Computational tool for the early screening of monoclonal antibodies for their viscosities.

Highly concentrated antibody solutions often exhibit high viscosities, which present a number of challenges for antibody-drug development, manufacturing and administration. The antibody sequence is a key determinant for high viscosity of highly concentrated solutions; therefore, a sequence- or structure-based tool that can identify highly viscous antibodies from their sequence would be effective in ensuring that only antibodies with low viscosity progress to the development phase. Here, we present a spatial charge map (SCM) tool that can accurately identify highly viscous antibodies from their sequence alone (using homology modeling to determine the 3-dimensional structures). The SCM tool has been extensively validated at 3 different organizations, and has proved successful in correctly identifying highly viscous antibodies. As a quantitative tool, SCM is amenable to high-throughput automated analysis, and can be effectively implemented during the antibody screening or engineering phase for the selection of low-viscosity antibodies.

Toward Computationally Designed Self-Reporting Biosensors Using Leave-One-Out Green Fluorescent Protein.

Leave-one-out green fluorescent protein (LOOn-GFP) is a circularly permuted and truncated GFP lacking the nth ß-strand element. LOO7-GFP derived from the wild-type sequence (LOO7-WT) folds and reconstitutes fluorescence upon addition of ß-strand 7 (S7) as an exogenous peptide. Computational protein design may be used to modify the sequence of LOO7-GFP to fit a different peptide sequence, while retaining the reconstitution activity. Here we present a computationally designed leave-one-out GFP in which wild-type strand 7 has been replaced by a 12-residue peptide (HA) from the H5 antigenic region of the Thailand strain of H5N1 influenza virus hemagglutinin. The DEEdesign software was used to generate a sequence library with mutations at 13 positions around the peptide, coding for approximately 3 × 10(5) sequence combinations. The library was coexpressed with the HA peptide in E. coli and colonies were screened for in vivo fluorescence. Glowing colonies were sequenced, and one (LOO7-HA4) with 7 mutations was purified and characterized. LOO7-HA4 folds, fluoresces in vivo and in vitro, and binds HA. However, binding results in a decrease in fluorescence instead of the expected increase, caused by the peptide-induced dissociation of a novel, glowing oligomeric complex instead of the reconstitution of the native structure. Efforts to improve binding and recover reconstitution using in vitro evolution produced colonies that glowed brighter and matured faster. Two of these were characterized. One lost all affinity for the HA peptide but glowed more brightly in the unbound oligomeric state. The other increased in affinity to the HA peptide but still did not reconstitute the fully folded state. Despite failing to fold completely, peptide binding by computational design was observed and was improved by directed evolution. The ratio of HA to S7 binding increased from 0.0 for the wild-type sequence (no binding) to 0.01 after computational design (weak binding) and to 0.48 (comparable binding) after in vitro evolution. The novel oligomeric state is composed of an open barrel.

Rational design of viscosity reducing mutants of a monoclonal antibody: hydrophobic versus electrostatic inter-molecular interactions.

High viscosity of monoclonal antibody formulations at concentrations ≥100 mg/mL can impede their development as products suitable for subcutaneous delivery. The effects of hydrophobic and electrostatic intermolecular interactions on the solution behavior of MAB 1, which becomes unacceptably viscous at high concentrations, was studied by testing 5 single point mutants. The mutations were designed to reduce viscosity by disrupting either an aggregation prone region (APR), which also participates in 2 hydrophobic surface patches, or a negatively charged surface patch in the variable region. The disruption of an APR that lies at the interface of light and heavy chain variable domains, VH and VL, via L45K mutation destabilized MAB 1 and abolished antigen binding. However, mutation at the preceding residue (V44K), which also lies in the same APR, increased apparent solubility and reduced viscosity of MAB 1 without sacrificing antigen binding or thermal stability. Neutralizing the negatively charged surface patch (E59Y) also increased apparent solubility and reduced viscosity of MAB 1, but charge reversal at the same position (E59K/R) caused destabilization, decreased solubility and led to difficulties in sample manipulation that precluded their viscosity measurements at high concentrations. Both V44K and E59Y mutations showed similar increase in apparent solubility. However, the viscosity profile of E59Y was considerably better than that of the V44K, providing evidence that inter-molecular interactions in MAB 1 are electrostatically driven. In conclusion, neutralizing negatively charged surface patches may be more beneficial toward reducing viscosity of highly concentrated antibody solutions than charge reversal or aggregation prone motif disruption.

Highly viscous antibody solutions are a consequence of network formation caused by domain-domain electrostatic complementarities: insights from coarse-grained simulations.

Therapeutic monoclonal antibody (mAb) candidates that form highly viscous solutions at concentrations above 100 mg/mL can lead to challenges in bioprocessing, formulation development, and subcutaneous drug delivery. Earlier studies of mAbs with concentration-dependent high viscosity have indicated that mAbs with negatively charged Fv regions have a dipole-like quality that increases the likelihood of reversible self-association. This suggests that weak electrostatic intermolecular interactions can form transient antibody networks that participate in resistance to solution deformation under shear stress. Here this hypothesis is explored by parametrizing a coarse-grained (CG) model of an antibody using the domain charges from four different mAbs that have had their concentration-dependent viscosity behaviors previously determined. Multicopy molecular dynamics simulations were performed for these four CG mAbs at several concentrations to understand the effect of surface charge on mass diffusivity, pairwise interactions, and electrostatic network formation. Diffusion coefficients computed from simulations were in qualitative agreement with experimentally determined viscosities for all four mAbs. Contact analysis revealed an overall greater number of pairwise interactions for the two mAbs in this study with high concentration viscosity issues. Further, using equilibrated solution trajectories, the two mAbs with high concentration viscosity issues quantitatively formed more features of an electrostatic network than the other mAbs. The change in the number of these network features as a function of concentration is related to the number of pairwise interactions formed by electrostatic complementarities between antibody domains. Thus, transient antibody network formation caused by domain-domain electrostatic complementarities is the most probable origin of high concentration viscosity for mAbs in this study.

Concentration dependent viscosity of monoclonal antibody solutions: explaining experimental behavior in terms of molecular properties.

PURPOSE: Early identification of monoclonal antibody candidates whose development, as high concentration (≥100 mg/mL) drug products, could prove challenging, due to high viscosity, can help define strategies for candidate engineering and selection. METHODS: Concentration dependent viscosities of 11 proprietary mAbs were measured. Sequence and structural features of the variable (Fv) regions were analyzed to understand viscosity behavior of the mAbs. Coarse-grained molecular simulations of two problematic mAbs were compared with that of a well behaved mAb. RESULTS: Net charge, ξ-potential and pI of Fv regions were found to correlate with viscosities of highly concentrated antibody solutions. Negative net charges on the Fv regions of two mAbs with poor viscosity behaviors facilitate attractive self-associations, causing them to diffuse slower than a well-behaved mAb with positive net charge on its Fv region. An empirically derived equation that connects aggregation propensity and pI of the Fv region with high concentration viscosity of the whole mAb was developed. CONCLUSIONS: An Fv region-based qualitative screening profile was devised to flag mAb candidates whose development, as high concentration drug products, could prove challenging. This screen can facilitate developability risk assessment and mitigation strategies for antibody based therapeutics via rapid high throughput material-free screening.

Agitation-induced aggregation and subvisible particulate formation in model proteins.

The kinetics of agitation-induced subvisible particle formation was investigated for a few model proteins - human serum albumin (HSA), hen egg white lysozyme (HEWL), and a monoclonal antibody (IgG2). Experiments were carried out for the first time under relatively low protein concentration and low agitation speed to investigate the details of subvisible particle formation at the initial phase of aggregation (<2%) process. Upon agitation, both soluble higher molecular mass species (HMMS) and subvisible particles (SbVPs) formed at different rates, and via different mechanisms. Agitation enhanced exposure of hydrophobic sites in HSA but did not cause detectable structural changes in HEWL and IgG2. SbVPs from HSA partially dissociates in a neutral pH buffer (SEC mobile phase) but does not upon dilution in the same formulation buffer. Opposite results were obtained for SbVPs from IgG2 and HEWL. Neither the relative hydrophobic surface area nor the Tm of the model proteins seems to be an indicator of tendency for agitation-mediated SbVP formation. Taken together, our data suggests that agitation-induced SbVP formation can occur through different mechanisms and can vary, depending on the protein and solution conditions.

On the role of aggregation prone regions in protein evolution, stability, and enzymatic catalysis: insights from diverse analyses.

The various roles that aggregation prone regions (APRs) are capable of playing in proteins are investigated here via comprehensive analyses of multiple non-redundant datasets containing randomly generated amino acid sequences, monomeric proteins, intrinsically disordered proteins (IDPs) and catalytic residues. Results from this study indicate that the aggregation propensities of monomeric protein sequences have been minimized compared to random sequences with uniform and natural amino acid compositions, as observed by a lower average aggregation propensity and fewer APRs that are shorter in length and more often punctuated by gate-keeper residues. However, evidence for evolutionary selective pressure to disrupt these sequence regions among homologous proteins is inconsistent. APRs are less conserved than average sequence identity among closely related homologues (≥80% sequence identity with a parent) but APRs are more conserved than average sequence identity among homologues that have at least 50% sequence identity with a parent. Structural analyses of APRs indicate that APRs are three times more likely to contain ordered versus disordered residues and that APRs frequently contribute more towards stabilizing proteins than equal length segments from the same protein. Catalytic residues and APRs were also found to be in structural contact significantly more often than expected by random chance. Our findings suggest that proteins have evolved by optimizing their risk of aggregation for cellular environments by both minimizing aggregation prone regions and by conserving those that are important for folding and function. In many cases, these sequence optimizations are insufficient to develop recombinant proteins into commercial products. Rational design strategies aimed at improving protein solubility for biotechnological purposes should carefully evaluate the contributions made by candidate APRs, targeted for disruption, towards protein structure and activity.

Impact of deglycosylation and thermal stress on conformational stability of a full length murine IgG2a monoclonal antibody: observations from molecular dynamics simulations.

With the rise of antibody based therapeutics as successful medicines, there is an emerging need to understand the fundamental antibody conformational dynamics and its implications towards stability of these medicines. Both deglycosylation and thermal stress have been shown to cause conformational destabilization and aggregation in monoclonal antibodies. Here, we study instabilities caused by deglycosylation and by elevated temperature (400 K) by performing molecular dynamic simulations on a full length murine IgG2a mAb whose crystal structure is available in the Protein Data bank. C(α)-atom root mean square deviation and backbone root mean square fluctuation calculations show that deglycosylation perturbs quaternary and tertiary structures in the C(H) 2 domains. In contrast, thermal stress pervades throughout the antibody structure and both Fabs and Fc regions are destabilized. The thermal stress applied in this study was not sufficient to cause large scale unfolding within the simulation time and most amino acid residues showed similar average solvent accessible surface area and secondary structural conformations in all trajectories. C(H) 3 domains were the most successful at resisting the conformational destabilization. The simulations helped identify aggregation prone regions, which may initiate cross-ß motif formation upon deglycosylation and upon applying thermal stress. Deglycosylation leads to increased backbone fluctuations and solvent exposure of a highly conserved APR located in the edge ß-strand A of the C(H) 2 domains. Aggregation upon thermal stress is most likely initiated by two APRs that overlap with the complementarity determining regions. This study has important implications for rational design of antibody based therapeutics that are resistant towards aggregation.

Insights into the potential aggregation liabilities of the b12 Fab fragment via elevated temperature molecular dynamics.

Aggregation is a common hurdle faced during the development of antibody therapeutics. In this study, we explore the potential aggregation liabilities of the Fab (fragment antigen-binding) from a human IgG1κ antibody via multiple elevated temperature molecular dynamic simulations, analogous to accelerated stability studies performed during formulation development. Deformation and solvent exposure changes in response to thermal stress were monitored for individual structural domains (V(H), V(L), C(H)1 and C(L)), their interfaces (V(H):V(L) and C(H)1:C(L)), edge beta-strands and sequence-predicted aggregation-prone regions (APRs). During simulations, domain interfaces deformed prior to the unfolding of individual domains. However, interfacial beta-strands retained their secondary structure and remained solvent protected longer than all other strands or loops. Thus, APRs located in interfacial beta-strands are effectively blocked from self-association. Structural deformations were also observed in complementarity-determining regions, edge beta-strands and adjoining framework beta-strands, which increased their solvent-accessible surface area and exposed APRs in these regions. From the analysis of these structural changes, two potential aggregation liabilities were identified in the V(H) domain of this Fab. Insights gained from this investigation should be useful in devising a rational structure-based strategy for the design and selection of antibody candidates with high potency and improved developability.

Consequences of glycan truncation on Fc structural integrity.

Effective characterization of protein-based therapeutic candidates such as monoclonal antibodies (mAbs) is important to facilitate their successful progression from early discovery and development stages to marketing approval. One challenge relevant to biopharmaceutical development is, understanding how the stability of a protein is affected by the presence of an attached oligosaccharide, termed a glycan. To explore the utility of molecular dynamics simulations as a complementary technique to currently available experimental methods, the Fc fragment was employed as a model system to improve our understanding of protein stabilization by glycan attachment. Long molecular dynamics simulations were performed on three Fc glycoform variants modeled using the crystal structure of a human IgG1 mAb. Two of these three glycoform variants have their glycan carbohydrates partially or completely removed. Structural differences among the glycoform variants during simulations suggest that glycan truncation and/or removal can cause quaternary structural deformation of the Fc as a result of the loss or disruption of a significant number of inter-glycan contacts that are not formed in the human IgG1 crystal structure, but do form during simulations described here. Glycan truncation/removal can also increase the tertiary structural deformation of CH2 domains, demonstrating the importance of specific carbohydrates toward stabilizing individual CH2 domains. At elevated temperatures, glycan truncation can also differentially affect structural deformation in locations (Helix-1 and Helix-2) that are far from the oligosaccharide attachment point. Deformation of these helices, which form part of the FcRn, could affect binding if these regions are unable to refold after temperature normalization. During elevated temperature simulations of the deglycosylated variant, CH2 domains collapsed onto CH3 domains. Observations from these glycan truncation/removal simulations have improved our understanding on how glycan composition can affect mAb stability.

Computational methods to predict therapeutic protein aggregation.

Protein based biotherapeutics have emerged as a successful class of pharmaceuticals. However, these macromolecules endure a variety of physicochemical degradations during manufacturing, shipping, and storage, which may adversely impact the drug product quality. Of these degradations, the irreversible self-association of therapeutic proteins to form aggregates is a major challenge in the formulation of these molecules. Tools to predict and mitigate protein aggregation are, therefore, of great interest to biopharmaceutical research and development. In this chapter, a number of such computational tools developed to understand and predict the various steps involved in protein aggregation are described. These tools can be grouped into three general classes: unfolding kinetics and native state thermal stability, colloidal stability, and sequence/structure based aggregation liabilities. Chapter sections introduce each class by discussing how these predictive tools provide insight into the molecular events leading to protein aggregation. The computational methods are then explained in detail along with their advantages and limitations.

Constraining local structure can speed up folding by promoting structural polarization of the folding pathway.

The pathway which proteins take to fold can be influenced from the earliest events of structure formation. In this light, it was both predicted and confirmed that increasing the stiffness of a beta hairpin turn decreased the size of the transition state ensemble (TSE), while increasing the folding rate. Thus, there appears to be a relationship between conformationally restricting the TSE and increasing the folding rate, at least for beta hairpin turns. In this study, we hypothesize that the enormous sampling necessary to fold even two-state folding proteins in silico could be reduced if local structure constraints were used to restrict structural heterogeneity by polarizing folding pathways or forcing folding into preferred routes. Using a Go model, we fold Chymotrypsin Inhibitor 2 (CI-2) and the src SH3 domain after constraining local sequence windows to their native structure by rigid body dynamics (RBD). Trajectories were monitored for any changes to the folding pathway and differences in the kinetics compared with unconstrained simulations. Constraining local structure decreases folding time two-fold for 41% of src SH3 windows and 45% of CI-2 windows. For both proteins, folding times are never significantly increased after constraining any window. Structural polarization of the folding pathway appears to explain these rate increases. Folding rate enhancements are consistent with the goal to reduce sampling time necessary to reach native structures during folding simulations. As anticipated, not all constrained windows showed an equal decrease in folding time. We conclude by analyzing these differences and explain why RBD may be the preferred way to constrain structure.

Simulating protein folding initiation sites using an alpha-carbon-only knowledge-based force field.

Protein folding is a hierarchical process where structure forms locally first, then globally. Some short sequence segments initiate folding through strong structural preferences that are independent of their three-dimensional context in proteins. We have constructed a knowledge-based force field in which the energy functions are conditional on local sequence patterns, as expressed in the hidden Markov model for local structure (HMMSTR). Carbon-alpha force field (CALF) builds sequence specific statistical potentials based on database frequencies for alpha-carbon virtual bond opening and dihedral angles, pair-wise contacts and hydrogen bond donor-acceptor pairs, and simulates folding via Brownian dynamics. We introduce hydrogen bond donor and acceptor potentials as alpha-carbon probability fields that are conditional on the predicted local sequence. Constant temperature simulations were carried out using 27 peptides selected as putative folding initiation sites, each 12 residues in length, representing several different local structure motifs. Each 0.6 micros trajectory was clustered based on structure. Simulation convergence or representativeness was assessed by subdividing trajectories and comparing clusters. For 21 of the 27 sequences, the largest cluster made up more than half of the total trajectory. Of these 21 sequences, 14 had cluster centers that were at most 2.6 A root mean square deviation (RMSD) from their native structure in the corresponding full-length protein. To assess the adequacy of the energy function on nonlocal interactions, 11 full length native structures were relaxed using Brownian dynamics simulations. Equilibrated structures deviated from their native states but retained their overall topology and compactness. A simple potential that folds proteins locally and stabilizes proteins globally may enable a more realistic understanding of hierarchical folding pathways.