Differential gene expression in a coculture model of angiogenesis reveals modulation of select pathways and a role for Notch signaling

Abstract

Communication between endothelial and mural cells (smooth muscle cells, pericytes, and fibroblasts) can dictate blood vessel size and shape during angiogenesis, and control the functional aspects of mature blood vessels, by determining things such as contractile properties. The ability of these different cell types to regulate each other's activities led us to ask how their interactions directly modulate gene expression. To address this, we utilized a three-dimensional model of angiogenesis and screened for genes whose expression was altered under coculture conditions. Using a BeadChip array, we identified 323 genes that were uniquely regulated when endothelial cells and mural cells (fibroblasts) were cultured together. Data mining tools revealed that differential expression of genes from the integrin, blood coagulation, and angiogenesis pathways were overrepresented in coculture conditions. Scans of the promoters of these differentially modulated genes identified a multitude of conserved C promoter binding factor (CBF)1/CSL elements, implicating Notch signaling in their regulation. Accordingly, inhibition of the Notch pathway with gamma-secretase inhibitor DAPT or NOTCH3-specific small interfering RNA blocked the coculture-induced regulation of several of these genes in fibroblasts. These data show that coculturing of endothelial cells and fibroblasts causes profound changes in gene expression and suggest that Notch signaling is a critical mediator of the resultant transcription.

Fig. 1.

Angiogenesis assays reveal cell-cell communication. A: human umbilical vein endothelial cells (HUVECs) and human dermal neonatal fibroblasts (HDFNs) were cultured alone or cocultured (Coculture) in a 3-dimensional collagen I matrix for 5 days, fixed, and stained with hematoxylin or tetramethylrhodamine isothiocyanate (TRITC)-labeled endothelium-specific lectin (red) and DAPI (blue). B: triple labeling of cocultured cells with lectin (red), a preloaded cell tracker dye (green) to visualize HDFNs surrounding vessels, and DAPI (blue).

Fig. 2.

Microarray sample preparation and graph. A: diagram of method used to produce equivalent RNA samples for cells cultured alone and together. The microgram ratio of endothelial to fibroblast RNA in cocultured samples was determined to be 1:2. After angiogenesis assays, cells were collected and RNA was purified. Endothelial cell and fibroblast RNA from cells cultured separately were then mixed at a 1-to-2 ratio to produce a sample that was comparative to the cocultured sample. B: graph of the average signal intensities of genes detected on Illumina BeadChips from cRNA of cocultured samples (y-axis) vs. cRNA derived from cells cultured alone (x-axis). Gray dots represent all genes on the BeadChip, and black dots mark only those having intensities >0.99 and 2-fold differential expression with P < 0.01.

Fig. 3.

Quantitative PCR (qPCR) analysis of genes expressed in HUVECs and HDFNs cultured alone or together. To evaluate transcript levels in individual cell types that had been cocultured, cells were separated with anti-platelet endothelial cell adhesion molecule (PECAM)-1-conjugated magnetic beads. Verification of separation efficiency was determined by using the cell-specific marker genes PECAM-1 for endothelial cells and PDGF receptor β (PDGFRβ) to detect fibroblasts. Note that there are barely detectable levels of expression of these marker genes in the other cell type fraction after coculture and separation. A, cells cultured alone; CO, cells cocultured and separated. *P < 0.05.

Fig. 4.

Conditioned medium modulates select genes. Expression analysis by qPCR shows that endothelium-conditioned medium (HUVECs) added to fibroblasts causes a decrease in PDGFD and an increase in HSPB2 expression, similar to that observed from the array data. Fibroblast-conditioned medium (HDFNs) was used as control. In contrast, endothelium-conditioned medium reduced COL5A3 expression in fibroblasts, while fibroblast-conditioned medium downregulated the expression of LAMB3 in endothelial cells. *P < 0.05.

Fig. 5.

Inhibition of Notch signaling by DAPT. qPCR analysis shows expression of a representative gene that was modulated in HUVECs (F3/TF) and HDFNs (HEYL) under coculture conditions. Cocultured cells were treated with vehicle or DAPT and separated, and RNA was extracted for qPCR. Untreated cells cultured alone were used as a reference for the effects of DAPT treatment. The upregulation of HEYL was inhibited in fibroblasts by DAPT, while the increase of F3/TF in endothelial cells was not affected. *P < 0.01.

Fig. 6.

Small interfering RNA (siRNA) to block NOTCH3 in fibroblasts. NOTCH3-specific siRNA (N3) or control siRNA (C) was transfected into HDFNs, and 24 h later HUVECs were added to indicated samples and incubated for an additional 48 h. RNA was harvested, and qPCR was performed. Top: effectiveness of the NOTCH3 siRNA to abolish its expression and that of a known target gene, HEYL. Bottom: effect of NOTCH3 knockdown on DAPT-sensitive gene expression. *P < 0.05; ns, not significant.