Machine Learning Applied to Determine the Molecular Descriptors Responsible for the Viscosity Behavior of Concentrated Therapeutic Antibodies.

Predicting the solution viscosity of monoclonal antibody (mAb) drug products remains as one of the main challenges in antibody drug design, manufacturing, and delivery. In this work, the concentration-dependent solution viscosity of 27 FDA-approved mAbs was measured at pH 6.0 in 10 mM histidine-HCl. Six mAbs exhibited high viscosity (>30 cP) in solutions at 150 mg/mL mAb concentration. Combining molecular modeling and machine learning feature selection, we found that the net charge in the mAbs and the amino acid composition in the Fv region are key features which govern the viscosity behavior. For mAbs whose behavior was not dominated by charge effects, we observed that high viscosity is correlated with more hydrophilic and fewer hydrophobic residues in the Fv region. A predictive model based on the net charges of mAbs and a high viscosity index is presented as a fast screening tool for classifying low- and high-viscosity mAbs.

Machine Learning Feature Selection for Predicting High Concentration Therapeutic Antibody Aggregation.

Protein aggregation can hinder the development, safety and efficacy of therapeutic antibody-based drugs. Developing a predictive model that evaluates aggregation behaviors during early stage development is therefore desirable. Machine learning is a widely used tool to train models that predict data with different attributes. However, most machine learning techniques require more data than is typically available in antibody development. In this work, we describe a rational feature selection framework to develop accurate models with a small number of features. We applied this framework to predict aggregation behaviors of 21 approved monospecific monoclonal antibodies at high concentration (150 mg/mL), yielding a correlation coefficient of 0.71 on validation tests with only two features using a linear model. The nearest neighbors and support vector regression models further improved the performance, which have correlation coefficients of 0.86 and 0.80, respectively. This framework can be extended to train other models that predict different physical properties.

Molecular computations of preferential interactions of proline, arginine.HCl, and NaCl with IgG1 antibodies and their impact on aggregation and viscosity.

Preferential interactions of excipients with the antibody surface govern their effect on the stability of antibodies in solution. We probed the preferential interactions of proline, arginine.HCl (Arg.HCl), and NaCl with three therapeutically relevant IgG1 antibodies via experiment and simulation. With simulations, we examined how excipients interacted with different types of surface patches in the variable region (Fv). For example, proline interacted most strongly with aromatic surfaces, Arg.HCl was included near negative residues, and NaCl was excluded from negative residues and certain hydrophobic regions. The differences in interaction of different excipients with the same surface patch on an antibody may be responsible for variations in the antibody's aggregation, viscosity, and self-association behaviors in each excipient. Proline reduced self-association for all three antibodies and reduced aggregation for the antibody with an association-limited aggregation mechanism. The effects of Arg.HCl and NaCl on aggregation and viscosity were highly dependent on the surface charge distribution and the extent of exclusion from highly hydrophobic patches. At pH 5.5, both tended to increase the aggregation of an antibody with a strongly positive charge on the Fv, while only NaCl reduced the aggregation of the antibody with a large negative charge patch on the Fv. Arg.HCl reduced the viscosities of antibodies with either a hydrophobicity-driven mechanism or a charge-driven mechanism. Analysis of this data presents a framework for understanding how amino acid and ionic excipients interact with different protein surfaces, and how these interactions translate to the observed stability behavior.

Machine Learning Models of Antibody-Excipient Preferential Interactions for Use in Computational Formulation Design.

Preferential interactions of formulation excipients govern their impact on the stability properties of proteins in solution. The ability to predict these interactions without the need to perform experiments would enable formulation design to begin early in the development of a new antibody therapeutic. With that in mind, we developed a feature set to numerically describe local regions of an antibody's surface for use in machine learning applications. Then, we used these features to train machine learning models for local antibody-excipient preferential interactions for the excipients sorbitol, sucrose, trehalose, proline, arginine·HCl, and NaCl. Our models had accuracies of up to about 85%. We also used linear (elastic net) models to quantify the contribution of antibody surface features to the preferential interaction coefficients, finding that the carbohydrates and proline tend to have similar important features, while the interactions of arginine·HCl and NaCl are governed by charge features. We present several case studies demonstrating how these machine learning models could be used to predict experimental aggregation and viscosity behavior in solution. Finally, we propose an approach to computational formulation design wherein a panel of excipients may be considered while designing an antibody sequence.

Molecular Computations of Preferential Interaction Coefficients of IgG1 Monoclonal Antibodies with Sorbitol, Sucrose, and Trehalose and the Impact of These Excipients on Aggregation and Viscosity.

Preferential interactions of formulation excipients govern their overall interactions with protein molecules, and molecular dynamics simulations allow for the examination of the interactions at the molecular level. We used molecular dynamics simulations to examine the interactions of sorbitol, sucrose, and trehalose with three different IgG1 antibodies to gain insight into how these excipients impact aggregation and viscosity. We found that sucrose and trehalose reduce aggregation more than sorbitol because of their larger size and their stronger interactions with high-spatial aggregation propensity residues compared to sorbitol. Two of the antibodies had high viscosity in sodium acetate buffer, and for these, we found that sucrose and trehalose tended to have opposite effects on viscosity. The data presented here provide further insight into the mechanisms of interactions of these three carbohydrate excipients with the antibody surface and thus their impact on excipient stabilization of antibody formulations.

Understanding the Role of Preferential Exclusion of Sugars and Polyols from Native State IgG1 Monoclonal Antibodies and its Effect on Aggregation and Reversible Self-Association.

PURPOSE: To investigate differences in the preferential exclusion of trehalose, sucrose, sorbitol and mannitol from the surface of three IgG1 monoclonal antibodies (mAbs) and understand its effect on the aggregation and reversible self-association of mAbs at high-concentrations. METHODS: Preferential exclusion was measured using vapor pressure osmometry. Effect of excipient addition on accelerated aggregation kinetics was quantified using size exclusion chromatography and on reversible self-association was quantified using dynamic light scattering. RESULTS: The doubling of excipient concentration in the 0 to 0.5 m range resulted in a doubling of the mAb transfer free energy for all excipients and antibodies tested in this study. Solution pH and choice of buffering agent did not significantly affect the magnitude of preferential exclusion. We find that aggregation suppression for trehalose, sucrose and sorbitol (but not mannitol) correlates with the magnitude of their preferential exclusion from the native state of the three IgG1 mAbs. We also find that addition of sugars and polyols reduced the tendency for reversible self-association in two mAbs that had weakly repulsive or neutral self-interactions in the presence of buffer alone. CONCLUSIONS: The magnitude of preferential exclusion for trehalose, sucrose and sorbitol correlates well with their partial molar volumes in solution. Mannitol is excluded to a greater extent than that expected from its partial molar volume, suggesting specific interactions of mannitol that might be different than the other sugars and polyols tested in this study. Local interactions play a role in the effect of excipient addition on the reversible self-association of mAbs. These results provide further insights into the stabilization of high-concentration mAb formulations by sugars and polyols.

Kirkwood-Buff-Derived Alcohol Parameters for Aqueous Carbohydrates and Their Application to Preferential Interaction Coefficient Calculations of Proteins.

The CHARMM36 carbohydrate parameter set did not adequately reproduce experimental thermodynamic data of carbohydrate interactions with water or proteins or carbohydrate self-association; thus, a new nonbonded parameter set for carbohydrates was developed. The parameters were developed to reproduce experimental Kirkwood-Buff integral values, defined by the Kirkwood-Buff theory of solutions, and applied to simulations of glycerol, sorbitol, glucose, sucrose, and trehalose. Compared to the CHARMM36 carbohydrate parameters, these new Kirkwood-Buff-based parameters reproduced accurately carbohydrate self-association and the trend of activity coefficient derivative changes with concentration. When using these parameters, preferential interaction coefficients calculated from simulations of these carbohydrates and the proteins lysozyme, bovine serum albumin, α-chymotrypsinogen A, and RNase A agreed well with the experimental data, whereas use of the CHARMM36 parameters indicated preferential inclusion of carbohydrates, in disagreement with the experiment. Thus, calculating preferential interaction coefficients from simulations requires using a force field that accurately reproduces trends in the thermodynamic properties of binary excipient-water solutions, and in particular the trend in the activity coefficient derivative. Finally, the carbohydrate-protein simulations using the new parameters indicated that the carbohydrate size was a major factor in the distribution of different carbohydrates around a protein surface.

Preferential interactions of trehalose, L-arginine.HCl and sodium chloride with therapeutically relevant IgG1 monoclonal antibodies.

Preferential interactions of weakly interacting formulation excipients govern their effect on the equilibrium and kinetics of several reactions of protein molecules in solution. Using vapor pressure osmometry, we characterized the preferential interactions of commonly used excipients trehalose, L-arginine.HCl and NaCl with three therapeutically-relevant, IgG1 monoclonal antibodies that have similar size and shape, but differ in their surface hydrophobicity and net charge. We further characterized the effect of these excipients on the reversible self-association, aggregation and viscosity behavior of these antibody molecules. We report that trehalose, L-arginine.HCl and NaCl are all excluded from the surface of the three IgG1 monoclonal antibodies, and that the exclusion behavior is linearly related to the excipient molality in the case of trehalose and NaCl, whereas a non-linear behavior is observed for L-arginine.HCl. Interestingly, we find that the magnitude of trehalose exclusion depends upon the nature of the protein surface. Such behavior is not observed in case of NaCl and L-arginine.HCl as they are excluded to the same extent from the surface of all three antibody molecules tested in this study. Analysis of data presented in this study provides further insight into the mechanisms governing excipient-mediated stabilization of mAb formulations.