The elusive relationship between protein sequence, expression, and solubility necessitates costly and time-consuming analogue screening for protein therapeutics R&D. In this presentation, I will share our work on applying machine learning to predict protein expression and to accelerate protein solubility assessment, using internal and publicly available data sources.

I will present novel methodologies for the de novo
design of proteins and antibodies, including a protein design platform based on modern convolutional neural network architectures and methods for de novo design of H3 loops.

Integral Molecular has a 95% success rate of discovering high-affinity monoclonal antibodies against multipass membrane proteins despite the structural complexity and low immunogenicity of these targets. Here, we present some of the lessons we have learned to enable discovery campaigns against these historically intractable targets, including GPCRs, ion channels and transporters. Case studies for CB1, P2X7, SLC2A4 and Claudin18.2 will be discussed.

Recent advances in paired heavy-light sequencing, native antibody display, and computational analysis of NGS datasets have opened up new possibilities for mining antibody responses. We have developed new strategies in library design and screening to enable comprehensive antibody developmental pathway assessment and to survey the entire landscape of possible single mutations. We are also exploring new features related to the genetic regulation of antibody development.  We will discuss applications of these technologies to better understand antibody development and to identify improved antibody therapeutics.

Traditional approaches for engineering T cell receptors (TCR) with increased affinity can be largely unsuitable for T cell therapy applications, as they can yield TCR variants that promote T cell exhaustion, preclude serial TCR triggering and, in some instances, cause fatal cross-reactivity. Using multi-step CRISPR-Cas9 genome editing, we have developed a novel cellular TCR display platform supporting high-throughput selection of optimal TCRs by means of functional screening and deep sequencing.

Monoclonal antibodies in their many divergent formats have revolutionized medicine and today represent >$100B/year in pharmaceutical sales. ATUM has built an integrated pipeline from generation of antigens via affinity maturation, developability, engineering, and humanization all the way through scale-up and stable cell line generations. The presentation will include case studies highlighting the process; each step uses technological breakthroughs in synthetic biology, machine learning, LIMS data integration, robotics, and engineered transposases to ensure maximum efficiency.

There is increasing demand for sophisticated specificity profiles for next-generation biotherapeutics. Gone are the days when we simply needed to generate high affinity to the target of interest. We often need to achieve extremely high specificity, as well as cross-reactivity among relevant targets and epitopes. I will illustrate critical roles that the design of library sorting schemes plays in achieving exquisite binding specificity using several case studies.

VISTA is a checkpoint inhibitor that preferentially engages T cells at low pH. We engineered potent, acidic pH-selective mAbs that bind to VISTA and block its interaction with the receptor, PSGL-1. Further characterization using revertants, also a structure of VISTA:Fab complex, revealed the mechanism for the acidic pH selectivity. The acidic pH-selective mAbs have improved PK and superior tumor targeting compared to a pH-independent mAb, while retaining in vivo efficacy.

We identified hundreds of target-specific antibodies using high-throughput, single-cell screening from humanized mice. To reveal greater diversity, we sequenced the immunoglobulin repertoire of immunized animals and superimposed validated single-cell-derived sequences using bioinformatics to reconstruct clonal trees. This versatile approach dramatically expands the total available number of fully human, in vivo-generated antibodies against any therapeutic target.

The presentation will describe how analyses of the Fc receptor, FcRn, using a combination of protein engineering, cellular biological studies, and mouse disease models, have been used to design therapeutics to modulate the dynamic behavior of antibodies. The generation of engineered antibody-drug conjugates that are designed to deliver their cytotoxic payload more efficiently through altered subcellular trafficking behavior to tumor cells will also be discussed.

To allow biologics to reach intracellular targets, Feldan Therapeutics generated peptides that can deliver proteins, peptides, and protein complexes in cells. Using intranasal instillations, Feldan and collaborators showed that these carrier peptides (“Feldan Shuttles”) efficiently deliver CRISPR complexes in lungs, leading to genetic modifications of epithelial cells in vivo. This example proves the therapeutic interest of the Shuttle platform for pulmonary diseases by showing that it can lead to functional changes in lungs.

As the complexity of therapeutic targets continues to evolve, more efficient and predictive identification of pertinent antibody candidates becomes increasingly valuable. Incorporating state-of-the-art, high-resolution characterization tools and methods into discovery workflows enables rapid and reliable selection of lead candidates more effectively than traditional techniques. The presentation will focus on elucidating function early during screening and development to enhance candidate triage, as opposed to relying on brute-force methods to characterize large panels containing primarily non-functional clones.

The early selection of biotherapeutics requires the meticulous assessment of a variety of molecule properties to ensure their desired quality, safety, and efficacy. The increased complexity of non-standard formats requires further molecular design and engineering strategies to optimize their properties, as well as new screening approaches to select the best candidate molecules. In this presentation, I will discuss applications and current engineering challenges to optimize the developability profile of molecules during early discovery and accelerate their development.

The anti-PCSK9 antibody, Bococizumab, was terminated after Phase 3 due to high immunogenicity. We assessed its intrinsic immunogenicity risk using our current tools, consisting of in silico analyses, MAPPS, and T cell activation. We observed significant presentation of non-germline residues and Bococizumab-driven T cell activation in 90% and 56% of donors, respectively. In contrast, anti-PCSK9 antibodies with low immunogenicity (Evolocumab and Alirocumab) demonstrated low-to-moderate presentation of non-germline residues and no T cell activation.

Antibody engineering is now more streamlined than ever, but several factors contribute to its complexity such as concerns around immunogenicity, increasing diversity of antibody formats, and the need for comprehensive characterization. I will highlight how Benchling provides a modern and fully configurable informatics platform to address complex antibody R&D needs.

Rapid Novor has perfected the art of monoclonal antibody sequencing, and is now ready to demonstrate its ability to sequence mAbs from polyclonal mixtures. Further, they have coupled this technology with the ability to track the relative abundance of these antibodies over time using NovorIg™ to profile the immune response.

Antibody molecules circulating in blood and neutralizing different pathogens is a key element of the immune system. Combining high-resolution proteomics with Next-Gen sequencing of B cell receptor transcripts, we have developed a quantitative approach for delineating and characterizing the antigen-specific antibody repertoire. Here, I will present insights gained from characterizing serological repertoires and highlight recent work demonstrating how this knowledge can guide the engineering efforts for next-generation therapeutics and vaccines.

A high-throughput, automated platform for production of large molecules makes it feasible to generate purified, high-quality material for screening potential therapeutic candidates in assays. We have developed a highly productive, preparative HPLC workflow to rapidly purify complex-engineered, antibody-like molecules for use in assays designed to evaluate novel large-molecule formats for desired functionality, thereby overcoming some of the limitations of traditional purification techniques.