Knockdown of PSMC2 contributes to suppression of cholangiocarcinoma development by regulating CDK1

PSMC2 was upregulated in CCA tissues and expressed in CCA cells

IHC was used to map the expression of PSMC2 in tissues from patients with CCA in comparison with tissues from normal controls. These analyses clearly demonstrated that the expression levels of PSMC2 were upregulated in CCA (Figure 1A). Statistical analyses of 74 CCA tissues and 5 normal tissues showed that the expression levels of PSMC2 were significantly higher in the CCA tissues (P < 0.001, Table 1). We also performed correlation analysis to investigate the relationship between the expression levels of PSMC2 and specific clinical characteristics. These analyses showed the expression levels of PSMC2 were significantly upregulated in patients suffering from advanced tumor grades (P < 0.05, Figure 1A and Table 2). These findings were further validated by Spearman’s rank correlation analysis (Supplementary Table 4). Our analyses demonstrated clear that more serious cases of CCA exhibited higher levels of PSMC2, thus suggesting that PSMC2 may be associated with the progression of CCA. We also determined the endogenous expression of PSMC2 in several CCA cell lines, including HUCCT1, QBC939, RBE, and HCCC-9810. Our analyses showed that QBC939 and HCCC-9810 cells exhibited high expression levels of PSMC2 expression and were therefore selected to construct a cell model of PSMC2 knockdown for subsequent research (Figure 1B).

Figure 1. PSMC2 was upregulated in CCA tissues and abundantly expressed in CCA cells. (A) The expression level of PSMC2 was detected by IHC analysis in CCA tissues and normal tissues (scale bar = 50 μm). (B) The endogenous expression of PSMC2 in CCA cell lines including HUCCT1, QBC939, RBE and HCCC-9810 was evaluated by qPCR. Data was shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

The depletion of PSMC2 inhibited the development of CCA in vitro

We constructed a cell model of PSMC2 deficiency by transfecting a lentivirus that was specifically designed to silence PSMC2. Our intention was to use this model to investigate CCA and the mechanisms that underlie the progression of this disease. Over 80% of the cells tested emitted fluorescence, thus indicating that transfection had been successful (Supplementary Figure 1). qPCR and western blotting were also used to demonstrate that the expression levels of PSMC2 had been downregulated at the mRNA level (P < 0.001 for HCCC-9810 cells, P < 0.01 for QBC939 cells) and at the protein level (Figure 2A). These data confirmed that the cellular model of PSMC2 knockdown had been successfully created in both HCCC-9810 and QBC939 cells. Furthermore, MTT assays demonstrated that cells in which PSMC2 had been depleted (shPSMC2) grew significantly more slowly than those without PSMC2 depletion (shCtrl) (P < 0.001, Figure 2B).

Figure 2. PSMC2 knockdown inhibited CCA development in vitro. (A) Cell models with or without PSMC2 knockdown were constructed by transfecting shPSMC2 or shCtrl. The knockdown efficiency of PSMC2 in HCCC-9810 and QBC939 cells was assessed by qPCR and western blotting. (B) MTT assay was employed to show the effects of PSMC2 on cell proliferation of HCCC-9810 and QBC939 cells. (C, D) Flow cytometry was performed to detect cell apoptosis (C) and cell cycle distribution (D) of HCCC-9810 and QBC939 cells with or without PSMC2 knockdown. (E) Human Apoptosis Antibody Array was utilized to analyze the regulatory ability of PSMC2 on expression of apoptosis-related proteins in HCCC-9810 cells. (F) Wound-healing assay was performed to distinguish cell migration of HCCC-9810 and QBC939 cells with or without PSMC2 knockdown. (G) WB was used to detect the expression of EMT related proteins in CCA cell models. The representative images were selected from at least 3 independent experiments. Data was shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Flow cytometry demonstrated that cell proliferation was influenced by cell apoptosis in CCA cells, either with or without the knockdown of PSMC2. Cells that were deficient in PSMC2 showed a significantly greater proportion of apoptotic cells than the shCtrl group (P < 0.001, Figure 2C). Cell cycle analyses further showed that the downregulation of PSMC2 in HCCC-9810 and QBC939 cells was associated with significant cell cycle arrest in the G2 phase (P < 0.001, Figure 2D). Next, we used an antibody array to investigate the effect of PSMC2 on apoptosis-related proteins. This array was designed to help us identify the specific mechanisms underlying the ability of PSMC2 to regulate cellular apoptosis. The antibody array demonstrated that PSMC2 knockdown led to an upregulation in the expression levels of BID and caspase 3, and a downregulation in the levels of Bcl-2, CD40, cIAP-2, HSP27, IGF-II, Survivin, and sTNF-R1 (Figure 2E). In addition, wound-healing assays were used to investigate the mobility of HCCC-9810 and QBC939 cells in control and shPSMC2 groups. We observed a significant reduction of cell mobility in the shPSMC2 groups for both cell lines (P < 0.001, Figure 2F). Data further suggested that these effects were caused by the upregulation of E-cadherin and the downregulation of both N-cadherin and Vimentin (Figure 2G). Collectively, our findings suggest that PSMC2 may play a key functional role in the development of CCA by regulating cellular migration, colony formation, and apoptosis.

PSMC2 regulates CCA by targeting CDK1

Prior to this study, we knew very little about how PSMC2 was associated with the regulation of CCA. In the present study, we used 3 v 3 RNA-seq to identify differentially expressed genes (DEGs) between HCCC-9810 cells in the shPSMC2 group and the shCtrl group. We identified 1106 DEGs that were upregulated, and 1261 DEGs that were downregulated, in the shPSMC2 cells when compared with the shCtrl cells (Supplementary Figure 2A and Figure 3A), when using a |Fold Change| ≥ 1.3 and an FDR < 0.05 (the P value after Benjamini-Hochberg analysis) as threshold criteria. Next, we used IPA analysis to investigate how the changes in the 2367 DEGs might influence canonical signaling pathways or IPA disease and function (Supplementary Figure 2B, 2C). Based on IPA analysis, and the bioinformatic analysis of a PSMC2 interaction network, we identified several DEGs that showed maximal levels of fold-change; these findings were then verified by qPCR and western blotting in HCCC-9810 cells. The most promising candidate target for PSMC2 was CDK1, a core component in the pathway that regulates cyclin and the cell cycle (Supplementary Figures 2B, 3) (Figure 3B–3D). CDK1 expression levels exhibited a similar profile to those of PSMC2 that expression levels were significantly higher in CCA tissues than in normal control tissues (Figure 3E). qPCR also confirmed the significant upregulation of CDK1 levels in CCA cells (Figure 3F). Collectively, these findings demonstrated that CDK1 may represent a potential target of PSMC2 in the regulation of CCA, although this now needs to be validated in vitro.

Figure 3. The exploration and verification of downstream underlying PSMC2 induced regulation of CCA. (A) A PrimeView Human Gene Expression Array was performed to identify the differentially expressed genes (DEGs) between shPSMC2 and shCtrl groups of HCCC-9810 cells. (B) A PSMC2-induced interaction network was established based on IPA analysis. qPCR (C) and western blotting (D) were used to detect the expression of several selected DEGs in HCCC-9810 cells with or without PSMC2 knockdown. (E) The expression of CDK1 in CCA tissues and normal tissues was evaluated by IHC analysis (scale bar = 50 μm in 200 magnification, scale bar = 20 μm in 400 magnification). (F) The mRNA expression of PSMC2 in CCA cell lines was detected by qPCR. The representative images were selected from at least 3 independent experiments. Data was shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

The knockdown of PSMC2 aggravated the inhibition of CCA progression induced by the depletion of CDK1

Next, we transfected HCCC-9810 cells with shCDK1, or both shPSMC2 and shCDK1, to investigate their combined action in terms of CCA. First, we evaluated the transfection efficiency as described earlier. In brief, we identified the most effective shRNA for silencing CDK1 by performing qPCR (Supplementary Figure 4). As shown in Figure 4A, 4B, it was clear that there was a relationship between the expression levels of PSMC2 and CDK1. The depletion of PSMC2 downregulated the levels of CDK1 while a deficiency in CDK1 led to a downregulation of PSMC2. Our analyses demonstrated that the knockdown of CDK1 exhibited significant effects on cell proliferation, colony formation, and cellular apoptosis (P < 0.001, Figure 4C–4E), as also seen with PSMC2. Cell migration may be the main mechanism underlying tumor metastasis and was also investigated. As shown in Figure 4F, 4G, a reduction in CDK1 levels significantly inhibited the cell migration ability of HCCC-9810 cells, as determined by both wound-healing and Transwell assays. We also demonstrated that the additional knockdown of PSMC2 in CDK1 knockdown cells had deleterious effects on a range of cellular functions, including colony formation, cell proliferation, cell apoptosis, and migration (P < 0.001, Figure 4C–4G). The combination of CDK1 overexpression and PSMC2 knockdown had no significant effects on cell apoptosis and cell migration in CCA cells (Supplementary Figure 5). Collectively, these results indicated the crucial role of the PSMC2/CDK1 axis in the development of CCA.

Figure 4. Knockdown of PSMC2 deepens the effects on CCA cells by CDK1 knockdown. (A, B) The expression of PSMC2 and CDK1 in HCCC-9810 cells transfected with shCtrl, shCDK1 and simultaneous shPSMC2 and shCDK1 were detected by qPCR (A) and western blotting (B). Cell models were subjected to the detection of cell proliferation by Celigo cell counting assay (C), colony formation (D), cell apoptosis (E), cell migration by wound-healing assay (F) and Transwell assay (G). The representative images were selected from at least 3 independent experiments. Data was shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Identifying the mechanisms underlying the suppressive effect of PSMC2 on tumor growth in vivo

We successfully constructed mouse models by injecting HUCCT1 cells (with or without PSMC2 knockdown). Subsequently, in vivo bioluminescence imaging demonstrated that the total bioluminescence intensity was significantly weaker in the shPSMC2 group (P < 0.001, Figure 5A) and that the tumor burden was significantly smaller. In addition, the solid tumors in the shPSMC2 group had a reduced volume and smaller weight, thus highlighting the suppressive effects of PSMC2 silencing on tumor growth (P < 0.001, Figure 5B–5D). Tumors removed from mice in the shPSMC2 group, also had a lower Ki67 index and exhibited lower proliferative activity (Figure 5E).

Figure 5. PSMC2 knockdown inhibited CCA development in vivo. (A) In vivo imaging was performed to evaluate the tumor burden in mice of shPSMC2 and shCtrl groups at day 67 post tumor-inoculation. The bioluminescence intensity was scanned and used as a representation of tumor burden in mice of shPSMC2 and shCtrl groups. (B) 24 days post injection of HCCC-9810 cells with or without PSMC2 knockdown, the volume of tumors formed in mice was measured and calculated at indicated time intervals. (C, D) Mice were sacrificed at day 67 post injection, and the tumors were removed for collecting photos (C) and weighing (D). (E) After removing the tumors, the Ki67 index was evaluated by IHC staining as a representative of tumor growth activity (scale bar = 100 μm in 100 magnification, scale bar = 50 μm in 200 magnification). Data was shown as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Immunohistochemistry (IHC)

Paraffin-embedded cancer/normal tissues microarray chip (Xi’an Alenabio Co., Ltd, China) was applied for immunohistochemistry staining to identify PSMC2 and CDK1. Antigen of the tissue slide chip was retrieved, next slides were blocked by 3% H₂O₂ and appropriate serum, then PSMC2 and CDK1 antibodies were incubated with the slides overnight at 4° C. After washing, appropriate second antibody was added and cultured at room temperature for 2 h. After staining, tissues pictures were collected and viewed using ImageScope software and CaseViewer. Antibodies used were detailed in Supplementary Table 1.

Cell lines and lentiviral vector cell transfection

HCCC-9810, HUCCT1, RBE, and QBC939 cells lines (BeNa Technology, China) were cultured in an incubator 37° C 5% CO₂ as cell protocols described. shRNAs of human PSMC2 and CDK1 were designed and lentivirus was produced in Shanghai Bioscienceres, Co., Ltd. and Supplementary Table 2 showed the detailed RNA sequences. Cells were transfected with lentivirus and stable expression cells were collected after three days and used for our study.

qRT-PCR

RNA was isolated using TRIzol (Sigma) and the quality was evaluated by Thermo Nanodrop 2000/2000C spectrophotometer. M-MLV RT kit (Promega) was used to obtain cDNA and SYBR Green Master Mix Kit (Vazyme) was applied for qPCR to value target genes expression levels using 2^−ΔΔCt method. Primer sequences were showed in Supplementary Table 3 with GAPDH served as housekeeping gene.

Western blotting (WB) and human apoptosis antibody array

Total protein was extracted using ice-cold RIPA buffer (Millipore) for further electrophoresis. In every WB analysis, equal amount protein was separated by SDS-polyacrylamide gel electrophoresis (PAGE) (10%, Invitrogen) and electro-transferred onto Poly-(vinylidene fluoride) membranes. Antibodies (Supplementary Table 1) were used to identify the target protein with corresponding second antibodies. For human apoptosis protein detection, the array membrane was blocked and then incubated with protein samples overnight at 4° C and proceeded to incubate with HRP linked streptavidin for another 1 h. Signals were captured by ECL-plus^TM WB system (Amersham) and protein gray images were analyzed by ImageJ.

MTT and Celigo assay

Transfected HUCCT1 cells were seeded into a plate with 96 wells (2,000 cells per well). Cell viability and growth was measured using MTT solution (GenView). CD490 value were obtained to draw the cell growth curve (five days) using a microplate reader (Tecan). Live cells in five consecutive days were counted by Celigo image cytometer (Nexcelom Bioscience).

Fluorescence activated cell sorting assay

Lentivirus transfected cells were collected for apoptosis and cycle assay detecting. Annexin V-APC (eBioscience) staining was used for cell apoptosis, and PI (BD Biosciences) staining was used for cell cycle. Cells were measured using FACS Calibur (BD Biosciences).

Scratch assay

Transfected cells were grown to 90% confluence, and scratches across monolayer were made by a VP scientific wounding replicator. The migration distance was recorded for consecutive two days post scratching using ImageJ software.

Colony formation assay

Transfected HUCCT1 cells were cultured in a plate with six well in triplicate. After ten days growing, 4% paraformaldehyde was added into each well for cells fixing, next, clones were washed and stained with Giemsa and fluorescence pictures was captured under an Olympus microscope. The colonies number was counted.

RNA-sequencing

Following the manufacturer's instruction manual, total RNA in transfected HCCC-9810 cells was extracted and RNA concentration and quality was determined. Gene expression profile detection was accomplished using Affymetrix human GeneChip PrimeView gene expression array. Welch t-test and Benjamini-Hochberg FDR were used to assess raw data statistical significance (|Fold Change| ≥ 1.3 and FDR < 0.05 as significant). Next, significant data was used for Ingenuity Pathway Analysis (IPA) (Qiagen) executing (|Z - score| > 2 as significant).

Xenograft mouse model

Female BALB/c nude mice (Beijing Vital River Laboratory Animal Technology Co., Ltd (Beijing, China)) were grown for xenograft mouse model experiments. 20 4-week-old mice were randomly divided into shCtrl group and shPSMC2 group and were injected subcutaneously with LV-shCtrl or LV-shPSMC2 transfected HUCCT1 cells, then housed in a 12/12 day/night cycle environment at 24° C with 60% humidity. The weight of mouse and the size of tumors were monitored. All mice were anesthetized used 0.7% Pentobarbital Sodium (10 μL/g) by intraperitoneal injection 67 days after the injection. The Berthold Technologies living imaging system was used to collect the in vivo bioluminescence images. Then all mice were sacrificed and tumors were harvested and the volumes were measured using a Vernier caliper. Mice tumor tissues were stained by Ki-67 staining assay. All animal studies were approved by Ethics committee of Hunan Provincial People’s Hospital.

Statistical analyses

Data in triplicate are expressed as mean ± SD. Statistical significance (calculated by SPSS 22.0 (IBM) and Graphpad Software (GraphPad Prism 6.01)) was assessed at P < 0.05 and analyzed using Student’s t test or One-way ANOVA. Sign test was used to assess the difference of PSMC2 expression in cholangiocarcinoma tissues and normal tissues revealed in immunohistochemistry assay. Mann-Whitney U analysis and Spearman rank correlation analysis was used to analyze the relationships between PSMC2 expression and tumor characteristics.