Abstract Small virus‐retentive filtration (VRF) is a critical unit operation utilized in the viral clearance package to ensure the viral safety of biotherapeutics. This step is relied upon to provide effective parvovirus retention of ≥4 log reduction value (LRV) (e.g., minute virus of mice; MVM), and robust retention of larger viruses. Biomolecules, including multispecific antibodies and recombinant proteins, present purification challenges that require multiple chromatography columns to achieve final purity specifications. These additional polishing steps may utilize high ionic strength solution conditions to meet purity targets, and it is critical to assess if those process conditions impact the viral retention performance of the subsequent VRF operations. In one instance, retention performance was lower than expected (≤4 LRV) and was attributed to the use of a high ionic strength buffer. Two possible mechanisms were investigated, that is, reduction in the effective size of the MVM virus (~25 nm) and/or an increase in the membrane filter pore size, as observed with increase in the virus filter permeability. Based on modeling and experimental data presented here, at high ionic strength solution conditions the hydrodynamic size of MVM is reduced while virus‐retentive membrane effective pore size is larger, as compared to low ionic strength conditions. Additionally, MVM LRV and membrane permeability of multiple commercial‐grade virus filters were impacted by high ionic strength and process interruptions. These results demonstrate that high ionic strength solution conditions could impact parvovirus retention performance, and these findings may guide process development for biotherapeutics operating under similar atypical solution conditions.

In vitro transcription (IVT) is a powerful method to generate RNA which not only facilitates RNA research but also plays a key role in the development and manufacture of RNA-based vaccines. mRNA is produced via the IVT process with a DNA template that contains information for the target antigen. However, as many disease-causing viruses mutate quickly and the cost of raw materials is high for the IVT reaction, there is a need for a system to develop a cost-effective and efficient IVT process platform. In this paper, we showed how total nucleoside-5'-triphosphates (NTPs) input, Mg2+ concentration and NTP preparation methods can influence IVT reaction yield and purity level of the final RNA constructs of different lengths and sequences. We propose an IVT design that will result in high RNA yield, high RNA integrity and low double-stranded RNA (dsRNA) concentrations for multiple RNA sequences. The approach presented here could significantly contribute to the development of a cost-effective, easy-to-adopt IVT process platform for RNA manufacturing.

Recent advances in gene editing technologies have transformed the genetic engineering of Chinese hamster ovary (CHO) hosts, enabling the development of cell lines with improved stability and productivity. In this study, we employed the programmable nuclease (PN) Cas‐CLOVER to precisely target the Glutamine synthetase (GS) locus in CHO cells. A total of 30 unique serum‐free, suspension‐adapted CHO‐K1 candidate host cell lines were subjected to Cas‐CLOVER‐mediated gene editing, generating over one hundred potential GS knockout (GSKO) clones. A subset of the GSKO clones was subsequently validated using three orthogonal methods to confirm complete knockout of the GS gene in 98 clones. Randomly selected GSKO clones were utilized to produce standard monoclonal antibodies. The resulting pools demonstrated enhanced productivity, with up to a 14.5‐fold increase in titer compared to their wild‐type parental hosts. These findings highlight the potential of gene editing approaches to significantly improve recombinant protein production in CHO expression systems, offering valuable insights for biopharmaceutical manufacturing applications.

Red blood cells (RBCs) play a critical role in oxygen and carbon dioxide transport, which is facilitated by RBC-encapsulated hemoglobin (Hb) and carbonic anhydrase (CA). In addition, RBCs are constantly exposed to oxidative stress due to the intracellular reactive oxygen species (ROS) generated during Hb auto-oxidation. Antioxidant enzymes within RBCs, such as superoxide dismutase (SOD), catalase (CAT), and peroxiredoxin (Prx), counteract ROS generation to protect the RBC from oxidative stress. Therefore, this study presents a scaled-up method to extract an enzyme cocktail from lysed human RBCs, enriched with the major RBC enzymes with minimal Hb contamination. Using ethanol-chloroform precipitation and multiple biophysical analyses (SDS-PAGE, SEC-HPLC, MALDI-TOF, and LC-MS/MS), the RBC enzymes were successfully separated from Hb in the hemolysate. The purified enzyme cocktail exhibited minimal Hb contamination and retained a significant amount of CA, and antioxidative enzymes like SOD and CAT. Therefore, this scalable RBC enzyme purification method provides an efficient approach for isolating RBC enzymes with broad biomedical relevance.

N-linked glycosylation stands as a pivotal quality attribute for monoclonal antibodies (mAbs), particularly the high mannose (Man5) variant, which significantly influences the pharmacokinetics of mAbs. Traditional approaches to modulate Man5 have frequently resulted in suboptimal outcomes. In this investigation, we introduced a novel additive, N-acetyl-d-mannosamine (ManNAc), which selectively targeted and reduced Man5 without compromising other product quality attributes (PQAs). The study further examined optimal concentrations and timing for the incorporation of ManNAc in the mAbs expression process utilizing CHO-K1 cells within a fed-batch shaker flask culture mode. In the ManNAc titration experiments, we established groups at concentrations of 5, 10, 15, 20, 40, 60, 80, and 100 mM. The findings revealed a concentration-dependent decrease in Man5, with reductions reaching as low as 2.9% from an initial 8.9%. Importantly, cellular growth, metabolism, and other PQAs remained unaffected. Regarding the timing of ManNAc addition, groups were set for days N-1, 0, 5, and 11. The results indicated that ManNAc addition on Day 11 did not affect Man5 levels, whereas earlier additions proved effective. A full factorial design was employed to assess the interplay between ManNAc concentration and addition timing, revealing no significant interaction. Consequently, it is recommended to administer 20-40 mM ManNAc prior to Day 4. The strategy of introducing 20 mM ManNAc on Day 0 has been successfully implemented across 12 clones, achieving an average Man5 reduction of 46%. Collectively, these findings delineate a novel and efficacious strategy for the Man5 modulation, promising enhanced control over this critical quality attribute in mAbs production.

In the downstream processing of antibody-based therapeutics, ultrafiltration/diafiltration (UF/DF) is commonly applied for concentration and buffer exchange in the final formulation. For a given molecule, various factors such as membrane type, feed flux, and transmembrane pressure (TMP) can significantly influence the performance of UF/DF, impacting yield, buffer exchange efficiency, and product quality. Conventional membrane pore size selection is based on product molecular weight to ensure high retention. While working on an Fc-fusion protein, we found that the pH of load material had a critical effect on the retention of the molecule due to conformational changes at different pH values, as evidenced by the size-exclusion chromatography (SEC). Meanwhile, optimization of the UF/DF process underscored the importance of concentration polarization to protein retention. Approaches to reduce concentration polarization, such as increasing feed flux and lowering TMP, resulted in less protein loss in the permeate stream. High retention of this Fc-fusion protein during the UF/DF step can be achieved not only by utilizing a 5 kDa membrane but also by employing a 10 kDa membrane with optimized process parameters such as load conditions, feed flux, and TMP. These observations provide important insights on the factors impacting protein retention beyond the molecular weight cutoff (MWCO) of UF/DF membrane.

Digital twins (DT) are sophisticated mathematical models representing real-world physical processes, equipped with predictive capabilities that adapt alongside the physical system. The successful implementation of DT in bioprocessing offers numerous advantages, including enhanced understanding of processes, accelerated overall development timelines, and effective monitoring of critical process parameters (CPPs). A comprehensive end-to-end DT can facilitate informed control decisions and forecast how disturbances within the process may affect the final output, accelerating the overall development timelines while optimizing process efficiency and productivity. Tangential flow filtration (TFF) is a standard methodology in bioprocessing, commonly employed to concentrate and exchange buffers for bioproducts. The advancement of continuous process technologies has led to the emergence of alternative TFF methods, notably single-pass tangential flow filtration (SPTFF), which streamlines the process by eliminating the need for stream recirculation. Here, we present the development of a live DT of the SPTFF concentration step within the downstream continuous manufacturing line for a monoclonal antibody (mAb) process. A live DT, equipped with a state estimation tool, was implemented via the Siemens' gPROMS Digital Applications (gDAP) platform. The DT demonstrated the ability to monitor changes in membrane resistance, a typical process parameter that is not directly measured. This parameter is crucial for SPTFF control, as it allows for the constant setting of the concentration factor (CF) by adjusting the retentate flow rate based on the measured resistance and calculated transmembrane pressure (TMP). This achievement illustrates the potential of DT as effective tools for accurately tracking the complete state of the bioprocess.

Industrial fermentation continually improves biological process control for a wide range of microorganisms used in multi‐billion‐dollar industries including industrial enzymes, pharmaceuticals, foods, beverages, commodity chemicals, and bioenergy. In the case of recombinant protein production, batch and fed‐batch phases of fermentation are usually followed by an induction phase, where chemical or thermal induction initiates the expression of a target protein. Fed‐batch processes are usually automated, whereas “out‐of‐the‐box” distributed control systems (DCS) are often unable to define the threshold for induction and respond accordingly. The present study demonstrates the integration of optical density (OD) process analytical technology (PAT) and Lucullus®, a process information management system (PIMS), to enable end‐to‐end automated fermentation at bench and pilot scale. Data aggregated from tens of fermenter runs and hundreds of offline training measurements enabled the development of an accurate multivariate model to predict OD in real‐time. This eliminated the requirement to generate offline correlation models for each OD probe, allowed for model transfer, and incorporated additional predictor terms such as antifoam usage. Automating the induction phase enabled end‐to‐end fermentation, reducing labor and operational costs while increasing yield through higher reactor utilization within the same time period.

Variable lymphocyte receptors (VLRs) are the antigen receptors of jawless vertebrates such as lamprey. VLRs are of growing biotechnological interest for their ability to bind certain antigenic targets with higher affinity than traditional immunoglobulins. However, VLRs are disulfide‐bonded proteins that are often challenging to produce requiring genetic modifications, fusion partners, non‐scalable host cell lines or inclusion body formation and refolding. As a potential VLR expression platform option, the SHuffle Escherichia coli strain has been genetically altered to allow cytoplasmic disulfide bond formation by mutations to thioredoxin reductase (trxB) and glutathione reductase (gor) to create an oxidative cytoplasm. Furthermore, the SHuffle strain expresses disulfide bond isomerase DsbC in the cytoplasm to promote correct disulfide bond pairing. Here, we demonstrate that the SHuffle strain can produce high yield VLRs with titers ranging from 2 to 32 mg of VLR per liter of SHuffle culture. Three VLRs (P1C10, RBC36, VLRA.R2.1) were expressed in SHuffle E. coli and the products were compared directly to those generated using the Rosetta E. coli strain. All VLRs were validated for correct sequence, purity, and activity. For all VLRs, SHuffle E. coli produced 2–9 times more soluble VLRs than Rosetta E. coli. Furthermore, the soluble protein fraction was 2–6 times greater in SHuffle E. coli than Rosetta E. coli for all VLRs. Overall, these results suggest that the E. coli SHuffle strain is a convenient and effective expression system for producing large amounts of VLRs.

Chimeric antigen receptor (CAR) T cell therapy has tremendous potential for the treatment of cancer and other diseases. To manufacture cells of the desired quantity and quality, it is important to expand the CAR T cells ex vivo for an optimal duration. However, identifying the optimal harvest time requires knowledge of the cell concentration during the expansion period. To address this challenge, we have developed a digital shadow of CAR T cell expansion that provides a soft sensor of cell concentration in real‐time. Specifically, a novel mechanistic mathematical model of cell growth within a proportional‐integral‐derivative (PID) controlled perfusion bioreactor has been developed using nonlinear ordinary differential equations. The model is fitted to data generated via bioreactor runs of the Aglaris FACER, in which both donor and patient cells have been expanded in two different media. Off‐line data includes the initial and final cell concentrations, and online data includes the glucose and lactate concentrations as well as the perfusion rate. Training the digital shadow utilizes all the off‐line and online data for each run. In contrast, real‐time testing utilizes only the initial cell concentration and the available online data at the time of model fitting. Real‐time testing shows that with at least 2.5 days of online data, the final cell concentration up to 2.5 days later is predicted with a mean relative error of 13% (standard deviation ≈ 6%). Informative real‐time predictions of cell concentration via the digital shadow can guide decisions regarding the optimal harvest time of CAR T cells.