MiR-106b-5p regulates esophageal squamous cell carcinoma progression by binding to HPGD

Two mRNA (GSE38129 and GSE17351) and one miRNA (GSE114110) expression profile was downloaded from the GEO DataSet (https://​www.​ncbi.​nlm.​nih.​gov/​gds/​). GSE38129 included ESCC and adjacent normal samples from 30 Chinese patients, whereas GSE17351 included ESCC and adjacent normal samples from five American patients. With adjusted P-value (adj. P) < 0.05, and log fold change (logFC) < − 1, the differentially expressed genes (DEGs) were screened using the limma 3.26.8 package. For the miRNAs, Limma 3.26.8 was applied to analyze the differentially expressed miRNAs in ESCC with adj. P < 0.05, and logFC > 1. Metascape (https://​metascape.​org/​gp/​index.​html) was used to analyze the key biological processes associated with these DEGs. Their expression in normal and esophageal carcinoma (ESCA) tissues was analyzed using UALCAN, with data obtained from the cancer genome atlas (TCGA). StarBase (http://​starbase.​sysu.​edu.​cn) was used to analyze miRNA expression in ESCA tissues and to predict the miRNAs that bind to HPGD. TargetScan was also used to predict the miRNAs that could bind to HPGD. Finally, DEGs and miRNAs were overlapped using the Venny 2.1.0.

ESCC tissue specimens (n = 45) and paired adjacent normal esophageal samples (n = 45) were collected from patients at Huangshi Central Hospital, Wuhan, China between October 2016 and February 2019. The protocol was approved by the Ethics Committee of the Huangshi Central Hospital, and all patients read and completed the informed consent forms to participate in the study. All tissues were stored at − 80 °C, and the clinical data pertaining to individual patients with ESCC are listed in Table 1.

Human ESCC cell lines (KYSE30, KYSE180, KYSE450, and KYSE510) and normal esophageal epithelial cells (Het-1A) were purchased from ATCC (Manassas, VA, USA). All cell lines used in this study were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Cat#: A4192101, Gibco, Waltham, MA, USA) containing 10% fetal bovine serum (Cat#: 10099141, Gibco) and incubated at 37 °C in an incubator containing 5% CO₂.

RT-qPCR and RNA isolation were performed according to methods described previously [27]. RNA isolation from the cell samples was performed using TRIzol Reagent (Invitrogen, Waltham, MA, USA). cDNA was obtained from the extracted RNA using the PrimeScript First Strand cDNA Synthesis Kit (Takara, Dalian, China). Gene expression was quantitated by RT-qPCR using SYBR Premix Ex Taq (Takara) on an ABI Prism 7900 Detector System (Life Technologies Inc., Waltham, MA, USA). miRNA and mRNA expressions were normalized to that of U6 and β-actin, respectively. The data were analyzed using the 2^−ΔΔCT method. The primer sequences used are listed in Table 2.

MiR-106b-5p inhibitor, miR-106b-5p mimic, and their corresponding negative controls, were purchased from Shanghai Tuoran Co., Ltd. The corresponding sequences are listed in the Supplemental Table 1. A full-length HPGD cDNA was synthesized (Shanghai Tuoran Co. Ltd.) and cloned into pcDNA3.1 plasmid. Puromycin (4 μg/mL) was used to select stably transfected cells. KYSE450 and KYSE510 cells (3 × 10⁵ cells) were transfected with 50 nM miR-106b-5p inhibitor, miR-106b-5p mimic, or miR-NC using Lipofectamine 3000 Reagent (Invitrogen, Waltham, MA, USA). The cells were cultured for 48 h before performing subsequent experiments as described previously [28].

CCK-8 assay was performed to investigate cell viability [29] (Cat#: K1018; APExBIO, China). Briefly, KYSE450 and KYSE510 cells (3 × 10³ cells) were seeded in a 96-well plate. At the indicated times, CCK-8 (10 μL) reagent was added to the wells with cells, and the cells were incubated further for 2 h. Finally, the optical density (OD) was measured at 450 nm with a multimode-plate-reader (Tecan, Switzerland).

BrdU assay was performed according to a previously reported method [30]. First, KYSE450 and KYSE510 cells (3 × 10³ cells) were cultured in 96-well plates for 24 h followed by 12 h of serum starvation. After another 8 h, serum was added back to the cells. The BrdU Cell Proliferation Assay Kit (Cat#: 6813, CST, Danvers, MA, USA) was used to label cells for 8–12 h without removing the treatment media. Finally, the OD was measured at 450 nm.

Cell adhesion assays were performed based on the methodology used in a previous study [22]. KYSE450 and KYSE510 cells (3 × 10³ cells) were cultured in 96-well plates. Collagen I solution (40 μg/mL; Cat#: C7661, Sigma-Aldrich, St. Louis, MO, USA) was added to the wells and the plates were stored overnight at 4 °C. The transfected cells were cultured in serum-free DMEM for 8 h. Cells were treated with 10 mM EDTA (in DMEM) for 10 min to dissociate them from the dishes. After collecting and resuspending the cells in DMEM with 0.1% BSA (2 × 10⁵ cells/mL), the cells suspension (100 μL) was added to a air-dried 96-well plate for another 30 or 60 min. After incubation, 100 μL DMEM was added to remove the non-adherent cells and the dishes were further incubated for 4 h. Subsequently, the MTT substrate (Cat#: CT01, Sigma) (10 μL/well) was applied to the treated cells for 2 h at 30 °C. Next, 100 μL of DMSO was added to each well containing the lysed cells. Finally, the absorbance was measured at 570 nm.

Cells were disassociated, suspended, and plated in a 6-well culture plate at a density of 100 cells/well. After culturing for 14 days at 37 °C, the cells were washed twice with PBS and stained with Giemsa solution. Colonies larger than 75 μm in diameter or containing more than 50 cells were counted as a positive colony.

Cells were plated in a 6-well plate and once they reached more than 90% confluence, the cell monolayer was scratched with a 200 μL pipette tip to produce a wound. After removing the floating cells, fresh medium (without serum) was added to the wells and the cells were cultured at 37 °C for 24 h. Images of cells at the same location were captured at 0 h and 24 h using an Olympus IX51 inverted microscope (Olympus, Tokyo, Japan), and the wound healing rate was calculated as percent wound width covered = (width at 0 h - width at 24 h)/width at 0 h × 100%.

Transwell inserts coated with Matrigel (40 μg/well, BD Biosciences, San Jose, CA, USA) were used to assess cell invasion. Cells (2 × 10⁵) in serum-free medium were added to the upper chamber of the Transwell, and 600 μL of medium containing 10% FBS was added to the lower chamber. After 48 h of culture at 37 °C, the cells that had penetrated the membrane and adhered to the surface of the lower membrane were fixed with methanol, stained with Giemsa, and photographed under a light microscope (Leica Microsystems, Wetzlar, Germany).

Cells were collected 48 h after transfection and fixed overnight in 70% ethanol at 4 °C. Next, the DNA was stained using a cell cycle detection kit (KeyGen, Nanjing, China). Briefly, the cells were treated with RNase A and stained with 50 μg/mL propidium iodide. The DNA content was analyzed by flow cytometry using a FACS Calibur system (Becton Dickinson, Franklin, NJ, USA). Data were collected and processed using FlowJo FACS analysis software (Tree Star, Ashland, OR, USA).

The apoptotic ability of ESCC cells was assessed using the caspase-3/7 Assay Kit (Cat#: G8090, Promega, Madison, Wisconsin, USA) as described previously [27]. According to the manufacturer’s guidelines, both the ESCC cell lines (3 × 10³ cells) were cultured in 96-well plates. Next, the detection solution was prepared, and caspase 3/7 buffer was added to the bottle containing the caspase 3/7 substrate. After adding the mixed solution (100 μL/well) to the 96-well plate containing cells at 80% cell density, the mixture was incubated at room temperature for 2 h. Finally, OD values were measured at 450 nm.

Five-week-old female nude mice (5-weeks old) purchased from Shanghai SIPPR-BK Laboratory Animal Co. Ltd. (Shanghai, China) for the in vivo studies. Animal experiments were carried out in strict accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals. KYSE510 cells (2 × 10⁶) from the inhibitor-NC and miRNA inhibitor transfected groups were collected and resuspended in 2 mL PBS. Cells were injected subcutaneously into the backs of the nude mice. Tumor size was measured with calipers every fifth day. The tumor volume (V) was calculated using the following formula: V = length × Width² × 1/2. All the mice were euthanized 25 days after implantation by asphyxiation with carbon dioxide. The mice were placed into the euthanasia chamber and filled with CO₂ at 30% chamber volume /min. When the mice were unconscious and stopped breathing, the CO₂ flow was maintained for 1 min. The mice death was confirmed as cardiac arrest and did not respond to the toe-pinching reflex [28]. Tumors from the resected mice were weighed and photographed immediately.

The lung tissues were resected, fixed in 10% formaldehyde solution, dehydrated in an ethanol gradient, embedded in paraffin, and cut into slices of 4 μm thickness. After deparaffinization, the samples were stained with haematoxylin and eosin. The slices were then mounted and observed under a light microscope (Leica Microsystems).

The pmiRGLO-HPGD3′-UTR-Wt and pmiRGLO-HPGD3′-UTR-mutated vectors were constructed by Guangzhou Boxin Biotechnology Co. Ltd. (China). KYSE450 and KYSE510 cells (3 × 10⁵ cells) were co-transfected with pmiRGLO HPGD 3′-UTR-Wt or HPGD 3’UTR-Mut and either miR-NC or miR-106b-5p. At post 48 h post-transfection, luciferase was assessed with the Luciferase Assay Kit (Cat#: #16185, Thermo Scientific, Waltham, MA, USA) according to the manufacturer’s instructions and normalized to the firefly luciferase levels used as an internal control.

Biotin-labeled negative control (Bio-NC) and miR-106b-5p (Bio-miR-106b-5p) used in this study were provided by RiboBio (China). The two Bio-miRNA mimics were first transfected into KYSE450 and KYSE510 cells. After for 48 h lysis buffer supplemented with the protease and RNase inhibitors was added to the cells. Streptavidin beads (Cat#: #88817, Thermo Scientific) were washed and added to the cell lysates. The cells were incubated overnight at 4 °C. Subsequently, the beads were washed twice, and the RNA was eluted and purified using the RNeasy Mini Kit (Cat#: 74104, QIAGEN, Germany). Finally, HPGD enrichment was detected by RT-qPCR analysis.

Proteins from the cells were denatured and quantified. First, proteins (30 μg) were separated on a 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel. Next, the gels were electroblotted onto polyvinylidene difluoride (PVDF) membranes for hybridization. After blocking with 5% milk, the membranes were incubated with the following antibodies: anti-HPGD (1:1000, Cat#: ab187160, Abcam, UK), anti-Bax (1:2000, Cat#: ab32503, Abcam), anti-Bcl-2 (1:2000, Cat#: ab182858, Abcam), and β-actin (1:2000, Cat#: ab8226, Abcam) overnight at 4 °C. Then, the membranes were incubated with the corresponding secondary antibodies; goat anti-rabbit IgG H&L (HRP) (1:10000, Cat#: ab97051, Abcam) for HPGD and goat anti-mouse IgG H&L (HRP) (1:10000, Cat#: ab175783, Abcam, UK) for β-actin, at room temperature for 3 h. Next, ECL reagents (Bio-Rad, California, USA) were used to detect the protein bands. Finally, the density in each band was quantified using Image J software (ImageJ 1.48v, NIH, Maryland, USA).

Experimental data presented as the means ± standard deviation (SD) were analyzed using the paired Student’s t-tests for two-group comparisons and one-way ANOVA with Dunnett’s post hoc for multiple group comparisons using SPSS software (version 19.0; IBM Corp., Armonk, NY, USA). Three independent repeats were performed for each experiment. Significance was set at p < 0.05.

HPGD and miR-106b-5p were screened using bioinformatics analysis, and a total of 452 and 218 DEGs were identified in the GSE38129 and GSE17351 databases, respectively. Using Venny 2.1.0 (Fig. 1A), 85 overlapping DEGs were identified in GSE38129 and GSE17351. Metascape was used to process the 85 DEGs, and the positive regulation of cell death containing nine DEGs was selected as the key biological process (Fig. 1B). HPGD and MAL RNA transcripts were significantly downregulated in ESCA (Fig. 1C). Analysis of the mRNA levels of HPGD and MAL in clinical tissues revealed that HPGD expression was lower compared to that of MAL in the ESCC tissues (Fig. 1D). Additionally, HPGD expression was found to be significantly associated with the histological differentiation, tumor-node-metastasis (TNM) stage, and karnofsky performance status (KPS), but was not associated with gender, age, and weight loss in 45 ESCC cases (Table 1). Therefore, HPGD was selected as the gene of interest and further investigated in ESCC. TargetScan and starBase, used to predict the miRNAs that target HPGD, identified 749 and 94 miRNAs, respectively. Based on Venny 2.1.0 analysis miR-106b-5p and miR-31-5p were shortlisted as molecules of interest due to their overlap in the GSE114110 miRNA microarray, TargetScan, and starBase (Fig. 1E). Compared to miR-31-5p, miR-106b-5p expression was significantly upregulated in ESCA samples, and therefore we focused on miR-106b-5p for further experiments (Fig. 1F).

Aberrant upregulation of miR-106b-5p was detected by RT-qPCR analysis in ESCC tissues and cell lines. As shown in Fig. 2A, miR-106b-5p expression in the tumor tissues was nearly 2-fold higher than that in the corresponding normal tissues. The relationships between miR-106b-5p expression levels and the clinicopathological characteristics of individuals with ESCC are summarized in Table 1. We did not find a significant association between miR-106b-5p expression levels and patient gender, age, and weight loss in 45 ESCC cases. However, we found that the expression level of miR-106b-5p was positively correlated with histological differentiation, TNM stage, and KPS in ESCC patients. Examination of miR-106b-5p expression in ESCC cell lines revealed robust expression of miR-106b-5p in all ESCC cell lines compared to that in Het-1A cells (Fig. 2B). As KYSE450 and KYSE510 cells showed the highest miR-106b-5p expression levels, they were used in the subsequent experiments.

Synthetic target miRNA analogs were transfected into KYSE450 and KYSE510 cells to examine the effect of miR-106b-5p on ESCC cells. The transfection efficiency was determined by RT-qPCR (Fig. 3A and Supplemental Fig. 1). Further cellular functional assays demonstrated that the miR-106b-5p mimic increased cell viability and proliferation; whereas the miR-106b-5p inhibitor produced the opposite effect (Fig. 3B and C). Additionally, cell adhesion and colony formation experiments showed that upregulation of miR-106b-5p promoted adhesion and colony formation, and inhibition of miR-106b-5p reduced adhesion and colony formation (Fig. 3D and E). Likewise, an increase in cell migration and invasion was observed in wound healing and Transwell assays following transfection with the miR-106b-5p mimic. In contrast, the miR-106b-5p inhibitor impaired cell migration and invasion (Fig. 3F and G). Flow cytometry was performed to analyze the cell cycle and apoptosis rate. As shown in Fig. 4A and B, overexpression of miR-106b-5p decreased the proportion of cells in G1 phase and the apoptosis rate, and increased the proportion of cells in S-phase. In constrast, knockdown of miR-106b-5p showed an opposite effect on cell cycle and apoptosis. Similarly, analysis of the caspase-3/7 activity revealed that treatment with the miR-106b-5p inhibitor resulted in increased caspase-3/7 activity at 24 h, whereas the activity was reduced in the miR-106b-5p mimic group (Fig. 4C). Treatment with the miR-106b-5p mimic also reduced Bax protein levels and elevated Bcl-2 protein levels compared to that in the untransfected cells, whereas the miR-106b-5p inhibitor showed the opposite effect in the two ESCC cell lines (Fig. 4D).

To analyze the effect of miR-106b-5p on tumor growth and pulmonary metastasis of ESCC cells, KYSE510 cells that with miR-106b-5p inhibition were injected into nude mice. The results showed that tumor volume increased with time in the control group, whereas inhibition of miR-106b-5p reduced the tumor volume (Fig. 5A). This was further confirmed by visual observation of the tumor sizes following excision (Fig. 5B). Moreover, H&E staining was performed on the lung tissues following resection from the mice to observe the degree of metastasis, which indicated that reduced expression of miR-106b-5p inhibited tumor metastasis compared with to that in the control group (Fig. 5C).

The potential binding sequence of miR-106b-5p in the 3’UTR of HPGD was predicted using TargetScan Human 7.2 (Fig. 6A). Luciferase reporter assays revealed that transfection with the miR-106b-5p-mimic reduced the luciferase signals in both the ESCC cell lines with wild-type HPGD 3’-UTR plasmid; however, it did not affect the luciferase activity in the cells transfected with the mutant HPGD3’-UTR plasmid (Fig. 6B), indicating that miR-106b-5p directly targets HPGD. The physical interaction between miR-106b-5p and HPGD was also demonstrated using an RNA pull-down assay (Fig. 6C). Furthermore, HPGD expression was reduced by 30–50% in both the ESCC cells compared to that in Het-1A cells, as evidenced by RT-qPCR and western blotting (Fig. 6D and E). Taken together, these results indicated that HPGD is potential a target gene of miR-106b-5p.

KYSE450 and KYSE510 cells were transfected with HPGD overexpression plasmid (OE-HPGD) and miR-106b-5p-mimic to further explore the underlying mechanism of the miR-106b-5p/HPGD axis. Prior to the functional assays, the transfection efficiency of OE-HPGD was validated by RT-qPCR and western blotting (Supplemental Fig. 2A and B). RT-qPCR confirmed that HPGD expression increased by 4-fold in the cells transfected with OE-HPGD, and decreased by 50% following co-transfection with miR-106b-5p mimic. However, this increased HPGD protein expression was abrogated by the miR-106b-5p-mimic. The miR-106b-5p mimic increased miR-106b-5p expression by 5-fold; however, OE-HPGD had no effect on miR-106b-5p expression (Fig. 7A). Western blot analysis of HPGD protein levels showed similar results as that observed in RT-qPCR (Fig. 7B). The viability and proliferation of cells transfected with OE-HPGD were significantly inhibited, whereas that of cells transfected with miR-106b-5p mimic were markedly enhanced. Furthermore, the miR-106b-5p-mimic relieved the inhibitory effect of OE-HPGD on cell proliferation (Fig. 7C and D). Transfection with miR-106b-5p mimic enhanced adhesion and colony formation ability, whereas OE-HPGD decreased the adhesion and colony formation ability of the cells. Additionally, the miR-106b-5p-mimic reversed the OE-HPGD-induced changes in cell adhesion and colony formation (Fig. 7E and F). In addition, wound healing and Transwell assays showed that overexpression of HPGD inhibited the migration and invasion of the cancer cells, and reversed the ability of miR-106b-5p mimic to promote migration and invasion of the cancer cells (Fig. 7G and H). Flow cytometric analysis showed that upregulation of HPGD partially eliminated the effect of miR-106b-5p overexpression on the cell cycle and apoptosis in KYSE450 and KYSE510 cells, and decreased the proportion of cells in G1 phase and apoptosis, and increased the proportion of cells in G2 phase (Fig. 8A and B). Moreover, results of the caspase-3/7 activity and western blot assays revealed that miR-106b-5p overexpression markedly reduced caspase-3/7 activity and Bax protein level and enhanced Bcl-2 protein level. In contrast, HPGD overexpression increased caspase-3/7 activity and Bax protein level and decreased Bcl-2 protein level. Furthermore, we found that the miR-106b-5p-mimic inhibited the elevated caspase-3/7 activity and Bax protein level, and decreased Bcl-2 protein level induced by OE-HPGD (Fig. 8C and D). Overall, these data indicated that miR-106b-5p enhanced ESCC progression by targeting HPGD.