As well as having appropriate binding affinity, it is important that clinical drug candidates are non-polyreactive and have optimal biophysical properties allowing formulation at high concentrations. Our mammalian display platform has allowed direct selection of antibody variants with reduced polyreactivity and aggregation propensity from large mammalian display libraries. Thus, mammalian display addresses developability issues during the earliest stages of lead discovery and significantly de-risks the future development of antibody drugs.

Whole cell selections extend discovery of yeast-presented antibodies to integral membrane proteins, such as GPCRs. A case study from initial binder identification to functional characterization of affinity-optimized antagonist leads against a GPCR will be presented. Affinity optimization by selection and high-throughput selection-free screening of antibody variants will be discussed, along with takeaways for leveraging multiple cell lines in NGS-based enrichment analysis of highly diverse library selection outputs.

Antibody display libraries have served as a rich source of therapeutic antibodies. However, antibody leads selected from display libraries usually require downstream affinity and developability optimization, extending lead development timelines and costs. Specifica has established a unique antibody discovery display platform based on natural antibody sequences in which sub-nanomolar antibodies, requiring minimal optimization, are routinely selected.

Deciding which platforms to use for new campaigns is an important step in the antibody discovery process. Some considerations include the purpose of the antibodies (reagents vs. therapeutics), as well as the difficulty of the target. Recent campaigns at Biogen will be discussed that highlight the advantages and disadvantages of our in vivo and in vitro platforms.

We take advantage of our universal HybriFree antibody discovery engine to efficiently discover therapeutic antibody candidates by direct cloning from COVID-19-recovered patients’ blood samples. HybriFree method is further powered by our patented QMCF protein expression platform, permitting high-quality recombinant antigen and antibody production for pre-clinical research (including afucosylated antibodies for enhanced ADCC). Both technologies and relevant case studies will be presented, and discussed.

One important goal of therapeutic selection campaigns is to generate diverse antibody panels that comprehensively explore epitope space. While antibodies with very different sequences are expected to bind different epitopes, it is not clear what constitutes a “different antibody”. This seminar will describe a semi-supervised machine learning tool that utilizes low-to-moderate-trained screening data coupled to unsupervised learning built around high-throughput, next-generation sequence information to improve lead prediction accuracy.

Utilizing its proprietary DNA technology to write synthetic libraries, Twist Biopharma provides end-to-end antibody discovery and optimization solutions for the biotechnology industry.  This solution includes (1) a panel of highly diverse synthetic naïve antibody phage display libraries, (2) several target class specific antibody phage display libraries against difficult-to-drug targets, (3) a Twist Antibody Optimization (TAO) platform for antibody affinity and developability optimization and (4) a high-throughput antibody expression service.

Engineering of antibodies and related receptors is most commonly performed using phage or yeast display, but mammalian cells are used for production. To streamline engineering of these and other complex proteins, we developed a mammalian screening platform which allows for variant selection in the manufacturing host with near-native glycosylation. We have used this to identify human T cell receptors with sub-nanomolar affinities and modulate Fc binding to Fc receptors. Efforts to select for SARS-CoV-2 spike variants with increased stability and modified glycosylation sites to focus the immune response will be reported.

In 2013, we unveiled the cryoEM method, Microcrystal Electron Diffraction (MicroED). The CryoEM is used in diffraction mode for structural-using crystals that are a billion times smaller than what is used for X-ray crystallography. In this seminar, I will describe the basics of this method, from concept to data collection, analysis and structure determination, and illustrate how samples that were previously unattainable can now be studied by MicroED.

Syngene provides a one-stop solution for large molecule discovery and development using multiple modalities such as classical hybridoma, phage display, and yeast display. Yeast display combines the advantages of both hybridoma and phage methods viz. ease of identification of high affinity, manufacturability, and a higher number of sequences. Our novel methodologies to generate the desired immune response in animals and expertise in yeast display has yielded great success in antibody discovery/development for various disease condition.

Dengue virus is a mosquito-transmitted flavivirus that causes an estimated 390 million human infections each year. There are four serotypes of Dengue (DENV1-4) that can co-circulate in hyperendemic regions. Primary infection by a single serotype results in febrile illness and subsequent durable immunity to that serotype. Secondary infections by heterotypic serotypes can lead to severe shock syndrome and death. Severe Dengue disease is caused in part by cross-reactive antibodies elicited during primary infection that can bind heterologous DENV serotypes but cannot neutralize them. Instead, these non-neutralizing antibodies facilitate entry and infection in Fcg receptor positive cells thus causing "antibody-dependent enhancement" (ADE) of infection. The recent global emergence of Zika virus (ZIKV), a flavivirus that shares structural features with DENV, and the potential for ADE between DENV and ZIKV, raises concerns for vaccine strategies containing most or all epitopes in the E glycoprotein. The flavivirus E glycoprotein domain III (EDIII) is an attractive template for subunit vaccine design because it is relatively small (~100 residues) and is the target of potently neutralizing and protective antibodies for both DENV and ZIKV. However, there are limitations to the use of DENV and ZIKV EDIIIs as vaccines. For DENV EDIII, immunodominant regions lie outside of critical neutralizing epitopes, such as the broadly-neutralizing A/G-strand. Immunization of mice with ZIKV EDIII results in antibody responses to the neutralizing lateral ridge (LR) epitope, but also to the non-neutralizing ABDE sheet and CC'-loop epitopes. We have utilized structure-based protein engineering and phage display to mask unproductive epitopes of flavivirus EDIIIs by mutation while maintaining neutralizing epitopes. We have engineered these “resurfaced” EDIIIs from both DENV and ZIKV. Furthermore, we have developed protein nanoparticles containing EDIIIs from DENV and ZIKV using Spycatcher/Spytag conjugation technology. We will describe these design strategies and their potential for development of novel subunit vaccines.

Next-generation sequencing of antibody repertoires has become a promising and powerful tool in the drug discovery process. However, its practical applications are often limited since selecting potential leads from sequencing data is challenging. Common approaches focus on screening the most abundant variants in a dataset or closely related variants of already known, functional antibodies. Thus, in many cases only a small fraction of the actual data is utilized. Modern machine learning methods are uniquely suited to unlock the true potential of antibody repertoire sequencing as they can fully utilize these large datasets to discover novel, previously hidden features. Here, we present a deep learning model that can be used to extract antigen-specific sequence patterns, that act as highly accurate, immunological biomarkers. Besides its diagnostic value, we show how this model not only facilitates antibody discovery, but also generates a large amount of rationally designed, novel in silico variants which can be used to identify promising lead candidates at an early stage during drug discovery.

The cell surface proteome (surfaceome) is the primary hub for cells to communicate with the outside world. Oncogenes are known to cause huge changes in cells and we hypothesize transformation will lead to changes in the cancer surfaceome. Our lab uses proteomic technologies, both mass spectroscopy-based and a new multiplexed antibody method to systematically understand how cancer cells remodel their membrane proteomes. We then generate recombinant antibodies to detect and then attack these cells by toxifying the antibodies or recruiting immune cells to kill. I will describe our current studies in this area of attacking cancer from the outside.

G protein-coupled receptors (GPCRs) are critical regulators of most aspects of human physiology, including control of cardiovascular function and metabolic homeostasis. New methods for synthetic antibody fragment discovery allow targeting of GPCRs, offering new insights into receptor structure, function, and signaling. In particular, GPCR-targeted antibodies now enable structural studies of receptor activation mechanisms, revealing in atomic detail how agonists trigger conformational changes and thereby activate cellular signaling pathways.

Yeast have a remarkable capacity to functionally display a large fraction of the human exoproteome on their surface, including cytokines, growth-factors, and immunoreceptors. Using curated, genetically barcoded yeast libraries of these proteins, we have developed high-throughput approaches for antibody discovery, detection of protein-protein interactions, and characterization of functional humoral responses in patient samples.