Computational chemists have long used quantitative structure-property relationships (QSPR) to elucidate design strategies to improve the drug-like properties of small-molecule therapeutics. However, duplicating such efforts in the context of biologics modalities remains challenging. In this presentation, two case studies describe the application of QSPR in the context of biologics drug discovery. In both cases, the QSPR methodology notably improved the drug-like properties found in both sets of affinity-matured antibodies.

A mechanism-based biopanning strategy that specifically selects for antibodies that interfere with the interaction between the the SARS-CoV-2 Spike protein and ACE2 was used to identify antibodies that potently block ACE2 binding, yet exhibit divergent neutralization efficacy against live virus and in reconstituted human nasal and bronchial epithelium models. Cryo-EM structures of three different Spike-antibody complexes reveal distinct binding modes that alter the functional cycle of Spike conformations.

We used protein engineering to develop tetravalent synthetic neutralizing antibodies for SARS-CoV-2. We show that these antibodies can be produced at large scale and possess stability and specificity comparable to approved antibody drugs. The best antibody targets the host receptor binding site of the virus spike protein, and thus, its tetravalent version can block virus infection with a potency that far exceeds that of the bivalent IgG.

In three weeks, AbCellera discovered, characterized and selected hundreds of antibodies against SARS-CoV-2 from one of the first U.S. patients to recover from COVID-19. AbCellera’s technology stack combines AI-assisted high-throughput single B cell screening with immune repertoire profiling of natural immune responses. Bioinformatic analysis of the resulting panels of antibodies allowed for the rapid characterization of neutralizing antibodies and the identification of therapeutic lead candidates including bamlanivimab.

Target selection is a key feature in cancer immunotherapy. In this respect, gangliosides, a broad family of structurally related glycolipids, represent potential targets based on their higher abundance in tumors when compared with the matched normal tissues. This presentation will discuss the relevance of O-acetyl-GD2 as a novel target and  presents the available preclinical data of O-acetyl-GD2-specific immunotherapies.

One of the remaining issues of antibody therapeutics is on-target, but off-tumor, toxicity induced by binding to target antigens expressed in normal tissues. To overcome this problem, we have established novel antibody engineering to enable antibody binding to the antigen selectively at tumor site but not at normal tissues. In this presentation the mechanism of selective binding and its application onto therapeutics will be discussed.

Built on a de novo designed three-helix bundle protein, D domains constitute a robust targeting domain. D domains are utilized in both conventional chimeric antigen receptors as well as our ARC-sparX platform, a next generation CAR T cell therapy, which separates antigen recognition from the T cell effector function. We describe the D domain library design, selection and screening approaches as well as the characterization and manufacture of these novel therapeutics.

We describe Chain Selectivity Assessment, a high-throughput method to rationally select parental monoclonal antibodies to make bispecific antibodies requiring correct heavy/light chain pairing. With CSA, we have successfully identified mAbs that exhibit a native preference towards cognate chain pairing that enables the production of hetero-IgGs without additional engineering. CSA also identified rare light chains that permit positive binding of the non-cognate arm in the common LC hetero-IgGs, also without engineering.

We accomplished antibody discoveries against the nucleocapsid (N) protein of SARS-CoV-2 by working with phage-displayed synthetic antibody libraries designed with artificial intelligence models trained on antibody-antigen interactions. This work establishes a technological platform for rapidly developing lateral flow immunoassay (LFIA) devices in responding not only to the current COVID-19 pandemic but also in managing other infectious disease outbreaks in humans and animals.

Therapeutic proteins have an inherent-sequence based immunogenicity risk. In silico methods are used to predict MHC class II binding epitopes derived from therapeutic proteins and estimate immunogenicity risk. Further engineering of antibodies and other therapeutic proteins can modify this risk. In this talk, I will discuss challenges, current methods, and potential solutions of how to assess the immunogenicity risk of engineered antibodies and proteins.

Biotherapeutics undergo several critical steps to elicit CD4+ T cell dependent anti-drug antibody responses. Current preclinical immunogenicity risk assessment focuses on these steps using a panel of tools and assays, including in silico prediction, dendritic cell internalization, MAPPS, T cell activation, and pre-existing reactivity. All data shall be considered holistically to enable molecule selection and de-immunization engineering. Herein, an integrated immunogenicity risk assessment approach was presented with case examples.

Modern Machine Learning search algorithms in conjunction with the ability to synthesize sets of systematically varied antibodies allow the search of very large space (>1015) with just a few hundred antibody variants, enabling multidimensional optimization of 'hard-to-measure' antibody functions. We integrate this engineering process directly into Leap-In transposon-mediated stable cell lines for rapid generation of gram quantities of target antibodies.