A syndecan-4 hair trigger initiates wound healing through caveolin- and RhoG-regulated integrin endocytosis

Abstract

Cell migration during wound healing requires adhesion receptor turnover to enable the formation and disassembly of cell-extracellular matrix contacts. Although recent advances have improved our understanding of integrin trafficking pathways, it is not known how extracellular ligand engagement controls receptor dynamics. Using atomic force microscopy, we have measured cell avidity for fibronectin and defined a mechanism for the outside-in regulation of α(5)β(1)-integrin. Surprisingly, adhesive strength was attenuated by the syndecan-4-binding domain of fibronectin due to a rapid triggering of α(5)β(1)-integrin endocytosis. Association of syndecan-4 with PKCα was found to trigger RhoG activation and subsequent dynamin- and caveolin-dependent integrin uptake. Like disruption of syndecan-4 or caveolin, gene disruption of RhoG in mice was found to retard closure of dermal wounds due to a migration defect of the fibroblasts and keratinocytes of RhoG null mice. Thus, this syndecan-4-regulated integrin endocytic pathway appears to play a key role in tissue repair.

Graphical abstract

Figure 1

Engagement of Syndecan-4 Reduces Cell Avidity for Fibronectin (A) Individual, cantilever-mounted fibroblasts were contacted with fibronectin fragment-coated surfaces and then retracted. (B) During retraction, deflection of the calibrated cantilever was recorded to obtain measurements of the maximum applicable force, number of individual tethers, and the total work required to detach the cell from the substrate. (C) ELISA analysis of coated ligands to ensure equivalent density of integrin-binding motifs between the integrin-binding fragment of fibronectin (50K, closed bars) and fibronectin (open bars). (D) Unbinding force curves as a fibroblast was withdrawn from 50K or fibronectin in the absence or presence of a soluble syndecan-4-binding fragment of fibronectin (H/0). (E and F) Quantification of the energy (E) or maximum force (F) required to detach a fibroblast. (G) H/0, preincubated with 10 μg/ml heparin, was used as an inactivated syndecan-4 ligand. (H–J) The specific role of syndecan-4 was tested by comparing responses of syndecan-4 null MEFs (H), wild-type MEFs (I), and null MEFs reexpressing the syndecan-4 cDNA (J). (K) Individual unbinding energy measurements as fibroblasts were contacted sequentially with 50K (circles), fibronectin (crosses), and then 50K again. Graph depicts pooled data from three separate cells. (L) Time course of unbinding energy from 50K upon addition of soluble H/0. All values represent at least 80 measurements per condition from five to nine experiments. Error bars represent standard error of the mean; significance was tested by ANOVA. See also Figure S1.

Figure 2

Syndecan-4 Causes Endocytosis Rather than Inactivation of Integrin (A) Frequency distribution of the force required to unbind individual tethers as a cell is retracted from 50K (black), fibronectin (gray), or 50K in the presence of H/0 (white). The graph represents 337 unbinding steps from a single cell. The experiment was repeated on four separate occasions. (B) Average force required during individual unbinding steps. Values represent 1069 unbinding events from four separate experiments. (C) Number of unbinding steps per retraction curve. Values represent the average of 40 curves per condition from 4 separate experiments. (D and E) Response of cells to syndecan-4 engagement after surface integrin is forced into the active conformation using integrin-activating antibodies: 9EG7 (D) or TS2/16 Fab fragment (E). (F–H) Time-lapse TIRF imaging of β₁-integrin-GFP in the adhesion plane, comparing 5 min prestimulation with 5 min poststimulation when cells were treated with H/0 (F) or H/0 complexed with heparin (G). Scale bars represent 5 μm. (H) Intensity change of 20 adhesions per cell, 9 cells per condition. Error bars represent standard error of the mean; significance was tested by ANOVA. See also Movie S1.

Figure 3

Syndecan-4 Engagement Drives Reduction of Cell Avidity through Dynamin-Dependent, Caveolin-Dependent Endocytosis of α₅β₁-Integrin (A, B, D, and E) The effect of syndecan-4 engagement on unbinding energy using fibroblasts transfected with control (Ctrl) (A), dynamin-2-targeted (Dnm2) (B), clathrin heavy-chain-targeted (D), or caveolin-1-targeted (E) siRNA. Western blots compare expression of appropriate molecules between control and siRNA-transfected populations. Knockdown of caveolin-1 caused concomitant loss of caveolin-2, as previously reported (Razani et al., 2001). (C) Unbinding energy of fibroblasts treated with soluble H/0, and subsequently treated with the dynamin inhibitor MiTMAB to block the endocytic signal. (F) Time-lapse TIRF imaging of β₁-integrin-GFP in the adhesion plane, comparing MEFs transfected with control or caveolin-targeted RNAi oligos. Images represent 5 min pre- or post-H/0 stimulation. Scale bars represent 5 μm. Values represent at least 25 measurements per condition from 4 separate experiments. TIRF analysis summarizes intensity changes of 20 adhesions per cell, 9 cells per condition. Error bars represent standard error of the mean; significance was tested by ANOVA. See also Figure S2 and Movie S2.

Figure 4

Syndecan-4 Engagement Triggers Biphasic Redistribution of α₅β₁-Integrin between the Plasma Membrane and Vesicles (A) Fibroblast membranes were separated by centrifugation into a plasma membrane fraction (Mem) and a small vesicle/soluble (Sol) fraction. Segregation was verified by blotting for vesicle markers (EEA1 and Rab4), soluble marker (tubulin), and plasma membrane marker (Na⁺ K⁺ ATPase transporter). Int, integrin. (B) Western blot analysis of the redistribution of molecules when fibroblasts prespread on 50K were stimulated with H/0 over a time course. (C and D) Quantification of plasma membrane-associated β₁-integrin (C) and caveolin-1 (Cav-1) (D). (E) Redistribution of plasma membrane-associated β₁-integrin when fibroblasts in suspension were stimulated with H/0. Error bars represent standard error of the mean of experiments repeated on four to eight separate occasions; significance was tested by ANOVA.

Figure 5

Regulation of PKCα and RhoG Is Responsible for Caveolin-Dependent Endocytosis in Response to Syndecan-4 Engagement (A and B) The effect of syndecan-4 engagement on unbinding energy using MEFs transfected with PKCα-targeted siRNA (A) or expressing a PKCα-binding mutant of syndecan-4 (B). The sequence of signals was resolved by direct activation of PKCα with 100 nM PMA. (C–E) PMA treatment (C), control vehicle solvent (D), and PMA treatment (E) of fibroblasts transfected with caveolin (Cav)-targeted siRNA. (F) Time course of H/0-stimulated activation of RhoG by effector pull-down assay, followed by quantitative western blotting. Gels represent six separate experiments. (G–I) The effect of syndecan-4 engagement on unbinding energy (G), removal of β₁-integrin-GFP from the TIRF plane (H), and association of β₁-integrin with the plasma membrane (I), following reduction of RhoG expression by RNAi. Scale bars represent 5 μm. All atomic force measurements represent at least 20 measurements per condition, obtained on 3 separate occasions. (J) Multimolecular complexes associated with active (12G10, white bar) or inactive (mAb13, black bar) β₁-integrin or transferrin receptor (OKT9) were isolated from fibroblasts expressing GFP-RhoG; recruitment of GFP-RhoG and caveolin relative to β₁-integrin was determined by quantitative western blotting. Blots were reprobed for talin as a control for selective recruitment to active integrin. (K) H/0-regulated multimolecular complexes associated with β₁-integrin-GFP were isolated by GFP-Trap from MEFs spread on 50K. Atomic force measurements represent at least 25 measurements per condition from at least 3 separate experiments. Gels represent five separate experiments. Error bars represent standard error of the mean; significance was tested by ANOVA. Arrowheads mark immunoglobulin bands. Ctrl, control; Tub, tubulin. See also Figure S3 and Movie S2.

Figure 6

RhoG Expression Is Required for Efficient Wound Closure (A) Closure of 4 mm punch wounds in 7-week-old mice was compared between Rhog^−/− mice (cross), wild-type (closed triangle), and heterozygous (open square) littermates. (B) Trichrome staining of skin sections to show comparable tissue morphology but compromised wound contraction of Rhog^−/− mice; squares indicate the regions analyzed for myofibroblast recruitment. (C–G) Myofibroblast recruitment was measured by α-smooth muscle actin staining (C), generation of intensity threshold masks (D), and automated quantification of positive cells (per 0.23 mm² image) (E) from wild-type and Rhog^−/− mice. Density of macrophages was compared between unwounded (F) and wounded skin (G) of wild-type and Rhog^−/− mice. Data represent 14 wounds per genotype. Error bars represent standard error of the mean; significance was tested by ANOVA. Scale bars represent 100 μm.

Figure 7

Migration of RhoG or Syndecan-4-Defient Fibroblasts over a Fibrillar Matrix (A–L) Migration paths (A, E, and H), average speed (C, F, I, K, and L), and average persistence (= displacement/total distance moved) (D, G, and J–L) of cells migrating over a cell-derived matrix. Gray boxes indicate the minimum possible persistence values when cells migrate randomly on homogeneous matrix. Cells tracked were as follows: immortalized human fibroblasts transfected with control or RhoG-targeted siRNA (A–D), including analysis of expression of RhoG by western blot (B); primary E13.5 MEFs from Rhog^−/− mice, wild-type, and heterozygous littermates (E–G); immortalized MEFs from wild-type and Sdc4^−/− littermates and MEFs rescued by endogenous syndecan-4 expression (H–J); and wild-type (K) and Sdc4^−/− (L) MEFs following transfection with caveolin- or RhoG-targeted siRNA. Data represent analysis of over 100 cells per condition, from 3 separate experiments. (M) Ten hour scratch assay of primary keratinocytes isolated from neonatal Rhog^−/− mice, wild-type, and heterozygous littermates. Error bars represent standard error of the mean; significance was tested by Kruskal-Wallis tests for nonparametric data. Scale bar represents 100 μm. Arrowheads mark immunoglobulin bands. See also Figure S4 and Movies S3 and S4.