Figure 5
BEC- and LEC-defining gene expression is lost in vitro. (A) BEC and LEC transcriptomes are most divergent in vivo but assimilate in vitro. The 60 genes showing the highest coefficient of variation of expression in normalized datasets from freshly isolated (f)–BECs, f-LECs, cultured (c)–BECs, and c-LECs harvested at the indicated passages (passage 1 to 6) were subjected to unsupervised hierarchical clustering. Each column shows the expression of one gene in the various samples analyzed and color ranges indicate the relative expression level (red, high; green, low). Note that 2 distinctive clusters of expressed genes define f-BECs and f-LECs. With few exceptions (eg, PDPN), both gene expression patterns are lost in vitro. (B-C) Preferential loss of in vivo BEC- and LEC-defining gene expression but maintenance of pan-EC gene expression in vitro. Pan-EC–specific genes (black symbols), BEC-specific genes (red symbols), and LEC-specific genes (green symbols) were ranked according to their power to discriminate ECs from non-ECs, BECs from LECs, and LECs from BECs. The percentage of genes that were down-regulated at least 5-fold in the comparison of fresh and cultured ECs was plotted against the number of ranked genes analyzed. Note that a high percentage of the 40 genes that best discriminate BECs or LECs are down-regulated in vitro. The next 160 ranked genes are lost with much lower prevalence. In contrast, few pan-EC genes are down-regulated in vitro. (C) The alteration of expression of down-regulated pan-EC genes is of similar magnitude in cultured BECs and in cultured LECs. Shown is the expression of genes that are at least 5-fold down-regulated on culture of LECs and BECs. Log2-transformed ratios of mean expression values obtained with fresh and cultured EC subsets are shown. The degree of down-regulation of gene expression is similar in LECs (y-axis) and BECs (x-axis); R2 = 0.65.

BEC- and LEC-defining gene expression is lost in vitro. (A) BEC and LEC transcriptomes are most divergent in vivo but assimilate in vitro. The 60 genes showing the highest coefficient of variation of expression in normalized datasets from freshly isolated (f)–BECs, f-LECs, cultured (c)–BECs, and c-LECs harvested at the indicated passages (passage 1 to 6) were subjected to unsupervised hierarchical clustering. Each column shows the expression of one gene in the various samples analyzed and color ranges indicate the relative expression level (red, high; green, low). Note that 2 distinctive clusters of expressed genes define f-BECs and f-LECs. With few exceptions (eg, PDPN), both gene expression patterns are lost in vitro. (B-C) Preferential loss of in vivo BEC- and LEC-defining gene expression but maintenance of pan-EC gene expression in vitro. Pan-EC–specific genes (black symbols), BEC-specific genes (red symbols), and LEC-specific genes (green symbols) were ranked according to their power to discriminate ECs from non-ECs, BECs from LECs, and LECs from BECs. The percentage of genes that were down-regulated at least 5-fold in the comparison of fresh and cultured ECs was plotted against the number of ranked genes analyzed. Note that a high percentage of the 40 genes that best discriminate BECs or LECs are down-regulated in vitro. The next 160 ranked genes are lost with much lower prevalence. In contrast, few pan-EC genes are down-regulated in vitro. (C) The alteration of expression of down-regulated pan-EC genes is of similar magnitude in cultured BECs and in cultured LECs. Shown is the expression of genes that are at least 5-fold down-regulated on culture of LECs and BECs. Log2-transformed ratios of mean expression values obtained with fresh and cultured EC subsets are shown. The degree of down-regulation of gene expression is similar in LECs (y-axis) and BECs (x-axis); R2 = 0.65.

Close Modal

or Create an Account

Close Modal
Close Modal