In vitro induction of COX-2 (A and B) and VEGF (C and D) expression in GCs by EDN2 and hCG. Cells were incubated with basal media (Cont) or EDN2 (10 Ϫ 7 M ), hCG (10 U/ml), or both for 24 h. Cells 

Context in source publication

Context 1

... essential role of EDN1 in the demise of CL in several species (20). However, whether other members of the EDN family are present and may be involved in the function of the bovine CL is not yet known. Hypoxia was shown to be a strong trigger of EDN2 expression in several cancer cells and recently in GCs as well (21–25). Hypoxic conditions prevail in CL at early stages of its development before sufficient blood supply is achieved. It is therefore plausible that EDN2 is expressed and that it affects CL function. Two recent publications showed that EDN2 is transiently expressed in GC immediately before ovulation in pregnant mare serum gonadotropin and human chorionic gonadotropin (hCG)-induced rat and mouse models (26, 27). Hence, we examined here the pattern and regulation of EDN2 expression and its putative roles in bovine ovaries. The present study reveals that EDN2 is expressed for a short period of time, during CL formation, by the luteinizing GC. The in vitro data suggest that LH and hypoxia are the physiological triggers of EDN2 induction in the early CL. Finally, we investigated the biological actions of EDN2, demonstrating that it induced in GCs the characteristics of early CL: GC proliferation as well as elevated vascular endothelial growth factor (VEGF) and cyclooxygenase (COX)-2 expressions. Initially we determined the in vivo expression pattern of EDN2 during the folliculoluteal transition as follows: heifers bearing preovulatory follicles were exposed to GnRH-induced gonadotropin surge, and EDN2 mRNA levels were determined in whole follicles collected before or 4, 10, 20, 25, and 60 h after administering GnRH (Fig. 1A). This model had already been validated by monitoring other genes (37, 38), which showed the expected temporal profile (34, 37, 39). Steady-state EDN2 mRNA levels (Fig. 1B) were low before GnRH, with only a transient increase 4 h later, which did not reach statistical significance. However, a marked, significant, induction was observed 60 h after GnRH (early CL), reaching a value about 30-fold higher than that found before ovulation. Because this last experiment contained only very early CL samples (60 h after GnRH and ϳ 30 h after ovulation), EDN2 mRNA levels were also determined in CL (of noninduced cycles) collected at early luteal phase (before d 5 of the cycle) and at midluteal stage (Fig. 1C). In accordance with data from induced cycles, EDN2 mRNA was high in CLs collected on d 1–5 of the cycle; in fact, its levels in these samples were of the same order of magnitude as those found 60 h after GnRH (Fig. 1, B and C). EDN2 mRNA was undetectable at a later stage of the cycle (midluteal phase, Fig. 1C). To determine which cell population in the CL expresses EDN2 , early stage CL were enzymatically dispersed and separated by lectin-coated magnetic beads to generate enriched endothelial and steroidogenic luteal cell fractions. The results shown in Fig. 1D indicate that EDN2 mRNA was more abundant in the luteal steroidogenic cells than in the endothelial cells. In cells enriched from early CL, there was a 10 –15% contamination of luteal cells within the endothelial cell fraction. Therefore, the residual amount of EDN2 mRNA found in endothelial cells could be the result of contamination by luteal cells. As expected, endothelial cells had higher levels of EDN1 , indicating a different cell source for these two EDN peptides (Fig. 1D). Similarly, in preovulatory follicles, EDN2 was expressed in the GC layer, whereas in the vascular theca interna layer, the mRNA levels were 100-fold less (data not shown). Incubation of GC derived from large healthy follicles with bLH (10 and 100 ng/ml) for 42 h significantly stimulated EDN2 mRNA levels, with 10 ng/ml already pro- ducing a maximal response (Fig. 2). Interestingly, this stimulation could not be observed in cells incubated for shorter periods of time (3 and 24 h; data not shown). Progesterone on its own or in combination with LH did not affect EDN2 mRNA levels (Fig. 2). The effect of hypoxia on EDN2 expression was studied in two cell types: bovine GC and the human granulose cell line (SVOG; Fig. 3). Hypoxia was induced by either the hypoxia mimetic agent CoCl 2 or low oxygen tension. CoCl 2 elevated EDN2 mRNA levels in bovine GC in a dose-dependent manner (Fig. 3A). In SVOG cells, low oxygen tension also induced EDN2 mRNA as compared with cells incubated under normoxic conditions (Fig. 3B). In these experimental models, hypoxia, either in the form of the mimetic agent or oxygen pressure, induced VEGF expression in parallel to EDN2 , as was expected (Fig. 3, C and D). The reciprocal experiments, i.e. CoCl 2 in SVOG and in a low-oxygen environment in bovine GC culture, yielded similar results (data not shown). The effects of exogenously added EDN2 on GC functions were studied (Figs. 4 and 5): 24 h after this peptide (10 Ϫ 7 M ) was added to the culture medium, COX-2 and VEGF expression (mRNA and protein levels) were significantly enhanced (Fig. 4). hCG, known to induce these two genes, served as a positive control in these experiments. Combining both agents, EDN2 and hCG, tended to further increase the expression of these two genes, although this increase was not statistically significant. In another experiment, varying concentrations of EDN2 (10 Ϫ 8 to 10 Ϫ 6 M ) dose dependently elevated the number of viable GCs after 4 d in culture (Fig. 5). The highest OD readings in the XTT assay (an indication of viable cell numbers) was observed with a dose of 10 Ϫ 7 M . To manipulate endogenous EDN2 levels, we used two siRNA molecules targeting ECE-1: si-bECE-1 and si- hECE-1. Preliminary experiments were conducted to exam- ine siRNA transfection efficiency in bGCs. For that purpose, cells were transfected with different concentrations of FAM- labeled siRNA and labeled cells were visualized under a mi- croscope. We found that use of 25–50 nmol/liter of fluores- cent siRNA labeled about 75– 85% of the cells. Data presented in Fig. 6 show that both siRNA molecules markedly silenced (up to 80% inhibition) ECE-1 mRNA 48 h after transfection, as compared with cells transfected with noncoding siRNA sequence (Fig. 6A). Effective silencing persisted for at least 96 h after transfection (Fig. 6A), compared with negative controls for each time point. There was no statistical difference in the results obtained by the two scrambled siRNA sequences, and they were therefore combined. The average values (arbitrary units) of ECE-1 and VEGF mRNA in the negative controls were 38.4 Ϯ 5.5 and 7.3 Ϯ 2.6, respectively. Importantly, 96 h after introducing ECE-1 siRNA molecules into the cells, about 50% reduction in VEGF mRNA levels was also observed (Fig. 6B). However, this treatment did not affect steroidogenic acute regulatory protein mRNA (data not shown). Data presented in Fig. 6C indicate that each siRNA targeting ECE-1 could significantly inhibit viable GC numbers. The effects of ECE-1 siRNA on VEGF and GC cell numbers were reversed by adding back EDN2 to the culture medium (Fig. 6, B and C). In the present study, we described the expression pattern and potential roles of EDN2 in the bovine ovary. EDN2 is made by the young CL, most likely by luteinizing GCs and exogenous EDN2 or endogenously ablated peptide-modulated GC functions. The findings therefore portray EDN2 as a novel autocrine factor in CL development. EDN2 mRNA displayed a transient pattern of expression during the luteal phase; highest EDN2 mRNA levels were observed in early CL (60 h after GnRH and ϳ 30 h after ovulation) in the induced ovulation and in naturally occurring CL collected before d 5 of the cycle. EDN2 mRNA then declined to basal levels in midluteal phase. Notably, unlike EDN2, EDN1 was not elevated at any time point during folliculoluteal transition induced by GnRH (40). The transition of a preovulatory follicle into a CL is a complex process involving mechanisms similar to wound healing and tumor formation. Robust angiogenesis takes place and various cell types undergo proliferation and differentiation (39, 41). LH and hypoxia are two important stimulants of these processes. The ovulatory surge of LH triggers extensive structural and molecular changes in the preovulatory follicle leading to its ovulation and CL formation (39, 41). At this stage, because of insufficient blood supply to this fast-growing tissue, hypoxic conditions exist in the CL (42). The data presented here show that LH and hypoxia (either as hypoxia mimicking reagent, CoCl 2 , or low oxygen tension) each induced EDN2 mRNA, suggesting that these may indeed be the physiological stimuli of EDN2 expression in the luteinizing gland. Hypoxia-induced up-regulation of EDN2 mRNA was also observed in the human GC line (SVOG). It is note- worthy that these same factors (LH and hypoxia) are known to induce VEGF in the CL (42, 43). VEGF, an important growth and permeability factor for endothelial cells, is highly expressed in the developing CL (42, 44 – 46). In the ovary, unlike in many other tissues, LH/hCG was shown to be a major stimulant of VEGF expression (43, 47). Hypoxia is another strong inducer of VEGF in the CL (42) as well as many tissues, most notably in growing cancers (48, 49). Hypoxia-induced up-regulation occurs at the transcriptional level by activating hypoxia-inducible factor (HIF)-1 ␣ , which binds to the hypoxia response element region in the VEGF promoter (48, 49). Similar to VEGF, EDN2 was found to be one of the seven most hypoxia-inducible genes in several tumor cells (50). For instance, in breast cancer cells, EDN2 mRNA was induced by cobalt and inhibited by diphenylene iodonium (an inhibitor of the enzyme nicotinamide adenine dinucleotide phosphate oxidase) (24). The kinetics of this process suggested that up-regulation of EDN2 mRNA during hypoxia was HIF-1 dependent (24). MatInspect (Geno- matics, Munich, Germany) and MATCH (TRANSFAC) identified several putative HIF-1 ␣ response elements (in po- sitions ...