Pseudogene UBE2MP1 derived transcript enhances in vitro cell proliferation and apoptosis resistance of hepatocellular carcinoma cells through miR-145-5p/RGS3 axis

Cell lines

Three typical HCC cell lines (Huh7, HepG2, and Hep3B) were recruited, and the normal human hepatic cell LO2 was used as control (Shanghai Institutes for Biological Sciences, Chinese Academy of Science, Shanghai, China)). All the cell lines were cultured by RPMI 1640, supplemented with 10% heat-inactivated fetal bovine serum (FBS), incubated at 37° C environment temperature, with 100 ug/ml streptomycin, and 100U/ml Penicillin in a humidified cell, with an atmosphere of 5% CO₂. Specifically, for the transfected cells, a medium mixed with G418 (Santa Cruz Biotechnology, Inc; 400 μg/ml) was used for selection.

Clinicopathological specimens

Nighty-three paired specimens including the tumor tissue and the adjacent non-cancerous liver tissues were collected from the patients diagnosed and conducted radical resection with no preoperative treatment, during 2016–2019, at the Department of General Surgery, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine. The corresponding clinicopathologic parameters of the patients were obtained including gender, age, tumor size, number of lesions, grades et al. The informed consent was obtained, and the study was approved by the Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University School of Medicine.

Datasets preparation

The gene expression data for 110 normal liver samples from the GTEx and TCGA along with clinical information for 369 liver tumors and 50 normal samples from the UCSC Xena database were intensively explored by using the random walk-based multi-graphic (RWMG) model algorithm developed by our team [13]. In brief, the RWMG model was developed from the biophysical interaction networks and the co-expression profiles within a single analytical framework, by which it integrates sophisticated biological connections among lncRNA targets, including the transcription factors (TF), microRNAs (miRNAs), and the alternative splice factors. This model presents flexible and scalable characteristics in ranking a subset of lncRNAs based on the literature survey. The RWMG model provides more accurate results than the previous network-based algorithms defined as the ‘shortest path’ or the conventional random walk algorithms and can avoid the ‘noise’ from the dimensional heterogenicity of the data. And the related and comprehensive information of the involved pseudogene-derived lncRNAs was described in the mentioned LCLE online. The starBase datasets (https://starbase.sysu.edu.cn/) and the dreamBase (https://rna.sysu.edu.cn/dreamBase/) datasets were used for providing supplementary information on the expression and relationship of the candidate genes in this study.

RT-qPCR assay and immunohistochemistry assay

For RNA isolation, from either tissues or cells, conducted according to the instruction of the TRIzol reagent (Invitrogen, MA, USA). The first-strand cDNA was synthesized via High-Capacity cDNA Reverse Transcription Kit (ABI, NY, USA). All the primers were synthesized (Jike Biotech Company, Shanghai, China) (Supplementary Table 1). Real-time quantitative polymerase chain reaction (RT-qPCR) was operated following the TaqMan Gene Expression Assays protocol (ABI, NY, USA). The relative quantification of RNA in cell lines was normalized using GAPDH by the 2−ΔCT method. And, the relative quantification of miR-145-5p in tissue specimens and cell lines was measured by using the mirVANATM miRNA Isolation Kit (ABI, MA USA). The PCR program was set as follows: 95° C for 10 min, followed by 35 cycles of 95° C for 15s, 60° C for the 30s, and 72° C for 45s.

Antibody against RGS3 was prepared (Abcam, MA, USA). The immunohistochemistry assay complied with our previously described methods [14]. The protein expression levels detected by IHC were assigned to two experienced pathologists independently for blind examination and were separated into two groups by staining intensity grade: no to low staining (0–1+) and moderate to high staining (2+–3+).

Preparation of the plasmid and cell transfection

The lentiviral vectors pLKO.1 (Addgene, Cambridge, MA, USA) containing shRNA were transfected into cultured HepG2 and Hep3B cells at exponential phase (JIKE Biochemistry, Shanghai, China) for suppressing the expression of UBE2MP1 transcript. Meanwhile, the control vectors were set up. The transfected cells were selected by using a medium mixed with G418 (Santa Cruz Biotechnology, Inc; 400 μg/ml). The mimic was used to transfect HCC cells for ectopically introducing miR-145-5p (HepG2/miR-145-5p; Hep3B/miR-145-5p), and the negative controls (HepG2/NigmiR; Hep3B/NigmiR) were set. And the rescue experiment by knock-out miR-145-5p was set up by using the siRNA method by using the protocol and siRNA vector tools designed by Jike Biotech Company, (Shanghai, China), and the validation was implemented by the RT-qPCR assay. The lentiviral vector pLV (Addgene, Cambridge, MA, USA) was applied for ectopically re-expressing either RGS3 (pLV- RGS3) for the rescue experiments, and the pLV-Null was set as control.

Cell proliferation and cell cycle detection

The treated HCC cells (1x10⁶) were cultured in 96-well microtiter plates triplicated and incubated at an atmosphere of 5% CO₂ and 37° C for 5 days. Microplate computer software (Bio-Rad Laboratories, Inc., Hercules, CA, USA) was applied for measuring the OD following the Cell Counting Kit-8 (CCK-8) assay kit protocol (Dojindo, Tokyo, Japan). Then, we plotted the cell proliferation curves. Meanwhile, the cells were treated with ethanol fixation, followed by RNase A treatment and propidium iodide staining. Flow cytometry detection was carried out using FACSCalibur (Becton-Dickinson, Franklin Lakes, NJ, USA) for quantifying cell populations at the G0/G1, S, and G2/M phases, and ModFit software e (Becton-Dickinson) was used. The debris and fixation artifacts of the cells were excluded.

Cell apoptosis analysis

Cell apoptosis rate was calculated by using PE-Annexin V Apoptosis Detection Kit I (BD Pharmingen, USA) following the instructions. Transfected cells were resuspended in the concentration of 1×10⁶ cells/ml by the 1×Binding Buffer. 5μl of FITC and 5μl of PI were added into 100μl of the cell suspension, followed by a 15 minutes incubation in darkness, added with 400μl×Binding Buffer. The apoptosis rate was calculated through flow cytometry (Becton-Dickinson). Both Annexin V-FITC-positive and PI-negative cells were considered apoptosis cells.

The dual-luciferase reporter assay

By using the online tools of microcosm (http://mirecords.biolead.org) and dreamBase (http://rna.sysu.edu.cn/dreamBase/), miR-145-5p was respectively predicted directly binding to the sequences of either the 3’-untranslated region (3’-UTR) of RGS3 mRNA or the UBE2MP1 transcript. Both of the sequences above were intercepted for the 202 bp sections along with the corresponding mutative ones for further detection by the dual-luciferase reporter assay. (Supplementary Table 2). The sequences were respectively cloned into the pMIR-Report luciferase vector, which contains firefly luciferase, and the pRL-TK vector luciferase was set as control (Promega, Madison, WI, USA). These two sets of vectors were co-transfected into both HepG2 and Hep3B cells transfected miR-145a-5p mimics or the control ones. The luciferase activity was measured via the Dual-Glo Luciferase assay system (Promega) 48 hours after the transfection.

Statistical analysis

Statistical analysis was carried out by using SPSS 20.0. P-values were calculated using an unpaired Student's t-test and Fisher's exact test and the one-way ANOVA as the statistical methods. Differences were considered statistically significant at P-values < 0.05.

Pseudogene UBE2MP1 was anomalously transcripted in HCC cell lines and tissues

The exploration and analysis of the datasets from the dreamBase databases indicated a universal transcription of UBE2MP1 in either pan-cancers or HCC (Figure 1A). On basis of a random walk-based multi-graphic (RWMG) model algorithm we developed, a series of pseudogenes generating lncRNA-like transcripts in HCC were screened out from the analysis tools of the Liver Cancer lncRNA Explore (LCLE) online tool (https://datasciences123.shinyapps.io/LCLE/) [13], and UBE2MP1 and the volcano plot that presents the expression profile of the pseudogene transcript indicated a similar result (Figure 1B). As we observed, merely no detectable UBE2MP1 transcript was found in the normal organs and tissues, including the liver. Whereas, the up-regulation of the UBE2MP1 transcript was detected in different tumor tissues. Especially for HCC, the remarkable elevation of the UBE2MP1 transcript was observed.

Figure 1. Pseudogene UBE2MP1 was anomalously transcripted in HCC cell lines and tissues. (A) Analysis of the dreamBase datasets. The heatmap indicated the expression profile of the transcript of UBE2MP1 in pan-cancer. (B) Analysis of the data from the LCLE. Pseudogene UBE2MP1 is transcriptionally activated and presents a high expression in HCC. (C) RT-qPCR assay demonstrated a significantly high expression of UBE2MP1 transcript in three HCC cell lines, in comparison with the control LO2 cells barely transcribed this pseudogene (**P<0.01). (D) RT-qPCR assay was conducted on 93 real patients’ specimens. The UBE2MP1 transcript was significantly highly expressed in tumor tissues, and only very few detectable UBE2MP1 transcript was found in the non-cancerous tissues (**P<0.01). (E) Statistic of the number of cases concerning the expression of UBE2MP1 transcript in HCC specimens. UBE2MP1 transcript is detectable and highly expressed in most of the tumor tissues (79/93), and was not detected in most of the adjacent non-cancerous tissues (88/93) (P<0.01).

The UBE2MP1 transcript expression in three HCC cell lines (Huh7, HepG2, and Hep3B) was detected by RT-qPCR assay. The level of UBE2MP1 transcript in three HCC cell lines is significantly higher than in the control LO2 cells (Figure 1C). For the real patients’ specimens from our medical center, a high level of UBE2MP1 transcript was detected in HCC tumor samples, and the adjacent non-cancerous liver tissues presented nearly no sign of UBE2MP1 transcription. As Figure 1D, 1E show, the UBE2MP1 transcript is detectable in all of the tumor specimens, and 84.95% (79/93) of the HCC specimens presented a high level of UBE2MP1 transcript expression, and the rest 15.05% (14/93) presented a relatively lower level UBE2MP1 expression. On the contrary, only 5.38% (5/93) of the adjacent non-cancerous liver tissues presented detectable but low expression of UBE2MP1. Simultaneously, we also validated the high expression of the parental gene UBE2M by RT-qPCR assay and the IHC assay in the real patient specimens and cell lines, which suggests a co-expression inclination between these two genes (Supplementary Figure 1).

Transcription of UBE2MP1 is correlated with the HCC patients’ clinicopathologic features

The correlation between the transcription of UBE2MP1 in HCC and the clinicopathologic features of the 93 real HCC patients was analyzed statistically by Fisher's exact test and the one-way ANOVA. There is no significant correlation between UBE2MP1 and the patient’s age, gender, and virus control status. Whereas, UBE2MP1 transcription shows an adverse relationship with the tumor size (P<0.05), serum Alpha-fetoprotein (AFP) quantity (P<0.05), more advanced TNM stages (P<0.05), tumor microsatellite formation (P<0.05), invasion of venous (P<0.05), and liver cirrhosis stages (P<0.05) (Table 1). The findings here strongly suggest that the UBE2MP1 transcript plays a positive role in HCC development.

UBE2MP1 depletion impairs cell proliferation and arrests cell cycles in HCC

The RT-qPCR assay was conducted to validate the effect on UBE2MP1 depletion, in both HepG2 and Hep3B cells, and the expression of the parental gene was not affected (Figure 2A). The cell proliferation was significantly suppressed in both the two cell lines by abrogating the UBE2MP1 transcript (unpaired Student's t-test; *P<0.05; **P<0.01) (Figure 2B, 2C).

Figure 2. UBE2MP1 Depletion impairs cell proliferation and arrests cell cycles in HCC. (A) UBE2MP1 depletion was conducted in HepG2 and Hep3B cells by using shRNA transfection. The RT-qPCR assay demonstrated the significant abrogation of UBE2MP1 transcript in the two treated cell lines, and no significant change of the parental gene UBE2M was induced (**P<0.01). (B, C) The CCK8 assay was conducted to describe cell proliferation status. The cell proliferation in both the two cell lines was significantly suppressed by abrogating the UBE2MP1 transcript (*P<0.05; **P<0.01). (D, E) Flow cytometry was used for analyzing the cell cycle distribution. And the representative histograms were shown. The cell cycle of both HepG2 and Hep3B cells was significantly arrested in G0/G1 phase by depleting the UBE2MP1 transcript. The results are means of three independent experiments ±SD. (*P< 0.05, **P< 0.01).

The flow cytometric analysis showed a significant cell cycle arrest at the G0/G1 phases in the HCC cells when UBE2MP1 was depleted (Figure 2D, 2E). The percentage of the HepG2 and Hep3B cells in the G0/G1 phase increased respectively from 50.2% to 64.8% (P<0.01) and from 49.9% to 63.0% (P<0.01) (Student's t-test). Meanwhile, the obvious percentage decrease was observed in the S phase (HepG2: from 25.0% to 18.5%, P<0.05; Hep3B: from 25.7% to 17.5%, P<0.01) and the G2/M phase (HepG2: from 24.8% to 16.7%, P<0.01; Hep3B: from 24.4% to 19.5%, P<0.05).

Depletion of UBE2MP1 enhances HCC cell apoptosis

The flow cytometric analysis was applied for calculating the statuses of cell apoptosis by an unpaired Student's t-test. The cell apoptosis rates in both HepG2 and Hep3B cells were significantly increased (HepG2: from 13.01% to 27.59%, P<0.01; Hep3B: from 13.27% to 25.71%, P<0.01) following the depletion of UBE2MP. This finding suggested that the transcription of UBE2MP1 in HCC could enhance cell maintenance by inhibiting cell apoptosis (Figure 3A–3D).

Figure 3. Depletion of UBE2MP1 enhances HCC cell apoptosis. (A) Cell apoptosis was detected by flow cytometry. The representative histograms demonstrate the apoptosis rate in HepG2 cells. (B) The counting number of the apoptotic cells was significantly increased in HepG2 cells by depleting UBE2MP1 (**P<0.01). The results are means of three independent experiments ±SD. (C) The representative histograms demonstrate the apoptosis rate in Hep3B cells. (D) The counting number of the apoptotic cells was significantly increased in HepG2 cells by depleting UBE2MP1 (**P<0.01). The results are means of three independent experiments ±SD.

UBE2MP1 transcript sponges miR-145-5p and was positively correlated with RGS3

We explored the UBE2MP1 transcript and found that a sequence from 490 bp to 507 bp to the 3’ end of the UBE2MP1 transcript is probably a specific binding site that matches the seed region of miR-145-5p, which prompts a ceRNA effect of UBE2MP1 on miR-145a-5p (The minimum free energy, Mfe: -21.6 kcal/mol) (Figure 4A). The exploration of TCGA datasets also indicated a significant decline of miR-145-5p in HCC and was associated with poor clinical survival outcomes (Figure 4B, 4C). And according to the RT-qPCR assay, miR-145-5p was expressed at a relatively lower level in the HCC cell lines compared with the control LO2 cells (Figure 4D). Moreover, the expression of miR-145-5p was consequentially increased after the depletion of UBE2MP1 in the two HCC cell lines (Figure 4E).

Figure 4. UBE2MP1 transcript sponges miR-145-5p and was positively correlated with RGS3. (A) Predicted binding sequence of UBE2MP1 transcript with the seed sequence of miR-145-5p. The minimum free energy (Mfe) hybridization is calculated as: -21.6 kcal/mol. The comprehensive expression profile of miR-145-5p in pan-cancers (**P<0.001). (B) The expression profile of miR-145-5p in HCC through analysis of the dataset from the TCGA database (**P<0.01). (C) Exploration of the TCGA datasets indicated a significant decline of miR-145-5p in HCC associated with shortened overall survival. (D) The RT-qPCR assay demonstrated a significant decrease of miR-145-5p in three HCC cell lines, in comparison with the control LO2 cells (**P<0.01). (E) The expression of miR-145-5p in both HepG2 and Hep3B cells was significantly elevated by depleting UBE2MP1 (**P<0.01). (F) The dual-luciferase reporter assay verified the direct interaction of UBE2MP1 to miR-145-5p, as a molecular sponge (**P<0.01).

Based on this, we constructed the mutated binding site on UBE2MP1, and we carried out the dual-luciferase reporter assay to verify the direct interaction between UBE2MP1 and miR-145-5p. As discovered, the luciferase signal in either HepG2 or Hep3B cells transfected with miR-145-5p mimics was decreased significantly, after the transfection of the UBE2MP1/pMIR/WT vector, in comparison with the control ones. On the contrary, the transfection of the UBE2MP1/pMIR/MUT vector did not induce signal changes with significance (Figure 4F). Thus, the findings convincingly illustrated the ceRNA effect of UBE2MP1 on sponging miR-145-5p.

RGS3 mRNA is targeted by miR-145-5p post-transcriptionally

Interestingly, we occasionally noticed that RGS3 is highly expressed in both HCC cell lines and HCC tissues (Figure 5A–5D). And notably, RGS3 shared a positively correlated with UBE2MP1 expression in the HCC tissues from our center, and the expression of RGS3 presented a remarkable decrease in HCC cells when UBE2MP1 was depleted (Figure 5E). We wondered if there exists a co-expression or regulation between UBE2MP1 and RGS3. On basis of this point, we tried to explore the role that miR-145-5p plays in this axis. The microcosm an online prediction software (https://www.microcosm.com/) indicated that miR-145-5p was an upstream regulator potentially interacting with the 3’-untranslated region (3’-UTR) of RGS3 mRNA with a high recommendation (Mfe: -28.2 kcal/mol) (Figure 5F). The dual-luciferase reporter assay was used for validating this assumpted direct binding between miR-145-5p and RGS3 mRNA. Similar to the above technique, we constructed the vectors containing a 202 bp sequence intercepted from the 3'UTR from the RGS3 mRNA (WT-UTR), and also the control luciferase vectors containing a mutated miR-145-5p targeting site (MUT-UTR). Both HepG2 and Hep3B cells were transfected with either the above two kinds of vectors and also mimicked by miR-145-5p. Here, taking HepG2 cells, for example, the miR-145-5p mimics (HepG2/miR-145-5p) significantly defected the luciferase signal of RGS3/pMIR/WT in comparison with the negative control (HepG2/NigmiR). The signal suppressive effect induced by miR-145-5p was abrogated in HepG2 cells transfected with a mutated miR-145-5p binding site (Figure 5G). And similar results were obtained from the Hep3B cell line too.

Figure 5. RGS3 mRNA is targeted by miR-145-5p post-transcriptionally. (A) The RT-qPCR assay demonstrated a significant up-regulation of RGS3 mRNA in three HCC cell lines, in comparison with the control LO2 cells (**P<0.01). (B) RT-qPCR assay was conducted on 93 real patients’ specimens. The RGS3 was significantly highly expressed in tumor tissues (**P<0.01). (C) Statistic of the number of cases concerning the expression of RGS3 in HCC specimens. RGS3 is highly expressed in most of the tumor tissues (78/93) (P<0.01). (D) Representative graph of immunohistochemistry analysis (400×) of the HCC cases. The IgG antibody was used for staining the specimens as a control. RGS3 expression in tumor specimens was significantly higher than in adjacent non-cancerous tissues. (E) RGS3 shared a positively correlated with UBE2MP1 expression in the HCC tissues from our center, and the expression of RGS3 presented a remarkable decrease in HCC cells when UBE2MP1 was depleted (**P<0.01). (F) Predicted binding sequence of the 3’UTR of RGS3 mRNA with miR-145-5p. The Mfe value is calculated as: -28.2 kcal/mol. (G) The dual-luciferase reporter assay verified the direct interaction between miR-145-5p and RGS3 mRNA (**P<0.01).

Pseudogene UBE2MP1 enhances HCC tumor growth by modulating the miR-145-5p/RGS3 axis

Since the verification of the post-transcriptional degeneration effect of miR-145-5p on RGS3 mRNA, we prompt that the phenotypes induced by UBE2MP1 in HCC are due to the modulation of the potential miR-145-5p/RGS3 axis. Take HepG2 cells as an example, firstly we lower the miR-145-5p expression in HepG2 cells treated with UBE2MP1 depletion again (Figure 6A). The suppressed cell proliferation was significantly recovered along with the escape of G0/G1 stage arrest (Figure 6B, 6C). Simultaneously, the number of apoptotic cells was decreased (Figure 6D). Secondly, we re-introduced RGS3, which had been down-regulated through UBE2MP1 depletion, in HepG2 cells. As expected, the knock-down of miR-145-5p led to an obvious increase in RGS3 expression at both mRNA and protein stages (Figure 7A). Interestingly, even though the expression of either UBE2MP1 or miR-145-5p was not impacted by re-introducing RGS3 in HCC cells, the cell proliferation and apoptosis statuses induced by UBE2MP1 depletion were effectively recovered RGS3 (Figure 7B–7F).

Figure 6. Depletion of miR-145-5p rescued the phenotype induced by UBE2MP1 depletion. (A) The RT-qPCR assay validated the effect of knocking out miR-145-5p in HepG2 cells (**P<0.01). (B) No significant change was induced on the UBE2MP1 transcript by miR-145-5p depleting. (C) The suppressed cell proliferation was significantly recovered by depleting miR-145-5p (*P<0.05, **P<0.01). (D) The arrested cell cycle induced by UBE2MP1 depletion in HepG2 cells was reversed when miR-145-5p was depleted (*P<0.05, **P<0.01). (E) The number of apoptotic cells was decreased in HepG2 cells with UBE2MP1 depletion when RGS3 was elevated (**P<0.01).

Figure 7. Re-introducing RGS3 rescued the phenotype induced by UBE2MP1 depletion. (A) Depletion of miR-145-5p significantly increases the expression of RGS3 in HepG2 cells (*P<0.05, **P<0.01). (B) Validation of the re-introduction of RGS3 in HepG2 cells treated with UBE2MP1 depletion (**P<0.01). (C) No significant changes were induced on miR-145-5p and UBE2MP1 transcript expression by re-introducing RGS3. (D, E) The suppressed cell proliferation was significantly recovered along with the escape of the G0/G1 stage arrest by re-introducing RGS3 (*P<0.05, **P<0.01). (F) The increased number of apoptotic cells declined in HepG2 cells with UBE2MP1 depletion when RGS3 was re-introduced (**P<0.01).

Last but not least, we simultaneously obtained similar results from the rescue experiments in Hep3B cell lines (Supplementary Figures 2, 3). All these findings above present an HCC-promoting axis of miR-145-5p/RGS3 under the control of UBE2MP1 through its ceRNA effect.