Role of Tumor Necrosis Factor-α and TRAIL in High-Dose Radiation–Induced Bystander Signaling in Lung Adenocarcinoma

A central radiobiological paradigm contended that the biological effects of exposure to ionizing radiation (IR) occur only in directly irradiated cells (1). However, important biological effects may arise in cells which themselves receive no radiation exposure as a consequence of signals transmitted from neighboring cells secreting extracellularly or through gap junctions a phenomenon known as radiation-induced bystander effect (RIBE; refs. 2–5). RIBEs were shown at low doses of low LET radiation (<50 cGy of X-rays), and doses as low as 1 cGy can generate significant bystander cytotoxicity (6). Furthermore, exposure of one cell in a population to an α particle is sufficient to promote a RIBE (7). Most of the bystander factors have been elucidated using low-dose radiation. However, there are very few studies that have analyzed the RIBEs and factors released during high-dose radiation.

The nature, as well as the mechanisms by which these factors cause bystander effects, has been elusive. One such mechanism might involve radiation-induced early genes, such as early growth response-1 (Egr-1) gene that has a radiation-inducible promoter (8) and promotes the elevation of growth factors or their receptors, such as transforming growth factor-β1, tumor necrosis factor-α (TNF-α), or epidermal growth factor receptor (9). Ectopically expressed EGR-1 synergized IR-inducible TNF-α expression and apoptosis (10, 11). TNF-α and Apo2 ligand (Apo2L)/TNF-related apoptosis-inducing ligand (TRAIL) are directly involved in apoptosis and are induced by IR (10–14). TRAIL possesses bystander and antitumor properties that lead to selective killing of malignant cells (15). To understand the molecular mechanisms involved in high-dose RIBEs and to identify the putative clastogenic factors, we used non–small lung carcinoma A549 (EGR-1–positive) and H460 (EGR-1–negative) cell lines and media transfer experiments. The findings presented here show a distinct EGR-1–dependent and EGR-1–independent role in conferring the cell killing effects of IR-induced bystander signaling regulated by TNF-α and the soluble form of TRAIL (sTRAIL).

Transfections. The reporter construct EBS-CAT was described earlier (16). pSilencer control and pSilencer vector used for generation of transfectants stably expressing EGR-1 small interfering RNA (siRNA) were from Panomics. Adenoviral EGR-1 construct (EGRI293F) was a gift from Dr. Jeffery Milbrandt (17, 18). Transient or stable transfections were performed by Effectene transfection reagent (Qiagen, Inc.).

Cell lines, culture, and generation of stable clones expressing si-Egr-1. The lung cancer cell lines A549 and H460 were obtained from American Type Culture Collection and cultured in RPMI 1640 with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Primary cultures of mouse embryonic fibroblast (MEF) cells from homozygous (−/−) and heterozygous (+/−) Egr1 knockout mice were grown in DMEM with 10% FBS and 1% penicillin/streptomycin at 37°C. To generate Egr-1 siRNA stable cell lines, the plasmids pSilencer-si-Egr-1 and pSilencer vector DNA were transfected separately in A549 cells. The stable clones selected with 100 μg neomycin per milliliter were maintained in a medium containing 50 μg neomycin per milliliter.

Generation of radiation conditioned media. Exponentially growing cells at 80% confluency were irradiated (1–10 Gy) in freshly transferred media using a 100-kVp X-ray machine (Phillips) at room temperature. The supernatant from the medium removed after 24 h and centrifuged at 3,000 rpm for 3 min, hereby called as radiation-conditioned medium (RCM; 12 mL), was used for subsequent experiments.

Colony-forming assay and quantitation of apoptosis. For clonogenic cell survival studies, A549 and H460 cells were either untreated, exposed to 1 to 6 Gy IR, or exposed to RCM generated from 1 to 6 Gy exposures. Subsequent studies were performed with 2 Gy–derived or 10 Gy–derived RCM. For blocking experiments, neutralizing antibody against TNF-α or TRAIL (R&D System) was added prior and postirradiation as described below. One treatment: Cells were exposed to anti–TNF-α or TRAIL antibody (2.5 μg/mL) and irradiated with 2 or 10 Gy. After 24 h, medium was transferred to unirradiated cells, and colony-forming assay was performed. Two treatments: Cells were treated as above with a second treatment of anti–TNF-α or TRAIL antibody (1.25 μg/mL), and colony-forming assay was performed. Three treatments: Cells were treated as above (two treatments) followed by irradiation with 2 Gy and were immediately exposed to third treatment of anti–TNF-α or anti-TRAIL antibody (2.5 μg/mL). Surviving fraction was calculated as described previously (11). To quantify apoptosis, the TUNEL labeling (Roche) was used (11). Mean and SE were calculated from four independent experiments.

Enzyme-linked immunosorbent assay and Western blot analysis. Microtiter ELISA plates coated with anti–TNF-α or anti-TRAIL monoclonal antibody (R&D Systems) were used to estimate levels of TNF-α and TRAIL in RCM (19). Total proteins isolated using Laemmli buffer were subjected to Western analysis using anti–EGR-1 antibody (Santa Cruz Biotechnology, Inc.) or anti–β-actin antibody (Sigma Chemical Co.) as an internal loading control.

Immunofluorescence. A549 or H460 cells cultured on Lab-Tek chamber slides (Nunc, Inc.) were either untreated or exposed to 10 Gy IR or exposed to RCM or RCM exposure followed by 2 Gy IR. After blocking with 3% bovine serum albumin in PBS (pH 7.4) with 0.25% Tween 20 slides were incubated overnight at 4°C in primary antibodies, mouse anti-p65, and/or rabbit p50 or anti-TRAIL or anti-PAR4. After washes in PBS (pH 7.4) with 0.25% Tween 20, the cells were overlaid with secondary antibodies, Cy2-conjugated anti-mouse IgG, and Cy3-conjugated anti-rabbit IgG (1:1,000). After washes, the slides were mounted using antifade, counterstained with 4′,6-diamidino-2-phenylindole (DAPI; VectaSheild, Vector).

Electrophoretic mobility shift assay and chromatin immunoprecipitation. Nuclear extract from A549 and H460 cells were subjected to electrophoretic mobility shift assay using a consensus probe that contains Egr-1 binding site (underlined): 5′-GGATCCAGCGGGGGCGAGCGGGGGCGAACG-3′ (Geneka Biotechnology). EGR-1 antibody (Geneka Biotechnology) was used for supershift experiments. Binding reactions were electrophoresed on a 5% polyacrylamide gel. Chromatin immunoprecipitation assay was performed as per Orlando et al. with some modifications (20) using 2 to 5 × 10⁹ A549 cells fixed with 1% formaldehyde for 10 min. Cell lysates were subjected to sonication and incubated with rabbit anti–EGR-1 antibody for 3 h at 4°C. The cross-links were reversed at 65°C for 12 h. Proteinase K digestion of the extracts was performed at 37°C for 7 h. DNA was isolated by phenol/chloroform method. ³²P-PCR was performed using specific primers for TNF-α promoter (5′-CCACCCAGCCTTTCCTGAG-3′, 5′-CGGAAAACTTCCTTGGTGGAG-3′) that yielded ≈500 bp product. Human glucose-6-phosphate dehydrogenase (HG6PD) gene primers were used as a negative control.

Differential radiation sensitivity of A549 and H460 cells. Cell survival curves show that A549 cells were more radio-resistant (D₀ = 159 cGy) when compared with H460 (D₀ = 81 cGy) as also reported earlier (21, 22). The surviving fractions at direct 2-Gy dose were 0.656 ± 0.04 and 0.255 ± 0.02 for A549 and H460, respectively.

When compared with the effect of conditioned media obtained from untreated cultures of A549 and H460, increased clonogenic inhibition was observed with RCM obtained after 24 or 48 h of 10 Gy dose treatment for A549 (P < 0.034) and 24 h of 10-Gy dose treatment for H460 cells (P < 0.0056; Fig. 1A and B). Furthermore, a reduced survival was observed in both A549 and H460 cells when 2 Gy dose was given to cells exposed to RCM obtained after 3, 6, 24, and 48 h (Fig. 1A and B), and this effect was significant when compared with conditioned media plus 2 Gy dose (P < 0.0001). Based on these results, all the experiments were planned using RCM obtained after 24 h of 10-Gy treatment.

Using the RCM obtained after 24 h of 2 Gy radiation treatment, A549 and H460 showed a surviving fraction of 1.11 ± 0.107 and 0.78 ± 0.2, respectively. These results indicate that the direct radiation sensitivities were similar to that of the bystander sensitivities (RCM responses) for both cell lines (Fig. 1C; Supplementary Figs. S1 and S2). The degree of RCM-mediated clonogenic inhibition was greater in H460 cells than A549 cells, and a proportionate increase in clonogenic inhibition was observed in both the cell lines when a 2 Gy direct radiation was given after RCM exposure (Fig. 1A; Supplementary Figs. S1 and S2). RCM plus 2 Gy mimics the GRID radiation therapy application in patients. High doses of radiation exposure by GRID (induces bystander signaling) were followed by clinically relevant doses (2 Gy).

TNF-α mediates the radiation-induced bystander signaling in A549 cells. To identify the factors responsible for differential direct and bystander responses to high-dose radiation, we investigated the release of TNF-α in response to IR because TNF-α is a most common radio-inducible cytokine. A549 cells exposed to direct IR, RCM, and RCM + 2 Gy IR showed significant induction of TNF-α as assessed by ELISA. In particular, IR (10 Gy) alone caused an increase in TNF-α at 3 h with a peak elevation at 24 h and decline at 48 and 72 h. However, 10 Gy RCM caused a significant early induction at 3 h (P = 0.00023) followed by a decline at 6, 24, 48, and 72 h. Furthermore, when 10 Gy RCM–treated cells were exposed to 2 Gy dose, a significant sustained increase in TNF-α was observed at 3, 6, and 24 h (P < 0.0001) with a rapid decline at 48 and 72 h (Fig. 2A). In H460 cells, no significant changes (P < 0.167) in the levels of TNF-α were observed in different treatments compared with untreated control (Fig. 2A). These results show that TNF-α may be a high-dose radiation–induced bystander cytokine in A549 cells but not in H460 cells.

To ascertain the specificity of TNF-α in regulating the RIBEs, radiation-induced TNF-α function was blocked by using neutralizing antibodies and cell survival of A549 and H460 cells was assessed. Surviving fractions at 2 Gy, 10 Gy RCM, and 10 Gy RCM + 2 Gy were 0.823, 0.737, and 0.505, respectively. With two treatments of neutralizing antibody against TNF-α, the surviving fractions of 2 Gy RCM and 10 Gy RCM increased to 0.931 (P < 0.05) and 0.951 (P < 0.0002), respectively. Furthermore, the 10 Gy RCM + 2 Gy group showed significant enhancement in surviving fraction to 0.714 (P < 0.00012) after three treatments of neutralizing antibody against TNF-α (Fig. 2B). Overall, the effective reversal effect was significant in 10 Gy RCM and 10 Gy RCM + 2 Gy, but not in 2 Gy RCM group. A modest increase in surviving fraction was observed in 2 Gy RCM and 10 Gy RCM H460 groups, whereas, the 10 Gy RCM + 2 Gy group failed to show such increase in surviving fraction when exposed to neutralizing antibody against TNF-α (Fig. 2B). These results show that TNF-α plays a role in eliciting bystander mediated clonogenic inhibition in A549 cells but not in H460 cells.

TRAIL mediates the radiation-induced bystander responses in H460 cells. Since clonogenic inhibition in H460 cells was independent of TNF-α (Fig. 2A and B), we searched for the genes mediating such responses in H460 cells based on the following contentions: should be (a) inducible by IR, (b) involved in apoptosis, and (c) released into the medium. One of the genes that met all the above criteria was the Apo2L/TRAIL (23, 24). To ascertain the potential involvement of TRAIL in RIBEs, we analyzed the kinetics of TRAIL in response to direct and RCM treatments. In A549 cells, modest induction was evident in 10 Gy RCM + 2 Gy group with a peak at 72 h, whereas in 10 Gy direct and 10 Gy RCM group, TRAIL was not induced (Fig. 2C). In H460 cells, significant suprainduction of TRAIL was observed in 10 Gy RCM group when compared with 10 Gy group (P < 0.000045), with a peak at 24 h followed by decline in 48 h, reaching basal levels at 72 h. A relatively significant induction of TRAIL was observed in 10 Gy RCM + 2 Gy group with peak levels at 6 h (Fig. 2C) when compared with 10 Gy group (P < 0.0064). The 10 Gy direct radiation group showed a modest induction of TRAIL at all the time points assessed. These results show that the TRAIL may be an important bystander factor in eliciting the bystander response in H460 cells but not in A549 cells.

Blocking the activity of TRAIL using neutralizing antibodies did not alter the cell survival of A549 cells in 2 Gy RCM, 10 Gy RCM, and 10 Gy RCM + 2 Gy groups (Fig. 2D). However, H460 cells showed a significant increase in cell survival when treatment-induced TRAIL was blocked in the 10 Gy RCM (P < 0.002) and 10 Gy RCM + 2 Gy (P < 0.000047) groups compared with the IgG control (Fig. 2D). These results show that TRAIL is a potent bystander soluble factor mediating high-dose RIBEs in H460 cells.

TNF-α causes the activation of nuclear factor-κB in A549 cells. TNF-α initiates nuclear factor-κB (NF-κB) nuclear translocation by causing dissociation of the inhibitory protein IκBα from NF-κB dimeric complex (25). Because IR-induced TNF-α levels were elevated in A549, we reasoned that this TNF-α would mobilize the subunits of NF-κB (p65 and p50) into the nucleus, leading to the activation of the NF-κB signaling. To assess this activation, double immunofluorescence was performed in A549 and H460 cells exposed to either 2 Gy, 10 Gy IR, RCM, and RCM + 2 Gy direct IR. A549 but not H460 cells showed the translocation of NF-κB subunits into the nucleus (Fig. 3A). Furthermore, RCM from A549 cells caused translocation of NF-κB subunits into the nucleus of H460 cells (Fig. 3A). To confirm the specificity of this mechanism, TNF-α was blocked using a neutralizing antibody, and the translocation of NF-κB was assessed. Anti–TNF-α neutralizing antibody blocked the nuclear translocation of p50 in both A549 and H460 cells exposed to RCM derived from A549 cells (Fig. 3B). These results show the specificity of TNF-α present in the RCM of A549.

TRAIL mobilizes PAR4, a proapoptotic protein, into the nucleus. The Par4 gene induced during the process of apoptosis (26) needs nuclear translocation for apoptosis (27). Furthermore, ectopic expression of PAR4 in neoplastic lymphocytes augments sensitivity to TRAIL-induced cell death (28). To understand the molecular mechanism involved in TRAIL-mediated cell death, we analyzed the PAR4 localization in cells exposed to recombinant TRAIL (rTRAIL) or sTRAIL in RCM from H460 cells by immunofluorescence. We observed that the PAR4 translocated to the nucleus in both the A549 and H460 cells in response to IR, although there was no evidence showing the release of TRAIL into medium of A549 cells (Fig. 3C). In addition, both the cell lines showed the translocation of PAR4 to the nucleus in response to the treatment with rTRAIL (Fig. 3C). Thus, TRAIL-mediated effects might involve PAR4 translocation to the nucleus that is essential to execute cell death in H460 cells.

Differential induction of membrane-bound form of TRAIL in A549 and H460 cells. TRAIL was significantly elevated in RCM from H460 cells when compared with that from A549 cells (Fig. 2C). However, the mobilization of PAR4 to the nucleus was observed in both the A549 and H460 cells in response to radiation or rTRAIL (Fig. 3C). Because both sTRAIL and membrane-bound form of TRAIL (mTRAIL) have been described in transformed T and B cells, it was possible that ELISA results showed the presence of only the sTRAIL. The conversion of the mTRAIL to sTRAIL is regulated by cysteine protease (29). Thus, the differential release of TRAIL into the media could be due to the differences in the expression of cysteine protease regulating the expression of different forms of TRAIL. To test this hypothesis, we analyzed the mTRAIL expression in the A549 and H460 cells by immunofluorescence. While there is a significant increase in the mTRAIL in response to IR in A549 cells (Fig. 3C), a significant decrease was seen in mTRAIL in H460 cells exposed to IR (Fig. 3C). A549 and H460 cells were exposed to the RCM from A549 and H460 cells, respectively, and the localization of PAR4 was analyzed. RCM from H460, but not that from A549, cells caused translocation of PAR4 into the nucleus in both A549 and H460 cells (Fig. 3D). Furthermore, neutralizing antibody against TRAIL abrogated the effects of H460 RCM on PAR4 mobilization to the nucleus (Fig. 3D). These findings indicate that the differential bystander responses in these two cell lines could be due to the induction of membrane versus the soluble form of TRAIL. Thus, the proapoptotic effects of RCM from H460 cells are regulated by TRAIL and PAR4.

RCM swapping in A549 and H460 cells reverses the clonogenic response of the cells. To understand if the bystander-mediated cell survival responses are unique to the cell type or the soluble factors released in the RCM, we exposed the RCM from A549 and H460 cells to H460 and A549 cells, respectively, and assessed the surviving fraction. Isotransfer of RCM resulted in H460 cells being more sensitive to bystander effects when compared with A549 cells. However, survival effect was reversed when RCM was swapped (heterotransfer), in which A549 cells showed increased sensitivity. These observations show the specificity of soluble factors in RCM of both A549 and H460 in mediating the bystander responses (Fig. 4A and B). Similar results were observed in MEF cells with none or a single copy of Egr-1 gene when the RCM was swapped between these cells, suggestive of a role for EGR-1 in high-radiation dose–induced bystander signaling (Supplementary Fig. S3).

Radiation induces EGR-1 in A549 cells. Because TNF-α is a known target of EGR-1 (9), we investigated if the differences in bystander responses are due to EGR-1 function. Western analysis of the EGR-1 protein in A549 and H460 cells exposed to direct or RCM or RCM plus 2 Gy IR showed that EGR-1 protein level was significantly induced in A549 but was absent in H460 (Fig. 5A). Maximal binding of EGR-1 consensus probe was observed in RCM + 2 Gy dose compared with RCM exposure by electrophoretic mobility shift assay in A549 cells. This kind of specific binding was absent in H460 cells (Fig. 5B). The transactivation of EGR-1, measured using EBS-CAT reporter, was present in A549 cells but absent in H460 (Fig. 5C). CAT activity increased in A549 cells by 5.6-fold, 8.3-fold, and 8.3-fold in cells treated with RCM, 10 Gy IR, and RCM plus 2 Gy IR, respectively, indicating that A549 cells harbor functional EGR-1. To ascertain if EGR-1 in A549 cells regulates the expression of TNF-α in A549 cells, the binding of EGR-1 in TNF-α promoter was assessed by chromatin immunoprecipitation assay. Significant binding of EGR-1 to TNF-α promoter was detected in untreated and irradiated A549 cells compared with the absence of PCR product for HG6PD gene in immunoprecipitates (Fig. 6A), confirming the specificity of EGR-1 targeting the TNF-α promoter. In addition, slightly higher enrichment was observed for TNF-α promoter in cells treated with 10 Gy radiation compared with the untreated cells (Fig. 6A) indicating that radiation-induced EGR-1 protein is directed toward binding to TNF-α promoter. In addition, EGR-1 siRNA (Supplementary Fig. S4) caused significant decrease in expression of TNF-α in all treatment exposures (Fig. 6B). These observations strongly suggest that the TNF-α induction is mediated by EGR-1 in bystander responses in A549 cells.

EGR-1 function is essential in eliciting bystander-mediated apoptotic responses in fibroblast and A549 cells. Functional studies have shown that EGR-1 acts to increase the potency of apoptotic agents (9, 30, 31). In contrast, the studies using A549 and H460 cells clearly show that EGR-1 function might play an opposite role in regulating direct and bystander responses of A549 and H460 cells. Because the cell lines tested here were genetically unstable cells with complex aberrations, we used genetically matched isogenic MEF cells to study EGR-1 function. When compared with Egr-1^+/− MEF cells, the Egr-1^−/− MEF showed 50% diminished apoptosis in response to RCM or RCM plus 2 Gy IR (Fig. 6C). When EGR-1 protein expression was restored in Egr-1^−/− MEF cells by Ad/GFP-Egr-1 infection, the apoptotic response was reversed, resulting in a greater response than that observed in parental MEF cells harboring one allele of EGR-1 (Supplementary Fig. S5). These observations strongly indicate that EGR-1 function is pivotal in eliciting high-dose RIBEs. In addition, A549 cells, stably expressing EGR-1 siRNA, showed diminished apoptosis when compared with the control siRNA expressing A549 cells in response to 2 Gy IR, RCM, or RCM plus 2 Gy exposures (Fig. 6D). These findings suggest that EGR-1 regulates the high-dose bystander response in both epithelial and fibroblast cell type.

This study is the first direct evidence that high-dose X-ray irradiation can induce the release of factors into the cell culture to mediate bystander responses as measured by cell survival and apoptosis in lung adenocarcinoma. Previously, experiments using high-LET microbeam and biophysical modeling have shown that bystander effects are observed even at high doses of IR (32). Whereas the low-dose bystander effect in a biological system is of relevance to carcinogenesis and could have serious implications for radiation risk assessment, the bystander signaling induced by a high dose has therapeutic implications. High-dose RIBEs could be the underlying mechanism involved in a novel treatment modality, such as high-dose spatially fractionated radiation GRID therapy for management of advanced cancers (33–37). Significant induction of LDL-enriched ceramide, secretory SMase, and TNF-α in serum from patients treated with spatially fractionated radiation GRID therapy is suggestive of involvement of bystander signaling mechanisms in this therapeutic approach (19, 38). Bystander effects have been studied using either coculture or media transfer experiments. Whereas coculture experiments allow the study of effects before, during, and after radiation, bystander effects by media transfer experiments take into consideration the postirradiation effects. The events mediated by bystander signaling can be studied in isolation. However, studies on genomic instability after high LET radiation in murine primary hematopoeitic stem cells showed no difference, irrespective of the method used (39).

It has been shown that a protective adaptive response was elicited, where bystander cells (human lung fibroblasts exposed to media from 1 cGy) that are subsequently irradiated at doses 2 or 4 Gy were more radio resistant than cells not exposed to bystander signals (40). In our studies, the RCM from 10 Gy was effective in providing significant supersensitization of the effects of subsequent 2 Gy dose.

Using the inhibitors of reactive oxygen species (ROS), Yang et al. (41) showed that ROS has no effect on the survival of bystander human fibroblast cells, suggesting that irradiated cells (at 0.1–10 Gy) release toxic factors other than ROS into the medium (41). Several non-ROS factors that we had deciphered as bystander signals in clinical samples, such as TNF-α, SMase, and Ceramide, could be involved in mediating such bystander effects at high doses of radiation (19, 38). Thus, our study is the first one to report the induction of TNF-α and TRAIL as high-dose IR-induced bystander proapoptotic factors in lung adenocarcinoma cells.

TNF-α, bound to TNF receptor 1, may cause either survival or apoptosis, depending on the biochemical modification that determines the type of complex formed; one complex causes NF-κB activation (prosurvival), whereas the other complex recruits caspases and causes apoptosis (42). In this study, we found mobilization of NF-κB complex into the nucleus in A549 cells subjected to high-dose radiation direct and bystander exposures. These findings show that although TNF-α renders proapoptotic response to high-dose bystander effects, it might at the same time cause prosurvival signaling due to increased activation of NF-κB.

Activation of TRAIL receptors presents an alternative opportunity to destroy cancer cells as the TRAIL induces cell death predominantly in transformed cells (43). Despite the advantages of TRAIL over TNF-α–mediated therapy (44), constitutive activation of NF-κB dependent genes critically determines the susceptibility toward apoptosis induction by TRAIL receptors R1 and R2 (45). Induction of TRAIL was observed in lymphoma patients and several cell lines suggesting that TRAIL is a potential bystander signal in mediating abscopal effects in heamatopoietic malignancies (13). The enhanced release of TRAIL in RCM explained the increased cell death in RCM-exposed H460 cells. The decrease in cell death observed in A549 cells with the RCM from A549 cells could also be due to increased activation of NF-κB cells. Hence, the swapping of the RCM from the A549 and H460 cells reversed the cell survival effects, demonstrating a role of specific factors in mediating the cell death and cell survival signals (Fig. 4A and B). Similar experiments of swapping the medium in MEF cells (EGR-1^−/− and EGR-1^+/−; Supplementary Figs. S3 and S5) show that the effects of RCM derived from EGR-1–positive MEF cells compared with EGR-1–negative cells on cell survival shows a direct positive correlation of EGR-1 function and cell death. The decreased levels of mTRAIL observed in H460 cell membrane in response to IR could be due to induction of specific cysteine proteinase, which cleaves the mTRAIL to produce the soluble form. Identification of the factor regulating the expression of cysteine protease could explain the differential expression of TRAIL isoforms. Whereas sTRAIL and mTRAIL are known to act through TRAIL R1 receptor, TRAIL R2 receptor functions exclusively through the mTRAIL (46). Furthermore, it has been shown that nuclear PAR4 initiates programmed cell death (27). We found increased mobilization of PAR4 to the nucleus in response to the RCM from H460, but not A549, cells, and this was due to sTRAIL (Fig. 3C and D). This may be one of the mechanism by which PAR4-expressing cells are more prone to TRAIL effects as shown by Boehrer et al. (28). The sTRAIL could be mediating the PAR4 mobilization to the nucleus through TRAIL R1 receptor, because this receptor is activated by both forms of TRAIL (46). TRAIL signaling involves the activation of initiator and effector caspases that mediate the apoptotic events of the cell (47). Although it is not clear how TRAIL mobilizes PAR4, it is likely that TRAIL signaling leads to the cleavage of 14-3-3 proteins by caspases, which mobilizes PAR4 to the nucleus, analogous to association of BAD with BCL-xL (48, 49).

Since in this study we found that TNF-α is a crucial player in RIBEs, we reasoned that an upstream gene (EGR-1) that regulates TNF-α gene expression might play a role in bystander response. The functional role of Egr-1 in radiation-induced signaling is pivotal because the promoter of Egr-1 contains radiation-inducible CArG (CC(A/T)₆GG) DNA sequences (50). EGR-1 is regulated differentially in response to radiation in A549 and H460 cells. Although the molecular mechanism responsible for this differential regulation is not clear, it is possible that the different epigenetic regulatory mechanisms may be involved. Thus, our studies clearly show a role for EGR-1 in bystander-mediated apoptosis, both in the MEF and A549 cells.

In summary, IR induces EGR-1 expression in A549 cells that leads to increased expression of TNF-α protein. IR also increases the expression of TRAIL in both A549 and H460 cells through an unknown mechanism. However, the release of sTRAIL from H460 but not A549 cells may be due to selective induction of an uncharacterized cysteine protease in H460 cells by IR. mTRAIL, as well as the sTRAIL, is capable of nuclear translocation of PAR4 in A549 and H460 cells, respectively. However, activation of NF-κB by TNF-α in A549 cells may lead to abrogation of proapoptotic effects of both TNF-α and TRAIL, explaining the differential sensitivity of these two lung adenocarcinoma cells.

In conclusion, the high-dose RIBEs, like that induced by the low doses, depends on the genetic makeup of the target, as well as the bystander, cells. The functionality of EGR-1 in mediating the expression of TNF-α is a pivotal genetic factor in the bystander signaling events mediated by high-dose radiation in lung adenocarcinoma cells. Thus, understanding the genetic and epigenetic modifications and their interactions leading to differential bystander signaling could be useful in designing appropriate therapeutic approaches as adjuvants of high-dose GRID radiation therapy that is currently being used in the treatment of radio-resistant, bulky tumors.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.