Reshaping Undruggable Membrane-Obligate Targets: AI Antigen Design to Drive Antibody Discovery and Optimization

Membrane proteins such as G protein-coupled receptors (GPCRs) and ion channels remain among the most challenging therapeutic discovery targets, primarily due to the lack of soluble, native-like reagents for immunization and screening with display-based discovery technologies. To address this, Adimab scientists developed a machine learning (ML)-guided membrane protein solubilization pipeline to convert wild-type membrane-obligate targets (MOTs) into defined, native-like soluble antigens. Two case studies show the resulting reagents applied to antibody discovery against an ion channel and to optimization of a class-A GPCR agonist.

Approach and outcomes

• Starting from an experimental or predicted 3D structure of the wild-type protein, the pipeline uses ML-guided sequence optimization to generate solubilized MOT antigens, which are then screened in yeast and expressed in mammalian cells and further assessed for quality
• In the first case study targeting an ion channel protein,
  • This approach increased binder recovery to 72%, compared to 7% using the standard method, a statistically significant improvement in target-specific recovery.
  • covered distinct epitope bins as determined by Octet kinetics and cell binding, and demonstrated favorable developability profiles as assessed by hydrophobicity retention time and polyspecificity reagent (PSR) score.
• In the second case study, a solubilized antigen was used to optimize an existing IgG agonist lead targeting the class-A GPCR CCR5.
  • In earlier work, an affinity maturation library of 10⁸ diversity was selected against CCR5 nanodiscs. Its progeny antibodies reached up to 85% Gz agonist efficacy relative to the natural ligand CCL5, compared with 57.5% for the parental antibody.
  • Building on this, a deep mutational scanning (DMS) library of 10³ diversity was selected in parallel against nanodisc and solubilized CCR5 antigens.
  • Both formats produced affinity-improved outputs and pointed to the same key amino acid positions across the heavy- and light-chain CDRs. The outputs kept favorable cell binding and PSR profiles.

Why it matters

Solubilizing membrane-obligate targets removes the central reagent bottleneck that has kept ion channels and GPCRs out of reach for display-based and immunization-driven antibody campaigns. Adimab’s ML-guided solubilization pipeline opens these targets to the full range of antibody discovery and optimization workflows.

View poster