A hallmark of many, sometimes life-threatening, inflammatory diseases and disorders is vascular leakage. The extent and severity of vascular leakage is broadly mediated by the integrity of the endothelial cell (EC) monolayer, which is in turn governed by three major interactions: cell-cell and cell-substrate contacts, soluble mediators, and biomechanical forces. A potentially critical but essentially uninvestigated component mediating these interactions is the stiffness of the substrate to which the endothelial monolayer is adherent. Accordingly, we investigated the extent to which substrate stiffening influences endothelial monolayer disruption and the role of cell-cell and cell-substrate contacts, soluble mediators, and physical forces in that process. Traction force microscopy showed that forces between cell and cell and between cell and substrate were greater on stiffer substrates. On stiffer substrates, these forces were substantially enhanced by a hyperpermeability stimulus (thrombin, 1 U/ml), and gaps formed between cells. On softer substrates, by contrast, these forces were increased far less by thrombin, and gaps did not form between cells. This stiffness-dependent force enhancement was associated with increased Rho kinase activity, whereas inhibition of Rho kinase attenuated baseline forces and lessened thrombin-induced inter-EC gap formation. Our findings demonstrate a central role of physical forces in EC gap formation and highlight a novel physiological mechanism. Integrity of the endothelial monolayer is governed by its physical microenvironment, which in normal circumstances is compliant but during pathology becomes stiffer.
- human umbilical vein endothelial cells
- traction force
- cell-cell contact
- cell-substrate contact
- substrate stiffness
- Rho kinase
- vascular endothelial cadherin
the overall integrity and barrier properties of the endothelial cell (EC) monolayer are governed by three main categories of inputs: cell-cell and cell-substrate contacts, soluble mediators (e.g., thrombin, histamine, spingosphine 1-phosphate, and nitric oxide), and biomechanics (e.g., innate monolayer forces, shear forces, and stretch) (33). In vivo, these inputs are integrated by the EC monolayer to regulate its overall integrity and responses to inflammatory stimuli. Characterizing these inputs and their interrelationships is thus of central importance for understanding vascular biology and inflammation as a whole.
A potentially critical, but entirely ignored, component of the EC environment that may influence the aforementioned inputs and their interactions is stiffness of the substrate to which the EC monolayer is adherent. This substrate stiffness varies greatly among diverse physiological settings (12, 13, 32, 59), is enhanced with aging (21, 40), and is exacerbated by risk factors associated with common disorders such as diabetes, hypertension, cancer, atherosclerosis, and renal disease (28, 32, 34, 37, 44, 58, 60).
Although its effects on the EC monolayer are largely unknown, substrate stiffness is recognized as a powerful effector of migration, morphology, spreading, and growth in the single EC (5, 6, 8, 19, 25, 31, 35, 41, 42). Regulating these effects are the cell's innate physical forces, called traction forces (4, 11, 14, 23, 38, 41, 42, 46). These forces originate from actomyosin interactions within the cell, are balanced predominantly by the actin cytoskeleton, and are exerted by the single cell in a substrate stiffness dependent manner (6, 8, 41, 42).
Compared with the limited cases of single cells studied in isolation (5, 6, 8, 19, 25, 31, 35, 41, 42) or cells studied in pairs (30), cohesive groups of cells, as in a continuous monolayer that constitutes the vascular endothelium, have far more physiological relevance. However, because experimental tools were absent, forces in EC monolayers have remained hidden from observation, and, necessarily, their substrate dependent effects have remained inferential and obscure.
For the first time, we present maps of those monolayer physical forces between cell and cell and between cell and substrate, their spatial and temporal changes in response to a hyperpermeability stimulus, and the role of substrate stiffness in modulating monolayer forces and overall monolayer integrity. We further present the role of Rho kinase activity in monolayer force effects. Our major finding is that stiffness of the EC microenvironment has a profound and previously unanticipated role in regulation of endothelial monolayer integrity. Our secondary finding is that these substrate stiffness-dependent effects are mediated by basal force differences in the EC monolayer.
MATERIALS AND METHODS
Preparation of polyacrylamide gel substrates.
Polyacrylamide gel substrates were prepared according to a previously described method (11, 23, 57). Briefly, in stage 1, 0.1 M NaOH was added to 20-mm glass-bottom dishes (MatTek; uncoated), and the dishes were then air-dried overnight. In stage 2, each pretreated dish was covered with 97% aminopropyltrimethoxysilane (Sigma) for 5 min, rinsed thoroughly with distilled water, treated with 0.5% glutaraldehyde in PBS for 30 min, rinsed again, and air-dried overnight. In stage 3, solutions of acrylamide (Ac), bis-acrylamide (Bis; Bio-Rad, Hercules, CA), and 0.2-μm-diameter yellow fluorescent beads (Invitrogen, Eugene, OR) were dissolved in ultrapure water over a range of concentrations to yield a wide range of gel stiffnesses (1.2 kPa: 5.5% Ac and 0.05% Bis; 4 kPa: 5% Ac and 0.1% Bis; 11 kPa: 10% Ac and 0.07% Bis; 90 kPa: 12% Ac and 0.3% Bis). Each gel mixture was placed in a custom vacuum chamber for 20 min and mixed with 0.05% ammonia persulfate and 0.05% N,N,N-tetramethylethylenediamine (Bio-Rad). The ensuing mixture (20 μl) was added as a drop to the center of each pretreated dish from stage 2. The drop was covered with a glass coverslip (18-mm diameter; VWR, West Chester, PA), and the gel dishes were centrifuged with the top side facing down at 500 rpm for 10 min. The centrifugation favors fluorescent bead migration to the top plane of the gel during gelation. After centrifugation, the dishes were set aside for 30 min to yield gels with a final thickness of ∼100 μm. Next, 200 μl of a 1 mM sulfosuccinimidyl-6-[4-azido-2-nitrophenylamino]hexanoate (Pierce, Rockford, IL) solution were added to each gel surface, and the surfaces were exposed to UV light for 10 min. This step photoactivated the gel surfaces to promote ligand binding. The gels were then thoroughly washed with 0.1 M HEPES and PBS. Micropatterned polydimethylsiloxane (PDMS) membranes were placed in conformal contact with the gel surface, and 400 μl of type I collagen solution (0.1 mg/ml; Inamed Biomaterials, Fremont, CA) were added on top of the micropatterned membrane. The gels were then incubated overnight at 4°C. On the following day, the gels were washed with PBS, hydrated with 2 ml of medium 199 (M199) solution, and stored in an incubator at 37°C and 5% CO2 until the start of the experiment.
Preparation of micropatterned membranes.
Repeated patterns of 150-μm-diameter circular islands (spaced 300 μm apart) were fabricated on PDMS membranes using membrane patterning technology. Briefly, a master containing raised features in photoresist was fabricated by photolithography from high-resolution masks (CAD/Art Services, Bandon, OR). PDMS prepolymer was spin-coated on the photoresist master, cured at 70°C for 90 min, and peeled from the master to obtain PDMS membranes. The PDMS membrane was then stored in ethanol and dried overnight.
Human umbilical vein endothelial cells (HUVECs) were isolated and characterized as previously described (51). The primary cells were cultured on plastic flasks coated with 1% gelatin in M199 culture medium (Lonza) supplemented with 10% human serum, 10% newborn calf serum, 2 mM glutamine, 5 U/ml heparin, 100 IU penicillin, 100 μg/ml streptomycin, and 50 μg/ml crude endothelial cell growth factor. We refer to this supplemented medium as CM199. For experiments, cells in passages 1–3 were trypsinized, and 50 μl of concentrated cell suspension were added on top of the previously prepared PDMS membrane-gel surface. After 5 min, 2 ml of CM199 were carefully added to the dish, and the samples were placed in the incubator for an additional 45 min. The PDMS membrane was then carefully peeled off, leaving micropatterned ECs on the gel. The dishes were placed in the incubator at 37°C and 5% CO2 for 48 h. The medium was replaced once after 24 h.
The serum-containing cell medium (CM199) was replaced with 1 ml of serum-deprived medium (M199) supplemented with 1% human serum albumin and 1% penicillin-streptomycin for 5 min before experiments. The following pharmacological interventions were then used at the specific concentrations: thrombin (hyperpermeability inducer), final concentration 1 U/ml; cadherin 5 [vascular endothelial (VE)-cadherin monoclonal blocking antibody], final concentration 12.5 μg/ml; Y-27632 (Rho kinase inhibitor), final concentration 10 μM, preincubation time 30 min; and PBS time control.
Time-lapse experiments were performed on an inverted Leica microscope equipped with a climate-controlled chamber at 37°C and 5% CO2 (humidified). Phase contrast images of the micropatterned cell monolayer and fluorescent images of the nanobeads embedded in the substrate directly underneath the cells were taken at baseline, at several time points over 600 s following drug treatment, and after the cells were trypsinized at the end of the experiment.
Traction force microscopy.
Cell tractions were computed using constrained Fourier transform traction microscopy (4). Briefly, the displacement field was computed by comparing fluorescent nanobead images obtained during the experiment with the reference image obtained at the end of the experiment subsequent to detaching the cells from the substrate. The projected area was calculated based on a manual trace of the contour of the island of cells determined from a phase contrast image obtained at the start of the experiment. From the displacement field we calculated the traction field, and from the traction field we computed a scalar measure of contractility called the net contractile moment. The net contractile moment provides a scalar measure of the contractile strength of the EC island.
Over the last decade, we and others have established that this contractile moment is a measure of cytoskeletal retraction. Irrespective of cell type, agonist, treatment, and experimental protocol, changes in cell contractile moment were found to mirror changes in cytoskeletal dynamics (1–4, 24, 48, 49, 57).
Computation of intercellular forces.
As described above, the net contractile moment is now well established as a robust measure of cell contractility(4), and this concept extends to monolayers of cells plated on islands. We argue that the difference in contractile moment with treatment leading to gap formation is a measure of the intercellular forces borne within the monolayer as opposed to those supported by cell-substrate interaction. This can be shown by an elementary example: consider a monolayer consisting of just two cells in contact with one another, A and B, ordered from left to right. For simplicity, we consider only the forces in one dimension. Let the force exerted by the left-hand free edge of cell A on the substrate be F. Because the net force on every cell must sum to zero, the net force exerted by its right-hand edge must match this (with a negative sign). Suppose a portion of this force, −αF, is exerted on the neighboring cell B, and thus a complementary force, (1 − α)F, is exerted on the substrate. Similarly, the forces exerted by cell B at its left-hand edge (in contact with cell A) are +αF on cell A and +(1 − α)F on the substrate. Finally, the right-hand free edge of cell B exerts a force of −F on the substrate.
The contractile moment of this two-cell construct is given by F(LA + LB), where LA and LB are the lengths of cells A and B, respectively. This follows from the fact that the interior forces exerted by both cells on the substrate cancel. Now suppose a treatment is given, such as the VE-cadherin blocking antibody, which breaks the (A, B) junction but leaves the cell-substrate interactions intact. In this case, the net contractile moment is given by the sum of the two individual cells, since moments are additive and are indifferent to coordinate system origin. This results in a net moment of (1 − α)FLA + (1 − α)FLB = F(LA + LB) − αF(LA + LB). The difference, or loss, of contractile moment pre- vs. posttreatment is thus simply αF(LA + LB), which in turn is the fraction of forces borne by intercellular contact and, by inference, mediated by the intercellular junctions broken by lack of VE-cadherin adhesion molecules.
Note that this argument assumes that on gap formation, the lengths of the cells do not change significantly relative to their baseline lengths. To the extent that the gaps represent a substantial change in overall cellular dimensions, the resulting moment will reflect an additional loss associated with relaxation of the individual cells, but this is not expected to be a major confounding effect. If necessary, we note that this effect can be quantified by direct measurements of the change in contractile moments with stretching of the islands following gap formation.
Evaluation of endothelial permeability.
Confluent HUVEC monolayers (1st and 2nd passage) were cultured on fibronectin-coated polycarbonate filters of a Transwell system (pore size 3.0 μm). Permeability measurements were performed within the next 4 and 6 days, as described previously (52, 54). Briefly, macromolecular passage across HUVEC monolayers was investigated from measurements of horseradish peroxidase transfer before and after disruption of the EC barrier with a VE-cadherin blocking antibody (6.25 μg/ml). Effects of Rho kinase inhibition were investigated by preincubating HUVEC monolayers with 10 μM Rho kinase inhibitor Y-27632 for 30 min before experiments.
Evaluation of RhoA activity.
Expression of RhoA was measured with a RhoA G-LISA Biochem kit according to the manufacturer's protocol (Cytoskeleton, Denver, CO).
Imaging for immunofluorescence.
ECs were fixed at room temperature with 2% formaldehyde in PBS (incubation time 10 min) and stained for VE-cadherin or vinculin, as previously described (53). F-actin was visualized by staining with rhodamine-phalloidin. Cell nuclei were visualized with 4,6-diamidino-2-phenylindole. All images were obtained using three-dimensional digital imaging technology as previously described.
Statistical differences between groups of data were assessed using one-tailed paired t-tests. Differences with P ≤ 0.05 were considered significant. All error bars plot standard errors, unless otherwise indicated.
RESULTS AND DISCUSSION
Substrate stiffening enhances EC monolayer traction forces.
HUVEC monolayers were cultured within 150- to 200-μm-diameter micropatterned circular collagen islands. Each collagen island was patterned on a polyacrylamide gel of defined stiffness. Gel stiffness (Young's modulus) spanned a wide range (1.2–90 kPa) that encompassed physiologically relevant values. Micropatterning cells, as just described, enabled us to view changes in the entire monolayer within the microscope field of view.
Phase contrast images of the monolayers revealed 6–15 ECs per circular collagen island (Fig. 1, A and B). This cell count did not vary with substrate stiffness, nor did morphology(59), basal distribution of vinculin in focal adhesions, or VE-cadherin in EC junctions (Supplemental Fig. S1). (Supplemental data for this article is available online at the American Journal of Physiology-Cell Physiology website.) However, F-actin content in the EC cytoskeleton was markedly increased on stiffer substrates (Fig. 1D) in a manner comparable to dependencies of F-actin content on substrate stiffness reported in epithelial cell groups (10).
To measure the tractions that the EC monolayer exerts on its substrate, we used traction force microscopy (4). These tractions were higher in monolayer ECs than in single ECs (Supplemental Fig. S2). Furthermore, the tractions were characterized by punctate force hot spots located prominently at the monolayer edge. These traction distributions were similar across substrates of all stiffnesses but were enhanced substantially with progressive substrate stiffening (Fig. 1C). This behavior was observed not only in HUVECs but also in rat pulmonary microvascular endothelial cells and rat brain endothelial cells (data not shown).
Substrate stiffening enhances VE-cadherin-mediated forces.
Next, we probed the effects of substrate stiffening on EC junctional adhesiveness and intercellular force transmission. VE-cadherin-mediated adherens junctions are generally considered the dominant and most universal structure for stabilizing interendothelial junctions (15, 26, 56). We therefore focused on its role in mediating EC intercellular forces.
EC monolayers were plated across a range of substrate stiffnesses and treated with a monoclonal antibody that blocked VE-cadherin homotypic interactions (9). The VE-cadherin blocking antibody induced junctional breakage and gap formation on all substrates (Fig. 2). At the same time, net monolayer contraction, as quantified by the contractile moment of the overall EC monolayer, decreased (Fig. 2B). Forces borne by VE-cadherin interactions were greater when the substrate was stiffer (Fig. 2C), but when scaled by overall monolayer contraction, the fraction of forces supported by VE-cadherin was similar across all substrates (Fig. 2D). These findings imply that it is the overall force in the EC monolayer rather than VE-cadherin that sets the level of forces borne by VE-cadherin interactions. These findings also establish that VE-cadherin-mediated junctions bear approximately one-half of the overall contractile forces in intact EC monolayers.
To exclude nonspecific effects of the VE-cadherin blocking antibody on EC monolayer response, we tested the effects of the VE-cadherin blocking antibody on single ECs. No appreciable changes in single EC forces were observed (Supplemental Fig. S3).
Substrate stiffening promotes thrombin-induced gap formation.
To investigate how substrate stiffening regulates EC monolayer integrity, we exposed EC monolayers to thrombin (1 U/ml), a hyperpermeability stimulus (17, 51). In response to thrombin's effect on cells on soft substrates, cells comprising the monolayer contracted collectively, visible gaps did not form, and the monolayer stayed intact (Fig. 1A). In response to thrombin's effect on cells on stiff substrates, by contrast, cells comprising the monolayer contracted individually, large gaps arose between adjacent cells, and the monolayer became severely disrupted (Fig. 1B).
On a soft substrate, thrombin induced the formation of diffuse F-actin filaments. On a stiff substrate, by contrast, thrombin induced the formation of prominent F-actin stress fibers (Fig. 1D).
With thrombin, gaps form where force enhancements are largest.
Next, we used traction force microscopy to measure the underlying physical forces. In response to thrombin, the traction force magnitudes were enhanced by about a factor of three for 1.2- and 4-kPa substrates and by a factor of two for 11-kPa substrates with reference to their respective baseline values (Fig. 1C). On softer substrates, force enhancement occurred predominately at preexisting hot spots at the monolayer edge, whereas on stiffer substrates, force enhancement occurred within central regions of the monolayer (Fig. 1, A and B). The locus of enhanced forces correlated spatially with the formation of the largest gaps (Fig. 1B, arrows).
With thrombin, gap formation is associated with Rho kinase activity.
A key downstream effector of thrombin-induced effects is RhoA, a member of the Rho family of small GTPases (17, 53#x2013;55). RhoA activates Rho kinase, which in turn leads to the phosphorylation of the myosin binding subunit of myosin light chain phosphatase (PP1), thus enhancing EC actomyosin-mediated contractility. Together, these changes promote endothelial barrier disruption (7, 15, 17, 22, 43, 53–55). We therefore hypothesized that thrombin-induced endothelial barrier disruption, as observed on stiffer substrates (Fig. 1B), is associated with enhanced Rho kinase activity.
Western blot analysis revealed that thrombin induced a robust increase in Rho kinase activity on the soft 1.2-kPa substrate that was further enhanced on the stiffer 11-kPa substrate by a factor of 1.5 (Fig. 3A) and on the more extremely stiff 90-kPa substrate by a factor of 2 (Fig. 3B). Together with force changes (Fig. 1), these findings implicate a mechanism in which Rho kinase enhances monolayer forces and promotes ensuing gap formation.
Rho kinase inhibition lowers basal forces and prevents gap formation.
To further investigate the relationship between Rho kinase activity and monolayer forces, we treated monolayers plated across a range of substrate stiffnesses with a Rho kinase inhibitor, Y-27632 (20), and probed for underlying force changes. Rho kinase inhibition substantially decreased basal EC monolayer forces (Fig. 4). In response to thrombin, these decreased basal forces favored monolayer integrity, even on the stiffest substrate (compare Fig. 4, insets, with Fig. 1B).
With Rho kinase inhibition we could detect no effect on basal barrier permeability but observed significantly reduced hyperpermeability associated with disruption of VE-cadherin junctions (Fig. 5). In untreated controls, disruption of VE-cadherin was not associated with the formation of new F-actin stress fibers, focal adhesions (Supplemental Fig. S4), or an increase in RhoA/Rho kinase activity (Supplemental Fig. S5). These results suggest that Rho kinase inhibition may be reducing hyperpermeability through reduced basal presence of F-actin filaments and reduced Rho kinase activity, as well as through reduced basal EC monolayer forces.
In this context, it has been reported previously that disrupting endothelial cell contacts alone does not automatically lead to the opening of cell junctions when the Rac, ROS, and Pyk signaling pathways are disrupted (50). These data further implicate a Rho-like small GTPase in EC monolayer disruption, as well as an EC monolayer force-driven mechanism.
Implications for vascular biology.
The physiological range of cell substrate stiffness in the body varies from ∼1 kPa in the brain (18), ∼5 kPa in the alveolar wall (45), and ∼10 kPa in the corneal basement membrane (27) to ∼30 kPa in precalcified bone (12, 13, 16). These values depend on tissue source and species, and in a given tissue, they can vary with location (12, 13, 59). These substrate stiffness values are also enhanced with scarring or lesions. In atherosclerotic rabbit thoracic arteries, for example, substrate stiffness can be >100 kPa in calcified sites (32).
Correlating permeability changes with monolayer forces on physiological stiffness substrates is of great significance. Compared with this physiological stiffness range (kPa), existing techniques in vascular biology, including hydraulic conductivity assays that measure passage of macromolecules (Transwell assays) and transendothelial electrical resistance assays such as ECIS (electric cell-substrate impedance sensing) that measure monolayer electric resistance (47, 54), are performed on substrates that are virtually rigid (in the GPa range). On the basis of our current findings, we predict in those assays abnormally high basal cell contractile forces, altered cytoskeletal structure, and modified levels of cell signaling. Thus barrier permeability measurements reported previously on rigid substrates may not reflect in situ values.
Our findings also have strong implications for barrier permeability with respect to leukocyte transendothelial migration. As with thrombin, leukocyte binding to endothelium drives activation of RhoA and actin stress fiber formation and facilitates leukocyte paracellular transendothelial migration (29). Our studies posit that enhanced forces exerted by ECs on plastic or glass substrates may be important, but hidden modulators of paracellular permeability, which in turn may affect leukocyte migration. Moreover, these forces may be perceived by the interacting leukocytes as apparent substrate stiffening, which may directly influence their morphology, adhesion, spreading, and migration (36, 39). These important physiological effects remain to be directly verified.
In summary, we present in this study experimental evidence establishing a powerful and previously unanticipated role of substrate stiffness in EC monolayer integrity. By studying micropatterned HUVEC monolayers, we elucidated underlying physical forces and discovered that enhancement of these forces by stiffer substrates promotes monolayer gap formation (Fig. 6). We further elucidated that lowering baseline traction forces provides a novel mechanism by which Rho kinase inhibitors protect the endothelium from hyperpermeability challenges. Finally, our finding that enhanced substrate stiffness promotes endothelial gap formation also sheds new light on the hitherto unexplained observation that although gross similarities exist between in situ and in vitro endothelial responses, EC monolayers in situ do not respond to permeability inducers with obvious gap formation (55), as observed in vitro in monolayers plated on rigid plastic or glass substrates.
G. P. van Nieuw Amerongen was supported by The Netherlands Heart Foundation (The Hague) Grant 2003T032. V. W. M. van Hinsbergh was supported by The Netherlands Initiative for Regenerative Medicine. R. Krishnan was supported in part by the Parker Francis Foundation. C. V. Carman was supported by the American Heart Association and the Roche Organ Transplant Research Foundation. X. Trepat was supported by the European Research Council and the Spanish Ministry of Science and Innovation.
No conflicts of interest, financial or otherwise, are declared by the author(s).
- Copyright © 2011 the American Physiological Society