A gastrin precursor, gastrin-gly, upregulates VEGF expression in colonic epithelial cells through an HIF-1-independent mechanism

Abstract

One of the major angiogenic factor released by tumor cells is VEGF. Its high expression is correlated with poor prognosis in colorectal tumors. In colon cancer, gastrin gene expression is also upregulated. In these tumors, gastrin precursors are mainly produced and act as growth factors. Recently, a study has also shown that the gastrin precursor, G-gly induced in vitro tubules formation by vascular endothelial cells suggesting a potential proangiogenic role. Here, we demonstrate that stimulation of human colorectal cancer cell lines with G-gly increases the expression of the proangiogenic factor VEGF at the mRNA and protein levels. In addition, blocking the progastrin autocrine loop leads to a downregulation of VEGF. Although HIF-1 is a major transcriptional activator for VEGF our results suggest an alternative mechanism for VEGF regulation in normoxic conditions, independent of HIF-1 that involves the PI3K/AKT pathway. Indeed we show that G-gly does not lead to HIF-1 accumulation in colon cancer cells. Moreover, we found that G-gly activates the PI3K/AKT pathway and inhibition of this pathway reverses the effects of G-gly observed on VEGF mRNA and protein levels. In correlation with these results, we observed in vivo, on colon tissue sections from transgenic mice overexpressing G-gly, an increase in VEGF expression in absence of HIF-1 accumulation. In conclusion, our study demonstrates that gastrin precursors, known to promote colon epithelial cells proliferation and survival can also contribute to the angiogenesis process by stimulating the expression of the proangiogenic factor VEGF via the PI3K pathway and independently of hypoxia conditions.

Figure 1

G-gly regulates VEGF expression in human colon cancer cells. DLD1 and HT29 cells were treated or not with 1 nM of G-gly, 250 μM CoCl₂ or exposed to hypoxia (1%) for 6 hr as indicated. (a) Total RNA was isolated and VEGF mRNA expression was determined by real time PCR as described in methods. Quantifications of 3 experiments are presented as means ± SE. (b) Cells were grown for 24 hr on 12-well plates containing cover slides. Following treatments, cells were then similarly fixed and stained with anti-VEGF antibodies using standard immunofluorescence techniques as described in “Methods.” Representative micrographs from 3 independent experiments are shown. (c) Expression of VEGF protein was also examined by Western analysis following treatment of the cells with G-gly or CoCl₂. Blots were also probed with an antibody against tubuline to ensure equal loading of proteins. (d) VEGF expression in cell supernates was also measured by ELISA. Results were normalized by cell counting. Representative data from 3 experiments are shown. Quantifications of 3 experiments are presented as means ± SE. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 2

Immuno-staining of VEGF and blood vessels number in colonic mucosa of FVB/N and MTI/G-Gly transgenic mice. (a) Sections of paraffin-embedded colonic mucosa were immunostained with VEGF antibodies using standard immunofluorescence techniques as described in “Methods.” Micrographs from representative fields of the stained sections were taken (original magnification ×40). Immunofluorescent intensity was analyzed in at least 4 independent experiments using the image analyzer ImageJ. Results of immunofluorescence quantification are expressed as percentages of the control values (FVB/N). (b) Blood vessels were visualized by immunofluorescent staining as described in “Methods,” using a CD31 specific antibody. Immunostained blood vessels were counted using the microscope software Axio Vision from Zeiss. The total sections were screened and the results were normalized to tissue area. Quantifications of 3 experiments are presented as means ± SE. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 3

G-gly does not increase HIF-1α protein levels. DLD1 and HT29 cells were treated or not (control) with 1 nM of G-gly or exposed to hypoxia (1%) for the time indicated. (a) Cells were grown for 24 hr on 12-well plates containing cover slides. Following treatments, cells were then similarly fixed and stained with anti-HIF1α antibodies using standard immunofluorescence techniques as described in “Methods.” Representative micrographs from 3 independent experiments are shown. (b) Expression of HIF1α protein was also examined by Western analysis. Blots were also probed with an antibody against tubuline to ensure equal loading of proteins. Representative data from 3 experiments are shown. (c) Immunohistochemistry was performed on sections of paraffin-embedded colonic mucosa with anti-HIF1α antibodies. Micrographs from representative fields were taken (original magnification ×40). Results are representative of 3 experiments. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 4

G-gly induces AKT activation in human colon cancer cells. DLD1 and HT29 cells were treated or not with 1 nM of G-gly for the time indicated. AKT activation was determined on cell lysates by immunoblot using antibodies directed against the activated phosphorylated forms or AKT. Blots were also probed with tubuline to ensure equal loading of protein. Representative data from 3 experiments are shown. Quantifications of 3 experiments are presented as means ± SE.

Figure 5

Role of the PI3-kinase/AKT pathway in VEGF expression stimulated by G-gly. Cells were pretreated for 30 min. with a specific PI3-kinase inhibitor (LY294002, 20 μM) prior to G-gly stimulation. (a) Total RNA was isolated and VEGF mRNA expression was determined by real time PCR as described in methods. Quantifications of 3 experiments are presented as means ± SE. (b) Cells were grown for 24 hr on 12-well plates containing cover slides. Following treatments, cells were then similarly fixed and stained with anti-VEGF antibodies using standard immunofluorescence techniques as described in “Methods.” Representative micrographs from 3 independent experiments are shown. Immunofluorescent intensity was analyzed in at least 3 independent experiments using the image analyzer ImageQuant Quantifications of 3 experiments are presented as means ± SE. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Figure 6

Blocking autocrine gastrins decreases VEGF production in the human colon cancer cell line Lovo. (a) Gastrin mRNA expression from different human colon cancer cell lines (HT29, DLD1, Lovo) or normal epithelium was determined by real time PCR as described in methods. (b) Lovo cells were transiently transfected with Silencer Negative control siRNA or Gastrin silencer siRNA as described in “Methods.” Gastrin mRNA expression was controlled 48, 72 and 96 hr after transfection using real time PCR. (c–e) After 72 hr of transfection with Silencer Negative control siRNA or Gastrin silencer siRNA, proliferation rates in Lovo cells (c) were determined by MTT assay, AKT phosphorylation (d) and VEGF protein expression (e) were determined on cell lysates by immunoblot. Blots were also probed with GAPDH or tubuline (as indicated) to ensure equal loading of protein. Representative data from 3 experiments are shown. (a–e) Quantifications of 3 experiments are presented as means ± SE.

Figure 7

Blocking autocrine gastrins decreases VEGF production in the human colon cancer cell line HT29. HT29 cells were stably transfected with a shRNA directed against the gastrin gene or a scrambled control. Stably transfected cell pools were used for the experiments. (a,b) Gastrin or VEGF mRNA expression was measured using real time PCR. (c) VEGF expression in cell supernates was measured by ELISA. Results were normalized by cells counting. (d) Proliferation rates of HT29 in absence or presence of VEGF (10 ng ml⁻¹) were determined by cell counting. Representative data from 3 to 4 experiments are shown. Quantifications of 3 to 4 experiments are presented as means ± SE.