Using the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo), five healthy and eight tumor samples were obtained from a CSCC microarray (GSE66359); the healthy samples were used as the control. The limma package in R was used to screen the differentially expressed genes using a |logFC| >1 and a P<0.05 as the screening standards. The upstream miRNAs of EZH2 were predicted using the TargetScan 7.1 (http://www.targetscan.org/vert_71), mirDIP 4.1 (http://ophid.utoronto.ca/mirDIP/index.jsp#r), miRSearch V3.0 (https://www.exiqon.com/miRSearch), and miRDB (http://mirdb.org) databases. The upstream lncRNAs of miR-138-5p were predicted using the RNA22 2.0 database (https://cm.jefferson.edu/rna22).

Between October 2016 and October 2018, cancer tissues and healthy skin tissues were collected from 60 patients with CSCC (33 male; 27 female; age, 53.6±8.1 years; body mass index, 22.61±1.08) admitted to The First Affiliated Hospital of Zhengzhou University (Zhengzhou, China). Patients were excluded if they had incomplete clinical data, mental or consciousness disorders, other primary malignant tumors, autoimmune diseases, serious organic diseases, important organ dysfunction or coagulation dysfunction, if they were pregnant or lactating women, and if they had an allergic constitution or related contraindications.

CSCC cell lines (A431, COLO-16, SCC13, SCL-1, HSC-1, and HSC-5) and the human immortalized keratinocyte HaCaT cell line (all purchased from American Type Culture Collection) were cultured in DMEM supplemented with 10% FBS, 100 U/ml penicillin, and 100 U/ml streptomycin at 37°C and 5% CO₂.

EZH2 cDNA and lncRNA HCP5 were cloned into pcDNA3.1 (Invitrogen; Thermo Fisher Scientific, Inc.) to construct the overexpression vectors. miR-138-5p mimics, mimics negative control (NC), miR-138-5p inhibitor, inhibitor NC (inhi-NC), small interfering RNA (si)-HCP5-1, si-HCP5-2 and si-NC were designed and synthesized by Shanghai GenePharma Co., Ltd. The details are provided in Table I.

si-HCP5-1, si-HCP5-2, si-NC, miR-138-5p mimics and mimics-NC, miR-138-5p inhibitor and inhi-NC were transfected into A431 cells. pcDNA3.1-NC and pcDNA3.1-HCP5 were transfected into SCL-1 cells. miR-138-5p mimics and mimics-NC were transfected into SCL-1 cells and SCL-1 cells overexpressing HCP5. pcDNA3.1-NC and pcDNA3.1-EZH2 were co-transfected into SCL-1 cells and A431 cells with si-HCP-1 and si-HCP5-2; SCL-1 cells transfected with pcDNA3.1-EZH2 were pre-treated with STAT3/VEGFR2 pathway inhibitor AG-490 (10 nM; AmyJet Scientific Co., Ltd.) for 24 h at 37°C (17). Briefly, cells to be transfected were seeded in 6-well plates (1.0×10⁵ cells/well) and grown overnight for 18 h. Cells at 80% confluence were transfected with 100 pmol siRNAs, miRNA mimics, miRNA inhibitors or the respective controls using Lipofectamine^® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) following the manufacturer's instructions. After 48 h of incubation at 37°C, cells were collected for subsequent experiments.

Total RNA was obtained by the one-step method using TRIzol^® (Invitrogen; Thermo Fisher Scientific, Inc.), and high-quality RNA was confirmed using ultraviolet analysis (A260/280) and formaldehyde denaturation electrophoresis. qPCR was performed following the manufacturer's instructions of TaqMan™ Sample-to-SNP™ kit (cat. no. 4403313; Applied Biosystems; Thermo Fisher Scientific, Inc.). PCR primers were designed and synthesized by Sangon Biotech Co., Ltd. and are listed in Table II. GAPDH was used as the internal reference for lncRNAs and U6 was used as the internal reference for miRNAs; expression data were normalized to GAPH or U6. Amplification and dissolution curves were confirmed after the reaction. The relative expression was calculated using the 2^−ΔΔCq method (18).

Cells were lysed for 30 min using a cold radio-immunoprecipitation assay buffer containing a protease inhibitor mixture (Sigma-Aldrich, Merck KGaA). Lysates were centrifuged at 4°C and 16,000 × g for 20 min, and the supernatant was obtained. Protein concentration was determined using the bicinchoninic acid kit (Beyotime Biotechnology Co., Ltd.). Proteins were separated by 10% SDS-PAGE and transferred onto polyvinylidene fluoride membranes. The membranes were blocked with 5% skim milk for 2 h at room temperature. After incubation with the primary antibodies (Table III) overnight at 4°C, the membranes were incubated with goat anti-rabbit IgG (1:2,000; cat. no. ab205718; Abcam) or goat anti-mouse IgG (1:2,000; cat. no. ab205719; Abcam) secondary antibody for 1 h at room temperature, were developed using the Immobilon Western Chemiluminescent HRP substrate (cat. no. WBKLS0100; MilliporeSigma), and visualized using an imager (Bio-Rad, Inc.). Protein expression levels were semi-quantified using ImageJ Pro Plus 6.0 [National Institutes of Health (NIH)].

The CCK-8 assay was used to determine cell viability following the manufacturer's instructions (cat. no. C0038; Beyotime Institute of Biotechnology). CSCC cells with different transfections were seeded onto 96-well plates at a density of 5,000 cells/well. After incubation at 37°C for 2 h, the media was replaced with 100 µl CCK-8 solution (90 µl serum-free DMEM and 10 µl CCK-8) at 0, 12, 24, 36 and 48 h incubation. The absorbance at 450 nm was then measured for each well.

A 24-well Transwell system (8 µm pore size membrane; Corning, Inc.) and a filter membrane pre-coated with Matrigel (BD Biosciences) were used to evaluate cell invasion. Suspensions of the transfected CSCC cells (1×10⁵ cells/ml) were prepared in serum-free DMEM (0.1% bovine serum albumin) and placed onto the apical chamber of the Transwell insert; the basolateral chamber was filled with DMEM containing 10% FBS. The Transwell chamber was incubated for 24 h at 37°C and 5% CO₂. Subsequently, the filter membrane was removed and rinsed with PBS, fixed with 0.5% glutaraldehyde for 30 min at room temperature and stained with crystal violet for 30 min at 37°C and 5% CO₂. Five random fields of vision were selected for cell counting under a light microscope (magnification, ×200).

Transfected CSCC cells were detached and collected using 0.25% trypsin, and then resuspended to 2×10⁵ cells/ml in DMEM containing 10% FBS. The cells were seeded onto 6-well plates (2 ml/well) and cultured at 37°C and 5% CO₂ for 12 h. Following cell adherence, a straight line perpendicular to the center was scratched on the cell monolayer using a pipette tip. Subsequently, the floating cells were washed away using PBS, and cells were continuously cultured for 12 h. Images were captured under a phase contrast microscope at 0 and 12 h incubation. ImageJ Pro Plus 6.0 (NIH) was utilized to analyze the scratch width of cells in different groups and time periods. Cell migration=scratch width at 0 h-scratch width at 12 h.

When the transfected cells were cultured to 85% confluence, the suspended cells in the culture supernatant were collected into the centrifuge tube. The adherent cells were detached with 0.25% trypsin, washed three times with PBS, and collected into the centrifuge tube in the previous step. Next, the cells were centrifuged at 4°C and 150 × g for 5 min. The cells were resuspended with cold PBS and centrifuged for 5 min at 150 × g at 4°C to collect cell precipitates. This procedure was repeated twice to remove residual trypsin. Following centrifugation, the cells were resuspended in PBS and adjusted to 1×10⁶ cells/ml. According to the instructions of the Annexin V-FITC Apoptosis Detection kit (BD Biosciences), 500 µl cell suspension was added into the centrifuge tube at 4°C and 150 × g for 5 min, followed by the addition of 195 µl Annexin V Binding Buffer for resuspension. Then, 5 µl Annexin V-FITC and 10 µl PI were added into each tube, mixed well and incubated for 10 min at room temperature in the dark. Cells were evaluated using a MoFlo Astrios EQ flow cytometer (Beckman Coulter, Inc.) admd t Flowjo 7.6 software (FlowJo LLC) was used for data analysis. Total apoptotic rates were calculated as the total early- and late-stage apoptosis.

LC3 double-labeled autophagy adenovirus [Monomeric red fluorescent protein (mRFP)-green fluorescent protein (GFP)-LC3] was purchased from Hanbio Biotechnology Co., Ltd. Transfected cells were cultured in 24-well plates (1×10⁵ cells/ml) and incubated with mRFP-GFP-LC3 adenovirus (multiplicity of infection, 100) for 2 h at 37°C, when cell confluence reached 70-80%. The cells were washed with PBS and cultivated in DMEM containing 10% FBS overnight. The transfection efficiency of mRFP-GFP-LC3 adenovirus was evaluated using a BX51 fluorescence microscope (Olympus Corporation). ImageJ Pro Plus 6.0 software (NIH) was used for image analysis and counting the number of GFP and RFP fluorescent punctae. In the merge image, autophagosomes are represented by yellow dots and autolysosomes are represented by red dots.

A431 cells in logarithmic growth phase were collected and centrifuged at 4°C and 150 × g for 5 min and then mixed with Pierce IP Lysis Buffer (Thermo Scientific, Inc.). After lysis on ice for 30 min, the cells were centrifuged at 4°C and 1,200 × g for 30 min to collect the supernatant and stored on ice. RNA-RNA interaction was assessed by RNA pull-down using Pierce Magnetic RNA-Protein Pull-Down kit (Thermo Fisher Scientific, Inc.) following the manufacturer's protocol. Biotinylated HCP5 and NC probes (50 pmol; made using the Pierce RNA 3′ End Desthiobiotinylation Kit; cat. no. 20163; Thermo Fisher Scientific, Inc.) were dissolved in washing/binding buffer (contained in the pull-down kit) and cultured with streptavidin-coupled magnetic beads for 2 h at 4°C. Next, the probes were added to the cell lysates for 2 h to elute the RNA complex conjugated with the magnetic beads, and miR expression was determined using RT-qPCR.

The binding sites of lncRNA-HCP5 and miR-138-5p, and for EZH2 and miR-138-5p were predicted using RNA22 v2 (https://cm.jefferson.edu/rna22/Interactive) and TargetScan (http://www.targetscan.org) online databases. The HCP5 fragment containing the binding site of miR-138-5p was cloned into the pmirGLO oligosaccharide enzyme vector (Promega Corporation) to construct the pmirGLO-HCP5-wild-type (WT), and the pmirGLO-HCP5-mutant type (MUT) vector was constructed using the mutant binding site of miR-138-5p. The pmirGLO-EZH2-WT and pmirGLO-EZH2-MUT vectors were constructed using the same methods. The sequences were synthesized by Shanghai GenePharma Co., Ltd. The constructed vectors (50 ng) were transfected into A431 cells (at 80-90% confluence), and then co-transfected with miR-138-5p or miR-NC (20 nM) according to the instructions of Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.). After incubation at 37°C for 48 h, the luciferase activity was evaluated using a Dual Luciferase Reporter Assay System (Promega Corporation), and the relative activity was determined as the ratio of firefly luciferase activity to Renilla luciferase activity.

A total of 20 specific pathogen-free BALB/c nude male mice (weight, 20±2 g; age, 6-8 weeks) were provided by Beijing Vital River Laboratory Animal Technology Co., Ltd. [license SYXK (Beijing) 2015-0001], numbered according to body weight and allocated into three groups (n=10 mice/group) using the random number method. All mice were reared in an animal room at 25°C temperature and 50-60% humidity under a 12 h light/dark cycle. All mice were allowed free access to food and water. In the A431 group, each mouse was injected subcutaneously (on the back near the right armpit) with 2×10⁶ A431 cells. In the si-HCP5-1 group, each mouse was injected subcutaneously with 2×10⁶ A431 cells transfected with si-HCP5-1. Tumor volumes were measured every 3 days, and mice were euthanized using pentobarbital (>200 mg/kg intraperitoneal injection) 28 days later (19,20). Death was confirmed by examining pupillary dilation and complete cardiac arrest, and the tumors of five random mice from each group were embedded in paraffin and the tumors from the remaining five mice were ground into a homogenate for subsequent experiments.

A Histostain-Plus Kit (cat. no. 85-6643; Invitrogen; Thermo Fisher Scientific, Inc.) was used for immunohistochemical staining. Tumor tissue was fixed with 4% formaldehyde for 6 h at room temperature and embedded in paraffin. The paraffin-embedded tumor tissue was cut into 3 µm sections; the sections were baked at 45°C for 3 h. After dewaxing with xylene and rehydration, the sections were treated with 3% H₂O₂ and kept at room temperature for 10 min to eliminate the activity of endogenous peroxidase. After washing with PBS + 0.05% Tween-20 (PBST) three times, the sections were placed in a heat-resistant glass container, submerged in the citrate sodium buffer (10 mM, pH 6.0), heated to boiling in the microwave for antigen retrieval and cooled naturally (5-10 min). These steps were repeated three times and then the sections were cooled down to room temperature. The sections were blocked with 10% goat serum (reagent A in the kit) at 37°C for 1 h in a humidified chamber and incubated with anti-Ki67 primary antibody (1:500; cat. no. ab15580; Abcam) at 4°C overnight. The next day, sections were washed three times with PBST and incubated with biotinylated secondary antibody working solution (reagent B in the kit), incubated at 37°C for 15 min and washed three times with PBST times to wash away the unbound antibody. The sections were subsequently incubated with horseradish peroxidase-conjugated streptavidin working solution (reagent C in the kit), incubated in a wet box at 37°C for 15 min, and washed with PBST three times. The percentage of Ki67-positive cells was counted using a CellInsight laser confocal microscope (Thermo Fisher Scientific, Inc.) and analyzed using ImageJ-Pro Plus 6.0 (NIH).

Tumor sections were treated and stained with TUNEL reaction mixture in accordance with the manufacturer's instructions (Wanleibio Co., Ltd.). DAPI solution was added to the media (1:10) to stain the nuclei, and the cells were cultured at 37°C for 15 min. Sections were washed with PBS, observed under a fluorescence microscope, and an anti-fluorescence quenching agent (Wanleibio Co., Ltd.) was added. Within 24 h, images of the sections were captured under a fluorescence microscope; the apoptotic cells were counted using ImageJ Pro Plus 6.0 (NIH), and the results were analyzed.

Statistical analyses were conducted using SPSS v21.0 (IBM Corp.). The Kolmogorov-Smirnov test confirmed that the data were normally distributed. Data are presented as the mean ± standard deviation. Paired Student's t-test or Mann-Whitney U test was used for comparisons between two groups, whereas one-way ANOVA was used to compare multiple groups; Tukey's multiple comparisons test was applied for pairwise comparisons after ANOVA analyses. The P-value was obtained using a two-tailed test, and a P<0.05 was used to indicate a statistically significance difference.

Through the differential analysis of GSE66359 microarray, 1,252 genes with significant differential expression were obtained (Fig. 1A). Among these genes, EZH2 was screened based on a literature search that suggested EZH2 can be used as a diagnostic index to differentiate squamous cell carcinoma from normal skin (14). EZH2 expression was significantly increased in CSCC compared with normal tissue (Fig. 1B), which was consistent with a previous report (14). EZH2 mRNA and protein expression levels in 60 patients with CSCC collected from The First Affiliated Hospital of Zhengzhou University also verified these results (Fig. 1C and D, respectively). To further understand the mechanism of EZH2 in CSCC, the upstream regulatory miRNAs of EZH2 were predicted using four online databases. The combined prediction results identified five miRNAs that may target EZH2 (Fig. 1E). RT-qPCR was used to determine the expression levels of the five miRNAs, which revealed that the relative miR-138-5p expression was the lowest in CSCC compared with its expression in normal tissue (Fig. 1F). miR-138-5p has been shown to inhibit tumor development (21,22), but its role in CSCC remains unclear. Using the RNA22 database, lncRNAs that may interact with miR-138-5p were predicted. The prediction results of RNA22 were intersected with the upregulated gene expression results obtained from GSE66359 microarray (Fig. 1G), and eight putative lncRNAs were identified that may regulate miR-138-5p; these lncRNAs were shown to be highly expressed in CSCC compared with normal adjacent tissue (Fig. 1H). Furthermore, the expression of these eight lncRNAs were screened in the GSE66359 microarray (Table IV) and found that HCP5 had the greatest upregulation in CSCC (|logFC|=1.8) and the highest relative expression levels in patient CSCC compared with the normal adjacent tissue (Fig. 1H). These data suggested that lncRNA HCP5 may regulate EZH2 expression through miR-138-5p.

The expression levels of lncRNA HCP5 were detected in cultured CSCC cell lines and the HaCaT human immortalized keratinocyte cell line. A431 cells exhibited the highest relative HCP5 expression and SCL-1 cells exhibited the lowest HCP5 expression levels among the CSCC cell lines (Fig. 2A); therefore, these two lines were used for subsequent experiments. Two siRNAs against HCP5-1 were transfected into A431 cells, of which si-HCP5-1 exhibited the best transfection efficiency and was selected for the following experiment (Fig. 2B). An HCP5 overexpression vector was transfected into SCL-1 cells, and the transfection efficiency was verified using RT-qPCR (Fig. 2C). Silencing of HCP5 expression resulted in significant decreases in A431 cell viability, invasion and migration, whereas SCL-1 cells overexpressing HCP5 exhibited opposite trends (Fig. 2D-F).

LncRNAs can regulate apoptosis and autophagy of CSCC cells (23). Therefore, autophagy and apoptosis of CSCC cells were measured following HCP5 knockdown and overexpression. After HCP5 silencing, the number of autophagosomes (yellow punctae) decreased (Fig. 2G), the LC3II/LC3I protein expression ratio decreased and the expression of p62 protein increased (Fig. 2H). In addition, the apoptotic rate significantly increased (Fig. 2I), apoptotic protein Bax expression level increased and Bcl-2 expression level significantly decreased (Fig. 2J). Conversely, the number of autophagosomes in CSCC cells increased following HCP5 overexpression, the protein expression levels of LC3 II/LC3I increased, whereas the expression of p62, the apoptotic rate and Bax protein expression decreased; Bcl-2 expression increased significantly. In summary, HCP5 silencing inhibited the malignant behaviors and autophagy of CSCC cells and promoted apoptosis, whereas overexpression of HCP5 promoted the malignant behaviors and autophagy of CSCC cells and inhibited apoptosis.

Combined with the microarray analysis results, the binding sites of lncRNA HCP5 and miR-138-5p as well as EZH2 and miR-138-5p were predicted using online databases RNA22 and TargetScan. Dual luciferase reporter assays were used to verify the binding relationships (Fig. 3A-B). In addition, lncRNA HCP5 significantly enriched miR-138-5p using an RNA pull-down assay (Fig. 3C). As shown in Fig. 3D-F, after HCP5 silencing in A431 cells, miR-138-5p expression levels increased, whereas the EZH2 mRNA and protein expression levels decreased significantly; the opposite trends were observed in cells overexpressing HCP5. After miR-138-5p mimics transfection, EZH2 expression levels decreased, whereas the results of miR-138-5p inhibitor transfection were the opposite. These results indicated negative regulatory relationships between HCP5 and miR-138-5p, and between EZH2 and miR-138-5p. HCP5 may function as a ceRNA to compete with EZH2 to bind to miR-138-5p.

miR-138-5p inhibitor was co-transfected into A431 cells with si-HCP5 to conduct a functional rescue experiment (Fig. 4A). Cell viability was reduced upon upregulation of miR-138-5p or downregulation of HCP5, whereas downregulation of miR-138-5p partially counteracted the effect of HCP5 silencing on CSCC viability (Fig. 4B). HCP5 silencing and upregulation of miR-138-5p significantly decreased the LC3II/LC3I ratio and increased the protein expression of p62 in CSCC cells; downregulation of miR-138-5p partially reversed the inhibitory effects of HCP5 silencing in CSCC cells (all P<0.05; Fig. 4C). Additionally, si-HCP5 or miR-138-5p mimics alone significantly increased the apoptotic rate, increased the protein expression levels of Bax and decreased Bcl-2; downregulation of miR-138-5p partially abolished the promoting effect of HCP5 silencing on cell apoptosis (Fig. 4D-E). These results indicated that upregulation of miR-138-5p may inhibit autophagy and promote apoptosis of CSCC cells.

An EZH2 overexpression vector was transfected into CSCC cells (Fig. 5A and B). EZH2 overexpression increased the viability of SCL-1 cells and offset the inhibitory effect of silencing HCP5 on A431 cell viability (Fig. 5C). Western blotting analysis revealed that overexpression of EZH2 significantly increased the protein expression ratio of LC3II/LC3I, decreased the expression of p62 in SCL-1 cells and counteracted the inhibitory effects of HCP5 silencing on the autophagy of A431 cells (Fig. 5D). Additionally, overexpression of EZH2 significantly decreased the level of apoptosis (Fig. 5E), decreased the protein expression levels of Bax and increased the expression of Bcl-2 in SCL-1 cells (Fig. 5F), and counteracted the effects of HCP5 silencing in of A431 cells.

EZH2 was previously reported to serve a regulatory role in head and neck squamous cell carcinoma through the STAT3/VEGFR2 pathway (24). Therefore, whether the STAT3/VEGFR2 pathway is involved in CSCC development was investigated. Western blotting results revealed that p-STAT3/total (t)-STAT3 ratio and VEGFR2 protein expression levels in CSCC cells overexpressing EZH2 were significantly increased compared with the control (Fig. 6A). Co-treatment with AG-490, an inhibitor of the STAT3/VEGFR2 pathway, significantly inhibited the levels of p-STAT3/t-STAT3 and VEGFR2 in EZH2-overexpressing CSCC cells (Fig. 6A). In addition, AG-490 co-treatments also resulted in decreased autophagy, as determined by western blot analysis of autophagy-related proteins (Fig. 6B), and the apoptotic rate increased (Fig. 6C). Taken together, EZH2 may regulate autophagy and apoptosis in CSCC cells through the STAT3/VEGFR2 pathway.

si-HCP5 or si-NC A431 cells were subcutaneously injected into mice. Tumor volume and weight measurements indicated that tumor growth was limited after silencing HCP5 compared with the control group (Fig. 7A-C). In addition, the protein expression levels of EZH2, p-STAT3/t-STAT3 ratio and VEGFR2 were decreased; autophagy-related proteins LC3II/LC3I decreased significantly, whereas p62 protein expression increased (Fig. 7D). After HCP5 silencing, the Ki67-positive rate was significantly reduced (Fig. 7E), and TUNEL staining indicated that apoptosis was increased (Fig. 7F). These data suggested that HCP5 may promote EZH2 expression, increase proliferation and autophagy, and reduce apoptosis in CSCC cells through the STAT3/VEGFR2 pathway.