Mixed-mode size-exclusion silica resin for polishing human antibodies in flow-through mode

The polishing step in the downstream processing of therapeutic antibodies removes residual impurities from Protein A eluates. Among the various classes of impurities, antibody fragments are especially challenging to remove due to the broad biomolecular diversity generated by a multitude of fragmentation patterns. The current approach to fragment removal relies on ion exchange or mixed-mode adsorbents operated in bind-and-gradient-elution mode. However, fragments that bear strong similarity to the intact product or whose biophysical features deviate from the ensemble average can elude these adsorbents, and the lack of a chromatographic technology enabling robust antibody polishing is recognized as a major gap in downstream bioprocessing. Responding to this challenge, this study introduces size-exclusion mixed-mode (SEMM) silica resins as a novel chromatographic adsorbent for the capture of antibody fragments irrespective of their biomolecular features. The pore diameter of the silica beads features a narrow distribution and is selected to exclude monomeric antibodies, while allowing their fragments to access the pores where they are captured by the mixed-mode ligands. The static and dynamic binding capacity of the adsorbent ranged respectively between 30-45 and 25-33 gs of antibody fragments per liter of resin. Selected SEMM-silica resins also demonstrated the ability to capture antibody aggregates, which adsorb on the outer layer of the beads. Optimization of the SEMM-silica design and operation conditions - namely, pore size (10 nm) and ligand composition (quaternary amine and alkyl chain) as well as the linear velocity (100 cm/h), ionic strength (5.7 mS/cm), and pH (7) of the mobile phase - afforded a significant reduction of both fragments and aggregates, resulting into a final antibody yield up to 80% and monomeric purity above 97%.

Factors affecting product association as a mechanism of host-cell protein persistence in bioprocessing

Product association of host-cell proteins (HCPs) to monoclonal antibodies (mAbs) is widely regarded as a mechanism that can enable HCP persistence through multiple purification steps and even into the final drug substance. Discussion of this mechanism often implies that the existence or extent of persistence is directly related to the strength of binding but actual measurements of the binding affinity of such interactions remain sparse. Two separate avenues of investigation of HCP-mAb binding are reported here. One is the measurement of the affinity of binding of individual, commonly persistent Chinese hamster ovary (CHO) HCPs to each of a set of mAbs, and the other uses quantitative proteomic measurements to assess binding of HCPs in a null CHO harvested cell culture fluid (HCCF) to mAbs produced in the same cell line. The individual HCP measurements show that the binding affinities of individual HCPs to different mAbs can vary appreciably but are rarely very high, with only weak pH dependence. The measurements on the null HCCF allow estimation of individual HCP-mAb affinities; these are typically weaker than those seen in affinity measurements on isolated HCPs. Instead, the extent of binding appears correlated with the initial abundance of individual HCPs in the HCCF and the forms of the HCPs in the solution, i.e., whether HCPs are present as free molecules or as parts of large aggregates. Separate protein A chromatography experiments performed by feeding different fractions of a mAb-containing HCCF obtained by size-exclusion chromatography (SEC) showed clear differences in the number and identity of HCPs found in the protein A eluate. These results indicate a significant role for HCP-mAb association in determining HCP persistence through protein A chromatography, presumably through binding of HCP-mAb complexes to the resin. Overall, the results illustrate the importance of considering more fully the biophysical context of HCP-product association in assessing the factors that may affect the phenomenon and determine its implications. Knowledge of the abundances and the forms of individual or aggregated HCPs in HCCF are particularly significant, emphasizing the integration of upstream and downstream bioprocessing and the importance of understanding the collective properties of HCPs in addition to just the biophysical properties of individual HCPs.

Mechanistic model-based characterization of size-exclusion-mixed-mode resins for removal of monoclonal antibody fragments

Although antibody fragments are a critical impurity to remove from process streams, few platformable purification techniques have been developed to this end. In this work, a novel size-exclusion-mixed-mode (SEMM) resin was characterized with respect to its efficacy in mAb fragment removal. Inverse size-exclusion chromatography showed that the silica-based resin had a narrow pore size distribution and a median pore radius of roughly 6.2 nm. Model-based characterization was carried out with Chromatography Analysis and Design Toolkit (CADET), using the general rate model and the multicomponent Langmuir isotherm. Model parameters were obtained from fitting breakthrough curves, performed at multiple residence times, for a mixture of mAb, aggregates, and an array of fragments (varying in size). Accurate fits were obtained to the frontal chromatographic data across a range of residence times. Model validation was then performed with a scaled-up column, altering residence time and feed composition from the calibration run. Accurate predictions were obtained, thereby illustrating the model's interpolative and extrapolative capabilities. Additionally, the SEMM resin achieved 90% mAb yield, 37% aggregate removal, 29% [Formula: see text] removal, 54% Fab/Fc removal, 100% Fc fragments removal, and a productivity of 72.3 g mAbL×h. Model predictions for these statistics were all within 5%. Simulated batch uptake experiments showed that resin penetration depth was directly related to protein size, with the exception of the aggregate species, and that separation was governed by differential pore diffusion rates. Additional simulations were performed to characterize the dependence of fragment removal on column dimension, load density, and feed composition. Fragment removal was found to be highly dependent on column load density, where optimal purification was achieved below 100 mg protein/mL column. Furthermore, fragment removal was dependent on column volume (constant load mass), but agnostic to whether column length or diameter was changed. Lastly, the dependence on feed composition was shown to be complex. While fragment removal was inversely related to fragment mass fraction in the feed, the extent depended on fragment size. Overall, the results from this study illustrated the efficacy of the SEMM resin in fragment and aggregate removal and elucidated relationships with key operational parameters through model-based characterization.

Exploring preferred binding domains of IgG1 mAbs to multimodal adsorbents using a combined biophysics and simulation approach

In this work, we employ a recently developed biophysical technique that uses diethylpyrocarbonate (DEPC) covalent labeling and mass spectrometry for the identification of mAb binding patches to two multimodal cation exchange resins at different pH. This approach compares the labeling results obtained in the bound and unbound states to identify residues that are sterically shielded and thus located in the mAb binding domains. The results at pH 6 for one mAb (mAb B) indicated that while the complementarity determining region (CDR) had minimal interactions with both resins, the FC domain was actively involved in binding. In contrast, DEPC/MS data with another mAb (mAb C) indicated that both the CDR and FC domains were actively involved in binding. These results corroborated chromatographic retention data with these two mAbs and their fragments and helped to explain the significantly stronger retention of both the intact mAb C and its Fab fragment. In contrast, labeling results with mAb C at pH 7, indicated that only the CDR played a significant role in resin binding, again corroborating chromatographic data. The binding domains identified from the DEPC/MS experiments were also examined using protein surface hydrophobicity maps obtained using a recently developed sparse sampling molecular dynamics (MD) approach in concert with electrostatic potential maps. These results demonstrate that the DEPC covalent labeling/mass spectrometry technique can provide important information about the domain contributions of multidomain proteins such as monoclonal antibodies when interacting with multimodal resins over a range of pH conditions.

Identification and characterization of CHO host-cell proteins in monoclonal antibody bioprocessing

Host-cell proteins (HCPs) are the foremost class of process-related impurities to be controlled and removed in downstream processing steps in monoclonal antibody (mAb) manufacturing. However, some HCPs may evade clearance in multiple purification steps and reach the final drug product, potentially threatening drug stability and patient safety. This study extends prior work on HCP characterization and persistence in mAb process streams by using mass spectrometry (MS)-based methods to track HCPs through downstream processing steps for seven mAbs that were generated by five different cell lines. The results show considerable variability in HCP identities in the processing steps but extensive commonality in the identities and quantities of the most abundant HCPs in the harvests for different processes. Analysis of HCP abundance in the harvests shows a likely relationship between abundance and the reproducibility of quantification measurements and suggests that some groups of HCPs may hinder the characterization. Quantitative monitoring of HCPs persisting through purification steps coupled with the findings from the harvest analysis suggest that multiple factors, including HCP abundance and mAb-HCP interactions, can contribute to the persistence of individual HCPs and the identification of groups of common, persistent HCPs in mAb manufacturing.

Comparative Analysis of Protein Surface Hydrophobicity Maps Determined by Sparse Sampling INDUS and Spatial Aggregation Propensity

Protein surface hydrophobicity plays a central role in various biological processes such as protein folding and aggregation, as well as in the design and manufacturing of biotherapeutics. While the hydrophobicity of protein surface patches has been linked to their constituent residue hydropathies, recent research has shown that protein surface hydrophobicity is more complex and characterized by the response of water to these surfaces. In this work, we employ water density perturbations to map the surface hydrophobicity of a set of model proteins using sparse indirect umbrella sampling simulations (SSI). This technique is used to identify hydrophobic surface patches for the set of model proteins, and the results are compared to those obtained from the widely adopted spatial aggregation propensity (SAP) technique. While SAP-based calculations show agreement with SSI in some cases, there are several examples of disagreement. We identify four general classes of difference in behavior and study factors that contribute to these differences. We find that the SAP method can sometimes mask the effect of weakly nonpolar or isolated nonpolar residues that can lead to strong hydrophobic patches on the protein surface. In addition, hydrophobic patches identified by SAP can exhibit shifts in both position and strength on the SSI map. Our results demonstrate that the combination of topography and chemical context controls the hydrophobicity of a given patch above and beyond the intrinsic polarity of the residues present on the patch surface. The availability of more accurate protein hydrophobicity maps in concert with new classes of hydrophobic molecular descriptors may create significant opportunities for in silico prediction of protein behavior for a range of applications, such as protein design, biomanufacturability, and downstream bioprocessing.

Continuous precipitation-filtration process for initial capture of a monoclonal antibody product using a four-stage countercurrent hollow fiber membrane washing step

The significant increase in product titers, coupled with the growing focus on continuous bioprocessing, has renewed interest in using precipitation as a low-cost alternative to Protein A chromatography for the primary capture of monoclonal antibody (mAb) products. In this work, a commercially relevant mAb was purified from clarified cell culture fluid using a tubular flow precipitation reactor with dewatering and washing provided by tangential flow microfiltration. The particle morphology was evaluated using an inline high-resolution optical probe, providing quantitative data on the particle size distribution throughout the precipitation process. Data were obtained in both a lab-built 2-stage countercurrent washing system and a commercial countercurrent contacting skid that provided 4 stages of continuous washing. The processes were operated continuously for 2 h with overall mAb yield of 92 ± 3% and DNA removal of nearly 3 logs in the 4-stage system. The high DNA clearance was achieved by selective redissolution of the mAb using a low pH acetate buffer. Host cell protein clearance was 0.59 ± 0.08 logs, comparable to that based on model predictions. The process mass intensity was slightly better than typical Protein A processes and could be significantly improved by preconcentration of the antibody feed material.

Implications of AAV affinity column reuse and vector stability on product quality attributes

In this work, the implications of AAV9 capsid design and column reuse on AAV9 vector product quality were assessed with POROS CaptureSelect (PCS) AAVX and AAV9 resins using sf9 insect cell-derived model AAV9 vectors with varying viral protein (VP) ratios. Chromatographic experiments with purified drug substance AAV9 model feeds indicated consistent vector elution profiles, independent of adeno-associated virus (AAV) VP ratio, or cycle number. In contrast, the presence of process impurities in the clarified lysate feeds resulted in clear changes in the elution patterns. This included increased aggregate content in the vector eluates over multiple cycles as well as clear differences in the performance of these affinity resin systems. The AAV9-serotype specific PCS AAV9 column, with lower vector elution pH, resulted in higher aggregate content over multiple cycles as compared to the serotype-independent PCS AAVX column. Further, the results with vectors of varying VP ratio indicated that while one vector type eluate displayed higher aggregation in both affinity columns over column reuse, the eluate with the other vector type did not exhibit changes in the aggregation profile. Interestingly, vector aggregates in the affinity eluates also contained double-stranded DNA impurities and histone proteins, with similar trends to the aggregate levels. This behavior upon column reuse indicates that these host cell impurities are likely carried over to subsequent runs due to incomplete clean-in-place (CIP). These results indicate that feed impurities, affinity resin characteristics, elution pH, column CIP, and vector stability can impact the reusability of AAV affinity columns and product quality.

Characterization and implications of host-cell protein aggregates in biopharmaceutical processing

In the production of biopharmaceuticals such as monoclonal antibodies (mAbs) and vaccines, the residual amounts of host-cell proteins (HCPs) are among the critical quality attributes. In addition to overall HCP levels, individual HCPs may elude purification, potentially causing issues in product stability or patient safety. Such HCP persistence has been attributed mainly to biophysical interactions between individual HCPs and the product, resin media, or residual chromatin particles. Based on measurements on process streams from seven mAb processes, we have found that HCPs in aggregates, not necessarily chromatin-derived, may play a significant role in the persistence of many HCPs. Such aggregates may also hinder accurate detection of HCPs using existing proteomics methods. The findings also highlight that certain HCPs may be difficult to remove because of their functional complementarity to the product; specifically, chaperones and other proteins involved in the unfolded protein response (UPR) are disproportionately present in the aggregates. The methods and findings described here expand our understanding of the origins and potential behavior of HCPs in cell-based biopharmaceutical processes and may be instrumental in improving existing techniques for HCP detection and clearance.

Isotherm model discrimination for multimodal chromatography using mechanistic models derived from high-throughput batch isotherm data

In this work, we have examined an array of isotherm formalisms and characterized them based on their relative complexities and predictive abilities with multimodal chromatography. The set of isotherm models studied were all based on the stoichiometric displacement framework, with considerations for electrostatic interactions, hydrophobic interactions, and thermodynamic activities. Isotherm parameters for each model were first determined through twenty repeated fits to a set of mAb - Capto MMC batch isotherm data spanning a range of loading, ionic strength, and pH as well as a set of mAb - Capto Adhere batch data at constant pH. The batch isotherm data were used in two ways-spanning the full range of loading or consisting of only the high concentration data points. Predictive ability was defined through the model's capacity to capture prominent changes in salt gradient elution behavior with respect to pH for Capto MMC or unique elution patterns and yield losses with respect to gradient slope for Capto Adhere. In both cases, model performance was quantified using a scoring metric based on agreement in peak characteristics for column predictions and accuracy of fit for the batch data. These scores were evaluated for all twenty isotherm fits and their corresponding column predictions, thereby producing a statistical distribution of model performances. Model complexity (number of isotherm parameters) was then considered through use of the Akaike information criterion (AIC) calculated from the score distributions. While model performance for Capto MMC benefitted substantially from removal of low protein concentration data, this was not the case for Capto Adhere; this difference was likely due to the qualitatively different shapes of the isotherms between the two resins. Surprisingly, the top-performing (high accuracy with minimal number of parameters) isotherm model was the same for both resins. The extended steric mass action (SMA) isotherm (containing both protein-salt and protein-protein activity terms) accurately captured both the pH-dependent elution behavior for Capto MMC as well as loss in protein recovery with increasing gradient slope for Capto Adhere. In addition, this isotherm model achieved the highest median score in both resin systems, despite it lacking any explicit hydrophobic stoichiometric terms. The more complex isotherm models, which explicitly accounted for both electrostatic and hydrophobic interaction stoichiometries, were ill-suited for Capto MMC and had lower AIC model likelihoods for Capto Adhere due to their increased complexity. Interestingly, the ability of the extended SMA isotherm to predict the Capto Adhere results was largely due to the protein-salt activity coefficient, as determined via isotherm parameter sensitivity analyses. Further, parametric studies on this parameter demonstrated that it had a major impact on both binding affinity and elution behavior, therein fully capturing the impact of hydrophobic interactions. In summary, we were able to determine the isotherm formalisms most capable of consistently predicting a wide range of column behavior for both a multimodal cation-exchange and multimodal anion-exchange resin with high accuracy, while containing a minimized set of model parameters.

Evaluation of preferred binding regions on ubiquitin and IgG1 FC for interacting with multimodal cation exchange resins using DEPC labeling/mass spectrometry

There is significant interest in identifying the preferred binding domains of biological products to various chromatographic materials. In this work, we develop a biophysical technique that uses diethyl pyrocarbonate (DEPC) based covalent labeling in concert with enzymatic digestion and mass spectrometry to identify the binding patches for proteins bound to commercially available multimodal (MM) cation exchange chromatography resins. The technique compares the changes in covalent labeling of the protein in solution and in the bound state and uses the differences in this labeling to identify residues that are sterically shielded upon resin binding and, therefore, potentially involved in the resin binding process. Importantly, this approach enables the labeling of many amino acids and can be carried out over a pH range of 5.5-7.5, thus enabling the protein surface mapping at conditions of interest in MM cation exchange systems. The protocol is first developed using the model protein ubiquitin and the results indicate that lysine residues located on the front face of the protein show dramatic changes in DEPC labeling while residues present on other regions have minimal or no reductions. This indicates that the front face of ubiquitin is likely involved in resin binding. In addition, surface property maps indicate that the hypothesized front face binding region consists of overlapping positively charged and hydrophobic patches. The technique is then employed with an IgG1 F_C and the results indicate that residues on the C_H 2-C_H 3 interface and the hinge are significantly sterically shielded upon binding to the resin. Further, these regions are again associated with significant overlap of positively charged and hydrophobic patches. On the other hand, while, residues on the C_H 2 and the front face of the IgG1 F_C also exhibited some changes in DEPC labeling upon binding, these regions have less distinct charged and hydrophobic patches. Importantly, the hypothesized binding patches identified for both ubiquitin and F_C using this approach are shown to be consistent with previously reported NMR studies. In contrast to NMR, this new approach enables the identification of preferred binding regions without the need for isotopically labeled proteins or chemical shift assignments. The technique developed in this work sets the stage for the evaluation of the binding domains of a wide range of biological products to chromatographic surfaces, with important implications for designing biomolecules with improved biomanufacturability properties.

Connecting Non-Gaussian Water Density Fluctuations to the Lengthscale Dependent Crossover in Hydrophobic Hydration

We connect density fluctuations in liquid water to lengthscale dependent crossover in hydrophobic hydration. Specifically, we employ indirect umbrella sampling (INDUS) simulations to characterize density fluctuations in observation volumes of various sizes and shapes in water and as a function of temperature and salt concentration. Consistent with previous observations, density fluctuations are Gaussian in small molecular scale volumes, but they display non-Gaussian "low-density fat tails" in larger volumes. These non-Gaussian tails are indicative of the proximity of water to its liquid to vapor phase transition and have implications on biomolecular interactions and function. We show that the onset of non-Gaussian fluctuations in large volumes is accompanied by the formation of a cavity in the observation volume. We develop a model that uses the physics of cavity-water interface formation as a key ingredient and show that it captures the nature of non-Gaussian density fluctuations over a broad region in water and in salt solutions. We discuss the limitations of this model in the very low density region of the distribution. Our calculations provide new insights into the origins of non-Gaussian density fluctuations in water and their connections to lengthscale dependent crossover in hydrophobic hydration.

Association of Frailty, Age, and Biological Sex With Severe Acute Respiratory Syndrome Coronavirus 2 Messenger RNA Vaccine-Induced Immunity in Older Adults

BACKGROUND

Male sex and old age are risk factors for severe coronavirus disease 2019, but the intersection of sex and aging on antibody responses to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) vaccines has not been characterized.

METHODS

Plasma samples were collected from older adults (aged 75-98 years) before and after 3 doses of SARS-CoV-2 mRNA vaccination, and from younger adults (aged 18-74 years) post-dose 2, for comparison. Antibody binding to SARS-CoV-2 antigens (spike protein [S], S receptor-binding domain, and nucleocapsid), functional activity against S, and live-virus neutralization were measured against the vaccine virus and the Alpha, Delta, and Omicron variants of concern (VOCs).

RESULTS

Vaccination induced greater antibody titers in older females than in older males, with both age and frailty associated with reduced antibody responses in males but not females. Responses declined significantly in the 6 months after the second dose. The third dose restored functional antibody responses and eliminated disparities caused by sex, age, and frailty in older adults. Responses to the VOCs, particularly the Omicron variant, were significantly reduced relative to the vaccine virus, with older males having lower titers to the VOCs than older females. Older adults had lower responses to the vaccine and VOC viruses than younger adults, with greater disparities in males than in females.

CONCLUSIONS

Older and frail males may be more vulnerable to breakthrough infections owing to low antibody responses before receipt of a third vaccine dose. Promoting third dose coverage in older adults, especially males, is crucial to protecting this vulnerable population.

Evaluation of Lentiviral Vector Stability and Development of Ion Exchange Purification Processes

In this manuscript we employ parallel batch stability and chromatographic screens in concert with linear and step gradient experiments to develop a high yield, HCP clearance anion exchange capture process for lentiviral vector (LVV) purification. An initial broad resin screen is carried out to determine anion exchange-based resins that exhibit high recovery of LVV. LVV stability is then evaluated and conditions are established where the vector exhibits good stability, namely phosphate buffer at pH 6.5-7.5, with low to moderate salt concentrations. A subsequent high-throughput batch screen is then carried out with a subset of resins selected from the first screen under stable conditions to identify optimal wash and elution steps to further improve product yield and protein clearance. Linear gradient experiments are also conducted in mini-column format to refine the operating conditions and final step gradient processes are established that exhibit greater than 70% yield of infectious LVV while also achieving up to 2.89 log reduction values (LRV) of HCPs during the process. The large set of stability and chromatographic data provided in this work represent an important contribution to knowledge in the field about the chromatographic efficacy of a wide range of resins for LVV bioprocessing under stable conditions.

Towards continuous mAb purification: clearance of host cell proteins from CHO cell culture harvests via "flow-through affinity chromatography" using peptide-based adsorbents

The growth of advanced analytics in manufacturing monoclonal antibodies (mAb) has highlighted the challenges associated with the clearance of host cell proteins (HCPs). Of special concern is the removal of "persistent" HCPs, including immunogenic and mAb-degrading proteins, that co-elute from the Protein A resin and can escape the polishing steps. Responding to this challenge, we introduced an ensemble of peptide ligands that target the HCPs in Chinese hamster ovary (CHO) cell culture fluids and enable mAb purification via flow-through affinity chromatography. This work describes their integration into LigaGuardTM, an affinity adsorbent featuring an equilibrium binding capacity of ~30 mg of HCPs per mL of resin as well as dynamic capacities up to 16 and 22 mg/mL at 1- and 2-minute residence times, respectively. When evaluated against cell culture harvests with different mAb and HCP titers and properties, LigaGuardTM afforded high HCP clearance, with logarithmic removal values (LRVs) up to 1.5, and mAb yield above 90%. Proteomic analysis of the effluents confirmed the removal of high-risk HCPs, including cathepsins, histones, glutathione-S transferase, and lipoprotein lipases. Finally, combining LigaGuardTM for HCP removal with affinity adsorbents for product capture afforded a global mAb yield of 85%, and HCP and DNA LRVs > 4. This article is protected by copyright. All rights reserved.

Adaptive immune responses in vaccinated patients with symptomatic SARS-CoV-2 Alpha infection

Benchmarks for protective immunity from infection or severe disease after SARS-CoV-2 vaccination are still being defined. Here, we characterized virus neutralizing and ELISA antibody levels, cellular immune responses, and viral variants in 4 separate groups: healthy controls (HCs) weeks (early) or months (late) following vaccination in comparison with symptomatic patients with SARS-CoV-2 after partial or full mRNA vaccination. During the period of the study, most symptomatic breakthrough infections were caused by the SARS-CoV-2 Alpha variant. Neutralizing antibody levels in the HCs were sustained over time against the vaccine parent virus but decreased against the Alpha variant, whereas IgG titers and T cell responses against the parent virus and Alpha variant declined over time. Both partially and fully vaccinated patients with symptomatic infections had lower virus neutralizing antibody levels against the parent virus than the HCs, similar IgG antibody titers, and similar virus-specific T cell responses measured by IFN-γ. Compared with HCs, neutralization activity against the Alpha variant was lower in the partially vaccinated infected patients and tended to be lower in the fully vaccinated infected patients. In this cohort of breakthrough infections, parent virus neutralization was the superior predictor of breakthrough infections with the Alpha variant of SARS-CoV-2.

Development of QSAR models for in silico screening of antibody solubility

Although monoclonal antibodies (mAbs) have been shown to be extremely effective in treating a number of diseases, they often suffer from poor developability attributes, such as high viscosity and low solubility at elevated concentrations. Since experimental candidate screening is often materials and labor intensive, there is substantial interest in developing in silico tools for expediting mAb design. Here, we present a strategy using machine learning-based QSAR models for the a priori estimation of mAb solubility. The extrapolated protein solubilities of a set of 111 antibodies in a histidine buffer were determined using a high throughput PEG precipitation assay. 3D homology models of the antibodies were determined, and a large set of in house and commercially available molecular descriptors were then calculated. The resulting experimental and descriptor data were then used for the development of QSAR models of mAb solubilities. After feature selection and training with different machine learning algorithms, the models were evaluated with external test sets. The resulting regression models were able to estimate the solubility values of external test set data with R² of 0.81 and 0.85 for the two regression models developed. In addition, three class and binary classification models were developed and shown to be good estimators of mAb solubility behavior, with overall test set accuracies of 0.70 and 0.95, respectively. The analysis of the selected molecular descriptors in these models was also found to be informative and suggested that several charge-based descriptors and isotype may play important roles in mAb solubility. The combination of high throughput relative solubility experimental techniques in concert with efficient machine learning QSAR models offers an opportunity to rapidly screen potential mAb candidates and to design therapeutics with improved solubility characteristics.

Systematic workflow for studying domain contributions of bispecific antibodies to selectivity in multimodal chromatography

In this article, a systematic workflow was formulated and implemented to understand selectivity differences and preferred binding patches for bispecific monoclonal antibodies (mAbs) and their parental mAbs on three multimodal cation exchange resin systems. This workflow incorporates chromatographic screening of the parent mAbs and their fragments at various pH followed by surface property mapping and protein footprinting using covalent labeling followed by liquid chromatography-mass spectrometry analysis. The chromatography screens on multimodal resins with the intact mAbs indicated enhanced selectivity as compared to single-mode interaction systems. While the bispecific antibody (bsAb) eluted between the two parental mAbs on most of the resins, the retention of the bispecific transitioned from co-eluting with one parental mAb to the other parental mAb on Capto MMC. To investigate the contribution of different domains, mAb fragments were evaluated and the results indicated that the interactions were likely dominated by the Fab domain at higher pH. Protein surface property maps were then employed to hypothesize the potential preferred binding patches in the solvent-exposed regions of the parental Fabs. Finally, protein footprinting was carried out with the parental mAbs and the bsAb in the bound and unbound states at pH 7.5 to identify the preferred binding patches. Results with the intact mAb analysis supported the hypothesis that interactions with the resins were primarily driven by the residues in the Fab fragments and not the Fc. Furthermore, peptide mapping data indicated that the light chain may be playing a more important role in the higher binding of Parent A as compared with Parent B in these resin systems. Finally, results with the bsAb indicated that both halves of the molecule contributed to binding with the resins, albeit with subtle differences as compared to the parental mAbs. The workflow presented in this paper lays the foundation to systematically study the chromatographic selectivity of large multidomain molecules which can provide insights into improved biomanufacturability and expedited downstream bioprocess development.

Probing IgG1 FC-Multimodal Nanoparticle Interactions: A Combined Nuclear Magnetic Resonance and Molecular Dynamics Simulations Approach

In this study, NMR and molecular dynamics simulations were employed to study IgG1 F_C binding to multimodal surfaces. Gold nanoparticles functionalized with two multimodal cation-exchange ligands (Capto and Nuvia) were synthesized and employed to carry out solution-phase NMR experiments with the F_C. Experiments with perdeuterated ¹⁵N-labeled F_C and the multimodal surfaces revealed micromolar residue-level binding affinities as compared to millimolar binding affinities with these ligands in free solution, likely due to cooperativity and avidity effects. The binding of F_C with the Capto ligand nanoparticles was concentrated near an aliphatic cluster in the C_H2/C_H3 interface, which corresponded to a focused hydrophobic region. In contrast, binding with the Nuvia ligand nanoparticles was more diffuse and corresponded to a large contiguous positive electrostatic potential region on the side face of the F_C. Results with lower-ligand-density nanoparticles indicated a decrease in binding affinity for both systems. For the Capto ligand system, several aliphatic residues on the F_C that were important for binding to the higher-density surface did not interact with the lower-density nanoparticles. In contrast, no significant difference was observed in the interacting residues on the F_C to the high- and low-ligand density Nuvia surfaces. The binding affinities of F_C to both multimodal-functionalized nanoparticles decreased in the presence of salt due to the screening of multiple weak interactions of polar and positively charged residues. For the Capto ligand nanoparticle system, this resulted in an even more focused hydrophobic binding region in the interface of the C_H2 and C_H3 domains. Interestingly, for the Nuvia ligand nanoparticles, the presence of salt resulted in a large transition from a diffuse binding region to the same focused binding region determined for Capto nanoparticles at 150 mM salt. Molecular dynamics simulations corroborated the NMR results and provided important insights into the molecular basis of F_C binding to these different multimodal systems containing clustered (observed at high-ligand densities) and nonclustered ligand surfaces. This combined biophysical and simulation approach provided significant insights into the interactions of F_C with multimodal surfaces and sets the stage for future analyses with even more complex biotherapeutics.

Advancing a rapid, high throughput screening platform for optimization of lentivirus production

BACKGROUND

Lentiviral vectors (LVVs) hold great promise as delivery tools for gene therapy and chimeric antigen receptor T cell (CAR-T) therapy. Their ability to target difficult to transfect cells and deliver genetic payloads that integrate into the host genome makes them ideal delivery candidates. However, several challenges remain to be addressed before LVVs are more widely used as therapeutics including low viral vector concentrations and the absence of suitable scale-up methods for large-scale production. To address these challenges, we have developed a high throughput microscale HEK293 suspension culture platform that enables rapid screening of conditions for improving LVV productivity.

KEY RESULTS

High density culture (40 million cells mL^-1 ) of HEK293 suspension cells in commercially available media was achieved in microscale 96-deep well plate platform at liquid volumes of 200 μL. Comparable transfection and LVV production efficiencies were observed at the microscale, in conventional shake flasks and a 1-L bioreactor, indicating that significant scale-down does not affect LVV concentrations and predictivity of scale-up. Optimization of production step allowed for final yields of LVVs to reach 1.5 × 10⁷  TU mL^-1 .

CONCLUSIONS

The ability to test a large number of conditions simultaneously with minimal reagent use allows for the rapid optimization of LVV production in HEK293 suspension cells. Therefore, such a system may serve as a valuable tool in early stage process development and can be used as a screening tool to improve LVV concentrations for both batch and perfusion based systems.

Evaluation of guanidine-based multimodal anion exchangers for protein selectivity and orthogonality

In this paper, we examined the chromatographic behavior of a new class of guanidine-based multimodal anion exchange resins. The selectivities and protein recoveries on these resins were first evaluated using linear gradient chromatography with a model acidic protein library at pH 5, 6 and 7. While a single-guanidine based resin exhibited significant recovery issues at high ligand density, a bis-guanidine based resin showed high recoveries of all but two of the proteins evaluated in the study. In addition, the bis-guanidine resin showed a more pH dependent selectivity pattern as compared to the low density single-guanidine resin. The salt elution range for the low density single-guanidine and bis-guanidine resins was also observed to vary from 0.250 to 0.621 M and 0.162 to 0.828 M NaCl, respectively. A QSAR model was then developed to predict the elution behavior of these proteins on the guanidine prototypes at multiple pH with overall training and test scores of 0.88 and 0.85, respectively. In addition, molecular dynamics simulations were performed with these ligands immobilized on a self-assembled monolayer (SAM) to characterize their conformational preferences and to gain insight into the molecular basis of their chromatographic behavior. Finally, a recently developed framework was employed to evaluate the separability of the bis-guanidine resin as well as its orthogonality to the multimodal cation exchanger, Nuvia cPrime. This evaluation was carried out using a second model protein library which included both acidic and basic proteins. The results of this analysis indicated that the bis-guanidine prototype exhibited both higher pair separability (0.73) and pair enhancement (0.42) as compared to the less hydrophobic commercial Nuvia aPrime 4A with pair separability and enhancement factors of 0.57 and 0.22, respectively. The enhanced selectivity and orthogonality of this new multimodal anion exchange ligand may offer potential opportunities for bioprocessing applications.

Behavior of Water Near Multimodal Chromatography Ligands and Its Consequences for Modulating Protein-Ligand Interactions

Multimodal chromatography is a powerful approach for purifying proteins that uses ligands containing multiple modes of interaction. Recent studies have shown that selectivity in multimodal chromatographic separations is a function of the ligand structure and geometry. Here, we performed molecular dynamics simulations to explore how the ligand structure and geometry affect ligand-water interactions and how these differences in solution affect the nature of protein-ligand interactions. Our investigation focused on three chromatography ligands: Capto MMC, Nuvia cPrime, and Prototype 4, a structural variant of Nuvia cPrime. First, the solvation characteristics of each ligand were quantified via three metrics: average water density, fluctuations, and residence time. We then explored how solvation was perturbed when the ligand was bound to the protein surface and found that the probability of the phenyl ring dewetting followed the order: Capto MMC > Prototype 4 > Nuvia cPrime. To explore how these differences in dewetting affect protein-ligand interactions, we calculated the probability of each ligand binding to different types of residues on the protein surface and found that the probability of binding to a hydrophobic residue followed the same order as the dewetting behavior. This study illustrates the role that wetting and dewetting play in modulating protein-ligand interactions.

Rapid optimization of processes for the integrated purification of biopharmaceuticals

Straight‐through chromatography, wherein the eluate from one column passes directly onto another column without adjustment, is one strategy to integrate and intensify manufacturing processes for biologics. Development and optimization of such straight‐through chromatographic processes is a challenge, however. Conventional high‐throughput screening methods optimize each chromatographic step independently, with limited consideration for the connectivity of steps. Here, we demonstrate a method for the development and optimization of fully integrated, multi‐column processes for straight‐through purification. Selection of resins was performed using an in silico tool for the prediction of processes for straight‐through purification based on a one‐time characterization of host‐cell proteins combined with the chromatographic behavior of the product. A two‐step optimization was then conducted to determine the buffer conditions that maximized yield while minimizing process‐ and product‐related impurities. This optimization of buffer conditions included a series of range‐finding experiments on each individual column, similar to conventional screening, followed by the development of a statistical model for the fully integrated, multi‐column process using design of experiments. We used this methodology to develop and optimize integrated purification processes for a single‐domain antibody and a cytokine, obtaining yields of 88% and 86%, respectively, with process‐ and product‐related variants reduced to phase‐appropriate levels for nonclinical material.

Development of a platform process for the production and purification of single-domain antibodies

Single‐domain antibodies (sdAbs) offer the affinity and therapeutic value of conventional antibodies, with increased stability and solubility. Unlike conventional antibodies, however, sdAbs do not benefit from a platform manufacturing process. While successful production of a variety of sdAbs has been shown in numerous hosts, purification methods are often molecule specific or require affinity tags, which generally cannot be used in clinical manufacturing due to regulatory concerns. Here, we have developed a broadly applicable production and purification process for sdAbs in Komagataella phaffii (Pichia pastoris) and demonstrated the production of eight different sdAbs at a quality appropriate for nonclinical studies. We developed a two‐step, integrated purification process without the use of affinity resins and showed that modification of a single process parameter, pH of the bridging buffer, was required for the successful purification of a variety of sdAbs. Further, we determined that this parameter can be predicted based only on the biophysical characteristics of the target molecule. Using these methods, we produced nonclinical quality sdAbs as few as 5 weeks after identifying the product sequence. Nonclinical studies of three different sdAbs showed that molecules produced using our platform process conferred protection against viral shedding of rotavirus or H1N1 influenza and were equivalent to similar molecules produced in Escherichia coli and purified using affinity tags.

Identification of preferred multimodal ligand-binding regions on IgG1 FC using nuclear magnetic resonance and molecular dynamics simulations

In this study, the binding of multimodal chromatographic ligands to the IgG1 F_C domain were studied using nuclear magnetic resonance and molecular dynamics simulations. Nuclear magnetic resonance experiments carried out with chromatographic ligands and a perdeuterated ¹⁵ N-labeled F_C domain indicated that while single-mode ion exchange ligands interacted very weakly throughout the F_C surface, multimodal ligands containing negatively charged and aromatic moieties interacted with specific clusters of residues with relatively high affinity, forming distinct binding regions on the F_C . The multimodal ligand-binding sites on the F_C were concentrated in the hinge region and near the interface of the C_H 2 and C_H 3 domains. Furthermore, the multimodal binding sites were primarily composed of positively charged, polar, and aliphatic residues in these regions, with histidine residues exhibiting some of the strongest binding affinities with the multimodal ligand. Interestingly, comparison of protein surface property data with ligand interaction sites indicated that the patch analysis on F_C corroborated molecular-level binding information obtained from the nuclear magnetic resonance experiments. Finally, molecular dynamics simulation results were shown to be qualitatively consistent with the nuclear magnetic resonance results and to provide further insights into the binding mechanisms. An important contribution to multimodal ligand-F_C binding in these preferred regions was shown to be electrostatic interactions and π-π stacking of surface-exposed histidines with the ligands. This combined biophysical and simulation approach has provided a deeper molecular-level understanding of multimodal ligand-F_C interactions and sets the stage for future analyses of even more complex biotherapeutics.

The Role of Ligand-Ligand Interactions in Multimodal Ligand Conformational Equilibria and Surface Pattern Formation

Multimodal chromatography uses multiple modes of interaction such as charge, hydrophobic, or hydrogen bonding to separate proteins. Recently, we used molecular dynamics (MD) simulations to show that ligands immobilized on surfaces can interact and associate with neighboring ligands to form hydrophobic and charge patches, which may have important implications for the nature of protein-surface interactions. Here, we study interfacial systems of increasing complexity-from a single immobilized multimodal ligand to high density surfaces-to better understand how ligand behavior is affected by the presence of a surface and the presence of other ligands in the vicinity, and how this behavior scales to larger systems. We find that tethering a ligand to a surface restricts its conformations to a subset of those observed in free solution, yet the ligand maintains flexibility in the plane of the surface and can form contacts with neighboring ligands. We find that although the formation of a contact between two neighboring ligands is slightly unfavorable, three neighboring ligands exhibit a preference for the formation of a fully connected cluster. To explore how these trends in ligand association extend to a larger surface with high density of ligands, we performed coarse-grained Monte Carlo (MC) simulations of a 132-ligand surface using ligand interactions parametrized based on free energies obtained from the three-ligand MD simulations. Despite their simplicity, the coarse-grained simulations qualitatively capture the cluster size distribution of ligands observed in detailed MD simulations. Quantitative differences between the two suggest opportunities for improvements in the coarse-grained energy function for efficient predictions of cluster and pattern formations. Our approach presents a promising route to the engineering of multimodal patterns for future chromatographic resin design.

Mechanism of sequence-based separation of single-stranded DNA in capillary zone electrophoresis

Separation of DNA by length using CGE is a mature field. Separation of DNA by sequence, in contrast, is a more difficult problem. Existing techniques generally rely upon changes in intrinsic or induced differences in conformation. Previous work in our group showed that sets of ssDNA of the same length differing in sequence by as little as a single base could be separated by CZE using simple buffers at high ionic strength. Here, we explore the basis of the separation using circular dichroism spectroscopy, fluorescence anisotropy, and small angle X-ray scattering. The results reveal sequence-dependent differences among the same length strands, but the trends in the differences are not correlated to the migration order of the strands in the CZE separation. They also indicate that the separation is based on intrinsic differences among the strands that do not change with increasing ionic strength; rather, increasing ionic strength has a greater effect on electroosmotic mobility in the normal direction than on electrophoretic mobility of the strands in the reverse direction. This increases the migration time of the strands in the normal direction, allowing more time for the same-length strands to be teased apart based on very small differences in the intrinsic properties of the strands of different sequence. Regression analysis was used to model the intrinsic differences among DNA strands in order to gain insight into the relationship between mobility and sequence that underlies the separation.

Formation of Ligand Clusters on Multimodal Chromatographic Surfaces

Multimodal chromatography is a powerful tool which uses multiple modes of interaction, such as charge and hydrophobicity, to purify protein-based therapeutics. In this work, we performed molecular dynamics simulations of a series of multimodal cation-exchange ligands immobilized on a hydrophilic self-assembled monolayer surface at the commercially relevant surface density (1 ligand/nm²). We found that ligands that were flexible and terminated in a hydrophobic group had a propensity to aggregate on the surface, while less flexible ligands containing a hydrophobic group closer to the surface did not aggregate. For aggregating ligands, this resulted in the formation of a surface pattern that contained relatively large patches of hydrophobicity and charge whose sizes exceeded the length scale of the individual ligands. On the other hand, lowering the surface density to 1 ligand/3 nm² reduced or eliminated this aggregation behavior. In addition, the introduction of a flexible linker (corresponding to the commercially available ligand) enhanced cluster formation and allowed aggregation to occur at lower surface densities. Further, the use of flexible linkers enabled hydrophobic groups to collapse to the surface, reducing their accessibility. Finally, we developed an approach for quantifying differences in the observed surface patterns by calculating distributions of the patch size and patch length. This clustering phenomenon is likely to play a key role in governing protein-surface interactions in multimodal chromatography. This new understanding of multimodal surfaces has important implications for developing improved predictive models and designing new classes of multimodal separation materials.

Implementation of an experimental and computational tool set to study protein-mAb interactions

This work focused on the development of a combined experimental and computational tool set to study protein-mAb interactions. A model protein library was first screened using cross interaction chromatography to identify proteins with the strongest retention. Fluorescence polarization was then employed to study the interactions and thermodynamics of the selected proteins-lactoferrin, pyruvate kinase, and ribonuclease B with the mAb. Binding affinities of lactoferrin and pyruvate kinase to the mAb were seen to be relatively salt insensitive in the range examined. Further, a strong entropic contribution was observed, suggesting the importance of hydrophobic interactions. On the other hand, ribonuclease B-mAb binding was seen to be enthalpically driven and salt sensitive, indicating the importance of electrostatic interactions. Protein-protein docking was then carried out and the results identified the CDR region on the mAb as an important binding site for all three proteins. The binding interfaces identified for the mAb-lactoferrin and mAb-pyruvate kinase systems were found to contain complementary hydrophobic and oppositely charged clusters on the interacting regions which were indicative of both hydrophobic and electrostatic interactions. On the other hand, the binding site on ribonuclease B was predominantly positively charged with minimal hydrophobicity. This resulted in an alignment with negatively charged clusters on the mAb, supporting the contention that these interactions were primarily electrostatic in nature. Importantly, these computational results were found to be consistent with the fluorescence polarization data and this combined approach may have utility in examining mAb-HCP interactions which can often complicate the downstream processing of biologics. © 2019 American Institute of Chemical Engineers.

On-demand manufacturing of clinical-quality biopharmaceuticals

Conventional manufacturing of protein biopharmaceuticals in centralized, large-scale, single-product facilities is not well-suited to the agile production of drugs for small patient populations or individuals. Previous solutions for small-scale manufacturing are limited in both process reproducibility and product quality, owing to their complicated means of protein expression and purification. We describe an automated, benchtop, multiproduct manufacturing system, called Integrated Scalable Cyto-Technology (InSCyT), for the end-to-end production of hundreds to thousands of doses of clinical-quality protein biologics in about 3 d. Unlike previous systems, InSCyT includes fully integrated modules for sustained production, efficient purification without the use of affinity tags, and formulation to a final dosage form of recombinant biopharmaceuticals. We demonstrate that InSCyT can accelerate process development from sequence to purified drug in 12 weeks. We used integrated design to produce human growth hormone, interferon α-2b and granulocyte colony-stimulating factor with highly similar processes on this system and show that their purity and potency are comparable to those of marketed reference products.

Single Molecule Force Spectroscopy and Molecular Dynamics Simulations as a Combined Platform for Probing Protein Face-Specific Binding

Biomolecular interactions frequently occur in orientation-specific manner. For example, prior nuclear magnetic resonance spectroscopy experiments in our lab have suggested the presence of a group of strongly binding residues on a particular face of the protein ubiquitin for interactions with Capto MMC multimodal ligands ("Capto" ligands) (Srinivasan, K.; et al. Langmuir 2014, 30 (44), 13205-13216). We present a clear confirmation of those studies by performing single-molecule force spectroscopy (SMFS) measurements of unbinding complemented with molecular dynamics (MD) calculations of the adsorption free energy of ubiquitin in two distinct orientations with self-assembled monolayers (SAMs) functionalized with "Capto" ligands. These orientations were maintained in the SMFS experiments by tethering ubiquitin mutants to SAM surfaces through strategically located cysteines, thus exposing the desired faces of the protein. Analogous orientations were maintained in MD simulations using suitable constraining methods. Remarkably, despite differences between the finer details of experimental and simulation methodologies, they confirm a clear preference for the previously hypothesized binding face of ubiquitin. Furthermore, MD simulations provided significant insights into the mechanism of protein binding onto this multimodal surface. Because SMFS and MD simulations both directly probe protein-surface interactions, this work establishes a key link between experiments and simulations at molecular scale through the determination of protein face-specific binding energetics. Our approach may have direct applications in biophysical systems where face- or orientation-specific interactions are important, such as biomaterials, sensors, and biomanufacturing.

Interactions of Multimodal Ligands with Proteins: Insights into Selectivity Using Molecular Dynamics Simulations

Fundamental understanding of protein-ligand interactions is important to the development of efficient bioseparations in multimodal chromatography. Here we employ molecular dynamics (MD) simulations to investigate the interactions of three different proteins--ubiquitin, cytochrome C, and α-chymotrypsinogen A, sampling a range of charge from +1e to +9e--with two multimodal chromatographic ligands containing similar chemical moieties--aromatic, carboxyl, and amide--in different structural arrangements. We use a spherical harmonic expansion to analyze ligand and individual moiety density profiles around the proteins. We find that the Capto MMC ligand, which contains an additional aliphatic group, displays stronger interactions than Nuvia CPrime ligand with all three proteins. Studying the ligand densities at the moiety level suggests that hydrophobic interactions play a major role in determining the locations of high ligand densities. Finally, the greater structural flexibility of the Capto MMC ligand compared to that of the Nuvia cPrime ligand allows for stronger structural complementarity and enables stronger hydrophobic interactions. These subtle and not-so-subtle differences in binding affinities and modalities for multimodal ligands can result in significantly different binding behavior towards proteins with important implications for bioprocessing.

Investigation into the molecular and thermodynamic basis of protein interactions in multimodal chromatography using functionalized nanoparticles

Although multimodal chromatography offers significant potential for bioseparations, there is a lack of molecular level understanding of the nature of protein binding in these systems. In this study a nanoparticle system is employed that can simulate a chromatographic resin surface while also being amenable to isothermal titration calorimetry (ITC) and solution NMR. ITC and NMR titration experiments are carried out with (15)N-labeled ubiquitin to investigate the interactions of ubiquitin with nanoparticles functionalized with two industrially important multimodal ligands. The ITC results suggest that binding to both multimodal ligand surfaces is entropically driven over a range of temperatures and that this is due primarily to the release of surface bound waters. In order to reveal structural details of the interaction process, binding-induced chemical shift changes obtained from the NMR experiments are employed to obtain dissociation constants of individual amino acid residues on the protein surface. The residue level information obtained from NMR is then used to identify a preferred binding face on ubiquitin for interaction to both multimodal ligand surfaces. In addition, electrostatic potential and spatial aggregation propensity maps are used to determine important protein surface property data that are shown to correlate well with the molecular level information obtained from NMR. Importantly, the data demonstrate that the cluster of interacting residues on the protein surface act co-operatively to give rise to multimodal binding affinities several orders of magnitude greater than those obtained previously for interactions with free solution ligands. The use of NMR and ITC to study protein interactions with functionalized nanoparticles offers a new tool for obtaining important molecular and thermodynamic insights into protein affinity in multimodal chromatographic systems.