Radiation is used both from conventional external sources and internal radioisotope-targeted therapy as a major treatment for neoplasms with the main aim of maximizing tumor cell killing and minimizing damage to normal tissues. It is well known, however, that radiation is carcinogenic. For example, epidemiologic studies have shown that exposure to environmental radiation such as α-particles emitted from radon and its daughter products leads to a high incidence of radiation-induced lung cancer.1 It has been commonly accepted that radiation-induced effects for both normal and tumor cells are a result of energy deposited in the nuclear DNA of a cell.
Recently, however, mounting evidence has indicated that some biologic responses can occur in non-irradiated cells via the production of bystander effects.2, 3, 4 Such radiation-induced bystander effects include cell killing, the induction of micronuclei (MN), sister chromatid exchanges (SCEs), mutations, genomic instability, changes in gene expression and in cell growth.5, 6, 7, 8, 9, 10, 11, 12, 13 The mechanisms underlying radiation-induced bystander effects are still not well understood. For confluent cell monolayers and 3-dimensional cell clusters, gap junctional intercellular communication (GJIC) has been found to play a critical role in radiation-induced bystander effects.14, 15, 16, 17, 18 In other cases, soluble extracellular factors, including reactive oxygen species (ROS) and cytokines,19, 20, 21 which are released from irradiated cells and even human tissue,22 contribute to the bystander responses. We and others have found that radiation-induced nitric oxide (NO) has multiple bystander effects on non-irradiated cells.10, 23, 24 It has been suggested that the radiation-induced bystander effects, including genomic instability and increased mutation, may contribute to second cancer induction in normal cells after radiotherapy.25
Most of the previous studies of bystander effects focused on the radiation-induced responses within a cell population or cluster consisting of the same cell type. Recently, however, bystander responses have been reported in a 3-dimensional ureter primary tissue model after microbeam irradiation.26 Our previous studies have shown that, when a single cell is targeted, a significant bystander response is induced in the populations of either human glioma T98G cells27 or human primary fibroblast AG01522 cells.18 It is not known, however, whether there are direct bystander interactions between normal and tumor cell populations, and if so, what signaling factors are involved. Here, we explored these questions by using 2 cell lines, namely, human glioma cells and fibroblasts, and individually targeting a fraction of cells through their nuclei within one population with a precise number of microbeam helium particles. These 2 cell lines were selected as models because they have both been widely reported to produce bystander responses in their own rights and some mechanistic information is already known.8, 15, 28 Bystander signaling was observed, the mechanism of which was genotype-dependent.
Material and methods
Cell cultures and treatments
Human glioblastoma T98G cells and normal human diploid skin fibroblast AG01522 (hereafter AG0) cells were used for this study. T98G and AG0 have a mutated p53 (mp53) gene29 and a wild-type p53 (wtp53) gene,15 respectively. T98G cells were maintained in RPMI-1640 medium supplemented with 10% (v/v) fetal calf serum (FCS) and 0.01% sodium pyruvate, 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin. AG0 cells were grown in α-MEM containing ribonucleosides supplied with 18% FCS, 1% nonessential amino acids, 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin. All cells were maintained at 37°C in an atmosphere of 95% air and 5% CO2.
One day before microbeam irradiation, the cells were trypsinized and a 20 μl cell suspension with approximately 1,000 plateau phase T98G cells were seeded on one region of a specially designed microbeam dish consisting of a 3 μm thick Mylar film base. At the same time, a drop of AG0 cells containing about 1,000 cells was seeded on another separate region 5 mm away from the T98G population. The regions prepared for cell seeding had been pretreated with 1.7 μg/cm2 Cell-Tak adhesive (Collaborative Biomedical Products, Bedford, MA). After 2 hr of cell seeding, 2 ml α-MEM medium was supplied to the microbeam dish for overnight cell coculture so that the cell cytoplasm was well spread out. Pilot test showed that T98G cells also had optimal growth in the α-MEM medium.
One hour prior to irradiation, the cells were stained with 0.2 μg/ml Hoechst 33342, enabling individual nuclei to be identified by the microbeam system.30 Excess stain was removed by washing the cells with serum-free α-MEM containing 10 mM HEPES just before irradiation, and cells were maintained in this medium during microbeam irradiation in order to avoid any possibility of the production of short-lived SCE-inducing factor that might result from serum-containing culture medium being exposed to α-particles.20 Most of the cells were present as well-separated individual cells before irradiation. Immediately after irradiation, the medium was replaced with 2 ml complete α-MEM containing FCS to allow cell culture for 24 hr until further treatment for MN measurement. In some experiments, 20 μM 2-(4-carboxyphenyl)-4,4,5,5-tetramethyl-imidazoline-1-oxyl-3-oxide (c-PTIO; Molecular Probes, Eugene, OR), 0.5% DMSO, or the antioxidant mixture of 150 U/ml superoxide dismutase (SOD) and 150 U/ml catalase (CAT) was present in the medium during and after irradiation. C-PTIO and DMSO are commonly used scavengers of NO and ROS, respectively. SOD and CAT were used to metabolize superoxide anion radicals and hydrogen peroxide, respectively.
Cells in only one population of either T98G or AG0 were irradiated using the Gray Cancer Institute Charged Particle Microbeam. Details of the microbeam experimental setup have been described previously.30, 31 The nuclear position of each stained population was scanned by a computerized imaging system, and its coordinates were stored so that it could be revisited and irradiated automatically. A fraction, including one cell, 1%, 10% and 100% of the cells in the targeted population, was individually irradiated at the nuclear center by a precise number of helium-3 particles (3He2+, used as surrogates for α-particles) with a linear energy transfer (LET) of 100 keV/μm. Using these particles, 99% of cell nuclei could be precisely targeted with an accuracy of ± 2 μm.30 Control dishes were handled in the same way, including cell scanning but with no irradiation. The cell population neighboring the irradiated population in the microbeam dish, which was destined for bystander analyses, was not scanned.
The bystander responses of MN formation and free radical production were assayed 24 hr post-irradiation rather than the 1 hr we used previously.27 It has been found that the yields of irradiation-induced production of ROS,21 NO and its oxidization products32 and some cytokines33 increased with the cell culture time post-irradiation.
MN were assayed in situ by using the cytokinesis block technique.34 Following the cell coculture process, the culture medium in the microbeam dishes was replaced with 1 ml of medium containing 1 μg/ml cytochalasin-B. To get an optimum frequency of binucleated cells, the cultures were maintained at 37°C for 25 ± 1 hr and 47.5 ± 0.5 hr for T98G and AG0, respectively. The cells were fixed in situ with methanol:acetic acid (9:1 v/v) for 20 min. After air-drying, cells were stained with 10 μg/ml acridine orange for 5 min. MN were scored in binucleated cells of the population and classified according to standard criteria.35 The MN yield, YMN, was calculated as the ratio of the number of MN to the number of the binucleated cells. It was confirmed that treatment with c-PTIO or DMSO did not significantly influence the MN yields of either T98G or AG0 control populations.
After 24 hr of irradiation, the NO level, represented by its derivatives, in the cells was measured in situ by using DAF-FM diacetate (Molecular Probes) rather than using an iNOS assay because of the limited number of cells here. The details have been described previously.27 Briefly, after treatment with 3.5 μM DAF-FM diacetate for 45 min, cells were washed and incubated for an additional 20 min to allow complete de-esterification of the intracellular diacetates. The fluorescence images of at least 100 randomly selected cells per dish were captured using a 3 CCD-cooled color camera (Photonic Science, East Sussex, U.K.) attached to a fluorescent microscope (Axioskop; Zeiss, Oberkochen, Germany). The exposure conditions were standardized to allow quantitative comparisons of the relative fluorescence intensity of the cells between groups.
The ROS level in the cells was measured in situ by using CM-H2DCFDA (Molecular Probes) 24 hr post-irradiation. Cells were treated with 3.5 μM CM-H2DCFDA for 20 min and then washed and incubated for additional 10 min at 37°C in order to obtain stable fluorescence under the UV light of microscope. The fluorescence image of at least 100 randomly selected cells per dish was captured and analyzed with the same method as that used for the NO assay.
Statistical analysis was done on the means of the data obtained from at least 3 independent experiments. Two replicates were counted for each experimental point to assess the micronucleus formation. All results are presented as mean ± SEM. Statistical calculations were performed by use of SigmaStat (SSPS, Chicago, IL). A one-way analysis of variance (ANOVA) on ranks was performed by the Kruskal-Wallis method and p-values were obtained for multiple comparisons with the Dunnett's method. All statistical tests were 2-sided and the differences were considered to be statistically significant at p < 0.05.
MN induction in bystander non-irradiated populations
It was found that the yields of MN in both non-irradiated populations of AG0 and T98G increased after they were co-cultured for 24 hr, respectively, with T98G and AG0 populations, where a fraction of cells had been individually targeted with a precise number of 3He2+ particles. A typical example of this increase is illustrated in Figure 1, where different fractions of AG0 and T98G populations had been individually targeted with one 3He2+ particle. Even when only one cell in a population was targeted by a single particle, the MN yield in the neighboring non-irradiated population was significantly increased relative to the nontreated control. For the bystander AG0 cells, it was increased from 0.033 to 0.058 (p = 0.009; Fig. 1a); for the bystander T98G cells, it was increased from 0.138 to 0.175 (p = 0.032; Fig. 1b). This result gives direct evidence that single cell irradiation in one population of cells can induce a bystander response in another cell population with a different genotype. In previous studies we have performed, after a single cell had been irradiated with one 3He2+ particle, the yield of MN in the whole T98G cell population was increased by 16% 1 hr after irradiation and by 54% 24 hr after irradiation. Other workers have shown that bystander responses can occur very quickly after irradiation and that the overall level of effect can also increase with time after exposure.36, 37
Another finding was that the bystander response-induced formation of MN was not influenced by the number of irradiated cells. When the fraction of cells, which was individually irradiated with one 3He2+ particle, increased from ∼ 0.1% (one cell irradiation) to 100%, the yields of MN caused by the bystander response in the non-irradiated populations did not increase but had similar values. With respect to 4 different fractions of irradiated cells in Figure 1, the yield of MN was increased by 79% (p = 0.016) on average for the non-irradiated AG0 population and increased by 28% (p = 0.02) on average for the non-irradiated T98G population after coculturing with irradiated populations of T98G and AG0, respectively. Parallel experiments showed that within the irradiated cell population itself, targeting of a single cell increased the yield of MN by 20% and 54% in the whole population of AG0 and T98G, respectively. Therefore, T98G tumor cells appear much more effective in generating bystander signals than AG0 cells. Importantly, it was observed that the spontaneous control MN yield of the T98G cells was 4.2 times higher than that of the AG0 cells, which suggests that T98G tumor cells are also much more effective in generating intracellular damage signals than AG0 cells. Taken together, this predicts that intrinsic cell radiosensitivity may be important in defining the bystander response.
Dose response of bystander MN induction
The bystander induction of MN in the non-irradiated populations did not vary with the particle number when a fraction, 0.1%, 1%, 10% or 100%, of cells in the co-cultured population were individually targeted. A representative result is illustrated in Figure 2; the non-irradiated population of AG0 (or T98G) was co-cultured with the irradiated population of T98G (or AG0), where 100% of cells had been targeted with 1 or 5 3He2+ particles. It is seen that the bystander MN induction in the non-irradiated population does not significantly increase with the number of 3He2+ particles delivered to the targeted cell nuclei. Therefore, the bystander effect was independent of radiation dose delivered to the cell population. This result reinforces the finding that the microbeam-induced bystander effect in a confluent culture does not depend on the radiation dose and LET18 and is also in agreement with other reports of the dose independence of the bystander effect induced by conditioned medium harvested from irradiated cells.38
Influence of c-PTIO on bystander effect
Because the non-irradiated population was co-cultured with the irradiated population in 2 separate regions and there was no cell-cell contact between AG0 and T98G cells, the cellular damage in the non-irradiated populations should result from some reactive biologic factors generated from the irradiated cells. To determine what signaling factors were involved, the cell populations were treated with c-PTIO during irradiation and subsequent cell coculture. The c-PTIO treatment itself had no influence on the MN yield in the non-irradiated control cells, but it eliminated the bystander MN formation in the AG0 population that was co-cultured with the T98G population, where a fraction of cells had been individually irradiated with 1 or 5 3He2+ particles. A typical example of the effect of c-PTIO on the bystander MN induction in AG0 population is illustrated in Figure 3(a), where a fraction of T98G cells were individually irradiated with 5 3He2+ particles. Clearly, NO contributes to the bystander MN formation in the non-irradiated AG0 population co-cultured with the irradiated T98G cells. This was further confirmed by another experiment where the treatment of T98G cells with aminoguanidine, an NOS inhibitor, eliminated the MN production in the bystander AG0 population (data not shown).
However, with respect to the non-irradiated T98G population that had been co-cultured with the irradiated AG0 cells, the bystander MN induction was only partially and not significantly reduced by the c-PTIO treatment (p = 0.38; Fig. 3b). The yields of bystander-induced MN after c-PTIO treatment were also not significantly different from the non-irradiated control (p = 0.13). Thus, NO is not the main source of the irradiated AG0-induced bystander response in the T98G cells. In fact, it will be shown below that NO could not be induced in irradiated AG0 cells with a wtp53 gene. One possible reason for the above c-PTIO effect is that the recipient T98G cells could be attacked by some unknown factors, possibly ROS-dependent, from irradiated AG0 cells. Following this stress, an NO response may be induced and it further contributes partially to the bystander MN induction in the T98G cells.
Influence of DMSO and antioxidants on bystander response of T98G cells
To investigate the signaling factor involved in the MN induction of T98G cells adjacent to the irradiated AG0 cells, further experiments were performed by treating the populations with DMSO or an antioxidant mixture of SOD and CAT during and after irradiation. Figure 4 illustrates the relationship of the MN yield in the bystander T98G population to the fraction of AG0 cells targeted with one 3He2+ particle in the presence of DMSO or antioxidants. These treatments fully inhibited the occurrence of the bystander MN formation in the T98G population neighboring the irradiated AG0 population. Treatment with SOD and CAT even decreased the background MN level of the T98G cells probably because it also reduced endogenous cellular ROS. Thus, ROS play an important role in the bystander effect induced by the irradiated AG0 cells.
NO and ROS production
To confirm the above findings and obtain direct evidence of radiation-induced production of NO and ROS, we measured the NO level and ROS level in situ in the T98G and AG0 populations 24 hr after microbeam irradiation by using the fluorescent probes DAF-FM diacetate and CM-H2DCFDA, respectively. Figure 5 illustrates representative changes in the relative intensity of NO-induced fluorescence in the T98G population, where 1% of the cell nuclei had been individually irradiated with a single 3He2+ ion; here, the fluorescence intensity of non-irradiated control is normalized to 1.0. Even when this small fraction of the cells was targeted, the NO level in the whole population was significantly increased so that the NO-induced fluorescence intensity in the irradiated T98G population was increased by 31% (p = 0.029) 24 hr after irradiation. This increase in NO is greater than the 13% observed in cells 1 hr post-irradiation.27 Treatment of cells with 20 μM c-PTIO diminished the NO-induced fluorescence of the irradiated population to the control level of the non-irradiated cells. In contrast, there was no increase of the NO fluorescence intensity in the AG0 population that had been co-cultured with the irradiated T98G cells for 24 hr (data not shown), which indicates that NO is not produced in the bystander AG0 cells.
On the other hand, when 1% of cell nuclei in the AG0 population were individually irradiated with a single 3He2+ ion, subsequent to 24 hr of cell culture, the ROS level, represented by the relative fluorescence intensity of DCF, in the irradiated population was increased by 18% (p = 0.04) and this fluorescence enhancement was eliminated by treatment with either DMSO or antioxidants (Fig. 6). Treatment with SOD and CAT not only inhibited the generation of radiation-induced ROS but also reduced the endogenous ROS to 60% of that in the nontreated AG0 cells. Moreover, the ROS level in the bystander T98G population did not vary before and after coculturing with the irradiated AG0 population, although it was reduced when the cells were pretreated with SOD and CAT.
In the present study, 2 different types of human cells, radioresistant T98G glioblastoma and primary human AG0 fibroblasts, were seeded at low densities and co-cultured in 2 separate regions of a microbeam dish. It was found that a single cell irradiated with a single 3He2+ particle could produce a significant increase in MN induction in the neighboring non-irradiated cell population, and that irradiated T98G cells were much more effective in inducing a bystander response than irradiated AG0 cells. Obviously, a medium-mediated signaling pathway has been involved in these bystander responses. The results indicate that every cell within a population has the potential to release a bystander signal, which is independent of the phenotype of the cells studied here; however, the sensitivity of cells in responding to a bystander signal may depend on intrinsic cell characteristics.
To investigate which kind of signaling factors are involved in this radiation-induced bystander effect, the cell cultures were treated with either c-PTIO or aminoguanidine (NOS inhibitor). These treatments reduce the bystander induction of MN in the AG0 population that has been co-cultured for 24 hr with 3He2+ particle-irradiated T98G cells, demonstrating that NO contributes to the bystander MN induced by the irradiated T98G cells. This deduction is strengthened by the finding that the NO level is increased by 31% even when only 1% of cells in a T98G population are irradiated. Our result is in agreement with other work that iNOS can be extensively expressed in the irradiated T98G cells 10 hr after irradiation.39 NO is generated endogenously from L-arginine by inducible NO synthases40, 41 that can be stimulated by radiation in mammalian cells bearing an mp53 gene.39 Our previous study of the kinetics of NO production suggests that NO can be released from the damaged cells during programmed cell death.32 It is believed that NO itself does not induce DNA strand break, although one of its oxidization production, peroxynitrite, is toxic and can cause DNA damage by both attacking deoxyribose and direct oxidation of purines and pyrimidines.42
With respect to the bystander response induced by the irradiated AG0 cells, it is DMSO and antioxidants rather than c-PTIO that effectively abolishes the bystander induction of MN in the non-irradiated T98G population. Correspondingly, the intracellular ROS level in AG0 is obviously enhanced by irradiation. These results indicate that it is ROS rather than NO that is induced in the irradiated AG0 cells and subsequently involved in the production of MN in the neighboring T98G cells. Similar to the production process of NO, ROS could be released during programmed cell death of AG0.43 Our finding of an ROS-involved bystander response is in agreement with the report that ROS are major contributors to the bystander effect-induced upregulation of p53 and p21 as well as MN formation in the AG0 cells irradiated with low-dose α-particles.44
The radiation-induced NO and ROS have the possibility of attacking vicinal cells within their diffusion distance of less than 0.5 mm.45, 46 However, because of having very short half-lives, they may not be the direct contributors to the cellular damage in the bystander population, which is ∼ 5 mm away from the irradiated cell population. On the other hand, it has been reported that NO and ROS are involved in the bystander responses caused by the conditioned medium harvested from irradiated cells.27, 47 Therefore, some long-lived bioactive factors downstream of NO and ROS are most likely involved in the medium-mediated bystander responses. It has been reported that some cytokines such as tumor growth factor-beta (TGF-β) can be induced and mediated by the stress of ROS or NO,48 whereas the treatment of cells with TGF-β1 activates cell surface membrane-associated NADH oxidase, which in turn increases the production of ROS19 so that the cells undergo a time-dependent increase in DNA cleavage.49 Also, cytokines, including those induced by ROS, may trigger the production of NO50 in the T98G recipient cells, which is strengthened by the result in Figure 3b, where c-PTIO treatment partially attenuates the bystander MN formation in the T98G cells. Accordingly, some unknown cytokines are likely to be candidates for the long-lived signaling factors involved in the bystander responses.
It is found here that the irradiation-induced bystander effect is independent of the number of irradiated cells as well as the irradiation dose, which shows that bystander responses may predominate after low-dose exposure, including that of a single particle of relevance to environmental exposures. Our findings also suggest that, 24 hr post-irradiation, the signaling factors produced in the coculture medium may be independent of the fraction of the individually targeted cells and have a similar concentration and thus a signal amplification mechanism may be involved. When a few cells in a population are individually irradiated, some signaling factors mediated by ROS or NO can be released from damaged cells. These active molecules will react with a number of vicinal non-irradiated cells simultaneously; the new damaged cells will release biochemical factors and cause other cells to be damaged again. With such a series of cascade reactions, the original signaling factors generated from the directly irradiated cells are magnified so that a significant bystander effect is triggered by a single cell irradiation. The progress of these amplification reactions may require several hours as the yields of irradiation-induced bystander signaling factors, including ROS and NO, increase with the cell incubation time after irradiation.21, 23
In summary, the bystander signaling between 2 cell populations of T98G (mp53) and AG0 (wtp53), leading to the induction of micronuclei, can be triggered by individual 3He2+ particles, and the production of radiation-induced signaling molecules, NO and ROS, is dependent on the genotype of the irradiated cells. Via a cascade of reactions, NO is involved in the irradiated T98G-induced bystander response, whereas ROS contribute to the irradiated AG0-induced bystander effect. This differential in mechanisms between normal and tumor cells may allow potential exploitation of bystander responses if these are observed in tissue systems. For instance, inducible NO regulates tumor microenvironments, leading to angiogenesis and vasorelaxation,51 and then enhances the growth of tumors.52, 53 On the other hand, it has been reported that NO can act as an intrinsic radiosensitizer in vivo.54 Mechanistic control of the bystander effect could either reduce radiation damage and potential carcinogenesis in normal tissues by reducing bystander signaling or maximize cell sterilization during radiotherapy by amplifying bystander responses in tumors.
The authors are grateful to Mr. Stuart Gilchrist and Mr. Bob Sunderland for assistance with the microbeam irradiations and to Dr. Victoria Stewart for helpful discussions.