The human endometrial cancer HEC-1-A cell line was purchased from Shanghai GeneChem Co., Ltd. (Shanghai, China). The cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Corning Incorporated, Corning, NY, USA) supplemented with 10% fetal bovine serum (VS500T; Ausbian; Vian-Saga Co., Ltd, Shanghai, China) at 37°C in 5% CO₂. For transfections, cells were seeded onto a 6-well plate 24 h prior to the experiment at a cell density of 3–5×10⁴/ml in DMEM in an atmosphere of 5% CO₂ at 37°C. Each experiment was performed in triplicate.

ARID1A mutations in HEC-1-A cells were verified by Sanger sequencing using a 3730XL DNA analyzer (Thermo Fisher Scientific, Inc., Waltham, MA, USA) and specific primers targeting exons 1212, 3969, 5281 and 5503 of the ARID1A NM_006015 CDS region were used. The experimental workflow of the Sanger sequencing included the following 6 steps: Isolating the DNA, performing a polymerase chain reaction (PCR; incubate at 95°C for 10 min, denature at 96°C for 3 sec, anneal at 58°C for 15 sec, extend at 72°C for 30 sec and final extension at 72°C for 10 min for 35 cycles with DNA polymerase (BigDYe3.1; Applied Biosystems; Thermo Fisher Scientific, Inc.), then hold at 4°C), performing a sequencing reaction, purifying the sequencing reaction, performing capillary electrophoresis and analyzing the data using Applied Biosystems sequence scanner software v2.0 (Thermo Fisher Scientific, Inc.). PCR forward and reverse primers were designed according to the ARID1A gene fragment downloaded from the NCBI website (https://www.ncbi.nlm.nih.gov/nuccore/NC_000001.11?report=genbank&from=26696031&to=26782110) and (https://www.ncbi.nlm.nih.gov/tools/primer-blast/index.cgi?LINK_LOC=BlastHomeAd). Gene sequencing of HEC-1-A was performed using gene-specific primers. The following PCR primers designed for exon 1212 were amplified at 479 bp: GSPE91935-primer 1, GTTACTAGGTTGGTCTCATTGCTC and GSPE91935-primer 2, AGCCAACAGGTCTACATTCCTGTC. The following PCR primers designed for exon 3969 were amplified at 353 bp: GSPE91935-primer 3, TGAAGCTATAGTGGGCTCAATCTG and GSPE91935-primer 4, CTGTTGATACATTGTAGTCTGCTG. The following PCR primers designed for exons 5281 and 5503 were amplified at 573 bp: GSPE91935-primer 5, TCCTTGTAGAATATTTCCGACGATG and GSPE91935-primer 6, GTTTTTCTGGAGGTCCATCAGGTG.

Three short hairpin RNA (shRNA) sequences were designed by Shanghai GeneChem Co., Ltd. The shRNA expression cassettes were designed according to the small interfering (siRNA) sequences. The shRNA vectors were generated by inserting annealed oligo sequences into the digested GV248 vectors (Shanghai GeneChem Co., Ltd.) between the Phu6 and Pubi sites. HEC-1-A cells were infected with lentiviral vector GV248 or ARID1A-specific shRNA (Shanghai GeneChem Co., Ltd.) particles with a GV248 plasmid (2 µl at 1×10⁹ TU/ml, 4 µl at 5×10⁹ TU/ml and 2 µl at 1×10⁹ TU/ml) in 6-well plates in the presence of 6 mg/ml polybrene (Genomeditech Co., Ltd., Shanghai, China) and were then treated with 2 mg/ml puromycin (Clontech Laboratories, Inc., Mountainview, CA, USA) to generate stable clones. An empty vector was used as a control. ARID1A gene expression and protein levels were confirmed by reverse transcription-quantitative polymerase chain reaction (RT-qPCR).

To measure the abundance of ARID1A mRNA, primers were selected and tested at different primer concentrations. Total RNA was extracted from EC cells using the TRIzol (Thermo Fisher Scientific, Inc.) method. A reverse transcription kit (Promega Corporation, Madision, WI, USA) was used to reverse transcribe the RNA into cDNA at 42°C for 1 h. Subsequently, qPCR was performed using SYBR-Green (DRR041B; Takara Biotechnology Co., Ltd., Dalian, China) and fluorescence microscopy (model no. IX71; Olympus Corporation, Tokyo, Japan), The ARID1A siRNA sequences were as follows: 5′-GTCCCTCAAGTCTGGTCTCC-3′ and reverse, 5′-GATCTCAATCAGGCATCGTC-3′. The thermocycling conditions were as follows: Degeneration at 95°C for 30 sec, annealing at 60°C for 30 sec and extension at 72°C for 1 min, for 40 cycles. Relative gene expression levels were calculated using the 2^−ΔΔcq method (12) and were normalized to the expression of GAPDH (forward, 5′-TGACTTCAACAGCGACACCCA-3′ and reverse, 5′-CACCCTGTTGCTGTAGCCAAA).

The transfected cells were washed twice with phosphate-buffered saline, centrifuged at 4°C, 1,000 × g for 5 min and lysed on ice in radio immunoprecipitation assay (RIPA) lysis buffer (WB-0071; Thermo Fisher Scientific, Inc.) containing protease inhibitors for 30 min. Subsequently, the protein was quantified using the bicinchoninic acid method with BCA protein assay kit (P0010S; Beyotime Institute of Biotechnology, Haimen, China). The protein (20 µg per lane) was subjected to SDS-PAGE electrophoresis to separate the protein using a 5% stacking gel and 10% separating gel, prior to being transferred onto polyvinylidene fluoride membranes. Membranes were blocked by Tris-buffered saline containing 0.05% Tween-20 (TBST) and 5% skimmed milk for 1 h at room temperature and were incubated overnight with one of the following primary antibodies for 30 min at 4°C: Anti-rabbit eukaryotic translation initiation factor 4E (EIF4E; cat. no. 2067; dilution, 1:500; Cell Signaling Technology, Inc., Danvers, MA, USA), anti-FOS (cat. no. 8333; dilution, 1:500; Cell Signaling Technology, Inc.), anti-rabbit insulin receptor substrate 1 (IRS1; cat. no. ab52167; dilution, 1:500; Abcam, Cambridge, MA, USA), anti-rabbit phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit β (PIK3CB; cat. no. ab32569; dilution, 1:500; Abcam), anti-rabbit forkhead box protein O1 (FOXO1; cat. no. 2880; dilution, 1:500; Cell Signaling Technology, Inc.) and anti-rabbit insulin-like growth factor 1 receptor (IGF1R; cat. no. ab131476; dilution, 1:500; Abcam). Following washing, the horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (cat. no. sc-2004; dilution, 1:2,000; Santa Cruz Biotechnology, Inc., Dallas, TX, USA) was added for 2 h at room temperature. The anti-mouse GAPDH (cat. no. sc-32233; dilution, 1:2,000, Santa Cruz Biotechnology, Inc.) was used as a sample loading control. Pierce™ ECL Western Blotting Substrate kit (Thermo Fisher Scientific, Inc.) was used for enhanced chemiluminescence, and the membranes were then placed on X-ray film (Carestream, Health, Inc., Rochester, NY, USA). The X-ray film was placed in X-ray film imaging powder (P61-04-1; Guanlong, Shanghai, China) for 1 min and then in X-ray film fixing powder (Guanlong, Shanghai, China) for 2 min. The data were normalized to GAPDH expression by densitometry. Western blot automated quantitative analysis system Compass software (version WS-2471; Protein Simple, San Jose, CA, USA) was used to analyze the gradation test results. Quantitative results, including molecular weight, signal intensity (area), % area and signal-to-noise ratio were obtained for each immunodetected protein.

Human Gene Chip Prime View (cat. no. 901838; Affymetrix; Thermo Fisher Scientific, Inc.) was used to simultaneously investigate alternations in the activities of canonical signaling pathways in response to ARID1A depletion. The relative activity of each pathway was determined and normalized to that of untreated controls. Experiments were performed in triplicate and the values were calculated as the mean ± standard error of the mean. The sample quality was required to meet the following standards: NanoDrop 2000 (Thermo Fisher Scientific, Inc.), 1.7<A260/A280<2.2; Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA), RNA integrity number ≥7.0; and 28S/18S>0.7. Expression data were normalized through quantile normalization. All gene level files were imported into IPA^® software (v39473382; Qiagen GmbH, Hilden, Germany; http://www.ingenuity.com) for further analysis.

The distributions of the intensities of 6 samples and the similarities between ARID1A-knockdown (KD) and control (NC) groups were examined by three-dimensional principal component analysis (Fig. 1A) and Pearson's correlation of the signal value (Pearson's correlation coefficient >0.95; Fig. 1B). Signal value distribution (Fig. 1C) and relative signal box plot graphs (Fig. 1D) demonstrated the expression values of all microarray probe distribution statistics and all samples in the present study were reproducible. Scatterplot graphs (Fig. 1E) and a dendrogram (Fig. 1F) demonstrated differences in gene expression of all microarray probe distribution statistics between the KD and the NC group. Differentially expressed genes between 6 samples were identified through fold change filtering.

Statistical analysis was performed using SPSS 22.0 (IBM Corp., Armonk, NY, USA). Comparisons between two groups were performed using Fisher's exact test, Student's t-test and one-way analysis of variance followed by Tukey's honest significant difference test. P<0.05 or P<0.01 was considered to indicate a statistically significant difference. The data are presented as the mean ± standard deviation.

We first used exon capture sequencing to determine the ARID1A mutation frequencies in the HEC-1-A cell line and observed that all the ARID1A mutations were heterozygous: i) Missense mutations, transversion of A to T at codon 1212 on exon 2 and transversion of G to T at codon 5281 on exon 20; ii) synonymous mutations, transversion of C to A at codon 3969 on exon 16; or iii) nonsense mutations, transversion of C to T at codon 5503 on exon 20 (Fig. 2A).

To explore the association of mutated ARID1A with the behavior of EC cell lines, 3 pairs of ARID1A gene shRNA interference fragments; LV-ARID1A-RNA interference (i)-24485-1 (RNAi-1), LV-ARID1A-RNAi-24486-1 (RNAi-2) and LV-ARID1A-RNAi-24487-1 (RNAi-3), one negative control (NC) pair and one blank control (MOCK, empty vector-transfected cells) pair were designed and transfected into HEC-1-A cells. RT-qPCR was used to detect the expression of ARID1A mRNA transfected with interference fragments. ARID1A mRNA expression in HEC-1-A cells following transfection with RNAi-1 (P=0.0042), RNAi-2 (P=0.0006) and RNAi-3 (P=0.0428) was significantly decreased compared with that in the NC group. Three shRNA vectors suppressed ARID1A mRNA expression by 78.8, 55.0 and 55.6%, respectively, compared with that in the NC group. RNAi-1, with an increased transfection efficiency (78.8%) compared with the NC group, was selected for subsequent experiments as the KD group (Fig. 2B). Since the ARID1A gene shRNA interference fragment was designed based on codon 6795–6813, it did not fall on the mutation position as mentioned earlier. Therefore, mutated ARID1A may have been depleted by shARID1A knockdown plasmids in the KD group.

A total of 408 canonical signaling pathways, 39 disease states and functions, 25 networks, 1,085 molecular targets, 662 upstream and 512 downstream genes were identified in ARID1A-depleted HEC-1-A cells compared with control cells. A Z-score/fold change ≥1.0 represented an activated signaling pathway and disease functions, while a Z-score/fold change ≤-1.0 represented an inhibited signaling pathway and disease functions. Furthermore, a signal Z-score/fold change ≥1.5 or ≤-1.5 and P<0.05 indicated a significant change in signaling pathways and disease functions between the KD and NC groups.

To investigate specific signaling pathways and functional gene groups, IPA tools were used to identify 143 differentially expressed canonical pathways by defining the enrichment P-value of the pathway using Fisher's exact test ≤0.05, among which 17 pathways were significantly downregulated (Z-score ≤-1.5; P<0.05) and 13 were significantly upregulated (Z-score ≥1.0; P<0.05) in the KD group compared with those in the NC group. The extracellular signal-regulated kinase (ERK)/mitogen-activated protein kinase (MAPK; Z-score=−2.353<-2.0; log|P-value|=6) and IGF-1 (Z-score=−2.138 <-2.0; log|P-value|=4.06) were the signaling pathways were the 2 most significantly downregulated canonical pathways. Signaling pathways associated with ARID1A-depleted HEC-1-A cells were sequenced by their -Log(P-value). Genes associated with the pathways demonstrating differential expression were highlighted in different colors depending on their -Log(P-value) (Fig. 3A). The ratios presented in Table I represent the ratio of differentially expressed genes from a pathway to the total number of genes of that particular pathway. The majority of these pathways were associated with cancer development, proliferation and apoptosis. The volcano graph (Fig. 3B) demonstrates the distribution of differentially expressed genes by fold change between the KD group and the NC group. Genes presented with a red color indicate P<0.05 and fold change ≥1.5. The heat map demonstrates the differences in 32 genes directly associated with the ERK/MAPK and IGF-1 signaling pathways from the normalized data of 6 experimental groups (Fig. 3C). Microarray analysis indicated that there were 24 significantly downregulated (fold change ≤-1.5; P<0.05) and 8 significantly upregulated genes (fold change ≥1.5; P<0.05) in the KD group compared with the NC group (Fig. 3D).

The network diagram of gene interaction analyzed by IPA revealed interactions between ARID1A and molecules associated with the MAPK/ERK and IGF-1 signaling pathways; EIF4E, IGF1R, IRS1 and PIK3CB were downregulated, while FOS and FOXO1 were upregulated (Fig. 4A). To confirm these results, western blot analysis was used to investigate alternations in the activities of the MAPK/ERK and IGF-1 signaling pathways upon ARID1A depletion in HEC-1-A cells. Multiple cellular processes were potentially affected. IRS (Z-score=−1.546), PIK3CB (Z-score=−1.673) and IGF1R (Z-score=−1.952) were significantly downregulated in shARID1A-HEC-1-A cells compared with that in control cells. However, ARID1A knockdown significantly elevated the level of FOXO1 (Z-score=1.588) and there were no significant differences in EIF4E (Z-score=−1.531) or FOS (Z-score=1.975) expression in the shARID1A-HEC-1-A cells compared with that in the control cells (Fig. 4B and C).

To evaluate the association of disease states and functions with biological networks, 39 affected disease states and functions were identified. A heat map was produced demonstrating the association between activated or inhibited disease states or functions and different Z-scores in the KD group compared with those of the control group. A total of 9 disease states and functions were significantly upregulated, including carcinoma tumorigenesis, mitosis of tumor cell lines, cancer, adenocarcinoma, tumorigenesis of malignant tumors, mitochondria density, tumor necrosis, tumor cell death and cancer cell death; 5 functions were significantly downregulated, namely cell proliferation, cell survival, uptake of D-glucose, cell viability and tumor growth. The affected disease states and functions graph was constructed based on the -Log (P-value), and genes within the disease states and functions that exhibited differential expression were highlighted in different colors depending on their -Log (P-value) (Fig. 4D; Table II).