CYpHER: catalytic extracellular targeted protein degradation with high potency and durable effect

Abstract

Many disease-causing proteins have multiple pathogenic mechanisms, and conventional inhibitors struggle to reliably disrupt more than one. Targeted protein degradation (TPD) can eliminate the protein, and thus all its functions, by directing a cell's protein turnover machinery towards it. Two established strategies either engage catalytic E3 ligases or drive uptake towards the endolysosomal pathway. Here we describe CYpHER (CatalYtic pH-dependent Endolysosomal delivery with Recycling) technology with potency and durability from a catalytic mechanism that shares the specificity and straightforward modular design of endolysosomal uptake. By bestowing pH-dependent release on the target engager and using the rapid-cycling transferrin receptor as the uptake receptor, CYpHER induces endolysosomal delivery of surface and extracellular targets while re-using drug, potentially yielding increased potency and reduced off-target tissue exposure risks. The TfR-based approach allows targeting to tumors that overexpress this receptor and offers the potential for transport to the CNS. CYpHER function was demonstrated in vitro with EGFR and PD-L1, and in vivo with EGFR in a model of EGFR-driven non-small cell lung cancer.

Fig. 1. Basic principles of CYpHER and component binders.

a CYpHER design including a pH-independent TfR-binding domain and a pH-dependent target-binding domain separated by a linker. b CYpHER mechanism. CYpHER induces ternary complex formation with target and TfR. Upon TfR-mediated uptake and endosomal acidification, target is released for endolysosomal system trafficking. TfR and CYpHER recycle to the surface for engagement with another target molecule. c 293F cells displaying a high-affinity TfR-binding CDP were stained with TfR and rinsed at pH 7.4 or pH 5.5 for 10 min, showing similar binding via flow cytometry in both conditions. Mean fluorescence intensity [MFI] ±95% confidence interval [CI], pH 7.4, 80.1 ± 1.9; pH 5.5, 90.6 ± 2.0; precise N per sample unavailable. d 293F cells displaying medium or high affinity TfR-binding CDPs were stained with human TfR (hTfR) or mouse TfR (mTfR). Kruskal–Wallis test with Dunn’s correction: Med affn hTfR vs High affn hTfR was not significant (P > 0.9999), all others P < 0.0001. N cells per sample: GFP-, 479; Med affn hTfR, 317; Med affn mTfR, 295; High affn hTfR, 266; High affn mTfR, 236. e pH-dependent PD-L1 binding flow profile of 293F cells displaying a pool of histidine-doped variants of a PD-L1-binding CDP after four rounds of flow sorting; two for high binding after pH 7.4 rinse, two for low binding after pH 5.5 rinse. pH 7.4, N = 38,979 cells; pH 5.5, N = 37,450 cells. f Three His substitutions were tested as singletons and combinations for PD-L1-binding on 293F cells displaying a given binder after 10 min pH 7.4 or pH 5.5 rinse followed by flow cytometry quantitation. N cells per sample: Untransfected [UTF] 7.4, 19727; UTF 5.5, 17843; Parental 7.4, 41; Parental 5.5, 46; His sub 1 [HS1], 7.4, 27; HS1 5.5, 17; HS2 7.4, 29; HS2 5.5, 25; HS3 7.4, 58; HS3 5.5, 62; HS1 + 2 7.4, 62; HS1 + 2 5.5, 58; HS1 + 3 7.4, 25; HS1 + 3 5.5, 29; HS2 + 3 7.4, 45; HS2 + 3 5.5, 40; HS1 + 2 + 3 7.4, 87; HS1 + 2 + 3 5.5, 92. Non-Parental samples within a given pH (7.4 or 5.5) vs Parental sample by Kruskal–Wallis test with Dunn’s correction: at pH 7.4, His sub 1 (P > 0.9999), His sub 3 (P = 0.1161), and His sub 1 + 3 (P > 0.9999) were not significant, all others P < 0.0001. At pH 5.5, His sub 1 (P = 0.1666) and His sub 3 (P > 0.9999) were not significant, all others P < 0.0001. Variant with His substitutions 1 and 3 was chosen for further work. Each experiment in c–f was performed once. All box plots (d and f) feature a median (black line), 25^th and 75^th percentiles (box boundaries), and 5^th and 95^th percentiles (whiskers). See the Supplementary Data for full statistical breakdown. Source data are provided as a Source Data file.

Fig. 2. PD-L1 CYpHER design and target depletion in cell pools overexpressing PD-L1-GFP.

a Two designs of PD-L1 CYpHERs, named CT-4212-1 and CT-4212-3, using a high-affinity TfR-binding CDP and a pH-dependent PD-L1-binding CDP. b Illustration of PD-L1-GFP trafficking induced by CYpHER. Pools of 293T cells (c), H1650 cells (d), and MDA-MB-231 cells (e) transduced with lentivirus driving PD-L1-GFP were untreated or incubated with 10 nM CYpHER for 24 h before GFP-channel microscopy (above) and flow cytometry (below) after staining for surface PD-L1. Black contour in flow profiles: cells stained without PD-L1 antibody. Flow cytometry quantitation of normalized surface PD-L1 (f, h, and j) or total PD-L1-GFP (g, i, and k) signal in 293T-PDL1-GFP cells (f and g), H1650-PDL1-GFP cells (h and i), and MDA-MB-231-PDL1-GFP cells (j and k) with or without CYpHER treatment. N cells per sample as follows. c, f, g Untreated, 842; CT-4212-1, 808; CT-4212-3, 598. d, h, i Untreated, 1867; CT-4212-1, 2318; CT-4212-3, 2734. e, j, k Untreated, 823; CT-4212-1, 536; CT-4212-3, 893. For f–k, within each line and assay, significance by Kruskal–Wallis test with Dunn’s correction were all P < 0.0001 except: 293T-PDL1-GFP surface PD-L1, CT-4212-1 vs CT-4212-3 (P = 0.1647); 293T-PDL1-GFP total GFP, CT-4212-1 vs CT-4212-3 (P = 0.0027); H1650-PDL1-GFP surface PD-L1, CT-4212-1 vs CT-4212-3 (P = 0.0908); H1650-PDL1-GFP total GFP, Untreated vs CT-4212-3 (P = 0.0002); MDA-MB-231-PDL1-GFP total GFP, Untreated vs CT-4212-1 (P = 0.3139). Each experiment in c–k was performed once. All box plots (f–k) feature a median (black line), 25^th and 75^th percentiles (box boundaries), and 5^th and 95^th percentiles (whiskers). See the Supplementary Data for full statistical breakdown. Source data are provided as a Source Data file.

Fig. 3. EGFR CYpHER based on VHH nanobody.

a CT-1212-1 design. b CT-1212-1 SDS-PAGE Coomassie stain. NR: non-reduced. R: DTT-reduced. c SE-HPLC of CT-1212-1; right is zoomed. d EGFR-GFP trafficking by CYpHER in 293T-EGFR-GFP cells. e 293T-EGFR-GFP cells treated 24 h with PBS or 10 nM CT-1212-1 before either GFP microscopy (left) or flow cytometry after staining for surface EGFR (right). Black contour: unstained cells. f 293T-EGFR-GFP cells treated with PBS or 10 nM CT-1212-1 for 24 h and flow sorted for viable (DAPI-) cells prior to Western blotting. Full blot in Supplementary Fig. 13. g Same 293T-EGFR-GFP cells and treatment as e, stratified by surface EGFR quintile and showing normalized surface EGFR stain per cell. Two-tailed Kolmogorov–Smirnov test, PBS vs CT-1212-1 pairwise were all P < 0.0001, Quintile 1 vs Quintile 3 was P = 0.0062, all other CT-1212-1 quintile pairwise comparisons were P < 0.0001. h, i 293T-EGFR-GFP cells dosed with PBS or 10 nM CT-1212-1 for 30 min, 4 h, 24 h, or 24 h followed by 24 h without drug (“Withdrawal”) were flow analyzed for surface EGFR (h) or total EGFR-GFP (i) as in e, with EGFR stain (h) or total EGFR-GFP (i) fluorescence per cell shown. N cells per sample in h and i: PBS 30 min [m], 14806; 1212-1 30 m, 12117; PBS 4 hour [h], 15289; 1212-1 4 h, 13400; PBS 24 h, 14449; 1212-1 24 h, 12588; PBS Withdrawal [WD], 14487; 1212-1 WD, 13972. For surface EGFR (h), PBS vs CT-1212-1 pairwise were all P < 0.0001 via two-tailed Kolmogorov–Smirnov [KS] test. CT-1212-1 samples pairwise by Kruskal–Wallis test with Dunn’s correction [KWD]: 30 min vs Withdrawal was not significant (P > 0.9999), all others P < 0.0001. For total EGFR-GFP (i), PBS vs CT-1212-1 pairwise were all P < 0.0001 via two-tailed KS test. All four CT-1212-1 samples by KWD: 24 h vs Withdrawal was P = 0.0008, all others P < 0.0001. j 293T-EGFR-GFP Cells (24 well dish, 500 μL media per well) were treated with 50 μL CellLight Lysosomes-RFP (delivering gene for LAMP1-RFP) for 24 h, after which they were untreated or treated with 10 nM DyLight 755-labeled CT-1212-1 for 1 or 4 h and then imaged on the GFP, RFP, and near IR channels. Arrows indicate location of LAMP1-RFP foci (i.e., lysosomes). Each experiment in e–j was performed once (e, g–j) or twice (f) producing similar results. All box plots (h and i) feature a median (black line), 25^th and 75^th percentiles (box boundaries), and 5^th and 95^th percentiles (whiskers). See the Supplementary Data for full statistical breakdown. Source data are provided as a Source Data file.

Fig. 4. Performance comparison of different EGFR CYpHER designs.

a A549, H1975, H1650, and H358 cells were flow analyzed alongside calibration beads to quantitate surface EGFR and TfR protein levels. b Normalized surface EGFR levels in the four lines after 1 or 24 h treatment with 10 nM CT-1212-1 or cetuximab, showing surface EGFR stain per cell by flow. N cells per sample as follows. A549: Untreated, 17129; CT-1212-1 1 h, 17641; CT-1212-1 24 h, 19326; Cetuximab 1 h, 16815; Cetuximab 24 h, 18083. H1975: Untreated, 5544; CT-1212-1 1 h, 6091; CT-1212-1 24 h, 6593; Cetuximab 1 h, 4847; Cetuximab 24 h, 5400. H1650: Untreated, 4789; CT-1212-1 1 h, 5258; CT-1212-1 24 h, 5460; Cetuximab 1 h, 5548; Cetuximab 24 h, 5541. H358: Untreated, 7846; CT-1212-1 1 h, 8239; CT-1212-1 24 h, 7834; Cetuximab 1 h, 8120; Cetuximab 24 h, 8409. Within each of the four cell lines (A549, H1975, H1650, H358), all five samples were compared to one another by Kruskal–Wallis test with Dunn’s correction [KWD]; all comparisons for all cell lines were P < 0.0001. c A549 cells incubated with 10 nM CT-1212-1 or cetuximab for 2 h or for 2 h followed by 24 h without drug (“Withdrawal”) followed by staining for human IgG to quantitate surface drug levels, showing surface IgG stain per cell by flow. N cells per sample: CT-1212-1 2 h, 6336; Cetuximab 2 h, 6126; CT-1212-1 Withdrawal [WD], 7335; Cetuximab WD, 7187. All vs all by KWD; CT-1212-1 2 h vs Cetuximab 2 h was P = 0.0001, all others P < 0.0001. Furthermore, all samples were compared via one sample two-tailed T test to 0; Cetuximab Withdrawal was P = 0.37, all others P < 0.0001. d EGFR CYpHER designs. e Surface EGFR levels in A549 cells incubated with 10 nM CYpHER for 24 h or for 24 h followed by 24 h without CYpHER (“Withdrawal”), showing surface EGFR stain per cell by flow. N cells per sample – see Source Data for details. All treated samples vs Untreated by KWD were P < 0.0001. All 24 h vs Withdrawal samples compared by Kolmogorov–Smirnov [KS] test were P < 0.0001. f Same treatment as e but staining for human Fc to quantitate surface CYpHER levels, showing surface Fc stain per cell by flow. N cells per sample – see Source Data for details. All treated samples vs Untreated by KWD were P < 0.0001. All 24 h vs Withdrawal samples compared by Kolmogorov–Smirnov [KS] test were P < 0.0001. g A549, H1975, H1650, and H358 cells untreated or treated with 10 nM CT-1212-1 for 1 h, 1 day, 2 days, 3 days, or 1 day followed by 1 day without drug (“Withdrawal”) then analyzed by flow cytometry to see surface EGFR levels per cell. N cells per sample – see Source Data for details. All samples vs all within a cell line by KWD: A549, 1 h vs 1 day P = 0.0019, 1 h vs 2 day P = 0.9911, all others P < 0.0001; H1975, 1 day vs 2 day P = 0.8709, 1 day vs 3 day P = 0.0004, 2 day vs 3 day P = 0.3246, all others P < 0.0001; H1650, 1 h vs 3 day P = 0.0188, 1 day vs 2 day, P = 0.0069, all others P < 0.0001; H358, 1 day vs 2 day P > 0.9999, all others P < 0.0001. h A549 cells treated for 24 hr with 2 nM, 10 nM, 50 nM, or 200 nM CYpHER then analyzed by flow cytometry for surface EGFR levels to show surface EGFR levels per cell by flow. N cells per sample – see Source Data for details. Experiments in b, c, e, and f were performed once. All box plots (b, c, e–g) feature a median (black line), 25^th and 75^th percentiles (box boundaries), and 5^th and 95^th percentiles (whiskers). See the Supplementary Data for full statistical breakdown. Source data are provided as a Source Data file.

Fig. 5. EGFR trafficking upon CYpHER treatment.

a Designs of various EGFR CYpHERs and control molecules. b A549-EGFR-GFP (knockin) cells treated with PBS or 10 nM CYpHER for 48 h with drug and imaged for GFP localization. c A549 cells treated with 2 or 10 nM CYpHER for 24 hrs, or 10 nM CYpHER for 24 h followed by 24 h drug withdrawal, and assayed by flow cytometry for surface EGFR per cell. N cells per sample as follows: Untreated, 3598. CT-1212-1: 2 nM, 2979; 10 nM, 3288; Withdrawal [WD], 2795. CT-1112-1: 2 nM, 3527; 10 nM, 3286; WD, 3135. CT-1211-1: 2 nM, 2339; 10 nM, 3018; WD, 2250. CT-1111-1: 2 nM, 3412; 10 nM, 3532; WD, 2516. All-vs-all within treatments (including Untreated) by Kruskal–Wallis [KW] test: CT-1212-1, 2 nM vs 10 nM P = 0.1494, all others P < 0.0001; CT-1112-1, 2 vs 10 nM P = 0.0031, all others P < 0.0001; CT-1211-1; 2 vs 10 nM P = 0.0002, all others P < 0.0001; CT-1111-1, all P < 0.0001. d A549-EGFR-GFP cells imaged for GFP localization after 24 h treatment with 10 nM CYpHER (CT-1212-1) or control molecule (CT-1232-1 or CT-3212-1). e A549-EGFR-GFP cells imaged for GFP localization after 20 min treatment with PBS or 10 nM CT-1212-1. f A549-EGFR-GFP cells treated without or with 10 μM human holoTF for 15 min followed by addition of PBS or 10 nM CT-1212-1 for 4 h and imaged for GFP localization. g Same experimental design as in f except altered amount of holoTF and analyzed by flow cytometry for surface EGFR. Dashed lines indicate quantitation of surface EGFR in untreated cells (upper) or cells treated with 10 nM CT-1212-1 but no holoTF (lower). Each treatment vs No CYpHER (100% line) by Kruskal–Wallis test with Dunn’s correction [KWD] was P < 0.0001. Each CYpHER + holoTF treatment vs CYpHER + no holoTF (“No holoTF” line at ~21%) by KWD: 100 nM holoTF P = 0.9831, all others P < 0.0001. Experiments in b–g were performed once (c, f, g) or twice (b, d, e) with similar results. All box plots (c) feature a median (black line), 25^th and 75^th percentiles (box boundaries), and 5^th and 95^th percentiles (whiskers). See the Supplementary Data for full statistical breakdown. Source data are provided as a Source Data file.

Fig. 6. Catalytic soluble cargo uptake by CYpHER.

a Experimental design to quantitate fluorescent soluble cargo uptake in cells pre-treated with cargo-saturated CYpHER. Step 1: CYpHER is saturated with target (2:1 target:binding moiety ratio), then applied to cells for 2 h. Step 2: After cells are thoroughly rinsed, new fluorescently-labeled soluble target is added to cells. Fluorescence accumulation is increased by CYpHER pre-treatment. b Designs and elements of CYpHERs used. c Soluble EGFRvIII uptake per cell after 24 h incubation with A549, H1650, H1975, or H358 cells either untreated (to quantitate passive uptake) or pre-treated for 2 h with unlabeled-EGFR-saturated 10 nM CT-1212-1, normalized to each cell line’s untreated uptake levels. N cells per sample as follows. H358: No CYpHER, 6309; CT-1212-1 Pre-treat, 5901. H1650: No CYpHER, 6622; CT-1212-1 Pre-treat, 6567. H1975: No CYpHER, 6087; CT-1212-1 Pre-treat, 5858. A549: No CYpHER, 7584; CT-1212-1 Pre-treat, 7848. Each line (H358, H1650, H1975, A549), pre-treatment vs Untreated via two-tailed Kolmogorov–Smirnov [KS] test were P < 0.0001. Pre-treatments all-vs-all by Kruskal–Wallis test with Dunn’s correction [KWD] were P < 0.0001. d Soluble EGFRvIII uptake per cell in H1975 cells as in c comparing CT-1212-1, CT-6212-1, and CT-5212-3. N cells per sample: CT-1212-1 Pre-treat, 5778; CT-6212-1 Pre-treat, 5277; CT-5212-3 Pre-treat, 5540. All-vs-all by KWD: CT-6212-1 Pre-treat vs CT-5212-3 Pre-treat P > 0.9999, all others P < 0.0001. e Soluble PD-L1 uptake per cell as in c except with soluble PD-L1 as cargo, comparing CT-4212-1 and CT-4212-3 in MDA-MB-231 (left) or H1650 (right) cells. N cells per sample as follows. MDA-MB-231: No CYpHER, 22023; CT-4212-1 Pre-treat, 6667; CT-4212-3 Pre-treat, 7398. H1650: No CYpHER, 20807; CT-4212-1 Pre-treat, 7546; CT-4212-3 Pre-treat, 7215. Within each cell line, all-vs-all KWD were P < 0.0001. Each experiment in c–e was performed once. All box plots (c–e) feature a median (black line), 25^th and 75^th percentiles (box boundaries), and 5^th and 95^th percentiles (whiskers). See the Supplementary Data for full statistical breakdown. Source data are provided as a Source Data file.

Fig. 7. Pharmacodynamic effects of CYpHER.

a Designs of various EGFR CYpHERs and control molecules. b and c A549 cells (unsorted, thus including live and dead cells) treated for 24 h with PBS, 10 nM CYpHER, or 10 nM control molecule followed by no treatment (“–”, EGFR, Actin) or addition of 50 ng/mL EGF for 30 min (“+”) and analyzed by Western blot for pY1068 EGFR (“–” and “+”), total EGFR, or actin. Full blots in Supplementary Figs. 14 and 15. d–h 96 well plate growth for 4 days (A431) or 7 days (all others) with single dose (no media exchange) of CT-1212-1, cetuximab, gefitinib, or osimertinib in A431 (d), H1975 (e), H1650 (f), A549 (g), and H358 (h) cells. After treatment, cell levels per well were quantitated by CellTiter-Glo 2.0 [CTG] assay. N = 3 wells per condition. See panel i for EC50 values and P values vs CT-1212-1; see Supplementary Data for details on statistical analysis. i EC50 values of the experiments in d–h from asymmetric sigmoidal (5PL) curve fit. Empty “X” box indicates no effect, as defined by failure to suppress growth by 20% at any dose tested. Experiments in b–h were conducted once (b, c, h) or twice (d–g) with similar results. See the Supplementary Data for full statistical breakdown. Source data are provided as a Source Data file.

Fig. 8. Pharmacokinetics and pharmacodynamics of CYpHER in mice.

a Designs of CYpHER and control molecules. b NCr nu/nu mice were dosed with 1.5 mg/kg CT-1212-1, CT-1211-1, CT-1222-1, or CT-1232-1 IV. Serum was taken after 10 min, 30 min, 1 h, 2 h, 4 h, 8 h, 24 h, 48 h, 96 h, or 168 h. Three mice per time point were analyzed. Samples were quantitated by ELISA for human Fc domain in technical triplicate. N = 3 mice per time point. Molecules exhibited a normal biphasic distribution curve, and as such, PK parameters were determined by non-compartmental analysis for IV bolus dosing using PKSolver 2.0. c Experimental design for tumor implantation and dosing. Female athymic nude mice (Foxn1^nu) were implanted (subcutaneous flank) with 5 ×10⁶ H1975 cells. After 21 days, mice were enrolled and dosed IV on days 0 (enrollment day), 3, and 7. On day 8, tumors from three mice per dosage group were harvested and split for Westerns or for histology. d Western blot of total EGFR and actin. N = 3 mice per condition, all seen in blot. Full blots in Supplementary Fig. 16. e Quantitation of the blots from d. f IHC (hematoxylin/DAB) for total EGFR (top) and Ki67 (bottom) in vehicle, CT-1212-1 450 μg, and CT-1222-1 150 μg groups. Full EGFR fields can be found in Supplementary Fig. 17. g Quantitation of Ki67 positivity, derived from 6-9 regions of interest (ROI) per tumor, three tumors per group, pooled for analysis N (ROI) per sample: Vehicle, 24; CT-1212-1 450 μg, 23; CT-1212-1 150 μg, 22; CT-1212-1 50 μg, 20; CT-1222-1 150 μg, 22; CT-5212-1 150 μg, 21. Two-tailed Welch’s T test vs Vehicle: CT-1212-1 450 μg, P = 0.0192; CT-1212-1, 150 μg, P = 0.0033; CT-1222-1 150 μg, P = 0.0013; CT-1212-1 50 μg and CT-5212-3 150 μg, P > 0.9999. See the Supplementary Data for full statistical breakdown. Each experiment was performed once. Source data are provided as a Source Data file.