A549 (human lung carcinoma), MCF7 (human breast carcinoma), and LLC1 (mouse lung carcinoma) cells were obtained from the American Type Culture Collection (Manassas, VA, United States). For monolayer cells culture, A549 cells were cultured in RPMI-1640 medium (Gibco, United States), and MCF7 and LLC1 cells were cultured in high-glucose Dulbecco’s modified Eagle’s medium (Gibco), supplemented with 10% fetal bovine serum (Hyclone, United States) and 1% penicillin/streptomycin (Amresco, United States). Construction of 3D cultured microenvironment using matrigel matrix (BD, United States) was performed as described previously (Asaithamby et al., 2011). Briefly, 30-μl trypsinized cells at a density of 1.5 × 10⁶ cells/ml were mixed with 250-μl pre-thawed matrigel and seeded into a single well of 12-well tissue culture plates. Two 2-ml media were added after incubating for 30 min at 37°C. All 3D-grown cells were cultured in matrigel for 5 days before subsequent experiments. The medium was changed every 2 days. Both 2D- and 3D-grown cells were cultured at 37°C in a humidified atmosphere containing 5% carbon dioxide.

X-ray irradiation was carried out with a Faxitron RX-650 facility (Faxitron Bioptics, United States) at a dose rate of 0.75 Gy/min, which was operated at 100 kVp 5 mA at room temperature. The dose rate was measured using the Ray meter (RX-650, Germany).

Samples of 2D and 3D cultured A549 cells were collected. Then, the microarray analysis of miRNAs was performed by CapitalBio Corporation following the protocol as previously described (Guo et al., 2008). In brief, all of the miRNA probe sequences were designed to be fully complementary to their cognate mature miRNA. Oligonucleotide probes were synthesized and were printed in triplicate using the SmartArraymicroarrayer (CapitalBio). Total RNAs were isolated from these cells with TRIzol reagent (Invitrogen, United States) 0.5 h after 5-Gy X-ray irradiation. RNA was labeled using the T4 RNA ligase, and the hybridization was performed in a hybridization cassette. Gene Spring Software (Agilent, Santa Clara, CA, United States) was used for data analysis. Values at least two times higher than the background were screened out for analysis.

Genomic DNA was extracted from the cells by using Wizard^® SV Genomic DNA Purification System (Promega, United States) according to the instruction. Extraction of total RNA from cells used for qRT-PCR was performed using the E.Z.N.A.^® Total RNA Kit (Omega, United States) following the manufacturer’s protocol.

Reverse transcription was conducted with the ALL-in-one^TM miRNA quantitative real-time polymerase chain reaction (qRT-PCR) detection kit (GeneCopoeia, United States). Primers for miR-29b-3p and the U6 internal control for humans and mice were purchased from GeneCopoeia (China). MiR-29b-3p-human (Ca: HmiRQP0373), miR-29b-3p-mice (Ca: MmiRQP0373), universal reverse primer, and U6 forward primer were from the kit. qRT-PCR was performed using a Bio-Rad Chromo4 System Real-Time PCR detector (Bio-Rad, United States). All procedures were conducted according to the manufacturers’ protocols under the following conditions: initiation for 10 min at 95°C, followed by 40 thermal cycles each at 95°C for 10 s, 60°C for 20 s, and 70°C for 10 s. Relative fold-change in miRNA expression was calculated using the 2^–ΔΔCT method with the following equation: RQ (Relative Quantitation) = 2^–ΔΔCt.

All patients’ clinical data were obtained from the KM plotter database¹, including multiple studies from Gene Expression Omnibus, The Cancer Genome Atlas, and Molecular Taxonomy of Breast Cancer International Consortium (Nagy et al., 2018). The analyzed cohort includes 504 LUADs and 1,262 BRCAs within the timeframe of our study (April 5, 2019 to June 10, 2020). For survival analysis, Kaplan–Meier survival curves were generated by using the statistical software GraphPad Prism. The non-parametric Mantel–Cox log-rank test was used to determine the statistical differences among different patient groups.

Cells were lysed in radioimmunoprecipitation assay buffer (Beyotime, China) with Protease Inhibitor Cocktail Tablets (Roche, Switzerland). The total protein concentrations of the lysates were determined using the Bio-Rad protein assay kit. Equal amounts of protein were denatured with loading buffer (Beyotime) at 100°C for 10 min, then loaded in 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis for electrophoresis and transferred to a methanol-activated polyvinylidene fluoride membrane (Millipore, United States). The membrane was blocked in tris-buffered saline containing 5% bovine serum albumin (MP Biomedical, United States) for 1 h at room temperature. Primary antibodies were incubated overnight at 4°C. The primary antibodies included AKT2, CD133, DNMT3B, MYC, PIK3R1, RBL1 (1:1,000, Proteintech, United States), Bcl-2 (1:1,000, Affinity Biosciences, United States), and GAPDH (1:2,000, ZSGB-BIO, China). After washing with tris-buffered saline twice, the membrane was incubated with the appropriate horseradish peroxidase-labeled secondary antibody for 1 h at room temperature. The secondary antibody conjugated with horseradish peroxidase is goat anti-rabbit immunoglobulin G (1:5,000, ZSGB-BIO). Immunoblots were visualized using enhanced chemiluminescence detection system according to the manufacturer’s protocol.

Si-miR-29b-3p (knockdown), Over-miR-29b-3p (overexpression), and negative control lentivirus were purchased from Genechm (China). For the construction of miR-29b-3p deficiency and overexpression stable cell lines, cells were plated on the day before the lentivirus infection at a confluence of 30–50%; the multiplicity of infection is 10. The medium was changed 24 h post-infection, and cells were selected for 7 days with 1 μg/ml puromycin.

Three-dimensional cultured cells were recovered from matrigel using recovery solution (BD) according to the manufacturer’s instructions as described previously (Lee et al., 2007). In brief, matrigel containing the 3D structures was first washed with ice-cold PBS and then removed from the well. After being transferred to a 15-ml tube containing the pre-chilled recovery solution (1 ml per well), the mixture was incubated on ice for 45 min with intermittent mixing and then centrifuged at 1,000 rpm for 10 min at 4°C. The supernatant containing the dissolved matrigel was discarded, and the 3D structures were washed once with PBS. To make a single-cell suspension of recovered 3D structures, cells were trypsinized using trypsin- ethylenediaminetetraacetic acid (0.25%, Invitrogen). Dissociated cells were used for colony formation assay.

Cells from 2D culture were washed with PBS buffer, trypsinized, and counted using a cell counter (Coulter) after irradiation. Cells from 3D culture converted to single-cell suspension as described earlier and resuspended in medium. An appropriate number of cells (0:100, 1:200, 2:500, 4:2,000, and 6:10,000)were seeded into each Φ60 dish in 5 ml of complete media. After 10 days of incubation, colonies were fixed with 10-ml fresh Carnoy’s fluid, stained with 0.5% crystal violet for 20 min. Colonies with more than 50 cells were recorded and counted manually under an inverted microscope. Plating efficiencies (PEs) were calculated as follows: numbers of colonies formed/numbers of cells plated. Cell surviving fractions were calculated as follows: PE (irradiated)/PE (unirradiated). All experiments were performed in triplicate.

The extracted genomic DNA was subjected to bisulfite modification using the EpiTect Fast DNA Bisulfite Kit (Qiagen) according to the manufacturer’s instruction. After that, the bisulfite-converted genomic DNA was amplified by a set of RBL1 primers for the unmethylated reaction and methylated reaction to the methylation-specific PCR as described previously (Pan et al., 2016a): Unmethylated forward primer (5′ GGAGGTATTTTATTATGTTGTATGA) and reverse primer (5′ TCCTTAACCCTTAACTAATCACAAA), methylated forward primer (5′ GGAGGTATTTTATTACGTTGTACGA), and reverse primer (5′ CTTAACCCTTAACTAATCGCGAA). PCR was carried out with MyCycler RCR (Bio-Rad) by using the following condition: 94°C for 5 min, followed by 40 cycles of 94°C for 30 s, 52°C (unmethylated primer) or 56°C (methylated primer) for 30 s, and 72°C for 30 s.

Cells (1 × 10⁵) were resuspended in 100 μl of serum-free medium and seeded in matrigel-coated transwell upper chambers (Millipore, United States) with 8.0 μm polycarbonate filter inserts in 24-well plates, whereas the bottom chambers were filled with 600-μl complete medium. After incubation for 24 h, non-migrated cells and the matrigel were scraped using a cotton swab. The bottom side of the membrane was fixed with ethanol and stained with Giemsa. The transwell chambers were washed three times with PBS. Images of migrated cells were obtained using a microscope (Carl Zeiss, Germany).

Cells were trypsinized, and 1 × 10⁵ cells per well were added to the 12-well plate with complete media. When cells reached approximately 90% confluency, scratch wounds were created on monolayers by a sterile 200-μl micropipette tip. Then, cells were washed twice with a fresh medium to remove any loosely held cells. Photographs were taken immediately (0 h) and 24 h later. Images at 0 and 24 h were compared, and the area of the wound closure was calculated using Image J software.

To detect 53BP1 and γ-H2AX foci that form at the DSB sites, 1 × 10⁴ cells were seeded on glass coverslips in each well of a 12-well plate, cultured for 24 h before radiation. Subsequent experimental procedures followed the previous description (van Oorschot et al., 2014). After the 1-Gy X-ray irradiation, cells were fixed with 4% paraformaldehyde for 20 min at room temperature and permeabilized with 0.5% Triton X-100 in PBS for 10 min. Non-specific binding sites were blocked with 5% bovine serum albumin (BSA) in PBS for 60 min at room temperature before probing with primary antibodies. The primary antibodies used for immunostaining include 53BP1 (1:3,000, Abcam, United States) and γ-H2AX (1:1,000, CST). Secondary antibodies (anti-mouse or rabbit conjugated with Alexa 488/633) were purchased from Beyotime (1:2,000). Then, the cells were incubated in the primary antibody diluted in 5% BSA/1 × PBS 1 h at room temperature. After incubation, cells were washed three times with PBS and then incubated with the appropriate Alexa Fluor secondary antibodies diluted in 5% BSA for 1 h. Cells were washed three times with PBS again. The nuclei were counterstained with 4′,6-diamidino-2-phenylindole (Beyotime). Digital image analysis was performed to determine the number of γH2AX and 53BP1 foci by fluorescence microscope (Axio Imager Z2) at 63 × magnification and confirmed by visual inspection of images. Quantification of foci per cell was carried out from images of 50 cells for every time point from at least three independent experiments.

Animal experiments conducted in this study were approved by Wenzhou Medical University Institutional Animal Use and Care Committee. C57BL/6J mice were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. Before tumor cell injection, age-matched 6-week-old female mice were shaved at the right hindlimbs. LLC1 cells (2 × 10⁵) were resuspended in 50-μl PBS and injected into shaved flanks subcutaneously with negative control or miR-29b-3p overexpression tumor cells. After 7 days, tumors were irradiated with a single 8-Gy dose of X-rays with the help of lead shielding. Tumor volumes were measured every other day and calculated by the formula: (length) × (width)²/2. The mice were killed when tumor volumes reached 2,000 mm³. Kaplan–Meier estimator and log-rank (Mantel–Cox) test were used for survival analysis among different tumor-bearing mice groups.

Flow cytometry was used to identify lung cancer stem cells with positive CD133 (Ferrandina et al., 2009). After dissociation with trypsin and subsequent neutralization, 2 × 10⁶ A549 cells were resuspended in 500-μl PBS, and 20-μl monoclonal antibody CD133 (Miltenyi Biotec, United States) was added. The mixture was incubated for 20 min at room temperature and protected from light. After incubation, cells were washed twice with PBS and resuspended in a 500-μl medium. Background fluorescence was estimated by substituting the specific primary antibody with specific isotype controls. Samples were acquired on a FACSCanto II flow cytometer (BD Biosciences) and analyzed with FACSDiva software (BD Biosciences). The fresh isolated CD133 + cells were cultured before assay in a stem cell medium containing a serum-free Dulbecco’s modified Eagle’s medium/F12 (1:1) medium (Gibco-BRL), 20 ng/ml epidermal growth factor, 20 ng/ml basic fibroblast growth factor, and 20 ng/ml leukemia inhibitor factor (LIF) (all were from Miltenyi Biotec).

Side population assessment by Hoechst 33342 analysis followed the previous description (Tirino et al., 2009). Both 2D and 3D cultured A549 cells were trypsinized and suspended at 2 × 10⁶ cells/ml in a medium with 5 mg/ml Hoechst 33342 (Sigma, United States) for 90 min at 37°C with intermittent stirring. After incubation, the cells were washed in PBS and kept on ice for 5 min until analyzed by a flow cytometer FlowSight (Amnis, United States). The Hoechst 33342 dye was excited at 350-nm ultraviolet, and the resultant fluorescence was measured at two different wavelengths using 424/44 BP and 675 LP filters for detection of Hoechst blue and red, respectively. The results were analyzed using Flowjo software.

All experiments in vitro were performed in triplicate and repeated at least three times. Statistical significance (P-values) of differences in means between two samples were evaluated using unpaired t-tests. Data are represented as individual values or as mean ± standard error of the mean. Group sizes (n) and applied statistical tests are indicated in figure legends. For in vivo experiments and clinical data analysis, log-rank (Mantel–Cox) tests were used for survival analysis, and tumor volume significances were assessed by two-way analysis of variance among different groups. Statistics were calculated using GraphPad Prism 8.2.1. P < 0.05 was considered statistically significant (^∗), P < 0.01 as highly significant (^∗∗), and P < 0.001 (^∗∗∗) as extremely significant.

The morphological features of A549 and MCF7 cells are different in 2D and 3D cultures (Figure 1A). Two-dimensional cultured cells were flat and formed a monolayer on the Petri dish, whereas cells in 3D culture in the matrigel formed microspheres. To identify the different characteristics of miRNAs in cells between 2D and 3D cultures, we analyzed the miRNA expression profiles of the 2D and 3D cell cultures of A549 cells 0.5 h after exposure to 0- and 5-Gy X-rays using miRNAs microarray analysis (Figure 1B). We found 10 miRNAs were upregulated, whereas 11 were downregulated with more than threefold changes between 2D and 3D cultures (Figure 1C). MiR-29b-3p had the biggest low fold change in 3D compared with 2D cultures (−10.43). For further investigation, we measured the expression of miR-29b-3p in A549 and MCF cells in the 2D and 3D cultures after 5-Gy X-ray irradiation by qRT-PCR. As shown in Figure 1D, the expression of miR-29b-3p was significantly lower in 3D culture compared with 2D culture both in irradiated and unirradiated cells. These results indicated that the miRNA, particularly miR-29b-3p, expression profiles in cells between 2D and 3D cultures are different.

To explore the potential association between miR-29b-3p expression and overall survival of cancer patients, we analyzed the miR-29b-3p expression status and overall survival in LUAD and BRCA patients. Published clinical and genomic data from the KM plotter database, including 504 late-stage LUAD and 1,262 late-stage BRCA patients, were used. The median miR-29b-3p expression was used as a cutoff value. It was found that the miR-29b-3p high expression group had a higher survival probability than the low expression group, which was 58 months in LUAD (vs. 42 months for low group, P = 0.012, log-rank test, Figure 2A) and 198 months in BRCA (vs. 180 months for low group, P = 0.0050, log-rank test, Figure 2B). These data suggested that the high expression of miR-29b-3p had significant advantages in cancer treatment.

Previous research reported that MYC Proto-Oncogene (c-MYC) directly suppresses miR-29b-3p (Chang et al., 2008). MiR-29b-3p is known to critically affect cancer progression by functioning as a tumor suppressor (Yan et al., 2015), and it targets several DNA methyltransferases (DNMTs, including DNMT1, DNMT3B, etc.) and regulates members of the DNA demethylation signaling pathway, leading to the downregulation of global DNA methylation in malignant cells (Garzon et al., 2009b). In addition, miR-29b-3p inhibits cell proliferation and promotes apoptosis by targeting B-cell lymphoma 2 (Bcl-2), phosphoinositide-3-kinase regulatory subunit 1 (PIK3R1), and AKT serine/threonine kinase 2 (AKT2) (Figure 2C). Because the miR-29b-3p expression is different between 3D and 2D cultured cells, we speculated that the expression of c-MYC and miR-29b-3p downstream oncogenes could be different as well. Figure 2D shows that the expression of the genes mentioned earlier was higher in unirradiated A549 cells cultured in 3D than in those of 2D culture. At the time point of 1, 6, and 12 h after 5-Gy X-ray irradiation, c-MYC, DNMT3B, PIK3R1, AKT2, and Bcl-2 maintained high expression in 3D cells. These results indicated that high expression of c-MYC suppresses miR-29b-3p and increases expression of DNMT3B, PIK3R1, AKT2, and Bcl-2. Recent reports have demonstrated that c-MYC contributes to chemotherapeutic resistance in various CSCs (Elbadawy et al., 2019; Zhang et al., 2019), and CSCs are generally resistant to conventional chemotherapy and radiotherapy through activation of DNMT1, DNMT3B, Bcl-2, and cellular pro-survival signaling pathways PI3K/AKT (Berghauser Pont et al., 2014; Wang et al., 2014; Liu et al., 2015; Xia and Xu, 2015; Chang et al., 2016; Zagorac et al., 2016). These results suggested that c-MYC/miR-29b-3p and downstream DNMT3B, Bcl-2, and PIK3R1/AKT2 may be involved in radioresistance in 3D cultured cells.

Because high expression of miR-29b-3p benefits the prognosis of LUAD and BRCA, we investigated whether miR-29b-3p was involved in the radioresistance. Firstly, we constructed miR-29b-3p knockdown stable cell lines in 2D cultured A549 and MCF7 cells by lentivirus. As shown in Figure 3A, the expression of miR-29b-3p significantly decreased in miR-29b-3p knockdown cell lines. In the miR-29b-3p knockdown groups, the expression of DNMT3B, PIK3R1, AKT2, and Bcl-2 was maintained at a high level after 5-Gy X-ray radiation. The miR-29b-3p level had no impact on c-MYC expression (Figure 3B). Our previous study revealed that promoter methylation of transcriptional corepressor like 1 (RBL1) enhances the radioresistance of 3D cells (Pan et al., 2016a). We detected the correlation between miR-29b-3p and RBL1 promoter methylation. The results show that a low level of miR-29b-3p decreased RBL1 expression (Figure 3B) and methylated promoter region of RBL1 in 2D cultured A549 cells. The promoter of RBL1 maintained the methylated status after 48 h in 5-Gy X-ray irradiated cells (Figure 3C), which accounts for the RBL1 high expression in the miR-29b-3p knockdown group. Figure 3D showed that in the miR-29b-3p knockdown groups, survival fractions of X-ray irradiated 2D cultured A549 and MCF7 cells were higher than those of negative control. These results suggested that the knockdown of miR-29b-3p enhances the radioresistance of 2D cultured cells.

Because knockdown of miR-29b-3p enhanced the radioresistance of 2D cultured A549 and MCF7 cells, we next examine whether the overexpression of miR-29b-3p decreases the radioresistance of 3D cells. As shown in Figure 4A, miR-29b-3p expression significantly increased in 3D cultured A549 and MCF7 cells infected with lentivirus containing miR-29b-3p overexpression vector, compared with the negative control groups. In the miR-29b-3p overexpression groups, DNMT3B, PIK3R1, AKT2, and Bcl-2 expression levels were lower than that of negative control groups after 5-Gy X-ray radiation. Meanwhile, RBL1 expression increased in the miR-29b-3p overexpressed groups, but c-MYC expression had no obvious change (Figure 4B). Figure 4C showed that the high level miR-29b-3p demethylated promoter region of RBL1 and resulted in RBL1 high expression in miR-29b-3p overexpressed groups. This is because the survival fractions of miR-29b-3p overexpressed A549 and MCF7 cells in 3D culture exposed to X-rays were lower than those of the negative control, suggesting that overexpression of miR-29b-3p sensitizes 3D cultured A549 and MCF7 cells to radiation (Figure 4D).

To further investigate the effect of miR-29b-3p on radiosensitivity in vivo, mice were injected with the miR-29b-3p overexpression or vector control LLC1 cells and exposed to 8-Gy X-ray after 7 days (Figures 5A,B). Overexpression miR-29b-3p resulted in significant inhibition on tumor growth and prolonged host survival in the lung cancer mouse model, with one of five mice had been cured (Figures 5C,D). These results suggested that overexpression of miR-29b-3p significantly enhances radiosensitivity in vivo.

We have demonstrated that miR-29b-3p influences cancer cells’ radiosensitivity; we further investigated whether miR-29b-3p impacts the kinetic process of DNA damage repair. We examined the 53BP1 and γH2AX foci that are surrogate markers of DNA damage repair by immunofluorescence staining in the 2D cultured A549 cells. In 1-Gy X-ray irradiated miR-29b-3p knockdown A549 cells, the kinetics of foci dissolution of 53BP1 and γH2AX were faster than in negative groups after 12 h (Figures 6A,B). Moreover, more 53BP1 and γH2AX foci remained in miR-29b-3p overexpressed A549 cells than in negative control cells after 6 h (Figures 6C,D). These results indicated that miR-29b-3p impacts the recruitment of the DNA repair complex.

Epithelial–mesenchymal transition (EMT) is a critical step in cancer cell invasion and metastasis, and it positively correlates with poor patient prognosis. In addition to its roles in cell proliferation, apoptosis, and differentiation, miR-29b-3p also exerts effects on cell migration and invasion in vitro by regulating EMT signaling (Ru et al., 2012; Kinoshita et al., 2013; Poudyal et al., 2013). According to the pieces of evidence discussed earlier, we hypothesized that miR-29b-3p knockdown might rescue the migratory and invasive capacity of irradiated cancer cells. Matrigel invasion assay showed that miR-29b-3p knockdown A549 and MCF7 cells exhibited more marginal invasion through the extracellular matrix after 48 h in both irradiated and unirradiated cells compared with control cells (Figures 7A,B). Figures 7C,D show that miR-29b-3p knockdown A549 and MCF7 cells had a remarkable ability on promoting cell migration in the scratch wound after 24 h in both irradiated and unirradiated cells compared with control cells. Thus, the effects of miR-29b-3p on cell migration and invasive properties may play a critical role in poor patient prognosis with radiation therapy.

CD133 is currently considered the most robust surface marker for CSCs in various tumor types (Glumac and LeBeau, 2018). Wang et al. (2015) reported low expression of miR-29b in CD133-positive A549 cells. MiR-29b-3p is significantly downregulated in CD133 + cells separated from the peripheral blood of HCC patients (Zekri et al., 2018). As our previous study revealed that 3D cultured cells simulate some stem cell characteristics and have high radioresistance (Xue et al., 2015b), we speculated that miR-29b-3p deficiency in 3D cells might account for these phenomena. Flow cytometry was used for CD133-positive A549 cell sorting (Figure 8A). Figure 8B shows a significantly low expression of miR-29b-3p in CD133-positive A549 cells by qRT-PCR. In the CD133-positive A549 cells, cMYC, RBL1, and oncogene expression at indicated time points after 5-Gy X-rays have a similar tendency with 3D cultured cells (Figure 8C). Hoechst 33342 side population (SP) analysis is a common method for identifying stem cells in mammalian tissues (Telford et al., 2007). Flow cytometry analysis shows that the number of SP cell proportions was higher in 3D cultured A549 and MCF7 cells compared with 2D cells. MiR-29b-3p knockdown in 2D cells could increase the SP cell proportion. Inversely, miR-29b-3p overexpression in 3D cells would alleviate the SP cell proportion (Figures 8D,E). These results demonstrate that miR-29b-3p is associated with stemness characters and oncogenes PIK3R1, AKT2, and Bcl-2 expression in CSC.