The human endometrial adenocarcinoma cell lines AN3CA, EFE-184, ETN-1, Hec1A, Hec108 and Nou-1 were obtained from the Dana-Farber/Harvard Cancer Center (Boston, MA, USA). AN3CA, EFE-184, ETN-1, Hec108 and Nou-1 cells were maintained at 37°C in a humidified normoxic atmosphere in the presence of 5% CO₂ in RPMI-1640 (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA), and Hec1A cells were maintained in McCoy's 5A (Gibco; Thermo Fisher Scientific, Inc.). HEK293T cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Gibco; Thermo Fisher Scientific, Inc.). All the media were supplemented with 10% fetal bovine serum (FBS; HyClone; GE Healthcare, Logan, UT, USA). Among the 6 cell lines, EFE-184 and ETN-1 cells were screened and used the two for the subsequent experiments. Palbociclib was purchased from Selleck Chemicals (Houston, TX, USA; cat. no. S1579), and GSK-J4 was obtained from MedChemExpress (Shanghai, China; cat. no. HY-15648B).

Cells were plated into 96-well plates at a density of 5×10³ cells/well, and incubated at 37°C with or without increasing concentrations of palbociclib (0, 0.25, 0.5, 1, 2.5, 5 and 10 µM). Cells were harvested at 72 h after palbociclib or vehicle [dimethyl sulfoxide (DMSO)] treatment, and evaluated with a Cell Counting Kit-8 assay (Dojindo Molecular Technologies, Inc., Kumamoto, Japan), according to the manufacturer's guidelines, as previously described (15). GraphPad Prism software 5.0 (GraphPad Software, Inc., San Diego, CA, USA) was used to calculate the half-maximal inhibitory concentration (IC₅₀) values.

The construction of shRNA P16INK4A was performed as previously described (16). Briefly, a doxycycline-inducible shRNA system was constructed and the plasmids (~150 ng/µl) were transfected into cells with a lentivirus. The following primers were designed and synthesized by Sangon Biotech Co., Ltd. (Shanghai, China), and used for P16INK4A knockdown: shP16INK4A-1, forward, 5′-CCGGCGCACATTCATGTGGGCATTTCTCGAGAAATGCCCACATGAATGTGCGTTTTTG-3′, and reverse, 5′-AATTCAAAAACGCACATTCATGTGGGCATTTCTCGAGAAATGCCCACATGAATGTGCG-3′; and shP16INK4A-2, forward, 5′-CCGGATCAGTCACCGAAGGTCCTACCTCGAGGTAGGACCTTCGGTGACTGATTTTTTG-3′, and reverse 5′-AATTCAAAAAATCAGTCACCGAAGGTCCTACCTCGAGGTAGGACCTTCGGTGACTGAT-3′. To generate pLKO-tet-on-shRNAs targeting human P16INK4A, oligonucleotides were designed and synthesized (Sangon Biotech Co., Ltd.). Following annealination, double-stranded oligonucleotides were directly ligated with pLKO-Tet-on vector that was digested with Agel (New England Biolabs, Inc., Ipswich, MA, USA) and EcoRI (Takara Biotechnology Co., Ltd., Dalian, China). Retroviruses were generated by transfecting HEK293T cells with pWzl plasmids and packaging DNA, A total of 1.6 µg pWzl DNA, 1.2 µg pCG-VSVG, 1.2 µg pCG-gap/pol and 12 µl Lipofectamine^® 3000 (Thermo Fisher Scientific, Inc.) were used. DNA and lipid were diluted in 300 µl DMEM and mixed, and following 15 min of incubation at room temperature, they were added to one 6-cm dish that was seeded with 3×10⁶ HEK293T cells 1 day earlier at room temperature. Viral supernatant was collected 48 and 72 h after transfection. After the supernatant was filtered through 0.45-µm membrane, it was added to target cells in the presence of 8 µg/ml polybrene (EMD Millipore, Billerica, MA, USA). Lentiviruses were generated with a similar approach with the exception of HEK293T cells that were transfected with 2 µg pLKO DNA, 1.5 µg pCMV-dR8.91 (Biovector, Beijing, China) and 0.5 µg pMD2-VSVG (Biovector). Cells were selected with antibiotics at 72 h after initial infection. Puromycin (Thermo Fisher Scientific, Inc.) and blasticidine (Thermo Fisher Scientific, Inc.) were used at the final concentrations of 1.5 and 4 µg/ml, respectively. The knockdown of P16INK4A was confirmed by reverse transcription-quantitative polymerase chain reaction (RT-qPCR) analysis and western blotting for at least 3 times each. The subsequent experiments were performed immediately following the transfection.

Following transfection with shRNA plasmids, cells were seeded into 12/24-well plates at a density of 3×10³/6×10³ cells/well, respectively, with or without doxycycline (1 µg/µl), and cultured at 37°C for 10 days. ETN-1 and EFE-184 cells were also plated in 6-well plates at a density of 1×10⁴ cells/well with GSK-J4 (30 µM) or DMSO, and were also cultured at 37°C for 10 days. The culture medium RPMI-1640, containing doxycycline (1 µg/µl), GSK-J4 (30 µM) or DMSO, was changed every 3 days. After 10 days of treatment, the cells were fixed and stained with methanol and crystal violet (Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) mixture (0.05% concentration at room temperature for 30 min), and then extracted with 10% glacial acetic acid. The optical density (OD) was measured at 570 nm by EnSpire^® Multimode Plate Readers (PerkinElmer, Inc., Waltham, MA, USA).

The cells not treated were collected at 80% confluence, while if treated with GSK-J4, the cells were collected at the 24 h after the treatment, and if treated with Palbociclib, they were collected at 12 h after treatment. The cells were seeded at 1×10⁴ cells/plate and maintained in 60-cm plates and were lysed on ice in radioimmunoprecipitation assay buffer supplemented with protease and phosphatase inhibitors (KGP2100; Nanjing KeyGen Biotech Co., Ltd., Nanjing, China), as previously described (17). Immunoblots were conducted using a Bio-Rad system (Bio-Rad Laboratories, Inc., Hercules, CA, USA). A total of 25 µg protein was loaded for each sample, the total protein was separated by 12% SDS-PAGE and transferred to polyvinylidene difluoride membrane (EMD Millipore). The membranes were blocked with 5% milk at room temperature for 1 h and the target proteins were probed by the following antibodies for 12 h at 4°C: P16INK4A (OriGene Technologies, Inc., Rockville, MD, USA; cat. no. ZS-0033; 1:500), KDM6B (ProteinTech Groups, Inc., Chicago, IL, USA; cat. no. 55354-1-AP; 1:1,000), H3K27-M3 (Immunoway Biotechnology Company, Plano, TX, USA; cat. no. YM3338; 1:1,000) and H3K27-M1 (Immunoway Biotechnology Company; cat. no. YM3336; 1:1,000). Vinculin (Cell Signaling Technology, Inc., Danvers, MA, USA; cat. no. #4650; 1:1,000) was used as loading control. The membrane was lastly stained with horseradish peroxidase (HRP)-conjugated rabbit anti-sheep IgG H&L (1:5,000; ab6747; Abcam, Shanghai, China) for 2 h at 4°C. The protein bands were then detected using a radioactive detection system (SuperLumia Enhanced Chemiluminescence kit; cat. no. K22020; Abbkine Scientific Co., Ltd., Redlands, CA, USA, www.abbkine.com). The absolute density of P16INK4A bands was later determined by ImageJ software 4.0 (National Institutes of Health, Bethesda, MD, USA). The relative density of P16INK4A was normalized to the corresponding vinculin bands. In vehicle-treated cells, the relative density of P16INK4A was assigned as 1.

Total RNA was isolated from the cells with an Illustra RNAspin kit (GE Healthcare) and reverse-transcribed using a iScript cDNA synthesis kit (Bio-Rad Laboratories, Inc.), according to the manufacturer's protocols. An universal SYBR^®-Green MasterMix and RT-qPCR was used to measure gene expression. Briefly, the samples were incubated for 60 min at 42°C, 15 min at 72°C, and stored at −20°C. For qPCR, 1 µl of diluted RT products were mixed with 10 µl of SYBR^®-Green MasterMix (Bio-Rad Laboratories, Inc.), 0.5 µl of forward and reverse primers, and 4 µl of nuclease-free water for a final volume of 20 µl, according to the manufacturer's protocols. The reactions were run on the ABI-7500 Real-Time PCR System (Applied Biosystems; Thermo Fisher Scientific, Inc.) using the conditions of 40 cycles at 95°C for 20 sec and 60°C for 45 sec. Actb was used as the loading control. The primers for P16INK4A were as follows: Forward, 5′-ATATGCCTTCCCCCACTACC-3′, and reverse, 5′-CCCCTGAGCTTCCCTAGTTC-3′. The primers for Actb were: forward, 5′-CCTAGAAGCATTTGCGGTGG-3′, and reverse, 5′-GAGCTACGAGCTGCCTGACG-3′. Cq values were generated using the default analysis settings. ΔCq was defined as Cq gene of interest - Cq β-actin. ΔΔCqT was defined as ΔCq treated sample - Cq control sample. Relative quantification was calculated as 2^−ΔΔCq, as described previously (18).

As previously described (19), the cells (2,000/well) were seeded on 96-well plates coated with Matrigel (BD Biosciences; Beckon, Dickinson and Company, Franklin Lakes, NJ, USA). The cells were grown in RPMI-1640 medium supplemented with 2% FBS and 2% Matrigel, and allowed to grow for 96 h at 37°C. The original medium was replaced with the fresh RPMI-1640 medium containing 2% FBS and 2% Matrigel additional with GSK-J4 (30 µM) or vehicle (DMSO) at this time point. For shRNA P16INK4A ETN-1 and EFE-184 cells, doxycycline was added when seeding. Over 100 colonies were scored for each condition. Quantitation of tumor spheres for structural integrity was performed after a 96-h culture.

A wound healing assay was used to evaluate the migration ability of ETN-1 and EFE-184 cells, as previously described (20). Cells were plated in 24-well plates at the density of 20,000/well and grown at 37°C in RPMI-1640 medium supplemented with 10% FBS until confluence. A scratch was created using sterile 200 µl pipette tips. PBS was used twice to remove cell debris and fresh RPMI-1640 medium supplemented with 2% FBS was added, with or without doxycycline. The mean width of each scratch was measured using Image-Pro Plus software 4.0 (Media Cybernetics, Inc., Rockville, MD, USA).

For H&E, tissues were first fixed in 4% paraformaldehyde solution at room temperature for 24 h. After gradient tissue dehydration (75% for 24 h; 85% for 3 h; 95% for 1 h; 100% for 1 h; and 100% for 1 h; ethanol solution at room temperature), followed by 100% xylene to remove alcohol, the tissues were embedded in paraffin. Subsequently, paraffin-embedded tissue sections (4-µm) were dewaxed with 100% xylene at room temperature for 30 min and gradient ethanol solution (100% for 10 min; 100% for 10 min; 95% for 10 min; 80% and 10 min). Subsequently, sections were immersed in 0.5% hematoxylin (cat. no. H8070; Beijing Solarbio Science & Technology Co., Ltd., Beijing, China) for 10 min followed by 5 quick dips in 0.3% acid alcohol at room temperature. The sections were then washed with running water for 60 min. Following this, 1% of eosin (cat. no. G1100; Beijing Solarbio Science & Technology Co., Ltd.) was used for 1 min at room temperature to stain the cytoplasm. IHC was performed using P16INK4A (OriGene Technologies, Inc.; cat. no. ZS-0033; 1:200), H3K27-M3 (Immunoway Biotechnology Company; cat. no. YM3338; 1:500) and H3K27-M1 (Immunoway Biotechnology Company; cat. no. YM3336; 1:500), as previously described (21). Briefly, the IHC stainning of paraffin-embedded samples was performed using a standard Biotin-Streptavidin HRP Detection method, as previously described (https://www.cellsignal.com/contents/resources-protocols/immunohistochemistry-protocol-(paraffin)/ihc-paraffin). The 4-µm sections were deparaffinized as aforementioned and antigens in the tissue were retrieved for 5 min. Following blocking endogenous peroxidase activity with 30% hydrogen peroxide formaldehyde solution at room temperature for 30 min, the sections were incubated with 20% normal goat serum to block non-specific binding sites for 30 min at room temperature. The primary antibody used was a polyclonal antibody against human P16INK4A protein. The samples were incubated overnight at 4°C in a moist chamber. Following 3 washes with PBS, the sections were incubated for 30 min at room temperature with goat anti-mouse secondary IgG (dilution 1:400; cat. no. SP-9001; OriGene Technologies, Inc.). The 3,5-diaminobenzidine detection kit (OriGene Technologies, Inc.) was used for staining. Negative controls with PBS (0.01 mol/l, pH 7.4) replacing the primary antibody were also included. Finally, the tissue sections were counterstained with 0.5% hematoxylin for 1 min at room temperature, dehydrated and mounted in resinous mountant at room temperature. Digital images were captured using a light microscope (Leica Microsystems GmbH, Wetzlar, German) at a ×400 magnification. The intensity of P16INK4A, H3K27M1 and H3K27M3 staining was quantified as the percentage of positive cells per high-power field. The formalin fixed paraffin-embedded tissues including malignant and normal endometrium were all obtained from First Affiliated Hospital of Dalian Medical University (Dalian, China). It was approved by the ethics committee of the hospital and written informed consent were also obtained from the patients.

Fresh tumor tissue and 121 FFPE tumor samples from other patients (age range, 38–77 years; mean age, 56 years) with endometrial cancer were collected from the Biobank of First Affiliated Hospital of Dalian Medical University. All were processed according to the protocol approved by the Institutional Review Board of the First Affiliated Hospital of Dalian Medical University. Written informed consent was obtained from the patients. A primary tumor tissue was obtained from a 57-year-old female during hysterectomy who had been diagnosed with endometrial serous carcinoma in dilatation and curettage, a pre-operative endometrium biopsy. The fresh tumor tissue was maintained in normal saline and transported to the laboratory in 4°C on ice. Prior to transplanting the tumor into the mice, the tumor tissue was cut into 5×5×5 mm tissue blocks. A total of 2 non obese-severe combined immunodeficiency (NOD-SCID) female mice maintained in a pathogen-free environment were used to establish the first generation of PDX mouse models. The mice, each weighing 25–30 g were provided by the Laboratory Animal Center of Dalian Medical University. The 6-week old NOD-SCID mice were raised in large plastic cages with a maximum of 6 mice/cage at 40–60% humidity and 19–23°C. The mice were maintained on a 12/12 h light/dark period with 12–15 air exchanges/h and were fed with food and water ad libitum. The mice bearing xenografts were housed under standard conditions and monitored closely. During the whole experiments, a total of 8 mice were used to establish models of different generations. The PDX mouse model was constructed as previously described (22). In brief, after the mice were anesthetized with intraperitoneal injection of Tribromoethanol (avertin 1.25%), 300 mg/kg, a 1-cm scalp incision was produced on the skin near each leg. Subsequently, tissue blocks were placed into the subcutaneous space and the incisions were closed. Tumor volume, body weight, living activity and food intake were monitored every 3 days. The timing of sacrifice depended on the following two conditions: If multiple tumors were present, the monitoring was terminated when the largest tumor volume reached 1 cm³ (with the combination of the two largest tumor size <2 cm); but if only a single tumor was observed, monitoring was terminated when the largest diameter of the tumor was near 2 cm. Therefore, sufficient tissue for the experiments could be ensured. After the mice were sacrificed with cervical dislocation and the tumor tissue was transplanted into other NOD-SCID mice, as a next generation model, the 2 mice for this experiment were the third generation. In total, 5 tumors from 8 transplanting sites (4 for each mouse) were obtained and the maximum tumor size was 1.8 cm (the single tumor appeared in that mouse). The mice were then sacrificed by decapitation and tumor specimens were used for ex vivo culture, as previously described (23). Briefly, the fresh tumor specimen was dissected into ~1-mm³ tissue blocks. The blocks were cultured on an absorbable gelatin sponge in the presence of vehicle control (DMSO) or GSK-J4 (50 µM) for 24 h. All animal procedures were conducted under the approval of the Institutional Animal Care and Use Committee of Dalian Medical University (Dalian, China) and strictly followed the guideline for tumor induction in mice and rats (http://igeam.dmu.edu.cn/jgsz1/SPFdwsyzx1/zcfg.htm).

Data are expressed as mean ± standard deviation or percentage. In analyzing quantitative results, Student's t-test and χ² test were used for two group comparison, analysis of variance and a post hoc test (Student-Newman-Keuls) were used for multiple comparisons. P<0.05 was considered to indicate a statistically significant difference. All statistical analyses were performed by GraphPad Prism 5.0 software.

To evaluate the levels of P16INK4A expression in endometrial cancer, IHC staining was performed on 121 tumor samples from patients (age range, 38–77 years, with the mean age of 56 years) with endometrial cancer at First Affiliated Hospital of Dalian Medical University. In line with the observations of previous studies (8,9,24–26), it was observed that P16INK4A was overexpressed in the majority of USC cases and was negative in the majority of UEC cases, with 75% (25/33) of USCs and 12% (11/88) of UECs being P16INK4A-positive (Fig. 1A and B, Table I). In normal endometrium from 8 individuals receiving hysterectomy due to of uterine leiomyoma, the expression of P16INK4A was negative. The representative image was depicted (Fig. 1A). Therefore, these results demonstrated that P16INK4A is overexpressed in the majority of USC samples.

As P16INK4A overexpression in tumor tissue contradicts its known role as a tumor suppressor, the aim was to investigate its function in endometrial cancer cells. Firstly, the expression of P16INK4A was tested in 6 endometrial cancer cell lines (AN3CA, EFE-184, ETN-1, Hec108, Hec-1A and Nou-1) and the expression level of P16INK4A was determined to be highest in ETN-1 and EFE-184 cells, whereas it was relatively low in AN3CA, Hec108, Hec-1A and Nou-1 cells (Fig. 1C and D).

Tumor suppressor retinoblastoma (RB) acts downstream of CDK4/6 and serves an important role in the cell cycle (4,5). RB-deficient tumors frequently express high levels of P16INK4A (4–6). A research group reported that, when P16INK4A is overexpressed, CDK4/6 expression is suppressed to prevent cancer cells from responding to CDK4/6 inhibitors (27). Thus, to investigate the function of CDK4/6 in the aforementioned endometrial cancer cell lines, their treatment response to the CDK4/6 inhibitor palbociclib were evaluated. Highly P16INK4A-expressing ETN-1 and EFE-184 cells were determined to be resistant to palbociclib. The remaining cell lines (AN3CA, Hec108, Hec-1A and Nou-1), in which P16INK4A expression was low or negative, were sensitive to the treatment, with an IC₅₀ of 0.066, 0.714, 1.741 and 1.863 µM, respectively (Fig. 1E). Therefore, ETN-1 and EFE-184 cells were selected for further experiments.

To determine whether P16INK4A acts as a driving factor in the proliferation of ETN-1 and EFE-184 cells, P16INK4A was transfected into ETN-1 and EFE-184 cells by transfecting two P16INK4A-specific shRNAs. The downregulation of P16INK4A was verified by RT-qPCR and western blot analyses (Fig. 2A and B). Following doxycycline-induced P16INK4A depletion, the growth potential of ETN-1 and EFE-184 cells was significantly decreased, as depicted in 2D colony-forming and 3D culture assays (Fig. 2C and D). A wound healing assay also demonstrated that cells with P16INK4A knocked down exhibit significantly reduced migration ability, compared with those with intact P16INK4A expression (Fig. 3). These results indicated that P16INK4A, in addition to its canonical role as a tumor suppressor, may also serve an oncogenic role in P16INK4A-positive cells, although the underlying mechanism remains elusive.

Previous studies demonstrated that the transcription of P16INK4A is primarily regulated by the methylation level of H3K27 in cervical cancer (14,28,29). To determine whether a similar mechanism is involved in the abundant expression of P16INK4A in USC, 20 cases of P16INK4A-positive USC tissue samples was investigated. Using IHC, 70% (14/20) samples in this cohort were identified as H3K27M3-negative and H3K27M1-positive (Fig. 4A and B). A similar result was also observed in the ETN-1 and EFE-184 cell lines. Western blot analysis revealed weak H3K27M3 and robust H3K27M1 expression in ETN-1 and EFE-184 cancer cells (Fig. 4C). These observations indicate that the upregulation of P16INK4A in USC may share the same mechanism as reported in cervical cancer aforementioned.

As histone KDM6B mediates H3K27 demethylation and increases the expression of P16INK4A (11,14), it was hypothesized that treatment with the KDM6B inhibitor GSK-J4 may be effectively used to target P16INK4A-positive cancer cells. As expected, GSK-J4 treatment significantly suppressed cell growth, compared with the control vehicle (Fig. 4D), which was also verified in 3D culture assays (Fig. 4E). Furthermore, following GSK-J4 treatment, P16INK4A expression was markedly inhibited with the increase of H3K27M3 and the decrease of KDM6B and H3K27M1 (Fig. 4C), supporting our hypothesis that target inhibition of H3K27 demethylation is a potential treatment for P16INK4A-positive endometrial cancer.

To further confirm the treatment effects ex vivo, a PDX mouse model of USC was then established (Fig. 5A). The model maintained a number of histopathological characteristics of the human primary tumor, including P16INK4A overexpression (Fig. 5B). In the ex vivo experiment, following GSK-J4 treatment, samples were subjected to histological examination. It was observed that, following treatment with GSK-J4, cellular integrity was markedly disrupted, without distinct USC-specific structures. However, explants treated with DMSO retained the architecture and cellularity of non-treated tumors. IHC analysis also demonstrated that the expression of P16INK4A and H3K27M1 was significantly reduced, while H3K27M3 staining was significantly increased in GSK-J4-treated explants (Fig. 5C). The results from the tumor explants were consistent with those obtained from cell lines, indicating that targeting KDM6B by GSK-J4 may be a viable treatment strategy for P16INK4A-positive endometrial cancer cases.