Alternative molecular mechanisms for force transmission at adherens junctions via β-catenin-vinculin interaction

Abstract

Force transmission through adherens junctions (AJs) is crucial for multicellular organization, wound healing and tissue regeneration. Recent studies shed light on the molecular mechanisms of mechanotransduction at the AJs. However, the canonical model fails to explain force transmission when essential proteins of the mechanotransduction module are mutated or missing. Here, we demonstrate that, in absence of α-catenin, β-catenin can directly and functionally interact with vinculin in its open conformation, bearing physiological forces. Furthermore, we found that β-catenin can prevent vinculin autoinhibition in the presence of α-catenin by occupying vinculin´s head-tail interaction site, thus preserving force transmission capability. Taken together, our findings suggest a multi-step force transmission process at AJs, where α-catenin and β-catenin can alternatively and cooperatively interact with vinculin. This can explain the graded responses needed to maintain tissue mechanical homeostasis and, importantly, unveils a force-bearing mechanism involving β-catenin and extended vinculin that can potentially explain the underlying process enabling collective invasion of metastatic cells lacking α-catenin.

Fig. 1. Compact vinculin loses its contact with the cadherin-catenin layer when α-catenin is depleted.

a Cartoon representation of full-length vinculin structure from AlphaFold database indicating the four domains (called D1-D4) in an auto-inhibited conformation with the tail (Vt). b Cartoon representation of vinculin D1 domain showing the three important binding sites (herein referred to as S1, S2, and S3) for vinculin function. c Cartooon representation of full-length α-catenin structure obtained from AlphaFold database. In different gray scales the N, M, and FABD domains are indicated, highlighting in green the vinculin binding site (VBS) segment (amino acids 305–355). d Cartoon representation of α-catenin VBS showing polar non-charged amino acid in green and the non-polar residues L348, L347, L344, V340, and I333 in white. These residues are those that partially match the conserved hydrophobic motif of vinculin binding site (VBS) segments LxxAAxxVAxxVxxLIxxA (alignment in Supplementary Fig. 1) previously reported by Izard T. and coworkers for VBS1, VBS2, and VBS3. e Cartoon representation of the MD simulation last frame of vinculin D1 (cyan) and αcatenin VBS (green) complex. The two insets illustrate the interfaces of the main surface interactions: intermolecular hydrogen bonds (S349, Y351, S323, E336 of α-catenin and Q19, R105, K170, N53 of vinculin), hydrophobic stabilization of conserved non-polar residues motif inside the groove of S1, and salt bridge interactions (E60, E66 of vinculin and R332, R326 of α-catenin). f Electrostatic surface potential of the vinculin D1 and α-catenin VBS contact interfaces. g MM-PBSA per-residue energetic contribution. The energetic contribution of arginine residues (R326 and R332) of α-catenin together with asparagine (N53) and glutamic acid (E60) residues of vinculin coupled with the short mean distance of interaction in N53/R326 and E60/R332 of close to 3 Å along the MD (Supplementary Fig. 3) support the involvement of these residues in the complex stabilization. Notched box and whisker plots indicating the median ^zcentre position of indicated proteins in control MDCK (h) and α-catenin KD MDCK (i) cells obtained by SAIM (statical analysis reported in Supplementary Table 1). Box represents median, 1st, and 3rd quartiles; whiskers, 5th and 95th percentiles; #Adh. indicates the number of adhesions measured. Dotted line in (h) corresponds to the C-terminal z-position of α-catenin (^zcentre = 51.95 nm). Inserts in (h) and (i) indicate Topographic Map of the z-position of vinculin n- and c- termini. MDCK wt and MDCK α-catenin KD expressing FP probes of vinculin n- and c-terminus (left and right, respectively) were cultured on biomimetic E-cadherin-Fc -coated (Supplementary Fig. 4a) silicon wafer and imaged by surface-generated structured illumination microscopy. Color bar, 40–140 nm. Scale Bar, 10 μm. j, k iPALM imaging for F-actin in MDCK wt (left) and when α-catenin is depleted in MDCK α-catenin KD (right) in cells on E-cadherin biomimetic substrate. F-actin in MDCK cell labeled by Alexa Fluor 647-phalloidin. j top view, and (k) side view of white box in (j). Colors (hue scale in I, 0–150 nm) indicate the vertical (z) coordinate, relative to the substrate surface (z = 0 nm, red). Scale bars: 1 μm (j), 250 nm (k). Associated plot in Supplementary Fig. 4h.

Fig. 2. Activated vinculin maintains its connection to the cadherin-catenin layer when α-catenin is depleted.

Activated vinculin spans the core of cadherin-based adhesions. a Notched box and whisker plots indicating the median zcentre position of vinculin wt and constitutively active vinculin T12 mutants in MDCK wt and α-catenin KD, obtained by SAIM (statical analysis reported in Supplementary Table 1). Box represents median, 1st, and 3rd quartiles; whiskers, 5th and 95th percentiles; #Adh. indicates the number of adhesions measured. b Localization of vinculin constructs to AJs in MDCK wt and MDCK α-catenin KD in monolayer. Maximum intensity projections of vinculin (GFP, green channel) and E-cadherin (antibody probes, red channel), and merged image. Scale bar, 10 μm. c, d Co- immunoprecipitation of β-catenin with vinculin. Representative western blots (c), probing for β-catenin pulled-down with GFP-vinculin constructs as indicated on the top lane and (d) quantification of β-catenin intensity and GFP (correspondent to vinculin) from 3 independent experiments (n = 3), expressed as mean ± SD. Control for whole cell lysate (WCL) included. e, f Cartooon representation of full-length β-catenin structure obtained from AlphaFold database. VBS segment (amino acids 1–30) is highlighted in purple. g Cartoon representation of α-catenin VBS (green) and β-catenin VBS (purple) which both overlap on S1 binding site of vinculin D1 domain (cyan). h Cartoon representation of the MD simulation last frame of vinculin D1 (cyan) and βcatenin VBS (purple) complex. Insets with solid lines represent the residue interface of the main surface interaction and the inset with dotted line show the β-catenin intermolecular interactions. i Electrostatic surface potential (right) of the β-catenin VBS (cartoon representation shown on the left) contact interface with vinculin (not shown). j MM-PBSA per-residue energetic contribution for main residues of vinculin and βcatenin pair residue.

Fig. 3. Molecular dynamics analysis of vinculin and α- or β-catenin complexes with mutated sequences.

The n-terminus of activated vinculin with the point mutation A50I (vinculin T12-A50I) shifts higher up with respect to active vinculin without mutation. a Notched box and whisker plots indicating the median z-center position of the n-terminus of vinculin mutants, T12 and T12-A50I in MDCK α-catenin KD cells, obtained by SAIM (statical analysis reported in Supplementary Table 1). Box represents median, 1st, and 3rd quartiles; whiskers, 5th and 95th percentiles; #Adh. indicates the number of adhesions measured. b Localization of vinculin T12-A50I constructs to AJs in MDCK α-catenin KD in monolayer. Maximum intensity projections of vinculin (GFP, green channel) and E-cadherin (antibody probes, red channel), and merged image. Scale bar, 10 μm. c Cartoon representation of vinculin D1 mutated A50I (cyan)/ β-catenin (purple) dimer. The mutated Isoleucine amino acid is highlighted in yellow. The inset shows the main interactions and on the side the Lateral view of vinculin A50I/β-catenin dimer (90° rotation). d MM-PBSA per-residue energetic contribution for vinculin D1 mutated A50I/ β-catenin dimer. e Cartoon representation of vinculin D1 mutated A50I (cyan)/ α-catenin (green) dimer. The mutated Isoleucine amino acid is highlighted in yellow. The inset show the main interactions and on the side the Lateral view of vinculin A50I/α-catenin dimer (90° rotation). f MM-PBSA per-residue energetic contribution for vinculin D1 mutated A50I/ α-catenin dimer. g Cartoon representation of D1 vinculin wt (cyan)/ β-catenin (M8P) (purple) dimer. h MM-PBSA per-residue energetic contribution for vinculin D1/ β-catenin (M8P) dimer. i D1 vinculin (cyan)/ α-catenin (L344P) (green) dimer with the corresponding rotations. j MM-PBSA per-residue energetic contribution for vinculin D1/ α-catenin (L344P) dimer. The mutated amino acid in each mutant is highlighted in yellow in (c, e, g, i).

Fig. 4. Junctional tension and stability are supported by activated vinculin interaction with β-catenin when α-catenin is depleted.

a Top, schematic representation of the construct used to conduct single protein manipulation by vertical magnetic tweezer to detect VBS-VD1 interaction and ability to functionally bear forces. The single molecule detector construct is composed of avi-tag, β-catenin VBS, formin FH1 domain, vinculin D1 domain, two repeats of titin I27 domain, and spy-tag. Bottom, schematic representation of the experimental design and possible outcomes. VBS-VD1 complex may form at low forces (looped). Upon application of pulling force to the magnetic bead by tweezer, dissociation events can be detected by measuring z position of the bead and resulting ΔH. Unlooping events correspond to instances of VBS-VD1 separation (unlooped). At higher forces, partial or complete VD1 unfolding events can be concomitantly observed. b 2D graph of the force-dependent steps sizes at a loading rate of 1 pN s⁻¹ illustrates the ability of β-catenin VBS to form a stable interaction with vinculin VD1 and sustain physiological forces ranging from 7 to 16 pN. 17dissociation events where beads height changes are significantly larger than 35 nm (horizontal dashed line—Supplementary Fig. 13) from 8 individual experiments are plotted. Large variability in ΔH observed results from complex behavior of FH1 unlooping (theoretical ΔH ≈ 50–60 nm) and partial (theoretical ΔH ≈ 60-90 nm) or full (theoretical ΔH > 100 nm) VD1 unfolding. Top and right histograms report the counts of single events as a function of the applied force and step size, respectively. The bars represent the relative uncertainty of force measurements generated by bead size variation and relative position of attachment of the tether on the bead. This has been estimated to be 20% of the measured force. Inset, representative experiment shows a significant ~120 nm step at ~16 pN, indicating VBS-VD1 dissociation and VD1 unfolding. Three cycles of pulling at 1 pN s⁻¹ (dark lines) and release at 0.1 pN s⁻¹ (light lines) are shown with 1st cycle in black, 2nd cycle in red and 3rd cycle in blue. Experiments were performed at 37 ± 2 °C. c Representative time-lapse montages of recoil upon laser ablation. MDCK cells were co-transfected by the indicated constructs as well as ZO-1 to mark the cell-cell junctions. The edges of junctions are denoted by red dots. Scale Bar, 10 µm. d Trajectory of junctional recoil following ablation. Error bars indicate s.e.m. e Plot of initial recoil rate after laser ablation for the indicated conditions. Box represents median, 1st, and 3rd quartiles; whiskers, 5th and 95th percentiles. Statistical analysis in Supplementary Table 4, following ordinary one-way ANOVA with Sidak correction for multiple comparisons.*: p<0.05; ***: p < 0.001; n.s.: not significant. Representative video of recoil upon laser ablation for each condition indicated in (d, e) can be found in Supplementary Movie 7. Number of cell-cell junctions measured for each condition is indicated in the plots. Statistics and statistical analysis are reported in Supplementary Tables 3, 4. f Representative time-lapse images from Fluorescence Recovery After Photobleaching (FRAP) of β-catenin. (Right) Montage (zoom 2X) of the bleached regions (yellow squares in the main figures). Scale bar, 5 μm. g, h FRAP analysis of β-catenin for the indicated conditions. g Normalized recovery trajectory was fitted to a single exponential model, with mobile fraction of β-catenin relative to total. Error bars indicate s.e.m. h Plot of mobile fraction of β-catenin relative to total. Box represents median, 1st, and 3rd quartiles; whiskers, 5th and 95th percentiles. Statistical analysis in Supplementary Table 6, following ordinary one-way ANOVA without correction for multiple comparisons *: p < 0.05; ***: p < 0.001; n.s. not significant. Representative video of FRAP for each condition indicated in (g, h) can be found in Supplementary Movie 8. Number of cell-cell junctions measured for each condition is indicated in the plots. Statistics and statistical analysis are reported in Supplementary Tables 5, 6. i, j Analysis of directionality of collective migration suggests a partial rescue effect of activated vinculin (vinculin T12) on α-catenin depletion. Rose plots show the angular frequency distribution of the integrated velocity (expressed as percentage of the total—i) and the relative angular speed magnitude (µm/hour—j) of migrating MDCK wt, MDCK α- catenin KD, MDCK α-catenin KD transfected with vinculin T12 cells. Red indicates the mean and pink indicates standard deviations. Representative video of migration assay for each condition can be found in Supplementary Movies 9–11 and statistical analysis in Supplementary Table 7.

Fig. 5. MD analysis of putative vinculin - α-catenin - β-catenin trimeric complex.

Cartoon type representation of the trimer vinculin D1 domain (cyan)/α-catenin VBS (green)/ β-catenin VBS (purple). a Focus on vinculin D1 domain (cyan)/α-catenin VBS (green) in S1 binding pocket. Insets show the intra and intermolecular residue interaction. b The complex is rotated 90° to show a cartoon and surface representation of lateral view of the trimer α-catenin/ vinculin/ β-catenin showing both catenin peptides coupled to vinculin D1. c Focus on vinculin D1 domain (cyan)/ β-catenin VBS (purple) in S3 binding site. The inset shows the new interaction generated by residue W25 that provides a key pi-stacking interaction with Y351 and establish the vinculin/α-catenin/β-catenin trimer. MM-PBSA per-residue energetic contribution for (d) vinculin D1/α-catenin VBS and (e) vinculin D1/β-catenin VBS to the trimer stability.

Fig. 6. Different conformational arrangements of vinculin-catenin complex.

Cartoon representations of full-length proteins of the mechanotransduction module (vinculin, α-catenin, and β-catenin). Vinculin is represented in its closed and open conformation. In colors, we highlighted the putative peptides/domains involved in the interactions. In the second row (Domains/Peptides), the interacting regions highlighted above (vinculin D1, VBS α-catenin, and VBS β-catenin) are individually visualized as surface representations. In vinculin D1 domain, the sites S1, S2, and S3 are indicated. In the third row (Dimers/pairs), interactions between couples of Domains/peptides are shown. In the last row, the proposed new trimer of vinculin/β- catenin/α-catenin is shown, with vinculin as central molecule. For dimers and the trimer, the calculated MM-PBSA per-residue energetic contribution are indicated.