Continuous chromatography for Protein A capture of monoclonal antibodies (mAbs) and related biotherapeutics is gaining popularity for higher throughput and reduced costs but introduces challenges for development and modeling. We present a streamlined, single-breakthrough-curve approach that predicts maximum and operating binding capacities (MBC and OBC, respectively) for continuous Protein A chromatography of monoclonal antibodies and derivatives. Our empirically-calibrated model-intended for early-stage, material-limited process development-enables accurate OBC prediction (root mean squared error [RMSE]≤ 5% for four distinct proteins) using less than 1 g of protein. This approach reduces the experimental timeline by ~2 weeks and is best suited for rapid screening and initial process optimization, not full design-space determination. Results confirm the model's accuracy across molecules, with clear practical advantages for accelerating preclinical pipeline advancement.

Perfusion bioreactor systems enable intensified bioprocessing with high cell densities and volumetric productivities in monoclonal antibody (mAb) manufacturing. However, their widespread adoption remains constrained by the high consumption of chemically defined cell culture media. In this study, we introduce a multi-stage electrokinetic waste separation process that regenerates spent media, recovering up to 87.5% of the spent-medium volume for reuse. Using a mock-perfusion model, the regeneration-recycle processes were shown to sustain high-density Chinese hamster ovary cell growth, compared with spent media recycling without regeneration. Integration into a self-recycling perfusion bioreactor (350 mL working volume) operating with 75% regenerated media sustained the target viable cell density and antibody titer, with limited reductions to cell-specific growth rate (~31%) and cell-specific productivity (~9%), supporting the feasibility of high-ratio media recycling. Techno-economic modeling of large-scale scenarios revealed that media regeneration could offer a path to $15/g COGs, thereby increasing the accessibility of mAbs. Techno-economic modeling of hypothetical large-scale manufacturing scenarios indicated that the experimentally demonstrated process could reduce cost of goods (COG) by approximately 9%, and identified a theoretical pathway toward ~$15/g COG contingent upon multi-cycle regeneration processes that could be developed in the future. These results support regenerative perfusion as a promising process-intensification strategy that, with further development, may be adaptable to other mammalian cell culture platforms where media cost and sustainability are critical bottlenecks.

Reliable assessment of cellular state remains a challenge in cell therapy manufacturing and regenerative medicine. Proliferative capacity, senescence, and metabolic activity are traditionally measured using labor-intensive assays, limiting timely and integrated assessment of cellular state. Here, we apply attenuated total reflection-Fourier transform infrared spectroscopy coupled with multivariate analysis to assess whether intrinsic vibrational fingerprints of mesenchymal stem cells reflect coordinated functional changes during in vitro expansion. By integrating proliferation kinetics, metabolite fluxes, and senescence-associated phenotype into a composite functional score, we mapped progressive functional decline across passages onto infrared spectral changes. Spectral alterations were associated with shifts in lipid-related and protein-associated regions as senescence accumulated. Despite strong donor-specific differences, we identified conserved spectral fingerprint regions that tracked functional decline across donor samples analyzed. These shared biochemical signatures were associated with the prediction of senescence progression and the estimation of functional passage in an independent donor sample. Together, our results show that intrinsic vibrational fingerprints capture coordinated biochemical changes during cellular decline and provide a label-free, rapid means to assess cellular state.

Predictive modeling of phototrophic cultures in photobioreactors remains challenging because growth emerges from the coupling between radiative transfer, intracellular bioenergetics, and physiological acclimation. Although existing approaches have progressively integrated light attenuation and reactor-scale heterogeneity, extending knowledge-based formulations to eukaryotic microalgae still raises difficulties, especially regarding the role of respiration under illumination and the dynamic adjustment of pigment content. In this work, we establish a biochemically structured predictive model of photoautotrophic growth for the eukaryotic microalga Chlamydomonas reinhardtii. The model is built from an explicit stoichiometric decomposition of the main metabolic functions involved in biomass synthesis, pigment synthesis, photosynthetic energy conversion, respiration, and ATP-consuming futile processes under redox regulation. Its structure is derived from intracellular conservation relationships and observability analysis, leading to a reduced formulation driven by two variables: the net conversion rate of photochemically productive photons within the biotic phase, q γ , and a dissipative ATP sink. The kinetic formulation explicitly couples growth to radiative transfer and accounts for dynamic pigment acclimation through variable partitioning of biomass formation between residual biomass and pigments. Model parameters were either fixed from previous physiological and bioenergetic analyses or identified from batch-culture experiments performed under different incident photon flux densities in a flat-panel photobioreactor illuminated from one side. The model satisfactorily predicts biomass and pigment dynamics in batch and continuous cultures over a broad range of light conditions and dilution rates, while comparisons with oxygen-exchange measurements under illumination provide additional support for its structural relevance. The proposed framework provides a predictive and mechanistically interpretable description of eukaryotic microalgal growth in photobioreactors.

Recombinant adeno-associated virus (rAAV) production commonly yields a high proportion of empty and partially filled capsids, necessitating downstream purification to enrich full capsids. Among purification approaches, anion-exchange chromatography (AEX) has emerged as a scalable alternative to ultracentrifugation. In this study, a structured and scalable development strategy for AAV5 AEX purification is presented. A high-throughput, ultra-scale-down platform using 200 μL RoboColumns™ was established to enable rapid screening of salt systems, additives, and operating conditions with minimal material requirements. Dual-salt systems were evaluated to decouple ionic strength from ion-specific effects, leading to the identification of a magnesium-only system (MgCl2/MgSO4) that improved both enrichment and recovery. Optimized conditions achieved >85% full capsids with robust recovery and >90% full capsids in peak fractions. Scalability was demonstrated through strong agreement between ultra scale-down (200 μL) and scale-down (1 mL) formats. A QbD-aligned framework was applied to define the purity-recovery design space and guide selection of operating conditions based on target product profile. The optimized gradient process was translated into a manufacturing-relevant isocratic step format, with systematic evaluation of buffer conductivity sensitivity and load robustness to address large-scale operational constraints.

Development of therapeutic bispecific antibodies (BsAbs) poses significant manufacturability challenges associated with correct chain pairing and an increased diversity of impurities. Here we address this with a product quality assessment during cell line development (CLD), forgoing the need for genetic characterization during cell line screening. Using Chinese Hamster Ovary (CHO) cell lines expressing a panel of monoclonal antibodies (mAb) and BsAbs, we demonstrate the highly specific binding of SpotLight™ Huλ, Huĸ and HuFc reagents to their target chain isotypes (lambda or kappa light chain and Fc regions) during single cell cloning workflows on a Bruker Beacon® Optofluidic System. We then demonstrate combinatorial use of SpotLight™ Huλ and Huĸ reagents as an early product quality indicator on a CHO cell line expressing a BsAb with dual kappa-lambda light chain configuration. Cell lines presenting equal binding profiles for each SpotLight assay exhibited platform-typical heterodimer levels with an average of 87.26% following fed-batch bioreactor production assessment. Conversely, cell lines biased towards either SpotLight™ assay resulted in lower than desired heterodimer (0%-84.23%) and increased associated impurities. This study demonstrates the application of dual SpotLight™ assays as an identifier of CHO cell lines expressing a kappa-lambda light chain BsAb with a high heterodimer percentage. This allows cell lines with optimal product quality to be identified earlier in CLD, resulting in fewer cell lines being progressed to resource-intensive fed-batch studies and subsequent product quality assessment. Furthermore, this assay provides an opportunity to characterize impurity expression biases prior to developing downstream purification processes.

The sustainable production of hydroxy fatty acids, such as 10-hydroxystearic acid (10-HSA), by biocatalysis is a promising alternative to petrochemical and castor oil-derived products. However, industrial implementation still requires an efficient and scalable process for biocatalyst production by fermentation, which precedes the biotransformation to 10-HSA. In general, industrial biocatalysis is commonly performed using whole cells, which are cheaper than purified enzymes. Still, the fermentation process itself is a major cost contributor requiring a high-yielding and scalable process. Here, we established and scaled a lactose-induced fed-batch process to produce an Escherichia coli BL21 (DE3) based whole-cell biocatalyst containing an oleate hydratase from Stenotrophomonas nitritireducens. The two-phased process employs an initial growth phase on glucose, followed by an induced feed phase using glycerol and lactose. When comparing two different growth rates during enzyme expression, a higher growth rate was found beneficial, resulting in a higher biomass concentration of 69.2 ± 0.5 g L[-1] and a yield increase of 80 % while maintaining biocatalyst activity at >85 % conversion. This highlights the importance of process design variables, such as growth rate settings, for a high-yielding and economic process. Feasibility of the process was demonstrated by scaling into a 150 L bioreactor, achieving a biomass concentration of 60.6 ± 0.4 g L[-1] with a yield of 0.44 gC,Biomass gC [-1] and 91.5 ± 1.4 % conversion. Supplemented with an initial test of suitable unit operations for technical biomass separation, this work provides a fermentation route for a whole-cell oleate hydratase biocatalyst, paving the way for further scaling toward industrial 10-HSA production.

Physiologically relevant lung-on-chip (LoC) systems remain constrained by polydimethylsiloxane (PDMS) limitations including small-molecule adsorption and mechanical deformation of device microchannels under perfusion, single-channel architectures limiting parallel tissue culture, and absence of organized stromal and immune compartments. Existing platforms predominantly model the distal airway, leaving the upper airway, the primary site of respiratory viral entry, insufficiently addressed as an in vitro tissue model for preclinical drug evaluation. We report a six-well, 3D-printed LoC platform fabricated using a photopolymer resin that addresses these limitations. Computational fluid dynamics (CFD)-guided design ensured uniform perfusion across all six wells, enabling parallel tissue culture within a single device. The photopolymer architecture eliminated mechanical deformation of the device microchannels under prolonged perfusion and demonstrated significantly reduced small-molecule adsorption relative to PDMS, improving quantitative reliability for drug evaluation workflows. Human tracheobronchial epithelial cells differentiated at air-liquid interface expressed MUC5AC and cytokeratin-14, formed continuous ZO-1 tight junctions, and achieved a transepithelial electrical resistance of ~800 Ω·cm[2], consistent with physiological barrier integrity. A GelMA-based stromal compartment, fabricated by extrusion bioprinting of lung fibroblast- and THP-1 monocyte-laden bioink, established a layered epithelial-stromal-immune architecture. Functional validation demonstrated IL-6 and IL-8 elevation following SARS-CoV-2 spike S1 stimulation, IFN-γ-driven modulation of apical ACE2 expression, and coordinated TNF-α, IL-6, and IL-8 release during Influenza A infection, all suppressed by oseltamivir treatment. This platform integrates CFD-optimized multi-well device fabrication, extrusion bioprinting of stromal constructs, and photopolymer resin-based microfluidic manufacturing into a reproducible, human-relevant upper airway model for respiratory infection research and preclinical antiviral drug evaluation.

Production of monoclonal antibodies (mAbs) in Chinese hamster ovary (CHO) cells is affected by the "Warburg effect" which is characterized by high lactate accumulation. Therefore, the present study explores the use of pyruvate dehydrogenase kinase (PDK) inhibitors dichloroacetate (DCA), metformin (M), (S)-3,3,3-trifluoro-2-hydroxy-2-methyl propionic acid (TPP), and 4-phenylbutyrate (PB) to counteract this metabolic bottleneck. To achieve optimal mAb yields, individual inhibitors (0.5-10 mM) were screened, followed by evaluation of their combinatorial effects. A combination of 5 mM DCA and 1 mM TPP (DCA + TPP) revealed ~2-fold enhancement in mAb yield (1.2 g/L) under fed-batch conditions, accompanied by a marked reduction in lactate accumulation as compared to control. The optimized DCA + TPP combination also exhibited 15%-20% enhancement in the mAb yield as compared to a single PDK inhibitor that is, DCA (5 mM) and TPP (1.0 mM) at the same concentration. Metabolic analysis revealed a 19.5-fold increase in pyruvate flux and a 2-fold increase in intracellular acetyl-CoA. The critical product quality attributes exhibited insignificant impact on charge variant (<3%), aggregation, and glycosylation of mAb. The amino acid consumption profiles predominantly showed depletion of glycine and tryptophan, whereas threonine and aspartic acid were primarily consumed in the control. Proteomic analysis exhibited upregulation and downregulation of various proteins, which significantly contribute to mAb's yield. Validation in Trastuzumab-producing CHO-S cells confirmed a 30% increase in mAb yield and a 10% improvement in cell density. These outcomes demonstrate that the combination of DCA + TPP is a promising metabolic intervention to mitigate the Warburg effect and enhance recombinant protein production in mammalian systems.

The presence of host cell proteins (HCPs) as process-related impurities in monoclonal antibody (mAb) therapeutics presents risks, including immunogenicity, product instability, and reduced therapeutic efficacy. Therefore, HCP levels are seen as a critical quality attribute which requires monitoring. Characterization of residual HCPs is essential during process development and after process changes in late project phases including post-submission changes. Mass spectrometry-based analysis of HCPs is increasingly employed to overcome the limitations of ELISAs, particularly in identifying individual low-abundance HCPs. In this study, we systematically compared three enrichment strategies: hexapeptide enrichment, immunoaffinity chromatography (IAC), and native digest (ND), to enhance LC-MS/MS-based detection of low-abundance HCPs across five recombinant products expressed in CHO cells. Our results show that both ND and IAC substantially expand HCP detection beyond hexapeptide enrichment and in-solution digest in total number of HCP identification. While IAC supported the detection of several potentially critical HCPs, ND uniquely identified additional low-abundance HCPs not captured by antibody-based methods. Furthermore, pull-down assays revealed product-specific HCPs likely interacting with the active pharmaceutical ingredient, suggesting that standard purification steps may be insufficient for their removal. Together, our findings emphasize the complementary nature of ND and IAC for in-depth HCP profiling. We propose the combined application of both methods to ensure comprehensive detection of potentially critical HCPs in biopharmaceutical products, particularly when facing issues with low product or polysorbate stability caused by HCPs with enzymatic activities. This integrative approach supports enhanced product safety and control of consistent HCP depletion in the process.

High performance countercurrent membrane purification (HPCMP) has emerged as a promising diffusive separation strategy for the purification of recombinant proteins and monoclonal antibodies (mAbs). However, prior studies employed dialysis membranes, which are designed for diffusive separations but have not been validated for biopharmaceutical manufacturing. This study evaluates the ability of commercial ultrafiltration (UF) membranes, engineered for pressure-driven filtration, to support highly selective diffusive separations. Scanning electron microscopy (SEM) analysis of the membrane cross-sections revealed marked morphological differences between UF and dialysis membranes. Yet, transport studies using polydisperse dextrans and model proteins demonstrated that UF membranes can deliver exceptional diffusive separation selectivity. In particular, a 100 kDa polysulfone membrane (Cytiva) provided >150-fold diffusive selectivity between bovine serum albumin (BSA) and immunoglobulin G (IgG). When utilized for mAb separation from a clarified CHO cell culture fluid, this membrane achieved >98% product yield, near-complete removal of low molecular weight species, and significant reduction in both DNA and host cell protein (HCP) levels. These results demonstrate that commercial UF membranes that are validated for bioprocessing applications can be repurposed for HPCMP, enabling scalable, continuous downstream processing of biotherapeutics.

Virus safety for biopharmaceutical products has been a well-established concern for several decades. While various virus reduction steps have been identified and implemented for traditional plasma-derived and monoclonal antibody products, there are various challenges to achieving the same virus safety in emerging gene therapy products. Virus removal filtration, or nanofiltration, has proven difficult to implement for larger gene therapy modalities, though there is a regulatory expectation to incorporate this step for smaller modalities such as adeno-associated virus (AAV) vector products. Here, we demonstrate the applicability of a virus removal filter with a 35 nm mean pore size distribution for AAV processes based on filtrations with porcine parvovirus (PPV) as a model for AAV vector and simian virus 40 (SV40) as a model contaminant virus. Filtration runs were performed at low pressure and included a defined process pause, two manufacturing conditions known to potentially impact viral clearance. All filtrations demonstrated SV40 reduction to below the limit of detection and no detectable reduction in PPV, indicating robust removal of large contaminating viruses without impacting product yields for AAV products. Thus, virus removal filtration is appropriate for the removal of adventitious viruses in gene therapy product manufacturing. Virus removal filters have an established history of improving patient safety and should be included in the manufacture of advanced gene therapy products whenever possible.

The vaccine and viral vector industry is growing at an accelerated rate. To improve harvest and purification processes, the development of continuous membrane-based operations, such as normal flow filtration (NFF) and single pass tangential flow filtration (SPTFF) for concentration were explored. This work was conducted using two model viruses, non-enveloped porcine parvovirus (PPV) and enveloped Suid herpesvirus (SuHV). The viruses are in the same family as the gene therapy vectors adeno associated virus and herpes simplex virus, respectively. SPTFF design started with batch TFF for membrane selection. Hollow fiber membranes with a 100 and 300 kDa molecular weight cut off were defined for PPV and SuHV SPTFF operations, respectively. The SPTFF runs for PPV did not provide any concentration of the virus and low protein and DNA removal, unlike batch TFF. Two hollow fiber membranes run at 10 mL/min and 2 psi were the best condition for SuHV concentration, with approximately 100-fold titer concentration and protein and DNA removal of 37% ± 5% and 32% ± 8%, respectively. This concentration was superior to the batch TFF and indicated a strong dependence on flow rate and transmembrane pressure. For NFF, filters selection and performance tests were carried out for NFF of PPV and SuHV, as well as cleaning protocols for hollow fiber membranes. The ultimate goal is to integrate this work into the continuous purification of viral vectors produced in mammalian cell cultures to reduce costs and increase throughput.

Maintaining consistent quality in the manufacturing of biotherapeutic proteins in mammalian cell culture is challenging, with unplanned deviations causing inconsistencies and potential batch failure. Current methods for monitoring and controlling critical process parameters (CPPs) rely on slow, labor-intensive offline analyses. This is particularly problematic in upstream manufacturing, where infrequent measurements hinder real-time CPP monitoring and result in suboptimal control. This study developed and implemented a robust, standardized framework that integrates Raman spectroscopy with real-time machine learning models and bioreactor control via OPC-UA, enabling seamless communication and effective control of mammalian cell culture CPPs. High-accuracy ML models (R[2] >0.92) enabled real-time monitoring of glucose, lactate, viable cell density, and antibody titre. The glucose model was integrated into an automated process analytical technology (PAT) control strategy to maintain glucose at a predefined setpoint. The PAT strategy was compared to a manual bolus glucose control under high and low glucose feeding regimes. The PAT control approach observed an increase in biotherapeutic product titre by up to 35% and reduced glycation by up to 27%. These results indicated that both glucose setpoints and fluctuations impacted cell culture performance. In conclusion, this study highlighted the value of PAT tools in automating control loops, consequently improving process performance with enhanced productivity and quality.

In biopharmaceutical manufacturing, protein aggregation is a critical quality attribute, necessitating rapid and reliable analytical strategies during downstream processes like anion-exchange chromatography (AEX). While Raman spectroscopy enables continuous monitoring of protein secondary structure, standard data-driven regression models struggle to decouple intrinsic structural changes from gradient-induced solvent and buffer drifts under dynamic chromatographic conditions. Addressing this methodological gap, this study establishes a constrained pseudo-Voigt hard modeling framework for the mechanistic deconvolution of bovine serum albumin (BSA) size variants during in-line Raman monitoring of AEX processes. By explicitly defining a parametric background model to capture salt-induced spectral drift, the methodology effectively isolates matrix variations from genuine protein-specific signals. The constrained hard model was applied to 285 in-line spectra across diverse chromatographic conditions, achieving reconstruction fidelity while maintaining stable, physically interpretable component identities. The mechanistically derived Amide I center of mass emerged as a robust, aggregation-sensitive descriptor that preserves structural information despite strong concentration dynamics. Furthermore, the extracted spectral features demonstrated strong predictive performance for monomer concentration and acceptable accuracy for high molecular weight components. Collectively, these results demonstrate that constrained spectral hard modeling provides a highly interpretable, robust, and calibration-light alternative to classical partial least squares approaches for the real-time monitoring of protein size variants.