Detection of an Integrin-Binding Mechanoswitch within Fibronectin during Tissue Formation and Fibrosis

Abstract

Fibronectin (Fn) is an extracellular matrix protein that orchestrates complex cell adhesion and signaling through cell surface integrin receptors during tissue development, remodeling, and disease, such as fibrosis. Fn is sensitive to mechanical forces in its tandem type III repeats, resulting in extensive molecular enlongation. As such, it has long been hypothesized that cell- and tissue-derived forces may activate an "integrin switch" within the critical integrin-binding ninth and 10th type III repeats-conferring differential integrin-binding specificity, leading to differential cell responses. Yet, no direct evidence exists to prove the hypothesis nor demonstrate the physiological existence of the switch. We report direct experimental evidence for the Fn integrin switch both in vitro and ex vivo using a scFv engineered to detect the transient, force-induced conformational change, representing an opportunity for detection and targeting of early molecular signatures of cell contractile forces in tissue repair and disease.

Keywords: angiogenesis; antibody phage display; fibronectin; fibrosis; integrins; mechano-biology.

Figure 1

Fibronectin strain drives differential integrin affinity. (A) PyMol structure predictions of engineered recombinant fragments of Fn’s integrin-binding domain. FnIII9*10, represents a stabilized native structure through a Leu-to-Pro point mutation at position 1408. In this conformation, the PHSRN-to-RGD distance is approximately 36 Å. FnIII9–4G-10 is a mutation of the FnII9*10 variant that contains a 4xGly insertion between the 9th and 10th type III repeats. This mutation increases domain separation and the PHSRN-to-RGD distance to approximately 43 Å. These fragments have been employed in the past to predict potential biological consequences of the theorized integrin switch. B) SPR binding characterization of integrins α5β1 and αVβ3 to the FnIII9*10 and FnIII9–4G-10 fragments demonstrate a nearly complete loss of binding of α5β1 upon domain separation, whereas αvβ3 binding is predictably unaffected. Black curves show experimental sensogram traces, and red curves show computationally fitted data. Equilibrium dissociation constants (K_D) are shown above the curves. (C) Fibroblasts cultured on FnIII9*10 and FnIII9-4G-10 fragments and immunostained for integrin α5 and αv demonstrate the fragments’ capability of skewing cellular binding toward specific integrins corroborating SPR analysis. Scale bar is 20 μm. (D) Schematic of cell attachment experiments designed to demonstrate the impact of strain of a decellularized Fn-rich matrix on integrin binding. (E) Quantitation of cell adhesion on strained and relaxed Fn-rich ECM demonstrates that strain of the ECM significantly impacts cell binding via α5β1 but not αvβ3. Error bars reflect SD, *p < 0.05, N = 6, ANOVA with Tukey’s post-test.

Figure 2

H5 antibody recognizes conformational change of the integrin-binding domain by binding an epitope specifically on the ninth type III repeat that can be exposed by denaturation. (A,B) SPR analysis of binding of H5 antibody to recombinant FnIII9*10 (A) and FnIII9-4G-10 (B) fragments demonstrates the conformation selectivity of its binding. H5 binds preferentially to the molecularly extended conformation. Black lines show experimental sensogram traces, and red lines show fitted data. Equilibrium dissociation constants (K_D) are shown above the curves. (C) Competitive binding of the H5 antibody to FnIII9-4G-10 in the presence of soluble FnIII9 (FnIII6-9) and FnIII10 (FnIII10-14) fragments. FnIII9-4G-10 was immobilized on ELISA plates, and the H5 antibody was co-incubated with the indicated soluble Fn fragments prior to incubation with FnIII9-4G-10. N = 8 for each group; error bars reflect SEM. (D,E) Domain mapping studies were performed by SPR using Fn fragments including either the ninth type III repeat (FnIII6-9) (B) or the 10th type III repeat (FnIII10-14) (C). Binding of H5 was only observed by SPR when fragments including the ninth Fn type III repeat were immobilized, suggesting that the epitope for H5 is located within FnIII9. (F) ELISA of H5 binding to full-length Fn and Fn fragments. H5 bound increasingly to increasing amounts of surface-adsorbed FnIII9-4G-10 to a significantly greater degree than FnIII9*10 or full-length Fn at all concentrations (p < 0.0001) (two-way ANOVA with Tukey’s post-test, N = 3). (G,H) Nitrocellulose dot blot (G) and Western blot (H) of H5 binding to full-length Fn and Fn fragments. Under these denaturing conditions, H5 binds FnIII9-4G-10 and FnIII9*10 to a similar degree. (I–K) ELISA (I), dot blot (J), and Western blot (K) of HFN7.1, a commercially available antibody targeting FnIII9-10. In the ELISA, HFN7.1 only bound FnIII9-4G-10 to a significantly greater degree than FnIII9*10 at 1 μM (p < 0.05) and was not significant at higher concentrations (two-way ANOVA with Tukey’s post-test, N = 3). HFN7.1 did not show preferential affinity for FnIII9*10 or FnIII9-4G-10 in the dot blot or Western blot.

Figure 3

H5 antibody is capable of discriminating conformational changes of Fn’s integrin-binding domain (i.e., the integrin switch) in multiple model systems. (A) Staining of H5 on in vitro Fn fibers deposited on PDMS membranes demonstrates the strain-dependent conformation of Fn’s integrin-binding domain. A discrete transition from modest to high binding occurs between an extension ratio of 1.25 and 1.5. N = 10, error bars reflect SD, ***p < 0.001, one-way ANOVA with Tukey’s post-test. (B) Staining of strained or relaxed decelluarized ECM assembled by human foreskin fibroblasts by H5 indicates that the Fn integrin-binding domain within complex (anisotropic) Fn-rich ECM undergoes a conformational change in response to strain. N = 6, error bars reflect SD, *p < 0.01, Wilcoxon sum-rank test. (C) Furthermore, the conformation of Fn’s integrin-binding domain within primary lung fibroblast-laden Fn-rich ECM is sensitive to modulation of fibroblast contractility through agonists (TGF-β) and inhibitors (blebbistatin). Scale bar, 50 μm.

Figure 4

Fn’s integrin mechano-switch displays spatially distinct patterns of activation in a model of resolving lung fibrosis. (A–D) Mouse lung tissue sections (scale bar = 50 μm) were immunostained for H5 (red), Fn (green), and DNA (blue) at the indicated time points after bleomycin-induced fibrotic injury. In this model, bleomycin induces fibrosis by 14–21 days, and is typically resolved by 56 days. H5/Fn ratio images displaying the heterogeneity of Fn fiber conformation within the ECM at the tissue scale are shown, along with H&E staining of a corresponding serial section at the same time points. The early repair (B) and fibrotic (C) time points show areas of higher H5/Fn ratio, indicating the unfolding of Fn’s integrin-binding domain, perhaps indicative of active fibrosis. By resolution (D), the H5/Fn ratiometric image resembles that of the saline control.

Figure 5

Fn’s integrin mechano-switch is activated at tip cells during retinal angiogenesis. (A) Whole mount immunostaining for retinal endothelial cells during postnatal retinal angiogenesis with tip cell area and capillary area (i.e., mature vessels) identified (scale bar = 500 μm). (B) Tip cell region (scale bar = 50 μm) and (C) mature vessels (scale bar = 80 μm) were immunostained with isolectin B4 (IB4, green) and H5 (red) and anti-Fn (blue) at postnatal day 6. Tip cells and blood vessels were visualized using isolectin B4. H5/Fn ratiometric images were generated for each region. The tip cell area (B) shows regions of high H5/Fn ratio, suggestive of endothelial tip cell force generation during angiogenesis. The mature vessel area, where forces are predicted to be low, displays a low H5/Fn ratio.