Transforming growth factor beta controls the directional migration of hepatocyte cohorts by modulating their adhesion to fibronectin

Abstract

Transforming growth factor beta (TGF-beta) has a strong impact on liver development and physiopathology, exercised through its pleiotropic effects on growth, differentiation, survival, and migration. When exposed to TGF-beta, the mhAT3F cells, immortalized, highly differentiated hepatocytes, maintained their epithelial morphology and underwent dramatic alterations of adhesion, leading to partial or complete detachment from a culture plate, followed by readhesion and spreading. These alterations of adhesive behavior were caused by sequential changes in expression of the alpha5beta1 integrin and of its ligand, the fibronectin. The altered specificity of anchorage to the extracellular matrix gave rise to changes in cells' collective motility: cohorts adhering to fibronectin maintained a persistent, directional motility, with ezrin-rich pathfinder cells protruding from the tips of the cohorts. The absence of adhesion to fibronectin prevented the appearance of polarized pathfinders and lead to random, oscillatory motility. Our data suggest a novel role for TGF-beta in the control of collective migration of epithelial cohorts.

Figure 1.

Changes in hepatocytes' morphology in response to TGF-β1. Subconfluent cultures of AML12 and mhAT3F cells were treated with TGF-β1 (5 ng/ml) for indicated times and analyzed by low-power phase-contrast microscopy (A and B, top) or fixed and stained for actin (phalloidin-FITC; green) and nuclei (Hoechst 33258; blue) and analyzed by confocal microscopy (B, bottom). Xy and xz planes are shown for the confocal images.

Figure 2.

Cell-cell interactions are conserved in mhAT3F after the TGF-β1 treatment. Subconfluent cultures of mhAT3F and AML12 cells were treated with TGF-β1 (5 ng/ml) for indicated times. RNA was assayed by RTqPCR for E-cadherin (A) and Snail (B) expression and normalized by expression of GAPDH. Mean ± SEM from three independent experiments is shown. Statistical significance was calculated relative to the 0 h time point; *p ≤ 0.05. Protein expression was analyzed in mhAT3F cells by immunoblotting (C) and protein subcellular localization by confocal immunofluorescence analysis before treatment and after 72 h culture in the presence of 5 ng/ml TGF-β1 (D) The nuclei were counterstained in blue with Hoechst 33258 dye. Xy and xz planes are shown for the confocal images.

Figure 3.

The mhAT3F undergo a cycle of contraction and respreading under the influence of TGF-β. Cells were grown on plastic dishes, treated with TGF-β1 (5 ng/ml), and observed by time-lapse microscopy for 96 h. Representative images of a low-power field and threefold enlargement of one cluster (indicated by an arrow in the top panel). Well-spread monolayer cell clusters (t = 0 h) contract soon after treatment (t = 12–48 h). A large proportion of clusters resume spreading at later time points (t = 72–96 h).

Figure 4.

The mhAT3F change the pattern of integrin gene expression in response to TGF-β. Exponentially growing cells were treated with TGF-β1 (5 ng/ml). At the times indicated, cells were collected and total RNA was subjected to RTqPCR analysis. The relative abundance of the mRNAs was estimated after normalization with primers specific for GAPDH. ■, TGF-β–treated samples; □, vehicle treated control. Results are presented as mean ± SEM of three independent experiments. Statistical significance was calculated relative to the 0 h time point; *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.005.

Figure 5.

Changes of the transcriptional profile of integrin expression are reflected at the protein level. Exponentially growing cells were treated with 5 ng/ml TGF-β1 or the vehicle only as a control for indicated times. (A) Total cell lysate was analyzed by immunoblotting. (B) The Western blots were scanned, and the signal intensity was quantified using the Gene Tools 3.06 software (SynGene, Frederick, MD). Data were normalized with respect to β-actin. □, control culture; ■, TGF-β–treated cells. (C) The surface expression of integrin α5 (top panels), β1 (middle panels) integrin subunits and of the extracellular fibronectin (bottom panels) was detected on fixed, nonpermeabilized cells. Nuclei are visualized in blue by the Hoechst 33258 staining. An increase in the surface expression of the α5 is detectable early after the TGF-β treatment, whereas the fibronectin secretion occurs later in the course of response. Orthogonal views of 3D-reconstructed images are also shown.

Figure 6.

Interference with integrin and fibronectin expression modulates the response to TGF-β. Stable populations of mhAT3F cells selected for expression of the control shRNA (shLuc), two independent shRNAs directed against integrin α5 (shα5A and shα5B) or two independent shRNAs directed against fibronectin (shFnA and shFnB) were grown as exponential cultures, treated with 5 ng/ml TGF-β and observed by time-lapse microscopy for 84 h. (A) Representative images of cell clusters undergoing morphological changes in response to treatment at 0, 42, and 84 h of treatment. (B) Five low-power microscopy fields for each experimental condition, each containing 30–40 cell clusters, were analyzed for changes of surface occupancy by cells, measured every 6 h. The initial coverage, similar for all conditions was arbitrarily set at 100%. The data represent the mean of three experiments ± SEM. (C) Dynamics of surface occupancy from experiments in B are represented as slopes (calculated by linear regression) of the curves from 0 to 36 h (shrinking) and 48 to 84 h (spreading). Statistical significance was calculated relative to control shRNA (shLuc); **p ≤ 0.01 and ***p ≤ 0.005.

Figure 7.

α5 integrin subunit is necessary for the maintenance of directional migration in response to TGF-β. Exponentially growing mhAT3F cells, stably expressing the indicated shRNAs, were seeded on a growth factor–reduced matrigel complemented with 40 ng/ml fibronectin, treated with TGF-β1, and analyzed by time-lapse microscopy using the LMC optics. (A) Spontaneous migration of two cell clusters, in the absence of TGF-β1, showing the position of cells at 0, 36, and 72 h and the paths traveled. (B) mhAT3F expressing either firefly luciferase shRNA (shLuc) or integrin α5 shRNA (shα5) were grown on fibronectin-supplemented matrigel and treated with 5 ng/ml TGF-β1 for 2 h, rinsed, and maintained in culture for an additional 72 h. RNA extracted at indicated times was assayed by RTPCR for changes of expression of integrin α5 and fibronectin. Expression of GAPDH served as a control. Velocity (C) and persistence (D) of cell clusters locomotion was analyzed in microscope fields containing ∼300 cell cohorts. Data are presented as the mean velocity (C) or the mean ratio (D) of distance to path over three periods of 24 h. Means ± SEM of four microscope fields for every condition are shown. (E) Representative tracings of paths traveled over 24 h by control (shLuc) or α5 shRNA expressing cells at day 3 after treatment with TGF-β or its vehicle as a control. (F) Data from D are represented as a measure of movement directionality (D/P; see text) for each individual cell cluster. Migration was measured during 24 h and starting 24 h after the TGF-β treatment.

Figure 8.

Protrusion of pathfinder cells with strong membrane activity correlates with the directional movement of cell cohorts. MhAT3F cells expressing either the control (shLuc) or the integrin α5 directed (shα5) shRNA were grown on Matrigel supplemented with fibronectin. The cells were treated with TGF-β for indicated times. (A) Rac1 activation was assayed by a GTPase pulldown assay using PAK1-GST as a bait. (B) The level of ezrin expression was assayed by immunoblotting and normalized by GAPDH expression. (C) Cells in panels a–p were fixed, permeabilized, and stained with an anti-ezrin antibody (green). F-actin was visualized by phalloidin staining (red) and nuclei by Hoechst 33258 (blue). Cells in panels q–t were fixed and labeled with anti-fibronectin antibody (purple), then permeabilized, and stained with phalloidin (red) to reveal cortical actin and Hoechst 33258 (blue) to visualize the nuclei. The cloud of fibronectin partially envelops the cohorts, leaving the protruding cells free of fibronectin staining. The arrows indicate the direction of movement deduced from the position of the protruding cells. Insets, glycoproteins (black), strongly enriched in the Golgi apparatus (arrowheads) were revealed with Alexa Fluor 488–conjugated lectin HPA. Two regions of the cohort (identified by squares) are shown. (D) Ezrin (green) and fibronectin (purple) labeling indicate distinct localization of the markers. Nonpermeabilized cells were first labeled with an anti-fibronectin primary antibody, followed by Cy5-conjugated anti-rabbit secondary antibody. The cells were then permeabilized and labeled with an anti-ezrin antibody and revealed with an anti-rabbit antibody coupled to the FITC fluorochrome. This staining protocol accounts for a high background of green labeling of the extracellular fibronectin. However, the membrane-associated ezrin labeling is fully specific of the anti-ezrin antibody (not shown). (E) A 3D reconstruction (top panel) and a section through a middle of a cohort (bottom panel), showing a distinct ezrin localization (green) in the top part as well as in the protruding cell. Actin is visualized in red and nuclei in blue. (F) A schematic representation of a 3D cohort shown in E. The section (indicated by brackets) reveals the cortical organization of actin, position of the Golgi apparatus, accumulation of ezrin in a subset of cells and the cloud of fibronectin in the rear of the cohort. Note the swelling of the matrix behind the cohort; it can be visualized in the Supplementary Fig 7 video and could be due either to a deformation of the matrix, as depicted in the drawing, or to the tail of fibronectin dragged behind the cohort.