The differential expression analysis of E2F1 and E2F3 between normal and cancer tissues were obtained from an Oncomine dataset (www.oncomine.org) by setting the following parameters: Gene, E2F1 or E2F3; analysis type, cancer vs. normal analysis; cancer type, breast carcinoma; data type, mRNA. Consequently, as indicated in Table S1, E2F1 (n=618) and E2F3 (n=295) samples from American, British and Canadian patients with invasive breast carcinoma and breast carcinoma were selected for a meta-analysis. The Kaplan-Meier plotter (http://kmplot.com/analysis) was used to construct Kaplan-Meier survival curves of E2F1, E2F3 and miR-34a expression in patients with breast cancer. Patients with high or low gene expression were divided by median expression level for E2F1/3 or by best cutoff value for miR-34a. Cutoff values were 216 for E2F1, 381 for E2F3 and 12.98 for miR-34a.

Human breast cancer cell lines T-47D, MDA-MB-231 and normal breast cell MCF-10A were purchased from the American Type Culture Collection (ATCC). The T-47D cell line is termed an ‘invasive ductal carcinoma’ on ExPASy (www.expasy.org). Cells of T-47D and MDA-MB-231 were cultured in RPMI-1640 Medium (ATCC) or Leibovitz's L-15 medium (LLM; ATCC) respectively, with 10% fetal bovine serum (ATCC); MCF-10A cells were cultured in MEBM (Lonza, Inc.) which was obtained by adding cholera toxin (Sigma-Aldrich; Merck KGaA) at a final concentration of 1 ng/ml into mammary epithelial growth medium (cat. no. CC-3150; Lonza, Inc.). All cells were cultured in a humidified incubator at 37°C, at 5% CO₂.

T-47D and MDA-MB-231 cells were seeded in a 96-well plate at a concentration of 3×10³ cells/100 µl (~90% in confluence), and mixed with 10 µl Opti-MEM media (Thermo Fisher Scientific, Inc.), 0.3 µl Lipofectamine RNAiMAX (Thermo Fisher Scientific, Inc.) and 0.3 µl RNA oligos of either 10 mM miR-34a mimic or 10 mM control mimic. Both miR-34a mimic (5′-UGGCAGUGUCUUAGCUGGUUGU-3′) and control mimic oligos (cat. no. 51-01-19-09) were purchased from Integrated DNA Technologies, Inc. The medium was replaced with a fresh medium on the next day of the seeding.

CMV-Firefly luciferase-IRES-Puro lentivirus (Cellomics; Thermo Fisher Scientific, Inc.) was transfected into both cells (T-47D and MDA-MB-231) with a multiplicity of infection of 5, after 8 h treatment with 6 µg/ml polybrene (Cellomics Technology, Inc.) in complete growth medium, at 37°C. Cell selection was conducted for ≥14 days with 1 µg/ml puromycin, and a stable fluorescence signal was confirmed using the 96 Microplate Luminometer (Promega Corporation).

Following the manufacturer's instruction, a cell proliferation MTS assay (Promega Corporation) was performed on miR-34a-treated cells or controls at different incubation time points (48, 72, 96 and 120 h), in triplicate, cells were incubated at 37°C. A Microplate Spectrophotometer (Biotek Instruments, Inc.) was used to detect the absorbance at the wavelength of 450 nm. The proliferation inhibition rate was then calculated as formula: Inhibition rate=[1-Absorbance of treated sample (or mock sample)/Absorbance of control sample (NC)] ×100.

In total, 2×10³ cells of either T-47D or MDA-MB-231 cells were used for colony formation assays in 6-well tissue culture plates. miR-34a-treated cells or control cells were incubated for 10 days before each well was gently washed with 1X PBS, and the cells were fixed using 4% paraformaldehyde (FD NeuroTechnologies, Inc.) for 15 min at room temperature and stained using crystal violet (0.1%; Sigma-Aldrich; Merck KGaA) for 15 min at room temperature. The number of colonies with >20 cells was counted.

In total, ~1×10⁶ miR-34a treated cells or controls were seeded in each well, and when the cells reached 90% confluence, a wound scratch was gently made using a 100 µl pipette tip. Cells were then cultured in 2 ml RPMI-1640 or Leibovitz's L-15 medium with 0.1% FBS at 37°C for 48 h and same medium was replaced each 24 h, as previously described (21). Cells were imaged at 0, 24 and 48 h post-wound for the wound closure measurement.

In total, 40 µl Matrigel solution [20 µl Matrigel (Corning, Inc.) and 20 µl serum-free medium mixed in 4°C atmosphere) was coated in the upper layer of each culture insert. After 1 h pre-coating of matrigel at 37°C, 1×10⁴ cells in 60 µl medium (RPMI-1640 medium for T-47D and Leibovitz's L-15 medium for MDA-MB-231) with 0.1% FBS were then seeded. Below the cell permeable membrane, 600 µl of 10% FBS medium (RPMI-1640 medium for T-47D and Leibovitz's L-15 medium for MDA-MB-231) was added to each chamber. After incubating T-47D (24 h) and MDA-MB-231 (48 h) at 37°C and 5% CO₂ atmosphere (25), migrated cells were fixed with 4% paraformaldehyde followed by crystal violet staining, both were performed at room temperature for 20 min, respectively. Subsequently, the surface of the upper layer of the membrane was gently cleaned using cotton swabs, the cells in three different fields of view were counted under an inverted microscope (Olympus Corporation, magification, ×100) and the average sum of cells was calculated.

After the cells (T-47D and MDA-MB-231) with the luciferase reporter system were transfected with miRNA, a 3D spheroid formation model was constructed using a hanging-drop approach (~200 cells per drop of 30 µl MammoCult™ human medium) (Stemcell Technologies, Inc.). One set was used for imaging at regular intervals between 24 and 120 h incubation time by using inverted microscope (Olympus Corporation; magnification, ×4; Scale bar, 100 µm), and another set was used for in vitro bioluminescence signal determination by transferring to a 96-well plate in the presence of D-luciferin (150 µl/ml) (PerkinElmer, Inc.) at each time point (24, 48, 72, 96 and 120 h) in triplicate. The software of ImageJ (v. 1.52a; imagej.nih.gov/ij) was used for counting cells. The average proliferation inhibition rate was calculated.

Total RNA was extracted from T-47D and MDA-MB-231 cells using the RNeasy mini kit (Qiagen, Inc.), according to the manufacturer's instructions. The concentration and purity of total RNA were determiend using an Epoch microplate spectrophotometer (Biotek Instruments, Inc.). cDNA was prepared using an AffinityScript multi temperature cDNA synthesis kit (Agilent Technologies, Inc.) following the manufacturer's protocol. The expression of E2F1, E2F3 and GAPDH genes was determined using the SYBR Green-based master mix (Qiagen) on a 7500 Fast Real-time PCR system (Thermo Fisher Scientific, Inc.). In addition, the relative expression level of miR-34a was tested in T-47D and MDA-MB-231 cells in different treatment groups, respectively. All the primer sequences used in this study are described in Table S2 (21). Each sample was analyzed in triplicate, and the qPCR reaction conditions included one cycle of 95°C for 15 min, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. The dissociation curve was run after the PCR amplification in each assay. GAPDH was used as an internal control for mRNA expression, and U6 was used as the reference gene for miR-34a expression. The relative expression levels of E2F1 and E2F3 mRNA, and miR-34a are calculated as a fold change using the 2^−∆∆Cq method (26).

The CaspACE™ assay was conducted following the manufacturer's protocol. Briefly, 2×10⁶ cells were treated with either 10 µmol/l miR-34a or control mimic for 72 h as the induced apoptosis group, and 3 ml Z-VAD-FMK inhibitor was added to the inhibited apoptosis groups. The mock groups were regarded as a normal control (NC). After 16 h incubation at 37°C, the cell supernatant fractions were harvested using centrifugation for CASP3 activity measurement at 450 × g for 10 min at 4°C (27). The protein concentration of each sample was determined using the bicinchoninic protein assay (Thermo Fisher Scientific, Inc.), and the pNA Calibration Curves were constructed using a colorimetric assay system. CASP3 specific activity (SA) was calculated as the following formulae:

X=[ΔA-(Y intercept of pNA std.curve)]/(incubation time in hours) x[100 µl (sample volume)]/[(slope of pNA std.curve (A405/pmol/µl)].

ΔA=induced apoptosis sample A405-inhibited apoptosis sample A405.

Data are presented as mean ± SD. One-way ANOVA was performed for group comparison, with post-hoc Bonferroni's correction used for multiple comparisons, as appropriate. The two-tailed unpaired Student's t-test was used for the comparison of differences between two groups. Normalized miRNA-seq and RNA-seq datasets of TCGA breast cancer (https://portal.gdc.cancer.gov/) were downloaded and combined to perform Spearman correlation analysis between the expression of E2F1 and E2F3, miR-34a and E2F1, and miR-34a and E2F3. P<0.05 was considered to indicate a statistically significant difference, or P<0.05/m (m, number of comparisons in Bonferroni correction; two-sided). All statistics and figures were generated using GraphPad Prism 8.0 software (www.graphpad.com). A random-effects model of meta-analysis was performed for the fold-change in expression of E2F1 and E2F3 between cancer and normal tissues using R package 3.5 (https://www.r-project.org).

The results of a random-effects model of meta-analysis revealed the significantly upregulated expression of E2F1 [Log2 (fold-change)=0.73; 95% confidence interval (CI), 0.20–1.25; fold-change=1.66, 95% CI, 1.15–2.38) (Fig. 1A), and E2F3 [log2 (fold-change)=0.83; 95% CI, 0.63–1.04; fold-change=2.46; 95% CI, 1.55–2.06] (Fig. 1B).

According to the binary category of either miR-34a, E2F1 or E2F3 expression level, Kaplan-Meier survival curve analysis was performed. Patients with high expression of either E2F1 [hazard ratio (HR), 1.54; 95% CI, 1.24–1.92, P=9.7×10⁻⁵; Fig. 1C] or E2F3 (HR, 1.46; 95% CI, 1.18–1.82; P=5.5×10⁻⁴; Fig. 1D) exhibited a less favorable prognosis compared with patients with low expression. By contrast, patients with high expression level of miR-34a exhibited a significantly longer survival time compared with patients with low expression (HR, 0.8; 95% CI, 0.65–0.98; P=0.028; Fig. 1E).

The results displayed in T-47D and MDA-MB-231 cells that the miR-34a level was significantly higher in the miR-34a mimic groups, compared with the control mimics (P<0.001), demonstrating the efficiency of miR-34a transfection (Fig. 2A). Moreover, at the time points of 72, 96 and 120 h, a significantly decreased cellular viability was observed in both cell lines in the miR-34a group (P<0.05) compared with the control mimic (Fig. 2B and C). However, the inhibition rate profile appeared different between T-47D (an invasive ER-positive ductal carcinoma) and MDA-MB-231 (triple negative breast cancer with expression of features associated with mammary cancer stem cells of CD44⁺/CD24^−/low phenotype) (28). For T-47D, the cell viability in the miR-34a-transfected group displayed a statistically significant difference when compared with the control group from 72 h (14.04±1.58%; P=4.78×10⁻⁴) (Fig. 2B) and continued decreasing throughout the whole experiment period. By contrast, the inhibition rate of MDA-MB-231 in the miR-34a group reached a maximum at 72 h, then decreased at 96 and 120 h (Fig. 2C).

A significant decrease in cell colonies was observed in the miR-34a-transfected group with a relative efficiency of 68.45±1.93% (P=2.07×10⁻⁵) for T-47D, and 79.45±5.19% (P=4.99×10⁻³) for MDA-MB-231, compared with their respective NC groups (control mimic) (Fig. 2D and E).

In both cell lines transfected with miR-34a, a decreased migration capacity and wound healing ability, was observed compared with the control and mock groups (Fig. 3A and B). For T-47D, at 24 h, the average wound gap width in the miR-34a group was 90.01±1.25% compared with the control and mock groups which had gap widths of 73.54±2.25% (P=8.31×10⁻⁴) and 75.88±3.72% (P=6.53×10⁻³), respectively. At 48 h, the width in the miR-34a group was 82.23±1.22% while its counterpart in the NC group dropped to 31.21±6.20% (P=1.57×10⁻³). For MDA-MB-231 cell, 75.80±5.16% (24 h) and 60.82±6.38% (48 h) of the initial width in the miR-34a group compared with 54.86±5.90% (24 h; P=0.019) and 31.21±6.20% (48 h; P=9.29×10⁻³) in the NC group, respectively (Fig. 3C and D).

Similarly, the miR-34a group displayed a decreased invasive ability compared with the control and mock groups (Fig. 4A and C). In T-47D cells, the average number of invaded cells in the NC group was 258±11.22, compared with 177±8.04 of the miR-34a group (P=1.15×10⁻³). Additionally, in MDA-MB-231 cells the number of invaded cells was 239±14.24 in the NC group, compared with 155.67±6.60 in the miR-34a group (P=1.68×10⁻³; Fig. 4B and D).

The dynamic changes of 3D spheroid formation are exhibited in Fig. 5A and D. In T-47D cells, the relative cell cross-sectional area of the miR-34a group increased by 127.08±15.90%, which was significantly different from the control (168.93±3.08%; P=0.022) and mock group (175.79±5.34%; P=0.015) at 72 h. This trend remained until 120 h at which an area of 203.65±12.70% in the miR-34a group was reached, compared with 250.89±10.4% in the NC (P=0.015) and 248.20±13.69% in the mock group (P=0.028) (Fig. 5B). For MDA-MB-231, significant differences in the average 3D spheroid area between the miR-34a group and the control and mock group were observed at 96 and 120 h (Fig. 5E). Again, the bioluminescence test of 3D spheroid cell formation revealed similar results to the cross-section area assay. The inhibition rates of T-47D cells in the miR-34a-transfected group were 21.60±3.99% (P=3.10×10⁻³) at 72 h, and 43.308±2.24% (P=6.5×10⁻³) at 120 h (Fig. 5C), and the rates for MDA-MB-231 were 26.61±3.20% at 96 h (P=2.15×10⁻³) and 22.38±2.00% (P=0.011) at 120 h (Fig. 5F).

RT-qPCR results revealed that both E2F1 and E2F3 expression levels were significantly higher in T-47D and MDA-MB-231 cell lines compared with the normal breast cell line MCF-10A, as reported previously (29,30). The expression levels of E2F1 and E2F3 were 1.59-fold and 1.67-fold larger in T-47D compared with in MCF-10A cells (P<0.001), respectively. The expression levels of E2F1 and E2F3 were 1.81-fold and 1.5-fold larger in MDA-MB-231 compared with MCF-10A (P<0.001), respectively (Fig. S1). Transfection with the miR-34a mimic significantly downregulated E2F1 and E2F3 expression in both cell lines in both 2D and 3D culture systems. For T-47D cells, the expression level change of E2F1 following transfection with a miRNA mimic was a 0.44-fold decrease in 2D (P<0.001) and 0.48-fold decrease in 3D (P<0.001) culture systems compared with the mimic control group. Furthermore, in MDA-MB-231 cells transfected with a miR-34a mimic, the E2F1 expression level change was 0.31-fold decrease in 2D conditions (P<0.001) and a 0.46-fold decrease in 3D cultured system (P=1.59×10⁻³), compared with the mimic control group. Moreover, in T-47D cells transfected with miR-34 a mimic, the relative expression level of E2F3 in 2D was 0.23-fold (P<0.001) and 0.54-fold decrease in the 3D group (P=1.25×10⁻³), compared with the mimic control. As for MDA-MB-231 miR-34a cells transfected with the miR-34a mimic, the relative expression level of E2F3 in 2D conditions was a 0.40-fold (P<0.001) and in 3D it was a 0.38 fold decrease compared with the mimic control group (P<0.001) (Fig. 6A and D).

The CASP3 activity in the miR-34a group was significantly higher compared with either the inhibited apoptosis or control groups in both T-47D and MDA-MB-231 cells (P<0.05; Fig. 6B and E). Moreover, CASP3 specific activities also indicated that the miR-34a group yielded a higher SA value than that of the control group (T-47D, P<0.001; MDA-MB-231, P=0.03; Fig. 6C).