The miR-93-5p target gene was identified using three well-known bioinformatics prediction algorithms: TargetScan (https://www.targetscan.org/vert_71/), microRNA (http://cbio.mskcc.org/miRNA2003/miranda.html and miRDB (http://www.mirdb.org/miRDB). The starBase database (http://starbase.sysu.edu.cn) was used to identify the potential target sequences of miR-93-5p and EphA4. The expression of miR-93-5p was analyzed in 1,085 breast cancer and 104 normal samples from healthy individuals. The expression of EphA4 and NF-κB was analyzed in 1,104 cancer and 113 normal samples. The expression data of genes in cancer were downloaded from The Cancer Genome Atlas project via the Genomic Data Commons Data Portal (https://portal.gdc.cancer.gov/). The starBase database was also employed to assess the correlations between miR-93-5p expression and EphA4/NF-κB expression.

The MDA-MB-231 cell line (cat. no. MXC234) was purchased from The Cell Bank of Type Culture Collection of The Chinese Academy of Sciences. This cell line was cultured in RPMI 1640 (Corning, Inc.) medium containing 10% fetal bovine serum (FBS; Gibco; Thermo Fisher Scientific, Inc.), 100 U/ml penicillin and 100 µg/ml streptomycin. MCF-7 cells (HTB-22; American Type Culture Collection) were cultured in Dulbecco's modified Eagle's medium (Cytiva) supplemented with 10% FBS, 100 U/ml penicillin and 100 µg/ml streptomycin. The cells were cultured in an atmosphere containing 5% CO₂ at 37°C.

The 2nd generation system was used to package lentiviruses. To induce transient expression of miR-93-5p' (5′-CTGGGGGCTCCAAAGTGCTGTTCGTGCAGGTAGTGTGATTACCCAACCTACTGCTGAGCTAGCACTTCCCGAGCCCCCGG-3′; Quanyang) the cloning vector (pCDH-CMV-MCS-EGFP-EF1-Puro; Quanyang) was constructed using BamHI (cat. no. NEB R0136) and EcoRI (cat. no. NEB R0101) restriction enzymes (both from New England BioLabs, Inc.). The lentiviral plasmid (5 µg; 1 µg/µl) (3.75 µg pH1 packaging plasmid and 1.25 µg pH2 envelope packaging plasmid; Beijing Yingmao Shengye Biotechnology Co., Ltd.) was transfected into 293T (cat. no. MXC006; Shanghai Meixuan Biotechnology Co.) cells using Lipofectamine^® 2000 reagent (Invitrogen; Thermo Fisher Scientific, Inc.) at 37°C for 4 h. The viral supernatant was collected, filtered and concentrated by ultracentrifugation (80,000 × g, 4°C, 2 h) after a 48-h incubation with new culture medium. MDA-MB-231 cells with stable overexpression (OE) of miR-93-5p were established by lentiviral infection. MDA-MB-231 cells were infected with lentivirus at a MOI of 10 with polybrene at 37°C for 24 h. Puromycin (2 µg/ml) was used to select a stable cell line. MDA-MB-231 cells were also infected with the empty lentiviral vector as negative control.

RNAiso PULS (Takara Bio, Inc.) was used to isolate total RNA from MDA-MB-231 cells and TNBC tissues. One-step RT-qPCR was performed using SYBR Premix Ex Taq (Takara Bio, Inc.) and a CFX 96 System (Bio-Rad Laboratories, Inc.). The RT-qPCR thermal cycling conditions were as follows: 50°C for 10 min and 95°C for 5 min, followed by 40 cycles at 95°C for 15 sec and 60°C for 30 sec. The dissociation stage (95°C for 15 sec, 60°C for 1 min and 95°C for 15 sec) was performed after amplification. U6 was used as an internal control for miRNA expression. The sequences of primers used in the present study are listed in Table I. The standard 2^−ΔΔCq method was used to calculate relative RNA abundance (18).

Wild-type (WT) or mutant (Mut) EphA4 was amplified and cloned into a pmirGLO (Beyotime Institute of Biotechnology) vector. Due to the poor survival of MDA-MB-231 cells post-transduction, the luciferase assay was only performed in MCF-7 cells. MCF-7 cells (2×10⁵) were seeded in 96-well plates and were co-transfected with miR-93-5p mimic or miR-negative control (NC) and EphA4 WT or Mut plasmids, using Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions. The miR mimic sequences were as follows: Hsa-miR-93-5p mimic, sense 5′-CAAAGUGCUGUUCGUGCAGGUAG-3′, anti-sense 5′-ACCUGCACGAACAGCACUUUGUU-3′; miR-NC (nonspecific scrambled RNA), sense 5′-UUCUCCGAACGUGUCACGUTT-3′ and anti-sense 5′-ACGUGACACGUUCGGAGAATT-3′. Cells were harvested 48 h after transfection. Finally, the luciferase activity was measured using a Dual-Luciferase Reporter Assay System (Promega Corporation). Using the Renilla luciferase as an internal control, the relative luciferase activity is presented as a ratio of firefly luciferase intensity to Renilla luciferase intensity.

MDA-MB-231 cells (5×10⁷) with or without miR-93-5p OE were injected subcutaneously into 4-week-old BALB/c female mice. All BALB/c nude mice (weight, 18±2 g) were purchased from Silaike Laboratory Animal Co., Ltd. A total of 24 mice were randomly divided into the following four groups (n=6 rats/group: negative control; miR-93-5p OE; radiation therapy (RT) and miR-93-5p OE + RT groups. All mice were housed in a specific pathogen-free sterile environment with a constant temperature of 25°C, under a 12-h light/dark cycle, a relative humidity of 50–70%, with free access to adequate food and water supply. Mice were acclimated for 1 week before tumor injection and tumors of ~1 mm in size were felt ~1 week after inoculation. The body weights of the mice were measured every 3 days. The NC (negative control) group mice were injected by MDA-MB-231 cells transfected with an empty vector with a lentivirus. In the RT and miR-93-5p OE + RT groups, 4 Gy RT was administered three times in 1 week on days 1, 4 and 7 and began when the tumor diameter was 5–7 mm (16 days after injection). The mice were sacrificed by cervical dislocation on the next day after the third irradiation. The humane endpoint was determined as a tumor volume of 1 cm³ and no animals reached the humane endpoint in the present study. All mice were humanely sacrificed 4 weeks after tumor injection. Death was verified by the cessation of breathing and heartbeat. After the mice were sacrificed, the maximum (L) and minimum (W) lengths and weights of tumors were determined. The tumor volume was estimated to be (L × W × W)/2. Hematoxylin and eosin (H&E) staining was performed to evaluate tissue morphology. Animal ethics approval was obtained from the Medical College of Yangzhou University of Animal Ethics Committee (approval no. YXYLL-2021-64).

The collected mouse tumor tissues were fixed in 4% paraformaldehyde for 24 h at room temperature, dehydrated, and immersed in paraffin. Subsequently, 4 µm sections were cut on a glass slide using a Leica RM2145 microtome (Leica Microsystems, Ltd.) and allowed to dry at 40°C for 2 h before staining. Xylene was used to dewax the slides and ethanol was used to rehydrate the slides. H&E staining was performed for 5 min in Mayer's hematoxylin and for 2 min in eosin at room temperature. Images of the slides were captured using a Cewei LW300LFT LED light microscope with a maximum magnification of ×200. H&E staining was used to observe pathological damage under a microscope.

Proteins were extracted from MDA-MB-231 cells using RIPA lysis buffer (Beyotime Institute of Biotechnology) and protein concentration was determined using a BCA Kit (Beyotime Institute of Biotechnology). Proteins (20 µg) were separated by SDS-PAGE on 12% gels and were then transferred to 0.45 µm PVDF membranes (MilliporeSigma). The membranes were blocked with 5% skim milk powder for 2 h at room temperature and incubated overnight at 4°C with the following primary antibodies: EphA4 1:1,000; cat. no. A8346; ABclonal Biotech Co., Ltd.), NF-κB (1:1,000; cat. no. 8242T; Cell Signaling Technology, Inc.), GAPDH (1:5,000; cat. no. 5174; Cell Signaling Technology, Inc.). The membranes were then incubated with a HRP-conjugated goat anti-rabbit IgG secondary antibody (1:5,000; cat. no. 111-035-003; Jackson ImmunoResearch Laboratories, Inc.) at room temperature for 2 h, and the protein bands were detected using an ECL detection system (Bio-Rad Laboratories, Inc.). GAPDH was used as a control. The relative expression levels were calculated as follows: (band intensity of target protein/band intensity of control protein); this was measured using Image-Pro Plus 6.0 software (Media Cybernetics, Inc.).

This prospective study investigated 43 TNBC formalin-fixed and paraffin-embedded (FFPE) samples and normal adjacent tissue (≥0.5 cm away from TNBC tissues) from patients who underwent surgery between January 2018 and March 2021 at the Jiangsu Taizhou People's Hospital (Taizhou, China). FFPE samples were fixed in 4% paraformaldehyde for 24 h at room temperature, dehydrated and embedded in paraffin and cut into 10 µm sections. None of the patients had received RT or chemotherapy before surgery. The mean age was 53.02 years (age range, 29–74 years), and the patients neither drank nor smoked. At the time of diagnosis, the current TNM status of each patient was classified according to the American Joint Committee on Cancer 8th edition (19). The Human Ethics Review Committee of Jiangsu Taizhou People's Hospital approved the present study (approval no. KY 2021-043-01). The requirement for informed consent was waived by the Human Ethics Review Committee of Jiangsu Taizhou People's Hospital.

FFPE sections were sliced to a thickness of 10 µm for subsequent RT-qPCR experiments. Finally, total RNA was extracted from FFPE using the mRNA prep Pure FFPE Kit (cat. no. DP502; Tiangen Biotech Co., Ltd.) according to the instructions provided by the manufacturer.

The present study investigated the relationships between miR-93/EphA4/NF-κB expression and clinicopathological characteristics using Spearman's correlation analysis and χ² test. The mean ± SD of three separate experiments were used to calculate all data. Differences between two groups were assessed using an unpaired Student's t-test, and between paired variables were assessed using a paired Student's t-test. Multiple group comparisons were assessed using one-way ANOVA followed by Bonferroni test to determine significance. P<0.05 was considered to indicate a statistically significant difference. SPSS version 18.0 (SPSS, Inc.) was used for all statistical analyses.

The miR-93-5p target gene was identified using three well-known bioinformatics prediction algorithms: TargetScan, microRNA and miRDB. Based on the results, 18 genes were predicted to be miR-93-5p candidates (Fig. 2A). EphA4 was one of the genes that was considered a potential target for miR-93-5p (Fig. 2B). The online bioinformatics tool starBase demonstrated that miR-93-5p expression was significantly higher in BC samples compared with those in normal controls (P<0.001; Fig. 2C), whereas EphA4 expression was significantly lower (P<0.001; Fig. 2D). The expression levels of NF-κB were also low in BC samples (P<0.01; Fig. 2E). In addition, a significant negative correlation was identified between miR-93-5p and EphA4 expression (P<0.001; Fig. 2F), and the expression of miR-93-5p was negatively correlated with NF-κB (P<0.001; Fig. 2G), whereas a positive correlation was identified between EphA4 and NF-κB (P<0.001; Fig. 2H); however, all correlations were weak (r-values were <0.3/-0.3).

Luciferase reporter vectors containing WT and Mut EphA4 3′-UTRs were generated to further validate the interaction between miR-93-5p and EphA4 (Fig. 3A). A dual-luciferase reporter assay confirmed that miR-93-5p can directly bind to the EphA4 3′-UTR. Transfection of MCF-7 cells with miR-93-5p mimic or miR-NC revealed that cells transfected with the miR-93-5p mimic had significantly increased miR-93-5p expression (miR-93-5p vs. miR-NC; miR-93-5p + EphA4 vs. miR-NC + EphA4; miR-93-5p + EphA4-Mut vs. miR-NC + EphA4-Mut; P<0.0001; Fig. 3B). When miR-93-5p and the p-mirGLO-EphA4-3′UTR were co-transfected into MCF-7 cells, the luciferase signal was significantly lower than that in cells transfected with miR-NC (P<0.001; Fig. 3C). In addition, miR-93-5p suppressed luciferase activity when co-transfected with a luciferase reporter containing Mut EphA4 3′-UTR (Fig. 3C). These findings indicated that the predicted binding site and the designed Mut EphA4 were not suitable. However, it still can be seen that the strength of the binding site in Mut EphA4 3′-UTR was weaker than that in WT EphA4 3′-UTR. These findings indicated that miR-93-5p directly targets the EphA4 3′-UTR.

Our previous study reported that miR-93-5p could improve the radiosensitivity of MDA-MB-231 cells (17); therefore, animal experiments were performed to validate the role of miR-93-5p in vivo. First, miR-93-5p OE vectors were constructed. RT-qPCR confirmed that miR-93-5p was overexpressed in the MDA-MB-231 OE group (Fig. 4A). Subsequently, experiments with tumor xenografts were conducted to examine whether miR-93-5p OE rendered TNBC tumors more susceptible to RT in vivo (Fig. 4B). Notably, the body weight of the tumor-bearing nude mice in each group was not significantly different (Fig. 4C). Xenograft tumor weights were measured after the nude mice were sacrificed (Fig. 4D). Comparing tumor sizes revealed that the OE of miR-93-5p in the MDA-MB-231 cells used to generate xenografts combined with concurrent RT significantly reduced tumor formation compared with that in the NC group (P<0.01; Fig. 4E).

The H&E-stained images are shown in Fig. 5A. H&E staining revealed that the tumor cells in the NC and OE groups grew well, with large and round cell nuclei. Dividing cells and abundant microvessels were also observed in the NC and OE groups. Distinct shrinkage of the cytoplasm, wrinkled cell nuclei, some lysed nuclei and a large necrotic area were visible in the RT group. Furthermore, necrosis in the OE + RT group was much more severe than in RT group.

To further understand the mechanisms underlying the effects of miR-93-5p, downstream regulatory genes were progressively explored. Based on the bioinformatics analysis, it was hypothesized that miR-93-5p targeting EphA4/NF-κB could improve the effects of RT on TNBC cells. The expression levels of miR-93-5p, EphA4 and NF-κB in each group (three tumors/group) are displayed in Fig. 5B. EphA4 and NF-κB expression levels were decreased in the miR-93-5p OE group compared with those in the NC group (P<0.001). Moreover, the expression levels of EphA4 and NF-κB were increased in the miR-93-5p OE + RT group compared with those in the OE group (P<0.001). However, the expression levels of EphA4 and NF-κB were not significantly different between the RT and OE + RT groups. In addition, western blotting was performed and the blots were semi-quantified (Fig. 6A and B). The relative expression levels of EphA4 and NF-κB were significantly lower in the miR-93-5p OE group compared with those in the NC group (P<0.001). Moreover, the relative expression levels of EphA4 and NF-κB were higher in the miR-93-5p OE + RT group compared with those in the OE group (P<0.001). Similar to the RT-qPCR results, the relative expression levels of EphA4 and NF-κB were not significantly different between the RT and OE + RT groups. These findings indicated that RT may prevent the miR-93-5p/EphA4/NF-κB pathway.

RT-qPCR was used to evaluate the expression levels of miR-93-5p, EphA4 and NF-κB in 43 pairs of TNBC and adjacent tissues. The clinicopathological associations between miR-93-5p, EphA4 and NF-κB expression in TNBC are presented in Table II. Notably, the expression levels of miR-93-5p were lower in patients with TNBC with positive lymph nodes (P=0.049). NF-κB expression was higher in patients with a larger tumor size (P=0.049) and later TNM stage (P=0.0498). miR-93-5p expression was upregulated (P=0.009), whereas EphA4 was downregulated in TNBC (P=0.036) (Fig. 7A and B). Correlations between expression levels were determined using the Spearman's test. The analysis revealed an inverse relationship (r=−0.33, P=0.031) between miR-93-5p and EphA4 (Fig. 7C). Although NF-κB expression did not differ between normal and cancerous tissues (P=0.164; Fig. 7D), a positive correlation was identified between NF-κB and EphA4 expression (r=0.336, P=0.034; Fig. 7E). The correlation analysis also found no significant differences between miR-93-5p and NF-κB (data not shown).