Abstract Background: LGR5 is a validated therapeutic target highly expressed in cancer stem cells and solid tumors. Petosemtamab, an EGFR/LGR5 bispecific antibody, confirms its druggability. Single-domain antibodies (VHHs) provide advantages over conventional antibodies through small size, high tissue penetration, and scalable manufacturing. We aimed to create a panel of VHH binders against LGR5 with broad epitope diversity and favorable biophysical traits for therapeutic use. Methods: VHHs were designed using AI-based structure and sequence-guided modeling targeting recombinant human LGR5 extracellular domain. Chain-of-thought optimization was applied to improve multiple parameters simultaneously, including affinity, thermal stability (Tm), expression yield, and hydrophobic interaction chromatography (HIC) behavior. Top candidates were evaluated for kinetics via surface plasmon resonance, epitope mapping by cryo-EM using N-glycan knockout variants, thermal stability by differential scanning fluorimetry, and expression in mammalian systems. Specificity was tested against related receptors LGR4 and LGR6. Results: Multiple VHH clones showed high-affinity binding to LGR5, with KD values in the nanomolar range. Cryo-EM mapping revealed distinct, non-overlapping epitopes across the LGR5 extracellular domain, providing full structural coverage of potential binding sites. The panel included both LGR5-selective and pan-LGR binders, offering flexibility in therapeutic design. R-spondin competition assays identified ligand-blocking and non-competing VHHs targeting alternate epitopes. All lead molecules displayed strong stability (Tm 65°C) and robust expression, supporting high manufacturability. These characteristics enable both pathway inhibition and targeted payload delivery approaches to eliminate LGR5+ cancer stem cells. Conclusions: AI-driven protein engineering produced a diverse, high-quality panel of LGR5-targeted VHHs with multi-epitope coverage, broad affinity range, and strong manufacturability. Chain-of-thought optimization improved multiple parameters in parallel, accelerating VHH development. This library enables rational design of bispecific and multispecific therapeutics for enhanced tumor targeting and represents a foundation for next-generation LGR5-directed ADCs, bispecifics, and CAR-T therapies. Citation Format: Chenrui Xu, Yifan Li, Trang Nguyen, Roger Shek, Longfei Chong, Tek H. Lee, Yuxiang Lang, Li Yi, Per Greisen. Diverse VHH antibodies targeting multiple LGR5 epitopes for cancer therapy abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4404.
Xu et al. (Fri,) studied this question.