Codon Optimization Using a Recurrent Neural Network.

Codon optimization of a DNA sequence can significantly increase efficiency of protein expression, reducing the cost to manufacture biologic pharmaceuticals. Although directed methods based on such factors as codon usage bias and GC nucleotide content are often used to optimize protein expression, undirected optimization using machine learning could further improve the process by capitalizing on undiscovered patterns that exist within real DNA sequences. To explore this hypothesis, Chinese hamster DNA sequences were used to train a recurrent neural network (RNN) model of codon optimization. The model was used to generate optimized DNA sequence based on an input amino acid sequence for the example receptor programmed death-ligand 1 and for an example monoclonal antibody. When RNN-optimized sequences were transfected transiently or stably into Chinese hamster ovary cells, the resulting protein expression was as high or higher than that produced by DNA sequences optimized by conventional algorithms.

ACE2-Fc fusion protein overcomes viral escape by potently neutralizing SARS-CoV-2 variants of concern.

COVID-19, an infectious disease caused by the SARS-CoV-2 virus, emerged globally in early 2020 and has remained a serious public health issue. To date, although several preventative vaccines have been approved by FDA and EMA, vaccinated individuals increasingly suffer from breakthrough infections. Therapeutic antibodies may provide an alternative strategy to neutralize viral infection and treat serious cases; however, the clinical data and our experiments show that some FDA-approved monoclonal antibodies lose function against COVID-19 variants such as Omicron. Therefore, in this study, we present a novel therapeutic agent, SI-F019, an ACE2-Fc fusion protein whose neutralization efficiency is not compromised, but actually strengthened, by the mutations of dominant variants including Omicron. Comprehensive biophysical analyses revealed the mechanism of increased inhibition to be enhanced interaction of SI-F019 with all the tested spike variants, in contrast to monoclonal antibodies which tended to show weaker binding to some variants. The results imply that SI-F019 may be a broadly useful agent for treatment of COVID-19.

Engineering an Enhanced EGFR Engager: Humanization of Cetuximab for Improved Developability.

The epidermal growth factor receptor (EGFR) is a receptor tyrosine kinase whose proliferative effects can contribute to the development of many types of solid tumors when overexpressed. For this reason, EGFR inhibitors such as cetuximab can play an important role in treating cancers such as colorectal cancer and head and neck cancer. Cetuximab is a chimeric monoclonal antibody containing mouse variable regions that bind to EGFR and prevent it from signaling. Although cetuximab has been used clinically since 2004 to successfully control solid tumors, advances in protein engineering have created the opportunity to address some of its shortcomings. In particular, the presence of mouse sequences could contribute to immunogenicity in the form of anti-cetuximab antibodies, and an occupied glycosylation site in FR3 can contribute to hypersensitivity reactions and product heterogeneity. Using simple framework graft or sequence-/structure-guided approaches, cetuximab was humanized onto 11 new frameworks. In addition to increasing humanness and removing the VH glycosylation site, dynamic light scattering revealed increases in stability, and bio-layer interferometry confirmed minimal changes in binding affinity, with patterns emerging across the humanization method. This work demonstrates the potential to improve the biophysical and clinical properties of first-generation protein therapeutics and highlights the advantages of computationally guided engineering.

Considerations for the Design of Antibody-Based Therapeutics.

Antibody-based proteins have become an important class of biologic therapeutics, due in large part to the stability, specificity, and adaptability of the antibody framework. Indeed, antibodies not only have the inherent ability to bind both antigens and endogenous immune receptors but also have proven extremely amenable to protein engineering. Thus, several derivatives of the monoclonal antibody format, including bispecific antibodies, antibody-drug conjugates, and antibody fragments, have demonstrated efficacy for treating human disease, particularly in the fields of immunology and oncology. Reviewed here are considerations for the design of antibody-based therapeutics, including immunological context, therapeutic mechanisms, and engineering strategies. First, characteristics of antibodies are introduced, with emphasis on structural domains, functionally important receptors, isotypic and allotypic differences, and modifications such as glycosylation. Then, aspects of therapeutic antibody design are discussed, including identification of antigen-specific variable regions, choice of expression system, use of multispecific formats, and design of antibody derivatives based on fragmentation, oligomerization, or conjugation to other functional moieties. Finally, strategies to enhance antibody function through protein engineering are reviewed while highlighting the impact of fundamental biophysical properties on protein developability.

Design and characterization of novel dual Fc antibody with enhanced avidity for Fc receptors.

Monoclonal antibodies (mAbs) have become an important class of therapeutics, particularly in the realm of anticancer immunotherapy. While the two antigen-binding fragments (Fabs) of an mAb allow for high-avidity binding to molecular targets, the crystallizable fragment (Fc) engages immune effector elements. mAbs of the IgG class are used for the treatment of autoimmune diseases and can elicit antitumor immune functions not only by several mechanisms including direct antigen engagement via their Fab arms but also by Fab binding to tumors combined with Fc engagement of complement component C1q and Fcγ receptors. Additionally, IgG binding to the neonatal Fc receptor (FcRn) allows for endosomal recycling and prolonged serum half-life. To augment the effector functions or half-life of an IgG1 mAb, we constructed a novel "2Fc" mAb containing two Fc domains in addition to the normal two Fab domains. Structural and functional characterization of this 2Fc mAb demonstrated that it exists in a tetrahedral-like geometry and retains binding capacity via the Fab domains. Furthermore, duplication of the Fc region significantly enhanced avidity for Fc receptors FcγRI, FcγRIIIa, and FcRn, which manifested as a decrease in complex dissociation rate that was more pronounced at higher densities of receptor. At intermediate receptor density, the dissociation rate for Fc receptors was decreased 6- to 130-fold, resulting in apparent affinity increases of 7- to 42-fold. Stoichiometric analysis confirmed that each 2Fc mAb may simultaneously bind two molecules of FcγRI or four molecules of FcRn, which is double the stoichiometry of a wild-type mAb. In summary, duplication of the IgG Fc region allows for increased avidity to Fc receptors that could translate into clinically relevant enhancement of effector functions or pharmacokinetics.

Antibody Structure and Function: The Basis for Engineering Therapeutics.

Antibodies and antibody-derived macromolecules have established themselves as the mainstay in protein-based therapeutic molecules (biologics). Our knowledge of the structure-function relationships of antibodies provides a platform for protein engineering that has been exploited to generate a wide range of biologics for a host of therapeutic indications. In this review, our basic understanding of the antibody structure is described along with how that knowledge has leveraged the engineering of antibody and antibody-related therapeutics having the appropriate antigen affinity, effector function, and biophysical properties. The platforms examined include the development of antibodies, antibody fragments, bispecific antibody, and antibody fusion products, whose efficacy and manufacturability can be improved via humanization, affinity modulation, and stability enhancement. We also review the design and selection of binding arms, and avidity modulation. Different strategies of preparing bispecific and multispecific molecules for an array of therapeutic applications are included.

Toward a Combinatorial Approach for the Prediction of IgG Half-Life and Clearance.

The serum half-life and clearance of therapeutic monoclonal antibodies (mAbs) are critical factors that impact their efficacy and optimal dosing regimen. The pH-dependent binding of an mAb to the neonatal Fc receptor (FcRn) has long been recognized as an important determinant of its pharmacokinetics. However, FcRn affinity alone is not a reliable predictor of mAb half-life, suggesting that other biologic or biophysical mechanisms must be accounted for. mAb thermal stability, which reflects its unfolding and aggregation propensities, may also relate to its pharmacokinetic properties. However, no rigorous statistical regression methods have been used to identify combinations of physical parameters that best predict biologic properties. In this work, a panel of eight mAbs with published human pharmacokinetic data were selected for biophysical analyses of FcRn binding and thermal stability. Biolayer interferometry was used to characterize FcRn/mAb binding at acidic and neutral pH, while differential scanning calorimetry was used to determine thermodynamic unfolding parameters. Individual binding or stability parameters were generally weakly correlated with half-life and clearance values. Least absolute shrinkage and selection operator regression was used to identify the combination of two parameters with the best correlation to half-life and clearance as being the FcRn binding response at pH 7.0 and the change in heat capacity. Leave-one-out subsampling yielded a root mean square difference between observed and predicted half-life of just 2.7 days (16%). Thus, the incorporation of multiple biophysical parameters into a cohesive model may facilitate early-stage prediction of in vivo half-life and clearance based on simple in vitro experiments.

Kinetic mechanism of controlled Fab-arm exchange for the formation of bispecific immunoglobulin G1 antibodies.

Bispecific antibodies (bsAbs) combine the antigen specificities of two distinct Abs and demonstrate therapeutic promise based on novel mechanisms of action. Among the many platforms for creating bsAbs, controlled Fab-arm exchange (cFAE) has proven useful based on minimal changes to native Ab structure and the simplicity with which bsAbs can be formed from two parental Abs. Despite a published protocol for cFAE and its widespread use in the pharmaceutical industry, the reaction mechanism has not been determined. Knowledge of the mechanism could lead to improved yields of bsAb at faster rates as well as foster adoption of process control. In this work, a combination of Förster resonance energy transfer (FRET), nonreducing SDS-PAGE, and strategic mutation of the Ab hinge region was employed to identify and characterize the individual steps of cFAE. Fluorescence correlation spectroscopy (FCS) was used to determine the affinity of parental (homodimer) and bispecific (heterodimer) interactions within the CH3 domain, further clarifying the thermodynamic basis for bsAb formation. The result is a clear sequence of events with rate constants that vary with experimental conditions, where dissociation of the K409R parental Ab into half-Ab controls the rate of the reaction.

Rapid Purification of Human Bispecific Antibodies via Selective Modulation of Protein A Binding.

Methods to rapidly generate high quality bispecific antibodies (BsAb) having normal half-lives are critical for therapeutic programs. Here, we identify 3 mutations (T307P, L309Q, and Q311R or "TLQ") in the Fc region of human IgG1 which disrupt interaction with protein A while enhancing interaction with FcRn. The mutations are shown to incrementally alter the pH at which a mAb elutes from protein A affinity resin. A BsAb comprised of a TLQ mutant and a wild-type IgG1 can be efficiently separated from contaminating parental mAbs by differential protein A elution starting from either a) purified parental mAbs, b) in-supernatant crossed parental mAbs, or c) co-transfected mAbs. We show that the Q311R mutation confers enhanced FcRn interaction in vitro, and Abs harboring either the Q311R or TLQ mutations have serum half-lives as long as wild-type human IgG1. The mutant Abs have normal thermal stability and Fcγ receptor interactions. Together, the results lead to a method for high-throughput generation of BsAbs suitable for in vivo studies.

Modulation of protein A binding allows single-step purification of mouse bispecific antibodies that retain FcRn binding.

The increased number of bispecific antibodies (BsAb) under therapeutic development has resulted in a need for mouse surrogate BsAbs. Here, we describe a one-step method for generating highly pure mouse BsAbs suitable for in vitro and in vivo studies. We identify two mutations in the mouse IgG2a and IgG2b Fc region: one that eliminates protein A binding and one that enhances protein A binding by 8-fold. We show that BsAbs harboring these mutations can be purified from the residual parental monoclonal antibodies in one step using protein A affinity chromatography. The structural basis for the effects of these mutations was analyzed by X-ray crystallography. While the mutation that disrupted protein A binding also inhibited FcRn interaction, a bispecific mutant in which one subunit retained the ability to bind protein A could still interact with FcRn. Pharmacokinetic analysis of the serum half-lives of the mutants showed that the mutant BsAb had a serum half-life comparable to a wild-type Ab. The results describe a rapid method for generating panels of mouse BsAbs that could be used in mouse studies.

Diffusion of Soluble Aggregates of THIOMABs and Bispecific Antibodies in Serum.

Submicrometer aggregates are frequently present at low levels in antibody-based therapeutics. Although intuition suggests that the fraction of the aggregate or the size of the aggregate present might correlate with deleterious clinical properties or formulation difficulties, it has been challenging to demonstrate which aggregate states, if any, trigger specific biological effects. One source of uncertainty about the putative linkage between aggregation and safety or efficacy lies in the likelihood that noncovalent aggregation differs in ideal buffers versus in serum and biological tissues; self-association or association with other proteins may vary widely with environment. Therefore, methods for monitoring aggregation and aggregate behavior in biologically relevant matrices could provide a tool for better predicting aggregate-dependent clinical outcomes and provide a basis for antibody engineering prior to clinical studies. Here, we generate models for soluble aggregates of THIOMABs and a bispecific antibody (bsAb) of defined size and exploit fluorescence correlation spectroscopy to monitor their diffusion properties in serum and viscosity-matched buffers. The monomers, dimers, and trimers of both THIOMABs and a bsAb reveal a modest increase in diffusion time in serum greater than expected for an increase in viscosity alone. A mixture of larger aggregates containing mostly bsAb pentamers exhibits a marked increase in diffusion time in serum and much greater intrasample variability, consistent with significant aggregation or interactions with serum components. The results indicate that small aggregates of several IgG platforms are not likely to aggregate with serum components, but nanometer-scale aggregates larger than trimers can interact with the serum in an Ab-dependent manner.