ICAM-1 regulates the size of apical BCs in polarized HepG2 cells

(A) ICAM-1 concentrates in bile canaliculi (top images) and bile ducts (bottom images) from polarized hepatocytes and cholangioncytes, respectively (arrows), and in sinusoids in human livers from healthy donors. Right images show a two-fold enlargement of the boxed areas in left images. Inset cartoon represents the basic organization of polarized hepatocytes at the hepatic lobule. Their apical membranes form bile canaliculi and their basolateral membranes face the sinusoids. Scale bars, 10 μm (B) CRISPR-CAS9 mediated editing of the ICAM1 gene (ICAM-1_KO) in HepG2 cells. Several clones were selected and pooled to prevent clonal variations. WT, parental wild type cells (C-D) Polarized human epithelial HepG2 cells form bile canalicular structures (BCs). Control WT and ICAM-1_KO HepG2 cells were cultured on coverslips for 72 h, fixed and stained for F-actin and ICAM-1, which concentrate in apical BCs (C), for ZO-1 and F-actin in cells expressing MDR1-GFP (D, top images) and for CD59, ZO1 and F-actin (D, bottom images). Scale bars, 20 μm. (E) Plots represent the mean ± SD. BC area increased from 18.9 + 1.1 μm2 in WT cells to 51.2 + 7.2 μm2 in ICAM-1_KO cells. ***p<0.001; ns, not statistically significant. BC area and frequency were quantified in at least 70 cells per experiment. Nuclei were stained with DAPI. (F) WT and ICAM-1_KO cells were transfected with GFP or ICAM-1-GFP expression plasmids, grown for 48 h and processed for immunofluorescence and confocal microscopy analysis (left images) or lysed for western blot analysis (central images). Single channels from the boxed areas are enlarged on the right of the corresponding image. The right panel shows the quantification of BC areas and represents the mean ± SD. * p<0.05. n=3. Scale bar, 10 μm. (G) Cryo-SXT of WT and ICAM-1_KO cells stably expressing GFP-Rab11 to localize BCs. Cells were cultured on holey carbon grids for TEM for 48-72 h. BCs were identified in live cells with an epifluorescence microscope (dotted lines) and immediately vitrified and cryopreserved for cryo-SXT. SXT images show a slice the reconstructed tomogram from the boxed areas in the correlative Epifluo-SXT images. GFP-Rab11 panels display the epifluorescence images of the same area. Scale bars, 2 μm.

ICAM-1 regulates BC dynamics in polarized HepG2 cells.

(A) Correlative cryo-SXT of ICAM-1_KO cells stably expressing GFP-Rab11 showing two fusing BCs. The SXT image shows an slice of the reconstructed tomogram from the boxed area in the correlative Epifluo-SXT image (top left). The GFP-Rab11 panel displays the epifluorescence image of the same area. Scale bar, 2 μm. (B) HepG2 cell stably expressing MDR1-GFP were subjected to time-lapse fluorescence microscopy analysis. White arrows point at MDR1- GFP-positive BCs. Red arrows point at two fusing MDR-1-GFP-positive BCs. Central panel shows the quantification of BCs in which fusing events were observed during the time- lapse microscopy assays. Right panel quantifies variations in BC area between consecutive frames, expressed as positive values. Scale bars, 10 μm (C) WT and ICAM-1_KO hepatic cells were cultured for at least 96 h, fixed and stained for the indicated proteins. Scale bar, 20 μm. The left scatter-plot shows the quantification of BC area of cells cultured for the indicated periods and represents the mean ± SD. *p<0.05; ***p<0.001;****p<0.0001. 50 BCs per experiment, n=3. (D) Distribution of F-actin and ZO-1 in ICAM-1_KO cells cultured for 96 h suggest coalescence of small BCs into enlarged (e)BC. Enlarged areas show F-actin- enriched BCs (arrows) in contact with massive eBCs. Bottom right images show a graphical representation of the top right images. Scale bar, 10 μm. (E) ICAM-1_KO cells were grown for 96 h and incubated with sulfo-NHS-biotin for 30 min at 4°C, washed, fixed and permeabilized. Biotinylated surface proteins were detected with TRITC-conjugated streptavidin. XZ stack projections are shown to visualize the relative localization of surface proteins, accessible to sulfo-NHS-biotin from the extracellular milieu, and F-actin-rich BCs (left images) and eBC (right images), which are sealed and not accessible from the extracellular milieu. Nuclei were stained with DAPI. Bottom images show a graphical representation of the top images. Scale bars, 10 μm.

Upregulation of ICAM-1 expression in response to inflammatory cytokines reduces BC frequency.

(A-D) Polarized hepatic epithelial cells were stimulated or not for 48 h with 50 ng/ml TNF-α, 15 ng/ml IL-1β and IFN-γ 1000 U/ml as indicated. (A) cells were lysed and ICAM-1 expression levels detected by western blot. Tubulin was blotted as a loading control. (B-D) Cells were fixed and stained for ICAM-1, F-actin and nuclei. (B) Basolateral-to-apical ratio of ICAM-1 staining intensities. (C) Representative confocal images of the effect of cytokine treatment. Scale bar, 10 μm. (D) Quantification of BCs (left) and BC area (right). Bars represent the mean ± SD. n>3. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

ICAM-1 controls canalicular membrane expansion by regulating a distal canalicular actomyosin ring.

(A,B) BCs contain a ring of actomyosin. (A) Polarized HepG2 cells expressing GFP-MLC were incubated with SirActin for 2 h and subjected to time-lapse confocal microscopy. Left images show the image projection of the indicated frames. Right images show a single image of the stack corresponding to t=0 of the same time-lapse movie (B) Polarized hepatic epithelial cells were cultured on coverslips for 72 h, fixed and stained for F-actin, ICAM-1 and phosphorylated myosin light chain (pMLC) (left images). Central panel. Relative distribution of the staining intensity for the indicated proteins, starting from the distal parts of BCs, which contain actomyosin, towards the centers of the BCs, which contain the microvillar tips. Note that p-MLC staining is mostly distal, F-actin distribute all along the profile and ICAM-1 is mostly found in microvilli. Right panel. Manders’ analysis for the indicated pairs of staining. Scale bar, 3 μm (C,D) Control WT, ICAM-1_KO and ICAM- 1_KO cells expressing ICAM-1-GFP were cultured on coverslips for 72 h, fixed and stained for pMLC and F-actin (C) and for F-actin and non-muscle myosin heavy chain-IIb (MHCIIb) (D). Scale bar, 5 μm. (E) Quantification of relative apical and basolateral levels of F-actin, pMLC and MHCIIb detected in (C) and (D). n=3. At least 20 BCs (apical) or 20 cells (basolateral) were quantified in each experiment. (F) WT cells were exposed or not to 10 μM of the myosin inhibitor blebbistatin for the indicated times. Scale bar, 10 μm. Plots show the quantification of the effect of blebbistatin on BC size (top) and BC frequency (bottom). n=3. At least 20 BCs were quantified in each experiment. (G) WT and ICAM-1_KO cells were cultured, stimulated with IL-1β, fixed and stained as in Figure 3. When indicated, cells were incubated with 10 μM blebbistatin for the last 40 h. BC frequency was quantified. Quantifications represent the mean ± SD. n>3. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001. Except otherwise indicated, scale bars correspond to 5 μm.

ICAM-1 signals to actomyosin thereby reducing apicobasal polarity.

(A) Surface ICAM-1 engagement reduces BC morphogenesis. ICAM-1-GFP HepG2 cells were cultured for 24 h on coverslips precoated with the indicated antibodies (α), IgG control or fibronectin (FN). Cells were fixed and stained for F-actin to quantify the amount of BCs per cell (right). Nuclei were stained with DAPI. Bars represent the mean ± SD. n=3. Scale bar, 50 μm. TfR, Transferrin receptor. (B) Impairment of BC morphogenesis depends on myosin-II. ICAM-1-GFP cells were cultured for 24 h as in (A) on coverslips precoated with the indicated proteins. Cells were seeded for 4 h and incubated with blebbistatin when indicated (blebb.). As a control, ICAM-1_KO cells were plated in parallel on ICAM-1. Cells were fixed, stained for F-actin and pMLC and the amount of BCs per cell was quantified (right graph). Nuclei were stained with DAPI. Bottom images are enlargements of the boxed areas in the top images showing cells plated on αICAM-1 coverslips. Bars represent the mean ± SD. n=3. Scale bar, 20 μm; enlarged boxed area 10 μm. Note that blebbistatin inhibits myosin but does not reduce pMLC levels, as previously described. Scale bar, 10 μm. (C) Quantification of the stress fibers (top) and stellate stress fibers (bottom) in cells plated on the coverslips precoated with the indicated proteins as in (b). n= 4. (D) Stellate stress fibers concentrate F-actin and pMLC but not the canalicular marker MDR1. Hepatic epithelial cells expressing MDR1-GFP were plated on coverslips precoated with FN or anti- ICAM-1 antibody (αICAM-1). (E) Effect of ICAM-1 clustering (X-ICAM-1) on MLC phosphorylation. Cells were cultured for 24 h on plastic dishes and then were sequentially incubated with anti-ICAM-1 antibody (1st Ab) for 30 min and an specific secondary antibody (2nd) for the indicated times. Cells were lysed and proteins were detected by western blot. Bottom plot shows the quantification of pMLC levels upon receptor clustering. (F) Hepatic epithelial cells stably expressing ICAM-1-GFP were cultured for 72 h and then incubated with T-lymphocytes (1 T-cell : 2 hepatic cells) for the indicated periods. The plot shows the percentage of BCs quantified prior to exposure to T-lymphocytes (0 h). Bars represent the mean ± SD. n=5. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001. (G) Current model for the effect of ICAM-1 signaling on hepatic epithelial cells. ICAM-1 engagement signals through the Rho-ROCK axe and increases the formation of actomyosin fibers that prevent the formation of actomyosin-regulated BC structures. In polarized hepatic cells, leukocyte-mediated engagement of ICAM-1 activates pericanalicular actomyosin, which induces contraction at BCs, thereby reducing their frequency.

Proximal interaction of ICAM-1 with EBP50/NHERF1/SLC9A3R1 into nano-scale microvillar domains

(A). The ICAM-1 BioID interactome reveals the proximal interaction of ICAM-1 with EBP50. Parental polarized HepG2 cells (WT) and HepG2 cells stably expressing ICAM-1-BirA* were incubated with 50 μM biotin for 16 h, lysed and subjected to a pull- down (PD) assay with neutravidin-agarose. Western blots show ICAM-1 and EBP50 biotinylation. Exo70 and ERK are shown as negative controls. ICAM-1 and ICAM-1-BirA* were detected with anti-ICAM-1 antibody. (B) Polarized hepatic epithelial cells were fixed and stained for ICAM-1 and EBP50. BC, bile canaliculi. The Manders’ analyses for the indicated pairs of staining show the remarkable co-localization between ICAM-1 and EBP50. Scale bar, 5 μm. (C) EBP50 is localized in murine hepatic sinusoids (S) and in the liver bile canaliculi (red arrows). Scale bar, 50 μm. (D) Triple staining of EBP50, F-actin and pMLC in BCs. Central panel. Relative distribution of the staining intensity for the indicated proteins, starting from the distal parts of BCs, which are enriched in actomyosin, towards the centers of the BCs, which contain the microvillar tips. Right panel. The Manders’ analyses for the indicated pairs of staining show the remarkable co-localization of EBP50 with F-actin, but not with pMLC. Scale bars, 3 μm. (E) Hepatic epithelial cells were transfected with the indicated expression vectors coding for FLAG-tagged full-length EBP50 or its N-termina PDZ domains. 24 post-transfection, cells were lysed, subjected to immunoprecipitation with anti-FLAG antibodies and the immunoprecipitates and the lysates were analyzed by western blot for the indicated proteins. (F,G) Polarized hepatic epithelial cells were stained for the indicated proteins and analyzed by STED super-resolution microscopy. Only two different fluorophores could be subjected to simultaneous STED. (F) STED analysis of ICAM-1 and EBP50. Central and right images are a 2.5-fold enlargement of the boxed area in the left image, which corresponds to BCs. Bottom images are a 2-fold enlargement of the boxed area in top central and right images, which corresponds to a canalicular microvilli-rich area. (G) STED analysis of F-actin and EBP50 in a BC. Bottom images are a 2-fold enlargement of the boxed area in the top image, which corresponds to a canalicular microvilli-rich area. Scale bars, 5 μm. (H) STED analyses of non- polarized epithelial cells exposed to anti-ICAM-1 for 30 min in the cold and then incubated at 37°C for the indicated times. ICAM-1 and EBP50 did not overlap at t=0 in these non- canalicular regions. However, 90 min at 37°C with anti-ICAM-1 antibody induced a redistribution of EBP50 into ring-shaped macroclusters that overlapped with ICAM-1 aggregates. Scale bars, 5 μm. (I) Manders’ analyses for the indicated pairs of staining from (F) (apical) and (H) (non-polarized). (J) Polarized WT and ICAM-1_KO cells were fixed, stained and the intensity of EBP50 at the apical BCs was quantified by confocal microscopy (right). (K) Polarized WT and ICAM-1_KO cells were fixed, stained and analyzed by STED confocal microscopy. Intensity and nearest neighbor distance (nnd) of detected spots at BCs and eBCs were calculated.

EBP50 regulates ICAM-1-mediated signaling.

(A,B) HepG2 cells were transfected with siRNA control (siControl) or with siRNA targeting EBP50 (siEBP50) and cultured for 72 h. Cells were lysed and analyzed by western blot for the indicated antibodies (A) or fixed and stained for the indicated proteins (B). BC frequency and area and the intensity levels of the indicated protein at BCs were quantified. Bars represent the mean ± SD of at least 3 experiments. Scale bar, 10 μm. (C) HepG2 cells stably expressing ICAM-1-GFP were cultured for 24 h on coverslips precoated with anti-ICAM-1 antibody, fixed and stained for F-actin and EBP50. Central and right images are an enlargement of the boxed area in the left confocal image, which shows an region enriched in stress fibers and stellate stress fibers. Scale bar, 5 μm. (D) HepG2 cells were transfected with siControl or siEBP50 for 48 h, then cultured for 24 h on coverslips precoated with anti-ICAM-1 antibody. Cells were fixed and stained for F-actin and EBP50. Central and right images are enlargements of the boxed areas in the left images, which show regions enriched in stress fibers and stellate stress fibers. Arrows point at BCs. (E) Quantification of the number of BC per cell (top) and the percentage of stress fibers per cell with respect to siControl-transfected cells (bottom) of experiments shown in (i). Bars represent the mean ± SD of at least 4 experiments. *p<0.05. **p<0.01. ***p<0.001. ****p<0.0001. A.U. Arbitrary Units. Nuclei were stained with DAPI.

ICAM-1 and non muscle myosin-II regulate the size of canalicular-like cavities in hepatic organoids.

(A) ICAM-1 concentrates in bile canaliculi and bile ducts in livers of LPS- stimulated mice (arrows). S, sinusoids. Scale bars, 20 μm (left image) and 10 μm (right image). (B) ICAM-1, F-actin and nucleus staining of livers from WT and ICAM-1_KO mice. Scale bar, 20 μm. Right graph. Quantification of BC width from WT and ICAM_KO murine livers identified morphologically by F-actin staining more than 80 bile canaliculi from four WT and four ICAM-1_KO mice. (C) Immunohistochemical analyses of F-actin and MRP2 in bile canaliculi in WT and ICAM-1_KO mice. Insets show enlargements of the boxed areas. Scale bar, 6 μm. (D) Mouse liver organoids have a spherical morphology. Differentiated hepatic organoids were fixed and stained for the indicated antiboidies and for F-actin and nuclei to visualize cell morphology. A Z-stack projection of an organoid is shown in left an central image. A single confocal image is shown on the central right image. Enlargement of the boxed areas are shown on the right. Scale bar, 100 μm. (E) Mouse liver organoids were treated with 25 μM CFDA for 30 min. Left panel shows the Z-stack projections of different confocal planes of an organoid. Central and right panels show the distal and central confocal planes, respectively, of the organoid in the left images. Right images show a two- fold magnification of the boxed areas. Scale bars, 100 μm. (F) Mouse liver organoids were fixed and processed for transmission electron microscopy. Cells organized in sheets facing big central lumens (central left image) but also formed cavities between their lateral membranes containing cell-cell junctions and microvilli (lateral lumens) (top and central right image, bottom cartoon). Scale bars, 5 μm. (G) Mouse (top) and human (bottom) liver organoids were fixed, permeabilized and stained for ICAM-1, F-actin and nuclei. Left images show a confocal plane which crosses the central part of a cell sheet. F-actin and ICAM-1 accumulations were detected between cells (arrows). ZO-1 staining clearly surrounded these accumulations in the mouse organoid (bottom). Black lines on the side of left images mark the location of the X-Z reconstructions shown in the right images. Encircled regions show lateral areas of ICAM-1 accumulation. (H) Ultrastructural analysis of differentiated WT and ICAM-1_KO liver organoids. Quantification of lateral lumen areas for organoids from two WT and five ICAM-1_KO mice (right). Areas of more than twenty lateral lumens were analyzed for each mouse. Bars represent the mean ± SD. Scale bars, 5 μm. (I) Functional comparison of WT and ICAM-1_KO organoids. WT and ICAM-1_KO organoids equally processed and secreted CFDA into lateral lumens (top images) and secreted albumin (bottom graph). A.U. arbitrary units. (J) Ultrastructural analysis of differentiated WT liver organoids treated or not with 10 μM blebbistatin for 24 h. Scale bars, 5 μm. The right plot shows the quantification of lateral lumen areas. Quantifications represent the mean ± SD. *p<0.05; **p<0.01; ***p<0.001;****p<0.0001. Nuclei were stained with DAPI in confocal images.