 20 min and characterised material binding. Material predominantly collected on the bead exterior, but gradual penetration occurred into the internal pore network. Breakthrough experiments connected these results to trends in DBC. A 257 % DBC10 increase was observed from 1-10 min residence time, highlighting the importance of diffusive transport in these beaded resins. We also compared mass transfer behaviours between two RNAs with median lengths of ∼500 and ∼5000 nucleotides, observing significant differences in diffusive transport speed, with larger RNA entering the bead interior more slowly. With this work we aim to inform future resin design for mRNA purification.

Traditional poly(ether sulfone) (PES) filters, widely used for sterile, viral, and ultrafiltration, often exhibit restrictions in their selectivity-permeability profile due to their heterogeneous pore size distribution. This limitation has sparked interest in developing novel isoporous membrane materials and fabrication techniques. Among promising candidates, block copolymer (BCP) membranes produced via self-assembly and nonsolvent-induced phase separation (SNIPS) offer significant advantages, including tunable pore size, narrow pore size distribution, high porosity, and enhanced mechanical flexibility. However, optimizing the structure formation in SNIPS remains a complex and time-consuming process, making it unsuitable for rapidly screening new BCP candidates. In response, this study introduces an alternative fabrication approach based on the direct spin-coating of BCPs onto anodic aluminum oxide (AAO) supports. Using this method, a poly(styrene)-block-poly(methyl methacrylate) (PS-b-PMMA) thin film was directly cast onto a water-filled AAO support, enabling the formation of an isoporous membrane structure for filtration applications, significantly reducing the complexity of structure-application optimization. When compared to commercial PES membranes with similar molecular weight cut-offs, these novel PS-b-PMMA thin-film composite membranes exhibited comparable transmission rates for bovine serum albumin and a monoclonal antibody, while delivering a ninefold improvement for thyroglobulin rejection. This superior cutoff precession highlights their potential to remove viruses and antibody aggregates during the downstream processing of monoclonal antibody production. By reducing the burden of chromatographic polishing steps, this advance offers promise for enhancing efficiency and lowering costs in biopharmaceutical manufacturing.

High throughput process development (HTPD) has been widely adopted for efficient development and optimization of chromatographic operations in monoclonal antibody (mAb) purification. However, the integration of non-chromatographic unit operations, particularly depth filtration following protein A chromatography, which is essential for the removal of process- and product-related impurities prior to the ion exchange chromatography (IEX) operations, remains a challenge due to the absence of commercially available micro-scale depth filtration tools. This limits the integration of this unit operation within the purification sequence, restricting the analysis of process interactions and overall process understanding. In this study, a micro-scale HTPD platform was designed and evaluated to enable integration of a depth filtration mimic, Sartobind® Q anion exchange adsorber, within a mAb purification sequence. This was achieved by translating laboratory-scale protocols to the micro-scale using workflow design tools and executed on an automated liquid handling system. Step yields and impurity clearance were assessed to confirm the equivalence of scale-down. The Sartobind® Q membrane achieved effective removal of host cell DNA (hcDNA), while subsequent IEX operations removed host cell proteins (HCPs) and high molecular weight components (HMWC), meeting target product quality specifications. The platform demonstrated robustness across varying impurity profiles, supporting its applicability for diverse process intermediates. Comparative analysis with laboratory-scale operations confirmed the performance and scalability of the micro-scale system, reducing the total run time by greater than 50%. The integrated HTPD platform offers a resource-efficient, scalable approach for comprehensive mAb purification process development and is suitable for developability assessments during early-stage development.

Background aims: Extracellular vesicles (EVs) have gained traction as potential cell-free therapeutic candidates. Development of purification methods that are scalable and robust is a major focus of EV research. Yet there is still little in the literature that evaluates purification methods against potency of the EV product. In the present study, we examined two monolith chromatography methods with a focus on assessing the ability of purified EVs to retain stimulatory effects on fibroblasts to connect scalable purification methods with product outputs. Methods: We characterized EVs recovered from CTX0E03 (CTX) neural stem cell-conditioned medium in terms of biomarker distribution, functional capacity and purity. We evaluated the ability of EVs to promote wound closure in an in vitro scratch assay prior to and following two monolith chromatography steps (anion exchange and hydrophobic interaction) to determine whether these options may better serve EV bioprocessing. Results: EVs from CTX cells were successful in initiating wound repair in a fibroblast scratch assay over 72 h with a single 20-mg dose. EV preparations presented the markers CD9, CD81 and CD63 but also contained culture albumin and DNA as process impurities. EVs recovered by tangential flow filtration could be successfully purified further by both monolith chromatography steps. Post-monolith EV stimulation was conserved. Conclusions: The results indicate that monolith chromatography is a viable purification method for EVs derived from cell culture that does not detract from the product's ability to stimulate fibroblasts, suggesting that product functionality is conserved. Further work is needed in developing suitable downstream processes and analytics to achieve clinically relevant purities for injectable biologics. (c) 2025 Published by Elsevier Inc. on behalf of International Society for Cell & Gene Therapy.

The growing demand for lentiviral vectors (LVs) in cell and gene therapies has highlighted significant challenges in large-scale LV production, particularly low product recovery. Anion-exchange chromatography (AIEX) is widely used for LV capture. However, AIEX accounts for most of the product loss in downstream processing, due to excessive interaction strengths in current commercially available membranes causing irreversible LV binding. To address this, a new AIEX membrane structure that reduces LV interaction strength by lowering ligand density, using direct ligand grafting, and incorporating tighter pore size distributions is tested. Four prototypes were created with Q and D anion-exchange chemistries at 1.05 (R1) and 1.32 (R2) μm pore sizes. Prototypes minimized product loss from irreversible binding, with high total particle recoveries irrespective of time spent in the adsorbed state ( ∼ 90% at t = 3 min, ∼ 80% at t = 100 min). Narrower elution ranges were shown, with LV eluted <450 mM NaCl. The best-performing prototype, DR2, exhibited a ∼3-fold higher functional product recovery than standard Q-membranes for LV encoding GFP (50%) and CAR (73%) transgenes. At large-scale, downstream processes using DR2 membranes showed a 3.5-fold improvement in functional product recovery at drug substance (43%) compared to a standard Q-membrane process (12%). These results demonstrate that adsorbents designed for lentiviral vectors significantly enhance downstream recoveries.

On March 30, 2022, Inno4Vac, a public-private partnership funded by the IMI2/EU/EFPIA Joint Undertaking (IMI2 JU), organised a hybrid workshop, titled "Regulatory Dialogue for Road Maps of Implementation of New Tools in Chemistry, Manufacturing, and Controls Dossiers." This event brought together modellers, regulatory experts, and academic and industry professionals specialising in vaccine process and product development. The sessions discussed key parameters and requirements for model development and verification relevant to vaccine biomanufacturing and shelf life. The stability model was highlighted as having the most significant impact on the common technical document (CTD) due to its potential to streamline data requirements. Regulators are open to considering reliable reduced stability data packages (3-12 months) instead of the standard 36 months, potentially expediting product availability. Appropriate study design reduces uncertainty and therefore the risk of making poor decisions. Upstream models are further from the final product, and their role in the control strategy of the product will define their level of risk and, therefore, requirements for validation and inclusion of information in the file. Regulators may consider downstream models high risk as these can be associated with the monitoring and/or control of critical quality attributes and/or be involved in the release of a product. However, requirements for validation and/or dossier content should always be linked to the intended use of the model and its overall role in the control strategy as per the new EMA Quality Innovation Group Considerations regarding Pharmaceutical Process Models. The success of these models hinges on manufacturers providing enough quality data to prove their accuracy in representing real-world processes. Proactive engagement with regulators, supported by detailed evidence, can foster regulator understanding of new models and potentially lead to new guidelines and pathways for model acceptance.

Recombinant adeno-associated viruses (rAAVs) have emerged as important tools for gene therapy and, more recently, vaccine development. Nonetheless, manufacturing can be costly and time-consuming, emphasizing the importance of alternative production platforms. We investigate the potential of E. coli-based cell-free protein synthesis (CFPS) to produce recombinant AAV5 virus-like particles (VLPs). AAV5 virus protein 3 (VP3) constructs, both with and without Strep-tag II, were expressed with CFPS. Lower reaction temperatures resulted in increased solubility, with the untagged variant containing nearly 90% more soluble VLP VP3 protein at 18 °C than at 37 °C. Affinity chromatography of N-terminally Strep(II)-tagged VP3 enabled successful isolation with minimal processing. DLS and TEM confirmed the presence of ∼20 nm particles. Furthermore, the N-terminally tagged AAV5 VP3 VLPs were biologically active, successfully internalizing into HeLa cells. This study describes an innovative approach to AAV VLP production using E. coli-based CFPS, demonstrating its potential for rapid and biologically active AAV VLP synthesis.

The demand for Lentiviral Vector (LV) drug substance is increasing. However, primary capture using convective anion-exchange chromatography remains a significant manufacturing challenge. This stems from a poor understanding of the complex adsorption behaviors linked to LVs intricate and variable structure, such as high binding heterogeneity which is typically characterized by a gradient elution profile consisting of two peaks. Understanding which LV structural components drive these phenomena is therefore crucial for rational process design. This work identifies the key LV envelope components responsible for binding to quaternary-amine membrane adsorbents. Eliminating the pseudotype protein (Vesicular Stomatitis Virus G glycoprotein [VSV-G]) did not impact the heterogenous two-peak elution profile, suggesting it is not a major binding species. Digestion of envelope glycosaminoglycans (GAGs), present on proteoglycans, leads to a dramatic reduction in the proportion of vector eluted in peak 2, decreasing from 50% to 3.1%, and a threefold increase in peak 1 maximum. Data from reinjection experiments point towards interparticle envelope heterogeneity from discrete LV populations, where the two-peak profile emerges from a subpopulation of LVs interacting via highly charged GAGs (peak 2) along with a weaker binding population likely interacting through the phospholipid membrane and envelope protein (peak 1).

Continuously secreted by all cell types, extracellular vesicles (EVs) are small membrane-bound structures which shuttle bioactive cargo between cells across their external environment. Their central role as natural molecular messengers and ability to cross biological barriers has garnered significant attention in the use of EVs as therapeutic delivery vehicles. Still, harnessing the potential of EVs is faced with many obstacles. A cell line engineering approach can be used to exploit EVs to encapsulate a bespoke cargo of interest. However, full details regarding native EV-loading mechanisms remain under debate, making this a challenge. While Chinese hamster ovary (CHO) cells are well known to be the preferred host for recombinant therapeutic protein production, their application as an EV producer cell host has been largely overlooked. In this study, we engineered CHO DG44 cells to produce custom EVs with bespoke cargo. To this end, genetic constructs employing split green fluorescent protein technology were designed for tagging both CD81 and protein cargoes to enable EV loading via self-assembling activity. To demonstrate this, NanoLuc and mCherry were used as model reporter cargoes to validate engineered loading into EVs. Experimental findings indicated that our custom EV approach produced vesicles with up to 15-fold greater cargo compared with commonly used passive loading strategies. When applied to recipient cells, we observed a dose-dependent increase in cargo activity, suggesting successful delivery of engineered cargo via our custom CHO EVs.

Maximizing product quality attributes by optimizing process parameters and performance attributes is a crucial aspect of bioprocess chromatography process design. Process parameters include but are not limited to bed height, eluate cut points, and elution pH. An under-characterized chromatography process parameter for protein A chromatography is process temperature. Here, we present a mechanistic understanding of the effects of temperature on the protein A purification of a monoclonal antibody (mAb) using a commercial chromatography resin for batch and continuous counter-current systems. A self-designed 3D-printed heating jacket controlled the 1 mL chromatography process temperature during the loading, wash, elution, and cleaning-in-place (CIP) steps. Batch loading experiments at 10, 20, and 30 °C demonstrated increased dynamic binding capacity (DBC) with temperature. The experimental data were fit to mechanistic and correlation-based models that predicted the optimal operating conditions over a range of temperatures. These model-based predictions optimized the development of a 3-column temperature-controlled periodic counter-current chromatography (TCPCC) and were validated experimentally. Operating a 3-column TCPCC at 30 °C led to a 47% increase in DBC relative to 20 °C batch chromatography. The DBC increase resulted in a two-fold increase in productivity relative to 20 °C batch. Increasing the number of columns to the TCPCC to optimize for increasing feed concentration resulted in further improvements to productivity. The feed-optimized TCPCC showed a respective two, three, and four-fold increase in productivity at feed concentrations of 1, 5, and 15 mg/mL mAb, respectively. The derived and experimentally validated temperature-dependent models offer a valuable tool for optimizing both batch and continuous chromatography systems under various operating conditions.

Electrospun cellulose adsorbents are an emergent class of materials applied to a variety of bioprocess separations as an analogue to conventional packed bed chromatography. Electrospun adsorbents have proven to be effective as rapid cycling media, enabling high throughput separation of proteins and viral vectors without compromising selectivity and recovery. However, there is a current lack of knowledge in relation to the manipulation and control of electrospun adsorbent structure with function and performance to cater to the separation needs of emerging, diverse biological products. In this study, a series of electrospun cellulose adsorbents were fabricated by adjusting their manufacturing conditions. A range of fiber diameters (400 to 600 nm) was created by changing the electrospinning polymer solution. Additionally, a range of porosities (0.4 to 0.7 v/v) was achieved by varying the laminating pressures on the electrospun sheets. The adsorbents were functionalized with different degrees of quaternary amine ligand density to create 18 prototype anion exchangers. Their morphology was characterized by BET nitrogen adsorption surface area, X-ray computed tomography, capillary flow porometry and scanning electron microscopy measurements. The physical characteristics of the adsorbents were used in an adapted semi-empirical model and compared to measured permeability data. Permeabilities of prototypes ranged from 10-2 to 10-4 mDarcy. The measured data showed good adherence to modelled data with possible improvements in acquiring wet adsorbent characteristics instead of dried material. Finally, the electrospun adsorbents were characterized for their binding capacity of model proteins of different sizes (diameters of 3.5 nm and 8.9 nm) and plasmid DNA. Static binding capacities ranged from 5 mg/ml to 25 mg/ml for the proteins and plasmid DNA and showed <20 % deviation from monolayer coverage based on BET surface area. Therefore, it was concluded that the electrospun adsorbents most likely adsorb monolayers of proteins and plasmid DNA on the surface with minimal steric hindrance.

This review highlights the importance of model-based approaches in accelerating vaccine manufacturing process development. The challenges of scaling up from laboratory to commercial processes are addressed through the adoption of Process Analytical Technology frameworks and Quality by Design principles. The application of various modeling approaches beyond downstream and upstream processes in vaccine production is discussed in detail. These in silico process simulation approaches enable deeper understanding of manufacturing dynamics, identification of critical process parameters, and the development of well-defined design spaces, ultimately leading to accelerated vaccine development and improved product quality. The authors stress the significance of an integrated modeling platform for vaccine manufacturing, exemplified by the Inno4Vac project. This initiative seeks to develop a comprehensive computational platform for vaccine manufacturing and stability testing, with a particular focus on stakeholder engagement and collaboration with regulatory bodies to ensure the acceptance and implementation of the platform.

Systematic development of a temperature-controlled isocratic process for one-column low-salt hydrophobic interaction chromatography (HIC) of proteins employing a travelling cooling zone reactor (TCZR) system, is described. Batch binding and confocal scanning microscopy were employed to define process conditions for temperature-reversible binding of bovine serum albumin (BSA) which were validated in pulse-response temperature switching HIC experiments, before transferring to TCZR-HIC. A thin-walled stainless-steel column mounted with a movable assembly of copper blocks and Peltier elements (travelling cooling zone, TCZ) was used for TCZR-HIC. In pulse-response TCZR-HIC, 12 TCZ movements along the column desorbed 86.3% of the applied BSA monomers in 95.3% purity depleted >6-fold in 2-4 mers and nearly 260-fold in higher molecular weight (HMW) species. For continuous TCZR-HIC, the TCZ was moved 49-58 times during uninterrupted loading of BSA feeds at 0.25, 0.5 or 1 mgmL(-1). Each TCZ movement generated a sharp symmetrical elution peak. In the best case, (condition 1: 0.25 mgmL(-1) BSA; >17 mg BSA applied per mL of bed) the height of TCZ elution peaks approached pseudo-steady midway through the loading phase with no rise in baseline UV280 signal between peaks. Peak composition remained constant averaging 94.4% monomer, 5.6% 2-4 mers and <0.05% HMW. Monomers were recovered in quantitative yield depleted >3.1 fold in 2-4 mers and 92-fold in HMW species cf. the feed (63.6% monomers, 21.8% 2-4 mers, 14.6% HMW). However, increasing the BSA concentration to 1 mgmL(-1) (condition 2) or employing a fouled HIC column with 0.5 mgmL(-1) BSA (condition 3) compromised monomer purification performance.

High throughput process development (HTPD) is established for time- and resource- efficient chromatographic process development. However, integration with non-chromatographic operations within a monoclonal antibody (mAb) purification train is less developed. An area of importance is the development of low pH viral inactivation (VI) that follows protein A chromatography. However, the lack of pH measurement devices at the micro-scale represents a barrier to implementation, which prevents integration with the surrounding unit operations, limiting overall process knowledge. This study is based upon the design and testing of a HTPD platform for integration of the protein A and low pH VI operations. This was achieved by using a design and simulation software before execution on an automated liquid handler. The operations were successfully translated to the micro-scale, as assessed by analysis of recoveries and molecular weight content. The integrated platform was then used as a tool to assess the effect of pH on HMWC during low pH hold. The laboratory-scale and micro-scale elution pools showed comparable HMWC across the pH range 3.2-3.7. The investigative power of the platform is highlighted by evaluating the resources required to conduct a hypothetical experiment. This results in lower resource demands and increased labor efficiency relative to the laboratory-scale. For example, the experiment can be conducted in 7 h, compared to 105 h, translating to labor hours, 3 h and 28 h for the micro-scale and laboratory-scale, respectively. This presents the opportunity for further integration beyond chromatographic operations within the purification sequence, to establish a fit-to-platform assessment tool for mAb process development.