Proximity interactions of the ubiquitin ligase Mind bomb 1 reveal a role in regulation of epithelial polarity complex proteins

Abstract

MIB1 belongs to the RING domain containing family of E3 ubiquitin ligases. In vertebrates, MIB1 plays an essential role in activation of Notch signaling during development, through the ubiquitination and endocytosis of Notch ligands. More recently, Notch independent functions for MIB1 have been described in centriole homeostasis, dendritic spine outgrowth and directional cell migration. Here we use proximity-dependent biotin identification (BioID) to define the MIB1 interactome that included 163 high confidence interactions with polypeptides linked to centrosomes and cilia, endosomal trafficking, RNA and DNA processing, the ubiquitin system, and cell adhesion. Biochemical analysis identified several proteins within these groups including CCDC14 and EPS15 that were ubiquitinated but not degraded when co-expressed with MIB1. The MIB1 interactome included the epithelial cell polarity protein, EPB41L5. MIB1 binds to and ubiquitinates EPB41L5 resulting in its degradation. Furthermore, MIB1 ubiquitinates the EPB41L5-associated polarity protein CRB1, an important determinant of the apical membrane. In polarized cells, MIB1 localized to the lateral membrane with EPB41L5 and to the tight junction with CRB1, CRB3 and ZO1. Furthermore, over expression of MIB1 resulted in altered epithelial cell morphology and apical membrane expansion. These results support a role for MIB1 in regulation of polarized epithelial cell morphology.

Figure 1

Identification of MIB1 proximity interactome using BioID. (a) Venn diagram of the MIB1 BioID protein data set and published MIB1 interactors obtained from BioGRID (https://thebiogrid.org) where MIB1 was bait. (b) Graph showing GO functional annotation clusters for the MIB1 BioID dataset. MIB1 proximity interactors were uploaded to the online Database for Annotation, Visualization and Integrated Discovery, DAVID, and clustered using an EASE of 1.0 and medium stringency. (c) Schematic representation of 147 of 163 proteins identified using BioID with FlagBirA*-MIB1 expressed in HEK293 cells. SAINT data were uploaded to string-dp.org for network analysis. The resulting network was imported into Cytoscape 3.6.1 for figure preparation. Nodes represent unique proteins identified and categorized according to the indicated function based on information in Entrez Gene, UniProt and primary literature. Light grey edges represent protein-protein interactions determined by STRING (interaction score of 0.4 or greater). Nodes and edges in bold are previously published MIB1 interactors or substrates. The complete list of unique interactors is shown in Table S1.

Figure 2

The centrosome-associated proteins CCDC14, KIAA0753 and OFD1 are ubiquitinated when co-expressed with MIB1. (a–c) Flag-CCDC14, KIAA0753-Flag or Flag-OFD1 were individually co-transfected into HEK293T cells with HA-MIB1 WT, a MIB1 mutant that disrupts the last RING finger domain (CA), or a MIB1 mutant that has the 3 RING domains deleted (Δ3R). After 18 hours, cell proteins were extracted in 1xLaemmli lysis buffer and boiled. Equal concentrations of protein were separated by SDS-PAGE and immunoblotted with either anti-FlagM2 or anti-HA (for MIB1). (d,e) Flag-tagged CCDC14, KIAA0753 and OFD1 were individually co-transfected with HA-Ub and either T7-MIB1 or T7-MIB1CA mutant. At 18 hours, cells were treated for 4 hours with 10 μM MG132, followed by lysis is 1%SDS/RIPA lysis buffer and boiled. Flag-tagged proteins were immunoprecipitated from an equal concentration of cell lysate, and immunoblotted with anti-HA (Roche) to identify ubiquitinated proteins. Lysates (50 μg) were immunoblotted for Flag-tagged substrate (middle panels) and T7-MIB1 (lower panels). Each figure is representative of at least 3 independent experiments. For clarity some blots were cropped. Full-length blots are shown in Fig. S4.

Figure 3

The endocytic proteins EPS15 and FCHO2 are ubiquitinated when co-expressed with MIB1. (a–c) Eps15-Flag, Flag-FCHO2 or HA-AP2B1 were individually co-transfected into HEK293T cells with either HA-tagged MIB1 WT, or mutants MIB-CA or MIBΔ3 R (Eps15 and FCHO2 samples) or T7-tagged MIB WT and mutants (AP2B1 experiment). After 18 hours, cell proteins were extracted in 1xLaemmli lysis buffer and boiled. Equal concentrations of protein were separated by SDS-PAGE and immunoblotted with either anti-FlagM2, anti-HA (for MIB1 or AP2B1) or anti-T7 (MIB1). (d–f) Eps15-Flag, Flag-FCHO2, or HA-AP2B1 were co-transfected with T7-MIB1 or T7-MIB1CA mutant, and either HA-Ub (Eps15 and FCHO2) or HIS6-Ub (AP2B1). At 18 hours, cells were treated for 4 hours with 10 μM MG132, followed by lysis in 1%SDS/RIPA lysis buffer and boiled. Tagged substrates were were immunoprecipitated from 2 mg of cell lysate as indicated, and immunoblotted with either anti-HA (Roche) or anti-HIStag to identify ubiquitinated proteins. Lysates (50 μg) were immunoblotted to confirm substrate and T7-MIB1 expression. Each figure is representative of 3 (Eps15), 2 (FCHO2) and 1 (AP2B1) independent experiments. For clarity some blots were cropped. Full length blots are shown in Supplementary Fig. S4.

Figure 4

MIB1-mediates ubiquitination and degradation of the polarity protein EPB41L5. (a,b) HEK293T cells were transiently co-transfected with FLAG-EPB41L5 or ZO1-myc, and HA-MIB1 WT or RING mutant (HA-MIB1CA or HAMIB1Δ3R). Eighteen hours following transfection, 50 μg of cell lysates were resolved by SDS-PAGE and immunoblotted with anti-FlagM2 (a, EPB41L5), anti-myc (b, ZO1), or anti-HA (MIB1). (c) EPB41L5 is ubiquitinated by MIB1. HEK293T cells were transiently co-transfected with Flag-EPB41L5, HA-ubiquitin and either T7-MIB1WT, T7-MIB1CA, or T7-MIB1Δ3R. Eighteen hours post-transfection, cells were treated for 4 hours with 10μM MG132, and cell lysates prepared in 1%SDS/RIPA lysis buffer. Flag-EPB41L5 was immunoprecipitated from 2 mg of the protein lysate with anti-FlagM2 and immunoblotted with anti-HA to detect ubiquitinated proteins (Top). Lysates (50μg) were probed with anti-FlagM2 (middle) and anti-T7 (bottom panel) to confirm expression of EPB41L5 and MIB1. Each figure is representative of at least 3 independent experiments. For clarity some blots were cropped. Full length blots are shown in Supplementary Fig. S4.

Figure 5

Co-expression of MIB1 prevents EPB41L5 mediated cell shape changes. (a) HeLa cells were transiently transfected with (images left to right): Flag-EPB41L5/pIRES/EGFP plus T7/pcDNA3.1, Flag-EPB41L5/IRES/EGFP plus T7-MIB1, T7-MIB1 with empty pIRES2/EGFP, or Flag-EPB41L5/IRES/GFP plus T7-MIB1 CA mutant. After 18 hours cells were trypsinized and reseeded onto fibronectin coated coverslips and allowed to adhere for 2 hours. Cells were fixed, permeablized, and immunostained with anti-FlagM2 (EPB41L5; CY3 anti-mouse, upper row, left), or anti-MIB1 (MIB1; CY3 anti-guinea pig, upper row, 3 right images), anti-GFP (co-expressed GFP; AlexaFluor 488 anti-rabbit, middle row), and Alexa Fluor 635 phalloidin and DAPI (F-actin and nuclei; lower row). Images were acquired at 20X magnification on a Quorum Spinning Confocal microscope and analysed using Perkin Elmer Volocity software. Scale bar is 50 μm. Representative immunoblots showing protein expression are shown in Fig. S2. (b) Quantification of cell shape. The cell length to width ratio for each condition was measured: untransfected (UT: 152 cells over 4 experiments), empty IRES2/EGFP and T7/pcDNA3.1 vectors (137 cells over 3 experiments), EPB41L5/IRES/EGFP (219 cells over 4 experiments), EPB41L5/IRES/EGFP plus T7-MIB1 (186 cells over 4 experiments), IRES/EGFP plus T7-MIB1 (184 cells over 4 experiments) and EPB41L5/IRES/EGFP plus T7-MIB1 CA mutant (60 cells, 1 experiment). The summary graph shows individual cell measurements with their mean +/−SEM. for each condition. Statistical significance was measured using a One-Way ANOVA (Kruskal-Wallis) plus Dunn’s Multiple Comparisons test, p < 0.05.

Figure 6

MIB1 MZM/REP region is required for binding to EPB41L5. (a) Schematic of the MIB full length (FL) and truncation mutant proteins displaying various domains as follows: MZM [MIB/HERC2 domains (MH); Zinc finger (ZF)]; REP [MIB domain (MIB)]; Ankyrin repeats (Ank); RING fingers; HA, Hemagglutinin tag. MIB1-FL and truncated mutants were used to map the interaction site of MIB1 with EPB41L5 by co-immunoprecipitation experiments. (b) The N-terminus of MIB1 mediates association with EPB41L5. Co-immunoprecipitation experiments were performed on lysates from HEK293T cells co-transfected with FLAG-EPB41L5 and the indicated HA-tagged MIB1 mutant. EPB41L5 co-immunoprecipitates MIB1-FL, MIB1-Nterm, MIB1-∆3R and MIB1-∆RF3. The panel on the right shows the relative expression of EPB41L5 and MIB1 constructs respectively, in whole cell lysates. (c) EPB41L5 binds to FL and N-terminus of recombinant GST-tagged MIB1. GST pulldown experiments were performed using GST-tagged MIB1-FL and mutants as described in (a), with lysates prepared from HEK293T cells transiently transfected with Flag-EPB41L5. EPB41L5 bound to GST-fusion proteins was identified by immunoblotting with anti-EPB41L5. The amount of GST-fusion protein used is shown in the coomassie-stained blot. (d) Schematic representation of the EPB41L5 full length (FL) and truncation mutant proteins displaying the FERM domain and PDZ-binding motif (PDB). (e) EPB41L5 co-immunoprecipitates MIB1 and the N-term + FA region of EPB41L5 is not required to mediate this interaction. Co-immunoprecipitation experiments were performed on lysates from HEK293T cells co-transfected with the FLAG-EPB41L5-FL or truncation mutants, and HA-MIB1 CA RING mutant (MIB1CA). The MIB1CA mutant binds EPB41L5, and was used in these experiments because, unlike MIB1WT, it is not degraded. Anti-FlagM2 immunoprecipitates were immunoblotted for bound MIB1 using anti-HA (top panel). All EPB41L5 deletion mutants except that lacking the C-terminus (EPB41L5Nterm + FERM/FA) bound MIB1CA. The lower panels show the relative expression of Flag-EPB41L5 and HA-MIB1 constructs, in whole cell lysates. For clarity, some blots were cropped. The full length blots are shown in Supplementary Fig. S4.

Figure 7

Apical polarity protein CRB1 is associated with and ubiquitinated by MIB1. (a) CRB1 co-immunoprecipitates with MIB1. Co-immunoprecipitation experiments were performed on lysates from HEK293T cells co-transfected with WT or mutant constructs of CRB1 and MIB1. CRB1AAA represents a mutant construct where the FERM domain binding motif of CRB1 is mutated thereby abolishing its binding to EPB41L5. The panels below (Lysates) show the relative expression of CRB1, MIB1 and EPB41L5 constructs in cell lysates. The MIB1CA ligase dead mutant bound to CRB1 and binding was enhanced in the presence of EPB41L5. (b) CRB1 is a substrate for MIB1-mediated ubiquitination. HEK293T cells were transiently co-transfected with myc-CRB1, HA-ubiquitin and either T7-MIB1WT or the ligase mutant T7-MIB1CA. Equivalent amounts of protein lysates were immunoprecipitated with anti-CRB1 and immunoblotted with anti-HA to detect ubiquitinated proteins. Lysates (5% of that used for immunoprecipitation) were immunoblotted with anti-CRB1 and anti-MIB1 to confirm protein expression. For clarity, blots were cropped. The full length blots are shown in Supplementary Fig. S4.

Figure 8

MIB1 co-localizes with EPB41L5 at the lateral membrane and CRB1, CRB3 and ZO1 at tight junctions in polarized MDCK cells. (a) Endogenous MIB1 is relocalized from the cytosol/tight junction to the lateral membrane by overexpression of EPB41L5. Polarized MDCK cells were transiently transfected with Flag-EPB41L5 and fixed and stained for endogenous MIB1 (left panel) and Flag (EPB41L5; middle panel). Shown are XY sections captured midway between the basal and apical membranes. XZ sections were captured where the crosshair is positioned (indicated by white line). Images were acquired at 60X magnification on a Quorum spinning disk microscope. Scale bars indicate 10 μm. (b) MIB1 colocalizes with CRB1 at the tight junction region of polarized MDCK cells. MDCK-MycCRB1 cells, transiently transfected with HA-MIB1, were fixed and immunostained with anti-HA (MIB1, left image) and anti-myc (myc-CRB1, middle image). Shown are XY sections taken across the apical region of the cell monolayer and corresponding XZ section. Scale bars indicate 10 μm. (c) MIB1 colocalizes with endogenous CRB3. Polarized MDCK cells, transiently transfected with HA-MIB1, were fixed and immunostained with anti-HA (MIB1) and anti-CRB3. Shown are XY sections taken at the apical region of the cell monolayer with a corresponding XZ section. Scale bars indicate 10 μm. (d) MIB1 colocalizes with endogenous ZO1 at the tight junction. Polarized MDCK cells, transiently transfected with HA-MIB1, were fixed and immunostained with anti-HA (MIB1) and anti-ZO1. Shown are XY sections taken at the apical region of the cell monolayer and corresponding XZ section. Scale bars indicate 6 μm.

Figure 9

MIB1 expression in polarized MDCK cells regulates cell shape. (a,b) Overexpression of MIB1 results in the expansion of the apical membrane. MDCK cells were transfected with HA-MIB1, or the ubiquitin ligase mutants HA-MIB1CA or HA-MIB1ΔRF. Polarized cells were fixed and stained with anti-HA (MIB1-transfected cells) and either (i)anti- ZO1 to delineate the apical region or (ii) β-Catenin to show the lateral membrane. (i) XY extended focus projections of transfected cells co-stained for MIB1 and ZO1. (ii) XZ optical sections of transfected cells co-stained for MIB1 and β-catenin. Images were acquired at 60X magnification on a Quorum spinning disk microscope. Scale bar indicates 10 μm. (b) Quantification of the apical areas of polarized MDCK cells over expressing HA-MIB1 and adjacent cells. ZO1 staining was used to define the apical region and the area was measured using Volocity software. The individual cell measurements for 3 independent experiments are shown (HA-MIB1: 101 cells, adjacent untransfected: 601 cells; HA-MIB1CA: 112 cells, adjacent untransfected: 413 cells; HA-MIBΔRF1: 127 cells, adjacent untransfected: 477 cells) together with the corresponding mean value +/−SEM. Differences in area between transfected cells and corresponding adjacent cells were tested for statistical significance using a 2-tailed Mann Whitney test, p = < 0.0001. (c,d) Depletion of MIB1 causes the shortening of the lateral membrane in MDCK cells. (c) MDCK cells treated with either scrambled or MIB1 siRNAs were allowed to polarize, then fixed and stained with anti-E-Cadherin (a marker of the lateral membrane) and anti-MIB1, and counterstained with DAPI to visualize nuclei. Shown are XY optical sections taken across the middle of the cell monolayer and XZ sections taken where the crosshair is positioned (top and bottom of each panel, respectively). Images were acquired at 60X magnification on a Quorum spinning disk microscope. Scale bars indicate 10 µm. (d) Quantification of the effect of MIB1 depletion on lateral membrane height. Measurement of the lateral membrane length was performed using E-Cadherin stained samples and Volocity measurement software. The individual cell measurements for 2 independent experiments are shown (siScrambled: 43 cells, siMIB1: 49 cells) with the corresponding mean value +/- SEM. Statistical significance was tested using the 2-tailed Mann-Whitney test, p = <0.0001.