Transcriptional network governing the angiogenic switch in human pancreatic cancer

Abstract

A shift of the angiogenic balance to the proangiogenic state, termed the "angiogenic switch," is a hallmark of cancer progression. Here we devise a strategy for identifying genetic participants of the angiogenic switch based on inverse regulation of genes in human endothelial cells in response to key endogenous pro- and antiangiogenic proteins. This approach reveals a global network pattern for vascular homeostasis connecting known angiogenesis-related genes with previously unknown signaling components. We also demonstrate that the angiogenic switch is governed by simultaneous regulations of multiple genes organized as transcriptional circuitries. In pancreatic cancer patients, we validate the transcriptome-derived switch of the identified "angiogenic network:" The angiogenic state in chronic pancreatitis specimens is intermediate between the normal (angiogenesis off) and neoplastic (angiogenesis on) condition, suggesting that aberrant proangiogenic environment contributes to the increased cancer risk in patients with chronic pancreatitis. In knockout experiments in mice, we show that the targeted removal of a hub node (peroxisome proliferative-activated receptor delta) of the angiogenic network markedly impairs angiogenesis and tumor growth. Further, in tumor patients, we show that peroxisome proliferative-activated receptor delta expression levels are correlated with advanced pathological tumor stage, increased risk for tumor recurrence, and distant metastasis. Our results therefore also may contribute to the rational design of antiangiogenic cancer agents; whereas "narrow" targeted cancer drugs may fail to shift the robust angiogenic regulatory network toward antiangiogenesis, the network may be more vulnerable to multiple or broad-spectrum inhibitors or to the targeted removal of the identified angiogenic "hub" nodes.

Fig. 1.

Genetic participants of the angiogenic switch. (A) We hypothesized that global analysis of transcriptional perturbation induced by positive and negative regulators of angiogenic balance may provide a rational algorithm to define the genetic participants of the angiogenic switch. Human microvascular endothelial cells were treated for 4 h with the endogenous angiogenesis inhibitor endostatin (200 ng/ml) to mimic the shift of the angiogenic balance toward an antiangiogenic (Off) state. Conversely, the proangiogenic (On) state in endothelium was emulated by using proangiogenic stimulators VEGF (10 ng/ml), bFGF (20 ng/ml), or combined VEGF (10 ng/ml) and bFGF (20 ng/ml). The inverse-regulation pattern of genes after pro- and antiangiogenic treatment was used as the selection criteria to predict the genes involvement in the angiogenic process. (B) Using significance analysis of microarrays, 2,370 transcripts with significant inverse-expression patterns were selected (P < 0.01). From these 2,370 transcripts, 1,140 were down-regulated after endostatin treatment and up-regulated after VEGF/bFGF treatment (categorized to participate in proangiogenic signaling, example genes in green box), whereas the remaining 1,230 transcripts were oppositely regulated (categorized to participate in antiangiogenic signaling, red box; see also SI Text). Each row represents log² expression ratios of an individual gene (see color code) and the columns indicate each respective treatment (in quadruplicates, 1–4). (C) Real-time quantitative RT-PCR confirmation of inverse regulation of six selected example genes in endothelial cells after endostatin vs. VEGF/bFGF treatment. Bars are means ± SD from three independent measurements and show relative expression levels compared with untreated control (P < 0.01).

Fig. 2.

Angiogenic signaling network. A gene regulatory network constructed from inversely regulated proangiogenic genes. All presented genes are down-regulated after endostatin although up-regulated after VEGF/bFGF treatment (except APC gene; arrow demonstrates opposite regulation). The direction of gene regulation and the high degree of cooperative networking between the selected genes point to a switchable angiogenic network. The concerted up-regulation of the network genes indicates the proangiogenic state (On). Highlighted are gene interactions based on promotor-binding site (green connection lines), protein modification (yellow connection lines), protein–protein binding (violet connection lines), gene expression (blue connection lines), and gene regulation (black connection lines). Two signaling pathways, STAT3 (yellow circles) and PPARδ/β-catenin (red shadows), are highlighted and demonstrate the interconnectedness of the pathways within the angiogenic network.

Fig. 3.

Angiogenic switch in human PC. (A) In vivo expression profiles from pancreatic tissue in correlation to the in vitro data on the angiogenic switch after endostatin (E), VEGF (V), bFGF (B), and VEGF+bFGF (V+B). Included in the analysis are nine samples from human NP, nine samples from patients with CP, nine samples from patients with PC, and nine samples from patients with MP. In accordance with in vitro data, the predicted proangiogenic genes are increasingly up-regulated from normal (angiogenesis Off) to chronic inflammation to cancer (angiogenesis On). This provides genetic evidence for the angiogenic switch in human tumors. Expression ratios are colored according to the scale bar: Blue, >2-fold down-regulation; red, >2-fold up-regulation. (B) The expression levels of PPARδ protein are confirmed in human pancreatic specimens by using immunohistochemistry on high-density tissue microarrays. PPARδ staining intensity is enhanced in CP, PC, and MP compared with NP. The up-regulation of PPARδ is not restricted to, but is actually more enhanced in the tumor vasculature and in the tumor stroma. The black arrows point to the endothelial cells. (Top) ×200, (Middle) ×400; Bottom highlights the endothelial cells magnified from the square box within the ×400 view.

Fig. 4.

Critical involvement of PPARδ in angiogenic process. PPARδ silencing in the tumor microenvironment inhibits tumor growth (A) and reduces tumor microvascular density (B) in LLC and B16 melanoma, two syngeneic tumors growing s.c. in WT (wt) and PPARδ (−/−) mice. A shows tumor volumes assessed 20 (LLC) and 29 (B16) days after s.c. tumor injection. Bars are mean ± SD (*, P < 0.01 in −/− vs. WT). (B) Reduced vascular density in −/− mice is demonstrated in LLC tumors (view, ×200) by CD31 (green, Alexa488, tumor vascular endothelium) and nuclear (blue, DAPI) costaining.