Dgi

KTJa

b

c

a

ARRAA

KDXDR

1

rcrtpsi(aicn

h0

Mutation Research 765 (2014) 32–39

Contents lists available at ScienceDirect

Mutation Research/Fundamental and MolecularMechanisms of Mutagenesis

j ourna l h o mepa ge: www.elsev ier .com/ locate /molmutComm u ni t y add ress : www.elsev ier .com/ locate /mutres

ose- and time-dependent changes of micronucleus frequency andene expression in the progeny of irradiated cells: Two componentsn radiation-induced genomic instability?

atriina Huumonena, Merja Korkalainenb, Eeva Bomanc, Janne Heikkiläc, Anne Höytöa,apani Lahtinenc, Jukka Luukkonena, Matti Vilukselab, Jonne Naaralaa,ukka Juutilainena,∗

University of Eastern Finland, Department of Environmental Science, P.O. Box 1627, 70211 Kuopio, FinlandNational Institute for Health and Welfare, Department of Environmental Health, P.O. Box 95, 70701 Kuopio, FinlandKuopio University Hospital, Cancer Center, P.O. Box 1777, 70211 Kuopio, Finland

r t i c l e i n f o

rticle history:eceived 18 September 2013eceived in revised form 11 April 2014ccepted 12 April 2014vailable online 3 May 2014

eywords:ose–response-rays

a b s t r a c t

Murine embryonic C3H/10T½ fibroblasts were exposed to X-rays at doses of 0.2, 0.5, 1, 2 or 5 Gy. Tofollow the development of radiation-induced genomic instability (RIGI), the frequency of micronucleiwas measured with flow cytometry at 2 days after exposure and in the progeny of the irradiated cells at8 and 15 days after exposure. Gene expression was measured at the same points in time by PCR arraysprofiling the expression of 84 cancer-relevant genes. The micronucleus results showed a gradual decreasein the slope of the dose–response curve between days 2 and 15. The data were consistent with a modelassuming two components in RIGI. The first component is characterized by dose-dependent increase inmicronuclei. It may persist more than ten cell generations depending on dose, but eventually disappears.

elayed effectsadiation-induced genomic instability

The second component is more persistent and independent of dose above a threshold higher than 0.2 Gy.Gene expression analysis 2 days after irradiation at 5 Gy showed consistent changes in genes that typicallyrespond to DNA damage. However, the consistency of changes decreased with time, suggesting thatnon-specificity and increased heterogeneity of gene expression are characteristic to the second, morepersistent component of RIGI.

© 2014 Elsevier B.V. All rights reserved.

. Introduction

The classical paradigm of radiobiology assumed that directadiation-induced DNA damage is the main cause of the biologi-al effects of ionizing radiation. Such direct damage, if not correctlyepaired, was assumed to result in cell death or heritable changeshat are responsible for the late effects (such as cancer) in therogeny of the irradiated cells. However, numerous studies havehown that ionizing radiation is able to induce a state of genomicnstability in cultured cells. Radiation-induced genomic instabilityRIGI) is a concept used to describe the finding that delayed dam-ge (chromosomal aberrations, mutations, micronuclei, apoptosis)

s observed many cell generations later in the progeny of irradiatedells (for reviews, see [1–4]). This finding indicates that cells inheritot only direct mutations but an increased frequency of genetic

∗ Corresponding author. Tel.: +358 447163226.E-mail address: jukka.juutilainen@uef.fi (J. Juutilainen).

ttp://dx.doi.org/10.1016/j.mrfmmm.2014.04.010027-5107/© 2014 Elsevier B.V. All rights reserved.

damage. As the development of cancer requires accumulation ofmultiple genetic changes, RIGI is believed to be relevant to cancer[5–7], and also transgenerational RIGI in animals and humans hasbeen described [8,9]. The target size for RIGI is large, and the effectcannot be due to mutations in a single gene, or even a group of genes[1,3,10]. This indicates an epigenetic effect, but the mechanismsremain unclear [11].

The results of some studies suggest that the dose–response rela-tionship for RIGI is flat (with no or little dependence on dose)above a threshold [12], but the data are partly inconsistent [13].Time-dependent changes after irradiation may explain some of theinconsistencies, but evaluating this hypothesis is difficult, as onlya few studies have investigated dose–response at more than onepoint in time.

Studying gene expression changes in the exposed cells and

their progeny might be useful for understanding the biologicalprocesses that lead to RIGI. Immediate (up to 48 h after irradia-tion) transcriptional responses to radiation-induced DNA damagehave been investigated in many studies, and the findings naturally

tion Re

irgdsg(se1[akrwptsbttd

dRseewaariuedcitoslRa

2

2

iwbEfpUiCtmtyt[

K. Huumonen et al. / Muta

nclude changes in the expression of genes involved in DNA damageesponse [14–17]. However, limited data are available concerningene expression changes at delayed endpoints, corresponding toevelopment of RIGI. Mello et al. [18] reported changes in the tran-cription profile of human primary fibroblasts exposed to 0.5 Gy ofamma rays, but the analysis was done at only one point in time6 days). Two studies have analyzed gene expression changes ateveral time points after irradiation. Falt et al. [34] studied genexpression patterns in cultured human lymphocytes grown for 7,7 or 55 days after exposure to 3 Gy of gamma rays. Kruse et al.19] induced kidney damage in mice by in vivo exposure to X-rayst 16 Gy and performed microarray analysis of gene expression inidney tissue at 1, 10, 20 and 30 days after exposure. Both studieseported gene expression profiles that changed strongly over time,ith very few genes consistently up- or downregulated at all timeoints studied. Falt et al. [34] also reported a more diversified pat-ern of gene expression changes in the irradiated than control cells,uggesting that increased heterogeneity of gene expression mighte characteristic of RIGI. However, the study did not include simul-aneous measurement of any other endpoints than gene expressiono confirm RIGI by independent methods and to follow the co-evelopment of RIGI and gene expression changes.

The aim of the present study was to investigate theose–response for both direct effects (2 days after irradiation) andIGI assessed at two delayed points in time, under the hypothe-is that the dose–response changes with time. Furthermore, genexpression was analyzed at the same points in time to relate genexpression changes to the development of RIGI. Gene expressionas examined both to identify possible changes specific to RIGI

nd to test the hypothesis that the development of RIGI is associ-ted with increased heterogeneity of gene expression. A wide doseange from 0.2 to 5 Gy was used to include the high doses usedn many previous studies. Measurement of micronuclei (MN) wassed for monitoring the development of RIGI. Micronuclei are anasily detectable and widely used indicator of different types ofirect chromosomal injuries, but they have also been found to beonstantly elevated over multiple generations in the progeny ofrradiated cells [20] and they have been therefore used for detec-ing genomic instability induced by ionizing radiation as well asther exposures in a variety of cells types [21–23]. Gene expres-ion was measured by PCR arrays, which allow profiling only aimited number of genes but provide highly reproducible data. AsIGI is assumed to have a role in radiation carcinogenesis, a PCRrray profiling cancer-relevant genes was selected.

. Materials and methods

.1. Cell line and cell culture

Murine embryonic fibroblasts (C3H/10T½ clone 8) from Amer-can Type Culture Collection (Manassas, VA, USA) were used, as

e have previously shown chemically induced genomic insta-ility in the same cell line [24]. The cells were maintained inagle’s basal medium supplemented with 10% heat-inactivatedetal bovine serum, 2 mM l-glutamine and antibiotics (100 U/mlenicillin and 100 mg/ml streptomycin) (all from Gibco, Carlsbad,SA). The cells were grown at 37 ◦C with 5% CO2 and detached by

ncubating them for 5 min in 0.25% trypsin (Invitrogen, Carlsbad,A, USA) in 0.02% EDTA in PBS (w/o Ca2+, Mg2+), after which therypsin was inactivated by adding a double volume of fresh +37 ◦C

edium. The subcultured cell density was selected depending on

he X-ray dose, end-point to be analyzed and time point of the anal-sis, assuming a doubling time of 24 h (the actual doubling time ofhis cell line is approximately 16 h with some delay after the plating25]).

search 765 (2014) 32–39 33

In order to get 400,000 cells for the micronucleus assay con-ducted 2 days after the exposure, cells were seeded on 60 mm platesat a density of 2300 cells/cm2 24 h prior to the exposure. For thecells to be collected 8 days after the exposure, the following celldensities (based on previous viability experiments) were plated onT75 flasks 24 h before the exposure: 200 cells/cm2 (controls and0.2 Gy, 0.5 Gy and 1 Gy), 270 cells/cm2 (2 Gy), and 400 cells/cm2

(5 Gy). Six days after the exposure cells were counted again andseeded on plates at a density of 4700 cells/cm2. At this same timepoint plates with 1000 cells on each were seeded for the micro-nucleus analysis 15 days after the exposure.

For gene expression analysis 2 days after exposure,2700 cells/cm2 (control, and 1 Gy), or 4500 cells/cm2 (5 Gy)were plated on T75 flasks 24 h before the exposure. For the sameanalysis to be conducted 8 days after exposure, 330 cells/cm2

(controls, and 1 Gy) or 670 cells/cm2 (5 Gy) were plated on T75flasks 24 h prior exposure. 5 days after the exposure these cellswere subcultured on T75 flasks at the density of 2700 cells/cm2

(controls, and 1 Gy) or 4500 cells/cm2 (5 Gy). The cells to be ana-lyzed 15 days after exposure were also subcultured on T75 flasksat 5 days; the cell density was 330 cells/cm2 (controls, and 1 Gy) or530 cells/cm2 (5 Gy).

Viability and proliferation assay was performed using Alamar-Blue reagent (Bio-Source International, Inc., Camarillo, CA, USA)according to the manufacturer’s protocol. Briefly, cells were washedtwice with ice-cold PBS (phosphate buffered saline without Ca2+,Mg2+, Oy Reagena Ltd, Toivala, Finland), removed from plates byscraping in 1 ml of ice-cold PBS. Cells were counted and transferredin 450 �l of cell culture medium into 96-well plates (Nunc, Roskilde,Denmark) 24 h before irradiation at cell densities 2000 cells/well(0.5 Gy), 4000 cells/well (1 Gy), or 8000 cells/well (2 and 5 Gy). Con-trol well plates were made using all the cell densities used inthe exposures. Then 50 �l of AlamarBlue was added to each well.Culture medium was used as blank samples. Fluorescence was mea-sured immediately after the 96-well plates were ready, 1 h beforeexposure, 2 h after exposure, 12 h after exposure and once in every24 h for the 7 days following the exposure. The plates were exposedto an excitation wavelength of 540 nm and the emission at 595 nmwas recorded on a fluorometer (Perkin Elmer HTS 7000 Plus BioAssay Reader, Turku, Finland). The viability was expressed as a ratioof fluorescence emitted by treated cells compared to control cells.

2.2. Exposure

The cells were exposed to X-rays with a 4 MV Varian 600 C lin-ear accelerator (Varian, Palo Alto, CA, USA) at the Kuopio UniversityHospital. The cell cultures were placed on the treatment couchbetween two plexiglass sheets 1 cm above and 2 cm below the sam-ples. The control cultures were also taken into the hospital facilities.Doses of 0.2, 0.5, 1, 2 and 5 Gy were used. The maximum pho-ton energy was 4 MeV (average 1.5 MeV) and the dose rate about2.5 Gy/min.

2.3. Micronucleus assay

The frequency of micronuclei was measured by a flow cytomet-ric assay as described previously [26]. The principle of the assay isthat the nuclei of dying cells are stained with ethidium monoazidebromide (EMA), followed by lysis of the viable cells and staining oftheir nuclei with SYTOX Green dye. Finally, nuclei are assorted byflow cytometer.

In brief, medium was removed from the plates containing cell

cultures and the plates were incubated on ice for 20 min. Afterthe incubation, 1.5 ml of 8.5 �l/ml +4 ◦C EMA-solution (MolecularProbes, Eugene, USA) was added and the stain was light-activatedfor 30 min with a light bulb 15 cm above the plates (lids removed).

3 tion R

Apwba(hF10wfvr(GEtfDCpS

2

(gticf

Kc(RwwhaSdwSw

2

PCbcd

M

wacwvoon

4 K. Huumonen et al. / Muta

fter the light-activation of the stain the cell cultures wererotected from light during the whole analysis. Cell culturesere washed with 1.5 ml of cold (+4 ◦C) 2% FBS (Gibco, Carls-

ad, USA) in phosphate buffered saline (without Ca2+ and Mg2+),nd 500–700 �l (depending on visualized cell density) of cold+4 ◦C) lysis 1 solution (0.3 �l IGEPAL/ml (Sigma–Aldrich, Stein-eim, Germany), 0.584 mg NaCl/ml (FF Chemicals, Haukipudas,inland), 0.5 mg RNase A/ml (Sigma–Aldrich, Steinheim, Germany),

mg sodium citrate/ml (Riedel-de Haën, Seelze, Germany), and.4 �M SYTOX Green (Molecular Probes, Eugene, USA) in MilliQ-ater) was added following incubation of the plates at +37 ◦C

or 1 h. After the incubation, 500–700 �l (a volume equal to theolume of lysis 1 solution) of +20 ◦C lysis 2 solution (15 mg cit-ic acid/ml (Riedel-de Haën, Seelze, Germany), 85.6 mg sucrose/mlMP Biomedicals Inc., South Chillicothe, OH, USA), 0.4 �M SYTOXreen, and 1 drop of 6 �m fluorescent beads (Molecular Probes,ugene, USA)), was added and the plates were incubated at roomemperature for 30 min again. Finally, cell suspensions were trans-erred into flow cytometer tubes and analyzed with a Bectonickinson FACSCalibur flow cytometer (Becton Dickinson, San Jose,A) according to Bryce et al. [27]. A total of 2 × 105 gated nucleier sample were scored and the data were analyzed by CellQuestoftware (v.3.3 Becton Dickinson, San Jose, CA).

.4. Gene expression

The mouse Cancer Pathway FinderTM RT2 Profiler PCR ArraySABiosciences, a Qiagen Company) profiles the expression of 84enes representative of the six biological pathways involved inransformation and tumorigenesis. These functional gene groupsnclude cell cycle control and DNA damage repair, apoptosis andell senescence, signal transduction molecules and transcriptionactors, adhesion, angiogenesis, and invasion and metastasis.

RNA was isolated from the pelleted cells using RNeasy Miniit and RNAse free DNase (Qiagen, Hilden, Germany). The con-entration and purity of RNA were determined using Nano DropThermo Scientific). 1 �g of RNA was generated into cDNA usingT2 First Strand Kit (SABiosciences). cDNA samples were mixedith RT2 qPCR Master Mix (SABiosciences) and distributed in allells of the PCR array plate containing the 84 selected genes, fiveousekeeping genes, controls for RT reaction and PCR reaction and

genomic DNA control. Applied Biosystems 7000 Real-Time PCRystem (Applied Biosystems, Foster City, CA, USA) was used toetermine Ct-values of each well. The fold-changes in Ct-valuesere calculated using the web-based data analysis program from

ABiosciences. The average Ct-values of five housekeeping genesere used for normalization.

.5. Data analysis

Statistical analysis and modeling of the data were done with therism 5.0 for Windows software (GraphPad Software, San Diego,A, USA). Microsoft Office Excel 2007 was used for calculationsefore transferring the data into Prism. Time- and dose-dependenthanges in micronucleus frequency were modeled by fitting theata with an exponential decay function

N(t) = MN0 ∗ e−Kt + c, (1)

here MN0 is the initial frequency of micronuclei (at two daysfter irradiation), K is a decay constant (half-life is 0.69/K), and

is an asymptotic value that the micronucleus level approachesith time. The parameters K and c were assumed to have the same

alue independent of dose. This model assumes two components,ne following an exponential decay function with time, and thether being constant at all doses and points in time. The ratio-ale for using this model is described in Section 4. As the purpose

esearch 765 (2014) 32–39

was to model radiation-induced micronuclei without the constantbackground level occurring in all cells, radiation-related increase(irradiated minus controls) in micronucleus frequency was used asdata in the model. Therefore, a non-zero positive value of c indi-cates that the micronucleus frequency is higher in the irradiatedthan in the control cultures.

Gene expression data were analyzed by comparing the numberof changes by Fisher’s exact test (2 × 2 contingency tables) or chi-square test for trend (testing trends over time).

3. Results

Cell survival was reduced in a dose-dependent manner (about85% at 0.2 Gy, 65% at 5 Gy) in all irradiated cells at 48 h after expo-sure (data not shown), but it returned to normal or only slightlyreduced level (about 80% at 5 Gy) by 5 d after irradiation (by 4 d inthe 0.2 Gy group). The cells also grew as expected in the 8-d and 15-d cultures, indicating that there were no major long-term effects oncell viability or cell cycle duration.

The frequency of micronuclei in unexposed control cells wasgenerally close to 1% in all experiments (average 1.27, standarderror of mean 0.15). As there was some variation between experi-ments, all comparisons were done to within-experiment controls.A clear dose–response relationship was observed in micronucleusfrequency (Fig. 1). The increase of micronuclei with increasing dosewas statistically highly significant (p < 0.0001; linear regression) atall points in time, but there was a dramatic decrease in the slope ofthe curve from day 2 to day 15.

As the pattern of changes in the dose–response relationship sug-gested that there might be two different dose- and time-dependentcomponents, we decided to explore the changes of micronucleusfrequency with time at each dose. As shown in Fig. 2, the fre-quency of micronuclei remained persistently at an elevated level(compared to unexposed controls) at doses of 0.5 Gy or higher. At0.2 Gy, the frequency of micronuclei was clearly above the controllevel only at 2 days after irradiation. The 0.2 Gy data were thereforeexcluded when the micronucleus data were modeled (as describedin detail in Section 2) under the assumption that there are two com-ponents, one following an exponential decay function with time,and the other being constant at all doses and points in time. Thedata fitted well with the model, with an R2 value of 0.992 for thewhole model involving all doses from 0.5 to 5 Gy. The results ofthe model suggested a half-life of 2.3 days (95% confidence inter-val 1.8–3.2 d) for the decrease of micronucleus frequency. As canbe seen in Fig. 2, the micronucleus frequency has almost reachedthe asymptotic constant level (plateau) at 15 days in cells exposedto doses from 0.5 to 2 Gy, but is still elevated in cells exposed at5 Gy. The plateau level that fits with the data is 0.30 (95% confi-dence interval 0.13–0.48), i.e., the micronucleus frequency settlesat an about 0.3% higher level than the level (average 1.27%) foundin unexposed control cells.

The PCR array method used for gene expression analysis showedvery high reproducibility. The number of genes showing ≥2-foldchanges was 3 in two independent comparisons of unexposed con-trol samples and 2 in one such comparison. For ≥1.5-fold changes,the same comparisons yielded 8, 5 or 7 differentially expressedgenes. One gene (Tnf) showed ≥2-fold differences in all three com-parisons, and one (Serpinb2) in two of the three comparisons. Thesegenes had very low expression levels and the comparison of theirexpression is therefore inaccurate.

Gene expression analysis was performed at 1 and 5 Gy. In the

limited number of genes measured, irradiation at 1 Gy resultedin so few gene expression changes (not essentially higher thanthe number of differences observed in the comparisons of unex-posed controls) that no meaningful analysis of the data could be

K. Huumonen et al. / Mutation Re

Fig. 1. Dose dependency of micronucleus frequency at 2 days (A), 8 days (B) or 15days (C) after irradiation. Mean (±SEM) difference from control cultures is shown.N = 11 (A), 9 (B), or 2 (C).

Fig. 2. Changes of micronucleus frequency as a function of time. The points are mean(±SEM) differences from corresponding control cultures and the curves describe theexponential decay model (see Section 2) fitted with the data. Day 0 is the time offirst measurement of micronucleus frequency (two days after exposure). N = 11, 9,or 2 for 0 days, 6 days, and 13 days, respectively.

search 765 (2014) 32–39 35

performed. Data on up- and downregulated genes at 2, 8, and15 days after irradiation at 5 Gy is presented in SupplementaryTable 1. When data from two independent replicate experimentsperformed at 5 Gy were pooled, the number of ≥2-fold changesdecreased with time, with only three genes showing altered expres-sion at 15 days after exposure (Fig. 3A). A different picture arosewhen the two independent replicate experiments were analyzedseparately (Fig. 3B). The exposed cells continued to show dif-ferences from controls at 8 and 15 days, but the consistency ofthe changes decreased with time so that none of the changesobserved at 15 days were similar in the two experiments. Thisdiversifying pattern was further explored as shown in Fig. 4. Forthis analysis, a less stringent criterion for consistency was used:a gene expression difference was considered consistent, if a ≥2-fold difference was observed in one experiment and a ≥1.5-folddifference in the same direction in another experiment. This anal-ysis confirmed that the number of consistent changes decreasedstrongly with time (number of changes differed significantly fromcontrol–control comparisons only at 2 and 8 days; the trend withtime was significant at p < 0.0001). Furthermore, the data shownin Fig. 4 also revealed that the number of inconsistent changes(≥2-fold change in one experiment only) did not decrease or evenincreased with time (no statistically significant trend with time;number of changes differed from control–control comparisons at8 and 15 days). The differences between the two experimentstended to increase with time, and the number of genes showing≥2-fold inter-experiment differences was significantly higher thanin control–control comparisons at 15 days.

4. Discussion

Both the micronucleus data and the gene expression changesobserved in the present study indicate that the delayed effects ofionizing radiation keep evolving in a dynamical way until at least15 days after irradiation (about 22 cell generations with cell cycleduration of 16 h and an assumed 1-day post-irradiation cell cyclearrest).

Frequency of micronuclei as a function of time and dose hasbeen previously measured in V79 Chinese hamster cells exposedto X-rays [20]. Consistently with the findings of the present study,the frequency of micronuclei decreased with time post-irradiationand at 7 days seemed to approach a constant level which wasdifferent from unirradiated controls but independent of dose. How-ever, the lowest dose used in the long-term experiments by Jamaliand Trott was 4 Gy. Similar findings were reported for micronucleiin experiments with 1–3 Gy of �-particles [28] and for dicentricchromosomes in experiments with 3–12 Gy of X-rays [12]. Moth-ersill et al. [29] reported that the level of chromosomal aberrationsremained at a constant and dose-independent level between 8 and72 population doublings after irradiation in human HPV-G kera-tinocytes exposed to �-rays at 1–3 Gy or to �-particles at 0.25 or0.5. In cells �-irradiated at 0.5 Gy, a statistically significant effectpersisted only up to 30 population doublings. Jamali and Trott [12]also reported changes in dose–response similar to those found inthe present study: the frequency of apoptosis was dose-dependentbetween 1 and 12 Gy at 48 h after irradiation, but developed towarda flat dose–response relationship at 14 days after irradiation. Thedose-dependence for delayed micronuclei and for delayed dicentricchromosomes at 7 days showed a plateau above 3 Gy but dose-dependent increase below 3 Gy. The rising dose–response below3 Gy is not necessarily inconsistent with our observations, as the

follow-up time was only 7 days; our data also still showed cleardose dependency at 8 days, and suggested that development ofa fully flat dose–response relationship takes a much longer time.Schwartz et al. [30] analyzed dicentrics in human lymphoblastoid

36 K. Huumonen et al. / Mutation R

Fig. 3. Venn diagram showing alterations in expression of genes in the mouse Can-cerFinder PCR array at 2, 8 and 15 days after exposure to 5 Gy of X-rays. Genes thatwere ≥2.0-fold up- or downregulated (compared to control cultures) are shown.(A) Pooled data from two experiments, comparison of changes at different points intime. (B) Comparison of the two independent experiments.

esearch 765 (2014) 32–39

cells at about 35 generations after �-irradiation, and reported a flatdose–response above a threshold at 0.05 Gy. The different cell lines(with different doubling times) used in the experiments describedabove may have affected the time dependency of the developmentof the dose–response. Overall, however, the results are consistentwith the results of the present study, suggesting flat dose responseat sufficiently late points in time after exposure.

The model we used for analysing the time-dependent changeof micronucleus frequency was inspired by the Monte Carlosimulation results reported by Andreev and Eidelman [31] forradiation-induced delayed dicentrics, showing evolution of thedose dependence from linear-quadratic to a flat dose–responserelationship within 400 h from irradiation, and time-dependentchanges very similar to those shown in Fig. 2. Andreev and Eidel-man discussed their model (and experimental findings fitting withthe model predictions) in terms of mechanisms that lead to for-mation of dicentrics. As our data do not allow deep discussion onmolecular level mechanisms, we consider here a simple model thatassumes two different components for the delayed effects of radi-ation. One of the components (Phase 1) is associated with severecellular damage, which is of course more common at high doses.We do not assume any specific type of damage in these severelydamaged cells, but the progeny of these cells is assumed to dis-play many different delayed effects (such as micronuclei, apoptosis,chromosomal aberrations, gene mutations). The severely damagedcells may proliferate for a limited number of cell cycles, but areassumed to have reduced viability and therefore eventually dis-appear from the population. This component corresponds to theexponential decay term (e−Kt) in our equation. The second com-ponent (Phase 2) describes cells that have not been damaged toan extent that would reduce their viability, but have neverthelessgained an instability phenotype. These cells are the ones that areresponsible for the reported persistent damage at very long timesafter irradiation [29,32]. This component is assumed to display theflat dose–response relationship (threshold and a plateau above it)reported for RIGI in some studies ([12,30,33]), and corresponds inour model to the constant term (c). To illustrate how this modelfits with the observed dose–responses shown in Fig. 1, theoreticaldose–response curves predicted by our model are shown in Fig. 5for different points in time. The first component showing a typicalrising dose–response dominates shortly after exposure (at 2 days),and the total increase in micronuclei is almost entirely explained bythis component. At 8 days, the first component has become weaker(corresponding to the exponential decay term in our equation), andthis component dominates only at high doses. At 15 days, the sec-ond component dominates, and the total increase of micronucleishows a nearly flat dose–response, with residual contribution ofthe first component only at high doses.

It should be noted that the exponential decay term in the equa-tion only means that the proportion of cells with micronucleidecreases – it does not mean that the total number of cells withmicronuclei would decrease with time. As no major effect on cellsurvival was observed and the surviving cells continue prolifer-ating, the number of cell divisions after the day 2 measurementscorresponds to at least a 200-fold increase in cell number at 8days and a 100,000-fold increase at 15 days, under the conser-vative assumption that cell survival is only 80% during the whole15-day period. A simple calculation from these cell numbers and theradiation-related micronucleus frequencies observed at the threetime points shows that, for example in cells irradiated at 5 Gy,the total number of radiation-related micronuclei has increasedabout 23-fold at 8 days and about 2500-fold at 15 days. It is there-

fore obvious that the observed micronucleus frequencies cannotbe explained by micronuclei that were directly induced by radia-tion, even if they all are (unrealistically) assumed to persist duringthe 15 days of cultivation; formation of new micronuclei must

K. Huumonen et al. / Mutation Research 765 (2014) 32–39 37

Fig. 4. Time dependence of gene expression differences. The left y-axis is for the number of consistent changes (the same difference from controls is observed in twoindependent experiments, a ≥2-fold difference in one experiment and a ≥1.5 difference in the same direction in the other experiment), number of inconsistent changes (a≥ ) betwb was sc

haedhoriadme

diwiwto

adMiivsbteaeAofihs

2-fold change in only one of the experiments) and number of differences (≥2-foldetween the two experiments. Asterisks indicate that the number of differencesontrol–control comparisons: ***p < 0.001; **p < 0.01; *p < 0.05.

ave occurred in the about 20 generations of cells between day 2nd day 15. Thus, also the first, decaying component of delayedffects fulfills the definition of RIGI, i.e., de novo appearance ofelayed damage in the progeny of irradiated cells. This component,owever, disappears after sufficient number of cell divisions, andnly the second, more persistent component of genomic instabilityemains. The two-component nature of RIGI may be one reason fornconsistent observations concerning the dose–response of RIGI,s different studies have assessed dose–response relationship atifferent points in time, and the relative duration of Phase 1 RIGIay also vary depending on cell line, type of radiation and other

xperimental variables.It should be noted that the present study was not specifically

esigned to test the existence of two components in RIGI. After see-ng the dose and time dependency of the micronucleus data (Fig. 1),

e generated the hypothesis of two components and formulatedt mathematically (Equation 1). As the model was found to fit well

ith the present data, we wish to put forward the hypothesis ofwo components, and suggest that it is independently tested withther data.

As expected, the gene expression changes observed at 2 daysfter irradiation included changes in genes associated with DNAamage response, cell cycle control and apoptosis (Ccnd1, Cdkn1a,dm2, Bcl2l1, Fas, Tnfrs10b). The findings of the two latter points

n time are consistent with earlier studies reporting that post-rradiation gene expression profiles change strongly over time, withery few genes consistently up- or downregulated at all time pointstudied [19,34]. As increased heterogeneity of gene expression haseen previously described in human lymphocytes cultured for mul-iple generations after irradiation [34] and in the progeny of C.legans nematodes tens of generations after irradiation [35], wessessed such heterogeneity in the present data by evaluating genexpression differences between two independent experiments.lthough we measured the expression of only a limited number

f cancer-relevant genes, our results are consistent with the earlierndings, and add to the evidence that non-specificity and increasedeterogeneity are characteristic to gene expression changes mea-ured multiple cell generations after irradiation. The present study

een the two experiments. The right y-axis is for geometric mean of fold differencestatistically significantly higher than the corresponding number of differences in

also included micronucleus measurements to detect RIGI, and wastherefore able to show that existence of RIGI at 15 days after expo-sure was not associated with consistent changes in gene expression.The diversifying pattern of gene expression changes during 15 daysafter irradiation might be related to the observation that diversify-ing changes observable also by cytogenetic assays continue to occurtens of generations after irradiation [32].

The gene expression data also support the two-componentinterpretation of the micronucleus data. Phase 1 RIGI is char-acterized by high but decreasing level of micronuclei, and isassociated with gene expression changes that are relatively con-sistent between experiments and include typical responses to DNAdamage. Phase 2 RIGI, on the other hand, is characterized bymoderate but persistently elevated level of micronuclei and is asso-ciated with gene expression changes that are apparently stochasticand exhibit increasing heterogeneity. This indicates that differentmechanisms may be responsible for the two different componentsof RIGI.

As stated above, we assume that Phase 1 RIGI is associatedwith severe cellular damage. According to classical radiobiology,the probability of severe damage in a “target” (such as, e.g., DNAdouble-strand breaks in the genome of a cell) increases with num-ber of hits (ionization events) in the target, so the proportion ofcells with Phase 1 RIGI should increase with increasing dose, con-sistently with our data. On the other hand, the flat dose–responserelationship seen in this and other studies at delayed time pointssuggests that Phase 2 RIGI is an all-or-nothing phenomenon. Thiskind of dose–response relationship is more difficult to explain bythe classical radiobiology approach. In a recent study, we stud-ied genomic instability induced by a non-genotoxic carcinogen,the dioxin model compound 2,3,7,8-tetrachlorodibenzo-p-dioxin(TCDD) [24]. Delayed increase of micronuclei was observed at 8days after exposure, with a flat dose–response relationship at theTCDD concentrations used (1, 10 and 100 nM). The similarity of the

TCDD-induced effect and Phase 2 RIGI suggests a common basicmechanism. As TCDD caused no genotoxicity directly after expo-sure, it seems likely that this common mechanism does not requiredirect damage to DNA.

38 K. Huumonen et al. / Mutation R

Fig. 5. Theoretical dose–response curves for different times after irradiation (2, 8and 15 days) as predicted by the two-component model of RIGI. The second com-ponent was assumed to exist also at 0.2 Gy, but it was assumed to be weaker thana

aosrltta

Cmdapdpttroem

[

[

[

[

[

[

[

[

[

on transcription profiles in human primary fibroblasts, Environ. Mol. Mutagen.52 (2) (2011) 117–129.

t higher doses.

The persistent Phase 2 RIGI is obviously relevant for long-termdverse effects of radiation. The flat dose–response relationshipbserved for this effect indicates that it may be necessary to recon-ider the relative importance of low doses in assessing the healthisks of radiation, particularly if the threshold for Phase 2 RIGI isow. As the present study included experiments with only one cellype, data from studies with other experimental models (other cellypes, in vivo models) will be needed to determine the thresholdnd other characteristics of the dose–response.

In conclusion, our study on dose-and time-dependent effects in3H/10T½ cells showed that X-ray induced delayed effects on bothicronucleus frequency and gene expression keep evolving in a

ynamical way until at least 15 days after irradiation. We propose hypothesis that there are two components in RIGI. The first com-onent consists of a dose-dependent increase of delayed geneticamage (measured as micronuclei in the present study) that mayersist more than ten generations depending on dose, but even-ually disappears. This component is associated with changes inhe expression of specific genes (e.g., genes regulating DNA damageesponses). The second component is more persistent, independentf dose above a threshold and associated with non-specific gene

xpression changes showing increased variability. A mathematicalodel based on this hypothesis fitted well with the micronucleus

[

esearch 765 (2014) 32–39

data. However, alternative explanations are possible, and thehypothesis should be independently tested in further studies.

Conflict of interest statement

The authors declare that there are no conflicts of interests.

Acknowledgements

We thank Arja Moilanen and Hanne Säppi for their excellenttechnical assistance. This work was supported by the Academy ofFinland (grant number 123439), Finnish Ministry of Education andCulture (Doctoral Programme in Environmental Health) and TheFinnish Cultural Foundation. The funding bodies did not take partin the study design or other phases of the study.

Appendix A. Supplementary data

Supplementary material related to this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.mrfmmm.2014.04.010.

References

[1] W.F. Morgan, J.P. Day, M.I. Kaplan, E.M. McGhee, C.L. Limoli, Genomic instabilityinduced by ionizing radiation, Radiat. Res. 146 (3) (1996) 247–258.

[2] J.B. Little, Radiation-induced genomic instability, Int. J. Radiat. Biol. 74 (6)(1998) 663–671.

[3] K. Baverstock, Radiation-induced genomic instability: a paradigm-breakingphenomenon and its relevance to environmentally induced cancer, Mutat. Res.454 (1–2) (2000) 89–109.

[4] C. Mothersill, C. Seymour, Changing paradigms in radiobiology, Mutat. Res. 750(2012) 85–95.

[5] R.L. Ullrich, B. Ponnaiya, Radiation-induced instability and its relation to radi-ation carcinogenesis, Int. J. Radiat. Biol. 74 (6) (1998) 747–754.

[6] S. Negrini, V.G. Gorgoulis, T.D. Halazonetis, Genomic instability—an evolvinghallmark of cancer, Nat. Rev. Mol. Cell Biol. 11 (3) (2010) 220–228.

[7] C. Streffer, Strong association between cancer and genomic instability, Radiat.Environ. Biophys. 49 (2) (2010) 125–131.

[8] R.C. Barber, R.J. Hardwick, M.E. Shanks, C.D. Glen, S.K. Mughal, M. Voutounou,Y.E. Dubrova, The effects of in utero irradiation on mutation induction andtransgenerational instability in mice, Mutat. Res. 664 (1–2) (2009) 6–12.

[9] A. Aghajanyan, N. Kuzmina, A. Sipyagyna, L. Baleva, I. Suskov, Analysis ofgenomic instability in the offspring of fathers exposed to low doses of ionizingradiation, Environ. Mol. Mutagen. 52 (7) (2011) 538–546.

10] K.R. Trott, M. Jamali, L. Manti, A. Teibe, Manifestations and mechanisms ofradiation-induced genomic instability in V-79 Chinese hamster cells, Int. J.Radiat. Biol. 74 (6) (1998) 787–791.

11] O. Kovalchuk, J.E. Baulch, Epigenetic changes and nontargeted radiationeffects—is there a link? Environ. Mol. Mutagen. 49 (1) (2008) 16–25.

12] M. Jamali, K.R. Trott, Persistent increase in the rates of apoptosis and dicentricchromosomes in surviving V79 cells after X-irradiation, Int. J. Radiat. Biol. 70(6) (1996) 705–709.

13] L.E. Smith, S. Nagar, G.J. Kim, W.F. Morgan, Radiation-induced genomic insta-bility: radiation quality and dose response, Health Phys. 85 (1) (2003) 23–29.

14] A.R. Snyder, W.F. Morgan, Gene expression profiling after irradiation: clues tounderstanding acute and persistent responses? Cancer Metastasis Rev. 23 (3–4)(2004) 259–268.

15] A.L. Fachin, S.S. Mello, P. Sandrin-Garcia, C.M. Junta, E.A. Donadi, G.A. Passos,E.T. Sakamoto-Hojo, Gene expression profiles in human lymphocytes irradiatedin vitro with low doses of gamma rays, Radiat. Res. 168 (6) (2007) 650–665.

16] Y.W. Jin, Y.J. Na, Y.J. Lee, S. An, J.E. Lee, M. Jung, H. Kim, S.Y. Nam, C.S. Kim,K.H. Yang, S.U. Kim, W.K. Kim, W.Y. Park, K.Y. Yoo, J.H. Kim, Comprehensiveanalysis of time- and dose-dependent patterns of gene expression in a humanmesenchymal stem cell line exposed to low-dose ionizing radiation, Oncol. Rep.19 (1) (2008) 135–144.

17] K.H. Kim, H.Y. Yoo, K.M. Joo, Y. Jung, J. Jin, Y. Kim, S.J. Yoon, S.H. Choi, H.J. Seol,W.Y. Park, H. Nam do, Time-course analysis of DNA damage response-relatedgenes after in vitro radiation in H460 and H1229 lung cancer cell lines, Exp.Mol. Med. 43 (7) (2011) 419–426.

18] S.S. Mello, A.L. Fachin, C.M. Junta, P. Sandrin-Garcia, E.A. Donadi, G.A. Passos,E.T. Sakamoto-Hojo, Delayed effects of exposure to a moderate radiation dose

19] J.J. Kruse, J.A. te Poele, A. Velds, R.M. Kerkhoven, L.J. Boersma, N.S. Russell, F.A.Stewart, Identification of differentially expressed genes in mouse kidney afterirradiation using microarray analysis, Radiat. Res. 161 (1) (2004) 28–38.

tion Re

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

K. Huumonen et al. / Muta

20] M. Jamali, K.R. Trott, Increased micronucleus frequency in the progenyof irradiated Chinese hamster cells, Int. J. Radiat. Biol. 69 (3) (1996)301–307.

21] A. Seoane, A. Guerci, F. Dulout, Genetic instability induced by low doses ofX-rays in hamster cells, Int. J. Radiat. Biol. 83 (2) (2007) 81–87.

22] Y. Yokota, T. Funayama, Y. Hase, N. Hamada, Y. Kobayashi, A. Tanaka, I. Narumi,Enhanced micronucleus formation in the descendants of gamma-ray-irradiatedtobacco cells: evidence for radiation-induced genomic instability in plant cells,Mutat. Res. 691 (1–2) (2010) 41–46.

23] G. Sciandrello, M. Mauro, I. Catanzaro, M. Saverini, F. Caradonna, G. Barbata,Long-lasting genomic instability following arsenite exposure in mammaliancells: the role of reactive oxygen species, Environ. Mol. Mutagen. 52 (7) (2011)562–568.

24] M. Korkalainen, K. Huumonen, J. Naarala, M. Viluksela, J. Juutilainen, Dioxininduces genomic instability in mouse embryonic fibroblasts, PLoS ONE 7 (5)(2012) e37895.

25] C.A. Reznikoff, D.W. Brankow, C. Heidelberger, Establishment and characteri-zation of a cloned line of C3H mouse embryo cells sensitive to postconfluenceinhibition of division, Cancer Res. 33 (12) (1973) 3231–3238.

26] J. Luukkonen, A. Liimatainen, A. Hoyto, J. Juutilainen, J. Naarala, Pre-exposure to

50 Hz magnetic fields modifies menadione-induced genotoxic effects in humanSH-SY5Y neuroblastoma cells, PLoS ONE 6 (3) (2011) e18021.

27] S.M. Bryce, J.C. Bemis, S.L. Avlasevich, S.D. Dertinger, In vitro micronucleus assayscored by flow cytometry provides a comprehensive evaluation of cytogeneticdamage and cytotoxicity, Mutat. Res. 630 (1–2) (2007) 78–91.

[

search 765 (2014) 32–39 39

28] L. Manti, M. Jamali, K.M. Prise, B.D. Michael, K.R. Trott, Genomic instability inChinese hamster cells after exposure to X rays or alpha particles of differentmean linear energy transfer, Radiat. Res. 147 (1) (1997) 22–28.

29] C. Mothersill, M.A. Kadhim, S. O’Reilly, D. Papworth, S.J. Marsden, C.B. Sey-mour, E.G. Wright, Dose- and time-response relationships for lethal mutationsand chromosomal instability induced by ionizing radiation in an immortalizedhuman keratinocyte cell line, Int. J. Radiat. Biol. 76 (6) (2000) 799–806.

30] J.L. Schwartz, R. Jordan, H.H. Evans, M. Lenarczyk, H. Liber, The TP53 dependenceof radiation-induced chromosome instability in human lymphoblastoid cells,Radiat. Res. 159 (6) (2003) 730–736.

31] S.G. Andreev, Y.A. Eidelman, Dose–response prediction for radiation-inducedchromosomal instability, Radiat. Prot. Dosim. 143 (2–4) (2011) 270–273.

32] B. Ponnaiya, C.L. Limoli, J. Corcoran, M.I. Kaplan, A. Hartmann, W.F. Morgan,The evolution of chromosomal instability in Chinese hamster cells: a changingpicture? Int. J. Radiat. Biol. 74 (6) (1998) 765–770.

33] J.B. Little, H. Nagasawa, T. Pfenning, H. Vetrovs, Radiation-induced genomicinstability: delayed mutagenic and cytogenetic effects of X rays and alphaparticles, Radiat. Res. 148 (4) (1997) 299–307.

34] S. Falt, K. Holmberg, B. Lambert, A. Wennborg, Long-term global gene expres-sion patterns in irradiated human lymphocytes, Carcinogenesis 24 (11) (2003)

1837–1845.

35] K. Huumonen, H.K. Immonen, K. Baverstock, M. Hiltunen, M. Korkalainen,T. Lahtinen, J. Parviainen, M. Viluksela, G. Wong, J. Naarala, J. Juutilainen,Radiation-induced genomic instability in Caenorhabditis elegans, Mutat. Res.748 (1–2) (2012) 36–41.