Subendothelial stiffness alters endothelial cell traction force generation while exerting a minimal effect on the transcriptome

Abstract

Endothelial cells respond to changes in subendothelial stiffness by altering their migration and mechanics, but whether those responses are due to transcriptional reprogramming remains largely unknown. We measured traction force generation and also performed gene expression profiling for two endothelial cell types grown in monolayers on soft or stiff matrices: primary human umbilical vein endothelial cells (HUVEC) and immortalized human microvascular endothelial cells (HMEC-1). Both cell types respond to changes in subendothelial stiffness by increasing the traction stresses they exert on stiffer as compared to softer matrices, and exhibit a range of altered protein phosphorylation or protein conformational changes previously implicated in mechanotransduction. However, the transcriptome has only a minimal role in this conserved biomechanical response. Only few genes were differentially expressed in each cell type in a stiffness-dependent manner, and none were shared between them. In contrast, thousands of genes were differentially regulated in HUVEC as compared to HMEC-1. HUVEC (but not HMEC-1) upregulate expression of TGF-β2 on stiffer matrices, and also respond to application of exogenous TGF-β2 by enhancing their endogenous TGF-β2 expression and their cell-matrix traction stresses. Altogether, these findings provide insights into the relationship between subendothelial stiffness, endothelial mechanics and variation of the endothelial cell transcriptome, and reveal that subendothelial stiffness, while critically altering endothelial cells' mechanical behavior, minimally affects their transcriptome.

Figure 1

Endothelial cells in monolayers exert higher cell-matrix traction forces onto stiff as compared to soft hydrogels. (a) Representative phase contrast images (phase, first column) and cell-matrix deformation maps (second column, color indicates deformation magnitude in μm) and traction stresses (third column, color indicates stress magnitude in Pa) exerted by confluent HUVEC adherent onto soft 3 kPa or stiff 35 kPa hydrogels. (b,c) Time evolution of the integral of the traction force magnitude over the whole field of view to its area (nN/μm²) (b) and of the total strain energy imparted by the cells per area of field of view (nN/μm) (c) calculated for two different regions within confluent HUVEC monolayers for cells residing on soft 3 kPa (blue) or stiff 35 kPa (red) matrices. (d–f) Same as in panels a-c but corresponding to HMEC-1 monolayers.

Figure 2

Post-transcriptional changes on ECs in monolayers grown on soft versus stiff hydrogels. (a) Western blots from whole HUVEC or HMEC-1 lysates of cells previously residing on soft gels (3 kPa) or stiff gels (70 kPa). Representative cropped blots are displayed and full-length blots can be found in the supplementary material (Supplementary Fig. S2). Each row shows a different protein whose expression, phosphorylation or conformation was probed, namely: active form of integrin β1, total integrin β1, p397FAK, total FAK, p-Vav2, total Vav2, p-p70S6K, total p70S6K, p-ERK, total ERK and GAPDH (used as loading control). Experiments were performed N = 3 times. (b) Bar plots show relative expression of the proteins probed in panel a for HUVEC cells residing on soft (blue) or stiff (red) gels. All measurements were normalized to GAPDH expression for each condition, and expressed as fold-change relative to the median expression level on soft substrates. One or two asterisks denote statistically significant differences between the medians of two distributions (<0.05 or <0.01 respectively; unpaired t-test) and non-significant differences are denoted as ns. (c) Same as panel b but for HMEC-1.

Figure 3

Endothelial origin but not matrix stiffness strongly determines the transcriptome of endothelial cells. (a,c,e,g) Scatter plots of expressed genes showing normalized counts of gene expression in the x and y axes for the indicated groups: (a) HUVEC on stiff 70 kPa (N = 6) versus soft 3 kPa matrices (N = 6), (c) HMEC-1 on stiff 70 kPa (N = 6) versus soft 3 kPa matrices (N = 6), and (d) HUVEC (N = 6) versus HMEC-1 on stiff 70 kPa matrices (N = 6) and (g) HUVEC (N = 6) versus HMEC-1 on soft 3 kPa matrices (N = 6). Light gray dots represent genes that are not differentially expressed while differentially expressed genes (DEGs) are shown as dots color-coded by their -log₁₀ p-values. (b,d,f,h) Volcano plots showing DEGs between the same groups compared as above. The -log₁₀ p-values (y-axis) are plotted against the average log₂ fold changes in expression (x-axis). Non DEGs are plotted in light gray. DEGs are color-coded depending on the log₁₀ of their mean normalized counts.

Figure 4

Principal component analysis (PCA) confirms that endothelial origin but not matrix stiffness is a major contributor of expression differences. (a,b) PCA on top 500 DEGs for HUVEC on soft 3 kPa (turquoise circles) and on stiff 70 kPa (purple circles) matrices and for HMEC-1 on soft 3 kPa (orange circles) and on stiff 70 kPa (green circles) matrices. PC1 versus PC2 is shown in panel A and PC3 versus PC4 is shown in panel B. The 95% confidence ellipse is shown for each group with the corresponding colors. (c) Scree plot showing in decreasing order the proportion of variance explained by each PCA mode up to PC10.

Figure 5

TGF-β2 modulates expression of some stiffness-sensitive genes in HUVEC. (a–p) Relative with respect to GAPDH expression levels of the indicated stiffness-sensitive DEGs obtained by RT-qPCR. For each boxplot N = 3 replicates are shown for each group treated with either vehicle control (#1, #4), or treated for 24 h with 1 ng/mL (#2, #5) or 10 ng/mL TGF-β2 (#3, #6). HUVEC either were residing on soft 3 kPa (blue) or stiff 70 kPa (red) matrices and the relative levels of expression in each treated sample (#2-#4 or #6-#8) are expressed relative to the vehicle control samples of cells residing on soft (#1) or stiff (#4) matrices respectively.

Figure 6

HUVEC but not HMEC-1 treated with recombinant TGF-β2 show a significant elevation in their traction stresses on both stiff and soft matrices. Boxplots of the strain energy imparted by confluent EC monolayers calculated during different instants of time and for multiple fields of view (N = 200). Boxplots refer to HUVEC or HMEC-1 residing on soft 3 kPa (blue) or stiff 35 kPa matrices (red) treated with vehicle control or 1 ng/mL TGF-β2 for 24 h prior to imaging. Each boxplot is normalized with respect to the mean value of the vehicle control case. Two asterisks denote statistically significant differences between the medians of two distributions (p < 0.01; Wilcoxon rank sum test).

Figure 7

F-actin and pMLC increase for HUVEC monolayers exposed to TGF-β2. (a–d) Representative images depicting the phase image of cells (first column), F-actin fluorescence (second column), anti-pMLC antibody fluorescence (third column) and the image of the nuclei (fourth column) for HUVEC residing on soft 3 kPa matrices and treated with vehicle control (a) or 1 ng/mL TGF-β2 for 24 h (b) or HUVEC residing on stiff 70 kPa matrices and treated with vehicle control (c) or 1 ng/mL TGF-β2 for 24 h (d). Scale bar is 50 μm. (e,f) Boxplots showing the integral of the F-actin fluorescence intensity (e) or pMLC (f) over different fields of view (N = 10) for HUVEC residing on soft 3 kPa matrices (blue) and treated with vehicle control or 1 ng/mL TGF-β2 for 24 h and for HUVEC residing on stiff (red) 70 kPa matrices and treated with vehicle control or 1 ng/mL TGF-β2 for 24 h. N = 400–500 cells were analyzed for each condition. One or two asterisks denote statistically significant differences between the medians of two distributions (<0.05 or <0.01 respectively; Wilcoxon rank sum test). (g,h) Same as panels e-f but for HMEC-1 cells.