Tumor-derived endothelial cells exhibit aberrant Rho-mediated mechanosensing and abnormal angiogenesis in vitro

Abstract

Tumor blood vessels exhibit abnormal structure and function that cause disturbed blood flow and high interstitial pressure, which impair delivery of anti-cancer agents. Past efforts to normalize the tumor vasculature have focused on inhibition of soluble angiogenic factors, such as VEGF; however, capillary endothelial (CE) cell growth and differentiation during angiogenesis are also influenced by mechanical forces conveyed by the extracellular matrix (ECM). Here, we explored the possibility that tumor CE cells form abnormal vessels because they lose their ability to sense and respond to these physical cues. These studies reveal that, in contrast to normal CE cells, tumor-derived CE cells fail to reorient their actin cytoskeleton when exposed to uniaxial cyclic strain, exhibit distinct shape sensitivity to variations in ECM elasticity, exert greater traction force, and display an enhanced ability to retract flexible ECM substrates and reorganize into tubular networks in vitro. These behaviors correlate with a constitutively high level of baseline activity of the small GTPase Rho and its downstream effector, Rho-associated kinase (ROCK). Moreover, decreasing Rho-mediated tension by using the ROCK inhibitor, Y27632, can reprogram the tumor CE cells so that they normalize their reorientation response to uniaxial cyclic strain and their ability to form tubular networks on ECM gels. Abnormal Rho-mediated sensing of mechanical cues in the tumor microenvironment may therefore contribute to the aberrant behaviors of tumor CE cells that result in the development of structural abnormalities in the cancer microvasculature.

Fig. 1.

Tumor CE cells reorient abnormally in response to uniaxial cyclic strain. (A) Fluorescent micrographs of normal (MDCE) and tumor CE cells cultured on FN-coated flexible silicone substrates and subjected to either no strain (Control) or 10% uniaxial cyclic strain (Strain) for 18 h at 1 Hz; arrow indicates the direction of applied strain. Actin stress fibers and nuclei were visualized by staining with Alexa4 Fluor 88-phalloidin and DAPI, respectively. (B) Computerized morphometric quantitation of the reorientation response in cells cultured in the absence (Control, white bars) or presence of 10% uniaxial cyclic strain (Strain, black bars). Data are presented as the percentage cells that reoriented 90° ± 30° relative to the direction of applied strain (**, P < 0.001). (Scale bar, 50 μm.)

Fig. 2.

Tumor CE cells respond abnormally to variations in ECM elasticity. (A) Phase contrast micrographs showing normal and tumor CE cell spreading after 6 h of culture on compliant transglutaminase-cross-linked gelatin hydrogels of varying stiffness (98, 370, or 2,280 Pa). (B) Average projected cell areas measured for cells cultured as described in A (*, P < 0.05; ** P < 0.001; each sample was compared with its immediately preceding data point). (Scale bar, 100 μm.)

Fig. 3.

Normal and tumor CE cells exhibit distinct angiogenic capabilities in vitro. Phase contrast micrographs showing normal and tumor CE cells plated either on the surface of 3D fibrin gels at the indicated plating densities and cultured for 18 h (A) or within 3D Matrigel at 5 × 10⁶ cells per milliliter and cultured for 2 wk (B). (Scale bar, 100 μm.)

Fig. 4.

Tumor CE cells exert stronger Rho/ROCK-mediated traction. (A) Traction force microscopy results showing that tumor CE cells are more contractile than normal CE cells. (Left) Phase contrast and traction field maps of normal and tumor CE cells cultured on FN-coated flexible polyacrylamide gels for 4–6 h before measurement of cell traction. White arrows in traction maps indicate the direction of cell-generated traction forces on the underlying substrate; the color scale indicates the magnitude of traction (Pa). (Right) Quantitation of traction field maps using a previously described Matlab algorithm (47) showing the average ± SEM of maximum cumulative traction force (nN) exerted by individual CE cells (n ≥ 7 per condition). (B) Tumor CE cells display higher baseline Rho activity than normal CE cells. Normal and tumor CE cells grown on FN-coated flexible silicone substrates were subjected to either no strain (Control) or 10% uniaxial cyclic strain (Strain) for 2 h, and Rho activity was analyzed by using the Rhotekin-RBD binding assay. (Upper) Representative western blot showing the levels of active GTP-Rho and total Rho for normal and tumor CE cells in the presence or absence of applied strain. (Lower) Relative changes in Rho activity under the various experimental conditions. GTP-Rho levels were measured as a percentage of total Rho levels and normalized to basal GTP-Rho levels in normal CE cells (**, P < 0.002 for comparison of baseline Rho levels in normal versus tumor CE cells). (C) ROCK activity in normal and tumor CE cells cultured under regular growth conditions, as determined by using a commercially available Rho-kinase assay. Bar graphs indicate the average OD450 value from triplicates.

Fig. 5.

Decreasing Rho-mediated tension significantly normalizes tumor CE behavior. (A) Inhibiting cell tension by pretreating CE cells with the ROCK inhibitor, Y27632, caused tumor CE cells to realign normally and reorient their actin stress fibers perpendicular to the applied strain direction, whereas it exerted an opposite effect on normal CE cells, as shown in the immunofluorescence micrographs. (Scale bar, 50 μm.) (B) Computerized morphometric quantitation of the reorientation response in cells (shown in A) cultured in the absence (Control, white bars) or presence of 10% uniaxial cyclic strain without (Strain, black bars) or with (Y27+Strain, gray bars) Y27632 pretreatment. Data are presented as the percentage of cells that reoriented 90° ± 30° relative to the direction of applied strain (*, P < 0.01; **, P < 0.001 for comparisons between strained samples treated with or without Y27632). (C) Phase contrast micrographs showing the opposing effects of Y27632-mediated reduction in cell tension on the angiogenic behavior of normal and tumor CE cells, when cultured on the surface of 3D fibrin gels. (Scale bar, 100 μm.)