Actin-rich lamellipodia-like protrusions contribute to the integrity of epithelial cell-cell junctions

Abstract

Metastasis-suppressor 1 (MTSS1) is a membrane-interacting scaffolding protein that regulates the integrity of epithelial cell-cell junctions and functions as a tumor suppressor in a wide range of carcinomas. MTSS1 binds phosphoinositide-rich membranes through its I-BAR domain and is capable of sensing and generating negative membrane curvature in vitro. However, the mechanisms by which MTSS1 localizes to intercellular junctions in epithelial cells and contributes to their integrity and maintenance have remained elusive. By carrying out EM and live-cell imaging on cultured Madin-Darby canine kidney cell monolayers, we provide evidence that adherens junctions of epithelial cells harbor lamellipodia-like, dynamic actin-driven membrane folds, which exhibit high negative membrane curvature at their distal edges. BioID proteomics and imaging experiments demonstrated that MTSS1 associates with an Arp2/3 complex activator, the WAVE-2 complex, in dynamic actin-rich protrusions at cell-cell junctions. Inhibition of Arp2/3 or WAVE-2 suppressed actin filament assembly at adherens junctions, decreased the dynamics of junctional membrane protrusions, and led to defects in epithelial integrity. Together, these results support a model in which membrane-associated MTSS1, together with the WAVE-2 and Arp2/3 complexes, promotes the formation of dynamic lamellipodia-like actin protrusions that contribute to the integrity of cell-cell junctions in epithelial monolayers.

Keywords: Arp2/3 complex; BAR domain; WAVE complex; actin; adherens junction; cell compartmentalization; epithelial cell; lipid-protein interaction; membrane curvature; protein self-assembly.

Figure 1

The I-BAR domain protein MTSS1 localizes to actin-rich foci at intercellular junctions.A, colocalization of MTSS1 (magenta) with Alexa Fluor 488 phalloidin-labeled F-actin (green) in polarized MDCK cells stably expressing HA-tagged MTSS1 (left), observed using confocal microscopy. The boxed regions are enlarged and shown (right). MTSS1 coclusters with F-actin (indicated by white) at adherens junctions (The scale bar represents 10 μm). B, immunofluorescence staining of the colocalization of MTSS1 (magenta) with nonmuscle myosin IIA (green) in polarized MDCK cells stably expressing HA-tagged MTSS1 (left), observed using confocal microscopy. The boxed regions are enlarged and are shown (right). Myosin II localizes further away from the intercellular junctions compared to MTSS1 at adherens junctions (The scale bar represents 10 μm). C, normalized fluorescence intensities along the boxed region in the model diagram (top) indicate that MTSS1 (middle) but not myosin II (bottom) coclusters with F-actin at adherens junctions, as shown by peaks overlapping at the same locations along the line profiles. D, normalized fluorescence intensities along the boxed region in the model diagram (top) indicate that myosin II (bottom, FWHM = 1.41) localizes further away from the intercellular junctions compared to MTSS1 (middle, FWHM = 0.93) at adherens junctions. The line profiles were fitted with Gaussian curves, and FWHM (Full Width at Half Maximum, μm) was estimated. E, MDCK cells expressing EGFP-tagged LifeAct and generating actin protrusions in a confluent monolayer. The boxed region is enlarged and displayed at below. The normalized fluorescence intensity of the line profile along adherens junctions indicates that membrane protrusions are generated from the F-actin clusters. Each number from one to five in the line profile corresponds to the number of each F-actin cluster at cell–cell junctions (The scale bar represents 10 μm). I-BAR, inverse BAR; MDCK, Madin-Darby canine kidney; MTSS1, Metastasis-suppressor 1.

Figure 2

MTSS1, F-actin, and E-cadherin colocalize at adherens junctions of epithelial cells.A, immunofluorescence images of MDCK cell monolayers stably expressing HA-tagged MTSS1. Adherens junctions were visualized with anti-E-cadherin antibody and F-actin with fluorescent phalloidin (The scale bar represents 10 μm). B, magnifications (from the regions indicated by yellow boxes in panel A) shown as composites and as orthogonal y-z and x-z projections along the yellow lines in the upper panel (The scale bar represents 2 μm). C, immunofluorescence images of MDCK cells stably expressing HA-tagged MTSS1. Tight junctions were visualized with anti-ZO-1 antibody and F-actin with fluorescent phalloidin (The scale bar represents 10 μm). D, magnifications (from the regions indicated by yellow boxes in panel C) shown as composites and orthogonal y-z and x-z projections along the yellow lines in the upper panel (The scale bars represent 2 μm). MDCK, Madin-Darby canine kidney; MTSS1, Metastasis-suppressor 1.

Figure 3

Adherens junctions contain highly curved lamellipodia-like membrane protrusions.A, the highly curved membrane structure (light blue) at cell–cell junctions in polarized MDCK cells imaged by serial block-face scanning electron microscopy (SBEM, left). The modeled membrane shown in the left image is rotated approximately 90 degrees (right) (The scale bar represents 500 nm). B, the immunoelectron microscopy image of polarized MDCK cells stably expressing HA-tagged MTSS1. Sections were labeled with the HA antibody against MTSS1 and then labeled with gold particles indicated with arrows (The scale bar represents 2 μm). C, the 3D reconstruction of membrane structures at cell–cell junctions in polarized MDCK cells, indicated in light and dark blue, respectively, by electron tomography (left). Neighboring membranes are enlarged and viewed from the side to show the lamellipodia-like highly curved membrane structures (middle and right) (The scale bar represents 500 nm). D, the width, length, and thickness of the lamellipodia-like membrane protrusions at cell–cell junctions are defined in the electron tomographic model (yellow) and quantified in the images taken by SBEM. Center lines show medians; box limits indicate 25th and 75th percentiles as determined by R software; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles (52). n = 22 (width), 22 (length), and 361 (thickness). E, live imaging of MDCK cells coexpressing EGFP-tagged full-length MTSS1 (green) and mCherry-tagged LifeAct (magenta) in a confluent monolayer, observed by confocal microscopy. The boxed regions are enlarged and shown (right). A kymograph was generated along the line profile (length = 4 min 49.795 s). Note that MTSS1 localizes closer to the leading edge of the cells than F-actin in lamellipodia-like membrane protrusions at cell–cell junctions (The scale bar represents 5 μm). MDCK, Madin-Darby canine kidney; MTSS1, Metastasis-suppressor 1.

Figure 4

Identification of MTSS1 interaction partners at adherens junctions.A, MTSS1 interaction partners at adherens junctions in polarized MDCK cells stably expressing BirA-tagged MTSS1 were identified using proteomics approach, BioID. Protein-protein interaction network analysis using stringApp in Cytoscape was performed in the result obtained from the BioID. Protein complexes (i.e., WAVE-2 and ENAH-VASP-RAPH1 complexes) in the protein-protein interaction network were identified using MCODE in Cytoscape. B, functional analysis of the identified protein–protein interactions by PANTHER indicates that a large number of proteins identified as the putative interaction partners of MTSS1 function close to the membrane by forming protein complexes. C, immunofluorescence staining of the colocalization of WAVE-2 (green) with MTSS1 (magenta) and Alexa Fluor 488 phalloidin-labeled F-actin (gray) in polarized MDCK cells (The scale bar represents 10 μm), observed using confocal microscopy. Note that MTSS1, WAVE-2, and F-actin cocluster at adherens junctions. MDCK, Madin-Darby canine kidney; MTSS1, Metastasis-suppressor 1.

Figure 5

The Arp2/3 complex is important for the integrity of adherens junctions.A, TEM micrographs of cell–cell junctions in polarized MDCK cells treated with DMSO (control) or CK-666 (Arp2/3 complex specific inhibitor). Note that the intercellular spaces (arrows) were observed in Arp2/3 complex–inactivated cells. Distances of the clearly identified intercellular spaces between adjacent MDCK cells treated with DMSO (control) or CK-666 were measured (right). Center lines show medians; box limits indicate the 25th and 75th percentiles as determined using R software; whiskers extend 1.5 times the interquartile range from the 25th and 75th percentiles. To facilitate comparison with the DMSO control, the Y-axis range was adjusted to display the interquartile range (IQR) clearly. n = 146 (DMSO) and 146 (CK-666) (The scale bar represents 2 μm). B, kymographs were generated along line profiles in DMSO- (control) or CK-666–treated MDCK cells expressing EGFP-tagged LifeAct (a marker to visualize F-actin) in a confluent monolayer, observed by time-lapse imaging using confocal microscopy. Control cells produce dynamic membrane protrusions, whereas Arp2/3 complex–inactivated cells have less membrane protrusions (The scale bar represents 10 μm). The number of membrane protrusions at cell–cell junctions was quantified in DMSO- (control) and CK-666–treated MDCK cells (right). n = 10 (DMSO) and 10 (CK-666) measurements. C, MDCK cells expressing EGFP-tagged actin were treated with DMSO (control) or CK-666, and actin dynamics were analyzed by fluorescence recovery after photobleaching (FRAP). The boxed regions (ROI) were photobleached, and fluorescence recovery was measured by confocal microscopy. n = 15 (DMSO) and 17 (CK-666). Note that the actin turnover in cell–cell junctions was suppressed in Arp2/3 complex–inactivated cells. (The scale bar represents 10 μm). D, F-actin intensities in CRISPR/Cas9 control and Wasf2 (gene that encodes WAVE-2) KO polarized MDCK cells mixed with each other. The left panel demonstrates the identification of Wasf2 KO cells by immunofluorescence staining with a WAVE-2 specific antibody, and the right panel shows the intensity of Alexa Fluor 488 phalloidin-labeled F-actin in these cells. Ratios (junction/cytoplasm) of F-actin fluorescence intensities at adherens junctions in CRISPR/Cas9 control, and Wasf2 KO–polarized MDCK cells indicate decreased F-actin assembly at adherens junctions when WAVE-2 is absent. n = 31 (WT) and 36 (KO). (The scale bar represents 10 μm). E, live imaging of F-actin at cell–cell junctions in EGFP-tagged LifeAct-expressing MDCK cells in a confluent monolayer in which Wasf2 was knocked out by CRISPR-Cas9, observed using confocal microscopy. The kymograph was generated along the line profile. Note that fewer actin protrusions were observed at cell–cell junctions in WAVE2-inactivated cells (The scale bar represents 10 μm). Frequencies of actin protrusions at cell–cell junctions were quantified in CRISPR/Cas9 control or Wasf2-KO MDCK cells. n = 10 (Control) and 10 (Wasf2 KO). MDCK, Madin-Darby canine kidney; TEM, transmission electron microscopy.