Fig 1.
The Nedd4 family of ligases contains conserved WW domains for interactor recognition.
(A) Nedd4 and related ligases contain 2–4 WW domains that recognize interactors containing a PY motif (PPxY, LPxY) or phosphorylated threonine or serine residues. (B) Alignment of the four WW domains from prototypical member Nedd4 shows moderate sequence similarity and highlights conserved residues, including the two characteristic tryptophan residues (indicated by red arrows). (C) Solution state NMR structure of the Nedd4 WW domain 3 (grey) in complex with a PY motif peptide (red) from a known Nedd4 substrate (PDB ID: 2KPZ, unpublished) reveals key residues (blue) involved in peptide binding.
Fig 2.
UpSet analysis and Gene Ontology annotation reveal that Nedd4 family ligases are functionally distinct.
(A) Cross-reference of annotated interactors in the BioGrid database reveals that approximately one quarter of all interactors of the Nedd4 family are recognized by 2+ ligases, revealing little overlap in the known interactomes of the Nedd4 family. Data analysis performed with the UpSet plot tool [76] and graphic annotated in Adobe Illustrator. (B) Gene ontology analysis via the PANTHER database [43, 44] reveals that each Nedd4 family ligase interactome has similar trends in biological process (top) but distinct patterns in protein class composition (bottom).
Table 1.
Number of previously identified interactors for each Nedd4 family ligase in the BioGrid protein-protein interaction database.
Fig 3.
Analysis of PY motifs in interactomes of the Nedd4 family of ubiquitin ligases.
(A) Prevalence of PPxY and LPxY motifs in the interactome (from BioGrid database) [48, 49] across the Nedd4 family of ubiquitin ligases. (B) Representative WebLogo depictions of PY motif consensus sequences from all PPxY and LPxY motifs ± 10 amino acids for Nedd4 and ITCH. The WebLogo [50] diagrams for the remainder of the Nedd4 family ligases are shown in S2 Fig.
Fig 4.
Relative solvent accessible surface area (RSA) of PY motifs from the Nedd4-1, SMURF1, and WWP2 interactome reveals that PPxY motifs are more solvent accessible than LPxY motifs.
RSA, calculated with NetSurfP-2.0 bioinformatic algorithm [51], is determined by the ratio of total accessible surface area of a residue in the protein relative to maximum accessible surface area of the residue itself. A score of 0.25 or lower is indicative of “buried” residues, or those that would be inaccessible for engaging in protein-protein interactions. Buried residues (RSA < 0.25) are indicated in blue, moderately accessible (0.25 < RSA < 0.5) in red, and highly accessible (RSA > 0.5) in green. Data visualized with Prism GraphPad.
Fig 5.
Prediction of relative order of PY-motif containing regions from the Nedd4 interactome.
PY motif sequences ± 20 amino acids were extracted from the Nedd4 interactome using PxYFinder script and analyzed using (A) IUPred2A [52] and (B) PPIIPred [53] bioinformatic tools to determine relative order and propensity to form polyproline II structure, respectively. Data shown as average ± S.D. of 109 (PPxY) and 41 (LPxY) sequences. Statistical analysis using a paired t-test (to compare at each residue, numbered 1–44 above) reveals a statistically significant difference in predicted order between the PPxY and LPxY sequences (p < 0.0001 for both IUPred and PPIIPred scores). Analysis across the sequence using an unpaired Welch’s t-test also shows significant differences (p < 0.0001 for IUPred; p < 0.002 for PPIIPred).
Fig 6.
Rational design and computational analysis of PY motif peptide library to predict residue-specific changes in binding affinity.
(A) Nedd4 WW domain (PDB ID: 2KPZ) in complex with PY motif peptides from previously resolved WW domain/PY peptide complexes (PDB IDs: 2KPZ, 2M3O, 2KQ0, 4N7H). Peptides aligned to the 2KPZ complex using PyMol [75], showing moderate conservation of peptide backbone conformation when bound to the WW domain. (B,C) Rational design of PY peptide library involved variation of residues in the x–1 and x positions (shown in pink, B) and was informed by PY motif consensus sequences for Nedd4 (shown in Fig 2C and S2 Fig), affording a 30 member library. (D) Computationally predicted binding affinities of PY motif peptides screened against Nedd4 WW domain (PDB ID: 2M3O). Binding affinities are presented as ΔΔGbinding relative to the native peptide substrate TAPPPAYATLG (ΔGbindingdesigned − ΔGbindingnative). ΔΔGbinding for the native peptide is presented in the upper-right corner of the left heatmap for reference. ΔΔGbinding energies presented as kcal/mol or % of ΔGbindingnative. Full energy properties provided as in S1 Data.
Fig 7.
Individual biomolecular interaction types have varying contributions to overall binding affinity.
(A) Energetic contributions of individual interactions to overall binding affinity are shown for the native ligand (PPPAY), a weaker predicate binder (PPPGY), and a stronger predicted binder (TLPFY). Linear regression analysis reveals a positive and negative correlation, respectively, between van der Waals forces or solvation energy with overall binding energy (ΔGbinding). (B) Analysis of all individual energy components (for ΔGbind, the optimized complex, ligand, and receptor) to overall binding affinity reveals factors that ligand and complex energies are more strongly correlated than receptor energies. Ligand efficiency is defined as the binding energy/# heavy atoms where “sa” accounts for solvent exposed surface area and ln is the natural log of ligand efficiency. * indicates lipophilic interactions in the ΔGbind,NS where NS indicates binding energy of the peptide without accounting for ligand strain energies. Correlations calculated using python as Pearson coefficients and visualized in Prism GraphPad.
Table 2.
Functional analysis of non-PY containing interactors involved in ubiquitination.