signature=b112cd0443bdd0bf8adb37c5f5f0fe24,Tumor-endothelial interaction links the CD44(+)/CD24(-) p...

Appendix Figure 1

GO terms. Graphical visualization…

Appendix Figure 1

GO terms. Graphical visualization of the output from GO::Termfinder for process ontology:…

Appendix Figure 1

GO terms. Graphical visualization of the output from GO::Termfinder for process ontology: GOgraph layout that includes the significant GO nodes annotated by DNA replication, derived from 44 clones froma background of 1140. The color of the nodes is an indication of their Bonferroni corrected P value (orange ≤ 1e - 10; yellow = 1e - 10 to 1e-8; green = 1e-8 to 1e-6; cyan = 1e-6 to 1e-4; blue = 1e-4 to 1e-2; tan > 0.01).

Appendix Figure 2

“Tumor-endothelial cell-induced M-phase cell…

Appendix Figure 2

“Tumor-endothelial cell-induced M-phase cell cycle” genes in advanced-stage breast cancer. The expression…

Appendix Figure 2

“Tumor-endothelial cell-induced M-phase cell cycle” genes in advanced-stage breast cancer. The expression values of genes in the “tumor-endothelial cell-induced M-phase cell cycle” gene signature were extracted from a published expression study of advanced-stage breast cancers from Norway/Stanford [33]. Genes and samples are organized by hierarchical clustering. The tumors segregated into two groups defined by high (red) or low (blue) expression levels of 29 genes matching the “tumor-endothelial cell-induced M-phase cell cycle” gene signature.

Figure 1

Effect of heterotypic interaction between…

Figure 1

Effect of heterotypic interaction between an endothelial cell and a breast cancer cell…

Figure 1

Effect of heterotypic interaction between an endothelial cell and a breast cancer cell line. (A) Biologically independent replicates of the monocultured HDMEC, the breast cancer cell line Hs578T, and the mixed coculture of HDMEC and Hs578T were grown for 48 hours at low serum conditions and characterized by DNA microarray hybridization. Hierarchical clustering of a total of 1140 elements that display a greater than three-fold variance in expression in more than two different experimental samples. Data from individual elements or genes are represented as single rows, and different experiments are shown as columns. Red and green denote the expression levels of the samples. The intensity of the color reflects the magnitude of the deviation from baseline. Unsupervised hierarchical clustering of the experiments grouped the biologic replicates together. Gene expression varied considerably between HDMEC and Hs578T cultures as expected for cells of mesenchymal or epithelial origin, respectively. The coculture profile showed mainly intermediate expression levels. However, the vertical black bar marks a cluster of genes that were induced in all cocultures when compared with both monocultures, which indicated that they were induced by the heterotypic interaction. Zooming in on the genes that were upregulated in coculture revealed that they were specific for proliferation and mitosis. (B) Correlation of the measured coculture gene expression levels and their estimated expression levels based on the proportional contribution of each cell type as determined by a linear regression fit of the monoculture to the coculture data. (C) Fold change of each gene that was associated with coculturing of HDMEC and Hs578T. Genes of the “proliferation and mitosis” cluster are indicated in red. Dashed lines indicate 95% confidence limits.

Figure 2

Gene expression changes in multiple…

Figure 2

Gene expression changes in multiple cocultures of breast cancer cell lines with endothelial…

Figure 2

Gene expression changes in multiple cocultures of breast cancer cell lines with endothelial cells. Overview of collapsed data from repeat coculture experiments of eight benign and malignant epithelial cells with two different endothelial cells (HUVEC and HDMEC). The 9 monocultures and the 14 cocultures were analyzed independently in duplicates on 46 HEEBO arrays. Raw data were filtered for technical quality as described in Materials and Methods section, leaving 37,773 spots. A data distribution filter eliminating spots with a log2-based red-green normalized ratio of less than 1 in at least two arrays removed 10,183 suid(s), leaving 27,590. Genes with missing data in more than 20% of the arrays were removed, leaving 14,565. On the basis of this data set, the calculation of the interaction factors was performed for all 14 cocultures separately as described in Materials and Methods section. The interaction factors were then further analyzed for their distribution, and factors with an SD of less than 0.5 were eliminated, leaving interaction factors for 8140 genes, which are shown as a heat map after unsupervised hierarchical clustering. Red and green denote relative changes in expression that were associated with heterotypic interaction. The magnitude of the relative change is given by color intensity. Zooming in on a cluster of genes that was consistently upregulated in more than two of the cocultures revealed that they were specific for the M-phase of the cell cycle.

Figure 3

Proliferation of tumor and endothelial…

Figure 3

Proliferation of tumor and endothelial cells is due to reciprocal stimulation. (A) Proliferation…

Figure 3

Proliferation of tumor and endothelial cells is due to reciprocal stimulation. (A) Proliferation of HUVECs and Hs578T monocultures and their 1:1 coculture as determined by measuring the increase in cell number by direct cell counting after 36 hours. (B) Box-and-whisker diagrams of relative cell numbers of MCF-7, Hs578T, and MDA-MB-231 after incubation with conditioned medium from HUVECs compared with a normalized negative control of the same cells incubated with autologous medium as measured by the colorimetric cell proliferation assay with the WST-1 compound. (C) Proliferation of HUVECs that was induced by conditioned medium from Hs578T and MCF-7 cells as measured by WST-1. Relative absorbance values of colorful formazan, which has been converted by HUVECs, correspond to relative cell numbers. A single column represents average absorbance values for a minimum of eight independent replicates. y Axis error bars correspond to SD. The HUVECs that were treated with the Hs578T supernatant grew significantly faster than the same cells that were treated with the MCF-7 supernatant, the fresh vehicle medium, or the autologous (HUVEC-derived) medium.

Figure 4

Genes associated with the tumor-endothelial…

Figure 4

Genes associated with the tumor-endothelial induced M-phase cell cycle gene signature. (A) Significance…

Figure 4

Genes associated with the tumor-endothelial induced M-phase cell cycle gene signature. (A) Significance analysis of microarray data to identify genes that show the largest expression differences between tumor cells that were inducing a proliferation response in coculture with the cells and those that were not inducing proliferation. The expression levels of the top 63 genes are shown on a heat map after unidimensional hierarchical clustering of the genes. (B) Relative expression of CD44 and CD24 in diverse breast cancer cell lines. The average expression of their mRNA over the cell lines corresponds to 0. (C) Working model of reciprocal tumor-endothelial signaling based on the higher expression in the cocultures that induced proliferation.

Figure 5

Stimulatory effect of Hs578T conditioned…

Figure 5

Stimulatory effect of Hs578T conditioned medium on endothelial cells can be partially blocked…

Figure 5

Stimulatory effect of Hs578T conditioned medium on endothelial cells can be partially blocked by bevacizumab. (A) Relative expression of the “tumor-endothelial cell-induced M-phase cell cycle” genes in different monocultures of breast cancer cell lines and endothelial cells and in their respective cocultures. (B) Blocking of the stimulatory effect of Hs578T conditioned medium by bevacizumab. Absolute absorbance values of formazan dye that was converted by HUVECs, which corresponds to cell numbers, are shown in columns, with the y axis bars corresponding to SD. Bevacizumab (100 ng/ml) depleted the stimulatory effect of Hs578T cell culture supernatant in a significant manner, whereas it had no effect on the MCF-7 cell culture supernatant. Recombinant VEGF-A (5 ng/ml) and 5% FBS served as positive and negative controls, respectively. (C) Dose-dependent blocking of HUVEC proliferation by bevacizumab and trastuzumab. Absolute absorbance values of formazan dye that was converted by HUVECs are shown. HUVECs that were treated with the Hs578T cell culture supernatant represent the baseline stimulatory effect. Bevacizumab depleted the stimulatory effect of the Hs578T conditioned medium in a significant, dose-dependent manner. Trastuzumab, which is a monoclonal antibody against HER2, did not influence HUVEC stimulation by the Hs578T conditioned medium.

Figure 6

“Tumor-endothelial cell-induced M-phase cell cycle”…

Figure 6

“Tumor-endothelial cell-induced M-phase cell cycle” genes in early-stage breast cancer. (A) The expression…

Figure 6

“Tumor-endothelial cell-induced M-phase cell cycle” genes in early-stage breast cancer. (A) The expression values of genes in the “tumor-endothelial cell-induced M-phase cell cycle” gene signature were extracted from a published expression study of 295 early-stage breast cancers from the Netherlands Cancer Institute [35]. Genes and samples are organized by hierarchical clustering. The tumors were segregated into two groups that were defined by high (red) or low (blue) expression levels of the 30 genes matching the M-phase cell cycle gene cluster. (B) Correlation of the “tumor-endothelial cell-induced M-phase cell cycle” gene signature status with distant metastasis-free and overall survival. Kaplan-Meier curves for the clinical outcomes of the indicated tumors that exhibit high (red curve) and low (blue curve) “tumor-endothelial cell-induced M-phase cell cycle” gene signature expression are shown.

Figure 7

Correlation to other prognostic gene…

Figure 7

Correlation to other prognostic gene signatures in early-stage breast cancer. Correlation of the…

Figure 7

Correlation to other prognostic gene signatures in early-stage breast cancer. Correlation of the 70-gene signature [50], the wound signature [49], the invasiveness gene signature [40], and the “tumor-endothelial cell-induced M-phase cell cycle” signature score in the NKI data set. Pairwise scatter plot matrix of the three gene signatures. Pearson correlations are shown in the lower part of each plot.