YY1 activates EMI2 and promotes the progression of cholangiocarcinoma through the PI3K/Akt signaling axis

Based on the gene expression data in CCA from the The Cancer Genome Atlas (TCGA) database (https://​tcga-data.​nci.​nih.​gov/​tcga), FPKM data of CCA were downloaded and analyzed for any differences, and the mRNA expression levels of EMI2 were presented using a box diagram. From the GEO database (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​) to download GSE32225 (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​query/​acc.​cgi?​acc=​GSE32225), and GSE22633 (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​query/​acc.​cgi?​acc=​GSE22633) data processing, and differential genes were screened by limMA package. The TIMER (http://​timer.​cistrome.​org/​) was used to check the quantity of EMI2 expressed in various tumors.

PROMO (http://​alggen.​lsi.​upc.​es/​cgi-bin/​promo_​v3/​promo/​promoinit.​cgi?​dirDB=​TF_​8.​3) and ChIPBase databases (http://​rna.​sysu.​edu.​cn/​chipbase/​) were used to predict the transcription factors of EMI2, then take intersection with Wayne figure, YY1 and STAT4 were found to be mating sites. JASPAR database (http://​www.​jaspar.​genereg.​net) was used to predict the possible binding sites of EMI2 and YY1.

Paired tumor and paracancerous tissue samples were collected from 8 patients with CCA who underwent surgical resection in the First Affiliated Hospital of the Bengbu Medical College from January 2020 to January 2021. The tissue samples were cryopreserved at − 80 °C. All patients signed informed consent, and the study was approved by the Ethics Committee of the First Affiliated Hospital of the Bengbu Medical College.

CCA cell lines (RBE, HuCCT1, HUCC-9810, and QBC-939) were purchased from the Shanghai Cell Bank, Chinese Academy of Sciences, and normal bile duct epithelial cell line (HIBEpiC) was purchased from the Shanghai Tongpei Biotechnology Co., Ltd. HuCCT1, RBE, HUCC-9810, QBC-939, and HIBEpiC cells were treated with RPMI-1640 medium (Gibco, USA), 10% fetal bovine serum (FBS, Gibco, USA), and 1% penicillin–streptomycin (Biosharp, China) and incubated at 37.5 °C with 5% CO₂.

To construct a stable transgenic vector using human EMI2 lentivirus, the single-stranded DNA oligonucleotide was synthesized by RNA interference target sequence technique, and the double-stranded DNA was annealed to form the double-stranded DNA. The restriction sites were 5ʹ-GGATCC(BamHI) and 3ʹ-GAATTC (EcoRI). The double-stranded DNA was inserted into the PLENR-GPH vector (Snap-gene sequencing sequences are shown in Additional file 1: Table S1) of RNAi lentivirus after digestion using multiple restriction enzymes. The interference sequence was 5ʹ-GCGTGAAATTGTTGTTCAAGA-3ʹ and the control sequence was 5ʹ-TTCTCCGAACGTGTCACGT-3ʹ (Shanghai Baioujing Biotechnology Co., Ltd.). The siRNA sequence of YY1 was 5ʹ-CCAAACAACUGGCAGAAUUTT-3ʹ (Shanghai Baioujing Biotechnology Co., Ltd.). The ligation products were then used to transform competent Escherichia coli cells, and the recombinants were detected using PCR. Lentiviral vectors were collected 72 h after transfection. Subsequently, the recombinant lentivirus (100 μL, 1 × 10⁷ units/mL) with green fluorescent protein were cultured at a density of 2 × 10⁵ cells/well for 24 h. After 72 h, stable lentiviral strains expressing shCtrl and shEMI2 were obtained. Meanwhile, YY1 and EMI2 overexpression plasmids were constructed using pEGFP-N1 vector. First, YY1 (Gene ID:7528) and EMI2 (Gene ID: 286151) sequences were obtained according to NCBI, and the sequences were synthesized with Shanghai General Biology Company. The target sequence was subcloned into pEGFP-N1 vector according to the designed enzyme cutting site. Subsequently, the recombinant plasmid was amplified into a monoclonal colony, and the plasmid was extracted using a plasmid extraction kit (Beijing Solaibao Biotechnology Co., LTD.).

Total RNA was extracted with Trizol reagent (Invitrogen, USA), and then reverse transcription was performed using the cDNA preparation kit according to the manufacturer’s instructions (Thermo, USA). The primers for all PCR primers, and their internal reference sequences were designed using Primer 5 (Additional file 1: Table S2). Then, quantitative PCR was performed using a real-time PCR machine (MX3000p, Agilent).

The transfected and control cells were digested using RIPA lysis buffer with a protease inhibitor and the total proteins were extracted. BCA kit (Beyotime Biotechnology, China) was used to detect the protein content. After quantification, the protein samples were subjected to the SDS-PAGE gel electrophoresis separation. Then, the target proteins were transferred to polyvinylidene fluoride (PVDF) membrane, which was sealed with protein-free rapid blocking solution (Beyotime Biotechnology, China) for 30 min. The primary antibody was added and the membranes were incubated at 4 °C overnight. Then use PBST to clean the membrane 3 times. They were imaged after incubation with the Tanon HighsigeCl western blotting substrate reagent for 2 h using a Biorad imaging instrument. The blots were analyzed by Image J software. The antibodies used are shown in Additional file 1: Table S3.

First, RBE and QBC9393 cells were planted in 6-well plates with a cell density of 4 * 104 according to different groups. The cells were treated according to the groups. When the growth reached 80%, the cells were digested and centrifuged, and inoculated into 96-well plates with a cell density of 5000/well. CCK-8 reagent (Beyotime Biotechnology) was added to each well after 1, 2, 3, 4, and 5 days of culture. After adding the CCK-8 reagent, the cells were incubated for 1–2 h at 37 °C. The optical density of cells within each group was detected using a microplate analyzer at 450 nm.

TMA included tissue samples from 75 patients with bile duct carcinoma (48 patients with extrahepatic bile duct adenocarcinoma, 27 patients with intrahepatic bile duct adenocarcinoma) and five cases had normal intrahepatic bile duct tissue. The microarray chips were purchased from Shanghai Kejie Biological Technology Co., Ltd. IHC antibodies used are shown in Additional file 1: Table S3. The patients’ age, gender, organ, pathology diagnosis, TNM stage, survival time, survival state and clinicopathological grade were noted (Additional file 1: Table S4). After immunohistochemical staining, two pathologists analyzed the results based on the intensity and range of staining using a previously described method [22]. The scores given by the pathologists were 0, no staining; 1, light yellow; 2, brown-yellow; and 3 brown for the standards. Positive cell percentage standards were 0–10%, 0 points; 10–25%, 1 point; 26–50%, 2 points; 51–75%, 3 points; and 76–100%, 4 points. The final result was calculated as the number of positive cells multiplied by the staining intensity.

The cells from each treatment group were inoculated in 6-well plates with an inoculation density of 1000 cells/well and a medium containing 10% FBS was added. The cultures were terminated after 2 W. After removing the supernatant, the cells were washed with PBS thrice and 4% paraformaldehyde was added for 15 min for fixing the cells. The cells were then washed with PBS twice. After staining with crystal violet for 15 min, the cells were washed with PBS twice.

The binding sites of EMI2 and YY1 were predicted using JAPAR data, and the first three significant hits were selected to construct the plasmids. All vectors were verified by sequencing them. The constructed plasmid was transfected into 293 T cells in 96-well plates. After transfection for 48 h, the medium was removed and the cells were washed once with PBS. After adding the lysate, pre-mixed LAR II was added and the 96-well plates were placed on a shaker at room temperature for 15 min. Then luciferase activity was detected.

The cells from each treatment group were inoculated in 6-well plates with a density of 30 × 10⁴ cells/well. After 24 h of culture, the cells were digested using trypsin (without EDTA). The cells were centrifuged at 1000 rpm for 10 min, collected, and washed with PBS twice. The centrifugation was continued for 10 min (at 1000 rpm), and the reagents from the APC single-dye apoptosis kit (Beyotime Biotechnology) were added successively according to the manufacturer’s instructions. The samples were tested by flow cytometry. After digestion and centrifugation, the cells from the different treatment groups were fixed with 70% ethanol at 4 °C for 12 h. Then the cells were washed with PBS once. After centrifugation, reagents from the cell cycle kit (Beyotime Biotechnology) were successively added according to the manufacturer’s instructions.

Four-week-old BALB/c female nude mice were purchased from Shanghai Lingchang Biotechnology Co., Ltd. The ethics approval to perform the animal experiments was obtained from the Bengbu Medical College. The QBC939 cells of stable transfected strains shEMI2/shCTRL were injected 1 cm into the armpit of the mice. Three weeks after the injection, the mice were sacrificed by neck removal under anesthesia, and the tumors were removed. The tumors were embedded in paraffin and immunohistochemical analysis was performed.

Statistical analyses were performed using Prism 8.0 (GraphPad Software). Data are expressed as mean ± standard deviation (\(\overline{x}\) ± s), and the differences between groups were analyzed using independent t-tests or one-way analyses of variance. P < 0.05 was considered statistically significant. Bioinformatic analysis was performed using R software (version 4.0). All experiments were repeated three times.

We combined the TCGA database with GSE3325 and GSE22633 datasets from GEO to search for differentially expressed genes, among which TCGA showed 4393 upregulated genes, GSE3325 showed 2460 upregulated genes, and GSE22633 showed 499 up-regulated genes (Additional file 2: Fig. S1A, B). Out of these, 11 upregulated genes were common to all three datasets (Fig. 1A). We found that the association between EMI2 (FBXO43) and cancer was not common in the literature. In the TIMER database, paired and unpaired analyses of the RNA sequencing files of CCA in the TCGA library showed that EMI2 was overexpressed in CCA (Fig. 1B, C). Subsequently, 8 pairs of CCA tissue samples (5 pairs of intrahepatic CCA and 3 pairs of extrahepatic CCA) were subjected to western blotting to detect the protein levels of EMI2. Compared to paracancerous tissues, EMI2 was highly expressed in the cancer tissues (Fig. 1D). IHC was used to detect the expression of EMI2 in TMA (Additional file 2: Fig. S1C), and the results showed that EMI2 was overexpressed in the cancer tissues (Fig. 1E). Results from TMA also showed that high expression of EMI2 correlated with higher N grade (P = 0.047) and M grade (P = 0.038) in CCA patients (Table 1). The above results showed that EMI2 was overexpressed in CCA.

Compared to the healthy intrahepatic bile duct epithelial HIBePic cells, EMI2 was overexpressed in bile duct cancer cell lines, and the highest expression was observed in the QBC939 and RBE cells (P < 0.05, Fig. 2A, B). Fluorescent images of lentiviral vector infection are shown in Additional file 2: Fig. S2. The silencing efficiency of EMI2 was detected using qRT-PCR and western blotting, and the difference was statistically significant (P < 0.05, Fig. 2C, D). Results of the CCK-8 assay showed that silencing of EMI2 inhibited the proliferation of QBC939 and RBE cells, especially on the 4th and 5th days (P < 0.05, Fig. 2E). Colony cloning experiment showed that the number of cell colonies in shEMI2 group was smaller (P < 0.05, Fig. 2F). The overexpression efficiency was verified in the recovery experiment (Fig. 2G, H). CCK-8 and colony cloning experiments showed that overexpression of EMI2 promoted the progression of QBC939 and RBE cells (Fig. 2I, J). These results suggested that EMI2 promoted the proliferation of CCA cells.

To determine whether EMI2 promotes the invasive ability of bile duct cancer cells, we performed the transwell assay. The results showed that compared to the shCTRL group, the invasion and migration abilities of the shEMI2 group decreased. After the overexpression of EMI2, the invasion and migration abilities of the oeEMI2 group increased (Fig. 3A, B). The wounding-healing experiment verified the changes in mobility of shCtrl, shEMI2, vector, and oeEMI2 groups, and the mobility decreased after silencing EMI2 and increased after overexpressing it (Fig. 3C). The cell cycle changes in the four groups were detected using flow cytometry, and the results showed that after silencing EMI2, the number of cells arrested at the G1 phase increased. Using the recovery experiment, it was found that overexpression of EMI2 decreased the number of cells arrested at the G1 phase (Fig. 4A). Subsequently, Annexin-V APC single staining technique was used to detect the apoptotic cells in the four groups, and the results showed that the early apoptosis rate increased after silencing EMI2, while the early apoptosis rate decreased after overexpressing EMI2 (Fig. 4B). To further predict the signaling pathway employed by EMI2, EMI2 was silenced in the RBE cell, and qRT-PCR was performed to detect the effects of silencing EMI2 on the mRNA levels of 18 star molecules, among which CDKN1B, P53, and TNF-a were overexpressed, while mTOR and MYC were downregulated. CDKN1B and mTOR were the most significantly downregulated and this result was further confirmed using western blotting (Additional file 2: Fig. S3A, B). CDKN1B and mTOR are key downstream molecules in the PI3K/Akt signaling pathway; therefore, we speculated that EMI2 may affect the PI3K/Akt signaling pathway. To further verify the relation between EMI2 and the PI3K/Akt signaling pathway, we performed western blotting to detect the expression of PI3K, p-PI3K, Akt, p-Akt, and MDM2, the key proteins of the PI3K/Akt signaling pathway, after silencing and overexpressing EMI2 separately. The results showed that the expression levels of PI3K, p-PI3K, AKT, p-AKT, and MDM2 decreased after silencing EMI2, and the result was verified using the recovery experiment (Fig. 4C, D). In summary, the results suggested that EMI2 inhibited apoptosis in the bile duct cancer cells and promoted cell cycle progression through the PI3K/Akt/mTOR signaling pathway.

To further confirm the function of EMI2, tumorigenesis experiments were performed in nude mice. First, transfected shCtrl and shEMI2 cells were subcutaneously injected to establish a nude mouse model. The fluorescence intensity of subcutaneous tumor in nude mice was detected by small animal living imager. Fluorescence measurement results showed that there was a significant difference between the two groups (P < 0.05, Fig. 5A). The subcutaneous tumors in the nude mice before dissection were imaged and the tumor growth curve was measured, which showed a statistically significant difference between the two groups (P < 0.05, Fig. 5B). After sacrificing the mice, the tumor weight was measured and the results showed that the mice in the shCtrl group (0.338 ± 0.256 g) had significantly larger tumors (P < 0.05) than those in the shEMI2 group (0.031 ± 0.035 g) (Fig. 5C). The expression levels of Ki67 in the tumor tissues of the two groups of mice were detected immunohistochemically, and the results showed that silencing EMI2 reduced the levels of Ki67 (Fig. 5D). These results suggested that silencing EMI2 reduced the growth of the bile duct cancer cells in mice, which further confirms that EMI2 promotes the progression of CCA cells.

To further study the regulation of EMI2, the upstream transcription factors of EMI2 were detected using PROMO (n = 75) and ChIPBase (n = 17) databases, and the common transcription factors found were YY1 and STAT4. TCGA data showed that YY1 was highly expressed in bile duct carcinoma (Fig. 6A). However, the overall survival of YY1 based on the TCGA database suggested that high and low YY1 was not correlated with the prognosis of CCA, and the combination of YY1 and FBXO43 was shown based on the TCGA database (Fig. 6B). Western blotting results showed that the expression of EMI2 decreased after silencing YY1, while the expression of EMI2 was restored after YY1 was overexpressed (Fig. 6C, D). The JASPAR database was used to predict the binding sites of YY1 and EMI2, the top three significant hits were selected (Additional file 2: Fig. S3C), and the sequence diagram of the binding sites was downloaded (Fig. 6E). Based on these results, we suspected that YY1 binds to the transcription initiation point of EMI2 and regulates the expression of EMI2. To verify whether YY1 binds EMI2, a dual-luciferase reporter gene assay was performed. The plasmids GV272-EMI2-WT and GV272-FBXO43-MUT were constructed (the mutation sites and base sequences are shown in Additional file 1: Table S5). The dual-luciferase reporter gene assay results indicated that compared to the negative control group, the wild-type oeEMI2 plasmid was significantly positively correlated with the expression of oeYY1, while the mutant plasmid showed no significant difference (Fig. 6F). The experiment was performed using eight different groups—NC (blank plasmid), siYY1 (siYY1 interference), siEMI2 (EMI2 interference), siYY1 + siEMI2 (co-transfected with YY1 and EMI2 interference), vector (invalid transfection sequence), oeYY1 (YY1-overexpressing plasmid), oeEMI2 (EMI2-overexpressing plasmid), and oeYY1 + oeEMI2 groups (co-transfected with YY1 + EMI2 plasmid). Cell cloning and CCK-8 assay were performed to detect the effects on the proliferation of CCA cells. The results showed that compared to the NC group, the bile duct cancer cells in the siYY1, siEMI2, and siYY1 + siEMI2 groups decreased. Out of these, the effect of siYY1 + siEMI2 was the most significant. The results were confirmed using an overexpression experiment (Fig. 6G–I). Meanwhile, we also further verified HUCCT1 and HIBEpiC cells with low EMI2 expression. Western blot, CCK8 and cloning experiments further confirmed the above hypothesis (Additional file 2: Fig. S4). These results suggested that YY1 activates EMI2, thereby, promoting the progression of CCA cells.

To further explore the combined regulation of metastasis, cell cycle progression, and apoptosis in bile duct carcinoma by YY1 and EMI2, transwell assay was performed to detect the migration and invasion of cells in each group. The results suggested that silencing YY1 and EMI2 together reduced the ability of the CCA cells to metastasize, and the effect of the siYY1 + siEMI2 group was the highest (P < 0.05). This result was confirmed using the overexpression experiment (Fig. 7A, B). We further verified it using the scratch assay, and the results showed that silencing YY1 and EMI2 inhibited the migration of the CCA, and the co-transfection YY1 + EMI2 showed the most significant effect. The overexpression of YY1 and EMI2 together promoted the migration of CCA cells (Fig. 7C, D). Flow cytometry was used to detect the effects of silencing and overexpressing YY1 and EMI2 on apoptosis in the bile duct cancer cells, and the results showed that YY1 activated EMI2, thus, inhibiting apoptosis (Fig. 7E). The cell cycle changes in each group were detected using flow cytometry, and the results indicated that compared to the NC group, the number of CCA cells in the G1 phase increased after YY1 and EMI2 were silenced, while the number of CCA cells in the S phase increased after YY1 and EMI2 were overexpressed (Fig. 7F). These results suggested that YY1 activates EMI2, promotes metastasis, and inhibits apoptosis in CCA cells. The silencing of YY1 also inhibited the expression of EMI2, arresting the CCA cells in the G1 phase.