Functional role and epithelial to mesenchymal transition of the miR-590-3p/MDM2 axis in hepatocellular carcinoma

The differential expression data of miR-590-3p and MDM2 between HCC and normal liver tissues were extracted from The Cancer Genome Atlas (TCGA) through Encyclopedia of RNA Interactomes (ENCORI) Pan-cancer analysis platform [26]. Potential downstream targets of miR-590-3p were predicted using the Condition-Specific miRNA Targets (CSmiRTar) database [27], which collects data from four common prediction tools (such as miRDB, TargetScan, microRNA.org and DIANA-microT) and then applies a functional filter to detect targets that are only expressed in a certain tissue and/ or a certain disease. TargetScan platform was used to feature the sequence alignment between miR-590-3p seed region and the downstream targets [28].

Two adherent hepatocellular carcinoma cell lines were used in this study: HepG2 and SNU449. HepG2 is a well differentiated, non-invasive HCC cell line. SNU449 cell line is a tumorigenic and more advanced stage of HCC with grade (II–III) [29]. HepG2 was obtained from NAWAH Scientific Inc., (Cairo, Egypt) and SNU449 was a kind gift from Dr Mehmet Ozturk from the Department of Molecular Biology and Genetics, Bilkent University, Turkey. HepG2 and SNU449 were grown in DMEM and RPMI 1640, respectively (LONZA, Bend, OR, USA), supplied with 10% fetal bovine serum (GIBCO, Grand Island, NY, USA) and 5% penicillin–streptomycin antibiotic (GIBCO, Grand Island, NY, USA). Cells were maintained in a humidified incubator supplied with 5% CO2 at 37 ̊Ϲ.

The knockdown of the MDM2 gene (NCBI Reference Sequence: NG_016708.1) was achieved using ON-TARGETplus SMARTpool siRNA (siMDM2 SMART Pool; L-003279–00-0020) (Dharmacon, Lafayette, CO, USA). Target sequences of the MDM2 siRNAs are presented in Table 1. AllStars Negative Control siRNA (SI027280) (Qiagen, Hilden, Germany), hidden sequence.

HepG2 cells were ectopically transfected with 40 nM of miR-590-3p mimics (Invitrogen, Waltham, MA, USA) or Allstar negative control siRNA (Qiagen, Hilden, Germany) using lipofectamine 3000 (Invitrogen, Waltham, MA, USA) in a 6-well plate with a seeding density of 4 × 10⁵. For MDM2 knockdown, cells were transfected with 60 nM of MDM2 siRNA or Allstar negative control siRNA in a 12-well plate with a seeding density of 7 × 10⁴. Cells were incubated for forty-eight hours before performing further experiments. All transfection experiments were done in an RNase-free environment.

To assess miR-590-3p expression levels, total RNA-enriched small RNAs were extracted using miRNeasy mini kit (Qiagen, Hilden, Germany). MiRCURY LNA SYBR Green qPCR System (Qiagen, Hilden, Germany) was carried out to assess the levels of miR-590-3p in HepG2 and SNU449 cells. In brief, 10 ng of mature miRNA was polyadenylated and reverse-transcribed parallelly into a universal one first-strand cDNA. A UniSp6 RNA spike-in was used in the RT reaction to control efficient cDNA synthesis. RT-qPCR was performed using Locked Nucleic Acid (LNA) SYBR Green qPCR assays. Mir-103a-3p served as the endogenous normalizing control.

For assessing mRNA levels of downstream genes used in this study, total RNA was extracted from transfected cells using Trizol reagent (Invitrogen, Waltham, MA, USA) as per the manufacturer’s protocol. RT reactions were made using 2 μg of total RNA samples using RevertAid First Strand cDNA synthesis kit (Thermo Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. 5 ng/reaction of cDNA along with 250–300 nM forward or reverse primers were used to carry out the qPCR using PowerUp SYBR Green kit (Applied Biosystems, Waltham, MA, USA). GAPDH was used for normalizing the expression of different genes. Table 2 lists the primers used in this study. All qPCR was done on Applied Biosystems 7500 system.

To test the effect of mir-590-3p overexpression on the migration of HepG2 cells, Transwell assay was performed. Cells were collected forty-eight hours post-transfection, and almost (5 × 10⁵) of mimic or negative control transfected cells were suspended in media supplemented with 1% FBS. For MDM2 knockdown, almost 150*10³ cells/well of MDM2 siRNA or negative control.transfected cells were resuspended in serum-free media. Cell suspensions were loaded on the upper chamber of 8 μm cell culture translucent inserts (Greiner Bio-One, Frickenhausen, Germany) placed in a 24-well plate. Inside the 24-well plate, complete media with 10% FBS was added to serve as a chemotactic agent to allow for cell migration through the microporous filter. After 10 h, cells in the upper chamber of the insert were removed with cotton swabs. Migrated cells at the bottom of the insert were fixed with 4% formaldehyde and stained with DAPI (KPL, Gaithersburg، MD, USA) (1:1000 in PBS). Cells were examined under a fluorescent microscope at 20X magnification, and five random fields per insert were captured. Cells were counted using Fiji Image J software.

A colony-formation assay was conducted to test if mir-590-3p has a role in regulating HCC clonogenicity. Forty-eight hours post-transfection, HepG2 cells transfected with either miR-590-3p mimic or negative control were detached and seeded in a 6-well plate at a density of 500 cells. Cells were incubated to colonize for seven days. For MDM2 knockdown, transfected cells were incubated for ten days. When visible colonies were seen under the microscope, the experiment was terminated, and formed colonies (more than 50 cells/ colony) were fixed with 4% formaldehyde, stained with crystal violet, and manually counted.

Wild type (WT) and mutant (MUT) constructs of the 3′ untranslated region (3′UTR) of MDM2 were chemically synthesized. For the WT constructs, the miR-590-3p binding site in the 3′UTR of MDM2 was cloned into the pmirGLO dual luciferase miRNA target expression vector (Promega, Madison, WI, USA). For the MUT constructs, the binding region nucleotides in the 3′UTR of MDM2 was completely deleted. The forward and reverse primer sequences of both designed constructs flanked by SalI and NheI restriction sites are shown in Table 3. HepG2 cells were co-transfected with 50 ng of the WT or MUT constructs along with miR-590-3p mimics or negative control using Lipofectamine 3000 (Invitrogen, Waltham, MA, USA). Twenty-four hours after transfection, cell lysates were prepared, and the relative luciferase activity was normalized to the total protein content and tested through a dual-luciferase reporter assay system (Promega, Madison, WI, USA).

Transfected cells were resuspended and lysed using the CelLytic™ M Cell lysis Reagent (Sigma Aldrish, Burlington, MA, USA). Equal amounts (15 μg) of proteins were loaded on 10% SDS–Polyacrylamide gel and separated by electrophoresis at 150 V for an hour. SDS gels were blotted to nitrocellulose membranes (GE Healthcare, Chicago, Illinois, USA). Membranes were then left for overnight blocking at 4 C in 5% nonfat dry milk diluted in 1X Tris-buffered saline with 0.01% Tween 20 (TBST). Following blocking, membranes were incubated with primary antibodies. Membranes were then washed by TBST and incubated with alkaline phosphatase-conjugated secondary goat anti-rabbit IgG or goat anti-mouse IgG antibodies (KPL, Gaithersburg، MD, USA) (1:10,000 in 5% non-fat milk). Finally, membranes were washed and incubated with the chromogenic substrate BCIP/NBT (KPL, Gaithersburg، MD, USA) until protein bands were visible with eye. Primary antibodies mentioned here are Anti-GAPDH (MA5-15,738, 1:10,000) (Invitrogen, Waltham, MA, USA) and MDM2 (E-AB-31995, 1:200) (Elabscience Biotechnology Inc., Houston, Texas, USA).

The relative expression of miR-590-3p in HepG2 and SNU449 and the differential expression levels of downstream genes in different transfection conditions were analyzed using the comparative ΔΔCT method [30]. Fiji ImageJ software [31] was used to analyze Transwell assay results. Data presented as the mean ± standard error of the mean (SEM) from three independent experiments unless specified otherwise. For statistical analysis, an unpaired Students’ t-test was used to measure differences between two experimental groups. A probability value (P-value) < 0.05 was considered a significant result. Statistical analyses were established using GraphPad Prism 6 statistical packages [32].

Comparison of miR-590-3p expression in HCC and non-cancerous liver tissues revealed a very significant (P > 0.0001) downregulation of miR-5909-3p expression in 370 HCC samples compared with 50 healthy counterparts, with a fold change of (0.75) (Fig. 1A). Examination of miR-590-3p levels in two cell lines representing an early and an advanced stage of HCC using RT qPCR revealed the expression of miR-590-3p to be significantly lower in SNU449 cells compared with HepG2 cells (P < 0.01) (Fig. 1B).

After establishing a miR-590-3p transient transfection using miR-590-3p mimics in the HepG2 cell line, miR-590-3p levels post-transfection were dramatically higher in miR-590-3p mimic transfected cells than in the negative control (Fig. 2A). Overexpression of miR-590-3p caused a significant decrease in the number of colonies of HepG2 cells compared with the negative control (Fig. 2B). The Transwell migration assay (Fig. 2C) revealed the cellular migration capacity of HepG2 cells to be significantly reduced when miR-590-3p was overexpressed. Having observed that miR-590-3p overexpression exerted a critical role in suppressing HepG2 cell motility, we investigated its effect on EMT. Bioinformatic analyses and RT qPCR revealed that the transcript levels of four EMT transcription factors: SNAIL, SLUG, ZEB1, and ZEB2 were dramatically decreased in miR-590-3p-mimics-transfected HepG2 cells compared with the control group (Figures S1 and 2D). In addition, a significant reduction was found in the levels of the mesenchymal marker N-cadherin, but not Vimentin, in HepG2 cells after overexpressing miR-590-3p (Fig. 2E). Consistently with qPCR results, a complementary binding between the seed region of miR-590-3p and N-cadherin was shown, but no sequence alignment was found with Vimentin (Fig. S2).

The CSmiRTar prediction algorithm identified five validated target genes that are liver-tissue and HCC-disease specific (Table 4). Among these, we selected MDM2 for further investigation based on a previous study that reported a functional network between MDM2 and miR-590-3p in HepG2 cells [33]. Specifically, miR-590-3p was identified as one of 33 P53-regulated miRNAs after doxorubicin treatment, and MDM2 was shown to be a direct target gene of miR-590-3p. However, our study was primarily interested in exploring the P53-independent effects of MDM2, particularly those related to EMT.

To further investigate the potential role of MDM2 in HCC, we analyzed TCGA data and found that MDM2 expression was significantly upregulated in HCC samples compared to healthy samples (fold change = 1.34, P < 0.001) (Fig. 3A). Additionally, we conducted functional experiments in HepG2 cells and found that overexpression of miR-590-3p significantly downregulated MDM2 mRNA levels (P < 0.05) (Fig. 3B). We confirmed the target site of miR-590-3p on MDM2 mRNA using TargetScan, which identified a sequence alignment in the seed region of miR-590-3p (Fig. 3C). Finally, to directly test if there is a mechanistic interaction between miR-590-3p and MDM2, we performed luciferase reporter assays and found that co-transfection of miR-590-3p mimics and WT constructs of the MDM2 3′UTR significantly decreased luciferase activity compared to the negative control (P < 0.01), while no change was observed in the MDM2 3′UTR MUT group (Fig. 3D). Taken together, our results provide evidence that miR-590-3p directly targets MDM2 and regulates its expression in HCC cells.

To characterize the role of MDM2 in HepG2 cells, we explored the effect of MDM2 silencing on cell proliferation, migration, and EMT progression of HepG2 cells. Figure 4A&B illustrates that the on-target smart pool siRNAs induced remarkably strong and detectable knockdown (more than 80% at the mRNA level and almost 70% at the protein level) (Full length blot S3 A&B). The clonogenic cell survival and Transwell assays revealed that MDM2 gene silencing impaired HepG2 cell proliferation and migration (Fig. 4C&D). The RT qPCR data indicated that MDM2 silencing significantly suppressed the transcript levels of EMT transcription factors such as SNAIL, SLUG, ZEB1, and ZEB2, and mesenchymal genes, such as N-cadherin and Vimentin (Fig. 4E&F).