Structure-based design of potent and selective inhibitors of the HECT ligase NEDD4

Abstract

Among the human E3 ubiquitin ligases, NEDD4 (Neural precursor cell expressed developmentally down-regulated 4) plays a critical role in development and cancer, making it a compelling therapeutic target. However, no specific NEDD4 inhibitors have advanced in drug development. In this study, we reveal the inhibitory mechanism of Norclomipramine, a tricyclic antidepressant, which inhibits NEDD4-mediated ubiquitin chain elongation by binding to a hydrophobic pocket in the Ub exosite of the N-lobe. Building on this mechanism, we conducted a focused medicinal chemistry campaign, resulting in the development of covalent inhibitors that specifically target the non-catalytic cysteine C627. These compounds exhibit selective binding to NEDD4 over other family members, effectively inhibiting NEDD4-mediated polyubiquitination while leaving monoubiquitinated substrates unaffected. Among these, compound 32 emerged as a potent lead (IC₅₀ = 0.12 µM) with favorable pharmacokinetic properties, including oral bioavailability, paving the way for future in vivo efficacy studies.

Fig. 1. Norclomipramine affects NEDD4 processivity and ubiquitin binding.

A Chemical structure of Norclomipramine. B Trans-thioesterification reaction performed with bacterially expressed HECT of NEDD4 in presence or absence of Norclomipramine (Norc). HECT thioester formation was monitored after quenching the reaction at different time points by addition of Laemmli buffer without reducing agent. Immunoblot (IB) and Coomassie as indicated. C In vitro ubiquitination assays of the HECT of NEDD4 pre-incubated with the indicated concentration of Norclomipramine (Norc). Reactions were carried out for the indicated time points. IB and Coomassie as indicated. D GST pull-down assays with the isolated NEDD4 HECT and K63-polyUb chains. GST-fusion proteins pre-incubated with Norclomipramine (Norc) at 0.2 mM for 1 hour at 4 °C were either washed or directly incubated with K63-polyUb chains as indicated. IB and Coomassie as indicated.

Fig. 2. Norclomipramine binding site overlaps with Ub exosite.

A Superposition of the HECT of NEDD4 in complex with Norclomipramine (Norc) or ubiquitin (Ub) in the exosite (from PBD 2XBB). The HECT of NEDD4 is depicted in surface representation, with N-lobe in blue and C-lobe in green. Norclomipramine, in yellow sticks, sits on a hydrophobic pocket which is also exploited by the C-terminal tail of Ub (gold ribbon representation, from PBD 2XBB). B Close-up view of HECT^NEDD4:ubiquitin interaction (PBD 2XBB, N-lobe in blue, Ub in yellow). C Close-up view of HECT^NEDD4:Norclomipramine interaction (PBD 9H9O, N-lobe in blue, Norclomipramine, in yellow sticks).

Fig. 3. Compound 15 is a covalent inhibitor of NEDD4.

A The modification of Norclomipramine’s nitrogen methyl groups with various electrophilic warheads led to the discovery of compound (cmpd) 5, which exhibits low micromolar potency. B Close-up view of HECT NEDD4 bound to compound 15 (PDB 9H9T). Side chains of residues involved in inhibitor binding as well as residue Cys627, covalently bound to the inhibitor are highlighted. C Ub-TR-FRET assay performed with full-length protein NEDD4 pre-incubated for 1 hour with increasing concentrations of compound 15 before ATP addition and measurement. Plots show single time-point measurements of FRET signal to background ratio (% Signal/Background) as function of time (minutes). Mean ± SEM (n = 4). D Ub-TR-FRET assay as in C using the HECT domain of NEDD4. Mean ± SEM (n = 4).

Fig. 4. The inhibitor binding pocket is unique among NEDD4 family members.

A Amino acids alignment, based on 3D structures, of NEDD4 and NEDD4 family members indicating the conserved and variable residues of the inhibitor binding pocket. Cys627, covalently bound to the inhibitor is highlighted in yellow, Phe residues corresponding to amino acid 553 in NEDD4 are highlighted in orange, Leu residues corresponding to the same position are in red. B Close-up view of the inhibitor binding pocket of ITCH (PDB 3TUG). Residue Cys610, positioned differently within the pocket, is highlighted. C Ub-TR-FRET assay performed using the HECT of NEDD4 (left panel) or NEDD4-Like (right panel) in absence or presence of compound 15 at the indicated concentrations. Plots show time-point measurements of FRET signal to background ratio (% S/B) as function of time (minutes). Mean ± SEM (n = 2). D Superposition of the N-lobes of HECT NEDD4 in complex with compound 15 (PDB 9H9T) and HECT NEDD4-Like structures (PDB 2ONI). The presence of Phe608 (orange) in NEDD4-Like hinders inhibitor binding. Leu residue corresponding to the same position in NEDD4 is in green. The respective Cys residues are highlighted using the same colour scheme. E Ub chain formation assay with the indicated HECT domains treated or not with compound 15 at 10 µM concentration. Reactions were stopped at 30 min by quenching the reaction with Laemmli buffer with reducing agent. IB as indicated. Mono-, di- and polyUb chains are labeled on the right. Bottom, Coomassie staining showing comparable loading of proteins. F Schematic representation of the target validation workflow using a mass spectrometry-based approach. A549 wild-type (WT) and NEDD4 knockout (KO) cell lines were treated with 1 µM compound 17^B for 4 h. A total of 2 mg of urea lysates were subjected to streptavidin-based immunoprecipitation. The bound proteins were eluted, separated by gel electrophoresis, and analysed by LC-MS/MS. G One-tenth of the protein complexes bound to streptavidin beads were analysed by IB, as indicated. For input controls, 30 µg of total cell lysates were loaded.

Fig. 5. Covalent inhibitors inhibit substrate polyubiquitination but not monoubiquitination.

A Substrate ubiquitination assay with the indicated E3s treated or not with compound 15 (10 µM). The complete reaction was loaded onto a gel. IB as indicated. Arrowhead indicates the unmodified WBP2 protein. B Same as in A, using GST-γENAC as substrate. At the end of the reaction, substrate was separated by centrifugation and loaded onto a gel. IB as indicated. Bottom, Coomassie staining showing comparable loading of proteins. C Same as in A, using GST-Eps15 as substrate. Top panel, experiment performed with NEDD4, bottom panel with NEDD4-Like. Coomassie staining showing comparable loading of proteins are reported. D B82L cells were serum starved overnight, treated with compound 15 (1 µM) for 4 h and then stimulated with EGF (100 ng/ml) for the indicated time points. Lysates were analyzed by IB with the indicated antibodies.