Stiff Substrates Increase Inflammation-Induced Endothelial Monolayer Tension and Permeability

Abstract

Arterial stiffness and inflammation are associated with atherosclerosis, and each have individually been shown to increase endothelial monolayer tension and permeability. The objective of this study was to determine if substrate stiffness enhanced endothelial monolayer tension and permeability in response to inflammatory cytokines. Porcine aortic endothelial cells were cultured at confluence on polyacrylamide gels of varying stiffness and treated with either tumor necrosis factor-α (TNFα) or thrombin. Monolayer tension was measured through vinculin localization at the cell membrane, traction force microscopy, and phosphorylated myosin light chain quantity and actin fiber colocalization. Cell permeability was measured by cell-cell junction confocal microscopy and a dextran permeability assay. When treated with TNFα or thrombin, endothelial monolayers on stiffer substrates showed increased traction forces, vinculin at the cell membrane, and vinculin phosphorylation, suggesting elevated monolayer tension. Interestingly, VE-cadherin shifted toward a smaller molecular weight in endothelial monolayers on softer substrates, which may relate to increased VE-cadherin endocytosis and degradation. Phosphorylated myosin light chain colocalization with actin stress fibers increased in endothelial monolayers treated with TNFα or thrombin on stiffer substrates, indicating elevated cell monolayer contractility. Endothelial monolayers also developed focal adherens intercellular junctions and became more permeable when cultured on stiffer substrates in the presence of the inflammatory cytokines. Whereas each of these effects was likely mitigated by Rho/ROCK, Rho/ROCK pathway inhibition via Y27632 disrupted cell-cell junction morphology, showing that cell contractility is required to maintain adherens junction integrity. These data suggest that stiff substrates change intercellular junction protein localization and degradation, which may counteract the inflammation-induced increase in endothelial monolayer tension and thereby moderate inflammation-induced junction loss and associated endothelial monolayer permeability on stiffer substrates.

Figure 1

Stiff substrates enhanced vinculin colocalization with VE-cadherin at apical adherens junctions in response to TNFα and thrombin and altered VE-cadherin molecular weight. (a) Cell monolayers on 6-, 14-, and 29-kPa substrates were treated with 10 ng/mL TNFα for 3 h or 10 U/mL thrombin for 30 min, fixed, and labeled for VE-cadherin (red, left), vinculin (green, right), and nuclei (blue). Confocal z stacks were acquired at 60× magnification, and the uppermost plane was selected as the apical cell surface. Representative cells from within confluent monolayers are shown to highlight differences in vinculin localization. Scale bars represent 10 μm. (b) Shown here is quantification of the fraction of VE-cadherin colocalized with vinculin from images in (a). Data are shown as average mean ± SE. #p < 0.05, ^∗p < 0.01, and ^∗∗p < 0.001 when compared to untreated sample at a given substrate stiffness (Tukey-Kramer posthoc test). (c) Given here are total vinculin, p-vinculin (Y-822), VE-cadherin (C19, intracellular), VE-cadherin (BV9, extracellular), and GAPDH protein levels in cells that were untreated, TNFα-treated (10 ng/mL, 3 h), or thrombin-treated (10 U/mL, 30 min) on 6-, 14-, and 29-kPa substrates or tissue culture plastic). (d) Quantification of VE-cadherin (80–105 kDa) protein levels normalized to GAPDH. To see this figure in color, go online.

Figure 2

Total pMLC was greater in endothelial monolayers on stiff substrates in response to TNFα and thrombin. (a) Endothelial monolayers on gels of varying stiffness were treated with 10 ng/mL TNFα for 3 h or 10 U/mL thrombin for 30 min, fixed, and labeled for pMLC. Linear pMLC represents localization along actin fibers. (b) Images were processed and quantified to assess total pMLC. By two-way ANOVA, TNFα and thrombin significantly increased pMLC labeling (p < 0.01). The interactions between substrate stiffness/TNFα and substrate stiffness/thrombin were significant (p < 0.05 and p < 0.01, respectively). ^∗p < 0.01, ^∗∗p < 0.001 (Tukey-Kramer posthoc test). Data are shown as average mean ± SE. Scale bar represents 25 μm.

Figure 3

Substrate stiffening exacerbated reticular adherens junction thinning in response to TNF-α and thrombin. Cells on 6-, 14-, and 29-kPa polyacrylamide gels were untreated, treated with 10 ng/mL TNF-α for 3 h, or treated with 10 U/mL thrombin for 30 min before fixation. (a) Images show β-catenin labeling in confluent monolayers. Scale bar represents 10 μm. Specific junctions are magnified in (b) to highlight differences in junction morphology. (c) Shown here is quantification of junction scoring by two independent observers, expressed as a ratio of linear/reticular junctions. ^∗p < 0.01 when compared to untreated sample at a given substrate stiffness (Tukey-Kramer posthoc test). Data are shown as average mean ± SE.

Figure 4

Endothelial monolayer permeability increased with substrate stiffness and TNFα and thrombin treatment. (a) Given here are representative xz images of permeability samples with and without thrombin treatment. Dotted lines indicate the approximate location of the endothelial monolayer; the gel substrate is beneath the dotted lines. Monolayer permeability was quantified by measuring fluorescence intensity in the gel. Fluorescence values for each sample were normalized to the fluorescence in gels without endothelial monolayers; this output is referred to as “normalized intensity”, which increases with monolayer permeability. (b) Monolayer permeability was measured in samples that were untreated, treated with 10 ng/mL TNFα for 3 h, or treated with 10 U/mL thrombin for 30 min. Data are shown as average mean ± SE. The effect of stiffness (p < 0.001) and treatment (p < 0.01) were significant by n-way ANOVA with a Tukey-Kramer posthoc test (^∗∗p < 0.001). The interaction between stiffness and thrombin was significant (p < 0.05); however, there was no significant interaction between stiffness and TNFα. To see this figure in color, go online.

Figure 5

ROCK inhibition abolished pMLC localization to actin fibers and vinculin localization to apical adherens junctions with TNFα or thrombin treatment in endothelial monolayers on all substrate stiffnesses. Endothelial monolayers on 6-, 14-, and 29-kPa substrates were pretreated with 5 μM Y-27632 for 30 min before the addition of 10 ng/mL TNFα (3 h) or 10 U/mL thrombin (30 min). Samples were fixed and labeled for (a) pMLC (green) and nuclei (blue); scale bars represent 25 μm or (b) VE-cadherin (red, left), vinculin (green, right), and nuclei (blue). Images are the apical plane of representative cells from confluent monolayers; scale bars represent 10 μm. To see this figure in color, go online.