Designed endocytosis-inducing proteins degrade targets and amplify signals

The study addresses how to trigger endocytosis and lysosomal trafficking of cell-surface receptors using bio-orthogonal, genetically encodable proteins to overcome limitations of native-ligand-based systems (e.g., LYTACs, KineTACs), which can compete with endogenous ligands, trigger off-target signalling, and require complex chemical modifications. The authors propose de novo designed endocytosis-triggering binding proteins (EndoTags) tailored to specific receptor mechanisms—orthogonal binding to constitutively cycling receptors (sortilin, transferrin receptor), induced conformational changes for IGF2R, and receptor clustering for ASGPR—to drive internalization without competing with natural ligands. They aim to demonstrate tissue-specific targeted degradation by leveraging receptors with distinct expression patterns and to show additional utilities such as enhanced signalling and local secretion.

Prior approaches to extracellular/membrane protein degradation leverage native endocytosis-inducing ligands: LYTACs using M6P or GalNAc engage IGF2R or ASGPR; KineTACs use cytokine receptors. These strategies can cause off-target signalling, face competition from endogenous ligands (e.g., M6P-tagged enzymes at IGF2R), and rely on multivalent chemical conjugation that complicates manufacturing and precludes genetic encodability. Antibodies that induce endocytosis exist but require extensive empirical screening per target. The literature also documents receptor endocytosis mechanisms (clathrin-mediated uptake, ligand-induced conformational changes or clustering) and tissue distributions of target receptors (sortilin, TfR, ASGPR, IGF2R), motivating receptor-specific designs.

• Design principles: three mechanisms were targeted: (1) orthogonal binding to constitutively cycling receptors (sortilin, TfR) at sites non-overlapping with natural ligands; (2) inducing receptor conformational changes (IGF2R) by bringing together specific extracellular domains; (3) receptor clustering (ASGPR) via multivalency.
• Computational design: Rosetta RIFdock to design minibinders to IGF2R domain 6 (D6) and domain 11 (D11); de novo binder design with Rosetta, ProteinMPNN for sequence optimization, and AlphaFold for structure validation; RFInpainting to build rigid fusions that position D6- and D11-binders.
• Library generation and screening: thousands of designs for ASGPR screened by yeast display and FACS coupled with next-generation sequencing; hits expressed in E. coli and purified.
• Affinity measurements: biolayer interferometry quantified KDs of designed binders (e.g., D6mb ~41 nM; D11mb improved to 6.5 nM; Sort_EndoTag ~21 nM to sortilin; ASmb1 ~2.7 µM to ASGPR).
• Construct engineering: IGF_EndoTags assembled as flexible or rigid fusions of D6mb and D11mb with varied linkers (GGS/GS) and orientation; AS_EndoTags built by tandemly duplicating ASmb1 (2C, 3C) via GS linkers; TfR_EndoTag leveraged a prior orthogonal TfR binder further optimized for solubility; Sort_EndoTag designed to avoid native ligand sites.
• Cell assays: uptake assessed by flow cytometry using AF647-labelled proteins across multiple lines (U-251MG, Jurkat, HeLa, Hep3B, K562). Confocal microscopy assessed lysosomal co-localization (LysoTracker, anti-LAMP2A). Competition/orthogonality tested by co-incubation with transferrin or IGF2; GNPTAB knockout tested M6P competition. Receptor dependency tested via receptor knockouts (IGF2R, TfR, sortilin) and binding-ablating mutations.
• pLYTAC construction: fused EndoTags to EGFR minibinder (EGFRn), cetuximab (CTX), PD-L1 antibody (atezolizumab, ATZ), or CTLA4 minibinder. Western blots quantified target depletion; phospho-flow measured downstream EGF→ERK signalling.
• Proteomics: quantitative mass spectrometry profiled global protein changes upon pLYTAC treatment.
• In vivo efficacy: BALB/c mice bearing A20 tumours received intratumoural injections (5 mg/kg on days 10, 13, 16) of ATZ-EndoTag variants vs ATZ; tumour growth, mass, survival (Mantel–Cox), and body weight monitored; tumour lysates analysed by western blot.
• Soluble protein clearance: heterodimer system (LHDA/LHDB) used to quantify uptake and supernatant depletion; protein G fusions (protein G-IGF_EndoTag3 vs protein G-M6Pn) assessed for IgG uptake/clearance and lysosomal co-localization; supernatant fluorescence quantified on a plate reader.
• Logic gating and secretion: Co-LOCKR system implemented as an AND gate combining EGFR and HER2 recognition with BCL2-IGF_EndoTag2 to degrade EGFR conditionally; secretion tested by transfecting IGF2R-KO HeLa producer cells with EGFRn-IGF_EndoTag constructs, collecting supernatants, and treating EGFR+ K562 target cells.
• Signalling enhancement: ortho-SNIPR system (LHDB receptor with Notch core) stimulated by LHDA fused to IGF_EndoTags; BFP reporter readout quantified activation; pharmacologic inhibitors (chloroquine, γ-secretase inhibitor, protease blockers) probed mechanism; time-lapse confocal microscopy tracked lysosomal co-localization over time.

• Orthogonal endocytosis via sortilin and TfR: • Sort_EndoTag bound sortilin (KD ~21 nM). U-251MG cells showed ~90-fold higher fluorescence vs control after 2 h; strong lysosomal co-localization at 24 h. • TfR_EndoTag yielded ~50-fold higher uptake in U-251MG; did not affect transferrin binding/uptake in HeLa.
• IGF2R conformational triggering: • Designed minibinders: D6mb KD ~41 nM; D11mb improved from ~190 nM to ~6.5 nM (D11mb2). • IGF_EndoTag1 (D11mb–GGS–D6mb) increased Jurkat uptake vs IGF2 or single binders; lysosomal targeting confirmed. Linker length/orientation critical; homodimeric fusions were inactive. • IGF_EndoTag2 (with D11mb2) doubled internalization vs IGF2 and was detectable in lysosomes within 30 min. Rigid fusions (e.g., IGF_EndoTag3) bound both domains (e.g., D6 KD ~6 nM; D11 KD ~190 nM) and internalized comparably to IGF2. • Orthogonality to M6P confirmed: GNPTAB knockout did not alter IGF_EndoTag2 binding; IGF_EndoTag4 did not block IGF2 uptake, while IGF_EndoTag2 partially reduced IGF2 uptake, indicating tunable orthogonality via domain 11 affinity.
• ASGPR clustering: • AS_EndoTag-2C and -3C increased uptake 2.5× and 4.5× over monomer; AS_EndoTag-3C co-localized strongly with lysosomes; induced trimeric ASGPR complexes by SEC.
• pLYTAC-mediated receptor degradation (tissue targeting): • Liver (ASGPR): EGFRn-AS_EndoTag-2C/3C reduced EGFR by ~40% in Hep3B; monomeric ASmb1 had little effect. • Brain/neural (TfR, sortilin): EGFRn-TfR_EndoTag reduced EGFR ~55% in HeLa; EGFRn-Sort_EndoTag reduced EGFR ~78%; receptor knockouts abolished effects; binding-ablating mutations eliminated activity. • Systemic (IGF2R): EGFRn-IGF_EndoTag2 achieved >80% EGFR clearance in H1975 and HeLa; effects were IGF2R-dependent and did not alter IGF2R abundance by proteomics; pre-treatment ablated EGF→ERK signalling. • CTX-IGF_EndoTag1 outperformed M6P-based LYTACs; as low as 10 nM CTX-IGF_EndoTag1 reduced EGFR by ~85%. • PD-L1: ATZ-EndoTag fusions reduced PD-L1 by ~77% in MDA-MB-231 at 48 h; CTLA4mb-EndoTag1 reduced CTLA4 by ~45% in Jurkat-CTLA4 cells (3 h).
• In vivo efficacy: • In BALB/c A20 tumours, ATZ-IGF_EndoTag3/4 reduced tumour size and mass vs ATZ; ATZ-IGF_EndoTag4 significantly improved survival (50% alive at day 55 vs 0% with ATZ); treatments were well tolerated by body weight.
• Soluble target clearance: • LHDA-IGF_EndoTag3 increased LHDB uptake up to ~40-fold MFI and cleared ~50% of 100 nM LHDB in 48 h (Jurkat). • Protein G-IGF_EndoTag3 boosted IgG uptake ~80-fold (K562) and ~360-fold (Jurkat) vs protein G alone; cleared ~70% of 133 nM IgG in 48 h; lysosomal co-localization and IGF2R dependence confirmed.
• Logic-gated and secreted degraders: • Co-LOCKR AND gate with BCL2-IGF_EndoTag2 induced ~80% EGFR degradation only in EGFR+HER2+ cells; no effect in EGFR+HER2− cells. • Supernatants from producer cells secreting EGFRn-IGF_EndoTag1/2 were as effective as purified proteins in degrading EGFR on target cells.
• Signalling amplification: • In the ortho-SNIPR system, IGF_EndoTag fusions to ligand enhanced signalling up to ~100-fold (IGF_EndoTag2); enhancement required endosomal acidification (blocked by chloroquine) and γ-secretase activity; rapid lysosomal co-localization observed.

The results demonstrate that rationally designed, bio-orthogonal EndoTags can trigger receptor endocytosis via three distinct mechanisms—orthogonal binding to constitutively cycling receptors, enforced conformational changes, and receptor clustering—without relying on native ligands. By leveraging receptors with distinct tissue distributions (ASGPR in liver, TfR/sortilin in neural tissues, IGF2R broadly expressed), pLYTACs can be directed to specific tissues. Compared to chemically modified LYTACs, all-protein EndoTags are genetically encodable, modular, and manufacturable, enabling fusions to antibodies or binders, logic-gated targeting (Co-LOCKR), and local secretion from engineered cells. EndoTags not only degrade membrane and soluble targets selectively and efficiently (with receptor dependence and minimal off-target proteomic changes), but also can amplify endocytosis-dependent signalling pathways, as shown with ortho-SNIPR. These capabilities expand therapeutic modalities for targeted degradation, enable integration into adoptive cell therapies, and provide tools to probe how receptor conformation and oligomerization regulate trafficking and signalling.

Designed EndoTags expand targeted degradation and signalling modulation by enabling bio-orthogonal, genetically encodable induction of receptor endocytosis via orthogonal binding, conformational triggering, or clustering. They avoid competition and off-target signalling associated with native ligands and simplify manufacturing versus multivalent chemical conjugations. EndoTag pLYTACs enhance the efficacy of antagonistic antibodies (e.g., anti-PD-L1) in vivo, can be logic-gated and locally secreted, and markedly enhance endocytosis-dependent signalling (e.g., SNIPR). Future work includes expanding to additional endocytosing receptors for finer tissue targeting and trafficking control, assessing and minimizing immunogenicity, engineering catalytic/recycling EndoTags for dose sparing, applying EndoTags to enhance uptake of nucleic acids and small-molecule conjugates, and testing whether similar signalling enhancements can be achieved in natural pathways.

• Immunogenicity of de novo designed EndoTags is unknown and requires assessment and potential deimmunization.
• Clearance efficiency can be limited by receptor availability and expression levels on target cells, potentially constraining maximal degradation of soluble targets.
• Most efficacy data are from in vitro cell lines and an intratumoural mouse model; broader in vivo pharmacokinetics, biodistribution, and systemic safety remain to be established.
• While orthogonality to native ligands was demonstrated (e.g., M6P), some IGF_EndoTag variants partially affect IGF2 uptake depending on domain 11 affinity, indicating a trade-off between potency and orthogonality.
• Generalizability to endogenous signalling pathways beyond the synthetic SNIPR system remains to be determined.
• Potential need for recycling/catalytic EndoTags to reduce dosing and avoid receptor downregulation over time.