Generative protein design is rapidly advancing the creation of de novo monospecific antibodies with drug-like properties. The next frontier is designing more complex therapeutic molecules, particularly multispecific antibodies. This talk will showcase examples of zero-shot generation and computational optimization of multispecific antibodies using generative models, and discuss the modeling, design, and experimental advances needed to make multispecific protein design as reliable and scalable as current monospecific approaches.

Despite rapid advances in AI-based protein design, predicting the function of multispecific antibodies remains a major challenge. This presentation will describe Aureka's closed-loop discovery platform, integrating AI-guided design with high-throughput single-cell functional screening to generate rich biological datasets that continuously improve predictive models. By learning the design principles underlying functional activity rather than binding alone, this approach aims to accelerate the discovery of best-in-class multispecific therapeutics.

In this work, multispecific antibodies in a variety of formats are produced using intein-mediated protein splicing, thus enabling the high-throughput production of combinatorial libraries of complex antibody modalities. We show that conventionally produced and intein-spliced bispecific molecules exhibit indistinguishable biochemical and functional properties. Dual splicing reactions facilitate the generation of bivalent molecules, and building off the DuetMab platform enables the creation of trispecific formats with unexpected functional consequences.

Artificial intelligence is expanding opportunities to design and optimize VHH therapeutics with increasingly complex functions. This presentation will explore how AI and computational approaches can support the discovery of mono- and multispecific VHHs, from sequence and structure design to optimization of binding, specificity, and developability. Integrated computational and experimental workflows can accelerate candidate selection and enable the design of next-generation multispecific biologics.

We present an integrated nanobody discovery and engineering platform that combines serum proteomics, repertoire sequencing, AI-enabled structure prediction, and modular multivalent design. The workflow rapidly identifies high-affinity binders, maps complementary epitopes, and assembles compact multispecific biologics with exceptional potency, breadth, and developability. Using Staphylococcus aureus toxins as a model, we demonstrate how this platform can generate scalable countermeasures against antigenic diversity and support rapid translation across infectious disease and oncology.

Bispecific antibody engagers combining an immunoglobulin kappa light chain-binding nanobody with CTLA-4- or PD-L1-targeting nanobodies demonstrated antitumour activity through recruitment of polyclonal immunoglobulins. The CTLA-4-targeting conjugate showed superior activity to a conventional antibody and reduced intratumoural regulatory T cells. PD-L1-targeting activity was further enhanced through conjugation with maytansine or a STING agonist, highlighting opportunities to improve immune checkpoint blockade through nanobody-based multispecific engineering.

This talk will outline engineering of a p40/TL1A bispecific antibody for IBD, designed to improve upon clinically validated monotherapies by targeting complementary IL-12/IL-23 and TL1A/DR3 inflammatory pathways. We will describe anti-TL1A domain optimization to minimize developability risks and predicted T cell epitopes, followed by systematic assessment of multiple bispecific formats for manufacturability, molecular quality, pharmacokinetics, and in vivo performance. Integration of these data guided selection of PF-07261271 (RO7837195) for clinical development.

Trispecific T cell engagers represent a next-generation approach to overcoming the challenges of solid tumor immunotherapy. This talk will examine emerging strategies for integrating tumor targeting, T cell engagement, and immune modulation within a single therapeutic format, with a focus on translational opportunities and key considerations for advancing trispecifics toward differentiated therapies for solid tumors.

Interleukin-10 possesses unique immunomodulatory activities with significant therapeutic potential, yet its clinical application has historically been limited by systemic toxicity and an unfavorable therapeutic window. This presentation will describe an engineered IL-10 mutant platform combined with antibody-directed delivery to selectively activate IL-10 biology at disease sites, illustrated by the PD-1×IL-10M immunocytokine FP008 for oncology and the TL1A×IL-10MM fusion protein FP012 for inflammatory diseases, highlighting a general strategy for designing safer and more effective multispecifics.

Next-generation bispecific ADCs overcome tumor heterogeneity and resistance by targeting dual antigens to accelerate endocytosis and combining dual-mechanism payloads. Utilizing OBI’s GlycOBIDUO site-specific conjugation and hydrophilic HYPrOBI linkers, the platform enables High DAR delivery without aggregation. Ultimately, enhanced linker stability minimizes off-target toxicity and expands the therapeutic window, offering a breakthrough for low-antigen-expressing tumors.

We describe a novel engineered multivalent multispecific therapeutic approach (SYNTHBODY) that enables targeting of combinatorial “synthetic” cancer targets. The SYNTHBODY ADC approach leads to log-order greater internalization and in vitro and in vivo potency, when compared with an approved benchmark or bispecific ADC approaches with component binders. We highlight the efficient discovery, layered logic-gated architecture, optimized developability, pharmacokinetics, and in vivo efficacy using a few example SYNTHBODY constructs.