Design of self-interaction chromatography as an analytical tool for predicting protein phase behavior

Utility of physiologically based pharmacokinetic modeling to predict inter-antibody variability in monoclonal antibody pharmacokinetics in mice

In this investigation, we tested the hypothesis that a physiologically based pharmacokinetic (PBPK) model incorporating measured in vitro metrics of off-target binding can largely explain the inter-antibody variability in monoclonal antibody (mAb) pharmacokinetics (PK). A diverse panel of 83 mAbs was evaluated for PK in wild-type mice and subjected to 10 in vitro assays to measure major physiochemical attributes. After excluding for target-mediated elimination and immunogenicity, 56 of the remaining mAbs with an eight-fold variability in the area under the curve (A U C 0 - 672 h : 1.74 × 106 -1.38 × 107 ng∙h/mL) and 10-fold difference in clearance (2.55-26.4 mL/day/kg) formed the training set for this investigation. Using a PBPK framework, mAb-dependent coefficients F1 and F2 modulating pinocytosis rate and convective transport, respectively, were estimated for each mAb with mostly good precision (coefficient of variation (CV%) <30%). F1 was estimated to be the mean and standard deviation of 0.961 ± 0.593, and F2 was estimated to be 2.13 ± 2.62. Using principal component analysis to correlate the regressed values of F1/F2 versus the multidimensional dataset composed of our panel of in vitro assays, we found that heparin chromatography retention time emerged as the predictive covariate to the mAb-specific F1, whereas F2 variability cannot be well explained by these assays. A sigmoidal relationship between F1 and the identified covariate was incorporated within the PBPK framework. A sensitivity analysis suggested plasma concentrations to be most sensitive to F1 when F1 > 1. The predictive utility of the developed PBPK model was evaluated against a separate panel of 14 mAbs biased toward high clearance, among which area under the curve of PK data of 12 mAbs was predicted within 2.5-fold error, and the positive and negative predictive values for clearance prediction were 85% and 100%, respectively. MAb heparin chromatography assay output allowed a priori identification of mAb candidates with unfavorable PK.

Chromatographic parameter determination for complex biological feedstocks

The application of mechanistic models for chromatography requires accurate model parameters. Especially for complex feedstocks such as a clarified cell harvest, this can still be an obstacle limiting the use of mechanistic models. Another commonly encountered obstacle is a limited amount of sample material and time to determine all needed parameters. Therefore, this study aimed at implementing an approach on a robotic liquid handling system that starts directly with a complex feedstock containing a monoclonal antibody. The approach was tested by comparing independent experimental data sets with predictions generated by the mechanistic model using all parameters determined in this study. An excellent agreement between prediction and experimental data was found verifying the approach. Thus, it can be concluded that RoboColumns with a bed volume of 200 μL can well be used to determine isotherm parameters for predictions of larger scale columns. Overall, this approach offers a new way to determine crucial model input parameters for mechanistic modelling of chromatography for complex biological feedstocks. © 2018 The Authors Biotechnology Progress published by Wiley Periodicals, Inc. on behalf of American Institute of Chemical Engineers Biotechnol. Prog., 34:1006–1018, 2018

From osmotic second virial coefficient (B22 ) to phase behavior of a monoclonal antibody

Antibodies are complex macromolecules and their phase behavior as well as interactions within different solvents and precipitants are still not understood. To shed some light into the processes on a molecular dimension, the occurring self-interactions between antibody molecules were analyzed by means of the osmotic second virial coefficient (B22 ). The determined B22 follows qualitatively the phenomenological Hofmeister series describing the aggregation probability of antibodies for the various solvent compositions. However, a direct correlation between crystallization probability and B22 in form of a crystallization slot does not seem to be feasible for antibodies since the phase behavior is strongly dependent on their anisotropy. Kinetic parameters have to be taken into account due to the molecular size and complexity of the molecules. This is confirmed by a comparison of experimental data with a theoretical phase diagram. On the other hand the solubility is thermodynamically driven and therefore the B22 could be used to establish a universal solubility line for the monoclonal antibody mAb04c and different solvent compositions by using thermodynamic models.

A comparative study of monoclonal antibodies. 1. Phase behavior and protein-protein interactions

Protein phase behavior is involved in numerous aspects of downstream processing, either by design as in crystallization or precipitation processes, or as an undesired effect, such as aggregation. This work explores the phase behavior of eight monoclonal antibodies (mAbs) that exhibit liquid-liquid separation, aggregation, gelation, and crystallization. The phase behavior has been studied systematically as a function of a number of factors, including solution composition and pH, in order to explore the degree of variability among different antibodies. Comparisons of the locations of phase boundaries show consistent trends as a function of solution composition; however, changing the solution pH has different effects on each of the antibodies studied. Furthermore, the types of dense phases formed varied among the antibodies. Protein-protein interactions, as reflected by values of the osmotic second virial coefficient, are used to correlate the phase behavior. The primary findings are that values of the osmotic second virial coefficient are useful for correlating phase boundary locations, though there is appreciable variability among the antibodies in the apparent strengths of the intrinsic protein-protein attraction manifested. However, the osmotic second virial coefficient does not provide a clear basis to predict the type of dense phase likely to result under a given set of solution conditions.

Protein solubilization: a novel approach

Formulation development presents significant challenges with respect to protein therapeutics. One component of these challenges is to attain high protein solubility (>50mg/ml for immunoglobulins) with minimal aggregation. Protein-protein interactions contribute to aggregation and the integral sum of these interactions can be quantified by a thermodynamic parameter known as the osmotic second virial coefficient (B-value). The method presented here utilizes high-throughput measurement of B-values to identify the influence of additives on protein-protein interactions. The experiment design uses three tiers of screens to arrive at final solution conditions that improve protein solubility. The first screen identifies individual additives that reduce protein interactions. A second set of B-values are then measured for different combinations of these additives via an incomplete factorial screen. Results from the incomplete factorial screen are used to train an artificial neural network (ANN). The "trained" ANN enables predictions of B-values for more than 4000 formulations that include additive combinations not previously experimentally measured. Validation steps are incorporated throughout the screening process to ensure that (1) the protein's thermal and aggregation stability characteristics are not reduced and (2) the artificial neural network predictive model is accurate. The ability of this approach to reduce aggregation and increase solubility is demonstrated using an IgG protein supplied by Minerva Biotechnologies, Inc.

Moving through three-dimensional phase diagrams of monoclonal antibodies

Protein phase behavior characterization is a multivariate problem due to the high amount of influencing parameters and the diversity of the proteins. Single influences on the protein are not understood and fundamental knowledge remains to be obtained. For this purpose, a systematic screening method was developed to characterize the influence of fluid phase conditions on the phase behavior of proteins in three-dimensional phase diagrams. This approach was applied to three monoclonal antibodies to investigate influences of pH, protein and salt concentrations, with five different salts being tested. Although differences exist between the antibodies, this extensive study confirmed the general applicability of the Hofmeister series over the broad parameter range analyzed. The influence of the different salts on the aggregation (crystallization and precipitation) probability was described qualitatively using this Hofmeister series, with a differentiation between crystallization and precipitation being impossible, however.

High-throughput screening for developability during early-stage antibody discovery using self-interaction nanoparticle spectroscopy

The discovery of monoclonal antibodies (mAbs) that bind to a particular molecular target is now regarded a routine exercise. However, the successful development of mAbs that (1) express well, (2) elicit a desirable biological effect upon binding, and (3) remain soluble and display low viscosity at high concentrations is often far more challenging. Therefore, high throughput screening assays that assess self-association and aggregation early in the selection process are likely to yield mAbs with superior biophysical properties. Here, we report an improved version of affinity-capture self-interaction nanoparticle spectroscopy (AC-SINS) that is capable of screening large panels of antibodies for their propensity to self-associate. AC-SINS is based on concentrating mAbs from dilute solutions around gold nanoparticles pre-coated with polyclonal capture (e.g., anti-Fc) antibodies. Interactions between immobilized mAbs lead to reduced inter-particle distances and increased plasmon wavelengths (wavelengths of maximum absorbance), which can be readily measured by optical means. This method is attractive because it is compatible with dilute and unpurified mAb solutions that are typical during early antibody discovery. In addition, we have improved multiple aspects of this assay for increased throughput and reproducibility. A data set comprising over 400 mAbs suggests that our modified assay yields self-interaction measurements that are well-correlated with other lower throughput assays such as cross-interaction chromatography. We expect that the simplicity and throughput of our improved AC-SINS method will lead to improved selection of mAbs with excellent biophysical properties during early antibody discovery.

Improved method for evaluating the dead volume and protein-protein interactions by self-interaction chromatography

Self-interaction chromatography (SIC) is a well-established method for studying protein-protein interactions. The second virial coefficient in SIC is evaluated directly from the measured retention coefficient for the protein using a column packed with resin on which the same protein has been immobilized on the pore surface. One of the challenges in determining the retention coefficient is the evaluation of the dead volume, which is the retention volume that would be measured for a noninteracting solute with the same effective size as the protein of interest. Previous studies of SIC have used a "dead column" packed with the same resin but without the immobilized protein to evaluate the dead volume, but this creates several experimental and theoretical challenges. We have developed a new approach using a dextran standard with effective size equal to that of the protein (as determined by size exclusion chromatography). The second virial coefficient was evaluated for a monoclonal antibody over a range of buffer conditions using this new approach. The data were in good agreement with independent measurements obtained by membrane osmometry under conditions dominated by repulsive interactions. The simplicity and accuracy of this method should facilitate the use of self-interaction chromatography for quantifying protein-protein interactions.

High throughput detection of antibody self-interaction by bio-layer interferometry

Self-interaction of an antibody may lead to aggregation, low solubility or high viscosity. Rapid identification of highly developable leads remains challenging, even though progress has been made with the introduction of techniques such as self-interaction chromatography (SIC) and cross-interaction chromatography (CIC). Here, we report a high throughput method to detect antibody clone self-interaction (CSI) using bio-layer interferometry (BLI) technology. Antibodies with strong self-interaction responses in the CSI-BLI assay also show delayed retention times in SIC and CIC. This method allows hundreds of candidates to be screened in a matter of hours with minimal material consumption.

The accurate measurement of second virial coefficients using self-interaction chromatography: experimental considerations

Measurement of B22, the second virial coefficient, is an important technique for describing the solution behaviour of proteins, especially as it relates to precipitation, aggregation and crystallisation phenomena. This paper describes the best practise for calculating B22 values from self-interaction chromatograms (SIC) for aqueous protein solutions. Detailed analysis of SIC peak shapes for lysozyme shows that non-Gaussian peaks are commonly encountered for SIC, with typical peak asymmetries of 10%. This asymmetry reflects a non-linear chromatographic retention process, in this case heterogeneity of the protein-protein interactions. Therefore, it is important to use the centre of mass calculations for determining accurate retention volumes and thus B22 values. Empirical peak maximum chromatogram analysis, often reported in the literature, can result in errors of up to 50% in B22 values. A methodology is reported here for determining both the mean and the variance in B22 from SIC experiments, includes a correction for normal longitudinal peak broadening. The variance in B22 due to chemical effects is quantified statistically and is a measure of the heterogeneity of protein-protein interactions in solution. In the case of lysozyme, a wide range of B22 values are measured which can vary significantly from the average B22 values.

Rapid analysis of antibody self-association in complex mixtures using immunogold conjugates

A key challenge in developing therapeutic antibodies is their highly variable propensities to self-associate at high antibody concentrations (>50 mg/mL) required for subcutaneous delivery. Identification of monoclonal antibodies (mAbs) in the initial discovery process that not only have high binding affinity but also have high solubility and low viscosity would simplify the development of safe and effective antibody therapeutics. Unfortunately, the low purities, small quantities and large numbers of antibody candidates during the early discovery process are incompatible with current methods of measuring antibody self-association. We report a method (affinity-capture self-interaction nanoparticle spectroscopy, AC-SINS) capable of identifying mAbs with low self-association propensity that is robust even at low mAb concentrations (5-50 μg/mL) and in the presence of cell culture media. Gold nanoparticles are coated with polyclonal antibodies specific for human antibodies, and then human mAbs are captured from dilute antibody solutions. We find that the wavelength of maximum absorbance (plasmon wavelength) of antibody-gold conjugates--which red-shifts as the distance between particles is reduced due to attractive mAb self-interactions--is well correlated with light scattering measurements conducted at several orders of magnitude higher antibody concentrations. The generality of AC-SINS makes it well suited for use in diverse settings ranging from antibody discovery to formulation development.

Self-interaction chromatography of proteins on a microfluidic monolith

A novel miniaturized system has been developed for measuring protein-protein interactions in solution with high efficiency and speed, and minimal use of protein. A chromatographic monolith synthesized in a capillary is used in the method to make interaction measurements by self-interaction chromatography (SIC) in a manner that, compared to column methods, is more efficient as well as more readily practicable even if only small amounts of protein are available. The microfluidic monolith requires much less protein for both column synthesis and the chromatographic measurements than a conventional SIC system, and in addition offers improved mass transfer and hence higher chromatographic efficiency than for previous SIC miniaturization systems. Protein self-interactions for catalase as a model protein, quantified by measurement of second virial coefficients, B(22), were determined by SIC and follow trends that are consistent with previously reported values. Different column derivatization conditions were studied in order to optimize the chromatographic behavior of the microfluidic system for SIC measurements. Chromatographic sensitivity can be further increased by using different column synthesis conditions.

Protein self-interaction chromatography on a microchip

This paper presents the development of a novel miniaturized experimental procedure for the measurement of protein-protein interactions through Self-Interaction Chromatography (SIC) on a microchip, without the use of chromatographic resins. SIC was recently demonstrated to be a relatively easy method to obtain quantitative thermodynamic information about protein-protein interactions, like the osmotic second virial coefficient B(22), which relates to protein phase behavior including protein crystallization. This successful miniaturization to microchip level of a measurement device for protein self-interaction data is a first key step to a complete microfluidic screening platform for the rational design of protein crystallizations, using substantially less expensive protein and experimentation time.

A simpler analysis for the measurement of second virial coefficients by self-interaction chromatography

We describe a thermodynamic approach that supports the adoption of a simplified procedure for the determination of protein second virial coefficients (B(2)) by self-interaction chromatography. Its major advantage over the original method is a decrease in the number of parameters to which magnitudes must be assigned for the determination of B(2). Improved correlation of virial coefficients obtained by the chromatographic procedure with those obtained by light scattering is achieved by taking into account the twofold larger magnitudes of the former because of the experimental distinction between free and immobilized protein molecules in self-interaction chromatography.

Phase behavior of an intact monoclonal antibody

Understanding protein phase behavior is important for purification, storage, and stable formulation of protein drugs in the biopharmaceutical industry. Glycoproteins, such as monoclonal antibodies (MAbs) are the most abundant biopharmaceuticals and probably the most difficult to crystallize among water-soluble proteins. This study explores the possibility of correlating osmotic second virial coefficient (B(22)) with the phase behavior of an intact MAb, which has so far proved impossible to crystallize. The phase diagram of the MAb is presented as a function of the concentration of different classes of precipitants, i.e., NaCl, (NH4)2SO4, and polyethylene glycol. All these precipitants show a similar behavior of decreasing solubility with increasing precipitant concentration. B(22) values were also measured as a function of the concentration of the different precipitants by self-interaction chromatography and correlated with the phase diagrams. Correlating phase diagrams with B(22) data provides useful information not only for a fundamental understanding of the phase behavior of MAbs, but also for understanding the reason why certain proteins are extremely difficult to crystallize. The scaling of the phase diagram in B(22) units also supports the existence of a universal phase diagram of a complex glycoprotein when it is recast in a protein interaction parameter.

Screening for physical stability of a Pseudomonas amylase using self-interaction chromatography

Formulation development is an integral step in the successful commercialization of protein-based products in both the biotechnology and pharmaceutical industries. As the number of these protein formulations increases, so does the need for innovative approaches to characterize physical and chemical product stability. In this study, the osmotic second virial coefficient (B) of a commercial amylase was evaluated by self-interaction chromatography (SIC) as an innovative approach to characterize physical protein stability. B was measured as a function of pH and several common formulation additives (cosolvents), including sodium chloride, sucrose, and sorbitol. Cosolvent- and pH-induced physical stabilization of amylase is discussed in terms of positive shifts in B. Liquid chromatographic measurements of total soluble amylase and enzymatic activity measurements correlated qualitatively with trends in B except near the pI of amylase, where physical stability was minimal.

Second virial coefficient determination of a therapeutic peptide by self-interaction chromatography

Self-interaction of macromolecules has been shown to play an important role in a number of physical processes, including crystallization, solubility, viscosity, and aggregation. Peptide self-interaction is not as well studied as for larger proteins, but should play an equally important role. The osmotic second virial coefficient, B, can be used to quantify peptide and protein self-interaction. B values are typically measured using static light scattering (SLS). Peptides, however, do not scatter enough light to allow such measurements. This study describes the first use of self-interaction chromatography (SIC) for the measurement of peptide B values because SIC does not have the molecular size limitations of SLS. In the present work, SIC was used to measure B for enfuvirtide, a 36-amino acid therapeutic peptide, as a function of salt concentration, salt type, and pH. B was found to correlate strongly with solubility and apparent molecular weight. In general, the solubility of enfuvirtide increases with pH from 6 to 10 and decreases as the salt concentration increases from 0 to 0.5M for three different salts. The effect of peptide concentration on B was also investigated and shown to have a significant effect, but only at high concentrations (>80 mg/mL).