The Role of the Nuclear Factor κB Pathway in the Cellular Response to Low and High Linear Energy Transfer Radiation

Astronauts are exposed to considerable doses of space radiation during long-term space missions. As complete shielding of the highly energetic particles is impracticable, the cellular response to space-relevant radiation qualities has to be understood in order to develop countermeasures and to reduce radiation risk uncertainties. The transcription factor Nuclear Factor κB (NF-κB) plays a fundamental role in the immune response and in the pathogenesis of many diseases. We have previously shown that heavy ions with a linear energy transfer (LET) of 100–300 keV/µm have a nine times higher potential to activate NF-κB compared to low-LET X-rays. Here, chemical inhibitor studies using human embryonic kidney cells (HEK) showed that the DNA damage sensor Ataxia telangiectasia mutated (ATM) and the proteasome were essential for NF-κB activation in response to X-rays and heavy ions. NF-κB’s role in cellular radiation response was determined by stable knock-down of the NF-κB subunit RelA. Transfection of a RelA short-hairpin RNA plasmid resulted in higher sensitivity towards X-rays, but not towards heavy ions. Reverse Transcriptase real-time quantitative PCR (RT-qPCR) showed that after exposure to X-rays and heavy ions, NF-κB predominantly upregulates genes involved in intercellular communication processes. This process is strictly NF-κB dependent as the response is completely absent in RelA knock-down cells. NF-κB’s role in the cellular radiation response depends on the radiation quality.


Introduction
Exposure to galactic cosmic radiation which consists of protons, α-particles and heavier nuclei is a major risk factor for long-term human space missions. On the International Space Station (ISS), the effective dose rate quantified by human phantom experiments amounted to 550-570 µSv/d inside the station and 690-720 µSv/d during extravehicular activities [1,2]. During a mission to Mars, astronauts will accumulate considerable doses of galactic cosmic radiation of about 1 Sv [3,4]. Due to mass limitations for spacecraft and production of secondary particles in shielding material, the radiation exposure cannot simply be avoided by radiation shielding of the spacecraft [5]. The potential space Figure 1. Kinetics of Nuclear Factor κB (NF-κB)-dependent destabilized enhanced green fluorescent protein (d2EGFP) expression after treatment with tumor necrosis factor α (TNF-α), X-rays and carbon ions, as detected in stably transfected human embryonic kidney cells (HEK) cells from d2EGFP reporter gene expression controlled by a promoter containing four NF-κB response elements (NRE) and the minimal promoter of thymidine kinase (TK) (HEK-pNF-κB-d2EGFP/Neo clone L2). HEK-pNF-κB-d2EGFP/Neo L2 cells were irradiated with X-rays (200 kV, linear energy transfer (LET) ~0.3-3 keV/µm) or 13 C-ions (75 MeV/n, LET 34 keV/µm) or incubated with 10 ng/mL TNF-α as a positive control. At different time points after induction, cells were trypsinated and fixed with 3.7% formaldehyde. d2EGFP expression was analyzed by flow cytometry.

Ataxia Telangiectasia Mutated (ATM) and the Proteasome Are Required for NF-κB Activation by X-rays and Heavy Ion Exposure
To determine whether ATM and the proteasome are involved in NF-κB activation by different ionizing radiation qualities, specific inhibitors were used. First, cytotoxicity of the inhibitors and the inhibitory concentration were determined (Figure 2A). 2 µmol/L MG-132 were sufficient to completely suppress TNF-α induced NF-κB activation ( Figure 2B). The ATM inhibitor KU-55933 had no effect on TNF-α triggered NF-κB activation, as it is not involved in the classical pathway ( Figure  2C). Therefore, the concentration was determined from the literature (10 µmol/L). Treatment with 2 µmol/L MG-132 or 10 µmol/L KU-55933 completely abolished the X-ray-induced NF-κB activation ( Figure 2D). Compared to X-rays in the same dose range, exposure with argon ions resulted in a higher NF-κB activation with a saturation at ~8 Gy ( Figure 2E). Additionally, this activation was completely suppressed by KU-55933 and by MG-132 ( Figure 2E). Kinetics of Nuclear Factor κB (NF-κB)-dependent destabilized enhanced green fluorescent protein (d2EGFP) expression after treatment with tumor necrosis factor α (TNF-α), X-rays and carbon ions, as detected in stably transfected human embryonic kidney cells (HEK) cells from d2EGFP reporter gene expression controlled by a promoter containing four NF-κB response elements (NRE) and the minimal promoter of thymidine kinase (TK) (HEK-pNF-κB-d2EGFP/Neo clone L2). HEK-pNF-κB-d2EGFP/Neo L2 cells were irradiated with X-rays (200 kV, linear energy transfer (LET) 0.3-3 keV/µm) or 13 C-ions (75 MeV/n, LET 34 keV/µm) or incubated with 10 ng/mL TNF-α as a positive control. At different time points after induction, cells were trypsinated and fixed with 3.7% formaldehyde. d2EGFP expression was analyzed by flow cytometry.

Ataxia Telangiectasia Mutated (ATM) and the Proteasome Are Required for NF-κB Activation by X-rays and Heavy Ion Exposure
To determine whether ATM and the proteasome are involved in NF-κB activation by different ionizing radiation qualities, specific inhibitors were used. First, cytotoxicity of the inhibitors and the inhibitory concentration were determined (Figure 2A). 2 µmol/L MG-132 were sufficient to completely suppress TNF-α induced NF-κB activation ( Figure 2B). The ATM inhibitor KU-55933 had no effect on TNF-α triggered NF-κB activation, as it is not involved in the classical pathway ( Figure 2C). Therefore, the concentration was determined from the literature (10 µmol/L). Treatment with 2 µmol/L MG-132 or 10 µmol/L KU-55933 completely abolished the X-ray-induced NF-κB activation ( Figure 2D). Compared to X-rays in the same dose range, exposure with argon ions resulted in a higher NF-κB activation with a saturation at~8 Gy ( Figure 2E). Additionally, this activation was completely suppressed by KU-55933 and by MG-132 ( Figure 2E). Ataxia telangiectasia mutated (ATM) and the proteasome are required for NF-κB activation by low and high LET radiation. Means and standard error (SE) of two to three independent experiments with three replicates each are shown. The viability of HEK-pNF-κB-d2EGFP/Neo L2 cells after treatment with increasing concentrations of ATM inhibitor KU-55933 and the proteasome inhibitor MG-132 was determined by means of the 3-(4,5-dimethylthiazole-2-yl-2,5-diphenyltetrazolium bromide (MTT) test 100 h after adding the inhibitor. The dotted line indicates 80% relative viability (A); the effect of MG-132 (B) and KU-55933 (C) on TNF-α-induced activation of the NF-κB pathway was determined in HEK-pNF-κB-d2EGFP/Neo L2 cells preincubated with the inhibitor for 1 h before addition of 10 ng/mL TNF-α. Cells were detached and fixed with 3.7% formaldehyde after 18 h. NF-κB-dependent d2EGFP-expression was measured by flow cytometry. The suppression of radiation-induced activation of the NF-κB pathway by KU-55933 and MG-132 was determined in HEK-pNF-κB-d2EGFP/Neo L2 cells after exposure to X-rays (200 kV, LET ~0.3-3 keV/µm, (D) and 36 Ar-ions (95 MeV/n, LET 270 keV/µm, (E) cells were preincubated with the inhibitor for 1 h and subsequently irradiated. After 18 h, NF-κB-dependent d2EGFP-expression was measured by flow cytometry. The percentage of d2EGFP(+) cells was normalized to the mock-irradiated control. For 36 Ar ions, doses (Gy) were calculated from fluences (P/cm 2 ) according to Equation (1) (Section 4.5.3.).

Figure 2.
Ataxia telangiectasia mutated (ATM) and the proteasome are required for NF-κB activation by low and high LET radiation. Means and standard error (SE) of two to three independent experiments with three replicates each are shown. The viability of HEK-pNF-κB-d2EGFP/Neo L2 cells after treatment with increasing concentrations of ATM inhibitor KU-55933 and the proteasome inhibitor MG-132 was determined by means of the 3-(4,5-dimethylthiazole-2-yl-2,5-diphenyltetra-zolium bromide (MTT) test 100 h after adding the inhibitor. The dotted line indicates 80% relative viability (A); the effect of MG-132 (B) and KU-55933 (C) on TNF-α-induced activation of the NF-κB pathway was determined in HEK-pNF-κB-d2EGFP/Neo L2 cells preincubated with the inhibitor for 1 h before addition of 10 ng/mL TNF-α. Cells were detached and fixed with 3.7% formaldehyde after 18 h. NF-κB-dependent d2EGFP-expression was measured by flow cytometry. The suppression of radiation-induced activation of the NF-κB pathway by KU-55933 and MG-132 was determined in HEK-pNF-κB-d2EGFP/Neo L2 cells after exposure to X-rays (200 kV, LET~0.3-3 keV/µm, (D) and 36 Ar-ions (95 MeV/n, LET 270 keV/µm, (E) cells were preincubated with the inhibitor for 1 h and subsequently irradiated. After 18 h, NF-κB-dependent d2EGFP-expression was measured by flow cytometry. The percentage of d2EGFP(+) cells was normalized to the mock-irradiated control. For 36 Ar ions, doses (Gy) were calculated from fluences (P/cm 2 ) according to Equation (1) (Section 4.5.3.).

Effect of RelA Knock-Down on NF-κB-Dependent Gene Expression
The reporter gene and inhibitor experiments had shown that ATM and the proteasome are involved in ionizing radiation-induced NF-κB activation. For specific downregulation of the NF-κB pathway, the key component RelA was knocked down by stable transfection of four different RelA shRNA plasmids carrying a hygromycin resistance gene. The knock-down level in polyclonal hygromycin-resistant cells resulting from transfection with RELA-1 and -4 plasmids was above 40% ( Figure 3A). Clones from these populations were grown, and three clones reached a knock-down level above 70%. The clone with the highest knock-down was selected for further experiments and designated HEK shRNA RelA ( Figure 3B, arrow). Int

Effect of RelA Knock-Down on NF-κB-Dependent Gene Expression
The reporter gene and inhibitor experiments had shown that ATM and the proteasome are involved in ionizing radiation-induced NF-κB activation. For specific downregulation of the NF-κB pathway, the key component RelA was knocked down by stable transfection of four different RelA shRNA plasmids carrying a hygromycin resistance gene. The knock-down level in polyclonal hygromycin-resistant cells resulting from transfection with RELA-1 and -4 plasmids was above 40% ( Figure 3A). Clones from these populations were grown, and three clones reached a knock-down level above 70%. The clone with the highest knock-down was selected for further experiments and designated HEK shRNA RelA ( Figure 3B, arrow). Figure 3. Verification of RelA knock-down on messenger RNA (mRNA) level after transfection of HEK-pNF-κB-d2EGFP/Neo L2 cells with RelA short-hairpin RNA (shRNA) plasmids. Polyclonal hygromycin-resistant cells resulting from transfection with the shRNA plasmids RELA-1 to 4 were grown in petri dishes and total RNA was collected after 3 days (A). Clones were grown from the cell population transfected with the shRNA plasmids RELA-1 and 4, as these were most effective in suppressing RelA expression, and the knock-down level of RelA mRNA was determined in these clones (B). RelA mRNA expression was determined by real-time RT-qPCR. Means and SD of three independent experiments with each three replicates are shown.  . Verification of RelA knock-down on messenger RNA (mRNA) level after transfection of HEK-pNF-κB-d2EGFP/Neo L2 cells with RelA short-hairpin RNA (shRNA) plasmids. Polyclonal hygromycin-resistant cells resulting from transfection with the shRNA plasmids RELA-1 to 4 were grown in petri dishes and total RNA was collected after 3 days (A). Clones were grown from the cell population transfected with the shRNA plasmids RELA-1 and 4, as these were most effective in suppressing RelA expression, and the knock-down level of RelA mRNA was determined in these clones (B). RelA mRNA expression was determined by real-time RT-qPCR. Means and SD of three independent experiments with each three replicates are shown.
As the selected HEK shRNA RelA clone showed a remaining RelA expression of~17%, the effect of this residual level on basal expression (Table 1) and on TNF-α-induced expression of NF-κB target genes (Table 2) was analyzed.  As expected, RelA mRNA was downregulated. In addition, cyclin D1 and IL-8 expression were reduced in HEK shRNA RelA cells compared to the parental cell line ( Table 1). The expression of CCL5, GADD45B and Jun was increased in cells with RelA downregulation.
Incubation with TNF-α induced a large set of NF-κB target genes in HEK cells, with highest expression of TNF, IL8, CXCL1, 2 and 10, and NF-kappa-B inhibitor alpha (NFKBIA) ( Table 2). RelA knock-down nearly completely abolishes the TNF-α-induced expression of all these target genes except CCL5. EXO1 which is not a NF-κB target gene is downregulated when RelA is knocked down.
Furthermore, NF-κB-dependent reporter gene expression was determined. The basal level of d2EGFP expression was unchanged compared to the parental cell lines and to cells stably transfected with a shRNA control plasmid with a random sequence ( Figure 4A). TNF-α-and X-ray-induced d2EGFP expression were significantly reduced in HEK shRNA RelA cells compared to the parental cell line as well as to the shRNA control cell line (p ≤ 0.05, Figure 4A). The reactions of the parental cell line and the shRNA control cell line were not significantly different in no case based on a p = 2α level of <0.05. Carbon ion induced d2EGFP expression was completely abolished ( Figure 4B).

Growth of RelA Knock-Down Cells
In order to determine whether RelA knock-down affects basic cellular functions such as growth, cell numbers were counted during a growth period of 10 days. HEK shRNA RelA cells showed a prolonged lag phase compared to HEK-pNF-κB-d2EGFP/Neo clone L2 cells ( Figure 5). Once proliferation starts, both cell lines grow with the same velocity.  . Effect of RelA knock-down on NF-κB activation by X-rays and TNF-α (A), and by carbon ions (B). HEK-pNF-κB-d2EGFP/Neo L2 cells, cells stably transfected with the shRNA control vector (HEK shRNA control) or the RelA shRNA plasmid (HEK shRNA RelA) were seeded in petri dishes, grown for two days, and exposed to X-rays (200 kV, LET~0.3-3 keV/µm), incubated with 10 ng/mL TNF-α (A) or irradiated with 13 C-ions (75 MeV/n, LET 34 keV/µm). 18 h after exposure, cells were harvested by trypsination, fixed with 3.5% formaldehyde and the percentage of d2EGFP (+) cells was determined by flow cytometry.

Growth of RelA Knock-Down Cells
In order to determine whether RelA knock-down affects basic cellular functions such as growth, cell numbers were counted during a growth period of 10 days. HEK shRNA RelA cells showed a prolonged lag phase compared to HEK-pNF-κB-d2EGFP/Neo clone L2 cells ( Figure 5). Once proliferation starts, both cell lines grow with the same velocity. . Effect of RelA knock-down on NF-κB activation by X-rays and TNF-α (A), and by carbon ions (B). HEK-pNF-κB-d2EGFP/Neo L2 cells, cells stably transfected with the shRNA control vector (HEK shRNA control) or the RelA shRNA plasmid (HEK shRNA RelA) were seeded in petri dishes, grown for two days, and exposed to X-rays (200 kV, LET ~0.3-3 keV/µm), incubated with 10 ng/mL TNF-α (A) or irradiated with 13 C-ions (75 MeV/n, LET 34 keV/µm). 18 h after exposure, cells were harvested by trypsination, fixed with 3.5% formaldehyde and the percentage of d2EGFP (+) cells was determined by flow cytometry.

Growth of RelA Knock-Down Cells
In order to determine whether RelA knock-down affects basic cellular functions such as growth, cell numbers were counted during a growth period of 10 days. HEK shRNA RelA cells showed a prolonged lag phase compared to HEK-pNF-κB-d2EGFP/Neo clone L2 cells ( Figure 5). Once proliferation starts, both cell lines grow with the same velocity.

Survival of RelA Knock-Down Cells After X-ray and after Heavy Ion Exposure
The survival curves after exposure of HEK-pNF-κB-d2EGFP/Neo clone L2 cells and HEK shRNA RelA cells were of curvilinear shape ( Figure 6). The curve of the RelA knock-down cells is significantly steeper, indicating a higher radiosensitivity. The D 0 indicating the dose necessary to reduce survival of HEK cells to 37% is 1.12 Gy for the parental cell line compared to 0.82 Gy for the RelA knock-down cells (Table 3).

Survival of RelA Knock-Down Cells After X-ray and after Heavy Ion Exposure
The survival curves after exposure of HEK-pNF-κB-d2EGFP/Neo clone L2 cells and HEK shRNA RelA cells were of curvilinear shape ( Figure 6). The curve of the RelA knock-down cells is significantly steeper, indicating a higher radiosensitivity. The D0 indicating the dose necessary to reduce survival of HEK cells to 37% is 1.12 Gy for the parental cell line compared to 0.82 Gy for the RelA knock-down cells (Table 3). High-LET radiation exposure of HEK cells results in purely exponential survival curves ( Figure  7). Based on energy dose, heavy ions with an LET of 55 keV/µm are most efficient in cell killing ( Figure 7A), while radiation qualities with an LET above or below this range are less efficient in cell killing ( Figure 7A,B). The D0 first decreases to 0.47 Gy for silicon ions, then increases with increasing LET to 0.72 Gy for argon ions (Table 3).   High-LET radiation exposure of HEK cells results in purely exponential survival curves ( Figure 7). Based on energy dose, heavy ions with an LET of 55 keV/µm are most efficient in cell killing ( Figure 7A), while radiation qualities with an LET above or below this range are less efficient in cell killing ( Figure 7A,B). The D 0 first decreases to 0.47 Gy for silicon ions, then increases with increasing LET to 0.72 Gy for argon ions (Table 3).

Induction of NF-κB Target Gene Expression by Exposure to Different Radiation Qualities
As NF-κB was weakly activated by X-rays and activated by heavy ions to a higher extent, dependent on LET, but only for X-rays, a reduction of survival in case of RelA downregulation was observed, the expression of 88 NF-κB target genes was profiled 6 h after exposure to X-rays and heavy ions ( Table 4).
The expression of chemokines and cytokines was upregulated by low-and high-LET radiation at higher doses. Exposure to 0.5 Gy X-rays or Ti ions did not significantly up-regulate the expression of the investigated NF-κB target genes. The chemokine and cytokine expression was in general higher after exposure to 4 Gy Ti ions compared to 4 Gy X-rays. The highest expression levels were reached for CXCL10 and TNF.
Adhesion molecules such as VCAM-1 and a costimulatory factor (CD83) were also induced. Components of the NF-κB pathway such as NFKB2 and the inhibitor NFKBIA were upregulated by the two radiation qualities.
RelA knock-down inhibited the ionizing radiation-induced NF-κB target gene expression except for TNFRSF1B.
Incubation with TNF-α resulted in strong secretion of IL-8 by HEK-pNF-κB-d2EGFP/Neo L2 cells within 12 h, and the IL-8 concentration in the cell culture supernatants further increased within the next 36 h ( Figure 8A). High doses of X-rays (8 and 16 Gy) are required to induce significant IL-8 secretion 12 h after irradiation, and this secretion is further enhanced at later time points (24 h and 48 h). In HEK shRNA RelA cells, the level of IL-8 secretion in mock-irradiated cells (0 Gy) is lower compared to HEK-pNF-κB-d2EGFP/Neo L2 cells, and only the highest X-ray dose (16 Gy) and incubation with TNF-α provoke a very slight time-dependent increase in IL-8 secretion ( Figure 8B).

Induction of NF-κB Target Gene Expression by Exposure to Different Radiation Qualities
As NF-κB was weakly activated by X-rays and activated by heavy ions to a higher extent, dependent on LET, but only for X-rays, a reduction of survival in case of RelA downregulation was observed, the expression of 88 NF-κB target genes was profiled 6 h after exposure to X-rays and heavy ions ( Table 4).
The expression of chemokines and cytokines was upregulated by low-and high-LET radiation at higher doses. Exposure to 0.5 Gy X-rays or Ti ions did not significantly up-regulate the expression of the investigated NF-κB target genes. The chemokine and cytokine expression was in general higher after exposure to 4 Gy Ti ions compared to 4 Gy X-rays. The highest expression levels were reached for CXCL10 and TNF.
Adhesion molecules such as VCAM-1 and a costimulatory factor (CD83) were also induced. Components of the NF-κB pathway such as NFKB2 and the inhibitor NFKBIA were upregulated by the two radiation qualities.
RelA knock-down inhibited the ionizing radiation-induced NF-κB target gene expression except for TNFRSF1B.
Incubation with TNF-α resulted in strong secretion of IL-8 by HEK-pNF-κB-d2EGFP/Neo L2 cells within 12 h, and the IL-8 concentration in the cell culture supernatants further increased within the next 36 h ( Figure 8A). High doses of X-rays (8 and 16 Gy) are required to induce significant IL-8 secretion 12 h after irradiation, and this secretion is further enhanced at later time points (24 h and 48 h). In HEK shRNA RelA cells, the level of IL-8 secretion in mock-irradiated cells (0 Gy) is lower compared to HEK-pNF-κB-d2EGFP/Neo L2 cells, and only the highest X-ray dose (16 Gy) and incubation with TNF-α provoke a very slight time-dependent increase in IL-8 secretion ( Figure 8B).  48 Ti-ions (1000 MeV/n, LET 108 keV/µm). ** The table shows all genes from the NF-κB signaling array, that are more than threefold up-or down-regulated (in bold) in irradiated cells compared to mock-irradiated cells. The mean of the fold up-or downregulation was calculated from up to two independent experiments with untreated cells. Prior to investigating gene expression with RT2 Profiler TM PCR arrays, a proper RNA and cDNA quality was confirmed. ACTB, B2M, GAPDH, HPRT1 and RPLP0 were used as reference genes to normalize the results. TNF-α secretion increased 12 h after exposure of HEK-pNF-κB-d2EGFP/Neo L2 cells to 16 Gy X-rays and remained elevated during the investigated time period without further significant increase ( Figure 8C). The total amount was much lower compared to IL-8. For HEK shRNA RelA cells, TNF-α secretion remained within background levels for all applied doses and incubation periods ( Figure 8D). TNF-α secretion induced by incubation with TNF-α is not shown, as the added TNF-α was present in the supernatants and disguised the TNF-α secretion of the cells.

Discussion
In this study, we show that knock-down of the NF-κB subunit RelA results in sensitization of HEK cells towards low-LET X-rays, but not towards heavy ions in a LET range of 10 to 272 keV/µm. It was previously demonstrated that heavy ions in the LET range of 91 to 272 keV/µm strongly activate NF-κB [20]. The NF-κB activation by low-and high-LET ionizing radiation was shown to be ATM dependent. Furthermore, IκB-degradation by the proteasome is essential for ionizing radiation-induced NF-κB-activation.
RelA knock-down also influenced cell proliferation, resulting in a prolonged lag phase, but it did not impact the proliferation rate once the exponential growth phase was reached. Without NF-κB-activation by TNF-α or ionizing radiation, the reduced RelA levels in the knock-down cells TNF-α secretion increased 12 h after exposure of HEK-pNF-κB-d2EGFP/Neo L2 cells to 16 Gy X-rays and remained elevated during the investigated time period without further significant increase ( Figure 8C). The total amount was much lower compared to IL-8. For HEK shRNA RelA cells, TNF-α secretion remained within background levels for all applied doses and incubation periods ( Figure 8D). TNF-α secretion induced by incubation with TNF-α is not shown, as the added TNF-α was present in the supernatants and disguised the TNF-α secretion of the cells.

Discussion
In this study, we show that knock-down of the NF-κB subunit RelA results in sensitization of HEK cells towards low-LET X-rays, but not towards heavy ions in a LET range of 10 to 272 keV/µm. It was previously demonstrated that heavy ions in the LET range of 91 to 272 keV/µm strongly activate NF-κB [20]. The NF-κB activation by low-and high-LET ionizing radiation was shown to be ATM dependent. Furthermore, IκB-degradation by the proteasome is essential for ionizing radiation-induced NF-κB-activation.
RelA knock-down also influenced cell proliferation, resulting in a prolonged lag phase, but it did not impact the proliferation rate once the exponential growth phase was reached. Without NF-κB-activation by TNF-α or ionizing radiation, the reduced RelA levels in the knock-down cells changed the expression of only a few NF-κB target genes (CCND1, CXCL8, JUN, TNF). After incubation with TNF-α or irradiation, the upregulation of CXCL1, 2, 8 and 10, NFKBIA and TNF was strongly reduced in RelA knock-down cells.

Activation of the NF-κB Pathway by Different Radiation Qualities
Our first study examining the effects of heavy ions on the NF-κB pathway revealed that high-LET argon ions (272 keV/µm) activated the sequence of all events in the NF-κB pathway from DNA binding via transcription to protein translation and maturation [18]. The study of the LET dependence of this cellular response to heavy ion exposure disclosed that indeed the maximal potency to activate NF-κB dependent d2EGFP gene expression in HEK cells occurs at 272 keV/µm (argon ions) have [20]. In accordance with the results presented here, other authors using a DNA binding assay report nuclear translocation NF-κB after exposure of normal human monocytes (MM6 cells) after exposure to 0.7 Gy of 56 Fe ions [29]. Activation of NF-κB was mediated through phosphorylation of IκBα via subsequent proteasomal degradation. The study on iron ion effects on monocytes revealed only binding of NF-κB to its consensus sequence of 5 -GGGGACTTTCC-3 , transcriptional activation itself was not shown.
Also in vivo, NF-κB activation was observed in the heart and bone marrow of mice one week, one month and six months after exposure to 300 MeV/n 28 Si ions (LET 77 keV/µm) [30]. This clearly indicates that high-LET ion exposure induces rapid and persistent NF-κB activation in vitro and in vivo.

NF-κB Activation by Ionizing Radiation Is ATM Dependent
KU-55933 is specific for ATM with respect to inhibition of other related kinases such as DNA-dependent protein kinase (DNA-PK) or phosphatidylinositol 3 -kinase (PI3K) which required ã 200 times higher concentration for their inhibition [31]. Results showed that while it was not possible to suppress TNF-α-induced NF-κB activation, KU-55933 suppressed ionizing radiation-induced NF-κB activity completely. Increasing doses of X-rays and 36 Ar ions were used to induce NF-κB, showing its dose and LET dependency, as the percentage of d2EGFP (+) cells increased with increasing doses and 36 Ar ions had a higher impact on d2EGFP-dependent NF-κB expression than X-rays.
These results support the assumption that ionizing radiation, including high-LET heavy ions, activates the genotoxic stress induced NF-κB pathway and proves that NF-κB activation induced by ionizing radiation is strictly ATM dependent. ATM foci form quickly in the nuclei of human cells after exposure to carbon or iron ions and are believed to facilitate or initiate DNA double strand break (DSB) repair by Non-Homologous Endjoining (NHEJ) and Homologous Recombination (HR) [32,33]. ATM was shown to be activated in human cells to a higher extent by carbon ions (LET 290 keV/µm) than by low-LET γ-rays [34]. This stronger ATM phosphorylation after heavy ion exposure might be important for the strong NF-κB activation that was observed in the LET range of 90-300 keV/µm.
In contrast, TNF-α induced NF-κB activation was not impaired by KU-55933, as ATM is not involved in the classical pathway, which is activated by TNF-α.

IκB Degradation by the Proteasome Is Essential for TNF-α and Ionizing Radiation-Induced NF-κB Activity
Within non-cytotoxic concentrations (viability above 80%), MG-132 suppressed TNF-α-and ionizing radiation-induced NF-κB activity, showing that IκB degradation by the proteasome is an essential step in both, the genotoxic stress and the TNF-α-induced pathway. However, it has to be considered, that disturbing proteasome functions with MG-132 leads to the loss of regulation of essential functions within the cell and these chemical inhibitors are not NF-κB specific.

Role of RelA in Cell Proliferation
Downregulation of RelA prolonged the lag phase. Furthermore, cells required a longer recovery period after freezing and thawing, indicating a role of RelA in cell proliferation. This is supported by the observation that transfection of NF-κB/RelA antisense oligodeoxynucleotide reduces the proliferation of SGC-7901 gastric cancer cells [35] and that siRNA-mediated RelA knock-down decreased the proliferation ability of human esophageal squamous cell cancer (ESCC) EC9706 cells [36]. In tumor cells, the inhibiting effect of RelA knock-down seems to be much more pronounced compared to HEK cells. This might be explained by a constitutive NF-κB activation in many tumors, e.g., some gliomas [37] and hepatomas [38], which is then suppressed. In breast cancer cells, transfection of RelA shRNA reduced proliferation [39].
The reduced expression of cyclin D1 (−3.0) is a possible explanation for the effects of RelA knock-down on proliferation, as RelA and JunD were shown to cooperate to activate the proximal κB site of the cyclin D1 promoter in presence of serum [40]. Cyclin D1 regulates the cyclin-dependent kinases CDK4 or CDK6, the activity of which is required for cell cycle G1/S transition. This connection of NF-κB, cyclin D1 and proliferation was also found in neural stem or progenitor cells (NSCs, NPCs) [41] and fibroblasts [42]. Upregulation of the activated protein 1 (AP-1) family member Jun (5.0) in HEK shRNA RelA cells might be a regulatory mechanism to compensate for the RelA loss, as c-Jun affects cell proliferation, migration and invasion and, like NF-κB, actively participates in tumorigenesis [43].

Role of RelA in Clonogenic Survival after Exposure to Different Radiation Qualities
As activation of the NF-κB pathway is supposed to play a role in the negative regulation of apoptosis, survival of cells with residual DNA damage might thus be favored after irradiation. Even if this happens only in rare cases after ionizing radiation exposure, it might result in initiation or promotion of cancer from such wise affected cells. Cellular responses which may be related to better survival of exposed cells would contribute considerably to astronauts' radiation risk, especially at low doses when transient induction of genes involved in maintaining DNA fidelity and in modulating cell cycle progression and cell death would occur [44]. As heavy ion exposure strongly activated NF-κB, we examined whether NF-κB gives HEK cells a survival advantage after heavy ion exposure. For this purpose, expression of the NF-κB subunit RelA was downregulated by stable transfection of RelA shRNA.
The colony forming ability test showed that HEK cells were more sensitive to X-irradiation when RelA was downregulated. Such sensitization to X-rays was also observed with chemical NF-κB inhibitors, e.g., the sesquiterpene lactone parthenolide, an IκB kinase (IKK) inhibitor, inhibited constitutive and radiation-induced NF-κB binding activity in p53 null PC-3 cells and enhanced their X-ray sensitivity by a dose modification factor of 1.7 [45]. RelA knock-out murine embryonic fibroblasts (MEF p65 −/− ) were 1.3-fold more sensitive towards X-rays than the p65 +/+ cells [46]. In many cancer cell types, NF-κB inhibition sensitized the cells to ionizing radiation [47][48][49][50].
In these MEF p65 −/− cells, the killing effect of UVB radiation was lower compared to MEF wildtype cells (p65 +/+ ) [51]. Therefore, in contrast to ionizing radiation, loss of RelA conferred a strong resistance to UVB-induced cell death which was accompanied by reduced Gadd45 gene expression [51].
In contrast to the sensitization to X-rays by RelA knock-down, the sensitivity to heavy ion exposure was equal or only slightly different for cells with and without RelA over a wide range of LET (10-271 keV/µm). This indicates that RelA and heavy ion induced NF-κB activation does not procure a survival advantage for the exposed cells. This might be explained by other pro-death signals of the heavy ion exposure. Interestingly, in human hematopoietic stem and progenitor cells (HSPC) and peripheral blood lymphocytes (PBL), inhibition of NF-κB activation by expression of a super-repressor variant of IκBα (IκBα-SR) altered DNA double-strand-break repair [52]. IκBα-SR expression reduced homologous DSB repair in PBL by 33% [52]. NHEJ of DNA DSBs was compromised in HSPC and PBL (reduction to 37% and 52%, respectively) [52]. These measurements were performed on enzymatically induced DNA DSBs, results for low-or even high-LET radiation-induced DSBs are not yet available. Nevertheless, reduced DNA DSB repair in case of NF-κB repression or down-regulation might play a role for survival after ionizing radiation exposure. Delayed or incomplete repair might override other NF-κB induced survival signals. Another explanation for the missing effect of RelA knockdown might be the high-LET which is in the range of the highest observed biological efficiency. The cell killing effect in the LET range of 100-200 keV/µm is already very high, so that disturbed cellular signaling pathways might not be able to further increase cell killing. This was observed for the pro-apoptotic p53 in some cell systems: apoptosis was independent of the p53 status [53,54].
Comparing ionizing radiation and chemotherapeutics such as the topoisomerase II inhibitor doxorubicin, knock-down of RelA by siRNA increased the killing effect of doxorubicin in HepG2 cells [38]. The authors of this publication explain the increased killing of hepatoma cells by lower NF-κB-mediated expression of multi-drug resistance (MDR1) gene [38]. A mechanism based on MDR1 is not likely to influence survival after ionizing radiation. In nude mice bearing HeLa cervical cancer xenografts, downregulation of p65 via photothermal transfection of p65 siRNA increased irinotecan-induced tumor apoptosis [55].
Furthermore, the NF-κB target gene expression profile might be different for low-and high-LET radiation exposure and might give an explanation for the different role in the cellular radiation response. Therefore, the expression of 88 target genes was determined after X-ray and heavy ion exposure. Genes that are regulated more than threefold and at the same time are controlled by RelA are discussed below.

Role of RelA in the Expression of NF-κB Target Genes after Exposure to Different Radiation Qualities
NF-κB target genes that are induced after irradiation affect intercellular communication and many of them are proinflammatory. e.g., TNF-α activates inflammation by inducing a proinflammatory cytokine cascade, mediated through expression of IL8 [56]. Incubation with TNF-α induces the expression of a larger subset of genes than low-or high-LET irradiation at a dose of 4 Gy ( Figure 9A). TNF, CXCL1 and −10 are upregulated by TNF-α, 4 Gy X-rays and 4 Gy Ti ions. A higher X-ray dose (8 Gy) increased the overlap in the subset of induced NF-κB target genes ( Figure 9B). The gene subsets show overlap with those determined in osteosarcoma cells [57].
The overlapping gene expression in HEK-pNF-κB-d2EGFP/Neo L2 cells after TNF-α and radiation treatment was not observed in RelA knock-down cells, confirming the role of the NF-κB subunit RelA in the expression of these genes.
After TNF-α treatment HEK-pNF-κB-d2EGFP/Neo L2 cells but not in HEK shRNA RelA cells showed a clear upregulation of C-X-C motif (CXC, C stands for cysteine, X for any other amino acid) chemokine genes, including CXCL1, CXCL10, CXCL2 (MIP-2) and CXCL8 (IL8). The CXC chemokines belong to one group of small cytokines, which, when they are are secreted by infected cells act as chemoattractants inducing chemotaxis or migration especially of neutrophils, which are the first responders during inflammation to migrate towards site of inflammation. Additionally, in wound healing the increased chemokine concentration of damaged cells results in attracting cells to site of damage. Loss of chemokine induction is a serious impairment for the organism as it leads to dysfunctional cell migration and wound healing deficiency. The essential role of TNF in the early induction of chemokines and subsequent recruitment of leukocytes have been shown using models of mycobacterial infection [58] as induction of chemokines and cellular recruitment were delayed in TNF −/− cells. Animal studies show that CXCL2 was upregulated in murine tumors after exposure to carbon ions (290 MeV/n, LET 50 keV/µm) [59].
As expected, TNF-α treatment induced high upregulation (60.0) of TNF, as TNF-α, after being released as a response to immunological stimuli, acts as a positive autocrine feedback signal to augment NF-κB activation [60]. In the knock-down cell line, no upregulation of TNF was observed, as NF-κB is a key transcription factor involved in the synthesis of TNF-α [61]. Upregulation of NFKBIA (encoding IκBα) in the original cell line was expected as well, as in addition to increasing the transcription of cytokines and adhesion proteins, NF-κB also increases the transcription of the inhibitor IκB, thus leading to its own inactivation and subsequent termination of the response [62]. Other studies showed that NFKBIA was also upregulated in six human malignant melanoma cell lines after carbon ion exposure (290 MeV/n, LET 50 keV/µm) [63].
Only three of the genes upregulated in HEK-pNF-κB-d2EGFP/Neo L2 cells were induced by TNF-α in the knock-down cell line: CCL5, TNF and TNFRSF1B. For the latter two genes, the induction factor was much lower compared to the RelA proficient cell line. Instead, downregulation of one NF-κB target gene was observed, EXO1. Exonuclease 1 (EXO1) is involved in DNA damage signaling [64].
The target gene array, investigating 88 genes after X-and 48 Ti ion irradiation, showed that NF-κB-dependent gene expression does not only depend on dose, but also on LET, complementing our recently published study [26] and studies with Caenorhabitidis elegans [65]. The extent of gene expression increases with increasing doses. Further, the degree of regulation is stronger after 4 Gy of 48 Ti ion exposure than after X-irradiation of the same dose.
In HEK-pNF-κB-d2EGFP/Neo L2 cells, the set of genes, that has already been shown to be upregulated after TNF-α treatment, was likewise upregulated after exposure to X-rays and heavy ions. It includes CXC chemokines, TNF and NFKBIA which encodes the NF-κB inhibitor IκBα. This Upregulation of NFKBIA (encoding IκBα) in the original cell line was expected as well, as in addition to increasing the transcription of cytokines and adhesion proteins, NF-κB also increases the transcription of the inhibitor IκB, thus leading to its own inactivation and subsequent termination of the response [62]. Other studies showed that NFKBIA was also upregulated in six human malignant melanoma cell lines after carbon ion exposure (290 MeV/n, LET 50 keV/µm) [63].
Only three of the genes upregulated in HEK-pNF-κB-d2EGFP/Neo L2 cells were induced by TNF-α in the knock-down cell line: CCL5, TNF and TNFRSF1B. For the latter two genes, the induction factor was much lower compared to the RelA proficient cell line. Instead, downregulation of one NF-κB target gene was observed, EXO1. Exonuclease 1 (EXO1) is involved in DNA damage signaling [64].
The target gene array, investigating 88 genes after X-and 48 Ti ion irradiation, showed that NF-κB-dependent gene expression does not only depend on dose, but also on LET, complementing our recently published study [26] and studies with Caenorhabitidis elegans [65]. The extent of gene expression increases with increasing doses. Further, the degree of regulation is stronger after 4 Gy of 48 Ti ion exposure than after X-irradiation of the same dose.
In HEK-pNF-κB-d2EGFP/Neo L2 cells, the set of genes, that has already been shown to be upregulated after TNF-α treatment, was likewise upregulated after exposure to X-rays and heavy ions. It includes CXC chemokines, TNF and NFKBIA which encodes the NF-κB inhibitor IκBα. This shows that expression of these genes is activated not only by the classical NF-κB pathway, but also the genotoxic stress induced, independent of the radiation quality.
In this work, both low-and high-LET radiation exposure resulted in upregulation of TNF, and with higher LET, the upregulation was stronger. Interestingly, in the tibial marrow of Fe ion (600 MeV/n, LET 174 keV/µm) irradiated mice, no significant increase in TNF expression was found three days after exposure to 2 Gy, and only a slight increase after 2 Gy γ-irradiation which remained elevated through day three after irradiation [66].
Especially the CXC chemokines are strongly upregulated, in case of CXCL10 up to 14.6 fold after irradiation with 8 Gy X-rays. None of the genes that are upregulated at higher doses in the original cell line exceeds the threshold of 3 after irradiation with 0.5 Gy of both, X-rays and 48 Ti ions. Though, after 4 Gy of X-rays, CXC chemokines reach expression values of 2.3 to 4.7 fold. For the same dose, these genes are expressed up to 7.6 fold after 48 Ti ion irradiation.
CCL5 has chemotactic activity for monocytes and basophils and binds to chemokine receptors CCR2 and CCR4. It was upregulated by high doses of X-rays. Such an upregulation was also observed in vivo in the bone marrow of C57BL/6J mice exposed to 2 Gy γ-rays and is suggested to promote ostoclastogenesis [66].
Apart from the genes expressed after both, TNF-α treatment and irradiation, additional genes are upregulated only upon X-irradiation, not after heavy ion exposure, proving the relation between radiation quality or LET and gene expression. These genes include CCL5, NFKB2, and VCAM1.
NFKB2 codes for the NF-κB subunit p100, which is involved in the non-canonical NF-κB pathway. It is processed to p52. This element of the non-canonical pathway seems to be upregulated only by X-rays and not by heavy ions. The RelB:p52 dimer was shown to confer radio-resistance to prostate cancer cells by upregulating the expression of the antioxidant enzyme manganese superoxide dismutase (MnSOD) [67][68][69][70]. An upregulation of RelB expression 8 h after X-ray exposure (0.1, 1 Gy) was also observed in a three-dimensional skin tissue model [71].
The different gene expression profile after 48 Ti ion irradiation indicates distinct responses, depending on radiation quality, eventually including different repair mechanisms, as indicated in the survival curves in this work. These genes are expressed after X-irradiation, but not after 48 Ti ion exposure. The missing expression of the protecting RelB gene after heavy ion exposure might be another explanation why RelA knock-down results in higher X-ray sensitivity, but not in higher heavy ion exposure sensitivity: Its protecting effect ceases to exist in HEK shRNA RelA cells after X-irradiation, but it is not induced after heavy exposure even in case of normal RelA levels.
CD83 expression was activated by 4 Gy Ti ions, TNF-α and 8 Gy X-rays. It is one of the central regulatory molecules in immune functions with anti-tumor effects [72]. It acts as a co-stimulator during activation of T-cells by antigen-presenting cells and is therefore essential for the induction of the adaptive immune response. NF-κB regulates inducible CD83 gene expression in activated T lymphocytes during an adaptive immune response [73]. Increased expression of this costimulatory molecule indicates that exposure to ionizing radiation does not only induce inflammation as a prominent innate immune response via proinflammatory cytokines such as TNF-α and via chemokines, but also induces selected steps towards activation of adaptive immunity.
In absence of RelA, only TNFRSF1B is upregulated after X-ray exposure. TNFRSF1B encodes for the TNF receptor TNF-R2. It was shown that TNFRSF1B receptor signaling resulted in the activation of anti-apoptotic survival proteins [74]. At the same time, TNFRSF1B promotes TNF-α induced apoptosis [75]. Baseline plasma TNFRSF1B, a marker of inflammation and a soluble TNF antagonist, was significantly associated with an increased risk of human colorectal cancer [76]. TNF-α bound to TNFRSF1B recruits intracellular adaptor proteins to activate multiple signal transduction pathways including NF-κB [77]. The activation of NF-κB in intestinal epithelial cells by TNF-α through TNFRSF1B receptor signaling has been previously linked to carcinogenesis. Silencing TNFRSF1B increased apoptosis [74]. However, NF-κB interferes with apoptotic signals at various levels. The best example is found in the TNF-R1 pathway [78]. It is therefore not certain whether increased apoptosis results from pro-apoptotic TNFRSF1A signaling. An upregulation of TNFRSF1B expression 8 h after X-ray exposure (0.1, 1 Gy) was also observed in a three-dimensional skin tissue model [71].
Overall, results revealed the outstanding role of NF-κB and its subunit RelA in cytokine expression after exposure to low-and high-LET radiation and therefore in the "sterile inflammation" after radiation exposure [57]. In contrast to other studies with cancer cells, no radiation-induced upregulation of antiapoptotic genes above the threefold threshold was found. BIRC3 encoding cIAP2 was upregulated 2.0 and 2.5 times after exposure to 4 and 8 Gy X-rays, respectively. In differentiated thyroid cancer (DTC) cells, incubation with (131)I-induced NF-κB activation, and knock-down of p65 by siRNA transfection increased (131)I-induced cell killing [79]. This effect was explained by lower expression of the antiapoptotic NF-κB target genes XIAP, cIAP1, and Bcl-xL when p65 expression was downregulated by siRNA [79]. In presence of RelA, IL-8 secretion was strongly induced by TNF-α and high X-ray doses. This cytokine is part of the CXC chemokine family and acts as proinflammatory mediator, further activating proinflammatory cytokines in a positive feedback loop. It attracts and activates neutrophils [80], causing them to migrate towards the site of infection. It was shown for several cell types, that the NF-κB binding site is essential for transcription of IL-8 [81,82]. Interleukin 8 secretion was also increased in response to oxidative stress, mediated by TNF-α, which further induces RelA-binding to the binding site of the IL-8 promoter [56]. In HTori-3 cells, a human thyroid epithelial cell line, exposed to either a non-toxic radiation dose (0.1 Gy) or a slightly toxic dose (0.2 Gy) from iron ions (1 GeV/n), IL-8 expression was altered [83]. In organotypic cultures of oral mucosa irradiated with X-rays or carbon ions, both radiation qualities increased the release of IL-8, with a saturation at 4 Gy for early time points, and higher early increases (4 h) after carbon ion exposure in the spread-out Bragg peak (SOBP, track-averaged LET~50 keV/µm) [84,85]. The IL-8 release remained increased for two days after carbon ion exposure (2, 4, 10 Gy) [84].
TNF-α secretion was increased after higher doses of X-rays (16 Gy), while its mRNA level was enhanced already at a dose of 4 Gy.

Implications for Countermeasure Development
Although caution should be taken to directly translate in vitro data to a clinical setting and the doses used here are quite high, the study suggests that due to NF-κB activation by heavy ion exposure astronauts would be prone to "sterile inflammation" in relevant target organs. Indeed, Parihar et al. found increased numbers of activated microglia in the perirhinal cortex of rodents exposed to low doses of helium and other space-relevant accelerated ions, indicating neuroinflammation [86,87].
For ensuring astronauts' health, countermeasures have to be developed which could rely not necessarily on therapeutically applied drugs but on natural inhibitors of the proteasome or other regulating factors in the NF-κB pathway as dietary supplements. The proteasome inhibitor MG132 which was used in this study is not such a candidate as it inhibits multiple proteases in addition to the proteasome. Furthermore, at least in tumor cells, treatment with MG-132 prior to irradiation lead to radiosensitization [88], which is an effect exploited for tumor therapy with the proteasome inhibitor bortezomib [89].
Possible candidates for protection against chronic inflammation could be members of the triterpenoid family, which inhibit the NF-κB pathway at various sites [90,91]. Additionally, in our earlier studies, capsaicin, an active component of the red pepper (Capsicum), partially inhibited NF-κB activation in HEK cells in a non-toxic concentration range [92]. However, NF-κB is a central element of various physiological functions and its inhibition might be associated with various side effects. Therefore, evaluation of its target genes that are upregulated by radiation exposure might lead to more specific toeholds to mitigate radiation-induced inflammation.
For high-dose exposure that could occur during a large solar particle event, activation of the NF-κB pathway might increase the survival of, e.g., bone marrow stem cells and intestinal epithelial cells and thereby ameliorate the acute radiation syndrome. For this purpose, Toll-like receptor 5 (TLR5) agonists were developed based on bacterial flagellin and successfully tested in animal models [93].

Conclusions
RelA is essential for cytokine and chemokine expression after exposure to both, low-and high-LET radiation and it thereby shapes the sterile inflammatory response, which comprises a smaller NF-κB target gene subset compared to TNF-α treatment. ATM and the proteasome are essential for NF-κB activation by different radiation qualities. NF-κB confers a survival advantage after X-ray, but not after heavy ion exposure.

Transformation and Preparation of Plasmid DNA
For plasmid propagation, 100 µL of competent E. coli DH5α-cells (Stratagene; Agilent Technologies, Karlsbrunn, Germany) were transformed with 2 µL of each SureSilencing™ shRNA Plasmid (~30-50 ng/µL) according to the manufacturer's protocol. The transformed bacteria were plated on selective LB-agar plates containing ampicillin (Biochrom GmbH, Berlin, Germany) with a final concentration of 50 µg/mL for selection and incubated at 37 • C over night. For plasmid preparation, single colonies were cultured for 16 h in 100 mL of LB-medium containing ampicillin under permanent shaking. For preparation of plasmid DNA the Plasmid Maxi Kit (Qiagen, Hilden, Germany) was used following the instructions of the manufacturer. For quality control, the optical density at 260 nm and 280 nm was determined in a spectrophotometer (Nanodrop 2000c, Thermo Scientific, Langenselbold, Germany). Only preparations with an OD260/OD280 of more than 1.8 were used for the transfection of human cells.

Cell Culture
The cell line HEK 293 was established by Graham et al. (1977) from human embryonic kidney cells immortalized by transfection with sheared fragments of adenovirus type 5 DNA [94]. The cells contain only the stably integrated region (E1a and E1b genes) of the human adenovirus genome and do not produce viral particles. Cells were obtained from the American Type Culture Collection (ATCC CRL-1573; now: LGC Standards, Wesel, Germany). Cloning of the plasmid pNF-κB-d2EGFP/Neo, stable transfection and selection of an appropriate cell clone (HEK-pNF-κB-d2EGFP/Neo L2) was already described [27]. HEK-pNF-κB-d2EGFP/Neo L2 and untransfected HEK cells have comparable radiosensitivity to X-rays and to accelerated carbon ions and show no difference in proliferation [19]. Cells were maintained in α-medium (Biochrom KG, Berlin, Germany) containing 0.6 mg/mL G418 and 10% fetal bovine serum (FBS) at standard conditions (37 • C, 95% air and 5% CO 2 atmosphere). Cells were passaged every week and seeded in a cell density of 3 × 10 4 cells/cm 2 in 80 cm 2 cell culture flasks (Nunc, Novodirect, Kehl, Germany). Medium was exchanged after four days. Cell culture vessels were coated with poly-D-lysine (10 µg/cm 2 , Sigma-Aldrich Chemie, Steinheim, Germany, in sterile deionized water) to allow a stronger binding of cells to the substrate. Therefore, the culture vessels were coated with poly-D-lysine for 15 min at room temperature and subsequently washed three times with sterile deionized water.

Stable Transfection
For knock-down of the NF-κB subunit RelA, HEK-pNF-κB-d2EGFP/Neo L2 were stably transfected with the SureSilencing™ shRNA plasmids RELA-1 to 4 and the control plasmid from SABiosciences (Frederick, MD, USA). Four of these plasmids encoded different targeting sequences of RelA mRNA, while the fifths plasmid was a negative control vector containing a scrambled artificial sequence which did not match any human, mouse or rat gene. HEK-pNF-κB-d2EGFP/Neo L2 cells were seeded in poly-D-lysine-coated 6-well-plates (Nunc). Transfection was conducted at a cell density of~50%. Therefore, 2 µg of DNA of each vector were mixed with the transfection reagent FuGENE 6 ® (Roche, Mannheim, Germany) and serum-free α-medium according to the protocol. After three days of incubation, cells were transferred into ∅ 6 cm culture dishes (Nunc) containing α-medium with 400 µg/mL hygromycin (US Biological, Salem, MA, USA). After approximately 10 days, cells were seeded for RNA isolation (Section 4.9.1) to measure the knock-down level of RelA in the mixed population by real-time quantitative Reverse Transcriptase Polymerase Chain Reaction (RT-qPCR). The strongest knock-down of RelA expression was achieved with the plasmids RELA-1 and RELA-4 ( Figure 4A). Therefore, clones grown from single cells of these two cell populations were grown by seeding 50 cells per microtiter plate containing hygromycin-medium. Cells were incubated for 2-3 weeks until colonies were grown. A single colony was washed with PBS, trypsinized and added to a 25 cm 2 cell culture flask (Nunc) containing 5 mL of hygromycin-medium. When cells had reached a density of 80-100%, their knock-down level was determined by seeding them in triplicates and proceeding according to 4.10. In clone 4-9, RelA mRNA expression was reduced by 83.1% and it was used for further experiments. The selected clone was named HEK shRNA RelA and is characterized by a RelA knock-down level of 83.1% and a resistance against hygromycin.

Growth Kinetics
To characterize differences in growth behavior between HEK-pNF-κB-d2EGFP/Neo L2 and HEK shRNA RelA, both cell lines were seeded with the same cell number (1 × 10 4 cells/cm 2 ) in poly-D-lysine-coated 6-well plates. The cell number of one well was determined daily until cells reached the stationary growth phase.

Treatment Modalities
For cytokine treatment or irradiation experiments, 3 × 10 4 cells/cm 2 were seeded into suitable culture vessels freshly coated with poly-D-lysine (10 µg/cm 2 , Sigma-Aldrich Chemie, Steinheim, Germany) for 15 min followed by washing with phosphate buffered saline (PBS). For exposure to TNF-α, X-rays and heavy ions, cells were seeded into 25 cm 2 flasks (Iwaki, Dunn Labortechnik GmbH, Asbach, Germany for GANIL, Nunc for GSI and HIMAC). Cells were incubated for two days before treatment.

Cytokine Treatment
Human recombinant TNF-α was obtained from Sigma-Aldrich Chemie. TNF-α (final concentration 10 ng/mL), was dissolved in PBS, and added in serum-containing medium. Cells were harvested for flow cytometry as described in 0 after different time points. Control dishes were solvent-treated.

X-Irradiation
X-irradiation was performed at DLR in Cologne using a Gulmay RS225 X-ray source (Xstrahl, Surrey, UK). Cells were irradiated at room temperature with 200 kV and 15 mA at a dose rate of 1 Gy/min. To eliminate soft X-rays, a copper filter (thickness 0.5 mm) was used. Cells were irradiated in petri dishes (Ø 3 or 6 cm). Doses up to 16 Gy were used, with an LET of 0.3-3 keV/µm. Control dishes were sham-irradiated. Doses were determined using the UNIDOS webline (PTW, Freiburg, Germany).

Heavy Ion Exposure
Irradiation with accelerated heavy ions was performed at the "Grand Accélérateur National d'Ions Lourds" (GANIL, Caen, France), at the GSI Helmholtzzentrum für Schwerionenforschung GmbH (GSI) in Darmstadt, Germany, or at the Heavy Ion Medical Accelerator in Chiba (HIMAC), Japan. Characteristics of the applied beams including the LET in water are shown in Table 5. Adherent cells were irradiated in sub-confluent stage and in upright position at room temperature. Therefore, flasks and petri dishes were completely filled with α-medium and closed. Dosimetry was performed by the staff at the accelerator facilities. Dose rates were adjusted to approximately 1 Gy/min. At GANIL, dosimetry was performed by the "Centre de recherches sur les Ions, les Matériaux et la Photonique" (CIMAP)/"Laboratoire d'Accueil en Radiobiologie avec les Ions Accélérés" (LARIA) staff, yielding the fluence in particles/cm 2 (P/cm 2 ) [95]. To convert fluence (F) to the energy dose, the following formula was applied [96]: After irradiation, serum-free α-medium was discarded and cells were further cultivated with fresh medium containing 10% FBS.

Cytotoxicity of Chemical Inhibitors
Cytotoxicity of chemical inhibitors KU-55933 and MG-132 was measured by the MTT assay. Reduction of MTT (3-(4,5-dimethylthiazole-2-yl)-2,5-diphenyltetrazolium bromide) to formazan is correlated to the activity of cellular enzymes and measured colorimetrically. HEK-pNF-κB-d2EGFP/Neo L2 cells were seeded in 96-well plates and incubated with each inhibitor for 20 and 72 hours. NF-κB was activated by adding TNF-α protected from light. 100% DMSO was added to one well to measure the background signal. After removing medium, cells were incubated for 1 h at 37 • C with serum-free α-medium containing 0.1 mg/mL MTT. Subsequently, cells were lysed with DMSO with 10% (w/v) sodium dodecyl sulfate (SDS) and 0.6% (v/v) acetic acid while shaking them for 20 min with 30 rpm. Finally, absorbance of formazan was measured with the fluorescence/luminescence microplate reader Lambda Fluoro 320 (MWG Biotech AG, Penzberg, Germany) at 562 nm. Cell survival was calculated as a ratio of formazan-absorbance of treated vs. untreated samples after subtraction of the background signal.

Colony Forming Ability
For investigating cellular survival after exposure to X-rays and heavy ions, cells were seeded and irradiated as already described. Immediately after irradiation adherent cells were trypsinized and re-plated in six petri dishes per dose. The cell density for seeding of irradiated cells was adjusted to compensate for the plating efficiency of the respective cell line and for the anticipated lethal effect of the treatment in order to allow growth of 40-60 colonies per dish.
After 21 days of incubation without medium change the resulting colonies were fixated and stained (1 mg/mL crystal violet in 3% formaldehyde solution, Sigma-Aldrich Chemie) and colonies containing more than 50 cells were counted. The experiments were performed with six dishes per dose and repeated if beam time was available.
All data from the irradiated samples were fitted by least-squares linear regression analysis to lnS = ln(PE D /PE D=0 ) versus dose, where the natural logarithm of S is the survival and PE are the plating efficiencies of treated (PE D ) or untreated cells (PE D=0 ). Differences in the survival curves were statistically verified by a two-sided Student's t-test of the regression lines. A p level of 0.05 was considered to reflect significance.

Flow Cytometry
For measurement of d2EGFP, cells were harvested at different time points after radiation or cytokine treatment by trypsinization and fixed with 4.5 mL cold 3.5% formaldehyde in PBS. For flow cytometric analysis, cells were centrifuged and resuspended in PBS. Forward and side scatter and green fluorescence (FL-1) of 20,000 cells from the samples were measured in a FACScan (Becton Dickinson, San Jose, CA, USA) with an argon laser (488 nm) as excitation source. The markers d2EGFP (−) and d2EGFP (+) were set by means of untreated and TNF-α treated cells. The percentage of d2EGFP (+) cells was used as measure of NF-κB-dependent d2EGFP expression.

RNA Isolation
Each sample was seeded in triplicates. At a density of~50% cells were lysed with RLT buffer containing 1% 14.3 mol/L β-mercaptoethanol and transferred into an RNase-free Eppendorf tube with an iced syringe. RNA was isolated using the RNeasy ® Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. As RNA was used for cDNA synthesis and subsequent RT-qPCR, the additional DNase treatment step was included to eliminate residual genomic DNA. RNA was eluted in 50 µL RNase-free H 2 O.
For gene expression analysis with the RT2 Profiler™ PCR Arrays (SABiosciences, Washington, DC, USA), further RNA quality control with RT2 RNA QC PCR array (SABiosciences) was performed.

cDNA Synthesis
Quality, integrity and quantity of isolated RNA were analyzed. Subsequently, RNA was transcribed into cDNA using the RT2 First Strand Kit (SABiosciences). A preliminary genomic DNA elimination step was conducted by incubation of 1 µg total RNA with gDNA Elimination Buffer, as recommended in the protocol. To detect possible remaining genomic DNA contamination in isolated RNA samples, one sample lacking reverse transcriptase was used as a negative control. The reaction mix (4 µL BC 3 (5X RT Buffer), 2 µL PC2 (Primer and External Control Mix 3), 2 µL RE3 (RT Enzyme Mix 3), 3 µL ddH 2 O) was incubated at 42 • C for 5 min. The RT reaction mix was then added to the Genomic DNA Elimination Mixture and incubated at 42 • C for 15 min. The reaction was stopped by heating at 95 • C for 5 min. cDNA concentration at this stage was 50 ng/µL. Finally, 91 µL of H 2 O were added to each 20 µL cDNA synthesis reaction to obtain a final cDNA concentration of 9 ng/µL. cDNA samples were stored at −20 • C.

Setting Up RT-qPCR Validation of Knock-Down-Level
The quantitative real-time PCR reaction mix for one PCR tube was composed referring to the RT2 qPCR Primer Assays protocol (SABiosciences) as follows: 12.5 µL RT2 SYBR Green qPCR Master Mix, 8.5 µL ddH 2 O, 3 µL Template cDNA (27 ng) and 1 µL 10 µmol/L PCR primer pair stock.
As a negative control DNA was replaced by H 2 O. The primer pairs SYBR ® Green Human RELA and SYBR ® Green Human HPRT 1 (RT2 qPCR Primer Assay, SABiosciences) were used for targeting the gene of interest (GOI) RelA and the housekeeping gene (HKG) HPRT (Table 6).   (Table A2) and one array focusing on genes targeted by NF-κB (human NF-κB target gene array, PAHS-225ZD, Table A1). An overview of all genes included in these arrays is listed in Appendix A. Overall, 162 genes, involved in cell cycle control, apoptosis, DNA damage, stress and toxicity and NF-κB signaling were investigated. Each well of these arrays contains primers for the respective gene.
Only cDNA samples, which had been approved by the RT2 RNA QC PCR Array (SA Biosciences) to prevent false positive signals by genomic DNA contamination, were used for further analysis. The RT-qPCR master mix for one 96-well plate was set up according to the RT2 RNA QC PCR Array protocol using 1350 µL RT2 SYBR Green qPCR Master Mix, 1248 µL ddH 2 O and 102 µL template cDNA (9 ng/µL).
Quantitative real-time analysis was performed in the Opticon2 thermocycler starting with an initial denaturation for 10 min at 95 • C, following by 40 cycles (denaturation for 15 s at 95 • C, annealing for 35 s at 55 • C, elongation for 30 s at 72 • C, plate read), a final incubation for 1 min at 95 • C and 2 min at 65 • C, and a melting curve from 65-95 • C with reads every 0.2 • C. The threshold cycle (Ct) was determined in the amplification plots. Results were analyzed according to the algorithm described in the RT2 RNA QC PCR Array protocol. It is based on the ∆∆Ct method: The relative expression of treated sample versus control was calculated by means of Equation (4): In a cautious data interpretation, the threshold for significant up-or downregulation was set to −3 or +3, respectively, to avoid misinterpretation of physiological fluctuation of mRNA levels.

Measurement of Cytokine Secretion
To measure the cytokine concentration in the cell culture supernatants (100 µL aliquots, frozen and stored at −80 • C after collection and once thawed, centrifuged for 5 min at a relative centrifugal field of 300× g) of HEK-pNF-κB-d2EGFP/Neo L2 and HEK shRNA RelA cells, the LUNARIS TM Human 11-Plex Cytokine Kit was used (LHCY-20110S, Ayoxxa Biosystems GmbH, Köln, Germany) which is based on beads-on-a-chip technology with a classical sandwich immunoassay principle and a fluorescence readout. The calibration curve was generated according to the kit handbook. Briefly, seven standards were prepared by serial dilution (1:4) of the Human Cytokine Standard using Assay Diluent 2. The standard contained the following analytes: GM-CSF, IFN-γ, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12(p70) and TNF-α. The LUNARIS TM BioChips were first washed and the 5 µL of standard, blank and samples which were diluted 1:2 in Assay Diluent 2 were loaded into the appropriate wells. After an incubation of at least 3 h at room temperature, the BioChips were washed and 10 µL of detection antibody solition was added and allowed to incubate for 60 min at RT. The BioChips were washed again and then 10 µL of Streptavidin-Phycoerythrin (SA-PE) reagent was added. After 30 min of incubation, the final washing step was performed and the BioChips were air-dried and imaged using the LUNARIS TM Reader (LRS-001). The quantification of the readout was performed using the LUNARIS TM Analysis Suite Software.

Statistics
Each experiment was repeated up to five times depending on the availability of beam time with one to six replicates each. In order to account for different numbers of replicates and repeats, the standard error was calculated. Means, standard errors and significance levels in the t-test were calculated with Microsoft ® Office Excel 2010 (Microsoft Deutschland GmbH, München, Germany). Regression analyses were performed using SigmaPlot 13.0. All authors have approved the submitted version and agree to be personally accountable for the author's own contributions and for ensuring that questions related to the accuracy or integrity of any part of the work, even ones in which the author was not personally involved, are appropriately investigated, resolved, and documented in the literature.