PGC-1α and ERRα in patients with endometrial cancer: a translational study for predicting myometrial invasion

PGC-1α and ERRα positively correlate with more advanced myometrial invasion in endometrial cancer

A total of 121 specimens from patients with endometrial cancer (N=81) or healthy subjects (N=40) were analyzed, and we found that PGC-1α was expressed in all tissue specimens examined. ERRα showed positive expression in all endometrial carcinoma samples and in 38 of 40 samples of normal endometrium. The immunoreactive score revealed that PGC-1α and ERRα expression was significantly higher in endometrial carcinoma than in normal endometrium (Figure 1A, 1B). Spearman’s correlation analysis showed that the expression of PGC-1α significantly associated with that of ERRα (Spearman’ s rank correlation 0.638, P < 0.001). We also analyzed the correlation between the expression of PGC-1α and ERRα and the clinicopathologic features of the subjects including FIGO stage, histologic grade, histology type, myometrial invasion, and nodal metastasis (Table 1). Higher expression of PGC-1α and ERRα positively correlated with more advanced myometrial invasion (P = 0.038 and 0.039, respectively). It was confirmed that the expression of PGC-1α and ERRα was higher in highly invasive endometrial cancer tissues than in less invasive endometrial cancer and significantly higher than in normal tissues (Figure 1C).

Figure 1. Expression of PGC-1α and ERRα in normal endometrium and endometrial cancer tissue. Expression of PGC-1α (A) and ERRα (B) in normal endometrium and different clinical stage endometrial cancer. (C) HE staining and immunohistochemical staining of PGC-1α, ERRα in normal endometrium (Case 1), superficial myometrial invasion EC (Case 2) and deep myometrial invasion EC (Case 3).

Analysis of factors related to advanced myometrial invasion in endometrial cancer

EC patients were divided into two groups: less and more than 1/2 myometrial invasion. Clinical characteristics were compared between the two groups, and no significant differences were observed in age, BMI, triglyceride, cholesterol, or electrolyte levels. In contrast, the serum glucose (P = 0.0407) and CA125 (P = 0.034) levels were higher in patients who had advanced myometrial invasion (Table 2). Factors such as blood glucose, serum CA125, PGC-1α, and ERRα with a P value of less than 0.05 after a t test or χ2 test were used as variables for single-factor logistic regression analysis. The results show that PGC-1α and ERRα were cogent predictors for myometrial invasion in patients with endometrial cancer (P <0.05) (Table 3).

Myometrial invasion in pre- and postmenopausal endometrial cancer is related to PGC-1α and vimentin

In this study, the frequency of deep 1/2 myometrial invasion of premenopausal endometrial cancer was 10/41 (24.4%), which was lower than that of postmenopausal endometrial cancer (15/40; 37.5%). Vimentin is an important indicator for cancer invasion and metastasis. Vimentin and PGC-1α expression levels were statistically different in pre- and postmenopausal endometrial cancer (P = 0.006 and P= 0.030, respectively). Spearman correlation analysis showed that the expression of PGC-1α in endometrial cancer was positively correlated with that of vimentin (r= 0.263, P = 0.018). Based on menopausal status combined with the depth of myometrial invasion of endometrial cancer, it was found that the expression of PGC-1α was related to the depth of myometrial invasion of premenopausal endometrial cancer (P = 0.022), but not to that of postmenopausal endometrial cancer (P = 0.056) (Table 4). Similarly, vimentin expression was related to the depth of myometrial invasion of premenopausal endometrial cancer (P = 0.009), but not to that of postmenopausal endometrial cancer (P = 0.064). Spearman correlation analysis showed that the expression of PGC-1α in premenopausal endometrial cancer patients was positively correlated with that of vimentin (r= 0.344, P = 0.027) (Table 4).

PGC- 1α and ERRα are mutually regulated in endometrial cancer cells

The relative mRNA expression of PGC-1α was (1 ± 0.0783), (1.600 ± 0.0435), (0.3227 ± 0.0359) and (0.3018 ± 0.0291) in RL-952, ECC-1, HEC-1A and HEC-1B cells, respectively. The relative mRNA expression of ERRα in the same cells was (1 ± 0.0413), (1.4857 ± 0.0329), (0.2608 ± 0.0356) and (0.163 3± 0.0436), respectively (Figure 2A). Western blot assays showed a similar expression pattern (Figure 2B, Supplementary Figure 1A). Spearman’s correlation analysis showed that the expression of PGC-1α was positively correlated with that of ERRα (r = 0.697, P < 0.01). In line with previous reports, it’s confirmed that PGC-1α and ERRα are overexpressed in RL-952 and ECC-1 cells, and are expressed at low levels in HEC-1A and HEC-1B cells. Thus, we choose the ECC-1 (high expression of PGC-1α and ERRα) and HEC-1A (low expression of PGC-1α and ERRα) cells for further experiments.

Figure 2. Expression of PGC-1α and ERRα in endometrial cancer cells. (A) related mRNA expression and (B) protein expression pattern of PGC-1α and ERRα in RL-952, ECC-1, HEC-1A and HEC-1B endometrial cancer cells. (C) PGC-1α and ERRα mRNA expression and (D) protein levels was significantly up-regulated after the infection with lentivirus targeted on OV-ERRα. Relative mRNA expression in ECC-1 (E) and HEC-1A (F) cells treated with XCT790 showed a downregulation of ERRα. (G) PGC-1α and ERRα mRNA expression (H) protein levels was significantly down-regulated after the infection with PGC-1α-siRNA, ERRα is consistent with PGC-1α regulation. Relative mRNA expression of PGC-1α is consistent with ERRα regulation. *mean P<0.05, # mean P>0.05.

In HEC-1A cells, ERRα levels were higher in the OV-ERRα group than in the BLANK and CON groups(BLANK, 1.005 ± 0.0305; CON, 0.985 ± 0.0288; OV-ERRα, 2.884 ± 0.032; P < 0.01). PGC-1α expression increased upon upregulation of ERRα (BLANK, 1.003 ± 0.040; CON, 0.981 ± 0.0324; OV-ERRα, 2.312 ± 0.026; P < 0.01; Figure 2C). Western blot assays confirmed these results (Figure 2D, Supplementary Figure 1B). Next, we treated ECC-1 and HEC-1A cells with XCT790, an ERRα inverse agonist. We observed the downregulation of ERRα levels in XCT790-treated cells compared with the untreated cells or those treated with DMSO (Figure 2E). Interestingly, we observed an ERRα-dependent inhibition of PGC-1α expression following XCT790 treatment (ECC-1: BLANK, 1.001 ± 0.027; DMSO, 0.975 ± 0.043; XCT790, 0.471 ± 0.051; P < 0.01 - HEC-1A: BLANK, 1.003 ± 0.040; DMSO, 0.978 ± 0.0372; XCT790, 0.132 ± 0.067; P < 0.01; Figure 2F). These data show that the expression of PGC-1α is regulated by ERRα. We next investigated whether PGC-1α affects ERRα expression by silencing PGC-1α expression. In ECC-1 cells, PGC-1α expression was significantly lower (89.5 %) in the siPGC-1α group than in the BLANK group, whereas no difference was observed between the BLANK and CON groups (BLANK, 1.001 ± 0.027; CON, 0.998 ± 0.034; siPGC-1α, 0.104 ± 0.0361; P < 0.01). Notably, ERRα levels decreased upon PGC-1α silencing (BLANK, 1.006 ± 0.022; CON, 0.997 ± 0.031; siPGC-1α, 0.0240 ± 0.0111; P < 0.01; Figure 2G). These results were confirmed via western blot analysis (Figure 2H, Supplementary Figure 1C) and suggest that PGC-1α induces ERRα expression in ECC-1 cells.

Downregulation of PGC-1α and ERRα inhibit the migration and invasion ability of endometrial cancer cells

We then assessed whether PGC-1α affects cell migration/invasion. In migration assays in ECC-1 cells, the width of the scratch was smaller in the BLANK and CON groups than that in the siPGC-1α group (BLANK, 409.32 ± 50.99; CON, 453.79 ± 22.98; siPGC-1α, 72.898 ± 21.32; F = 200.980, P < 0.001; Figure 3A and 3E). The results in the BLANK and CON groups were comparable (P > 0.05). Person’s correlation analysis showed that PGC-1α and ERRα levels were positively correlated with the migration distance (r_PGC-1α = 0.983 and r_ERRα = 0.966) (Supplementary Figure 3A, 3B). In invasion assays, the number of invading cells significantly decreased after PGC-1α silencing (BLANK, 174.67 ± 6.532; CON, 172.00 ± 7.176; siPGC-1α, 32.33 ± 6.623; F=847.642, P < 0.001; Figure 3C and 3F). Additionally, the number of invading cells in the BLANK and CON groups were comparable (P > 0.05). Pearson’s correlation analysis showed positive correlation between PGC-1α and ERRα levels and the number of invading cells (r_PGC-1α = 0.996 and r_ERRα = 0.992) (Supplementary Figure 2C, 2D).

Figure 3. Regulation of the migration and invasion ability of endometrial cancer cells by PGC-1α and ERRα. (A) Compared with blank groups (Blank), the 24-hour migrated distances of ECC-1 siPGC-1α group were significantly reduced. (B) Compared with blank groups (Blank), the 24-hour migrated distances of HEC-1A OV-ERRα group were significantly increased. (C) The endometrial cancer cells penetrated the Transwell membrane were significantly reduced in ECC-1 siPGC-1α group compared with Blank group. (D) the endometrial cancer cells penetrated the Transwell membrane were significantly increased in HEC-1A OV-ERRα group compared with Blank. Quantification of migration (E) and invasion (F) of ECC-1 cells. Quantification of migration (G) and invasion (H) of HEC-1A cells. *mean P<0.05, # mean P>0.05.

Enhanced PGC-1α/ERRα expression is associated with distant migration and invasion of endometrial cancer cells

In experiments similar to the ones described above, the migration capability of HEC-1A cells overexpressing ERRα was significantly higher than that of the BLANK and CON groups (BLANK, 87.376 ± 16.905; CON, 82.55 ±1 8.086; OV-ERRα, 235.712 ± 11.832; F = 152.652, P < 0.001; Figure 3B and 3G). The BLANK and CON groups showed similar results (P > 0.05). Person’s correlation analysis showed that PGC-1α and ERRα levels positively correlated with migration distance (r_PGC-1α= 0.979 and r_ERRα = 0.995) (Supplementary Figure 3A, 3B). In invasion assays, the number of invading cells was significantly higher in the OV-ERRα group than in the BLANK or CON groups (BLANK, 58.40 ± 6.269; CON, 56.67 ± 7.23; OV-ERRα, 129.00 ± 6.285; F = 198.471, P < 0.001). The invading capability of cells of the BLANK and CON groups was similar (P > 0.05; Figure 3D and 3H). Pearson’s correlation analysis showed a positive correlation between PGC-1α and ERRα levels and the number of invading cells (r_PGC-1α = 0.984 and r_ERRα = 0.974)(Supplementary Figure 3C, 3D).

Inhibition of PGC-1α/ERRα suppresses EMT in endometrial cancer cells

Increasing evidence shows that the progression of cancer cell motility is associated with EMT. We therefore hypothesized that PGC-1α plays a positive role in the progression of EMT. Immunofluorescence assays revealed increased expression of the epithelial cell marker E-cadherin (E-cad) and decreased expression of the mesenchymal cell marker vimentin (Vim) in the si-PGC-1α group compared with the BLANK and CON groups (Figure 4A). Western blotting analysis confirmed that inhibition of PGC-1α expression led to increased expression of E-cadherin (BLANK, 0.447 ± 0.125; CON, 0.443 ± 0.162; siPGC-1α, 0.884 ± 0.123; F = 2021.448, P < 0.001) and decreased expression of vimentin (BLANK, 0.700 ± 0.137; CON, 0.712 ± 0.111; siPGC-1α, 0.160 ± 0.026; F = 5533.24, P < 0.001; Figure 4C and 4D).

Figure 4. Up-and down-regulation of PGC-1α and ERRα alter the EMT process in EC cells. (A) the representative immunofluorescence images of E-cadherin (purple) and vimentin (red) in ECC-1cells and siPGC-1α-ECC-1cells. The nucleus is blue. optical microscope and fluorescence microscopic with a magnification of 400. (B) the representative immunofluorescence images of E-cadherin (purple) and vimentin (red) in HEC-1A cells and OV-ERRα cells. (C) The protein of E-cadherin and vimentin were measured by Western blot analysis in ECC-1cells and siPGC-1α-ECC-1cells. (D) Quantification of E-cadherin and vimentin protein in ECC-1 cells. (E) The protein of E-cadherin and vimentin were measured by Western blot analysis in HEC-1A cells and OV-ERRα cells. (F) Quantification of of E-cadherin and vimentin protein in HEC-1A cells. *mean P<0.05, # mean P>0.05.

Overexpression of PGC-1a/ERRα induces EMT in endometrial cancer cells

In similar experiments, immunofluorescence assays showed that ERRα overexpression in HEC-1A cells was associated with decreased expression of E-cadherin (E-cad) and increased expression of vimentin (Vim) (Figure 4B). Western blotting analysis also revealed that overexpression of ERRα led to decreased expression of E-cadherin (BLANK, 1.447 ± 0.182; CON, 1.486 ± 0.124; OV-ERRα, 0.855 ± 0.135; F = 3087.854, P < 0.001) and increased expression of vimentin (BLANK, 0.215 ± 0.134; CON, 0.236 ± 0.099; OV-ERRα, 0.746 ± 0.014; F = 2992.792, P < 0.001; Figure 4E and 4F). Therefore, the overexpression of ERRα is associated with a more aggressive phenotype.

Overexpression of ERRα increases TFEB transcription factor activity in endometrial cancer cells

High-throughput protein/DNA array analysis has previously shown that transcription factor EB (TFEB) is downregulated upon ERRα knockdown via PGC-1α. Therefore, we next examined whether TFEB activity was affected by overexpression of ERRα in endometrial cancer cells. TFEB activity was 0.346 ± 0.011 in the BLANK group, 0.345 ± 0.009 in the CON group and 0.469 ± 0.005 in the OV-ERRα group (P < 0.001).

Ethics committee approval

The study was conducted in accordance with ethical standards, the Declaration of Helsinki, and national and international guidelines, and has been approved by the Ethics Committee of Fujian Maternity and Child Health Hospital affiliated with Fujian Medical University (No.2018-014). An informed consent was obtained from all patients.

Study population and tissues

A total of 81 endometrial carcinoma and 40 normal endometrium samples with related clinical data were obtained from the patients who underwent surgical therapy in Fujian provincial Maternity & Children’s Health Hospital of Fujian Medical University, China. None of the patients received any chemotherapy, radiation, or hormonal therapy before surgery. An informed consent was obtained from all patients. This research protocol was approved by the Ethics Committee of Fujian Maternity and Child Health Hospital affiliated with the Fujian Medical University (No.2018-014).

Immunohistochemistry

All tissues were assembled into a tissue chip. Immunostaining for ERRα and PGC-1α was performed according to standard procedures. Rabbit polyclonal anti-ERRα (dilution 1:100; Abcam, UK) and rabbit polyclonal anti-PGC-1 alpha-N-terminal (dilution 1:200; Abcam) antibodies were used. The percentage of positive cells was scored as 0 (cells < 5%), 1 (5% to 25%), 2 (26% to 50%), 3 (51% to 75%), and 4 (76% to 100%). Positive staining intensity was scored as 0 (no staining), 1 (weak staining), 2 (moderate staining), and 3 (strong staining). The expression level of ERRα and PGC-1α was measured by the immunoreactive score (IRS) using the algorithm: IRS = Si × Pi (where Si and Pi represent the intensity and percentage of cells with each intensity, respectively). Samples were divided into four groups based on their IRS: 0, negative (-); 1-4, weakly positive (+); 5-8, positive (++); 9-12, strongly positive (+++).

Cell culture and treatment with XCT790

The endometrial cancer cell lines RL-952, ECC-1, HEC-1A and HEC-1B were purchased from the American Typical Culture Collection (ATCC, Rockville, USA). The cells were grown as previously described [33]. Cells treated with XCT790 (Sigma-Aldrich, St. Louis, MO, USA) were incubated in phenol red-free medium (Thermo Fisher) containing 1% Serum Replacement 2 (Sigma-Aldrich). ECC-1 and HEC-1A EC cells were incubated with 10 μM XCT790 (in dimethyl sulfoxide [DMSO]; Sigma-Aldrich) or DMSO (control) for 24 h.

Quantitative PCR (qPCR)

Total RNA was isolated according to the manufacturer’s protocol (Invitrogen, USA, Thermal), as previously described [33]. The mRNA was transcribed into cDNA using an Access RT-PCR System (Promega, USA). qPCR assays were performed using a LightCycler^® 480 SYBR Green I Master Mix (Roche, Germany). The following primers, synthesized by Sanggong Biotech (Shanghai, China), were used: PGC-1α, forward 5′-GAC ACA ACA CGG ACA GAA-3′ and reverse 5′- CAC AGG TAT AAC GGT AGG TAA -3′ (PCR product, 121 bp); ERRα, forward 5′- ACC GAG AGA TTG TGG TCA CCA -3′ and reverse 5′- CAT CCA CAC GCT CTG CAG TAC T- 3′ (101 bp); and GAPDH, forward 5′- GCA CCG TCA AGG CTG AGA AC- 3′ and reverse 5′- TGG TGA AGA CGC CAG TGG A -3′ (138 bp). The target gene mRNA levels were quantified using the comparative method (2^−ΔΔCT method) and normalized to GAPDH expression.

Western blotting

Whole-cell proteins were extracted according to the manufacturer’s protocol (Clontech, Palo Alto, USA) and their concentration was determined using an ELISA kit (Pierce), as previously described [33]. Thirty micrograms of whole-cell protein lysate was loaded to each lane of an 8%-polyacrylamide gel. Proteins were blotted onto nitrocellulose membranes. Membranes were incubated with a rabbit monoclonal antibody specific to ERRα (1:500; CST), PGC-1α (1:1,000; CST), or vimentin (1:1,000; CST) or E-cadherin (1:1,000; CST) in blocking buffer, followed by incubation with an alkaline phosphatase-conjugated secondary antibody (1:1,000; Abcam). Immunoreactive bands were visualized using the CDP star RTU luminescence system (Tropix).

Lentivirus-mediated silencing of PGC-1α and overexpression of ERRα

A lentiviral vector expressing small interfering RNAs (siRNA) targeting PGC-1α and named siRNA-PGC-1α (siPGC-1α) was constructed. The following siRNA target sequence in the PGC-1α gene (GenBank accession No. NM_013261) was selected: siPGC-1α, 5′- CAA CTT TAT CTC TTC CTC TGA -3′. A universal sequence (PSC-NC: 5′- TTC TCC GAA CGT GTC ACG T -3′ named CON) was used as the negative control. The lentiviral vector used to overexpress ERRα (named ESRRA, GenBank accession NM_004451) or the universal sequence (negative control) were purchased from Genechem (Shanghai, China). Three groups of cells were analyzed: cells transduced with lentiviruses expressing siRNAs targeting PGC-1α (siPGC-1α group) or overexpressing ERRα (OV-ERRα group), cells transduced with empty lentiviruses (CON group), and untreated cells (BLANK group). Lentivirus-based vectors carry the green fluorescent protein (GFP) gene (GV115, Genechem, Shanghai, China). After transduction for 72 h, GFP expression was detected to calculate the transduction efficiency (Supplementary Figure 1D–1E). EC cells were transduced at a multiplicity of infection (MOI) of 100. After 72 h of transduction, GFP expression was detected to calculate the infection efficiency.

In vitro cellular scratch assays

Cells were grown to confluence in 6-well plates and a 200-μL tip was used to introduce a scratch in the monolayer. The scratch areas in the wells were washed with PBS and 1 mmol/L R-flurbiprofen until the cells in those areas were removed thoroughly and imaged at 0 and 24 h post-scratching. The horizontal migration rate was calculated using the following formula: (width_{0 h} − width_{24 h})/width _{0 h} × 100% [34].

Transwell chamber invasion assays

Matrigel™ Basement Membrane Matrix (50 μL; BD, USA) was added to a Millicell Hanging Cell Culture Insert (Millipore, USA) to coat the membrane. Two hundreds microliters of cell suspension containing 0.5% FBS (5.0 × 10⁵ cells/mL) were added to the insert, placed in 24-well plates containing 1,300 μL of DMEM supplemented with 10% FBS. After incubation for 24 h, non-invading cells on the top of the filter were removed with a cotton swab, and the filters were fixed with methanol and stained with crystalline violet. The filters were removed from the inserts and mounted onto slides for imaging and quantification as described in a previous study [34].

TFEB transcription factor assay

Nuclear and cytoplasmic cell extracts were prepared using a Nuclear Extract kit. One hundred microliters of nuclear extract was added into a 96-well plate with immobilized oligonucleotides containing TFEB consensus binding sites. A positive control for TFEB activation was set, and 20 μL complete lysis buffer served as the blank. A diluted primary antibody (100 μL) was incubated for 1 h at room temperature. After three washes, an HRP-labeled secondary antibody was incubated. After 50 μL stop solution was added, the absorbance was measured by a microplate reader (RayBio).

Cell immunofluorescence

Cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100, and incubated with blocking buffer samples, and then with a rabbit monoclonal antibody specific to vimentin (1:100 dilution, CST) and mouse monoclonal antibody specific to E-cadherin (1:50 dilution, CST) at 4 °C overnight. After extensive wash, secondary anti-rabbit IgG conjugated with Alexa 647(CST) or anti-mouse IgG conjugated with Alexa 594 (Proteintech) were incubated. The samples were then fixed and analyzed via confocal microscopy. Cell fluorescence intensity was quantified with Image J.

Statistical analysis

Statistical analysis was performed using the average results of three experiments under identical conditions. Numerical data are presented as the mean ± SD. Differences between two means were compared using Student’s t-test, and related parameters were analyzed using Pearson’s correlation. Correlation coefficients for graded data were obtained using spearman correlation analysis. Data were analyzed using the SPSS 17.0 software for Windows (SPSS Inc., Chicago, IL, USA). Differences were considered significant at P< 0.05.