Back

Multivalent Ligand-Protein Interactions Using Polymeric Lysosome-Targeting Chimeras (PolyTACs) Leads to Lysosome-Targeting Receptor-Independent Degradation of Transmembrane Proteins

Dutta, R.; Alp, Y.; Gupta, P.; Singh, B.; Thayumanavan, S.

2025-05-11 bioengineering
10.1101/2025.05.06.652519 bioRxiv
Show abstract

Targeted protein degradation is growing rapidly as a therapeutic approach, with intracellular proteins degraded via the ubiquitin-proteasome or the autophagosome system and membrane proteins mainly through the lysosomal pathway. Current lysosomal degradation strategies rely on lysosome-targeting receptors (LTRs), limiting their applicability. We propose that multivalent non-covalent interactions on the cell membrane can drive lysosomal degradation of membrane proteins without the need of LTRs. To demonstrate this, we designed antibody-polymer conjugates, viz. Polymeric Lysosome-Targeting Chimeras (PolyTACs) functionalized with ligands that would non-covalently bind with transmembrane non-LTR proteins, viz., helper proteins on the cell surface in a polyvalent fashion. Cetuximab-based PolyTACs decorated with 4-(2-aminoethyl)benzenesulfonamide (ABS) ligands that cause multivalent interactions with membrane carbonic anhydrases induced degradation of EGFR, while atezolizumab- and trastuzumab-based PolyTACs effectively degraded PD-L1 and HER2, respectively. Additionally, PolyTACs using PD-L1 as the helper protein further improved degradation. Mechanistic studies confirmed clathrin- and caveolae-mediated endocytosis followed by lysosomal degradation of the target proteins. This LTR-independent nature of the approach offers opportunities for tissues targeting in membrane protein degradation that could open up new avenues in therapeutic strategies.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.