Perivascular cells function as key mediators of mechanical and structural changes in vascular capillaries

Abstract

A hallmark of chronic and inflammatory diseases is the formation of a fibrotic and stiff extracellular matrix (ECM), typically associated with abnormal, leaky microvascular capillaries. Mechanisms explaining how the microvasculature responds to ECM alterations remain unknown. Here, we used a microphysiological model of capillaries on a chip mimicking the characteristics of healthy or fibrotic collagen to test the hypothesis that perivascular cells mediate the response of vascular capillaries to mechanical and structural changes in the human ECM. Capillaries engineered in altered fibrotic collagen had abnormal migration of perivascular cells, reduced pericyte differentiation, increased leakage, and higher regulation of inflammatory/remodeling genes, all regulated via NOTCH3, a known mediator of endothelial-perivascular cell communication. Capillaries engineered either with endothelial cells alone or with perivascular cells silenced for NOTCH3 expression showed a minimal response to ECM alterations. These findings reveal a previously unknown mechanism of vascular response to changes in the ECM in health and disease.

Fig. 1.. Device engineering and collagen characterization.

(A) Schematic diagram showing the steps to engineer perivascularly supported capillaries on-a-chip with different collagen stiffness and microarchitectures. h, hours. (B) Collagen fibers became progressively thicker and more bundled with lower temperatures in a controllable manner as demonstrated by the second harmonic generation images. (C) Representative scanning electron microscopy images of collagen mesh after fibrillogenesis with different temperatures with gradual changes from soft reticular to stiff bundled. (D) Altering the fibrillogenesis temperatures did not change the bulk modulus but caused the average single collagen fibril elastic moduli to vary from 4.7 to 27.7 kPa, with progressive bundling of collagen fibers and heterogeneous pore distribution. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 2.. Vasculature characterization.

(A and C) Engineered capillaries with only ECs presented similar morphologies irrespective of collagen stiffness and microarchitecture. The presence of PCs resulted in more disrupted morphology and abluminal migrating cells in stiff bundled collagen (D) and (S). In addition, the presence of PCs (B and D) markedly changed the morphology and number of capillaries within stiff bundled collagen, leading to larger and more irregular vessels than in soft reticular collagen. (E to L and T) Capillaries engineered in both soft reticular and stiff bundled collagen showed angiogenic sprouts; however, only those in soft reticular fibrils had pericyte coverage [(E) to (H)]. (M to O) Capillaries engineered in soft reticular collagen also showed abundant pericyte coverage, while vasculature within stiff bundled collagen had fewer NG2-positive cells associated with the endothelial wall (P to R and U). *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 3.. Cell-cell junctions, basement membrane, and barrier function of the vasculature.

(A to D) Capillaries with only ECs expressed PECAM-1 at comparable levels, while PECAM-1 is increased in capillaries with PCs in soft reticular collagen. Heterogeneous and decreased expression of PECAM-1 is observed in capillaries engineered with PCs in stiff bundled collagen. (E and F) p-pax is overexpressed in vasculature engineered in stiff collagen. (G to J) Laminin (LAM) was expressed in both groups regardless of the absence of PCs; however, PCs in stiff bundled collagen showed irregular deposition of laminin across the capillary. (K to N) Vascular capillaries engineered in soft reticular collagen preserved the barrier function regardless of the presence of PCs, while stiff bundled collagen was associated with reduced barrier function even in the presence of PCs. *P < 0.05, **P < 0.01, and ***P < 0.001.

Fig. 4.. Gene analysis.

(A) Heatmap comparing the effect of collagen stiffness and architecture on gene expression in the presence of PCs. (B) Volcano plot showing that genes encoding for DCN, CXCL8, and TIMP1 were most highly expressed by the vasculature in a stiff environment. (C) Pairwise comparison of the nine most different log values for gene expression. (D) Enrichment score and pathways associated with capillaries engineered in either soft reticular or stiff bundled collagen. (E) Engineered bone marrow cells were seeded on top of the soft and stiff vasculature and allowed to interact for 3 days. In the presence of the vasculature in the stiff matrix, bone marrow cells tended to express more monocytic and myeloid markers consistent with IL-8 (inflammatory) stimulation.

Fig. 5.. Normalization of vasculature after NOTCH3 silencing.

Morphology (A, B, E, and F), pericyte coverage (C and G), and PECAM-1 expression (D and H) were similar between the vasculature engineered in a soft reticular or stiff bundled environment. (I to K) Migrating cells, pericyte coverage, and vessel leakage were around 10%, similar in both groups. (L) In addition, IL-8 secretion was comparable for both groups. Nonsilenced controls are shown in fig. S20.