Introduction

Epithelial ovarian cancer, the most common form of ovarian malignancy, is usually diagnosed at a late stage and this results in a poor prognosis 1. The lack of preventive strategies, early diagnostic tools and effective therapies to treat the recurrent epithelial ovarian tumors prompts an urgent need to a better understanding of its pathogenesis at the molecular level 2, 3.

PGC-1α is a multiple-function transcription coactivator that regulates the activity of multiple nuclear receptors and transcription factors involved in the mitochondrial biogenesis 4, 5. PGC-1α is widely expressed in brown fat, skeletal muscle, liver, heart, kidney and brain 6, 7. It plays important roles in the regulation of adaptive thermogenesis in brown fat and muscle 8, muscle fiber-type switching in muscle 9, gluconeogenesis in liver 10 and insulin secretion in islets 11. More recently, it has been found that the expression level of PGC-1α decreased in certain cancer tissues such as breast 12, 13 and colon cancers 14, suggesting that PGC-1α may be involved in the pathogenesis of cancer. In addition, the decrease of PGC-1α is always linked to the variation of peroxisome proliferator-activated receptor gamma (PPARγ) expression level 12, 13, 14. PPARγ, a member of the nuclear receptor superfamily, forms a complex with coactivator PGCs to regulate gene expression 8, 15, 16. PPARγ has a tumorigenic role in many kinds of cancers such as colon cancer 17, 18 and mammary gland cancer 19, and activation of PPARγ signaling exacerbates the development of these tumors. On the other hand, PPARγ ligands have antiproliferative activity and induce apoptosis in many cancers 20, 21, 22. Given that activation of PPARγ by its ligands can induce apoptosis in cancer and that PGC-1α also powerfully coactivates PPARγ, we speculate that there may be a potential relationship between PGC-1α and PPARγ in cancer development. A recent study has shown that the expression of PPARγ is indeed enhanced in epithelial ovarian carcinoma 23; however, the role of PGC-1α in the pathogenesis of epithelial ovarian carcinoma remains to be addressed.

In the present study, we showed that PGC-1α expression in the surface epithelium of epithelial ovarian tumors was downregulated compared with normal ovaries. Overexpressing PGC-1α in human ovarian epithelial carcinoma cell line Ho-8910 significantly and specifically induced cell apoptosis through decreasing Bcl-2/Bax expression ratio and through the release of cytochrome c; PGC-1α exerted its effect by a PPARγ-dependent pathway. Our results reveal a new role of PGC-1α in the pathogenesis and treatment of epithelial ovarian cancer.

Materials and Methods

Human ovarian specimens

Tissue samples from 10 patients (age 46-71 years; mean age 58.2 years) with malignant epithelial ovarian tumors (four stages II and six stages III) were from the first Affiliated Hospital of Nanjing Medical University and Tissue Banking Facility of Tianjin Medical University Cancer Institute & Hospital and National Foundation for Cancer Research. All tissue samples were from patients of ovarian serous cystadenocarcinoma, which is the major type of epithelial tumors. The tumors were histopathologically diagnosed and classified assessed according to the International Federation of Gynecology and Obstetrics 24 criteria by two gynecology pathologists (WRW and DAB). Seven cases with normal ovaries from hysterectomy specimens resected for non-ovarian diseases were added and normal human ovarian surface epithelium (OSE) (age 36-55 years; mean age 45.6 years) was derived from the surface scrapings of these normal ovaries. All cases were selected based on availability of tissue and were not stratified for known preoperative or pathological prognostic factors. Following surgery, tissues were snap-frozen in liquid nitrogen and stored at −80 °C for subsequent RNA extraction.

Cell line and culture conditions

Human epithelial ovarian carcinoma cell Ho-8910 25, 26, 27 was obtained from the Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences. They were cultured in RPMI 1640 medium supplemented with 10% fetal calf serum (GBICO BRL, Rockville, MD, USA) at 37 °C in a humidified atmosphere containing 5% CO2. Cells in the logarithmic growth phase were used for experiments.

Adenovirus infection

Recombinant adenoviruses expressing PGC-1α GFP fusion protein and GFP alone were provided by Dr Dan Kelly (Washington University, USA). Purified virus stocks were prepared through CsCl step gradient centrifugation as described previously 28. Cells were grown to confluence, induced to differentiate for 48 h and then infected with adenovirus at a multiplicities of infection (moi) of 50. The infection efficiency was determined by GFP. Cells were harvested for isolation of total RNA or protein 48 h after infection.

Semiquantitative RT-PCR and quantitative PCR

Total RNA was isolated from Ho-8910 cells with 48 h Ad-GFP or Ad-PGC-1α infection by an RNeasy mi kit (Qiagen, Hilden, Germany). One microgram of RNA was reverse-transcribed into cDNA using oligo (dT)18 primers and AMV reverse transcriptase (Gibco invitrogen, Carlsbad, CA, USA). PCR products of PGC-1α and house-keeping gene glyceraldehydes-3-phosphate dehydrogenase (GAPDH) were amplified with specific primers (Table 1) using the following parameters: 94 °C for 30 s, 56 °C (PGC-1α) and 59 °C (GAPDH) for 30 s and 72 °C for 30 s, with a final extension step at 72 °C for 7 min. PCR products were electrophoresed through 1.0% agarose gel, stained with ethidium bromide and visualized under ultraviolet illumination. Band intensity was calculated densitometri-cally using Quantity One-4.4.1 imaging software (Bio-Rad Laboratories). Levels of mRNA were expressed as the ratio of band intensity for PGC-1α relative to that for GAPDH.

Table 1 PCR primers used in this study
Full size table

Quantitative PCR was performed with the ABI Prism 7000 sequence detection system (Applied Biosystem, Foster City, CA) using Tagman probes (Shanghai Shinegene Molecular Biotechnology Co., Ltd), and threshold cycle numbers were obtained using ABI Prism 7000 SDS software version 1.0. The primers and probes sequences used in this study are in Table 1. Conditions for amplification were one cycle of 95 °C for 5 min followed by 40 cycles of 95 °C for 30 s, 60 °C for 1 min and final one cycle of 72 °C for 2 min.

Hoechest 33258 staining

Ho-8910 cells were infected with Ad-GFP and Ad-PGC-1α for 48 h, then cells were collected, washed with phosphate-buffered saline (PBS) and fixed by methanol-acetic acid (3:1, v/v) at 25 °C for 10min. Following centrifugation, the samples were washed with PBS three times, stained with Hoechest 33258 staining solution (Beyotime Biotechnology) in the dark for 5 min and then seeded separately onto cover slides. Once the treated samples had been dried in the dark at 25 °C, stained nuclei were observed under microscope (Nikon E800).

FACS analysis

Adenovirus-infected Ho-8910 cells were subjected to single parameter analysis for apoptosis and DNA fragmentation. In brief, cells were washed with PBS and suspended in 100% ethanol at 4 °C overnight. Fixed cells were then centrifuged at 1000 r.p.m. and treated with hypotonic fluorochrome solution, containing 0.1% sodium citrate, 0.1% Triton-X-100 and 50 mg/ml propidium iodide (PI) according to the procedure of Nicoletti et al. 29. Treated nuclei were stained 30 min at 4 °C, while protected from light and washed with PBS three times. The PI fluorescence of individual nuclei was measured by a FACSort flow cytometer (Becton-Dickinson, San Jose, CA,USA). Stained cells traversed the 488 nm argon laser at an event rate of approximately 400 nuclei/s and it was maintained for all samples analyzed. A 585/42 nm band pass filter was used to collect the red fluorescence from PI-stained nuclei and the data recorded in the linear scale. A minimum of 10 000 events were collected for each analysis. The percentage of apoptotic nuclei (sub-diploid DNA peak in the DNA fluorescence histogram) was calculated with Cell Quest software (Becton-Dickinson, San Jose, CA, USA).

Electron microscopy

Adenovirus-infected Ho-8910 cells were detached from the plates with a cell lifter and centrifuged at 1 000 r.p.m. for 5 min. The cell pellet was first fixed with 2.5% glutaraldehyde in 0.1 M PBS (pH7.4) for 2 h at 4 °C and then post-fixed in 1% osmium tetroxide, 0.1 M phosphate buffer (pH 7.4). After dehydration in ethanol and embedded in an Epon-Araldite mixture, thin sections were obtained with a Reichert-Jung Ultracut-E ultramicrotome and stained with lead citrate. Samples were examined with a JEM-EX electron microscope (JEOL, Tokyo, Japan).

Western blot analysis

Cytosolic-specific, mitochondria-free lysates were prepared as described in Miyake et al. 30. After 48 h adenovirus infection, Ho-8910 cells were collected by centrifugation at 600 × g for 5 min. The cell pellet was lysed on ice in 300 μl of a chilled hypotonic lysis solution (220 mM mannitol, 68 mM sucrose, 50 mM PIPES-KOH (pH 7.4), 50 mM KCl, 5 mM EDTA, 2 mM MgCl2, 1 mM dith-iothreitol and a mixture of protease inhibitors consisting of 100 μM 4-(2-amoethyl)-benzenesulfonyl fluoride, 80 nM aprotinin, 5 μM bestatin, 1.5 μM E-64 protease inhibitor, 2 μM leupeptin and 1 μM pepstatin A). After 45 min incubation on ice, the cell lysates were centrifuged at 600 × g for 10 min. The supernatant was collected in a microcentrifuge tube and centrifuged at 14 000 r.p.m. for 30min. Protein concentration was determined by the BCA method (Pierce, Rockford, USA). For electrophoresis, equal amounts of proteins (60μg) were loaded onto 15% SDS-PAGE gels. After electrophoresis, protein bands were transferred to nitrocellulose membranes. The membranes were blocked with TBST containing 5% non-fat milk (blocking buffer) for 1 h and then incubated with anti-cytochrome c antibody (Santa Cruz Biotechnology, Santa Cruz, USA) diluted 1:1000 in blocking buffer at 4 °C overnight. After washing three times with TBST, the membranes were incubated with HRP-conjugated goat anti-mouse IgG antibody (Santa Cruz; diluted 1:2000) for 1 h at room temperature. The immunoblots were visualized by an enhanced chemiluminescence detection system (NEN, Boston, USA).

Microarray assay

Total RNA of adenovirus-infected Ho-8910 cells were isolated using Trizol reagent (Invitrogen) and analyzed by a human genome 70-mer oligonucleotide microarray from CapitalBio Corporation (Beijing, China). Briefly, a Human Genome Oligo Set Version 2.0 consisting of 5′-amo-modified 70-mer probes and representing 21329 well-characterized Homo sapiens genes was purchased from Qiagen and printed on amo-silaned glass slides. A space- and intensity-dependent normalization based on a LOWESS program 31 was employed. Three independent experiments were performed, and for each test and control sample, two hybridizations were performed by using a reversal fluorescent strategy. Only genes whose alteration tendency kept consistency in both microarray assays were selected as differentially expressed genes.

siRNA interference

Three siRNA sequences targeting different sites of human PPARγ cDNA were designed and synthesized by Genesil (Wuhan, PR China). Control sequence that could not target PPARγ cDNA was also included as a negative control. siRNA was transfected into Ho-8910 cells using LipofectAmine reagent (Gibco-Invitrogen) according to the manufacturer's manual. At 48 h after medium replace-ment, cells were harvested for total RNA isolation, and PPARγ expression level was analyzed by quantitative PCR. The primer sequences used in the experiments are detailed in Table 1. The sequence with the best interfering effect (named S1) was selected. siRNA sequences were as follows: S1, AAGATGACAGACCTCAGGCAG; S2, AAACCCTTTACCACGGTTGAT; S3, AACTTCGG-AATCAGCTCTGTG; N (the negative control sequence), ACTACCGTTGTTATAGGTG.

Statistical analysis

All experiments were performed in triplicate and the results are expressed as means ± S.E.M. Statistical analyses were performed using SPSS/Windows version 12.5S statistical package (SPSS, Chicago, IL, USA). Statistical significance was accepted at the level of p < 0.05.

Results

Expression of PGC-1α is downregulated in epithelial ovarian cancer

To investigate PGC-1α expression in normal human ovaries and human ovarian epithelial tumors, quantitative RT-PCR was used to assess the levels of PGC-1α mRNA in normal ovaries (n = 7) and tumor tissues (n = 10), and the results were normalized against the GAPDH mRNA. Results showed that the PGC-1α mRNA level in surface epithelium decreased to 41 ± 8% in ovarian tumors as compared with normal tissues (Figure 1).

Figure 1
figure 1

PGC-1α mRNA levels in normal human ovaries and ovarian tumors. PGC-1α expression levels in the OSE were determined by quantitative RT-PCR and quantified as the ratio of PGC-1α/GAPDH. Data of normal ovaries (n = 7) and ovarian tumors (n = 10) were expressed as means ± S.E.M. PGC-1α expression levels were quantified as the ratio to GAPDH and the values of PGC-1α/GAPDH were normalized to that of control. **p < 0.01 compared with control.

Full size image

Overexpression of PGC-1α in Ho-8910 cells induces cell apoptosis

As shown in Figure 2A, moi of 50 for both Ad-GFP and Ad-PGC-1α resulted in 80-90% infection efficiency. RT-PCR analysis showed that PGC-1α expression was upregulated by 3.5-fold in Ad-PGC-1α-infected Ho-8910 cells as compared with the control group (Figure 2B).

Figure 2
figure 2

The expression levels of PGC-1α in recombinant adenovirus-infected Ho-8910 cells. Ho-8910 cells were infected with Ad-GFP and Ad-PGC-1α at a moi of 50 for 48 h. (A) Fluorescent photos of Ho-8910 cells after infection. (B) Representative RT-PCR analysis for PGC-1α in Ho-8910 cells. Lane 1, normal Ho-8910 cells, Lane2, Ho-8910-infected Ad-GFP, Lane 3, Ho-8910 infected by Ad-PGC-1α. (C) PGC-1α expression levels were quantified as the ratio to GAPDH and the values of PGC-1α/GAPDH were normalized to that of control. ***p < 0.001 compared with control.

Full size image

To investigate the effect of elevated PGC-1α on ovarian tumor cells, morphological changes were observed under phase contrast and light microscopes. At 48 h after infection, Ad-GFP-infected and control cells maintained normal morphology. In contrast, Ad-PGC-1α-infected Ho-8910 cells exhibited cell shrinkage, condensed chromatin, loss of cell-to-cell contact and had in part detached from the plate. Many of these characteristics were suggestive of apoptosis. We thus further characterized the phenomena by Hoechest staining assay, and our results showed that apoptotic bodies containing nuclear fragments were only found in Ad-PGC-1α-infected cells (73%) as compared with control and mock vector infected cells (Ad-GFP group) (Figure 3A).

Figure 3
figure 3

Detection of PGC-1α-induced apoptosis in Ho-8910 cells. (A) Hoechest 33258 staining of Ho-8910 cells. From Left to Right: Control, Ad-GFP, Ad-GFP-PGC-1α (fluorescence microscope, 400×). (B) Transmission electron microscopy of cells infected by Ad-PGC-1α. From Left to Right: Control, Ad-GFP, Ad-PGC-1α (original magnification, 3000×). (C) Representative flow cytometry profiles of either uninfected or infected Ho-8910 cells. From Top to Bottom: Control (0, 24 and 48 h), Ad-GFP (0, 24 and 48 h), Ad-GFP-PGC-1α (0, 24 and 48 h).

Full size image

To further confirm the effect of PGC-1α on apoptosis, cell slides were analyzed by electron microscopy. Results as shown in Figure 3B demonstrated that typical chromatin condensation and disappearance of nuclear ultrastructure occurred when a high level of PGC-1α was expressed in Ho-8910 cells. No such phenomenon was observed in cells infected with the control vector.

We subsequently used fluorescence-activated cell sorting (FACS) to investigate how PGC-1α affected cell apoptosis at various time points after its forced expression. In this study, as the expression of PGC-1α increased, majority of the cells entered apoptosis. From 0 to 24 h post-infection, the sub-diploid nuclei indicative of apoptosis appeared (32.7%) in Ad-PGC-1α-infected cells, and at 48 h the sub-diploid nuclei were further increased to 58.9%. In contrast, Ad-GFP-infected cells were not significantly influenced and showed a similar profile as control cells. Taken together, our results clearly showed a distinct apoptosis induced by PGC-1α in Ho-8910 cells compared with the control group. Thus, overexpression of PGC-1α suppressed cancer cell growth and induced massive cell death by apoptosis.

Differential gene expression induced by overexpressed PGC-1α in Ho-8910 cells

To further investigate genes that contribute to PGC-1α-induced apoptosis in Ho-8910 cells, we performed a global gene expression analysis by the standard microarray technology. A total of 1 998 genes were altered in Ho-8910 cells overexpressing PGC-1α. Among those altered genes, 1 057 genes were downregulated and 941 genes were upregulated. Many apoptosis-inducing genes were found in the upregulated group while apoptosis-restraining genes were in the downregulated group. For instance, the expression level of Bcl-2 was suppressed while Bax was increased (Figure 4A). Other pro-apoptotic genes of the Bcl-2 family including Bak1, Hrk, Bcl-2l11, Bcl-2l1, Bcl-2l13 were also upregulated (Figure 4A). The expression level of caspase family such as the apoptotic effector caspase 3 also changed (Figure 4B). Genes involved in DNA fragmentation were significantly upregulated. For example, the levels of CIDE3, CIDEA and DFFA were increased 8-, 4- and 2-, respectively (Figure 4C). In contrast, several IAP (inhibitor of apoptosis protein) family genes such as BIRC3, BIRC5 and BIRC1 were all downregulated (Figure 4D). Other genes involved in the apoptosis pathway showed altered expressions, too (Figure 4E). Our array data suggested that increased expression of PGC-1α in Ho-8910 cells affected many genes involved in apoptosis.

Figure 4
figure 4

Differential expression of apoptosis-related genes in PGC-1α-overexpressed Ho-8910 cells. (A) Bcl-2 family members. (B) Caspase family members. (C) DNA fragmentation gene. (D) IAP (inhibitor of apoptosis protein) family genes. (E) Others. Black bars are the genes upregulated with the ratio =1.5, while gray bars are the genes downregulated with the ratio =0.6. Three independent experiments were performed, and for each test and control sample, two hybridizations were performed by using a reversal fluorescent strategy.

Full size image

PGC-1α overexpression in Ho-8910 cells alters Bcl-2 and Bax expression and increases cytochrome c release

To validate the microarray results, the expression ratio of the anti-apoptotic gene Bcl-2 versus pro-apoptotic gene Bax was analyzed by quantitative RT-PCR. As shown in Figure 5A, Bcl-2 expression in Ad-PGC-1α infected Ho-8910 cells was significantly decreased (by ∼69%) compared with the control group, whereas Bax expression was increased by 1.5-fold after Ad-PGC-1α infection. In contrast, no changes of Bcl-2 and Bax expression were observed in the Ad-GFP group. We also examined the cytochrome c protein level in the cytoplasm. Overexpression of PGC-1α caused significant increase of cytochrome c level in Ho-8910 cells compared with the Ad-GFP group (Figure 5B). The significant decrease of Bcl-2/Bax ratio and the increase of cytochrome c level in the cytoplasm are consistent with the induction of apoptosis in Ad-PGC-1α-infected Ho-8910 cells.

Figure 5
figure 5

(A) Bcl-2 and Bax expression in Ad-PGC-1α-infected Ho-8910 cells. Total RNA was extracted from Ho-8910 cells either uninfected or infected for 48 h and subjected to quantitative RT-PCR. From Left to Right: PGC-1α, Bax, Bcl-2. Data (n = 9) are expressed as means ± S.E.M. Gene expression levels were quantified as the ratio to GAPDH and the ratio values were normalized to that of control.*p < 0.05, **p < 0.01, ***p < 0.001 compared with controls. (B) Effect of PGC-1α on cytochrome c release in Ho-8910 cells. Ho-8910 cells were treated with GFP or PGC-1α adenovirus vehicle for 48 h. Cytosol-specific, mitochondria-free lysates were prepared. An equivalent amount of protein (60 μg) from individual lysates were electrophoresed, and probed by Western blot with anti-cytochrome c antibody.

Full size image

PPARγ antagonist suppresses PGC-1α-induced apoptosis in Ho-8910 cells

Since PPARγ ligands induce apoptosis in many cancers 17, 18, 19 and PGC-1α powerfully coactivates PPARγ 8, we asked whether PGC-1α-induced apoptosis in Ho-8910 cells is mediated by a PPARγ-dependent pathway.

Ho-8910 cells infected with the GFP or PGC-1α adenovirus were immediately treated with 10 μM GW9662, an irreversible PPARγ antagonist, for 48 h followed by Hoechest staining. As shown in Figure 6, the number of apoptotic cells decreased by about 50-60% in Ad-PGC-1α-infected Ho-8910 cells after 10 μM GW9662 treatment compared with untreated cells, suggesting that PPARγ antagonist GW9662 partially, but not completely, blocked PGC-1α-induced Ho-8910 cell apoptosis. This result suggests that PGC-1α exerted its effect through PPARγ-dependent and -independent pathways.

Figure 6
figure 6

Effect of PPARγ antagonist on PGC-1α-induced apoptosis in Ho-8910 cells. Ho-8910 cells were infected with the GFP or PGC-1α adenovirus vehicles in either the absence (−) or presence (+) of 10 μM GW9662. After 48 h, apoptosis was assayed by Hoechest 33258 staining. From Top to Bottom: Control, Ad-GFP, Ad-PGC-1α. (fluorescence microscope, 100×).

Full size image

Suppression of PPARγ inhibits the PGC-1α-induced apoptosis in Ho-8910 cells

To further test our hypothesis that PGC-1α-induced apoptosis occurs through PPARγ-dependent pathway, we knocked down PPARγ expression in Ho-8910 cells by using siRNA interference. As shown in Figure 7, siRNA S1 inhibited PPARγ expression by 35-40% in Ho-8910 cells as compared to negative control (siRNA N). The decrease of PPARγ expression also inhibited PGC-1α-induced apoptosis in Ho-8910 cells, and the number of apoptotic cells decreased by about 70-80% compared with groups that did not receive siRNA S1 (Figure 8). This result supported our interpretation that PGC-1α-induced apoptosis in Ho-8910 cells occurs through a PPARγ-dependent pathway.

Figure 7
figure 7

The expression level of PPARγ in siRNA-infected Ho-8910 cells. Ho-8910 cells were transfected with siRNA S1 or the negative control (siRNA N). The interference effect was analyzed by quantitative PCR. Data were shown as the ratio of PPARγ/GAPDH. Individual data points in this figure represent the means ± S.E.M. of 18 determinants from three independently prepared samples each with six measurements. ***p < 0.001 compared with the control or N group.

Full size image
Figure 8
figure 8

Effect of suppressed PPARγ expression on PGC-1α-induced apoptosis in Ho-8910 cells. Ho-8910 cells were infected with the GFP or PGC-1α adenovirus vehicles 48 h after the treatment with siRNA S1 or the negative control (siRNA N). Apoptosis was assayed by Hoechest 33258 staining. From Top to Bottom: Control, Ad-GFP, Ad-PGC-1α; From Left to Right: Control, N, S1 (fluorescence microscope, 400×).

Full size image

Discussion

The OSE is thought to be the origin of most ovarian cancers in human. Greater than 90% of ovarian cancer arises from the transformation of OSE 32. Therefore, study of the pathogenesis of epithelial ovarian cancer is very important in ovarian cancer therapy. Recent report shows that the expression of PPARγ is significantly enhanced in the late-stage epithelial ovarian carcinoma, indicating that PPARγ contributes significantly to the onset and progression of ovarian cancer 33. Because the levels of PPARγ and the PPARγ coactivator, PGC-1α, are always linked to clinical outcome in patients with a variety of human cancers 12, 13, 14, it is reasonable to suspect that PGC-1α may also play a pivotal role in the pathogenic development of epithelial ovarian cancer. However, no correlation has been established for the possible involvement of PGC-1α in ovarian cancer. In this study, we first demonstrated that PGC-1α mRNA level was significantly decreased in late-stage human epithelial ovarian cancer compared with normal ovarian tissues. This finding is also consistent with others reports and our own studies showing that PGC-1α is downregulated in breast cancer 12, 13, colon cancer 14, liver cancer and hypothyroid cancer (data not shown). Therefore, a better understanding of PGC-1α's effect on epithelial ovarian cancer may have important implications for the treatment of this disease in clinics.

Our human epithelial ovarian cancer samples were collected from patients of ovarian serous cystadenocarcin-omas; thus, the Ho-8910 cell line, which also originated from human ovarian serous cystadenocarcinoma 34, would be a suitable model for in vitro study. Since PGC-1α was downregulated in epithelial ovarian cancer, we upregulated PGC-1α expression in epithelial ovarian cancer cell line Ho-8910 by adenovirus infection to examine the potential effect. Strikingly, we found that elevated PGC-1α induced cell apoptosis. PGC-1α has been shown to stimulate mitochondrial biogenesis and respiration in muscle cells 8, to stimulate the hepatic gluconeogenesis in liver 10 or to regulate insulin secretion in islets 11. There has been no previous reports showing that PGC-1α can induce apoptosis in cancer. The current findings uncovered a new function of PGC-1α in tumorigenesis, especially in ovarian cancer. The fact that PGC-1α expression decreases in ovarian cancer in vivo and the increase of PGC-1α expression promotes apoptosis of ovarian cancer cell in vitro suggests that PGC-1α might be closely involved in the pathogenesis of the epithelial ovarian cancer.

To address the mechanisms involved in PGC-1α-induced apoptosis in Ho-8910 cells, the mRNA levels of Bax and Bcl-2 were analyzed by quantitative PCR. Pro-apoptotic gene Bax and anti-apoptotic gene Bcl-2 are known as Bcl-2 family members, which are the central regulator of apoptosis 35, and the opposing functions of anti- and pro-apoptotic members arbitrate the life-or-death decision 36. The ratio of Bcl-2 to Bax is important in determining susceptibility to apoptosis and is usually regarded as a criterion in programmed cell death 37. Our results revealed that Bcl-2 was markedly downregulated while Bax was concurrently upregulated in PGC-1α-overexpressed Ho-8910 cells, suggesting that PGC-1α-induced apoptosis is mediated by a decreasing ratio of Bcl-2/Bax. The change of Bcl-2/Bax can destabilize mitochondria, leading to the release of several mitochondrial intermembrane space proteins such as cytochrome c into the cytosol where they actively trigger the process of programmed cell death 38, 39. Indeed, we found that exogenous PGC-1α can induce cytochrome c release from mitochondria into the cytosol in Ho-8910 cells.

In addition, our microarray analysis also indicated that the expressions of many apoptosis-related genes were markedly changed in PGC-1α-infected Ho-8910 cells, particularly the Bcl-2 family genes such as Bak, Hrk,Bcl2l1, Bcl2l11 and Bcl2l13. It was most likely that PGC-1α-induced apoptosis was mediated by the Bcl-2 family in Ho-8910 cells. The change of pro-apoptotic/anti-apoptotic member ratio then triggered the cytochrome c release from mitochondria to the cytosol, leading to downstream apoptosis. The mechanism by which this set of apoptosis-related genes was coordinately regulated by PGC-1α may be related with PPARγ, because PGC-1α powerfully coactivates PPARγ and the activation of PPARγ can also induce apoptosis through Bcl-2 family members in cancers 40, 41, 42, 43. To investigate whether PGC-1α-induced apoptosis is mediated through PPARγ, we utilized the PPARγ antigonist GW9662 to block the action of PPARγ. Results showed that GW9662 can partially inhibit PGC-1α-induced apoptosis in Ho-8910 cells, suggesting that PGC-1α may facilitate apoptosis through a PPARγ-dependent pathway. Since this antagonist has some off-target activity, we suppressed PPARγ expression by siRNA, and found that this also inhibited PGC-1α-induced apoptosis in Ho-8910 cells, supporting our interpretation that PGC-1α exerted its effect through a PPARγ-dependent pathway. Although PPARγ is upregulated in epithelial ovarian carcinoma 23, the downregulation of PGC-1α in epithelial ovarian cancer may restrain the activation of PPARγ-dependent apoptotic pathway, which could indirectly contribute to the development of ovarian tumor. However, the detailed molecular mechanisms need further study. In addition, due to difficulties to obtain normal human ovarian tissues or cells, the specificity of cytotoxic effects of PGC-1α on ovarian cancers remains to be addressed.

In summary, our current study provides the first evidence that the expression of PGC-1α is downregulated in human epithelial ovarian cancer and that the overexpre-ssion of PGC-1α can induce apoptosis in human ovarian epithelial cancer cell line Ho-8910. The cytotoxic effect of PGC-1α on ovarian cancer cells may be useful for cancer therapy.