Extracellular matrix fibronectin mediates an endothelial cell response to shear stress via the heparin-binding, matricryptic RWRPK sequence of FNIII1H

Our novel finding is that the RWRPK heparin-binding sequence in the first type III repeat of fibrillar fibronectin is a mechanosignaling region; it contributes to realignment of endothelial cells by flow. Mechanosensory responses in endothelial cells contribute importantly to integrated vascular function; hence identification of a new mechanism is significant.

Abstract

Endothelial cells (EC) respond to mechanical forces such as shear stress in a variety of ways, one of which is cytoskeletal realignment in the direction of flow. Our earlier studies implicated the extracellular matrix protein fibronectin in mechanosensory signaling to ECs in intact arterioles, via a signaling pathway dependent on the heparin-binding region of the first type III repeat of fibrillar fibronectin (FNIII1H). Here we test the hypothesis that FNIII1H is required for EC stress fiber realignment under flow. Human umbilical vein ECs (HUVECs) exposed to defined flow conditions were used as a well-characterized model of this stress fiber alignment response. Our results directly implicate FNIII1H in realignment of stress fibers in HUVECs and, importantly, show that the matricryptic heparin-binding RWRPK sequence located in FNIII1 is required for the response. Furthermore, we show that flow-mediated stress fiber realignment in ECs adhered via α₅β₁-integrin-specific ligands does not occur in the absence of FHIII1H, whereas, in contrast, α_vβ₃-integrin-mediated stress fiber realignment under flow does not require FNIII1H. Our findings thus indicate that there are two separate mechanosignaling pathways mediating the alignment of stress fibers after exposure of ECs to flow, one dependent on α_vβ₃-integrins and one dependent on FNIII1H. This study strongly supports the conclusion that the RWRPK region of FNIII1H may have broad capability as a mechanosensory signaling site.

NEW & NOTEWORTHY

Our novel finding is that the RWRPK heparin-binding sequence in the first type III repeat of fibrillar fibronectin is a mechanosignaling region; it contributes to realignment of endothelial cells by flow. Mechanosensory responses in endothelial cells contribute importantly to integrated vascular function; hence identification of a new mechanism is significant.

it is well established that endothelial cells (ECs) can respond to mechanical forces such as shear stress by activating a variety of vasoactive signaling mechanisms. For example, shear stress plays a significant role in early stage signaling events such as K⁺ channel activation (20) and also in longer term responses such as changes in EC morphology and actin cytoskeletal realignment (9). Transduction of mechanical stimuli into EC responses is known to depend on several different mechanosensors, including integrins (32), PECAM-1 (22), and G protein-coupled receptors (3) among others. Importantly, the underlying extracellular matrix (ECM) is a critical coregulator of many mechanically induced responses in ECs. Thus, for example, alterations in the ECM composition upon which ECs are cultured is known to affect flow-dependent (i.e., mechanical) responses in these cells, including cell alignment (21, 26) and stress fiber remodeling and orientation (13). This interdependence between the ECM and mechanosensory signaling in ECs is further illustrated by the finding that shear stress can alter the composition of the ECM upon which the ECs are growing (12, 29). Thus, overall, such studies as these establish that the EC mechanosensory response to flow is influenced significantly by the composition and arrangement of the ECM.

In recent work (15, 25), we described studies in which the ECM protein fibronectin was implicated in mechanosensory signaling at the wall of resistance arterioles. A considerable literature connects the mechanical event of skeletal muscle contraction to local dilation of the resistance vasculature (reviewed in Ref. 18). Our studies implicated fibrillar ECM fibronectin in this local vasodilation (15, 25). We showed that the fibronectin-dependent component of the vasodilation initiated by contracting skeletal muscle fibers was coupled to vascular smooth muscle activation via an EC-dependent pathway and was mediated via the matricryptic heparin-binding RWRPK sequence located in the first type III repeat of fibrillar fibronectin (15, 25). Moreover, this response was independent of integrin ligation and could be inhibited with antibodies and peptides directed against the matricryptic signaling site in fibrillar fibronectin. Thus our studies pointed towards the matricryptic signaling region of fibrillar fibronectin, FNIII1H, as having a mechanosensory function resulting in responses in ECs. To test this hypothesis directly, we used ECs grown on a series of engineered fibronectin fusion proteins (11, 14, 15, 23–25) to ask whether this signaling region in fibronectin had the capacity to directly modify an established EC-dependent mechanosensory response. In the present study, we provide data showing that the matricryptic RWRPK sequence of the first type III repeat of fibrillar fibronectin mediates the well-established response of ECs to realign their stress fiber orientation in response to the mechanical stimulus provided by shear stress. This constitutes direct evidence that the FNIII1H signaling region of fibrillar fibronectin acts as a mechanosensory mediator to activate EC function.

MATERIALS AND METHODS

Cell culture, proteins, and microslide perfusion system.

Human umbilical vein endothelial cells (HUVEC; VEC Technologies, Rensselaer, NY) were cultured in MCDB-131 Complete Growth Media (VEC Technologies) at 37°C and 5% CO₂. At 80–90% confluence, HUVECs were subcultured using 0.05% trypsin-EDTA (GIBCO, Grand Island, NY). Cells between passage 2 and 5 were used for flow experiments as described below.

The fusion proteins used in this study have been described in recent work (2, 11, 14, 15, 23–25) and are illustrated schematically in Fig. 1. Briefly, our strategy was to use a series of soluble, recombinant fibronectin fragments in which FNIII1H, the “open” heparin-binding fragment of FNIII1, was directly coupled to various portions of the integrin-binding domain: construction of these proteins and verification of their biological activity have been reported (2, 11, 14, 15, 23–25). The fusion proteins used in the present study were as follows: a fibronectin matrix mimetic (FNIII1H,8–10) that contains both the heparin-binding (FNIII1H) and the α₅β₁-integrin-binding (FNIII8-10) modules; a construct in which FNIII1H was replaced with the carboxyl terminal heparin-binding module FNIII13 (FNIII8-10,13); a construct in which the active heparin-binding sequence of FNIII1 (R⁶¹³WRPK; Ref. 24) was mutated to noncharged amino acids (FNIII1H,8–10ΔRRK); a construct lacking the heparin-binding region of FNIII1 but retaining the α₅β₁-integrin-binding region (FNIII8-10); and constructs that selectively ligate α_vβ₃-integrins in the presence (FNIII1H,8^RGD) or absence (FNIII8^RGD) of the FNIII1 heparin-binding module.

Fig. 1.

Schematic representation of a fibronectin subunit and the fibronectin fusion proteins used in this study. Bottom: summary of whether or not the fibronectin fusion protein contains the matricryptic heparin-binding site and which integrin is ligated by the protein. FNIII, type III repeat of fibrillar fibronectin. Schematic modified from Refs. 14, 24, 25.

Rectangular glass microslides (50 × 3 × 0.3 mm; Vitrocom, Mountain Lakes, NJ) were acid washed and coated either with 4% (3-aminopropyl)triethoxysilane (APTES; Sigma-Aldrich, St. Louis, MO) in acetone to modify the surface charge and allow for EC adhesion and growth (6), or, alternately, they were coated at saturating concentrations with one of the fibronectin fusion proteins (200 nM; Ref. 23, 24). To coat with APTES, microslides were washed with a 50% solution of nitric acid in distilled water and then coated with 4% APTES in acetone; microslides were rinsed in distilled water, oven-dried overnight, and then autoclaved. To coat the microslides with fibronectin proteins, they were first washed with a 3:1 solution of sulphuric acid and hydrogen peroxide, rinsed with distilled water and oven-dried overnight, and then autoclaved. The fibronectin proteins were applied to the microslide's cell culture surface (120 min at 37°C), washed with PBS to remove unbound protein, and blocked with 1% fatty acid free BSA in PBS for 60 min. HUVECs were then immediately seeded into the microslide as described below.

HUVECs were seeded into microslides at a predetermined cell density (3.0 × 10⁴ cells/cm² for APTES-coated slides and 1.2 × 10⁵ cells/cm² for fibronectin protein-coated slides). Cells were allowed to adhere and spread for 2 h. The microslides were then transferred into modified culture dishes (Ref. 6 and Fig. 2A) and MCDB-131 Complete Growth Media was added in excess. The culture dishes were placed in an incubator (37°C and 5% CO₂) and connected via sterile tubing to exteriorly located syringe pumps (Fig. 2A). To exchange the media within the microslide while the HUVECs were growing to confluence, the software-driven pumps were activated for 30 s every 2 h to exchange the media by slow withdrawal at low wall shear stress (0.89 dyn/cm²). Cells grown to confluence on APTES-coated slides under these conditions produce a fibronectin matrix.

Fig. 2.

Summary of key technical approaches for growth, perfusion and analysis of human umbilical vein endothelial cells (HUVECs). A: modified Petri dish for cell growth to confluence in rectangular glass microslides. Pumps exterior to the incubator enable brief, low flow exchange of media at 2-h intervals. B: for experimental perfusion at 10 dyn/cm², microslides were placed within a reservoir and attached via tubing to a calibrated pump exterior to the incubator. C: representative image obtained for F-actin analysis (stress fibers, red; nuclei, blue). Systematically sampled regions were analyzed (see materials and methods) to determine stress fiber angle relative to the flow direction. Bar = 25 μm. D: stress fiber angle measured in confluent monolayers exposed to flow at 0.5, 1.0, or 10 dyn/cm² for up to 4 h established that a significant change in stress fiber alignment can be quantified at 4 h exposure to 10 dyn/cm².

To expose confluent HUVECs to flow, microslides were transferred to a nonrecirculating flow system (Fig. 2B). Briefly, microslides were transferred to a sterile Erlenmeyer flask that was fitted with a manifold that enabled connection of the microslide(s) via appropriate tubing to an automated syringe pump located outside the incubator. Media was withdrawn through the microslide at predetermined flow rates for specific periods of time. Wall shear stress, Tω, was calculated as Tω = (6μQ)/(bh²), where Q = volumetric flow rate, μ = fluid viscosity, b = chamber width, and h = chamber height. To verify the pump settings, microslides were mounted on a microscope stage and the centerline velocity of fluorescent beads added to the media was tracked over specified time periods and used to verify agreement (±4.5%) between the calculated and preset volumetric flow rates.

HUVECs were subjected to shear stress in McCoy's 5A (+l-glutamine) media (GIBCO, Grand Island, NY) containing 10% fetal bovine serum (FBS; GIBCO) and 1% penicillin/streptomycin (10,000 U/ml and 10,000 μg/ml, respectively). In specified experiments, cells were subjected to flow after exposure to either the anti-FNIII1 blocking antibody 9D2 (25 μg/ml for 2 h or control nonimmune IgG; Ref. 4) or the fibronectin matrix assembly inhibitor FUD (125 nM, 24 h, or its control peptide Del29; Ref. 30). For protocols in which ECs were grown on fibronectin fusion proteins as adhesive substrates, fibronectin-depleted FBS was used and the inhibitory peptide FUD (125 nM) was added to the flow media to inhibit fibronectin matrix assembly. To deplete fibronectin, 400 ml FBS were rocked (×3, 45 min each) with gelatin-Sepharose resin (24-ml packed volume; GE Healthcare, Uppsala, Sweden) at 4°C. To confirm removal of fibronectin, serum samples pre- and posttreatment were analyzed using SDS/PAGE and immunoblotting. No fibronectin was detected after three consecutive incubations with gelatin-Sepharose resin.

General experimental strategy, data acquisition, and analyses.

Immunofluoresence microscopy of fixed, stained cells was used to visualize the actin cytoskeleton and the fibronectin matrices. HUVECs in the microslides were fixed with 3.7% methanol-free formaldehyde (Polysciences, Warrington PA) for 15 min at 37°C, permeabilized with 0.5% Triton X-100, and blocked with 1% BSA. Cells were then incubated with rabbit anti-human fibronectin polyclonal antibodies (Sigma-Aldrich) in 1% BSA followed by Alexa Fluor 488-conjugated secondary goat anti-rabbit IgG (Molecular Probes, Eugene, OR) and coincubated with rhodamine phalloidin (Molecular Probes). Lastly, cells were incubated with DAPI (Molecular Probes) and mounted using Fluoromount G (Southern Biotech, Birmingham, AL).

Changes in F-actin stress fiber alignment were measured as the primary response to flow; this response of ECs to flow has been well characterized (e.g., Refs. 5, 9, 16, 19). To quantify stress fiber alignment, stained HUVECs were imaged using an Olympus IX-70 inverted microscope (×40, 0.8 numerical aperture objective), coupled to a cooled CCD (Sensicam QE; PCO-Tech, Romulus, MI). Images were digitized using Slidebook software. For each microslide, images were captured from 10 locations along the centerline: these images were analyzed using National Institutes of Health ImageJ software. For each of the 10 images captured from each microslide, a predetermined grid system was used to systematically select 4 regions (each approximately the area of an EC); in each selected image the angle of 5–10 actin fibers relative to flow was measured, using a predetermined sampling protocol to avoid bias. Thus, for each microslide, the angles of 5–10 actin fibers within each of the 40 sampled regions were measured, to give 200–400 measurements. Figure 2C illustrates key aspects of the analysis. Stress fibers that were aligned with the flow direction were assigned a value of 0°, and those perpendicular to the flow were assigned 90°. A random distribution of stress fibers in nonoriented conditions will give an expected average orientation angle of 45°: a significant decrease from this angle indicates alignment with the flow direction.

Preliminary studies were undertaken to establish an appropriate shear stress and time of exposure for the planned protocols. For this, we exposed confluent HUVECs plated on APTES-coated surfaces to a range of flow rates that were calculated to produce Tω from 0–10 dyn/cm²: flow exposures were for 0–4 h. Figure 2D shows that significant stress fiber alignment (decrease in measured angle) was observable after 4 h at 10 dyn/cm² (29.7 ± 1.9°, P < 0.05): the alignment after 4 h at 0.5 dyn/cm² (44.2 ± 3.0°) was not different from static conditions (Fig. 2D). There was a small but significant decrease in alignment angle in cells exposed for 4 h to 1.0 dyn/cm² (39.2 ± 2.2°), indicating that our system has the capacity to identify small alignment changes at 4 h. From these findings, we selected a 4-h exposure to 10 dyn/cm² as a test condition capable of identifying early responses of actin stress fiber alignment to flow.

Each protocol was repeated a minimum of three times. Statistical analyses were performed using GraphPad Prism version 4 software. The data are presented as mean ± SE. Data sets were compared by ANOVA (with the Bonferroni post test), or Student's t-test, as appropriate. Differences were considered significant at P < 0.05.

RESULTS

ECM fibronectin and FNIII1H are necessary for shear stress-induced realignment of actin stress fibers.

ECM composition plays a role in EC responses to flow (12, 29); thus our first goal was to establish a role for fibrillar fibronectin in stress fiber realignment. To do this, fibrils were removed from confluent monolayers by treatment with the peptide FUD (30), which inhibits new fibril formation and promotes the turnover of preexisting fibronectin matrices (27). Confluent HUVEC monolayers that had been plated on APTES were treated for 24 h with either FUD (125 nM) or the control peptide Del29 (125 nM) and were then exposed to shear stress (10 dyn/cm²) for 4 h. Figure 3A shows that the fibrillar ECM fibronectin matrix is substantially reduced in the presence of FUD compared with treatment with the control peptide Del29 under both static and flow conditions. Importantly, the alignment of stress fibers in response to flow was significantly attenuated when the fibronectin matrix was reduced (Fig. 3, B and C). In the absence of ECM fibronectin, the average stress fiber angle relative to the flow direction was 40.2 ± 0.8°, compared with 33.5 ± 0.8° for cells having an intact fibronectin matrix (P < 0.05, Fig. 3C), implicating ECM fibronectin in the process of stress fiber realignment.

Fig. 3.

Extracellular matrix (ECM) fibronectin is implicated in shear stress-induced realignment of F-actin stress fibers. HUVECs were seeded (3 × 10⁴ cells/cm²) on 3-aminopropyltriethoxysilane (APTES)-coated microslides and grown to confluence (36 h). After 24 h treatment with 125 nM of either FUD or Del29, cells were either maintained static or exposed to flow (10 dyn/cm², 4 h). A: images of ECM fibronectin matrix in each condition. Note that fibrillar fibronectin is only in the Del29 treated group. Scale bar = 25 μm. B: images of F-actin stress fibers in each condition. Scale bar = 25 μm. Images in A and B are representative of 3 independent experiments. C: mean stress fiber alignment relative to flow for cells in the presence of the matrix assembly inhibitor FUD (white bars) or its control Del29 (black bars) in static conditions or after flow exposure. Data are presented as mean ± SE of 3 independent experiments. *P < 0.05, significantly different from corresponding static condition by ANOVA.

Our previous studies demonstrated a role for the RWRPK sequence of FNIII1 in the mechanosensory response initiated by skeletal muscle contraction (15, 25). To determine whether FNIII1 has a mechanosensory role in flow-dependent stress fiber realignment, we utilized the monoclonal antibody 9D2, which both recognizes the matricryptic heparin-binding site of FNIII1 and also blocks fibronectin matrix assembly (4, 11). HUVECs were grown to confluence on APTES to allow for the establishment of a fibronectin matrix and were treated for 2 h with 9D2 or its control IgG (both at 25 μg/ml). It has been established that turnover of fibrillar fibronectin upon inhibition of fibronectin matrix assembly is minimal in 2 h (28). This was confirmed in our studies by both immunofluorescence microscopy (Fig. 4A) and by immunoblot analysis (normalized intensity: 9D2, 1.63 ± 0.14; IgG, 1.28 ± 0.08; vehicle, 1.34 ± 0.10; n = 6, not different, ANOVA, P = 0.08). HUVECs were subjected to 10 dyn/cm² for 4 h. Figure 4 shows that after treatment with 9D2, the mean angle of stress fiber alignment after flow was significantly attenuated (42.0 ± 0.4°) relative to that for control IgG-treated cells (35.0 ± 0.4°). The continued presence of a fibronectin matrix in the presence of 9D2 suggests that the attenuated stress fiber alignment was more likely due to blockade of the matricryptic signaling site and not to reduction in fibronectin matrix assembly. These data therefore implicate the first type III repeat of fibrillar fibronectin in stress fiber alignment and strongly suggest a role for the matricryptic signaling site in this process.

Fig. 4.

FNIII1 is implicated in shear stress-induced realignment of F-actin stress fibers. HUVECs were seeded (3 × 10⁴ cells/cm²) on APTES-coated microslides and grown to confluence (36 h). After 2-h treatment with 25 μg/ml of either 9D2 (anti-FNIII1 mAb) or its control IgG, HUVECs were either maintained static or exposed to flow (10 dyn/cm², 4 h) and then fixed and stained for F-actin. A: HUVECs fixed and stained for fibronectin after 2 h exposure to the antibody 9D2 or its control IgG. Scale bar = 25 μm. B: images of F-actin stress fibers in each condition. Scale bar = 25 μm. Images in A and B are representative of 3 independent experiments. C: bars show mean ± SE of the stress fiber angle relative to flow in the presence of 9D2 (white bars) or its control IgG (black bars); data from 4 independent experiments. *P < 0.05, significantly different from corresponding static condition by ANOVA.

To ask directly whether the matricryptic FNIII1H site is required for the response of ECs to flow, we compared stress fiber alignment after exposure to flow in HUVECs grown to confluence on one of three fibronectin fusion proteins: an ECM fibronectin analog that contains both the matricryptic FNIII1H site and the α₅β₁-integrin-binding domain (FNIII1H,8–10); a fibronectin protein in which FNIII1H was replaced with an alternate heparin-binding module (FNIII8-10,13); and one in which the active heparin-binding sequence was mutated (FNIII1H,8–10ΔRRK). Each of these is described more fully in materials and methods. To eliminate signals from endogenous ECM fibronectin, cells were cultured in fibronectin-free media and in the presence of the fibronectin matrix assembly inhibitor FUD. Confluent HUVECs grown on the various substrates were exposed to 10 dyn/cm² for 4 h and F-actin stress fiber alignment was quantified.

Figure 5 shows that cells cultured on FNIII1H,8–10 in the absence of flow had an average stress fiber angle of 43.9 ± 0.9°, which is not significantly different from that for cells that had been plated in serum on APTES-treated glass (45.4 ± 0.4°). After exposure to flow, the average stress fiber angle (33.8 ± 0.9°) was significantly decreased compared with cells maintained in static conditions; this flow-induced alignment was similar to that induced by flow in cells that had been plated on APTES and allowed to assemble a fibronectin matrix (35.0 ± 0.3°). This indicates that FNIII1H,8–10 supports actin stress fiber realignment to a similar extent as observed under conditions that permit endogenous fibronectin matrix assembly (Figs. 3 and 4).

Fig. 5.

The matricryptic heparin binding site FNIII1H mediates shear stress-induced realignment of F-actin stress fibers. HUVECs were seeded (1.2 × 10⁵ cells/cm²) on acid washed microslides coated with ECM fibronectin mimetic proteins (200 nM) as specified below, grown to confluence (16 h), and then either maintained static or exposed to flow (10 dyn/cm², 4 h). Cells were grown in fibronectin-free media and in the presence of the fibronectin matrix assembly inhibitor, FUD. A: images of F-actin stress fibers in each condition. Images are representative of 3 independent experiments. Scale bar = 25 μm. B: bars show mean ± SE stress fiber alignment relative to flow under static (black bars) and flow (white bars) conditions for cells grown on FNIII1H, 8–10 (intact fibronectin matricryptic signaling sequence); FNIII1H,8–10ΔRRK (mutated matricryptic signaling sequence); and FNIII8-10,13 (alternate FNIII13 heparin binding module). Data are presented as mean ± SE of 3 independent experiments. *P < 0.05, significantly different from static FNIII1H,8–10 by ANOVA.

In contrast, stress fiber realignment in response to flow did not occur in cells adherent to FNIII1H,8–10ΔRRK, in which the active site of FNIII1 (R⁶¹³,R⁶¹⁵,K⁶¹⁷) was mutated. As shown in Fig. 5, cells grown on the mutated construct did not change their stress fiber alignment when subject to flow (45.0 ± 0.9° static vs. 42.0 ± 1.0° after 4 h at 10 dyn/cm²), indicating that mechanisms activating cell signaling in response to the mechanical stimulus produced by flow reside in the matricryptic RWRPK site of FNIII1.

To confirm the specificity of the mechanosensory signaling residing in FNIII1, vs. the possibility that this signaling might be characteristic of other heparin-binding regions of fibrillar fibronectin, we also measured the response to flow of HUVECs grown on FNIII8-10,13. Exposure of these cells to flow did not initiate an alignment response (Fig. 5), with static cells having a mean stress fiber angle (47.0 ± 1.0°) that was not different from the stress fiber angle measured in those that had been subjected to flow (45.0 ± 1.0°). Thus the data summarized in Fig. 5 indicate that the RWRPK site in FNIII1 is necessary for the response of ECs to realign their stress fibers upon exposure to flow and identify this region of FNIII1 as one which has mechanosignaling capabilities, without which cells adhered via α₅β₁-integrins do not respond.

Mechanosensory signaling via FNIII1H is coupled to α₅β₁-, but not α_vβ₃-, integrin ligation.

Both α_vβ₃- and α₅β₁-integrins can mediate EC adhesion and have been implicated in EC responses to flow (17, 32–34). We therefore undertook protocols designed to determine the involvement of each of these integrins in FNIII1H-dependent mechanosignaling in response to flow. To do this, we grew HUVECs to confluence on fibronectin substrates that selectively ligated either α₅β₁- or α_vβ₃-integrins and that either contained or lacked FNIII1H. Thus we used two substrates that bind α₅β₁-integrins, either with (FNIII1H,8–10) or without (FNIII8-10) FNIII1H, and two substrates that bind α_vβ₃-integrins but again with (FNIII1H,8^RGD) or without (FNIII8^RGD) FNIII1H (23). Figure 6, A and B, shows that F-actin stress fiber realignment occurred when cells were adherent to FNIII1H,8–10 but did not occur in response to flow when the HUVECs were grown on FNIII8-10 (43.2 ± 0.9 vs. 45.7 ± 1.0° under static conditions). These data thus indicate that for cells adherent via α₅β₁-integrins, the matricryptic heparin-binding site is necessary for the mechanosensory response to shear stress. In contrast, cells adherent via α_vβ₃-integrins responded to flow in the presence and also in the absence of FNIII1H (Fig. 6, C and D): mean stress fiber alignment decreased significantly from 44.6 ± 1.0° (static) to 30.3 ± 0.8° after flow exposure for FNIII1H,8^RGD and from 44.4 ± 1.0° (static) to 33.9 ± 0.9° after flow exposure for FNIII8^RGD. Thus the observations in Fig. 6, taken together, suggest that, for cells adhered via α₅β₁-integrins, stress fiber realignment under flow requires the presence of FNIII1H, whereas for cells adherent via α_vβ₃-integrins, realignment with flow is independent of the FNIII1H signaling region. Importantly, these data imply that there are two separate signaling pathways for this EC mechanosensory response to flow.

Fig. 6.

Stress fiber realignment due to flow on α₅β₁-integrin-binding substrates requires FNIII1H, whereas realignment on α_vβ₃-integrin binding-substrates does not. HUVECs were seeded (1.2 × 10⁵ cells/cm²) on acid washed microslides coated with ECM fibronectin mimetic proteins (200 nM) as specified below, grown to confluence (16 h), and then either maintained static or exposed to flow (10 dyn/cm², 8 h). Cells were grown in fibronectin-free media and in the presence of the fibronectin matrix assembly inhibitor FUD. A and C: images of F-actin stress fibers for cells grown on α₅β₁-integrin-binding substrates (A) or α_vβ₃-integrin-binding substrates (C). Images are representative of 3 independent experiments. Scale bar = 25 μm. B and D: bars show mean ± SE stress fiber alignment relative to flow under static (black bars) and flow (white bars) conditions for cells grown on α₅β₁-integrin-binding substrates (B) or α_vβ₃-integrin-binding substrates (D). In A–D, cells were grown on substrates that either did, or did not, contain the FNIII1H binding region (FNIII1H,8–10 vs. FNIII8-10; A and B) and (FNIIIH,8^RGD vs. FNIII8^RGD; C and D). Data are presented as mean ± SE of 3 independent experiments. *P < 0.05, significantly different from static control by ANOVA.

DISCUSSION

Our study has identified a mechanosensory signaling pathway in ECs exposed to physiological flow conditions that is dependent on the presence of a heparin-binding region in the first type III repeat of fibrillar fibronectin. This signaling is mediated by a matricryptic heparin-binding RWRPK sequence located in FNIII1. We have thus directly located this mechanosensory function to the same heparin-binding site in fibrillar fibronectin that we had previously implicated in mediating the vasodilatory response to the mechanical event of skeletal muscle contraction. This suggests that this FNIII1H site may have broad capability as a mechanosensory signaling site. Our study also provides initial evidence that indicates that this signaling pathway is independent from an apparently separate, α_vβ₃-integrin-dependent pathway that mediates stress fiber alignment in response to flow.

Our data show that the matricryptic heparin-binding site (FNIII1H) of fibrillar fibronectin mediates the response to shear stress. Several lines of evidence from our study support this conclusion. Firstly, we established a role for fibrillar fibronectin in the response by showing that a substantial reduction in the fibrillar ECM fibronectin matrix via treatment with FUD was associated with a significant attenuation in the alignment of stress fibers in response to flow (Fig. 3). Furthermore, we showed that 2-h treatment with the monoclonal antibody 9D2 was sufficient to block stress fiber realignment (Fig. 4). This antibody blocks both fibronectin matrix assembly and the matricryptic heparin-binding site of FNIII1(4, 11), but we showed in earlier work that brief application of 9D2 blocks FNIII1H-dependent responses independently of inhibition of fibronectin matrix assembly (15). Furthermore, as we show in Fig. 4, a fibronectin matrix was still present after 2-h treatment with 9D2, hence our finding that this treatment blocked stress fiber realignment strongly suggested involvement of the matricryptic signaling site. This was confirmed by our studies using HUVECs grown to confluence on selected fibronectin proteins; only cells adhered on α₅β₁-integrin-binding substrates that contain the heparin-binding site of FNIII1 showed stress fiber alignment on exposure to flow, and, furthermore, mutation of the active heparin-binding sequence was sufficient to abolish the response (Fig. 5). Thus our data identify this signaling sequence as required for mechanosensory responses, and, taken together with our earlier work in contracting skeletal muscle (15, 25), suggest that FNIII1H has broad capability in mechanosignaling. Our earlier work (15) identified a nitric oxide dependent component in the response to FNIII1H, but how this might relate to the current observations, or how the FNIII1H-dependent signaling pathway might relate to other mechanosensory responses, remains to be determined.

This study, combined with our earlier observations in arterioles (15, 25), shows clearly that ECs respond to FNIII1H-dependent mechanosignaling pathway(s). In the earlier work we summarized morphological and physiological findings that support the conclusion that ECs are in contact with the ECM in a manner that facilitates vascular responses (25). However, the identity of the cell surface receptor that binds to FNIII1H and thus facilitates signaling to ECs remains unknown. Previous work has suggested that heparan sulphate proteoglycans (HSPGs) are required for responses to FNIII1H and to the alternate heparin-binding site FNIII13 (1, 14). However, we note that in our study, the presence of FNIII13 was not sufficient to support flow-induced cytoskeletal reorganization (Fig. 5). Other studies exploring the EC response to flow have implicated HSPGs in the NO-dependent component of the response (8), and, furthermore, the heparan sulphate core proteins glypican-1 and syndecan-1 have both been identified as components of eNOS activation and EC alignment by flow (7, 33). How these observations relate to our current findings is clearly an important question that will require new studies; in particular, it will be important to determine which, if any, HSPGs might interact with the FNIII1H signaling region to modulate flow-dependent responses.

A noteworthy finding in the present work is that while we confirmed that α_vβ₃-integrins can mediate flow-dependent stress fiber realignment, in contrast, α₅β₁-integrins have no role independent of the presence of FNIII1H. Both of these integrins are activated upon stimulation by shear stress (17, 32) and appear to be required for the regulation of Rho and Rac activity, which in turn is required for cell alignment (31, 32, 34). Interestingly, other studies have shown that these two integrins activate different signaling pathways that link ECM to EC responses (10). An important new line of enquiry will be to dissect out how α₅β₁-integrin activation interacts with FNIII1H-dependent signaling, while α_vβ₃-integrins act apparently independently of any involvement of this heparin-binding site.

In summary, we have identified a role for the matricryptic, heparin-binding site located within the first type III repeat of ECM fibronectin (FNIII1H) in mediating EC responses to flow, that is, this site functions as a mechanosensory signaling region. We further show that this FNIII1H-dependent mechanosignaling pathway is separate from that mediating the same stress fiber alignment response via α_vβ₃-integrins. Thus, our data support the hypothesis that multiple signaling pathways arising from the ECM mediate EC responses to mechanical events: one of these encompasses the well-established role for α_vβ₃-integrins, but another, as we describe in this study, is via FHIII1H signaling, thus constituting a newly described, mechanosensing pathway dependent on fibrillar ECM fibronectin.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Grant R0-1HL-105909 (to I. H. Sarelius).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

W.O., K.M.A., J.M.K., D.C.H., and I.H.S. conception and design of research; W.O., K.M.A., and J.M.K. performed experiments; W.O., K.M.A., and J.M.K. analyzed data; W.O., D.C.H., and I.H.S. interpreted results of experiments; W.O. and I.H.S. prepared figures; W.O. and I.H.S. drafted manuscript; W.O., K.M.A., J.M.K., D.C.H., and I.H.S. approved final version of manuscript; D.C.H. and I.H.S. edited and revised manuscript.