Glass vial breakage during freezing and lyophilization processes poses a significant concern, as it can lead to product loss and potential safety hazards. In aqueous formulation buffers containing amorphous excipients, vial breakage has been correlated with the strains associated to the plugging-off of the frozen product from the inner vial surface during freezing. This study aimed to systematically identify parameters influencing this strain and to propose risk-mitigating options. The effects of formulation composition, vial type, and freezing process were investigated. The strain was assessed using strain gauge measurements and simulated via finite element analysis. Our findings show that the formulation composition and vial type had the greatest influence on the generated strains. Thermal contraction of the frozen formulation, and thus strain on the vial, occurred only below Tg'. The presence of a monoclonal antibody in trehalose-based formulation buffer led to an increased strain, with higher strain for higher protein concentrations. The strain was independent of the solid content of the formulation within the range probed in this study. The most effective way to mitigate the strain was to promote an earlier plugging-off, e.g. by modifying the vial surface using a hydrophobic coating (water contact angle > 120°). Bake-on siliconized vials showed the most substantial strain reduction, with a fivefold decrease compared to uncoated vials. In contrast, variation in freezing rate and freezing protocol had only marginal impact on the strain formed, as well as vial wall thickness or material. For formulations containing trehalose and mannitol at mass ratios of 2:1 and 1:2, the incorporation of an annealing step was beneficial due to a decreased plugging-off strain. Overall, this study provides valuable insights into the factors affecting the strain on vials during freezing and offers practical measures to minimize this strain and the associated risk of vial breakage, thereby enhancing the safety of frozen drug products.

Dissolved oxygen (DO) has been proposed to influence protein stability during freeze-thaw (F/T) processing, yet its direct impact remains unclear. This study evaluated whether DO levels or vacuum degassing affect the stability of human growth hormone (hGH) during F/T. Buffers were degassed with or without vacuum-induced surface freezing (VISF). Degassing itself did not induce particle formation. Upon freezing substantial aggregation occurred, independent of the presence or absence of dissolved gas. To isolate oxygen effects, hGH solutions equilibrated with nitrogen, air, or oxygen were subjected to F/T cycles or frozen storage at -40°C. DO levels did not significantly influence turbidity, particle counts, intact LC-MS profiles, while only minor differences were observed by SEC. Under metal-catalyzed oxidative conditions, however, oxygen modified aggregation pathways depending on the catalyst. These findings demonstrate that DO alone does not substantially affect hGH stability during F/T but becomes relevant in the presence of catalytic oxidative stress. Protein destabilization during freezing is instead dominated by interfacial stresses, particularly those associated with VISF. Together, these results help to distinguish chemical and physical contributions of dissolved gases in protein formulations.

Protein bulk drug substance (DS) is conventionally kept at -70 to -80 °C to reduce risks such as microbial growth, agitation-related stress, and degradation. This study investigates the long-term physical stability of four IgG-type monoclonal antibodies (mAbs) at -70 and -40 °C under unformulated and fully formulated conditions to probe the broader applicability of -40 °C storage across different antibody formats. For one representative mAb, we further screened minimal formulations and assessed stability at -10 °C. Across all mAbs, -40 °C preserved physical stability comparably to -70 °C. Minimal excipient formulations enabled reliable preservation even above Tg' at -10 °C. The glass transition temperature (Tg') serves as a useful guideline, but the difference between the storage temperature and Tg' is not predictive of stability. Our findings support the adoption of higher frozen-storage temperatures and help guide the optimization of formulations for frozen storage. With this, the study lays the foundation for sustainability, reducing energy consumption without compromising DS quality.

Mechanical stress of protein solutions in contact with a compressible interface can cause protein aggregation. This is a known problem for air-liquid and silicone-liquid interfaces, which occur during processing and handling of biopharmaceuticals. A systematic study comparing and unraveling the mechanism of particle formation at different compressible interfaces is lacking. To this end, we combined novel molecular dynamics simulations and established experimental setups that isolate and precisely define compression-decompression stress to elucidate and compare the mechanism of protein particle formation at the silicone-liquid interface, reflecting tubing used in pumping, and air-liquid interface. Simulations revealed that protein molecules bind rather loosely to the air-liquid interface and show high mobility. During interfacial compression, protein molecules therefore move from the air-liquid interface toward the bulk, reducing protein aggregation. At the silicone-liquid interface, strongly bound protein molecules are forced together upon compression of the adsorbed protein film, promoting particle formation already at little compression. Aggregates detach easily from the air-liquid interface, and compression further facilitates detachment. This enhanced detachment from the air-liquid interface renders similar particle counts in the bulk for both interface types at high interfacial compression, although simulations indicate less aggregate formation directly at the air-liquid interface. Clusters at the silicone-liquid interface break up during relaxation, whereas clusters at the air-liquid interface persist. This, in combination with more easy detachment leads to the formation of smaller particles at the air-liquid interface compared to the silicone-liquid interface. The simulations indicate that at high compression speed, the highly mobile protein molecules at the air-liquid interface do not have sufficient time to interact during compression and form fewer particles. Additionally, strong repulsive protein self-interaction resulting from high charge at low pH values reduced particle formation at the air-liquid interface more strongly due to the high molecular mobility at this interface as compared to the silicone-liquid interface. Our findings provide insights into the mechanisms of protein aggregation at different compressible interfaces, which is essential for developing strategies to mitigate particle formation in biopharmaceutical manufacturing and handling.

Frozen storage or lyophilization of biopharmaceuticals is often used to stabilize drug products. In most cases, the drug is formulated in an aqueous formulation in glass vials. During freezing, physicochemical changes of the formulation, such as excipient crystallization and differences in the coefficients of thermal expansion between glass and the aqueous phase can induce mechanical strain in the vial wall and ultimately lead to glass vial breakage. To systematically characterize strain formation during freezing and its spatial distribution in the vial walls, we used strain gauge measurements and finite element analysis (FEA) simulations. Strain gauge rosette measurements revealed that the direction of principal strains at the vial walls during strain peaks varied with excipients present in the aqueous phase and sensor position. These findings emphasize that correct alignment of linear strain gauges with the principal strain direction is crucial; misalignment can significantly diminish the measured signal and may lead to incorrect conclusions regarding strain distribution and magnitude. Using linear strain gauges, the influence of the gauge position and the vial size was subsequently investigated using a trehalose-based formulation buffer. The measured strain strongly depended on the position of the strain gauge with respect to the fill level, as strain was not evenly distributed throughout the vial wall. Highest strains occurred just below the fill level of the formulation. This was comparable across different vial sizes. Significantly higher strains were observed depending on the freezing method, with shelf cooling producing greater strain than air circulation cooling due to product temperature gradients along the vertical axis of the vial. While locally elevated strains are generated in colder regions, plug detachment is delayed by warmer regions that mechanically constrain the frozen plug. Also, the chamber pressure inside the lyophilizer affected the strain signal. FEA simulations confirmed that the regions of highest strain can be found just below and at the fill level. With increasing fill heights, additional areas of high strain in the lower third of the vial develop which is attributed to a constraining effect of the vial bottom to radial deformation. Overall, the combination of strain gauge measurements and finite element analysis allows systematic identification of parameters governing vial strain formation during freezing, enabling identification of critical parameters relevant to vial breakage during freezing.

This study investigates the long-term stability of a monoclonal antibody (mAb) in selected formulations containing histidine buffer, sucrose, and 2-hydroxypropyl-β-cyclodextrin (HPβCD). Samples were stored for up to 24 months at -80 °C, -40 °C, and - 10 °C to assess whether more energy-efficient storage temperatures can maintain drug quality. We analyzed key physical stability attributes by monitoring high-molecular-weight species, subvisible particles, and optical density, while chemical stability was assessed by monitoring methionine oxidation. A primary focus was evaluating whether physical stability in the frozen state correlates with the formulation's glass transition temperature (Tg') as an indicator of matrix mobility. We report that storage at -40 °C and, in many cases, even at -10 °C, offers stability comparable to -80 °C, making these temperatures viable and more sustainable long-term storage options. Formulations containing HPβCD as a novel approach for freeze-thaw and frozen storage demonstrated the best stability, likely due to its combined cryoprotective and interfacial-shielding properties, which reduce both aggregation and particle formation. Data demonstrate a drastic stability change at Tg', with substantially increased stability below Tg'. However, above and below Tg', the temperature difference relative to Tg' had only minor influence compared to formulation effects. We therefore hypothesize that the sharp change in matrix mobility primarily governs frozen-state stability. The observation that several formulations remained stable at -10 °C highlights that formulation-dependent mechanisms can compensate for increased matrix mobility, offering promising potential for more flexible, sustainable frozen storage conditions.

Intrinsic tryptophan (Trp) fluorescence emission (ITFE) spectroscopy is widely used in biologics development, yet interpretation remains challenging for proteins with more than one Trp, such as antibodies, and is often limited to bulk spectral shifts. For better understanding and more detailed characterization of mAbs by ITFE, we analyzed an in-house therapeutic IgG1 (mAbT) and the National Institute of Standards and Technology (NIST mAb) reference IgG1. Human serum albumin (HSA) was used as a single-Trp protein control. Steady-state, second-derivative and time-resolved fluorescence measurements combined with Molecular Dynamics simulations enabled site-resolved interpretation of ITFE signals. A limited number of dominant Trp residues, primarily near the complementary-determining regions (CDRs) in the VH domains, were identified as the main contributors to the overall emission. Long fluorescence lifetimes (>5 ns) correlated with moderate solvent accessibility (around 25%) and low quenching propensity. The direction of spectral shifts during unfolding and aggregation was governed by the environment of the individual VH domain Trp. The increase in fluorescence intensity during denaturation was primarily attributed to unfolding and was further enhanced by thermal stress induced aggregation. Protein aggregation induced by interfacial stress produced similar spectral signatures in mAbT and HSA, indicating comparable effects on ITFE. This study demonstrates that integration of ITFE with molecular dynamic simulations allows for mechanistic interpretation beyond just an empirical description. By thoroughly interpreting ITFE spectra, one can obtain information about spectral properties of variable Trps residues in IgG1 molecules and gain valuable insights into protein denaturation mechanisms.

Lyophilization is a key technology to improve the long-term stability of protein drug products, traditionally using the disaccharides sucrose and trehalose for cryo- and lyoprotection. Monosaccharides are less favored due to the low glass transition temperature and Maillard reaction potential. Additionally, trisaccharides and tetrasaccharides typically do not play significant roles, as they are often not approved for parenteral use and have been associated with lower protein stabilization. Key stability parameters include the preserved protein structure, solid-state accessibility, and monomer content. This study explores the long-term monomer retention of human serum albumin (HSA) in lyophilizates at 2-8, 25, and 40 °C by investigating the effect of a series of mono- to tetrasaccharides based on glucose (glucose, maltose, maltotriose, and maltotetraose) as well as glucose maltose and glucose maltotriose mixtures. We varied the residual moisture (RM) content (1, 1.5, and 2%) postlyophilization to understand the effects of water replacement, vitrification, and matrix mobility on protein stability. The molar ratios of maltose to HSA were set at 360:1 and 180:1 to investigate the impact of the sugar concentration at overall low sugar ratios. Solid-state hydrogen-deuterium exchange mass spectrometry (ssHDX MS) was performed on a QDa benchtop mass spectrometer to evaluate protein accessibility and structural preservation using RMs of 1% D2O, 2% D2O, and 1% D2O + 1% H2O. The larger the sugar, the lower its stabilizing potential and the higher the protein accessibility, indicating insufficient water replacement. Increasing the RM from 1 to 1.5 and 2% enhanced stability, highlighting the superiority of residual water molecules, which was especially the case for the tri- and tetrasaccharides. Mixtures of small and large sugars showed stabilization benefits in maintaining the monomer content and structural preservation, indicating a good balance of water replacement and vitrification. Overall, the ssHDX MS findings of samples with headspace-spiked D2O did correspond with monomer retention, indicating that it could be a valuable tool for characterization and understanding the stabilizing capacity of lyophilized formulations. Our findings highlight the importance of RM control for optimal stability as well as the importance of the sugar size on lyoprotection based on water replacement and the potential of sugar mixtures to optimize the stability of lyophilized proteins.

The paper investigates structure of freeze-dried formulations of a therapeutical monoclonal IgG1 with non-reducing disaccharides, sucrose and trehalose, as lyoprotectors. Formulations with variable sugar-to-protein ratios are manufactured using different freeze-drying protocols and post-drying temperature treatments. Small-angle X-ray scattering (SAXS) is applied to study packing arrangements of IgG1 molecules in the crowded solid-state environment. The retention of amorphous structure by the lyoprotectors is confirmed with the simultaneous wide-angle X-ray scattering (WAXS) tests. The new finding is the observation of two IgG1 SAXS peaks in the crowded environment, with position of one peak (peak A) changing with sugar-to-protein ratio, whereas the second peak (peak B) position is similar in all the formulations tested. The protein center of mass distance, which is calculated from the position of the peak A, increases with the increase in sugar content from 45 Å to 65 Å (sugar-to-protein mass ratio 0.79 to 7.95). Sugar molecules therefore serve as a physical barrier between IgG1 molecules. While the sugar-to-protein ratio significantly impacts protein separation, other factors, such as disaccharide type, presence of a surfactant, or drying process and post-drying thermal treatment, have minimal effect. The origin of the peak B has not been established yet, with three hypotheses considered. The results highlight applicability of SAXS for quantification of separation distances between protein molecules in freeze-dried formulations, thus improving the fundamental understanding of the stability of protein drugs.

The implementation of high-throughput methods for fuelling the design of effective nanocarriers for RNA delivery remains challenging. Traditional experimental screening is resource-intensive, while purely computational approaches face limitations, such as data scarcity for machine learning models and the high computational cost of molecular dynamics simulations. This work introduces a high-throughput virtual screening platform, ″Bits2Bonds,″ integrating coarse-grained molecular dynamics simulations with machine learning-driven optimization to design novel poly(β-amino ester) (PBAE) carriers for therapeutic siRNA delivery. The platform evaluates virtual polymers using MD-based ″challenges″ that simulate key hurdles in nucleic acid delivery, such as membrane and siRNA interaction (association/dissociation). The computational framework was calibrated and validated against experimental data, including synthesis and characterization of four distinct PBAEs, logP measurements, siRNA encapsulation assays, and cell culture knockdown experiments. This integrated approach provides a powerful tool for the de novo design and rapid virtual screening of optimized polymeric siRNA delivery systems.

The frozen storage of biopharmaceuticals brings new challenges to the primary packaging material. Due to an increasing demand and the downsides of standard type I glass vials, such as vial breakage, novel vial types for special applications of parenteral drug products have been introduced to the market in the past years. Mechanical stresses due to dimensional changes experienced during freezing and thawing could change the material properties, hence affecting the interaction with the drug product stored in the vial or functionality such as overall integrity. Therefore, we studied the suitability of different vial qualities related to the thermally induced mechanical stresses experienced during frozen drug product preparation and storage. First, the possible failure modes for each vial type were identified. The interaction between vial surface and drug product were investigated considering surface hydrophobicity, surface free energy and surface roughness as well as microscopically visible changes analyzed by confocal laser scanning microscopy. Differences in surface hydrophobicity, roughness and surface free energy between the vial types did not impact the performance upon freeze-thaw stress and did not change with the stress. Screening the vial content for particles originating from the container using light and electron scanning microscopy combined with energy-dispersive X-ray spectroscopy showed only rare cases of particles in coated glass vials. Under extreme stress conditions, including a drop-test in the frozen state, a low number of particles was also detected in coated polymer vials. No quality issues regarding the functionality were observed upon container closure integrity testing, while the oxygen permeability was slightly increased for uncoated and especially coated polymer vials. Overall, the results show that several vial types are appropriate for the frozen storage of drug products and selection should be based on the formulation and other product requirements.

During the pandemic, lipid nanoparticles (LNPs) became widely established as RNA nanocarriers, and hold the promise of future targeting of a broad variety of previously untreatable diseases. LNPs are mostly administered invasively via intramuscular or intravenous injections. Given the lung's large surface, high vascularization and low nuclease abundance, inhalation offers a promising alternative for both local and systemic delivery of LNPs. Vibrating mesh nebulizers present a patient-friendly, high-dose delivery platform. However, the nebulization process imposes thermal and mechanical stress on the LNP formulation. This study contributes to a better understanding of how nebulization affects the physicochemical properties and biological activity of LNPs, depending on formulation and process parameters. We investigated the impact of formulation and process variables such as temperature, concentration, buffer type, and RNA modality on LNP properties including particle size distribution, zeta potential, in vitro activity, and RNA integrity. While aggregating, siRNA LNPs protected the encapsulated RNA from degradation, and preserved biological function. In contrast, after the nebulization of mRNA LNPs the cargo was degraded and the biological function diminished. This observation can possibly be attributed both to the higher sensitivity of mRNA toward physical and chemical degradation, and the cargo-dependent morphology of LNPs. While demonstrating that siRNA LNPs preserved their most important characteristics, namely RNA integrity and biological function, our findings emphasize the need for route-specific optimization of LNPs, which need to meet different critical quality criteria when used for inhalation rather than injection.

Interfacial stress during peristaltic pumping can lead to particle formation in biopharmaceutical solutions. Since the impact of formulation on protein particle formation is not fully understood, we combined molecular dynamics (MD) simulations with experimental methods to investigate and understand the effects of pH, ionic strength, and protein type during peristaltic pumping. Building on our previous work, we improved the MD model to provide a more accurate representation of a protein solution at the polymer interface. Our results indicate that the pH value affects aggregate formation in a human growth hormone solution, both while protein molecules are adsorbed to the interface and during the detachment of aggregates into the bulk. Both steps were also directly influenced by protein-protein interactions. Studies at high ionic strength suggest that when protein self-interaction is similar, the amount of protein molecules adsorbed to the interface can be decisive of the extent of particle formation. Additional studies employing lysozyme as a second protein confirmed that protein-protein interactions are the key factor in protein aggregation at interfaces, validating the MD model and our findings across different low-molecular-weight proteins. Our study uncovers the specific points of action through which formulation parameters influence protein particle formation upon mechanical interfacial stress. Furthermore, our model enables the prognosis of protein particle formation in silico, potentially saving resources in formulation and process development.

Coarse-grained molecular dynamics simulations are highly valuable for studying protein-protein interactions. Unfortunately, commonly used force fields often overestimate these interactions. Here, we investigate the performance of the Martini 3 force field in predicting the self-interaction behavior of lysozyme and subtilisin using Metadynamics. The original Martini 3, despite improvements over its predecessor, overestimates interaction strength. Through reparameterization of bead interactions, we achieve good agreement with experimental data of the second virial coefficient and the diffusion coefficient. The new, refined force field enables more accurate CG-MD simulations, with potential applications in understanding and prediction of protein stability, aggregation tendencies, and solubility, with the possibility to aid in the development of protein-based drugs.

The aggregation of proteins is a major threat to the integrity of biopharmaceutical products. Typically the state of aggregation at a specific timepoint is evaluated via particle analysis and counting or turbidity. Backgrounded Membrane Imaging (BMI) is a recently introduced methodology that provides a low-volume, high-throughput alternative to be used in biopharmaceutical development. Recent work has successfully evaluated BMI as an orthogonal method regarding its counting and sizing accuracy for subvisible particle analysis. The work at hand shows that apart from background noise, stochastic variations need to be considered to define the lower limit of detection. A systematic evaluation of particle identification robustness shows that particles at the lower and upper size limit of the technique are not reliably detected. To overcome potential biases due to particle crowding and overlapping, novel evaluation parameters are introduced: the Total Area, the Total Intensity and the BMI-Z-Average to be reported alongside the particle count. Overall, we were able to refine root causes for loss in data quality in BMI and to showcase the use of additional reporting parameters to shift focus to more robustly-identified and quantified larger particles.

Protein-based drugs are prone to both physical and chemical instability in aqueous solutions. Surfactants, such as polysorbates (PS), are commonly employed to mitigate interfacial stress, thus preventing protein aggregation and particle formation. However, polysorbates can undergo enzymatic hydrolysis by residual host cell proteins and oxidation during long-term storage in parenteral formulations. This can lead to the generation of free fatty acid particles, inadequate protein stabilization, and protein oxidation. In this study, we investigated several monoacyl phospholipids (MAPLs) with varying fatty acid chains as potential alternative surfactants for monoclonal antibody (mAb) formulations and compared their efficacy to the industry standard, polysorbate 80. The hemolytic activity of MAPLs was tested using erythrocytes in 95 % plasma. All MAPLs prevented mAb particle formation during shaking and freezing-thawing at surfactant concentrations several orders of magnitude below the threshold for hemolysis, suggesting that the risk of erythrocyte damage from MAPLs is non-critical. Stabilization of mAbs occurred around the critical micelle concentration, which were comparable to that of PS80, but MAPLs achieved lower interfacial tension values. MAPLs were found to be more resistant to enzymatic hydrolysis by porcine liver esterase and forced oxidation than PS80. After long-term liquid storage, lyso-myristoyl-phosphatidylcholine (LPC 14:0) at low concentrations provided superior mAb stabilization to PS80, which exhibited substantial chemical degradation. At higher concentrations, both PS80 and LPC 14:0 showed a decrease in surfactant concentration. Lyophilization enhanced mAb stabilization relative to liquid formulations, with MAPLs performing as well as PS80 at high concentrations and outperforming PS80 at low concentrations. MAPLs also better preserve the siliconization in pre-filled syringe (PFS) barrels compared to PS80. In short, MAPLs demonstrate mAb stabilization and chemical stability comparable to, and in some cases superior to, PS80, making them a promising alternative as interfacial stabilizers in parenteral protein formulations and warranting further exploration.

There is still an insufficient understanding of how the characteristics of protein drugs are maintained in the solid state of lyophilizates, including aspects such as protein distances, local environment, and structural preservation. To this end, we evaluated protein folding and the molecules' nearest environment by electron paramagnetic resonance (EPR) spectroscopy. Double electron-electron resonance (DEER) probe distances of up to approximately 200 & Aring; and is suitable to investigate protein folding, local concentration, and aggregation, whereas electron spin echo envelope modulation (ESEEM) allows the study of the near environment within approximately 10 & Aring; of the spin label. We spin-labeled human serum albumin (HSA) and freeze-dried different concentrations with 100 g/L deuterated sucrose. DEER showed distinct local concentration behaviors for two folding states, directly correlating folding percentage with the interprotein distance, reaching 2 nm at an HSA concentration of 84 g/L. Interestingly, 50% of the HSA molecules showed partial structural perturbation already at 2.6 g/L, which corresponds to a molar ratio Suc/HSA of 7469. This percentage increased to 97% with an increase in the HSA concentration to 84 g/L. The degree of protein perturbation cannot be told, and no signs of unfolding are found after reconstitution. ESEEM demonstrated a higher sucrose concentration around the protein label compared to the HSA environment in highly concentrated sucrose solutions. The partial unfolding detected in DEER could lead to label exposure and explain the enhanced sucrose detection in the intimate shell. Our work provides new insights regarding sucrose enrichment in the nearest shell of proteins upon lyophilization. In addition, the results indicate substantial partial structural perturbation, even in the presence of enormous supplies of stabilizing sugars. Thus, pulse EPR spectroscopy allows additional understanding of the solid state of protein lyophilizates, which is complementary to SANS, FTIR, or ssNMR.

Repeated compression and dilation of a protein film adsorbed to an interface lead to aggregation and entry of film fragments into the bulk. This is a major mechanism for protein aggregate formation in drug products upon mechanical stress, such as shaking or pumping. To gain a better understanding of these events, we developed a molecular dynamics (MD) setup, which would, in a later stage, allow for in silico formulation optimization. In contrast to previous approaches, the molecules of our model protein human growth hormone displayed realistic shapes, surfaces, and interactions with each other and the interface. This enabled quantitative assessment of protein cluster formation. Simulation outcomes aligned with experimental data on subvisible particles and turbidity, thereby validating the model. Computational and experimental results indicated that compression speed does not affect the aggregation behavior of preformed protein films but rather their regeneration. Protein clusters that formed during compression disassembled upon relaxation, suggesting that the particles originate from a partly compressed state. Desorption studies via steered MD revealed that proteins from compressed systems are more likely to detach as clusters, implying that compression effects at the interface translate into aggregates present in the bulk solution. With the possibility of studying the impact of different variables upon compression and dilation at the interface on a molecular level, our model contributes to the understanding of the mechanisms of protein aggregation at moving interfaces. It also enables further studies to change formulation parameters, interfaces, or proteins.