Roles of a prostaglandin E-type receptor, EP3, in upregulation of matrix metalloproteinase-9 and vascular endothelial growth factor during enhancement of tumor metastasis

Abstract

Cyclooxygenase (COX)-2 is known to correlate with poor cancer prognosis and to contribute to tumor metastasis. However, the precise mechanism of this phenomenon remains unknown. We have previously reported that host stromal prostaglandin E(2) (PGE(2))-prostaglandin E2 receptor (EP)3 signaling appears critical for tumor-associated angiogenesis and tumor growth. Here we tested whether the EP3 receptor has a critical role in tumor metastasis. Lewis lung carcinoma (LLC) cells were intravenously injected into WT mice and mice treated with the COX-2 inhibitor NS-398. The nonselective COX inhibitor aspirin reduced lung metastasis, but the COX-1 inhibitor SC560 did not. The expression of matrix metalloproteinases (MMP)-9 and vascular endothelial growth factor (VEGF)-A was suppressed in NS-398-treated mice compared with PBS-treated mice. Lungs containing LLC colonies were markedly reduced in EP3 receptor knockout (EP3(-/-)) mice compared with WT mice. The expression of MMP-9 and VEGF-A was downregulated in metastatic lungs of EP3(-/-) mice. An immunohistochemical study revealed that MMP-9-expressing endothelial cells were markedly reduced in EP3(-/-) mice compared with WT mice. When HUVEC were treated with agonists for EP1, EP2, EP3, or EP4, only the EP3 agonist enhanced MMP-9 expression. These results suggested that EP3 receptor signaling on endothelial cells is essential for the MMP-9 upregulation that enhances tumor metastasis and angiogenesis. An EP3 receptor antagonist may be useful to protect against tumor metastasis.

Figure 1

Effect of COX‐2 inhibitors on tumor metastasis. A suspension of Lewis lung carcinoma cells that were syngeneic for C57Bl/6 was injected into the tail vein of wild‐type mice. COX‐2 inhibitors (NS‐398, 3 mg/mL; aspirin, 10 mg/mL) were administered orally in the suspension (0.1 mL/10 g of body mass) twice a day (every 12 h) beginning on the day of cell injection and continuing throughout the 28‐day experimental period. Data are means ± SD for the number of mice (all n = 6/group). *P < 0.05 versus vehicle‐treated mice (ANOVA).

Figure 2

Reduced pro‐MMP‐9 and vascular endothelial growth factor (VEGF)‐A expression in COX‐2 inhibitor‐treated mice. (a) Pro‐MMP‐9 levels and (b) VEGF‐A in lung tissue 14 days after injection of Lewis lung carcinoma cells determined by ELISA. Data are means ± SD for the number of animals (all n = 5/group). *P < 0.05 versus wild‐type mice (Student’s t‐test).

Figure 3

Reduced Lewis lung carcinoma colony formation in prostaglandin E2 receptor (EP)3^−/− mice. (a) The number of colonies in the lungs 28 days after intravenous injection of Lewis lung carcinoma cells. EP3^+/+, wild‐type mice; EP3^−/−, EP3 receptor knockout mice. Data are means ± SD for the number of mice (all n = 10/group). P < 0.05 versus wild‐type mice (Student’s t‐test). (b) Typical hematoxylin–eosin staining of metastatic lung tumors from EP3^+/+ and EP3^−/− mice 28 days after injection of Lewis lung carcinoma cells.

Figure 4

Reduced expression of MMP‐9 in tumor metastasis lesions in EP3^−/− mice. (a) The expression of MMP‐9 in HUVEC 24 h after treatment with selective agonists (final concentration, 0.1 nM) selective for EP1, EP2, EP3, and EP4. (b) Pro‐MMP‐9 levels in lung tissue 14 days after injection of Lewis lung carcinoma cells determined by ELISA. Data are means ± SD for the number of animals (all n = 5/group). *P < 0.05 versus wild‐type mice (Student’s t‐test). (c) Determination of MMP‐9 and MMP‐2 mRNA levels in lung tissues by real‐time PCR. Lung tissue was isolated 7, 14 days after injection of Lewis lung carcinoma. Real‐time PCR was carried out. Results were compared with WT and EP3^−/− mice. Data are means ± SD for the number of animals (7–14 animals). *P < 0.05 versus WT mice (Student’s t‐test). (d) Immunohistochemical localization of CD31 and MMP‐9 in EP3^+/+. Blue arrows indicate CD31‐ and MMP‐9‐positive endothelial cells. T, tumor. (e) Immunohistochemical localization of MMP‐9 in EP3^+/+ and EP3^−/−. Yellow arrows indicate endothelial cells and blue arrows indicated macrophage‐like interstitial cells.

Figure 5

Reduced expression of VEGF‐A in tumor metastasis lesions in EP3^−/− mice. (a) VEGF levels in the lungs homogenized with PBS. The levels were determined by ELISA. Data are means ± SD for the number of animals (all n = 5/group). *P < 0.05 versus wild‐type mice (Student’s t‐test). (b) The expression of VEGF‐A and CD31 mRNA in lung tissues. Lung tissue was isolated 2 weeks after injection of Lewis lung carcinoma. (c) Immunohistochemical localization of VEGF‐A and CD31 in EP3^+/+ and EP3^−/−. Blue arrows indicate VEGF‐positive endothelial cells near the metastatic locus, black circles indicate VEGF‐A‐positive stromal cells around tumor cells, and red arrows indicate CD31‐positive cells. T, tumor. (d) Immunohistochemical localization of CD31 and MMP‐9 in EP3^+/+. The localization of CD31‐positive cells (red arrows in the left panel) is coincident with that of MMP‐9‐positive cells in the serial sections (red arrows in the right panel). St, stroma; T, tumor.

Figure 6

Roles of EP3 signaling in enhancement of MMP‐9 and VEGF‐A expression in lungs containing Lewis lung carcinoma colonies in mice. The expression of MMP‐9 in the endothelial cells but not in tumor stromal cells around the metastatic lesions was dependent on EP3 signaling, whereas that of VEGF‐A both in the endothelial cells and in the stromal cells was dependent on EP3 signaling. The difference in metastatic colony formation between EP3^−/− and EP3^+/+ may be explained by EP3 dependency in these factors. PGE₂, prostaglandin E₂.