While many believe that antibody-mediated immunity is well understood, a variety of observations in recent decades implies the existence of major unresolved fundamental questions regarding the relationship between antibody structure and function. The observation that the constant region can affect specificity and affinity suggests an explanation for variable region restriction in antibody responses based on avoiding autoimmunity.  Antibodies have now been shown to modulate microbial function and to digest microbial antigens through catalytic activity. These findings suggest new ways to harness antibody-mediated immunity in the design of therapeutic immunoglobulins and protective vaccines.

Defining the exact sequences presented by human leudocyte antigen (HLA) Class II molecules is essential for understanding the immunogenicity potential of biotherapeutics. HLA heterozygosity complicates efforts to define both the precise sequences presented by various HLA alleles and the percentage of patients with potential to mount anti-drug responses that can negate clinical benefit and/or result in adverse events. This presentation will describe a diverse, robust, and reproducible monoallelic HLA-DRB1 system in professional antigen presenting cells capable of examining the immunogenic potential of any human IgG. A case study with adalimumab will be discussed.

Directed evolution is a powerful tool to optimize protein fitness for a specific application, but it can be inefficient when mutations exhibit non-additive, or epistatic, behavior. Machine learning (ML) methods have the potential to address this limitation by learning from assay-labeled data and suggesting ideal combinations of mutations to navigate the protein fitness landscape more efficiently. Here, I present frameworks for iterative, ML-assisted protein optimization: Active Learning-assisted Directed Evolution (ALDE) and Steered Generation for Protein Optimization (SGPO). Besides showing strong in silico performance, three rounds of wet-lab experimentation with ALDE enabled rapid optimization of five epistatic residues in the active site of an enzyme, yielding an ideal variant with a non-obvious (non-additive) combination of mutants. Overall, these ML-assisted methods are practical and broadly applicable strategies to unlock improved protein engineering outcomes.

Over the last decades of research on protein evolution and engineering, we have recognized that the success of protein engineering campaigns is overly depending on the systems. What are molecular determinants to dictate the evolution of new protein functions? How can we choose the right strategies to engineer new proteins?I will discuss key molecular properties that can be associated with evolvability of proteins, the ability of proteins to promptly evolve a new function. Especially, I will discuss the causes and consequences of mutational epistasis, interactions between mutational effects that affect the pathways and outcomes of evolution.

A central challenge is to predict the functional properties of a protein from its sequence, and thus discover new proteins with specific functionality. Experimental and computational data enable machine learning models that predict protein function directly from sequence to be trained and validated. However, the cost and latency of wet-lab experiments require methods that find good sequences in few experimental rounds, where each round contains many sequence designs. I will discuss model-based optimization approaches that find diverse sequence candidates for experimental evaluation, illustrated by case studies demonstrating the design and experimental validation of proteins for therapeutic applications.

The first naturally occurring cyclotide, a circularized disulfide-rich protein, was discovered as the active ingredient of a folk medicine brewed as a tea. These small cyclic proteins exhibit remarkable oral bioavailability. Combining this property with antibody-like recognition of arbitrary targets would enable new oral therapeutic modalities for interrupting inflammatory cytokine cascades. We will present our progress in designing pre-immune cyclotide repertoires, minimizing polyspecificity, and accelerating lead optimization.

Metastatic solid tumors represent a major unmet need, underscoring the limitations of therapies that focus solely on cancer cells. At Matrisome Bio, we are targeting the disease-associated extracellular matrix or ECM within tumors and metastases, which offers a compelling new therapeutic avenue. This talk will highlight the discovery of high-affinity nanobodies (via phage-display) against ECM proteins and their subsequent engineering as radioisotope carriers, enabling the precise delivery of highly differentiated radioligand therapies.

The design of multispecific protein therapeutics presents unique challenges that remain largely unaddressed by current computational approaches. We discuss critical data gaps in this field and present strategic approaches for generating fit-for-purpose datasets specifically tailored for multispecifics. Through practical examples and case studies, we demonstrate how targeted computational and machine-learning strategies can support the optimization of next-generation multispecific therapeutics.

Bicycle’s proprietary phage display-screening platform uses synthetic biology to rapidly screen large libraries of chemically-stabilized bi-cyclic peptides against a wide spectrum of targets, including those for which conventional small molecules and biologics have not proven optimal. High-quality hits are chemically optimized using structural and computation tools into leads with drug-like properties. Conjugation of payloads including toxins, radioisotopes, and oligonucleotides has created therapeutic modalities, some of which have progressed to the clinic.

Caplacizumab, a revolutionary anti-von Willebrand Factor NANOBODY compound, emerged from two decades of innovation at Sanofi. This presentation traces the entire story from discovery through clinical development to commercial success, chronicling the complete development journey from llama immunization in 2003 to first-in-class therapeutic approval, starting with EMA (2018) and FDA (2019). Clinical development demonstrated the successful translation of this innovative biologic into an effective therapy for acquired thrombotic thrombocytopenic purpura (aTTP). As the first approved NANOBODY therapeutic, caplacizumab's mechanism of targeting pathophysiological platelet adhesion has established a new treatment paradigm for this rare blood disorder.