Elucidation of Short Linear Motif-Based Interactions of the FERM Domains of Ezrin, Radixin, Moesin, and Merlin

Abstract

The ERM (ezrin, radixin, and moesin) family of proteins and the related protein merlin participate in scaffolding and signaling events at the cell cortex. The proteins share an N-terminal FERM [band four-point-one (4.1) ERM] domain composed of three subdomains (F1, F2, and F3) with binding sites for short linear peptide motifs. By screening the FERM domains of the ERMs and merlin against a phage library that displays peptides representing the intrinsically disordered regions of the human proteome, we identified a large number of novel ligands. We determined the affinities for the ERM and merlin FERM domains interacting with 18 peptides and validated interactions with full-length proteins through pull-down experiments. The majority of the peptides contained an apparent Yx[FILV] motif; others show alternative motifs. We defined distinct binding sites for two types of similar but distinct binding motifs (YxV and FYDF) using a combination of Rosetta FlexPepDock computational peptide docking protocols and mutational analysis. We provide a detailed molecular understanding of how the two types of peptides with distinct motifs bind to different sites on the moesin FERM phosphotyrosine binding-like subdomain and uncover interdependencies between the different types of ligands. The study expands the motif-based interactomes of the ERMs and merlin and suggests that the FERM domain acts as a switchable interaction hub.

Figure 1

Overview of the ERM and merlin subfamily of the FERM domain containing proteins. (A) Sequence identities between full-length proteins show that merlin is the most distant variant in terms of sequence divergence. (B) Modular architectures of the ERMs and merlin reflect a common N-terminal FERM domain, which is connected through an α-helical domain to the CTD. (C) Structure of moesin in complex with its CTD. The different subdomains and the binding sites are indicated (PDB: 1EF1(5)). (D) In its inactive, closed form, the CTD binds to the FERM domain. Upon phosphorylation and binding to PIP2, the two domains are free to interact with their partners: The FERM domain interacts with SLiM-containing target proteins, mainly in the plasma membrane’s vicinity, while the CTD domain binds actin filaments. Panel (B, D) inspired by Fehon, McClatchey, and Bretscher.

Figure 2

Overview of ProP-PD results, motifs, and ligands identified for the FERM domains of the ERMs and merlin. (A) Overlap of high/medium confidence peptide ligands identified through phage selections against the ERM proteins and merlin (Tables S1–S4). (B) Consensus binding motifs of the ERM ligands generated by PepTools. (C) Key amino acids in merlin binding FAM83G and NOP53 peptides identified through mutational analysis. The effects of the mutations on binding were evaluated by clonal phage ELISA against immobilized GST-tagged merlin FERM domain. The binding was assessed by the ratio of the A₄₅₀ values detected for the immobilized target protein (GST-tagged merlin) to that of the background (GST). The results were normalized to 100% binding of the respective WT peptide. As an extra negative control (indicated “control)”, a clonal phage ELISA was performed for the same proteins using an M13 phage displaying no peptide. (D) Network of a selected set of protein–protein interactions. Shown interactors include proteins that share biological functions with the baits that are unlikely to occur by chance based on GO term analysis (Tables S1–S4). The baits ezrin (EZN), radixin (RDX), moesin (MSN), and merlin (NF2) are indicated as blue octagon. Interactions supported by results from other studies are indicated with green edges. The binding motifs of the interactors are indicated by the node color.

Figure 3

Determination of affinities through direct binding FP experiments and validation of interactions with full-length proteins through GST-pulldowns. (A) FITC-labeled probe peptides (5 nM) were titrated with increasing concentrations of moesin, ezrin, radixin, and merlin FERM domains. The results were fitted to a quadratic equation for 1:1 binding (number of technical repeats (n) = 3). The raw FP data are available in Figure S2. (B) Summary of K_D values as determined in (A). Most of the tested ligands bind to ERM FERM domains with low-micromolar K_D values and to merlin with lower affinity. (C) GST-pulldowns of WT and motif-mutant full-length proteins. GST-tagged moesin or merlin FERM domains were used to pull down WT or mutant proteins transiently expressed in HEK293 cells. Mutated amino acids are indicated to the right (bold residues were mutated to alanine). Note that LATS1 has two moesin binding motifs. Results shown are representative of at least two replicated experiments. * indicates not tested.

Figure 4

MISP3, KIRREL3, and TBX4 top-scoring peptide models bind the F3b pocket adopting similar conformations as the crumbs (CRB) peptide in the F3b binding pocket. (A) Bound crumbs peptide (magenta, PDB 4YL8), MISP3 top-scoring models from FlexPepBind threading and PIPER-FlexPepDock global docking are shown in purple and cyan, respectively. (B and C) Top-scoring KIRREL3 peptide model from threading and TBX4 peptide model from global docking are shown in green and yellow, respectively. The supposedly crucial residues that interact with the peptides are labeled. (D and E) Threading binding energy landscapes of MISP and KIRREL, respectively. The identified motif is highlighted in red. (F) Global docking simulation binding energy landscape of TBX4, in which the lowest energy models converge toward the F3b site. Top 10 cluster representatives of the simulation are shown with red dots. The RMSD is calculated relative to the lowest scoring structure. (G) Affinities determined through competitive FP-based affinity measurements using variants of the KIRREL3 peptide (top) or the TBX4 peptide (bottom) for competition using FITC-KIRREL3_637–652 as probe. The mutational analysis validated the importance of the key YxV residues (mutated position in red), while N-terminal extensions of the peptides had no or minor effects on binding. (H) Fold change in the binding affinity of moesin FERM M285/H288A as compared to the WT (K_D^M285/H288A/K_D^wt) for six different peptides, as determined by direct binding of FITC-labeled peptides (see Table S7 for complete data).

Figure 5

Structural modeling suggests that ZNF622 and BTBD7 bind the moesin FERM domain at F3a using the same binding pocket as moesin CTD and EBP50. Models of the interactions of (A) ZNF622 and (B) BTBD7 suggest a perpendicular conformation, both using aromatic side-chains to fill the same hydrophobic pocket. As expected, this distinct conformation is only identified by a global docking simulation but not by threading. (C) Fragment of moesin CTD bound at the same site (PDB code: 1EF1). (D) Affinities determined through competitive FP-based measurements using WT FITC-labeled ZNF622 as probe. The results highlight the importance of core FYDF motif residues in peptide binding. (E) Fold change in affinity of the moesin K211A/I238A as compared to the WT-moesin for binding with five different peptides. Affinities were determined by direct binding using respective FITC-labeled probes.

Figure 6

Interplay among different ligands binding to moesin FERM domain F3a and F3b pockets. (A) Representative peptides from the ligand set that bind to F3a and F3b binding pockets show the presence of consensus motifs in the peptide sequence. (B) Schematic of the competition experiment design to evaluate the interplay between moesin FERM binding peptides. Binding of a FITC-labeled probe peptide was monitored in the presence of high saturating concentrations of unlabeled competing peptides, as shown in (C–F). (C and D) When the F3a binding peptides of EBP50 (C) or ZNF633 (D) are in excess, they block binding of all FITC-labeled probe peptides. (E and F) When the F3b binding peptides of KIRREL3 (E) or TBX4 (F) are in excess, they block binding of FITC-labeled probe peptides to the F3b site, but FITC-EBP50 and FITC-ZNZ622 can still bind to the F3a site. Schematics next to saturation plot (C–F) illustrate hypothetical explanations for the observed competition for each combination of titration of labeled peptide with moesin FERM domain in the presence of a saturating concentration of unlabeled peptide binding either to pocket F3a (EBP50 and ZNF622) or F3b (KIRREL3 and TBX4).

Figure 7

Energy landscapes of simulations using different receptor structures. (A, B) Energy landscape sampled by the ZNF622 peptide: The F3a site is accessible both in F3a bound (A) and F3b bound (B) structures. (C) Comparison of the F3a bound (green) and F3b bound (cyan) structures shows conformational change depending on the occupied binding site. (D, E) Energy landscapes demonstrate that the F3b site is inaccessible in the F3a bound FERM structure: Docking of the MISP peptide onto the F3b-bound (D), F3a-bound (E), and F3a-bound relaxed structure with the superimposed peptide at the F3b binding site (Figure S6) (F) highlights the importance of conformational change for the identification of the F3b binding site. Cluster centers are shown with red dots. Only the F3 subdomain was used in the simulations. The highlighted circles in (A) and (E) show reference models used for RMSD calculations (structures shown in Figures 5A and 4A, respectively).