Emergent mechanical control of vascular morphogenesis

Abstract

Vascularization is driven by morphogen signals and mechanical cues that coordinately regulate cellular force generation, migration, and shape change to sculpt the developing vascular network. However, it remains unclear whether developing vasculature actively regulates its own mechanical properties to achieve effective vascularization. We engineered tissue constructs containing endothelial cells and fibroblasts to investigate the mechanics of vascularization. Tissue stiffness increases during vascular morphogenesis resulting from emergent interactions between endothelial cells, fibroblasts, and ECM and correlates with enhanced vascular function. Contractile cellular forces are key to emergent tissue stiffening and synergize with ECM mechanical properties to modulate the mechanics of vascularization. Emergent tissue stiffening and vascular function rely on mechanotransduction signaling within fibroblasts, mediated by YAP1. Mouse embryos lacking YAP1 in fibroblasts exhibit both reduced tissue stiffness and develop lethal vascular defects. Translating our findings through biology-inspired vascular tissue engineering approaches will have substantial implications in regenerative medicine.

Fig. 1.. Emergence of enhanced stiffening through juxtacrine interactions.

(A) Seven-channel polydimethylsiloxane (PDMS) device to study microvasculature formation with three parallel hydrogel channels separated by growth medium perfusion channels. MC condition: ECs encapsulated in central vascularization hydrogel channel and cell-free hydrogels in side channels; PCC condition: ECs encapsulated in vascularization channel and fibroblasts encapsulated in side channels; JCC condition: ECs and fibroblasts encapsulated together in vascularization channel and cell-free hydrogels in side channels. (B) Fluorescent dextran perfusion assay. For fully formed vasculature, dextran flows through the vessel lumens and remains within vessel walls (i.e., JCC days 4 and 7). For unconnected vascular structures, dextran flows freely into the fibrin gel (i.e., MC days 1, 4, and 7). Scale bar, 200 μm. (C to E) Fold increases (means ± SEM) in length of the longest connected network, perfusability, and barrier function compared to MC day 1. (F) Manual stretching of fibrin gel or JCC vascularized tissue constructs (day 7) removed from device and anchored at one end while clamped and pulled at the other end. JCC tissue withstood extensive stretch until rupture (300% versus 100% for fibrin gel) and could be reversibly stretched by up to ~100%. Scale bar, 5 mm. (G) Apparent elastic modulus of the vascularized tissue, measured by AFM indentation. FB condition: fibroblasts encapsulated in central vascularization channel and cell-free hydrogels in side channels. Inset: Fold increases (means ± SEM) in elastic modulus compared to stiffness of MC day 1. At least 50 AFM indentation measurements on N = 3 independent devices were performed per condition. Kruskal-Wallis test followed by Dunn’s multiple comparison test: *P < 0.01 and ***P < 0.0001 (table S1). ns, not significant.

Fig. 2.. Temporal correlation of stiffness with morphological and functional characteristics.

(A) Three-channel PDMS device comprising a central vascularization channel (pink) and two growth medium perfusion channels (blue). (a and b) 2D and 3D views. (c) Channel dimensions. (d) PDMS lips extending from the top surface prevent leakage of prepolymerized hydrogel from central vascularization channel into medium perfusion channels via surface tension. (B) Confocal imaging of vascular morphogenesis progression for green fluorescent protein (GFP)–HUVECs and red fluorescent protein–fibroblasts in JCC condition. Scale bar, 100 μm. (C) Visualization of EC-fibroblast interactions. Scale bar, 100μm (top left). Bottom left: Magnified image of the yellow box showing fibroblasts extending along the outside of the EC vessel wall. Scale bar, 10 μm. Right images are 3D reconstructions of left images. Top right: A fibroblast extending toward multiple vessels. * denotes region of interaction. (D) Z-projected confocal image of GFP-HUVECs in JCC system perfused with blue fluorescent dextran on day 7. Bottom images are orthogonal views showing lateral (parallel to glass substrate) and transversal (perpendicular to glass substrate) cross sections with tracer contained within vascular lumens. Scale bars, 100 μm. (E) Distribution of lumen diameters on day 7, measured using images in (D). (F to H) Temporal percentage changes (means ± SEM) in network length, perfusability, and permeability, normalized to day 4 values. Inset in (H) shows the permeability values (means ± SEM). (I) Temporal fold increase (means ± SEM) in stiffness of the vascular tissue normalized to day 4. For (F to I), measurements were performed on N = 3 independent devices for each time point. Kruskal-Wallis test followed by Dunn’s multiple comparison test for day 4: *P < 0.01 and ***P < 0.0001 (table S2).

Fig. 3.. Origins of vascular tissue stiffening and evolution of strains.

(A) Vascular tissue stiffening over time for JCC control (CNT) and the softening effect of cytochalasin D (CytoD) treatment. Minimum of 50 AFM indentation measurements on N = 3 independent devices before and after treatment for each time point. Two-sided Mann Whitney U test, ***P < 0.0001. (B) Stiffness of the decellularized tissue. The inset shows percentage decrease (means ± SEM) in the stiffness of decellularized tissue compared to the stiffness of either untreated or cytochalasin D–treated vascular tissue at each time point. Two-sided Mann Whitney U test, ***P < 0.0001. (C) Fold increase (means ± SEM) in stiffness of vascularized tissue constructs for control (CNT), cytochalasin D, or decellularized treatments compared to cell-free fibrin gel. (D) Contribution of different factors to the tissue stiffness. (E) Absolute and (F) relative contribution of each factor. (G) 3D reconstructed vascularized region (size of 420 μm by 420 μm by 80 μm) showing fibroblasts (magenta) and HUVECs (gray) with superimposed, color-coded (0 to 8 μm) arrows that depict the magnitude and direction of 3D matrix displacements obtained by tracking fluorescently labeled fibrin gel. (H) Cross-sectional view of the 3D stack from (G) showing cellular structures (left; fibroblasts in magenta and HUVECs in gray), matrix strain field (center), and the overlay (right). Scale bar, 100 μm. (I) Confocal images of fibrin gel fiber structures acquired after decellularization. Thick bundles of fibers were observed at the edges of voids previously occupied by cellular structures. Scale bar, 100 μm. (J) Histograms of nonzero displacement magnitudes (induced by cytochalasin D treatment) and their relative frequencies for different time points (curves are averages of three regions).

Fig. 4.. Fibroblasts control vascular morphogenesis through YAP1.

(A) Fibroblasts were transfected with siRNAs targeting YAP1 before culture in the JCC system. Confocal images showing vasculature (GFP-HUVECs) perfused with dextran (red) in JCC control (CNT) and JCC fibroblast YAP knockdown (siRNA FBs YAP) conditions at day 7. The network perfusability and morphology were unaffected by scrambled siRNA (siRNA FBs CNT) but perturbed to varying degrees by siRNA knockdown. Scale bars, 200 μm and 50 μm (zoomed images). (B) Western blots of YAP protein in fibroblasts before siRNA transfection (Control) and at 3 and 7 days after transfection. The bar graph (bottom) shows quantification of transfection efficiency. Fibroblast YAP expression levels were normalized to β-actin values for each condition and time point before making comparisons. (C) Confocal images showing z projection and orthogonal slices of vasculature (GFP-HUVEC) perfused with dextran (red). Scale bar, 100 μm. (D) Quantification of lateral (unshaded bars) and transversal (shaded bars) diameters (means ± SD) comparing siRNA FBs YAP, CNT and siRNA FBs CNT at day 7 using images from (C). Measurements were performed on N = 4 devices from two separate experimental repeats. Linear analysis of variance (ANOVA) followed by Tukey-Kramer test, **P < 0.001 and ***P < 0.0001. (E) Stiffness of the vascular tissue constructs over time, comparing siRNA FBs YAP to CNT. Minimum of 50 AFM indentation measurements for N = 3 devices from two separate experimental repeats for each condition and time point. Kruskal-Wallis test followed by Dunn’s multiple comparison test, ***P < 0.0001. (F) Permeability (mean ± SD) at day 7 comparing siRNA FBs YAP, CNT and siRNA FBs CNT. Measurements on N = 4 devices from two separate experimental repeats. Linear ANOVA followed by Tukey-Kramer test, ***P < 0.0001.

Fig. 5.. Loss of YAP1 in fibroblasts is embryonic lethal because of vascular defects.

(A) Isolation of control (CNT), YAP^fl/+ PDGFRα-Cre positive (HET), and YAP^fl/+ PDGFRα-Cre positive (cKO) embryos at different embryonic stages shows a hemorrhage in cKO embryos at E11.5 and E12.5. Scale bars, 2 mm. (B) Hematoxylin and eosin images of E12.5 embryos show that defects in cKO embryos are restricted to the head, with high magnification panels showing vessel distension around the brain in cKO embryos. Black arrows point to erythrocytes. Scale bars, 1 mm and 100 μm in the low and high magnification panels, respectively. (C) Images show PDGFRα, GFP, and endomucin (EMCN) staining of the region of the head below the brain in control and PDGFRα-Cre × mTmG mice at E11.5. DAPI (4′,6-diamidino-2-phenylindole) stains nuclei. Scale bars, 100 μm. (D) Confocal microscopy of whole-mount immunofluorescence for endomucin (EMCN) with a focus on the central vessels in the head (white square; magnification to the right). Scale bars, 250 μm. (E) Quantification of vessel diameter (in μm) and vessel length (as defined by the distance between vessel branches in μm) on confocal whole-mount data using machine learning approaches. **** indicates different distribution of vessel sizes between cKO (N = 3, blue) versus HET (N = 2, turquoise) and CNT (N = 4, purple) (****P < 0.0001, Kolmogorov-Smirnov test). (F) Superplot of AFM data on head region of interest of freshly isolated CNT (N = 2, purple), HET (N = 3, turquoise), and cKO (N = 5, blue) embryos at E11.5. Elastic modulus of cKO was significantly different (***P < 0.001, Nested t test).