Host cell proteins (HCPs) complicate the purification of recombinant proteins produced in Chinese hamster ovary (CHO) cells, but most mechanistic knowledge derives from monoclonal antibody processes. This review examines how that knowledge applies to His-tagged proteins purified by immobilized metal affinity chromatography (IMAC). Secreted products and membrane-associated products enter purification through different routes, with extracellular release dominating the former and fractionation and co-extraction shaping the latter. HCP priorities are assessed from evidence of a relevant harmful effect, persistence or activity, and intended use. Four overlapping retention routes are considered: metal coordination, support or ligand adsorption, target association, and carriage within multicomponent assemblies. Direct CHO/IMAC studies address capture and process-level HCP reduction, whereas antibody and other-host studies provide mechanistic precedents requiring evaluation in the intended matrix. We distinguish product recovery from impurity clearance and relate analytical findings to feed preparation, chromatography, and selected upstream interventions. Particular attention is given to detergent- and lipid-containing preparations, for which solubility, association state, and analytical recovery can change together. The available evidence supports impurity-specific control linked to product quality, while highlighting the need for direct studies of individual HCPs in CHO/IMAC processes.

Sterile filtration represents the final critical step in biomanufacturing, where a sterilizing-grade membrane filter removes microbial contamination from the drug product - making filter integrity verification essential. While bubble point and gas-liquid diffusion tests are established non-destructive methods for hydrophilic membranes, diffusion testing offers high sensitivity to small defects and operates at lower test pressures. However, applying diffusion testing to small-area filters integrated into closed single-use assemblies remains challenging because pressure changes measured by automated integrity testers can arise not only from membrane diffusion, but also from assembly-related effects such as gas-permeable tubing, upstream volume and thermal equilibration. For low-area devices, these background contributions can become comparable to or greater than the membrane signal, reducing test sensitivity and complicating interpretation. This work presents a practical method for implementing diffusion-based integrity testing of small-format sterilizing-grade filters in final sealed single-use assemblies. The approach quantifies assembly-specific background contribution and incorporates it into an adjusted diffusion specification, allowing membrane-specific diffusion to be resolved without altering the assembled system. By accounting for system-derived measurement noise, the method enables reliable end-user integrity assessment of closed assemblies that cannot be reconfigured to reduce upstream volume or replace high-permeability tubing materials.

Chinese hamster ovary (CHO) cell culture underpins commercial manufacturing of most therapeutic monoclonal antibodies, yet productivity and product quality remain constrained by inhibitory metabolic by-products that accumulate across batch, fed-batch, and perfusion processes. While lactate and ammonium have historically dominated monitoring and control strategies, untargeted metabolomics and stable-isotope tracing have substantially expanded the inventory of bioprocess-relevant inhibitors. This review synthesizes current understanding of 17 inhibitory metabolites spanning eight metabolic pathways. These include branched-chain amino acid catabolites (isobutyrate, isovalerate, and 2-hydroxyisocaproate), aromatic amino acid derivatives (phenyllactate, indole-3-carboxylate), nucleotide-pathway by-products (cytidine monophosphate, guanosine monophosphate, and succinyladenosine), and apoptosis-correlated species such as glycerol-3-phosphocholine and oxidized glutathione. Reported concentration thresholds are compiled for better-studied inhibitors, while correlative evidence linking accumulation to apoptosis, reduced productivity, and altered glycosylation is summarized. We evaluate the analytical workflow combining LC-MS and NMR for discovery, Raman spectroscopy for in-line monitoring, and stable-isotope tracing for pathway validation. Control strategies are critically compared across three tiers: media and feed reformulation (amino acid limitation, dynamic pH, or oxygen-uptake-rate-driven feeding), targeted and multiplex genetic engineering (BCAT1 knockout, combined LDHA/PDK editing), and keto-acid precursor substitution. We close by identifying concrete gaps, chiefly the absence of cell-line-independent inhibition thresholds, and argue that coupling genome-scale metabolic models with PAT data is the logical next step toward predictive, closed-loop mAb manufacturing.

Raman spectroscopy is a promising in-situ Process Analytical Technology for upstream bioprocesses, but transferring trained calibration models to new, independent fermentation runs without re-calibration remains a challenge due to sparse sample references (SaR) and prediction drift caused by biomass-induced fluorescence and turbidity effects. We present SIM-CaliRam, a calibration workflow combining two physically motivated interventions. First, Simulation-Informed References (SiR) densify glucose, ammonium, and phosphate labels using a feed-driven, non-reactive mass balance on a dedicated uninoculated non-reactive feed run, where substrate dynamics are determined by feeding and dilution. SiR labels agreed closely with the SaR values. Additionally, Biomass-augmented Feature (BaF) appends Raman-predicted biomass as an additional substrate-model regressor, providing a proxy for residual biomass-induced spectral confounding in the feature space. The workflow is demonstrated on Saccharomyces cerevisiae fermentations and evaluated within a systematic grid-search over established preprocessing options and six calibration methods spanning linear, Bayesian nonparametric, and neural network approaches. Conventional SaR calibration serves as the baseline. Training uses two AV runs, with model selection on a separate CV run and final evaluation on a held-out TEST run. The best selected models achieve NRMSETEST of 0.203 (biomass, SaR), 0.206 (glucose, SiR), 0.350 (ammonium, SiR + BaF), and 0.598 (phosphate, SiR). SiR improved phosphate prediction (R[2] = 0.64) relative to SaR (R[2] = -1.2), although the remaining prediction error limits standalone process-control application. Selected preprocessing, learner, and augmentation configurations differed across analytes. These results support target-specific model selection rather than a universal calibration pipeline.

Chinese hamster ovary (CHO) cells are the predominant platform for biologics production. Expression of multi-subunit proteins within a single CHO cell requires balanced subunit expression to achieve optimal product quality and yield. Through systematic analysis of CHO expression data generated at the discovery stage using co-transfected monocistronic vectors, we found that subunit expression balance strongly correlates with product purity across both monoclonal and bispecific antibodies. These insights enabled the rational design of multicistronic vectors for production cell line development. For bispecific antibodies containing Fab and single-chain variable fragment (scFv) arms, scFv chains were frequently overexpressed, contributing to reduced product purity. Increasing heavy chain (HC) expression, for example through the introduction of an additional HC gene copy, improved both purity and titer across multiple programs. Likewise, optimizing the relative positioning of knob and hole HC genes within bicistronic vectors helped balance chain expression and enhance product purity in a complex IgG-like molecule. Overall, early-stage monocistronic expression analysis provides a framework for rational vector design of multi-subunit biologics, enabling more efficient cell line development with the potential to accelerate preclinical timelines and reduce resource utilization.

Highly intensified monoclonal antibody (mAb) production needs next-generation purification technologies for process streamlining. Cationic flocculants aid in clarification and can reduce DNA and host cell proteins (HCPs) decreasing the burden on purification. Using two different mAbs, we screened five different cationic flocculants for their potential to clarify cell culture fluid, while simultaneously removing host cell DNA and HCPs without affecting antibody yield and integrity. We identified Polymer X as the best performer, excelling in HCP reduction with 30%-50% reduction compared to an un-treated reference. Comparison to a commercially available polymer (Polymer D), showed a significantly higher and more robust reduction of HCPs for four different mAbs. Additionally, we investigated the reduction of high-risk HCPs, that are known to negatively impact final product quality or patient safety. We demonstrated that Polymer X excels in reduction of those compared to flocculation with Polymer D. Polymer X could be used to successfully (1) aid cell culture clarification, with achieved turbidities of less than 25 NTU, (2) maintain high mAb titers, with over 95% yield (in three out of four mAbs), (3) reduce host cell DNA by over 99% and (4) reduce total HCPs by up to 40%, including selected high-risk HCPs to be removed by over 90%. In a typical mAb purification scheme, an anion exchange chromatography is used to get rid of DNA and residual HCPs. The robust removal of HCPs and DNA shown by Polymer X might support streamlining of the purification.

Increasing protein therapeutics production by adopting intensified perfusion cultures places high demand on the mass transfer capabilities of single-use bioreactors (SUB). Here, we compare these capabilities for four SUB designs against worst-case O2 transfer requirements for supporting Chinese hamster ovary (CHO) perfusion cultures. Impact of power dissipation, gas flow rate, O2 enrichment, pCO2 and sparge-hole size on O2 and CO2 mass transfer (OTR and CTR) was characterized. Only one design delivered sufficient O2 and removed sufficient CO2 to support a worst-case cell line at 100 × 10[6] viable cells/mL with a single-sparger at both 50 and 1000 L scale. Larger-scale versions of 3 designs were also tested. Performance at a 50-L scale was not predictive of large-scale performance (200 or 1000 L) for any design. Changes in the volume and composition of the gas phase during residence in the vessels were characterized. Substantial changes were found. Calculations of predicted culture pCO2 were performed across the design space, assuming a respiratory quotient of 1. A response surface was constructed to predict culture pCO2 and used to examine the consequences of various strategies for increasing OTR. Most strategies disproportionately increase OTR relative to CTR, leading to an increase in culture pCO2. The implications on SUB design and scale-up for high-intensity perfusion cultures are discussed.

Vero cells (referred to as "Vero" henceforth) are a commonly used host for the production of a wide variety of vaccines, but due in part to the typically anchorage-dependent nature of the cells, Vero cells also pose multiple challenges during process development, scale-up, and commercialization. The adaptation of Vero to proliferate in a serum-free suspension culture while remaining non-tumorigenic can be a challenging endeavor, and where successful, has often utilized undisclosed and/or complex custom-designed medium formulations. Additional and expanded options for commercially available serum-free media that can support the robust expansion of suspension-adapted Vero cells for viral vaccine manufacturing would be a significant advantage for increasing the industrial adoption of suspension Vero, as virus production can also be highly medium-dependent and sensitive. In this work, we have successfully transitioned suspension-adapted Vero cells from a proprietary medium to a commercially-available base medium (ExCell VPRO) with supplementation of additional commercially-available components, and subsequently infected the resulting ExCell VPRO cell bank lineage with an attenuated vaccine strain Mumps virus. Like many suspension cell line media formulations, off-the-shelf commercially available ExCell VPRO medium contains dextran sulfate, which serves to minimize cell clumping, but is also well-known to directly impact the infectivity of some viruses. Unexpectedly, we have observed that the presence of dextran sulfate in the infection medium increased both culture longevity and infectious virus titers for attenuated Mumps, potentially eliminating the need for a medium exchange between cell growth and infection for a more streamlined and easily scalable attenuated Mumps virus production process.

Production of human papillomavirus (HPV) virus-like particles (VLPs) in Komagataella phaffii is often constrained by methanol-associated toxicity and suboptimal feeding strategies during induction. In this study, a microbioreactor-based approach was employed to systematically design and evaluate mixed-substrate fed-batch strategies for enhanced HPV16 L1 VLP production. Four methanol-based feeding compositions, including methanol-only and combinations with glycerol and sorbitol, were investigated under pulse and linear feeding modes. Among the tested conditions, a linear feeding strategy using 70% methanol and 30% glycerol (70 M + 30G) resulted in the highest biomass (DCW: 90.36 g/L) and VLP titer (42 mg/L), corresponding to a two-fold improvement over methanol-only feeding. Enhanced performance was attributed to reduced methanol-induced metabolic stress and improved substrate utilization, leading to more stable dissolved oxygen (DO) profiles and sustained cellular activity. In contrast, pulse feeding resulted in significant DO fluctuations, indicating transient metabolic imbalances and lower overall productivity. The optimized feeding strategy was successfully translated to a 1 L stirred-tank bioreactor, achieving a DCW of 120 g/L and a VLP titer of 49 mg/L, with a ~ 2.3-fold increase compared to the control. The consistency between microscale and bench-scale results demonstrates the predictive capability of microbioreactor systems for fed-batch process development. Overall, this study presents a rational and scalable feeding strategy that improves recombinant VLP production while reducing methanol usage, providing a practical framework for efficient bioprocess design in K. phaffii.

Regeneron's bispecific technology, based on a fully human IgG antibody construct, consists of two distinct heavy chains and a common light chain co-expressed in a single cell. The co-expression of the two distinct heavy chains, one of which contains a dipeptide substitution that ablates Protein A binding, leads to the formation of three species with varying degrees of avidity to Protein A: the heterodimer bispecific and two parental homodimer impurities. The avidity difference between the three species can be leveraged for both bulk capture of the bispecific and resolution of the bispecific from the homodimer impurities via a ternary resolving Protein A chromatographic separation on commercially available resins amenable to large scale Good Manufacturing Practice (GMP) production. This case study highlights the efficiency of a mechanistic modeling approach for the optimization and characterization of ternary resolving Protein A unit operations. A mechanistic model, comprised of a general rate transport model and colloidal particle adsorption (CPA) affinity isotherm, was developed to describe this ternary separation for a model bispecific. The model was calibrated and validated with 10 laboratory experiments and demonstrated an ability to predict bispecific yield within 3% and bispecific purity within 1% on average. The model was then used to optimize the process and characterize the design space, while providing mechanistic insight into the process.

The deployment of advanced modeling and machine learning in biopharmaceutical process control has been limited by legacy fragmented automation infrastructure and a lack of model lifecycle governance. We present BioOps, a modular automation framework for biopharmaceutical advanced process control (APC). The framework incorporates established Machine Learning Operations (MLOps) principles, including version-controlled model registries, containerized deployment, and continuous integration/continuous delivery (CI/CD) pipelines. BioOps explicitly decouples model development from model execution within the control architecture. This decoupling enables process scientists to deploy, test, and iteratively refine mathematical models and control strategies in a flexible manner. These workflows can be applied across heterogeneous bioreactor platforms and are executed entirely in-house, without reliance on vendor-specific solutions or custom system integration. In addition, BioOps embeds traceability, auditability, and model lifecycle management directly into the control stack. As a result, the framework aligns with regulatory expectations while continuing to support agile, in-house experimentation and development. Case studies across a geographically distributed set of bioreactors demonstrate BioOps' versatility in PAT-driven feedback control, hybrid model-based nutrient and glucose feeding, adaptive phase transitions, and cross-scale deployment. By operationalizing model-centric control through an MLOps-inspired architecture, BioOps provides a practical foundation for scalable, reproducible, and future-ready biomanufacturing.

Upstream continuous manufacturing of biologic therapeutics has been an attractive method to address growing demand. However, maintaining high cell densities (>100 million cells/mL) for long periods of time (>25 days) requires large amounts of media, complicating logistics and presenting a roadblock to increased adoption. To reduce volumetric media use, intensified media recipes have been developed. This study evaluates two such recipes to reduce perfusion rate by up to 30%: the first, a concentrated media formulation to increase nutritional depth, and the second, a commercial perfusion rate reducing supplement. We observe how the cell process, productivity, and product quality are impacted. Direct media intensification stalls cell growth by increasing osmolality, ultimately decreasing cell viability. Meanwhile, media with the perfusion rate reducing supplement maintains cell viability and productivity and has no significant impact on product quality compared to the control. We demonstrated the generalizability of the perfusion rate reducing supplement by testing with a different media and cell clone and observed similar results. Cell cycle analysis demonstrated that both intensified media strategies successfully stall cells in the G0/G1 phase, but the perfusion rate reducing supplement more effectively arrests cell proliferation, suggesting a larger shift from growth-driven cells in glycolysis to production-driven cells in oxidative metabolism, and likely contributing to its superior performance. These findings demonstrate important considerations in media intensification and introduce the perfusion rate reducing supplement as a viable media supplement to reduce volumetric media consumption while maintaining process performance.

Haptoglobin (Hp) is a polymorphic acute phase α-2 glycoprotein found in plasma that plays a critical role in binding, neutralizing, and removing cell-free hemoglobin (Hb) from the circulation. Under clinical conditions characterized by high levels of hemolysis, such as in patients with sickle cell disease (SCD), large quantities of cell-free Hb are released from lysed red blood cells (RBCs) into the circulation and bind to Hp. This interaction reduces the plasma Hp concentration below basal levels and diminishes its Hb-binding capacity. Therefore, plasma-derived Hp has the potential to be used therapeutically to scavenge, neutralize, and remove excess cell-free Hb from the blood, thus preventing Hb-mediated toxicity. This provides strong motivation to purify Hp at high purity using sustainable sources, such as waste plasma fractions from the Cohn plasma fractionation process used to produce human serum albumin from pooled plasma. Starting from human Cohn fraction IV, we first enriched an Hp-rich fraction using tangential flow filtration (TFF), which was then used as the starting material for purification. Hp was subsequently purified from this fraction using hydrophobic interaction chromatography (HIC) to homogeneity. The final Hp purity reached 98% by SDS-PAGE densitometry and 87% by trypsin digest LC-MS/MS analysis. The purified Hp was further characterized to determine its molecular weight, secondary structure, Hb-binding capacity, and binding kinetics using biophysical techniques including MALDI-TOF, circular dichroism, size exclusion HPLC, and UV-visible stopped-flow spectroscopy. More importantly, the simplicity and efficiency of the TFF-HIC workflow demonstrated strong potential for scalable Hp production.

Protein A chromatography is a key unit operation in biopharmaceutical purification, widely used to capture monoclonal antibodies (mAbs) from harvest cell culture fluid (HCCF). In addition to its primary function, protein A chromatography is also frequently tested for its ability to remove viruses. Although protein A specifically binds mAbs and should not interact with viruses, some studies have shown that logarithmic reduction values (LRVs) can vary between 1 and 4 log10 for retroviruses. This variance has been shown to depend on the mAb and feedstock compositions. However, it has not been evaluated so far which feedstock ingredients affect virus removal. To address this knowledge gap, a data set of virus spiking studies was analyzed and revealed a potential correlation between DNA content in HCCF and LRV of protein A chromatography. To evaluate this hypothesis, experiments with murine leukemia virus (MuLV) and retrovirus like particles (RVLP) spiked starting materials of three different mAbs were performed. In general, higher HCCF DNA levels were found to enhance LRV for both virus particles. A trend that was confirmed with RVLP spiked pure mAb. In addition, protein A chromatography washing buffers containing a detergent or a reducing agent were shown to improve LRV in RVLP spiked experiments. This study demonstrates the effect of DNA concentration on the virus removal capabilities of protein A chromatography. Findings that are especially useful when DNA is removed at early stages of the downstream process, for example by using functionalized depth filters or precipitation.

Recombinant adeno-associated virus production for gene therapy applications commonly relies on the Spodoptera frugiperda 9 (Sf9) baculovirus infection platform. The process encompasses cell expansion from flasks through intermediate-scale bioreactors to large-scale production. A critical challenge in this process is suboptimal cell proliferation during the expansion phase, which significantly impacts subsequent virus production. This study evaluates the effects of different single-use bioreactor bags on Sf9 cell growth, focusing on two potential issues: incompatibility between Sf9 cells and bag materials, and the presence of leachables and extractables from the bag. Single-use rocking motion bioreactor bag configuration, including material- and surface-related attributes, was associated with substantial differences in Sf9 cell expansion performance. While growth kinetics in shake flasks and the Pall Allegro XRS system were comparable, reduced growth rates and delayed proliferation were observed in Cytiva Cellbags. The comparable growth observed in shake flasks and the Pall XRS system, together with the lack of recovery after rocking-parameter adjustments in Cellbags, suggests that bag-associated material and surface-related factors, rather than rocking motion alone, contributed substantially to the observed growth differences. Lower bag occupancy further exacerbated growth inhibition. Washing studies showed no improvement in new bags, whereas partial recovery after initial use suggested transient bag conditioning or surface-related effects. Analytical extractables and leachables profiling was not performed; therefore, leachable-mediated contributions cannot be fully excluded. These findings highlight the importance of single-use bag selection and occupancy optimization for robust and scalable Sf9 cell expansion processes.