Targeted and Off-Target (Bystander and Abscopal) Effects of Radiation Therapy: Redox Mechanisms and Risk/Benefit Analysis

1. Direct effect

Through the direct effect, DNA molecules are directly ionized (loss of an electron), thus generating a DNA radical cation. For each nucleoside, the decomposition of the corresponding radical cation has been described in detail, but the chemistry is different in double-stranded DNA (dsDNA) (46). Indeed, among the DNA constituents, guanine has the lowest oxidation potential. Therefore, even if oxidation occurs on another base or sugar moiety, a fast electron transfer reaction occurs from guanine to the generated radical cation, thus repairing the initially produced radical and generating a guanine radical cation (G^•+).

Consequently, in dsDNA and in cells, the direct effect of radiation produces mostly unstable guanine radical cations that, after decomposition, give rise typically to two guanine chemical modifications (Fig. 2): 8-oxo-7′8-dihydro-2′-deoxyguanosine (8-oxodGuo) following oxidation and the corresponding formamidopyrimidine derivative FapydGuo on reduction. Interestingly, it has been shown that in irradiated cells, FapydGuo production is two times higher than that of 8-oxodGuo, suggesting that cellular DNA is in a reducing environment.

OPEN IN VIEWER

FIG. 2.

Mechanisms of 8-oxodGuo and FapydGuo formation through direct or indirect effects of irradiation. The direct effect produces a guanine radical cation that, following dehydration, produces a neutral radical 1 that can also be formed by addition of HO^• (produced through the indirect effect) onto the guanine moiety. Oxidation of 1 gives rise to 8-oxodGuo, whereas reduction of 1 leads to FapydGuo production. 8-oxodGuo, 8-oxo-7′8-dihydro-2′-deoxyguanosine; dGuo, 2′-deoxyguanosine. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

2. Indirect effect

The radiation-induced DNA lesions produced through the indirect effect are mediated by the initial formation of ROS due to water radiolysis. Exposure of water to ionizing radiation rapidly leads to the generation of HO^•, ionized water (H₂O⁺), hydrogen radicals, and hydrated electrons. Then, the reaction of the initially produced radicals generates hydrogen peroxide (H₂O₂) and superoxide anion (O₂ ^−•). As these ROS are also produced endogenously and their concentration is regulated by several antioxidant defense mechanisms, the radiation-induced increase of ROS cellular level could damage cellular constituents and also induce oxidative stress (113).

Among the ROS produced upon exposure to radiation, HO^• plays a predominant role because it can react, at a diffusion-controlled rate, with almost all biological constituents. Its reactivity within DNA is well documented (43), and about 30% of HO^• reacts with the sugar moiety or phosphate group of DNA. Such reaction (69) produces SSBs. HO^• reaction with the four different DNA bases has been studied in detail. This reaction produces a plethora of DNA lesions, but is limited by HO^• diffusion. Interestingly, the DNA lesions generated by such mechanism are not different from those produced via the direct effect, because similar radical intermediates are generated by the two mechanisms. For instance, concerning 8-oxodGuo, addition of HO^• to the position C8 of guanine produces radical 1 (Fig. 2), also generated following hydration of guanine radical cations arising from direct oxidation of DNA. Similar observations have been made for all DNA bases. One-electron oxidation of DNA bases gives rise to similar chemical modifications as those produced by the indirect effect of irradiation, the only difference concerns their relative yields (42).

The relative importance of the direct and indirect effects in the formation of radiation-induced DNA lesions is still a matter of debate. The general idea is that for low LET (linear energy transfer) radiation such as X- or γ-rays, the direct effect accounts for about 30% of DNA lesions (and thus 70% is attributed to the indirect effect), and that this proportion increases for higher LET radiations such as protons, carbons, and α-particles. A low mean LET of 0.2 keV/μm is observed with γ/X-rays and beta radiation, while LET increases up to 4–26 keV/μm with Auger electrons and up to 50–230 keV/μm with alpha particles. However, according to the chemical mechanism of formation of radiation-induced DNA lesions, the direct effect should mostly produce lesions on the guanine moiety (8-oxodGuo, e.g.) as explained above, due to fast electron transfer in DNA. Conversely, the indirect effect should generate lesions in all DNA bases and also strand breaks. Thus, an attempt has been made to determine the relative formation of 8-oxodGuo compared with other oxidative bases in cells after exposure to radiations with different LETs (78). To mimic the direct effect, a laser irradiation at 266 nm was used because it can produce direct DNA ionization following two-photon absorption. As expected, in these conditions, 8-oxodGuo was predominantly formed. This indicates that an efficient electron transfer reaction from guanine to other base radical cations (that are produced by two-photon ionization) takes place also within irradiated cells (vide supra). Conversely, the predominance of 8-oxodGuo formation relative to other oxidative DNA lesions in cellular DNA was not observed after gamma or high-LET radiation. Indeed, it has been shown that increasing the particle's LET does not increase the relative formation of 8-oxodGuo compared with other DNA lesions.

This suggests that the direct effect is not increased upon exposure to high LET radiation. Moreover, the strong correlation between HO^• radiolytic yield and the number of produced DNA lesions strongly suggests that DNA lesions are predominantly produced via the indirect effect (242). Additional work is needed to clarify this point. Particularly, recent studies have highlighted the fact that the guanine radical cation decomposition is affected by its environment. Therefore, the measurement of a specific product generated via the direct effect of radiation would be much more informative than measuring 8-oxodGuo formation, because the latter can be produced through the two mechanisms. Indeed, it has been shown that some amino acids chemically repair G^•+ (192, 193), and that addition of other amino acids as well as polyamines to the C8 position produces other DNA lesions (235, 285, 338), including DNA-protein crosslink. Thus, monitoring lesions specifically produced by the direct effect of radiation could provide additional information to better estimate the direct effect relative contribution.

Finally, it is interesting to note that the chemical nature of radiation-induced DNA lesions is not different from that of DNA lesions produced in cells subject to endogenous oxidative stress (45). This is not surprising because ROS produced by radiation through water radiolysis are similar to those produced endogenously, for example, HO^• through the Fenton reaction (204). However, the major difference concerns the lesion three-dimensional (3D) localization (212, 215, 330). Endogenous oxidative stress produces oxidative lesions that should be randomly distributed on the DNA macromolecules. These lesions, including modified DNA bases and SSBs, can be rapidly and efficiently repaired by the cell machinery, simply by removing the modification and using the complementary strand to resynthesize the original DNA sequence. After exposure to ionizing radiations, ROS are locally produced along the particle track and this could generate several DNA modifications at the same site (called “multiple damage sites,” MDS, or clustered DNA lesions) (Fig. 3). These lesions are defined as two or more modifications per helix turn (172, 219). DNA DSBs are one of the best examples of MDS. DSBs are produced in cells even after exposure to very low doses of radiations, and their formation yield increases almost linearly with the dose (102, 260, 329).

OPEN IN VIEWER

FIG. 3.

Radiation-induced DNA damage. Different types of DNA lesions that could be produced by ionizing radiations, including (A) base (B) modifications, abasic sites, SSB or DSB, intrastrand crosslink, tandem DNA lesions involving two adjacent modifications. (B) Types of DSB and non-DSB clustered DNA lesions involving the combination of all possible DNA lesions in one or two DNA helix turns (82). DSB, double-strand break; SSB, single-strand break. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

In addition, high-LET radiations, which induce more dense ionizations, produce, relatively to SSBs, more DSBs compared with low-LET radiations. In fact, when the LET of the radiation increases, the number of DNA lesions per unit dose (Gy) decreases (242) (due to an increased probability of radical recombination), but at the same time, MDS complexity increases (i.e., the MDS includes more DNA damage types, such as base modifications and strand breaks). MDS repair by the cell machinery is much more difficult as soon as the two DNA strands are modified (262). Thus, this explains why the harmful effect of radiation increases with higher LET, and also why the effect of radiation is mostly attributed to cluster DNA lesions (including DSBs and non-DSB cluster lesions) produced by several ionizations.

However, recent works have highlighted the fact that a single oxidation event also can produce several modifications (called tandem lesions), composed of two adjacent modifications. For example, as illustrated in Figure 4 for a dG-dC sequence, in the absence of oxygen, crosslinks between two adjacent DNA bases could be produced, generating intrastrand crosslinks (G[8-5]C adduct) (128). Within the same sequence, in the presence of oxygen, the reaction of the initially produced radical with O₂ produces a hydroperoxide radical that can add to the C8 position of an adjacent guanine (or adenine) base. The resulting unstable endoperoxide undergoes decomposition, producing tandem DNA lesions that involve two adjacent DNA modifications (253), such as 8-oxodGuo and a formylamine residue (8-oxodGuo-dF, Fig. 4) (29). The latter observation, although obtained using isolated DNA, indicates that a radical produced on a DNA base can react with surrounding molecules and particularly with adjacent bases, giving rise to possibly two adjacent modifications.

OPEN IN VIEWER

FIG. 4.

Mechanisms of formation of complex DNA lesions (including tandem 8-oxodGuo-dF lesions and G[8-5]C intrastrand crosslinks) at dG-dC sequences mediated by a single oxidation event. After HO^• reaction on the cytosine base, in the absence of oxygen, the produced radical can react with an adjacent guanine base, thus producing a G[8-6]C intrastrand crosslink. On the contrary, in the presence of oxygen, the cytosine radical is trapped by molecular oxygen, thus producing a peroxyl radical. This peroxyl radical can react with an adjacent purine base, thus generating an unstable endoperoxide that, on decomposition, gives rise to tandem lesions constituted of two adjacent oxidative DNA lesions. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Moreover, it has been shown that the repair of 8-oxodGuo involved in tandem lesions is significantly reduced compared with the repair of a single lesion (19). At the cellular level, the decomposition reactions of an initially produced radical will strongly depend on its environment. The fact that carbon-centered radicals can react efficiently with molecular oxygen provides a partial explanation for the well-known oxygen enhancement effect, observed in radiation biology. As illustrated in Figure 4, the DNA lesions generated in the absence of oxygen could be different from those created in oxic conditions. Low oxygen concentration could indirectly increase the lifetime of the initially produced radicals, thus favoring the possibility of radical recombination and ultimately decreasing the amount of radiation-induced DNA lesions (250). In addition, the increased lifetime of radicals could also promote repair, for example, by electron donation from mild reducing amino acids (193) or antioxidant molecules that could be used as radioprotectors (192).

This could provide a possible explanation for the reduced level of lesions produced under hypoxic or anoxic conditions (154). Nevertheless, increasing the lifetime of a radical could also favor its reaction with other cellular constituents, leading to the generation of more complex DNA lesions that could compromise DNA repair (154). Indeed, the presence of polyamines, which are highly concentrated in the nucleus of eukaryotic cells, or of a nucleophilic amino acid could lead, respectively, to the formation of DNA adducts (285) and DNA-protein crosslinks (206). These DNA damage types are repaired less efficiently than single lesions (19). These examples illustrate the complexity of the chemical reactions that can occur in cellular DNA following exposure of cells to ionizing radiation (47) and that can vary in oxic and hypoxic conditions.

Although the literature on damage to mitochondrial DNA (mtDNA) is less abundant, mtDNA also is affected by irradiation (358), and additional work is needed to better understand the consequences of this damage (143). This is also true for the nucleotide pool (110) and RNA that are more sensitive to oxidative stress (126) than DNA, which is compacted and protected within the nucleus.

1. Intercellular communications between irradiated and nonirradiated cells

Bystander effects involve signaling from irradiated cells to nonirradiated cells. The nature of the communications between irradiated and nonirradiated cells depends on whether cells are in contact or not. This cross talk can be mediated by paracrine secretion in the extracellular space of soluble factors via hemichannels formed by the transmembrane protein connexins (Cxs) and pannexins (Panxs) (182, 327). Cxs and Panxs are members of the tetraspan family in which proteins are classified according to their molecular weight (25–62 kDa for Cxs and 1–3 kDa for Panxs). These proteins allow the passage of ions (Ca²⁺, Na⁺) and of low-molecular-weight molecules (nicotinamide adenine dinucleotide, ATP, glutamate, GSH, prostaglandin E2 [PGE2] and inositol trisphosphate [IP₃]) between the intra- and extracellular environment, thus controlling autocrine and paracrine signaling (68, 327). The role in bystander effects of Cx43, the most abundant Cx in mammals, was identified very early and has been largely investigated (9).

Cxs and Panxs can also form full channels that are called gap junctions and mediate gap junction intercellular communication (GJIC). These intercellular channels show much higher selectivity and allow direct diffusion of factors between the cytoplasm of adjacent cells. The factors involved in bystander effects and transiting through intercellular gap junctions are small molecules (<1.5 kDa) and include ROS (9, 10), RNS (187), ions (Ca²⁺, K⁺, Na⁺), (178, 279), lipid peroxides (83), ATP (221), cyclic adenosine monophosphate (cAMP), glucose, glutamate, and GSH. As hemichannels are less selective than full gap junctions, it has been suggested that in normal conditions only GJIC plays a role. Conversely, in conditions of oxidative stress (e.g., on irradiation), both channel types might be involved (224, 279).

2. Reactive oxygen and nitrogen species initiate and propagate bystander effects

Early experiments highlighted the multiple roles of ROS and RNS in bystander effects. Specifically, they showed that radical scavengers (ascorbic acid, N-acetyl l-cysteine), NOS inhibitors (NG-nitro-l-arginine methyl ester [L-NAME]), NO scavenger (2-4-carboxyphenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide [c-PTIO] (281), antioxidant enzymes (superoxide dismutase [SOD], catalase) (10, 303), and DNA-binding antioxidants (methylproamine) (39) inhibit the appearance of bystander effects. For example, genomic instability due to bystander effects can be rescued by restoring mitochondrial function through overexpression of manganese-dependent superoxide dismutase (MnSOD) (103).

Oxidative processes that involve ROS and RNS continuously occur in cells because at low concentrations (0.01–0.001 nM for O₂ ^−• and 1–100 nM for H₂O₂), these molecules contribute to essential cellular functions (Fig. 8). At higher concentrations, they could be involved in the defense mechanisms against pathogens and they cause endogenous oxidative damage (22). It is thought that between ten and fifty DNA DSBs are produced per cell each day due to natural ionizing radiation, ROS, DNA replication errors, and unintended cleavage by nuclear enzymes (52, 262). In the cell, the concentrations of free radicals (O₂ ^−•, HO^•, NO^•) and nonradical reactive species (H₂O₂, ONO₂) are controlled by the balance between their production and clearance rates that are regulated by antioxidant compounds and enzymes, such as SOD, glutathione peroxidase (GPx), and catalase, as well as nonenzymatic compounds, such as α-tocopherol (vitamin E), β-carotene, ascorbate (vitamin C), and the cellular thiol/disulfide systems that include GSH/GSSH, thioredoxin (TRX1 [-SH2/-SS-]), and cysteine/cystine. Antioxidants can compete for oxidation with oxidizable substrates (at low concentration) and ultimately delay or inhibit the oxidation of these substrates (114). It must be noted that free amino acids, peptides, and proteins, some of which are present at high concentrations in cells (cysteine, tryptophan, histidine, tyrosine), also act as ROS scavengers (80). Reactions between ROS and redox active amino acid residues (e.g., cysteine) can modulate the activity of transcription factors (AP-1, NF-κB, and hypoxia-inducible factor 1 [HIF-1]) and enzymes (e.g., protein tyrosine phosphatases, acid sphingomyelinase [ASMase]) (168).

OPEN IN VIEWER

FIG. 8.

Endogenous sources of ROS and enzymatic antioxidant defenses. (A) O₂ ^−•, hydroxyl radical (HO^•), and H₂O₂ are produced by endogenous sources that reduce O₂. The main sources are mitochondria through ATP production and oxidative (oxygenase, dehydrogenase, oxidase) enzymes, such as NAD(P)H, and xanthine oxidase, lipoxygenase, myeloperoxidase. Oxygenase enzymes (e.g., lipoxygenase) oxidize substrates by transferring one electron, while oxidizing a cofactor [e.g., NAD(P)H] in the presence of oxygen. Dehydrogenases use organic substrates as an electron acceptor (e.g., quinones, NAD⁺). Oxidases just use O₂ as an electron acceptor. The yield of radiation-induced DNA lesions (/Gy/cell) is rather low compared with that produced by endogenous stress [averaged values from Goodhead (102), Pouget et al. (241, 245), Sage and Shikazono (262), and Ward (329)]. (B) Superoxide can be dismuted into H₂O₂ by the action of superoxide dismutase enzymes that possess a metal transition ion (Mn³⁺, Cu²⁺, Fe³⁺, or Ni³⁺) to catalyze the reaction. In the presence of M⁽ⁿ⁾⁺ metal ions, the resulting H₂O₂ can be broken down into HO^•+OH⁻ and M⁽ⁿ⁺¹⁾, according to the Fenton reaction. The latter reaction can also be mediated by catalase and GPx (22, 76, 80). GPx, glutathione peroxidase; NOS, nitric oxide synthase; SOD, superoxide dismutase. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

O₂ ^−• is one of the main endogenous cellular ROS compounds (Fig. 8). It is produced by reduction of ground-state molecular oxygen (³O₂) by enzymatic (nicotine adenine dinucleotide phosphate [NAD(P)H], xanthine oxidase) or nonenzymatic (semi-ubiquinone Q⁻, a compound of the mitochondrial electron transport chain [ETC]) systems (80).

Peroxisomes also contain enzymes, namely d-amino-acid-oxidase that contains flavin adenine dinucleotide (FAD) as cofactor and urate-oxidase oxidizing substrate by transferring H to O₂, leading to the formation of H₂O₂.

O₂ ^−•concentration is also regulated by three SODs, namely the cytoplasmic Cu–Zn-dependent superoxide dismutase (CuZnSOD or SOD1), the mitochondrial MnSOD (or SOD2), and the extracellular SOD (ECSOD or SOD3), that catalyze O₂ ^−•dismutation into the less reactive H₂O₂. This reaction requires reduced transition metals, such as iron or copper ions, and leads to the formation of the precursors of the highly damaging hydroxyl radicals (HO^•), according to the Fenton reaction (Fig. 8). In cells and in mitochondria, H₂O₂ can be decomposed into H₂O and O₂ by GPx, catalase, or peroxiredoxins (Fig. 8).

The initial production of ROS by water radiolysis is limited in time, space, and quantity. If one considers that 2000 ionization events are produced per cell per Gy (for low LET radiation), clinical doses of 2 Gy produce much less ROS than the amount generated during standard cell metabolism (329). For instance, a dose of 1 Gy would produce 1 × 10³ DNA breaks per cell (about 1000 SSBs and 40 DSBs) (102, 241, 245), while endogenous metabolism would produce more than 10 × 10³ DNA breaks per day (262, 318).

Therefore, radiotherapy efficacy in tumor cells can be partly explained by the amplification and prolongation in time (hours and weeks) of the initial radiation-induced ROS production. This would involve ROS- and RNS-mediated activation of signaling pathways and signal transmission to neighboring cells. The main sources of endogenous ROS that participate in this amplification process include mitochondria, NAD(P)H oxidases, and other oxidases, such as xanthine oxidase, lipoxygenases, and peroxisomes (80).

a. Mitochondria-dependent ROS production

Human-hamster hybrid AL cells with normal (ρ⁺) or depleted (ρ⁰) mtDNA were used to demonstrate the contribution of mitochondrial functions in the generation of bystander signals from irradiated cells (53, 127, 355).

(1) Mitochondria and ROS endogenous production

Mitochondria produce energy through the oxidative metabolism of carbohydrates, lipids, and amino acids that generate NADH via the ETC. They are also in charge of different metabolic processes, such as the synthesis of heme, nucleotides, lipids, and amino acids. They mediate the intracellular homeostasis of inorganic ions. The ETC is composed of four multisubunit enzyme complexes (Fig. 9): complex I with NADH coenzyme Q reductase activity, complex II with succinate dehydrogenase coenzyme Q activity, complex III with coenzyme Q cytochrome C reductase activity, and complex IV with cytochrome C oxidase activity. Coenzyme Q (or ubiquinone) allows the electron transfer from complex I to III, while cytochrome C is involved in the transfer between complex III and IV. Electrons are finally transferred to O₂, leading to water formation. Intermediate products, such as O₂ ^−• and H₂O₂, may also be liberated, particularly at complex I (NADH dehydrogenase) (107) and III (310), during or after premature leakage of electrons between ubiquinone and cytochrome C, and at complex II, to a lower extent (151, 348).

OPEN IN VIEWER

FIG. 9.

Mitochondrial electron transport chain and NAD(P)H oxidase as endogenous source of ROS. (A) NADH and succinate produced during the citric acid cycle are used as electron donors for ATP synthesis by the ETC. ETC consists of complex I (NADH coenzyme Q reductase), complex II (succinate dehydrogenase coenzyme Q), complex III (coenzyme Q cytochrome C reductase), and complex IV (cytochrome C oxidase). The final acceptor molecule O₂ is reduced to H₂O. However, a small percentage of electrons can leak at complex I and complex III and can reduce O₂ into O₂ ^−•. (B) The second major source of ROS is NOX. NOX contains membrane proteins (gp91^phox or NOX-2 and p22^phox that constitute the flavocytochrome b558, and the small G Rap1A protein). During NOX activation (for instance, in neutrophils, or by Ca²⁺ or radiation), cytosolic proteins (p40^phox, p47^phox, p67^phox and G Rac2) are recruited to the membrane and NADPH binds to NOX and transfer electrons to FAD and further across the membrane to O₂ (108, 208). ETC, electron transport chain; FAD, flavin adenine dinucleotide. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

(2) Mitochondria and irradiation

Mitochondria are essential in the intrinsic apoptosis pathway (but not in the death receptor-mediated extrinsic pathway). Specifically, mitochondrial membrane permeabilization and cytochrome C release are followed by caspase activation. The intrinsic apoptosis pathway involves PUMA, a proapoptotic protein that allows BAX and/or BAK translocation to mitochondria to signal apoptosis.

The total cellular mitochondrial volume is high (4–25% of the cell volume, depending on the cell type and state), and therefore, mitochondria represent a fairly substantial target for cytoplasmic irradiation. Radiation produces many different types of mitochondrial damage, including mtDNA damage. mtDNA is a circular double-stranded genome that encodes proteins, transfer RNA, and ribosomal RNA. It lacks protective histones and shows limited repair capability. Therefore, on radiation exposure, mtDNA could be preferentially damaged or lost (217), leading to loss of mitochondrial function (149, 316). Moreover, it has been shown that microbeam irradiation with carbon ions leads to depolarization of mitochondria (323). Exposure to direct gamma radiation causes mitochondrial mass increases (218), and alpha-particle microbeam irradiation leads to mitochondrial fragmentation, involving dynamin-related protein 1 (DRP1), a member of the dynamin family involved in mitochondrial fusion and fission (348).

Cytoplasmic irradiation also has been associated with altered protein synthesis and irregular mitochondrial oxidative phosphorylation (66, 217, 348), leading to persistent oxidative stress (162, 345). Yoshida et al. detected dysfunction of complex I at 12 h postgamma irradiation, after early and transient production of intracellular ROS (in the first minutes). This was followed by increased mitochondrial ROS levels and mtDNA damage at 24 h (345). These observations suggest that mitochondria participate in the amplification of the initial oxidative stress through transmission of oxidative events from one mitochondrion to the entire mitochondrial population of a given cell. This process might involve transient Ca²⁺-dependent mitochondrial permeability to adjacent mitochondria, resulting in enhanced ROS/RNS generation (53, 162).

Chen et al. showed that in mtDNA-depleted and wild-type AL cells treated with mitochondrial respiratory chain function inhibitors, γH2AX induction is attenuated in bystander cells (53). Incubation of HepG2 (liver cancer) cells with cyclosporin A, an inhibitor of cytochrome C release, before irradiation demonstrated that cytochrome C release regulates bystander effect-induced formation of micronuclei (MN) and production of NO, but not of ROS (117). This indicates that cytochrome C has an effect on inducible NOS (iNOS) that catalyzes NO production.

Zhou et al. showed that on alpha-particle microbeam irradiation, the frequency of DNA mutations is decreased in bystander cells in cultures of skin fibroblasts with reduced (ρ⁰) mitochondrial functions compared with wild-type (ρ⁺) cells. Moreover, incubation with a pharmacologic inhibitor of NF-κB activation (Bay 11-7082) or with the NO scavenger c-PTIO reduced the DNA mutation frequency in both mutant (ρ⁰) and wild-type (ρ⁺) cells, highlighting the role of NF-κB and of activation of its downstream NF-κB/iNOS and NF-κB/COX-2 signaling pathways (355).

Finally, mitochondria are affected also in nonirradiated cells grown in culture medium from irradiated cells. Specifically, changes of mitochondrial distribution, loss of mitochondrial membrane potential, and mitochondrial mass (218), increased ROS and RNS production, and increased apoptosis rates have been observed in such cells. These effects can be blocked by antioxidant treatments (180).

3. Cell membrane response to radiation

Compared with DNA-centered approaches, relatively few radiobiological studies have investigated the role of radiation targeted at the cell membrane that, for long time, was considered to be just an inactive phospholipid bilayer. However, in the 1990s, several studies showed the production of ceramide, a cellular second messenger of apoptosis involved in the sphingomyelin signaling pathway, during membrane irradiation (111). Using lymphoblasts from patients with Niemann–Pick disease (deficiency in ASMase activity), it was demonstrated that this enzyme is required for radiation-induced production of ceramide and apoptosis (263). Many studies now strongly support the ceramide-mediated cell membrane role in the biological effects of radiation, including cell death. Moreover, radiation resistance in Burkitt's lymphoma cells has been associated with defective ceramide signaling (189). Finally, structure-modulating agents (e.g., cholesterol) and antioxidants (e.g., tocopherol, eugenol) can modify the membrane response to stress (194, 230).

a. Cell membrane and lipid rafts

Cell membrane contains lipids (sphingolipids and glycerophospholipids), proteins, and sterols (Fig. 5B).

Lipids, such as PUFA, are susceptible to free radical-initiated oxidation (49), leading to lipid hydroperoxide formation that can then be reduced by peroxidases. If not efficiently reduced, lipid hydroperoxides can be degraded into hydroxyalkenals (such as 4-HNE) that show great reactivity toward DNA, proteins, and lipids (49).

Cholesterol and sphingolipids, among which sphingomyelin is the prevalent, mainly localize in the outer leaflet of the cell membrane. They play a crucial role in signal transduction and participate in cell growth, senescence, differentiation, and apoptosis (122). Sphingolipids contain a long-chain sphingoid base (such as sphingosine) linked via an amide to long-chain fatty acids and to one polar head group, making them amphipathic molecules (Fig. 5B). Head groups are different in sphingolipids (phosphorylcholine), ceramides (hydroxyl group), and glycerophospholipids (carbohydrates) (256). Ceramide is composed of D-erythro-sphingosine and of a fatty acid that contains 2–28 carbon atoms in the acyl chain. Similarly, a hydrophobic moiety forms the backbone of sphingomyelin and of other complex sphingolipids, such as cerebrosides and gangliosides.

The concept of the cell membrane as a fluid mosaic (289) is based on the finding that most physiological phospholipids exhibit low melting temperatures and, therefore, most likely exist in a liquid disordered phase. For instance, in Escherichia coli, the fatty acid composition of phospholipids depends on the temperature. Specifically, the proportion of unsaturated acids increases as the temperature decreases (67). At physiological temperature (37°C), 22% of the phospholipid molecules have two unsaturated acyl chain molecules, whereas this proportion is two times higher at 17°C (67).

However, the notion of fluid mosaic has been reconsidered because mammalian membranes contain very small domains that are in a liquid ordered phase (35, 287). Indeed, sphingolipids, which have a much higher melting temperature than other phospholipids in the cell membrane, interact with each other via hydrophilic interactions between the sphingolipid head groups. These complexes are stabilized by cholesterol that fills the gaps between the large sphingolipid molecules. The resulting domains are resistant to cold detergent extraction (2) or mechanical disruption and are called lipid rafts because they seem to float in the membrane. With a size of about 50 nm, they correspond to lateral subcompartments in the cell membrane and allow the cell membrane to exert its cellular functions by segregation of molecules, reorganization of receptor molecules and of membrane signaling and trafficking.

Lipid rafts can be converted into larger membrane platforms by ASMase activity that hydrolyzes sphingomyelin to ceramide in rafts (Fig. 5B). Ceramide molecules then spontaneously associate to form ceramide-enriched microdomains that fuse into large ceramide-enriched membrane platforms, thus altering the biophysical properties of these membrane domains (167) (Fig. 10). ASMase is activated by multiple stimuli, including CD95, CD40, DR5/TNF-related apoptosis-inducing ligand (TRAIL), CD20, FcgRII, CD5, LFA-1, CD28, cytokines, and chemotherapeutic drugs (doxorubicin, cisplatin), and also by ionizing radiation. Ceramide is considered the second messenger of many factors. For instance, ceramide acts as a messenger of inflammatory cytokines by binding and exerting a dual effect on the cytosolic phospholipase A2 (cPLA2) involved in eicosanoid formation (132). Ceramide is also involved in apoptosis. Its binding to the endosomal acidic aspartate protease cathepsin D results in the autocatalysis of the 52 kDa prepro-cathepsin D into the enzymatically active 48/32 kDa cathepsin D isoforms that mediate oxidative stress-induced apoptosis (120).

OPEN IN VIEWER

FIG. 10.

Lipid rafts: ceramide-enriched large platforms. (A) The cell membrane consists of a lipid bilayer that includes proteins and cholesterol. According to the fluid-mosaic model, lipids can rotate laterally and between bilayers. The lipid distribution in the cell membrane is associated with specific cellular functions. Specifically, the cell membrane can include glycolipids, phospholipids (e.g., phosphatidylcholine, phosphatidylserine, phosphatidylethanolamine), and sphingolipids, among which SM is predominant. Their resistance to disruption allowed the identification of membrane domains enriched in ceramide. Ceramide is produced via ASMase-mediated hydrolysis of SM following RNS and ROS activation and ASMase translocation to the outer layer of the cell membrane. Interaction between the resulting lipids and proteins leads to the coalescence of microdomains (lipid rafts) into ceramide-enriched large platforms. (B) These platforms can promote clustering of receptors and activation of signaling pathways. Besides its role in lipid rafts, ceramide is also a second messenger (58, 167, 346). CER, ceramide; GPL, glycophospholipids; TM, transmembrane. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

It has been proposed that oxidation of cysteine 629 in ASMase C-terminus by HO^• could be responsible for a possible mechanism leading to increased ASMase enzymatic activity (251). Moreover, the RNS ONOO⁻ specifically activates ASMase (48).

Ceramide is generated by ASMase in the outer leaflet of the cell membrane or within intracellular vesicles. It might also flip to the cytoplasmic leaflet and then interact with intracellular molecules. Although the mechanisms of ceramide translocation from one membrane leaflet to another are unknown, ceramide can interact and activate serine/threonine phosphatases, namely PP2A and PP1 (50, 58) (Fig. 10). Once activated, these phosphatases act on different signaling proteins, including MAPKs (AKT, c-JUN), PKC isoforms (PKCα and ζ), kinase suppressor of Ras (KSR), pRB, and BCL-2 (28, 349).

d. Ion channels and lipid rafts

Ceramide-enriched membrane platforms are involved in the regulation of potassium (27) and calcium (56) channels. Calcium ion cytoplasmic level is critical for many cellular functions via interaction with various signaling cascades, such as those that involve the activation of the calcium-dependent PKC (112) or Ca²⁺/calmodulin kinase (81).

Intracellular Ca²⁺ can activate NOX and iNOS, thereby promoting ROS and RNS formation, respectively. It can also modulate the activity of the transcription factors NF-κB and AP1 that are involved in COX-2 and iNOS transactivation and the subsequent release of ROS, RNS, and cytokines. Moreover, it activates kinases, including PI3K/AKT and MAPK.

In response to irradiation, Ca²⁺ levels increase (112, 319, 352). These changes can be described as oscillations or single transient changes within minutes to days after irradiation. The concentration of free Ca²⁺ in the cytoplasm is low (∼100 nM) in normal physiological conditions. It is controlled by modulating its cellular entry from the extracellular environment through, for example, ROS-mediated activation of plasma membrane Ca²⁺ channels, or by release from internal storages where it is present at higher concentrations (up to mM). ER is the main Ca²⁺ store (Fig. 11). As shown by experiments using the PLC gamma inhibitor U73122, Ca²⁺ release is regulated by PLC-mediated generation of the second messenger IP₃ that binds to the IP₃ receptors located at the ER membrane (20, 305) (Fig. 11). This process is mediated by activation of plasma membrane-associated receptor tyrosine kinases, such as ERBB3 and HER1. HO^• and exogenous NO induce the release of calcium from intracellular IP₃ receptor-sensitive stores and they modulate the open/closed status of Cx channels. Moreover, ROS can activate PLC (40, 229). An increase in IP₃ receptor level was observed after a dose of 3 Gy in lymphoblastoid cells (343). Moreover, IP₃ receptor phosphorylation by AKT is associated with reduced Ca²⁺ flux from the ER to mitochondria and subsequently with reduced apoptosis (299).

OPEN IN VIEWER

FIG. 11.

Interplay between Ca²⁺ and radiation-induced oxidative stress [extensively reviewed in Decrock et al. (68)]. Irradiation increases the intracellular Ca²⁺ level (oscillations or single transient changes occurring within minutes to days after irradiation). Radiation-induced ATP release by irradiated cells can activate ATP-gated P2X receptor cation channels (P2X receptors) present on the cell membrane, thus allowing Ca²⁺ entry into the cell. It can also activate P2Y receptors that have been identified as phospholipase C activators. Ca²⁺ can also be released from the ER through calcium-induced calcium release mechanisms that involve IP₃Rs or RyRs. IP₃ is produced (with diacylglycerol) during hydrolysis of phosphatidylinositol 4,5-bisphosphate (PIP2) by phospholipase C. Phospholipase C is activated by Ca²⁺, GPCRs, ROS, RNS, receptor and nonreceptor tyrosine kinase (e.g., HER1 and 3). Ca²⁺ can activate ion channels and binds to calmodulin before activation of the serine/threonine protein phosphatase calcineurin. Ca²⁺ can also activate protein kinase C that in turn activates, by phosphorylation, the MAPK pathway and phospholipase A2, at the origin of COX-2 activity modulation. It can activate transcription factors (NF-κB, AP1) that promote various downstream pathways (iNOS, COX-2). Released Ca²⁺ can also be taken up by mitochondria via VDAC and MCU and modulation by the ER-mitochondria-tethering proteins GRP75 (mitochondrial heat shot protein HSP70) and MFN 1 and 2 (involved in mitochondrial fusion). The increase in mitochondrial Ca²⁺ level is accompanied by ROS, an RNS increase, mtDNA damage, altered ATP synthesis, mitochondrial depolarization, and release of cytochrome C and caspase 3 that will amplify IP₃R activity. COX-2, cyclooxygenase-2; GPCRs, G-protein-coupled receptors; iNOS, inducible nitric oxide synthase; IP₃, inositol trisphosphate; IP₃Rs, IP₃ receptors; MAPK, mitogen-activated protein kinase; MCU, mitochondrial Ca²⁺ uniporter; MFN, mitofusin; NF-κB, nuclear factor kappa B; RyRs, ryanodine receptors; VDAC, voltage-dependent anion channel. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

f. ROS/RNS and growth factor receptor activation

Lipid raft-mediated membrane reorganization can lead to activation of receptor and nonreceptor tyrosine kinases, such as ERBB1, that can homodimerize or heterodimerize with other members of the ERBB receptor family (ERBB2, ERBB3, and ERBB4) (91), insulin-like growth factor-1 (IGF-1) receptors, and death receptors (Fig. 12). Binding of growth factors to their tyrosine kinase receptors leads to activation of MAPK and PI3K/AKT signaling pathways. These processes are mediated by similar signaling pathways. Indeed, activation of ERBB1, 2, and 3 leads to activation of RAS family members (K-RAS, N-RAs, H-RAS) and further downstream signaling pathways that involve MAP3K (MEKK2/3, RAF-1/B-RAF), MAP2K (MEK1/2; MEK5), and MAPK (ERK1/2) and in turn activate transcription factors, such as AP1, NF-κB, and CREB (315). Similarly, death receptors activate two other MAP3K/MAP2K/MAPK pathways resulting in the activation of p38 (α−γ) kinases and JNKs (1/2). Finally, IGF-1 receptor and its downstream kinases PI3K, PDK1, and AKT participate in mTOR activation and GSK-3 regulation. Ultimately, all these different pathways result in the activation of transcription factors involved in cell death or survival.

OPEN IN VIEWER

FIG. 12.

The MAPK and PI3K signaling pathways. The four major MAPK signaling pathways (ERK1/2, ERK5, JNK1/2, and p38) regulate cell survival (ERK1/2 and ERK5) and apoptosis (JNK1/2 and p38) via the expression of transcription factors. The PI3K signaling pathway is involved in cell growth via AKT1/2 proteins. Death (Fas/CD95, TNF-R, DR3-5) and growth (HER family and IGF-1 receptors) receptors are located in lipid rafts and can be activated during coalescence of ceramide-enriched raft platform. They facilitate the cross talk between death or growth signaling from the membrane environment and intracellular signaling cascades. FAK and PYK-2 serve as scaffold proteins to facilitate the functional integration of focal adhesion proteins, such as paxillin, involved in Ca²⁺ homeostasis and can phosphorylate PI3K. ERK, extracellular signal-related kinase; FAK, focal adhesion kinase; IGF-1, insulin-like growth factor 1; JNK, c-JUN N-terminal kinase; PTEN, prime time entertainment network. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Radiation can modulate the production and expression of cytokine and growth factor receptors (54, 270, 305). For example, a 2 Gy dose can modulate the expression of cell surface receptors, such as ERBB1 (HER1) (305). As described above, NO and ROS produced by radiation can lead to defective mitochondrial function due to leakage of mitochondrial membranes, allowing the amplification of the initial signal through massive release into the cytosol, via Ca²⁺-dependent mechanisms, of O₂ ^−• anions and RNS. These, in turn, may inhibit the activity of protein tyrosine phosphatase by oxidation and/or nitrosylation of a key cysteine residue in the active site, leading to increased phosphorylation of many proteins, such as ERBB1 (308). Cells lacking functional mitochondria (ρ⁰ cells) cannot inhibit such phosphatase activities (162). Similar results were reported on ROS scavenging by N-acetyl cysteine.

The PI3K pathway, which plays a role in the long-term effects of cell survival, can also be activated on irradiation (315).

It was shown that activation of growth receptors is modulated by radiation according to waves. For example, Dent et al. found that EGFR is first activated early and for few minutes (0–5 min) after irradiation. This is followed by a second prolonged activation at 90–240 min postirradiation (73). The MAPK pathways activate membrane-bound matrix metalloproteinase (MMP) activities that promote cleavage of proforms/zymogens of multiple growth factors into their functionally activated forms that can in turn activate, later after irradiation, cell surface receptors. For example, EGFR activation could in turn promote cleavage and release of presynthesized paracrine ligands, such as pro-TGFα (109, 315) that could contribute to sustained MAPK signaling via an autocrine positive feedback loop involving EGFR, the Ras-MAPK signaling pathway, and a ligand-releasing protease (284). Released factors could then bind to cells that express EGFR or be transmitted in a paracrine way for long-distance effects (abscopal effects). The observation that a broad-spectrum MMP inhibitor (marimastat, BB2516), known to affect tumor invasion, inhibits also growth of head and neck squamous cell carcinoma (HNSCC) cells that overexpress EGFR, highlighted the MMP role (e.g., MMP9) in cleaving HER ligands from membrane-anchored precursors into their functionally activated states (218a). Inhibition of ERK/MAPK (by PD98059 or U0126) and PI3K (by LY294002 or wortmannin) leads to a marked reduction of both basal and induced MMP9 activity, indicating that the MAPK signaling pathways are also required in an autocrine loop for HER1 activation (218a).

4. Central role of NF-κB in the nuclear and extranuclear responses to radiation

Experiments showing that ATM activates NF-κB, which is involved in iNOS and COX activity, demonstrated the link between inflammation and DDR (336). Therefore, as the bystander and inflammatory responses share homologies, studying the role of NF-κB in bystander effects can be of major relevance.

a. Nuclear factor kappa B

NF-κB is a redox-sensitive transcription factor made of five different subunits that are organized in homo- or heterodimers: p50 (e.g., p50/p105 constitutes NF-κB1), p52 (e.g., p52/p100 constitutes NF-κB2), p65^RelA, RelB, and c-Rel (Fig. 13). All subunits have a 300-amino acid N terminal sequence (Rel Homology Domain) that mediates dimerization, nuclear translocation, DNA binding, and I-κB binding. NF-κB is involved in innate and adaptive immunity and chronic inflammation, thereby linking the immune response to radiation response (121). Moreover, it is a transactivator of genes involved in cell proliferation and suppression of apoptosis induced by tumor necrosis factor (TNF) α by blocking caspase-8 activation (326) by induction of BCL2 (325, 326) and by transactivation of TNF receptor-associated factor 1 (TRAF1), TNF receptor-associated factor 2 (TRAF2), and the inhibitor-of-apoptosis (IAP) proteins c-IAP1 and c-IAP2. It is also involved in autophagy, angiogenesis, metastasis formation, tumor progression, and oncogenesis (121, 246), and is considered a molecular target for cancer therapy (322). It participates in the antioxidant defenses by promoting SOD expression (139). Its inactive form binds to the inhibitor I-κB that masks NF-κB nuclear localization signal and sequesters it in the cytoplasm. Upon irradiation, it can be released and transferred to the nucleus (226).

OPEN IN VIEWER

FIG. 13.

NF-κB activation mechanisms. The antiapoptotic NF-κB consists of five heterodimers, among which the most common form involves p65 and p50. It is kept in the cytoplasm by its interaction with IκB. On activation (e.g., by ATM, TNFα, IL1), the IKK complex, which is made of IKK-α (IKK1), IKK-β (IKK2), and the regulatory subunit IKK-γ/NEMO, phosphorylates and targets IκB for ubiquitination and degradation by the proteasome, while NF-κB can enter the nucleus and activate its target genes. IKK, IκB kinase; NEMO, NF-κB essential modulator. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Specifically, TNFα binding to its receptor (TNFR) leads to a conformational change of TNFR that is accompanied by binding of TNFR type 1-associated DEATH domain protein (TRADD) to TNFR death domain. Two proteins, TRAF2 and receptor-interacting protein (RIP), are then recruited and in turn recruit the IκB kinase (IKK) complex, which is made of IKK-α (IKK1), IKK-β (IKK2), and the regulatory subunit IKK-γ/NF-κB essential modulator (NEMO) (Fig. 13). IKK is then activated and phosphorylates IκB before ubiquitination and proteosomal degradation to release NF-κB (121, 124). Among the plethora of targets with NF-κB responsive element sequences, COX-2, iNOS, and also cytokines (TNFα, IL1, IL6, IL33), chemokines (IL8, MCP-1), VEGF, ICAM1, and VCAM1 have been identified as major factors that can modify the microenvironment and trigger inflammation (95). These molecules may also be involved in oncogenesis (246).

b. NF-κB and irradiation

NF-κB can be activated by doses as low as 0.1 Gy of X-rays, but participates in the RIAR (84) also at higher doses (2–50 Gy) (32). RIAR was demonstrated by the observation that a priming dose (generally >0.005 Gy) reduces the detrimental effects of the challenging dose administered a few hours later (186, 333). This can be a beneficial effect for low-dose radiation, but might induce radioresistance during radiotherapy. Indeed, RIAR has been associated with reduced chromosome aberrations, MN formation, and mutation induction (223, 344, 356). RIAR shares similarities with inflammation. It was shown that low irradiation doses lead to ATM phosphorylation that contributes to NF-κB activation and to cell survival, in a process involving also ERK (but not p38/JNK) (4, 151). ATM participates in NF-κB activation by phosphorylating IκB and NEMO (165) that mediate NF-κB inhibition. NF-κB activation by low-dose irradiation leads to the expression of MnSOD 15 min postirradiation, and small interfering RNAs (siRNAs) against MnSOD reduce RIAR (84).

RIAR is also observed in bystander cells. Indeed, a priming dose of 0.02 Gy of γ−rays, delivered 6 h before single-cell microbeam irradiation, inhibited 50% of the bystander effects observed in control cells (no priming dose) (265).

Optimal NF-κB activation is observed at doses between 7 and 10 Gy and with high LET (90–230 keV/μm). In these conditions, NF-κB activation is mediated by ATM recruited at DNA DSB sites (336) and is an alternative pathway leading to p53-mediated signaling. p53 participates in cellular redox status by transactivating p53-induced genes (PIGs) that encode antioxidant molecules (GPx). In irradiated cells, p53 transactivates genes encoding ROS-generating enzymes, such as quinone oxidoreductase (NQO1, PIG3) and proline oxidase (POX, PIG6), BAX, PUMA, and p66SHC, thereby leading to oxidative stress and apoptosis (168, 240). However, the ATM-activated p53 signaling pathway is not required for the bystander response (95), as indicated by the finding that bystander effects can be observed in p53-null cells exposed to gamma rays (350) or to Auger-emitting radionuclides (227).

Several studies reported that common p53-regulated radiation response genes, such as cyclin-dependent kinase inhibitor 1 (CDKN1A known as p21^Waf1), are upregulated in alpha particle-irradiated normal human lung fibroblasts, and also in bystander cells (8, 90). Genes regulated by NF-κB, such as COX-2, IL8, and BCL2A1, also are similarly expressed in bystander and irradiated cells (95).

Following gamma or alpha irradiation, activation of NF-κB signaling leads to the production of cytokines and chemokines, such as IL-1α and β, IL-6, TNFα, CXCL1, CXCL2, and CXCL8 (121, 158) (Fig. 14). Activation can occur within few hours following irradiation and might be ATM dependent. A second activation wave can be observed after 24 h and this might be related to receptor binding by secreted cytokines, such as TNFα (24). Therefore, NF-κB could modulate the response of directly irradiated cells and also alert neighboring nonirradiated cells via autocrine and paracrine pathways.

OPEN IN VIEWER

FIG. 14.

The NF-κB and cytokine receptor pathways in the bystander response. DNA DSBs in irradiated cells activate ATM and NEMO that in turn assemble IKK before IκB ubiquitination and proteosomal degradation. Released NF-κB enters the nucleus and induces the transcription of target genes, such as those encoding cytokines. Secreted cytokines can in turn bind to receptors on bystander cells. On binding, receptors will activate NF-κB-responsive element-containing molecules (e.g., cytokines, COX-2, and iNOS), thus contributing to ROS and RNS production and transmission of bystander signals in a self-sustained process. The MAPK and JAK2-STAT3 pathways are also activated by IL8 and IL6, respectively. The TGFβ receptor also can activate NOS. β-Cat, β-catenin; JAK2, Janus kinase 2; PGE2, prostaglandin E2; STAT3, signal transducer and activator of transcription 3; TGF, transforming growth factor. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

NF-κB regulates cytokine production in irradiated cells, but it is controlled by cytokine signaling in bystander cells (151). NF-κB-dependent expression of IL6 via IL6-receptor complex activates the Janus kinase 2 (JAK2)-signal transducer and activator of transcription 3 (STAT3) pathway and STAT3-dependent gene expression, thus establishing a link between ATM, NF-κB, and STAT3. STAT3 is a transcription factor located in the cytoplasm and transferred to nucleus on activation. It also maintains constitutive NF-κB activation by prolonging NF-κB nuclear retention (105, 106). NF-κB also transactivates IGF-1 receptor, thereby promoting PI3K/AKT-GSK3b-β-catenin signaling pathways in both irradiated and bystander cells (94, 119, 210) (Fig. 14).

NF-κB can also interfere with the main MAPK signaling. More specifically, it suppresses the JNK cascade and ROS activity (36, 232). It must be noted that TRADD, a TNFR1-associated signal transducer, can also bind to TRAF2 that activates NF-κB and Fas-associated protein with death domain (FADD), two proapoptotic factors, whereas NF-κB is a potent antiapoptotic agent (129).

NF-κB also controls IGF-1 receptor that can activate the downstream (PI3K)-AKT survival pathway in both directly irradiated and bystander fibroblasts (94). In bystander cells, GSK3β phosphorylation by AKT is accompanied by stabilization of beta catenin that acts as a nuclear activator of transcription after GSK3b phosphorylation by WNT signaling (86).

5. The COX-2 and iNOS

The NOS and COX systems include constitutive forms (NOS1, NOS3, and COX-1), which are expressed in many cell types and are mostly involved in housekeeping tasks, and inducible forms (iNOS known as NOS2 and COX-2), which are activated in stress conditions (294). For example, activated macrophages produce clastogenic factors, via superoxide and NO, and can induce gene mutations, DNA base modifications, DNA strand breaks, and cytogenetic damage in neighboring cells.

NOS1 and NOS3 are Ca²⁺/calmodulin dependent, while NOS2 is much less. Inducible forms lead to NO production from l-arginine and can be activated by irradiation (181). For example, iNOS activation has been observed as early as 3 h after X-ray irradiation and lasts over 24 h (187). Similarly, NOS1 can be activated by therapeutic doses (2 Gy) and blocks the cytoprotective effects of radiation-induced ERK1/2 activity (161).

COX-2, also known as prostaglandin endoperoxide synthase 2 (PTGS2), converts arachidonic acid into prostaglandin endoperoxide H2 (PGH2), a precursor of PGE2 involved in inflammation. During this reaction, singlet oxygen can also be released. COX-2 is an inducible enzyme responsible for the generation of ROS and of proinflammatory PGE2 (355).

Expression of iNOS (12) and COX-2 (133) is also controlled by NF-κB and is involved in secondary proinflammatory waves following irradiation. Conversely, cNOS can stimulate the early signaling effects of low-dose irradiation (161).

COX-2 is also upregulated in bystander normal human fibroblasts and its inhibition by NS-398 in bystander cells reduces mutagenesis and genetic instability (354). COX-2 is a downstream target of MAPK pathways, such as ERK, JNK, and p38 kinase. Zhou et al. showed that inhibition of ERK phosphorylation suppresses COX-2-mediated bystander response (354). Therefore, as a downstream target of the NF-κB and MAPK/AP1 pathways, COX-2 can be induced by a variety of molecules (TNFα, TGFβ, IL6, IL8, IL33, TRAIL, and IGF-1) (293). Finally, radiation-induced activation of NF-κB and of its downstream regulated genes that encode cytokines, such as IL-8, IL-6, COX-2-generated PGE2, and IL-33, can lead to COX-2 activation in bystander cells, via autocrine/paracrine stimulation of the NF-κB and MAPK pathways (134).

Zhou et al. also showed that radiation-induced bystander effects require mitochondria-dependent NF-κB/iNOS/NO and NF-κB/COX-2/PGE2 signaling pathways (355). While COX-2 leads to ROS production, it is interesting to note that MnSOD is also activated by NF-κB (139), thus providing an antioxidant role for NF-κB in agreement with its antiapoptotic and proliferation properties.

We mentioned above that post-translational modifications of p53, through phosphorylation by ATM and also by other kinases, such as CHK2 and homeodomain-interacting protein kinase 2 (HIPK2), contribute to its stabilization and subsequent induction of transcription of downstream genes involved in cell cycle arrest, programmed cell death, or cell metabolism. Accumulation of p53 also attenuates iNOS induction because p53 interacts with TATA binding protein and/or NF-κB that are essential for iNOS expression (85).

1. Immune response as the mediator of radiation-induced systemic effects

As discussed above, ionizing radiation can cause a variety of effects in cells directly exposed to radiation and also in neighboring and distant nonirradiated cells (i.e., off-target effects or systemic effects). Recent reviews on the topic describe extensively the biological and chemical molecules that induce and/or participate in the propagation of these various off-target effects. The growing number of data supports the necessity of using different approaches (i.e., bioinformatics and meta-analyses) to have a correct picture of the governing mechanisms and type of biological molecules that contribute to the different off-target effects. The most recent bioinformatic studies and meta-analyses have confirmed the interactions between mediators of systemic effects and DNA damage response/repair (DDR/R) components, as well as interactions between pivotal components of the innate immune response, such as pattern recognition receptors (PRRs), and DNA repair proteins (BRCA1, XRCC1, DNA-PK, Ku70/80, and others) (213).

Nikitaki et al. generated a detailed list of proteins involved in different categories of radiation-induced systemic effects, including the clinically relevant abscopal phenomenon, by using improved methodologies of literature text-mining and various bioinformatic tools. Many of these proteins belong to the DDR complex network and have been found as central hubs in the various protein/protein interaction (PPi) networks, indicating that the key pathways involved in off-target effects are apoptosis, TLR-like, and NOD-like receptor signaling pathways (213) (Fig. 15). PRRs are expressed by the cells of the innate immune system to recognize two classes of molecules and various “danger” signals: pathogen-associated molecular patterns (PAMPs), which are associated with microbial pathogens, and DAMPs, which are associated with cellular components that are produced after cell damage or death. In general, PRRs can recognize abnormal molecular complexes as a consequence of infection, inflammation, or other types of cellular stress (195). DAMPs seem to play an important role in the communication of this stress system-wide and in different organisms, from plants, fish, rats to humans, as reviewed in Ref. (188). Several studies have shown that radiation exposure results in the initiation of various triggering mechanisms associated with the inflammation and immune responses of the host organism, including the release of proinflammatory cytokines and chemokines by monocytes and macrophages. Radiation-induced cytokine gene upregulation is frequently observed, and several genes encoding inflammation-related cytokines (e.g., IFNs, IL-1, IL-6, IL-8, VEGF, EGFR, and TNFα) are considered to be “early response” genes that are activated within minutes to hours after irradiation. This process is considered a genuine “danger” signal in response to radiation injury and it is the main source of de novo ROS production. This secondary ROS production can partially explain the late increase in the expression of inflammatory cytokines in irradiated cells (268). Radiation-generated cytokines are responsible for the formation of inflammatory lesions and work together with DAMPs to create the proinflammatory, pro-oxidant microenvironment necessary to characterize this site as a “stress” site and to induce a systemic response. This is expected to promote maturation of dendritic cells and, in cancer treatment, the development of an effector T cell response (i.e., innate response to tumor-associated antigens) (Fig. 15). This process is completed by the attraction of cellular components of the immune system, such as neutrophils, and then macrophages and lymphocytes (268).

OPEN IN VIEWER

FIG. 15.

Complex DNA damage leads to immune signaling. (A) Repair of a clustered damaged DNA site: a challenging task. On induction by ionizing radiation of complex DNA damage, such as a DSB and two oxidative DNA lesions (a damaged base, red star, and an apurinic/apyrimidinic, AP, site), at least two DNA repair pathways and several DNA repair proteins will be activated. For the base damage, BER is the main repair pathway, while for the DSB, only NHEJ will be considered here for simplicity. In all cases, the most basic proteins and enzymes are described. For short-patch BER, a DNA glycosylase will recognize and remove the damaged base and the repair should be completed by the concerted activity of AP endonuclease 1 (APE1), a DNA polymerase, and ligase III to seal the broken ends. In the nearby DSB area (a few bp apart), the Ku heterodimer (Ku70/80) initiates NHEJ by binding to the free DNA ends and engaging other NHEJ factors, such as DNA-PK, XRCC4, and DNA ligase IV, to the DSB site. DNA-PK becomes activated on DNA binding and then phosphorylates a number of substrates, including p53, Ku, and the DNA ligase IV cofactor XRCC4. Phosphorylation of these factors is believed to further facilitate DSB processing. For ligation, the ends must be partially processed by the nucleases Artemis, MRE11/RAD50/NBS1 complex, and FEN-1. Moreover, as shown by advanced fluorescence microscopy, the formation of each DSB is rapidly accompanied by phosphorylation of thousands of histone H2AX molecules (γH2AX). The MRN complex functions as a sensor of DNA ends and activates the ATM kinase that phosphorylates CHK2, p53, and H2AX in flanking chromosomal regions. (B). Systemic effects. Processing of clustered DNA damage can lead to unrepaired and persistent DNA damage that can cause cell senescence or cell death (i.e., apoptosis). This can trigger the extracellular release of different “danger” signals or damage-associated molecular patterns (DAMPs: ATP, short DNAs/RNAs, ROS, heat shock proteins [HSPs], high-mobility group box [HMGB]-1, S100 proteins, and others). DAMPs activate different PRRs, such as TLRs and the formation of inflammasomes, a process that leads usually to inflammation and immune-related pathologies. Interestingly, recent evidence (see section IV.B.1) suggests a direct interaction between different PRRs and DNA repair proteins. Cell damage or death can also lead to the release of several cytokines and chemokines that can regulate immune responses. PRR activation usually results in NF-κB-mediated release of various proinflammatory cytokines, such as IFNs, IL-1, IL-6, IL-8, VEGF, EGFR, and TNFα. The activation of APCs, for instance, dendritic cells and macrophages, will induce primarily the innate immune response (activation of T cells) and most rarely the adaptive immune response (mediated by B cells). In all cases, the constant triggering of the immune system might generate many detrimental systemic effects for the organism. Positive immunomodulation is usually mediated by the action of regulatory (suppressor) T cells (i.e., Treg cells), suppressor macrophages, and immunosuppressive cytokines to maintain overall tissue homeostasis. APCs, antigen-presenting cells; BER, base excision repair; DNA-PK, DNA-dependent protein kinase; EGFR, epidermal growth factor receptor; PRRs, pattern recognition receptors; Treg cells, regulatory T cells; TLRs, Toll-like receptors. To see this illustration in color, the reader is referred to the web version of this article at www.liebertpub.com/ars

Overall, the immune response to radiation is highly related also to the radiation sensitivity (i.e., propensity to undergo apoptosis) of the various immune cell subsets, depending on the lineage, maturity, and activation status. B cells and naive T helper (Th) cells are radiation sensitive, whereas T memory cells, natural killer T cells, and regulatory T cells (Tregs) are more resistant (267). As presented in Figure 15, the final balance between controlled immune response and excessive, pathologically chronic immunogenic response is not easily understood. Certainly, the induction of repair-resistant complex DNA damage on exposure to ionizing radiation is expected to trigger a persistent immune signaling through the continuous activation of the various DNA damage checkpoints that freeze cell cycle until damage is repaired, or direct the cell toward senescence or apoptosis (92). For example, in mice in which the ERCC1-XPF DNA repair endonuclease has been knocked down in all cells or only in adipocytes, DNA damage signaling activates a chronic autoinflammatory response resulting in fat depletion. This response is cell autonomous and most importantly requires ATM, a well-known DNA damage signaling molecule (146).

The role of the innate immune system in coordinating these response types is a rapidly emerging research area (146). As discussed by Gasser et al. (93), the immune system uses mislocalized or damaged DNA to discover infected or otherwise diseased and malfunctioning cells. Conversely, genomic DNA and mtDNA in healthy cells are usually overlooked by the DNA recognition pathways. Furthermore, cells have a limited ability to degrade mislocalized DNA and to repair any damaged DNA, possibly to prevent the unplanned activation of DNA sensor pathways and of the inflammation and immune systems. It is generally accepted that there is a multivariable alliance between DDR and immune signaling and the modulating parameters are not always understood (234). One of the most coherent immune rebalancing activities involves shifting the T cell equilibrium toward regulatory (suppressor) T cells (i.e., Treg cells), and this can occur after radiotherapy (267). In addition, immunomodulation can be achieved also by the action of suppressor macrophages and immunosuppressive cytokines to secure overall tissue homeostasis (268).

2. Abscopal effects: historical changes and clinical evidence

Although modern radiotherapy is getting progressively closer to its ideal target size or volume, still there are unresolved questions; particularly, it is not clear whether radiation effects are be really limited and localized. The idea of distant, abscopal effects was introduced for the first time by Mole in 1953 (196). Abscopal is made of the “Ab” prefix, which means “away from,” and “scopos” (Latin), which means shooting mark or target. Therefore, this adjective describes the idea of “off-target” or “away from target” effects and supports the hypothesis of the interdependence of all body cells (196). As suggested by Mole, this practically means that damage to one cell will unavoidably affect the body as a whole. This initial hypothesis was based on the finding that in rats, the thyroid gland synthetic activity was reduced to 25% of the normal 3 days after localized irradiation with 6–10 Gy. However, this was not caused by whole body irradiation or direct irradiation of the thyroid or the pituitary gland. Moreover, the thyroid function reduction was observed only when a sufficiently large volume of the abdomen was irradiated (196).

Although the idea of abscopal effects seems logical, it has been disregarded in radiotherapy for many years. It took about 50 years to show clear mechanistic evidence, at least in mice, about the immune system's role in this phenomenon (72). Specifically, mice bearing one syngeneic mammary carcinoma (mouse 67NR cells derived from mammary gland malignant neoplasms) in each flank were treated with the growth factor Flt3-Ligand (Flt3-L; daily for 10 days) to boost dendritic cell production, after local radiotherapy (or not) to only one of the two tumors (single dose of 2 or 6 Gy). Growth of the nonirradiated tumor was impaired only in mice that received unilateral radiotherapy and Flt3-L, but not in animals treated only with Flt3-L. This result indicates that immunity is involved in the abscopal effect. Radiotherapy immunostimulatory effects have created a wide interest due to the preclinical and clinical observations that tumor-localized radiotherapy can occasionally prompt anticancer immune responses that facilitate the regression of distant and nonirradiated metastatic tumors, as recently reviewed in Wennerberg et al. (332).

Although circumstantial evidence existed since the 1950s, abscopal effects are still considered in some cases as unexplained, obscure, or of limited clinical value due to the high specificity and dependence on the type of host organism, tumor, radiation, and so on. In the 1960s, Law and Mole (159) showed direct (i.e., targeted) and abscopal effects of X-ray radiation on the thymus of weanling rats. More recently, many studies using primarily mice or rats described a variety of “off-target” effects (34, 57, 137, 155, 156, 184, 216, 320, 321). Additional work is needed to precisely evaluate the type and degree of abscopal effects in a variety of model organisms [mice (264), rats, earthworms (197)] and different targeted organs (head, thorax, etc.) (101, 152, 190, 236). Very recently, Ventura et al. exploited synchrotron radiation to study targeted radiation-induced “off-target” effects in C57BL/6 mice (317). Under different radiation settings, irradiation of a small leg area induced pronounced persistent systemic genotoxic effects (such as complex DNA damage and cell apoptosis) also in nonirradiated areas. These genotoxic events were accompanied by changes in the plasma concentrations of macrophage-derived cytokines, eotaxin, IL10, TIMP1, VEGF, TGFβ1, and TGFβ2, as well as changes in the tissue proportion of macrophages, neutrophils, and T lymphocytes, underlying the strong links between response to radiation and immune system.

It is now quite safe to suggest that local irradiation of a tumor leads to cell death and tissue damage and the release of ROS, cytokines, and danger signals, several of which can trigger an innate immune response (144), although clinical evidence is still rare. One of the first clinical evidences was the description in 2009 of an abscopal effect in a patient with chronic lymphocytic leukemia during radiation therapy (157). One week after X-ray radiotherapy, the lymph nodes in the neck that were not irradiated and distant from the irradiated area started to regress, and after 2 weeks of radiotherapy, they showed complete regression (157). The same year, another patient was diagnosed with acute radiation pneumonitis and other signs of radiation toxicity after spine irradiation. The authors suggested that this should alert clinicians on the toxicity risk for nontarget organs that receive low-dose or zero radiation (283). For a list of the reported clinical cases concerning abscopal effects in patients with nonhematological malignancies and treated by conventional radiation (patient characteristics, treatment strategy, and outcomes), the reader can refer to a relatively recent review by Siva et al. (291).

Currently, the idea that abscopal effects exist and can be modulated primarily via the immune system is progressively more accepted (87, 125, 174, 261). Encouraging results based on case reports and preclinical data suggest that radiotherapy and immunotherapy may synergize to create a scenario of off-target responses away from the radiation field that could be beneficial for the patient (2, 51, 60, 70, 71, 116, 171, 273, 301). DDR mechanisms and inflammatory responses have been recently observed in patients undergoing radiotherapy for nonsmall-cell lung cancer. The observed abscopal effect was also linked to changes in the plasma levels of MDC/CCL22 and MIP-1alpha/CCL3 cytokines (290). Undoubtedly, one of the major breakthroughs is the finding that the combination of radiation and immunotherapy to boost the immune system (e.g., with the human monoclonal anti-CTLA-4 antibody ipilimumab) induces immune-mediated abscopal effects in poorly immunogenic preclinical tumor models and in patients with metastatic melanoma or other cancer types (62, 100, 237). The suggested mechanism for this therapeutic type is that the anti-CTLA4 monoclonal antibody binds to CTLA4 expressed on the surface of T cells and inhibits the CTLA4-mediated downregulation of T cell activation (75). This consequently leads to a boosted cytotoxic T lymphocyte (CTL)-mediated synergistic immune response against cancer cells and high tumor immunity, especially when combined with radiation. Currently, the National Cancer Institute (NCI) lists only one Phase I clinical trial for solid cancers or lymphoma. This is a multicenter study to evaluate the safety of an anti-CTLA-4 human monoclonal antibody (AGEN1884) and to estimate the maximum tolerated dose in subjects with advanced or refractory cancer (Clinicaltrials.gov ID: NCT02694822).

Another early example of regression of hepatocellular carcinoma was observed after radiotherapy (total dose of 36 Gy) for bone metastasis (220). Doses of 2 and 6 Gy have been used in combination with the dendritic cell growth factor Fms-related tyrosine kinase 3 (FLT3) or with injection of dendritic cells in tumors as immune therapy (72, 150). In mice with melanoma tumors, a single fraction of 15 Gy showed similar results as 3 × 5 Gy and was accompanied by the generation of tumor antigen-specific effector cells that traffic to the tumor (175). The combination of anti-CTLA4 antibodies with fractionated irradiation (3 × 8 Gy or 5 × 6 Gy), but not single-dose irradiation (20 Gy), showed tumor growth delay outside the field of irradiation in preclinical models (300).