To determine the potential role of eIF3a in the cellular response to IR, we first knocked down eIF3a expression using siRNA in H1299 cells, which have a high level of endogenous eIF3a (Figure 1A) followed by analysis of cellular response to IR using colony formation assay. As shown in Figures 1B,C, H1299 cells with eIF3a knockdown (H1299/Si) are significantly more resistant with a 2-fold increase in relative resistance factor (RRF) than the control H1299 cells transfected with scrambled control siRNA (H1299/Scr). To confirm this observaiton, we performed a reverse experiment by using the stable NIH3T3 cells with eIF3a overexpression (NIH3T3/eIF3a) (Figure 1A) and tested their response to IR in comparison with the control cells transfected with the empty vector (NIH3T3/Vec). As shown in Figures 1B,C, NIH3T3/eIF3a cells are remarkably more sensitive than the control NIH3T3/Vec cells to IR with ∼2-fold reduction in RRF. Thus, eIF3a expression may affect cellular sensitivity to IR treatments.

To investigate how eIF3a affects the cellular response to IR, we tested the hypothesis that eIF3a may regulate the repair of DSBs induced by IR. For this purpose, we first tested the effect of eIF3a on gamma-H2A histone family member X (γ-H2AX) expression, a marker for DSB (Kuo and Yang, 2008), following IR treatment. As shown in Figure 2A, little γ-H2AX was detected in either H1299/Si or the control H1299/Scr cells without IR treatments. However, the γ-H2AX level drastically increased following IR treatment in these cells at 20 min after IR treatment. Interestingly, at 6 h after IR, γ-H2AX in H1299/Si cells returned essentially to the basal level while it remained high in the control H1299/Scr cells. We also performed similar experiments using NIH3T3/eIF3a and the control NIH3T3/Vec cells and found that eIF3a overexpression clearly delayed DNA repair, as indicated by the delayed disappearance of γ-H2AX (Figure 2A), consistent with the findings using H1299/Si and H1299/Scr cells.

To verify the above findings, we performed immunofluorescence staining of γ-H2AX in the nuclei of these cells at 2 and 6 h after IR exposure. As shown in Figure 2B, the high level of punctate staining of γ-H2AX in the nuclei of H1299/Si cells observed at 2 h following IR disappeared at 6 h. However, the punctate staining in the control H1299/Scr cells remained high at 6 h following IR. Similarly, NIH3T3/eIF3a cells retained high levels of γ-H2AX, whereas the control NIH3T3/Vec cells lost γ-H2AX staining at 6 h following IR. These observations are consistent with the results shown using Western blot analysis. Thus, it is possible that eIF3a suppresses the repair of DSB induced by IR.

In the above studies, we used γ-H2AX as a DNA damage marker to evaluate DNA damage and repair. To directly evaluate DNA damage induced by IR in the presence of different levels of eIF3a, we performed neutral comet assay at 2 and 6 h following IR treatment. Cells at 20 min following IR were not tested using this assay because the short time was difficult to control for the comet assay. As shown in Figure 3A, H1299/Si cells clearly have significantly lower Olive tail moment than the control H1299/Scr cells at both 2 and 6 h following IR treatment. It is noteworthy that the relative Olive tail moment was reduced at 6 h compared with 2 h following IR in H1299/Si cells while it remained high in H1299/Scr cells, suggesting that little DSBs were repaired in H1299/Scr cells compared with H1299/Si cells. Similarly, NIH3T3/eIF3a cells had significantly higher Olive tail moment than its control NIH3T3/Vec cells (Figure 3B) following IR treatment, and less DSBs were repaired in NIH3T3/eIF3a than in the control NIH3T3/Vec cells as indicated by the change in the relative Olive tail moment between 2 and 6 h following IR. Thus, eIF3a likely inhibits the repair of DSBs induced by IR, and cells with high levels of eIF3a retain higher levels of DSBs following IR while cells with lower eIF3a retain lower level of DSBs.

To determine if eIF3a regulates repairs of DSBs, we next performed host cell reactivation (HCR) assay of NHEJ activity since NHEJ is the main repair pathway of IR-induced DNA damages and it is independent of cell cycle stages as discused above. As shown in Figure 3C, H1299/Si cells had a 2-fold increase in NHEJ activity compared with the control H1299/Scr cells. Consistently, the NHEJ activity in NIH3T3/eIF3a cells was decreased by 40% compared with the control NIH3T3/Vec cells (Figure 3D). These findings suggest that eIF3a may play an important role in suppressing NHEJ repair of DSBs.

Because eIF3a is known to regulate the synthesis of proteins, we hypothesized that eIF3a may regulate NHEJ repair of DSBs by regulating the synthesis of NHEJ repair proteins. To test this hypothesis, we first performed a Western blot analysis of major proteins involved in NHEJ repair in H1299/Si vs. H1299/Scr and NIH3T3/eIF3a vs. NIH3T3/Vec cells. As shown in Figures 4A,B, the expression of DNA-PKcs, Ku70, and Ku80 in H1299/Si cells was drastically increased compared with the control H1299/Scr cells. Consistently, the expression of these genes in NIH3T3/eIF3a cells was dramatically reduced compared with the control NIH3T3/Vec cells. Interestingly, real-time reverse transcription (RT)-PCR analyses showed no change in the mRNA level of these genes, suggesting that the effect of eIF3a on the expression of DNA-PKcs, Ku70, and Ku80 is likely at the protein, not mRNA, level.

Next, we performed pulse labeling and cycloheximide chase experiments to determine the eIF3a effect on the synthesis and degradation of these DNA repair proteins, respectively. As shown in Figure 5, eIF3a knockdown using siRNA in H1299 cells (Figure 5A) or eIF3a overexpression in NIH3T3 cells (Figure 5B) had no effects on the decay of these DNA repair proteins. However, the synthesis of Ku70, Ku80, and DNA-PKcs was dramatically increased in H1299/Si (Figure 5A) and reduced in NIH3T3/eIF3a (Figure 5B) cells compared with their respective control H1299/Scr and NIH3T3/Vec cells. Thus, eIF3a likely inhibits the synthesis of Ku70, Ku80, and DNA-PKcs proteins, leading to a reduced repair of DSB by the NHEJ pathway.

The above studies suggest that patients with tumors that have a high level of eIF3a may be more sensitive to treatments that cause DSBs such as radiotherapy and chemotherapy with drugs that cause DSBs. To test this hypothesis, we performed overall survival analysis of breast, gastric, lung, and ovarian cancer patients using information freely available in Kaplan–Meier (KM) plotter. As shown in Figure 6, breast, gastric, lung, and ovarian cancer patients with a high eIF3a expression level all had better overall survival than patients with a low eIF3a level. These findings are consistent with the eIF3a function in sensitizing cancer cells to treatment that cause DNA damages.

Cell culture media and reagents were purchased from BioSources International (Camarillo, CA), Media Tech (Herndon, CA), or Cambrex (Walkersville, MD). All electrophoresis reagents and polyvinylidene difluoride (PVDF) membranes were purchased from Bio-Rad (Hercules, CA). SYBR Green PCR Master Mix for real-time PCR, the high-capacity cDNA reverse transcription kit, and all primers were purchased from Applied Biosystems-Thermo Fisher (Chicago, IL). Metafectene^® pro was purchased from Biontex (San Diego, CA). All other chemicals and reagents were of molecular biology grade from Sigma (St. Louis, MO) or Fisher Scientific (Chicago, IL).

Human lung cancer cell line H1299 and NIH3T3 cells were cultured at 37°C with 5% CO₂ in RPMI 1640 and Dulbecco’s modified Eagle’s medium (DMEM) medium, respectively, supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin. G418 at 400 μg/ml was also supplemented in the culture of the stable NIH3T3 cells overexpressing eIF3a and vector-transfected controls, which were established in our previous study (Yin et al., 2011b).

Colony formation survival assay was performed following IR treatments as previously described (Li et al., 2010; Wu et al., 2016). Briefly, 100 cells/well were seeded in six-well plates and cultured for 24 h followed by treatments with different doses of IR and continuous culture for 10–14 days with media changed every 2–3 days. At the end of the study, cells were washed with phosphate-buffered saline (PBS), and the colonies were stained with 0.05% (w/v) crystal violet in PBS containing 20% methanol for 15 min at room temperature and counted manually.

Western blot analysis and immunofluorescence staining were performed as previously described (Dong and Zhang, 2003; Wu et al., 2016). Briefly, for Western blot analysis, equal amount of proteins from different cells were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), transferred to PVDF membranes, and probed with antibodies specific to eIF3a, γ-H2AX, Ku70, Ku80, DNA-PKcs, and actin control followed by probing with horseradish peroxidase (HRP)-conjugated secondary antibody and enhanced chemiluminescence (ECL) reagents. The signals are captured on X-ray films.

Immunofluorescence staining was performed by culturing cells on glass coverslips, which were washed twice with cold PBS and fixed with acetone/methanol mixture (1:1) for 10 min and blocked in 1% bovine serum albumin (BSA) and 1% normal horse serum in PBS at 4°C for 30 min. The coverslips were then probed with the γ-H2AX-specific antibody for 30 min at 4°C followed by incubation with fluorescein isothiocyanate (FITC)-conjugated secondary antibody at room temperature for 30 min and washed twice with PBS. The cells were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) before viewing on a confocal microscope.

Immunoprecipitation was performed as previously described (Dong and Zhang, 2003). Briefly, cell lysates were first mixed with mouse immunoglobulin G (IgG) and incubated for 2 h at 4°C followed by addition of 50% protein G-agarose slurry and incubation for 3 h at 4°C to remove non-specifically bound proteins by centrifugation. The supernatant was transferred to a new tube and incubated with 5 μg primary antibodies against Ku70, Ku80, or DNA-PKcs at 4°C for 2 h. Finally, 50 μl 50% protein G-agarose beads was added to the mixtures and incubated at 4°C overnight. The immunoprecipitated materials were collected by centrifugation and extensive washing with lysis buffer followed by separation using SDS-PAGE.

Real-time RT-PCR was performed as described previously (Dong et al., 2004, 2005). Briefly, total RNA was extracted using RNeasy Mini Kit (Qiagen, Valencia, CA), and real-time RT-PCR was performed using Power SYBR Green RNA-to-CT 1-Step kit. Data were normalized to an internal control gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The sequences of primers for the PCRs were 5-CATGGCAACTCCAGAGCAG (forward) and GCTCCTTAAACTCATCCACC (reverse) for Ku70; AGAAGAAGGCCAGCTTTGAG (forward) and AGCTGTGACAGAACTTCCAG (reverse) for Ku80; CCGGACGGACCTACTACGACT (forward) and AGAACGACCTGGGCATCCT (reverse) for DNA-PKcs.

Neutral comet assay was performed as previously described (Wu et al., 2016). Briefly, cells treated with or without IR were embedded in low melting agarose on a microscope slide followed by lysis at neutral pH and electrophoresis in Tris-borate-ethylenediaminetetraacetic acid (TBE) buffer for 2 h. Comets were observed after staining the cells with SYBR Green I and scoring 120 cells in each sample to measure the Olive tail moment.

Host cell reactivation NHEJ assay was performed using pGL3 reporter plasmid with firefly luciferase gene driven by cytomegalovirus (CMV) promoter as previously described (Wu et al., 2016). Briefly, pGL3 was first linearized by HindIII digestion and co-transfected into cells along with the control circular pRL-TK reporter plasmid encoding Renilla luciferase followed by determination of both firefly and Renilla luciferase activity. Expression of firefly luciferase is dependent on the repair of the plasmid to regenerate the circular plasmid via NHEJ.

Pulse labeling using [³⁵S]methionine in combination with immunoprecipitation and SDS-PAGE analysis was used to determine the protein synthesis rate as previously described (Dong and Zhang, 2003; Dong et al., 2004). Briefly, cells were washed with PBS and methionine-free media before being subjected to labeling with [³⁵S]methionine (10 μCi/ml) in methionine-free media for 2 h. The cells were then washed with PBS and harvested for cell lysate preparation, immunoprecipitation, and separation by SDS-PAGE. The signals for immunoprecipitated proteins labeled by [³⁵S]methionine were captured by exposing to X-ray films.

The stability of specific proteins was determined using cycloheximide chasing as previously described (Liu et al., 2006). Briefly, cells were treated with 10 μg/ml cycloheximide for different times and harvested for preparation of cell lysates and Western blot analyses. The quantity of each protein at each time point of cycloheximide treatment was determined using the ImageJ software (NIH, USA) and plotted against the treatment time.

The KM plotter can assess the effect of 54,675 genes on survival using 10,461 cancer samples, including 5,143 breast, 1,816 ovarian, 2,437 lung, and 1,065 gastric cancer patients with a mean follow-up of 69/40/49/33 months (Lanczky et al., 2016). The analyses on all four cancers were conducted using default parameters, and the hazard ratio (HR) with 95% confidence intervals and log rank P < 0.05 considered significant were calculated and shown on the plot.