ABT-863 is a humanized VHH-Fc inverse agonist antibody that targets CCR8+ tumor-infiltrating Tregs. Discovered using Abilita’s EMP platform, it binds the receptor’s orthosteric pocket and inhibits both ligand-induced and basal signaling. In preclinical models, ABT-863 combined with anti-PD1 mediates tumor suppression independent of Fc-effector function. This mechanism may enable effective Treg modulation without systemic depletion, allowing a differentiated therapeutic approach with potential for improved safety and efficacy in cancer immunotherapy.

We have advanced multivalent miniprotein engineering platforms to achieve selective control of target engagement in a size-efficient manner. We will present the effects of different molecular formats for multivalent binding, including the impact of paratope linkages and monovalent affinity on resultant selectivity and utility across multiple applications.

Our Biologics Engineering and Discovery Sciences groups collaborate to characterize biologic–antigen complexes using mass photometry, mass spectrometry, and cryo-EM. These complementary structural biology tools reveal molecular-level insights that inform target selection, lead identification, and optimization. The resulting data also guide in-silico design efforts, improving selectivity and accelerating biologic drug discovery.

T cell engagers (TCEs) have shown promising clinical efficacy, but are still associated with significant toxicities, such as cytokine release syndrome (CRS). To address this, we have developed novel affinity attenuated and CD8-biased TCEs. A case study will be presented on the discovery and engineering of clinical lead candidates for CLDN18.2.

Targeting GPCRs for disease therapy has proven problematic with many drugs exhibiting off target side effects due to lack of receptor selectivity. We have targeted the cannabinoid GPCR CB2R and the glucagon-like 1-peptide receptor with libraries of stabilised and constrained peptides, called Selektides. We describe here our work on discovery and functional characterisation of candidate receptor agonists and antagonists with an emphasis on functional experimental design and receptor selectivity determination.

CXCR4 is a G-protein coupled receptor for the chemokine ligand CXCL12 and an established target for cancer and HIV. Here, we used cryoelectron microscopy (cryoEM) to capture CXCR4 in various states, including its complexes with G_i heterotrimer, CXCL12, the FDA-approved antagonist AMD3100, and monoclonal antibody REGN7663. We also determined the structures of homotrimeric and homotetrameric forms of CXCR4, which represent unique modes of GPCR oligomerisation.

Complex membrane proteins (CMPs) are major drug targets but difficult to study in native form. Virus-like particles (VLPs) enable CMP display in a biologically relevant context, yet characterization remains challenging. We present a flow cytometry–based assay, flow virometry (FV), to analyze VLPs displaying fluorescently labeled CMPs, enabling routine assessment of particle size, fluorescence, and surface expression as key quality indicators for biologics discovery.

CXCR4 is a chemokine receptor upregulated in various cancers, yet developing therapeutic antibodies against it has proven challenging. The best-in-class antibody, Ulocuplumab, was discontinued in clinical development. To develop next generation CXCR4 antibodies, we isolated wild-type oligomeric CXCR4 antigen and generated a panel of novel CXCR4 antibodies which exhibit higher affinity and improved antagonistic properties compared to Ulocuplumab.

• Antigen design: antigen strategies and engineering for optimal target presentation
• Antibody sources: pros and cons of naive and immune libraries, and the benefits of using divergent species for targets that are highly conserved
• Screening: best approaches to identify diverse antibody panels, and merits of B cell cloning vs display technologies
• Specificity: how to assess off-target binding​

• Customising antigens in formats to fit your discovery platforms and concerns
• How close are your targets to what is in the real world
• Examples on how to reduce the gaps with real world target presentation
• How to validate orphan targets with tools​

We present a generative protein design platform that enables fully-computational design of single and paired chain antibodies with therapeutic-grade properties. This system generates antibodies that achieve picomolar affinities, strong early-stage developability profiles, and precise targeting of functional epitopes without experimental optimisation. We demonstrate capabilities across multiple therapeutic contexts, including the first fully computationally-designed antibody agonists to a multipass membrane protein.

While cell painting is often used for drug screening, application to large-scale antibody data remains underexplored. We conducted a screen of thousands of combinatorial antibody treatments across multiple cell lines, then developed a data-driven computational pipeline to classify phenotypic responses. By analysing dual-target interventions, the pipeline enables stratification into phenotypic subclasses and direct comparisons between different therapeutic formats. This resource provides interpretable results suited for complex multifactorial phenotypic screening data.

This presentation details a case study on the discovery and optimisation of antagonistic antibodies targeting a challenging GPCR. We will present multiple approaches employed to enhance affinity, efficacy, and developability, including a comparison of degenerate codon libraries informed by deep learning models, structure data, and ddG calculations. We also highlight an example of modality optimisation through structure-guided ligand conjugation resulting in improved antibody functionality.

Targeting GPCRs with efficacious agonist antibodies is a significant challenge. Our function-first generative design platform employs AI/ML, initially proven targeting the GLP1 biology and now engineered for broader GPCR applications. ML models learn from live-cell functional data to guide the generative design of antibodies with desired activation profiles, allowing fine-tuning of potency, bias, and developability. This platform provides a robust engine to engineer functional antibodies for diverse, intractable GPCR targets.