The FASEB Journal • Research Communication

Cigarette smoke-induced transgenerational alterationsin genome stability in cord blood of humanF1 offspring

Julian Laubenthal,* Olga Zlobinskaya,†,1 Krzysztof Poterlowicz,‡,1 Adolf Baumgartner,*,§

Michal R. Gdula,� Eleni Fthenou,¶ Maria Keramarou,¶ Sarah J. Hepworth,#

Jos C. S. Kleinjans,** Frederik-Jan van Schooten,†† Gunnar Brunborg,‡‡

Roger W. Godschalk,†† Thomas E. Schmid,† and Diana Anderson*,2

*Division of Biomedical Sciences and ‡Centre of Skin Sciences, School of Life Sciences, University ofBradford, Bradford, UK; †Department of Radiation Therapy, Klinikum Rechts der Isar, MunichTechnical University, Munich, Germany; §Department of Paediatric Cardiology, Cardiac Centre,University of Leipzig, Leipzig, Germany; �Department of Dermatology, School of Medicine, Universityof Boston, Boston, Massachusetts, USA; ¶Department of Social Medicine, School of Medicine,University of Crete, Crete, Greece; #Division of Epidemiology, Leeds Institute of Genetics, Healthand Therapeutics, University of Leeds, Leeds, UK; **Department of Toxicogenomics and††Department of Toxicology, University of Maastricht, Maastricht, The Netherlands; and‡‡Department of Chemical Toxicology, Division of Environmental Medicine, Norwegian Institute ofPublic Health, Oslo, Norway

ABSTRACT The relevance of preconceptional andprenatal toxicant exposures for genomic stability inoffspring is difficult to analyze in human populations,because gestational exposures usually cannot be sepa-rated from preconceptional exposures. To analyze theroles of exposures during gestation and conception ongenomic stability in the offspring, stability was assessedvia the Comet assay and highly sensitive, semiauto-mated confocal laser scans of �H2AX foci in cord,maternal, and paternal blood as well as spermatozoafrom 39 families in Crete, Greece, and the UnitedKingdom. With use of multivariate linear regressionanalysis with backward selection, preconceptional pa-ternal smoking (% tail DNA: P>0.032; �H2AX foci:P>0.018) and gestational maternal (% tail DNA:P>0.033) smoking were found to statistically signifi-cantly predict DNA damage in the cord blood of F1offspring. Maternal passive smoke exposure was notidentified as a predictor of DNA damage in cord blood,indicating that the effect of paternal smoking may betransmitted via the spermatozoal genome. Taken to-gether, these studies reveal a role for cigarette smoke inthe induction of DNA alterations in human F1 offspringvia exposures of the fetus in utero or the paternalgermline. Moreover, the identification of transgenera-tional DNA alterations in the unexposed F1 offspringof smoking-exposed fathers supports the claim that

cigarette smoke is a human germ cell mutagen.—Laubenthal, J., Zlobinskaya, O., Poterlowicz, K., Baum-gartner, A., Gdula, M. R., Fthenou, E., Keramarou, M.,Hepworth, S. J., Kleinjans, J. C. S., van Schooten, F.-J.,Brunborg, G., Godschalk, R. W., Schmid, T. E., Ander-son, D. Cigarette smoke-induced transgenerational al-terations in genome stability in cord blood of human F1offspring. FASEB J. 26, 3946–3956 (2012). www.fasebj.org

Key Words: toxicants � DNA damage � male germline �transmission � germ cell mutagen

Individual phenotype and one’s susceptibility todisease are the combined result of genotype and itsinteractions with the environment, particularly duringthe very early stages of development. Therefore, threecritical windows of exposure have been proposed whenthe offspring are predominantly susceptible to a givenexposure, depending on the pattern of exposure andchemical properties of the toxicant: preconceptionalexposure of maternal and paternal gametes; prenatal(in utero) exposure of the fetus through the mother;and early postnatal exposures of the newborn throughthe early environment and mother during lactation(1–3). The two latter windows are increasingly beinginvestigated, as various mother-child birth cohort stud-ies have shown that the individual phenotype and one’s

1 These authors contributed equally to this work.2 Correspondence: School of Life Sciences, University of

Bradford, Richmond Rd., Bradford BD7 1DP, UK. E-mail:d.anderson1@bradford.ac.uk

doi: 10.1096/fj.11-201194

Abbreviations: BMI, body mass index; BSA, bovine serumalbumin; DAPI, 4=,6-diamidino-2-phenylindole; ESTR, ex-panded simple tandem repeat; ETS, environmental tobaccosmoke; �H2AX, histone H2AX phosphorylation on serine139; LSCM, laser scanning confocal microscopy; PBS, phos-phate-buffered saline

3946 0892-6638/12/0026-3946 © FASEB

susceptibility to disease in childhood can be signifi-cantly influenced by exposures to chemicals, environ-mental contaminants, and lifestyle and demographicfactors during gestation and/or shortly after birth.Mechanistically, adaptive responses of the fetus toexposures, resulting in genetic (e.g., DNA damage andgenomic stability) and/or epigenetic changes (e.g.,DNA methylation), which may subsequently lead totransformed physiological and metabolic conditions,have been suggested as a molecular basis (4, 5). On thecontrary, the first window, i.e., the possible role ofpreconceptional exposures in the induction of trans-generational genomic instability and DNA damage inthe offspring, is largely neglected in population studiesdespite a significant theoretical basis corroborated bycompelling data from rodent models (6–10).

Exposures to chemotherapy and irradiation, as wellas to environmental, occupational, and lifestyle toxi-cants, constitute major sources of mutations and DNAdamage that threaten the integrity of the genome.There is ample evidence that these factors can alsoresult in genetic or developmental defects in the unex-posed offspring via exposures of male or female germcells before conception or in utero during gestation(11–14). The ability of these factors to induce transgen-erational effects requires stable chromosomal altera-tions or epigenetic modifications of the DNA. However,these effects have only been demonstrated in rodentmodels, whereas direct evidence for the existence oftransgenerational DNA damage or a germline mutagenin humans has not been confirmed yet. Moreover, thelimitation in knowledge of transgenerational effects torodent studies only dampens the significance for publichealth policy makers as long as no striking evidence forthe existence of transgenerational effects in humans isgiven (7, 12, 15, 16). In laboratory experiments, onlythe male or female rodent is usually exposed to a singlemutagen at a certain concentration and period of time,making subsequent analyses of transgenerational ef-fects in the offspring feasible. In comparison, analysesof the relevance of preconceptional and/or gestationalexposures for the F1 offspring in human populationsare significantly more challenging and can only bereconstructed via questionnaires. Moreover, gestationalexposures usually cannot be separated from preconcep-tional exposures in F1 offspring, since different expo-sures occur at the same time in unknown concentra-tions, combinations, and periods of time, and variationsof demographical factors such as gender, age, race, andlifestyle exist (7, 17).

Smokers are considered a highly relevant example ofan exposed cohort in human population studies mainlybecause 35% of men and 30% of women (in the UnitedStates) are smokers of reproductive age, and cigarettesmoke is significantly correlated with adverse reproduc-tive outcomes, such as stillbirth, spontaneous abortion,birth defects, and severe genetic diseases (18, 19). Inaddition, the International Agency for Research on Can-cer correlated, for the first time, preconceptional mater-nal and paternal tobacco smoke exposure with an in-

creased risk for development of childhood cancer in theunexposed offspring (20). This is in agreement withexperimental data showing that in spermatozoa of smok-ers, increases in oxidative damage, DNA strand breaks,DNA adducts, chromosomal aberrations, and mutationsare consistently found (21, 22). Consequently, cigarettesmoke was recently considered to be the first identifiedgerm cell mutagen (7, 23).

In this context, the European Union FrameworkProgramme Newborns and Genotoxic Exposure Risk(NewGeneris) examined the possible roles of expo-sures to environmental and lifestyle toxicants duringgestation and analogous exposures at the time ofconception in the induction of adverse events in off-spring (5). Therefore, the present study investigatedtransgenerational alterations in DNA damage assessedvia the Comet assay for the detection of primarilysingle-strand breaks; double-strand breaks were de-tected by highly sensitive, semiautomated confocal laserscans of histone H2AX phosphorylation on serine 139(�H2AX) foci in cord, maternal, and paternal blood aswell as spermatozoa from 39 families of two well-documented NewGeneris cohorts (Rhea cohort, Crete,Greece, and cesarean-section born in Bradford cohort,UK). Paternal, maternal, and newborn lifestyle anddemographic factors that were significant predictorsfor the two different DNA damage biomarkers wereidentified by using multivariate linear regression anal-ysis with backward selection.

MATERIALS AND METHODS

Study population

All participants filled in a structured questionnaire, which wasused along with medical records to obtain data on pregesta-tional body mass index (BMI), age, origin, lifestyle duringpregnancy [smoking habits, alcohol intake, environmentaltobacco smoke (ETS) exposure, and supplement intake],delivery type, newborn length and head circumference, birthweight, newborn sex, and gestational age. The gestational agewas estimated using the time point of the last menstrualperiod reported at recruitment and was confirmed by the firstultrasound examination. For women with �7 d differencebetween reported last menstrual period and ultrasound-basedestimation, crown-rump length was used to estimate thegestational age. Written, informed consent was given by allsubjects after a complete description of the study (24, 25).Ethical approval had been granted by the Leeds East Re-search Ethics Committee (NewGeneris Work Package 3 Pa-ternal Impact, ref. no. 07/H1306/15); the University ofBradford Subcommittee on Ethics in Research InvolvingHuman Subjects, (0405/8); the Bradford Research EthicsCommittee Growing Up in Bradford (05/Q1202/54); andthe Ethical Committee of the University Hospital of Crete(Heraklion, Greece; 96/8-1-07).

Sample collection

Umbilical cord blood from 39 newborns and peripheralblood from their mothers were obtained in heparinizedVacutainers (BD, Plymouth, UK) by venipuncture at the

3947SMOKING-INDUCED TRANSGENERATIONAL INSTABILITY

Bradford Royal Infirmary (n�15) and the University of CreteMedical School (n�24) directly after delivery. Newborn cordblood was additionally supplemented with 0.5 ml of hepa-rin (Leo Pharmaceutical Products, Copenhagen, Den-mark) to avoid clotting. Lymphocytes were isolated fromperipheral blood using a Lymphoprep kit (Axis-Shield,Oslo, Norway), transferred to cryovials containing FBS/DMSO (9:1), and frozen at �20°C overnight and then at�80°C for storage. For the 39 fathers of the newborns,peripheral blood and semen samples were obtained at thetime after conception to delivery from 15 fathers, andperipheral blood only was obtained from the remaining 24fathers. Semen samples were obtained after 3 d of sexualabstinence and analyzed within 2 h for semen qualityparameters according to the World Health Organizationguidelines (26). Urine samples were portioned into ali-quots and stored at �20°C until analysis.

Cotinine analysis

[CD3]cotinine (500 ng; Sigma-Aldrich, St. Louis, MO, USA)was used as an internal standard added to urine (0.2 ml fromactive smokers and 1 ml from nonsmokers). The volume ofurine from the active smokers was adjusted to 1 ml, mixedwith 1 ml of 50% aqueous K2CO3, and then extracted with 2ml of CH2Cl2. The CH2Cl2 layer was separated and mixedwith 200 �l of CH3OH, concentrated under a stream of N2 toa total of 100–200 �l of CH3OH, and finally analyzed by gaschromatography-mass spectrometry-selected ion monitoring.The gas chromatograph was equipped with a Merlin septumand an HP-5 fused silica capillary column (30 m, 0.25-mminner diameter, and 0.25-�m film thickness). The injectionport temperature was 250°C, and the injection mode wassplitless. The oven temperature was 70°C for 0.5 min and thenincreased to 180°C at 10°C/min, held for 1 min, increased to275°C at 50°C/min, held for 5 min, and returned to initialconditions. The carrier gas was helium at a flow rate of 1ml/min. The mass spectrometry transfer line was heated at300°C. The retention time of cotinine was 12.78 min. Theinternal standards eluted 0.02 min before the analytes. Limitsof quantification for total cotinine were 0.5 ng/ml, whereasall samples were above the level of quantification here. Dataon precision (cumulative variance �10%) and accuracy ofthese assays were described earlier (27–29).

Cotinine in semen plasma was quantified using a commer-cially available enzyme-based immunoassay kit (Calbiotech,Spring Valley, CA, USA) in accordance with the manufacturer’sinstructions and modifications by Hammadeh et al. (30). Inbrief, semen plasma, standards, and controls along withenzyme conjugate were pipetted into the supplied 96-wellplate, incubated for 1 h at room temperature and washed withdouble-distilled H2O, and then substrate reagent was added.After a further incubation for 30 min at room temperature,stop solution was added, the absorbance was read at 450 nm,and cotinine concentration (ng/ml) was calculated againstthe standard curve.

Comet assay

The alkaline Comet assay protocol is based on Singh et al.(31), with modifications as provided by Sipinen et al. (32). Inbrief, samples in cryovials were thawed for 10 s at 37°C,centrifuged for 10 min at 450 g at room temperature,resuspended in cold phosphate-buffered saline (PBS), andkept on ice until mixed 1:1 with 2% low melting pointagarose. The samples were subsequently cast onto standardglass slides precoated with 1% normal melting point agarose.Cell lysis was achieved through submerging the slides in lysis

buffer (2.5 M NaCl, 100 mM Na2EDTA, 10 mM Trizma base,and 1% Triton X-100, pH 10) for 60 min at 4°C. After lysis,the slides were incubated in electrophoresis buffer (300 mMNaOH and 1 mM Na2EDTA, pH 13.2) at 4°C for 20 min tounwind DNA and then subjected to electrophoresis at 25 V(0.8 V/cm on platform, 300 mA) for 30 min at 4°C. Beforestaining with ethidium bromide (20 �g/ml) for 10 min, theslides were neutralized (0.4 M Trizma base, pH 7.5) 3 timesfor 5 min each. For the Comet assay on spermatozoa, lysis wasachieved through submerging the slides in 2 consecutive lysissolutions: lysis buffer containing 10 mM DTT for 60 min at4°C; and lysis buffer containing 0.05 mg/ml proteinase K for60 min at 4°C, and then the cells were electrophoresed for 30min at 4°C. Comet visualization and scoring were performedwith an epifluorescence microscope equipped with a charge-coupled device camera from Leica (Wetzlar, Germany) usingthe image analysis software Komet 6.0 (Andor, Belfast, UK).For each individual sample, 3 independent experiments wereperformed, and for each experiment, 50 cells were scoredblindly. To make the data more comparable among Cometassays performed on different days, lymphocytes and sperma-tozoa exposed to 50 �M H2O2 for 1 h at 37 and 32°C,respectively, were used as an internal standard, and the datawere corrected relative to this standard. All chemicals werepurchased from Sigma-Aldrich (Gillingham, Dorset, UK),unless otherwise stated.

Immunostaining

Immunostaining for �H2AX foci was based on Schmid et al.(33). In brief, the lymphocytes were fixed in 4% ice-coldparaformaldehyde dissolved in PBS for 10 min at 4°C. Cellswere subsequently washed twice with cold PBS and centri-fuged for 10 min at 450 g at room temperature. Cells weresubsequently loaded onto a Cytospin (Thermo Fisher Sci-entific, Walldorf, Germany) and centrifuged at 200 g for 2min onto the slide. Cell lysis was achieved through sub-merging the slides in lysis buffer [0.15% Triton X-100 and4% bovine serum albumin (BSA) in PBS] for 10 min. Afterlysis, slides were washed 3 times with PBS and submerged inblocking buffer (4% BSA in PBS) for 15 min. Then, �H2AXantibody (1:400 in blocking buffer; Abcam, Cambridge,UK) was applied, and slides were incubated for 16 h at 4°C.Slides were subsequently washed three times for 5 min eachwith PBS, followed by incubation with Alexa 488 (1:400 inblocking buffer; Invitrogen, Carlsbad, CA, USA) secondaryantibody for 1 h at room temperature. Finally, unbound anti-body was washed off with PBS, and the cells were embedded in4=,6-diamidino-2-phenylindole (DAPI)/Vectashield (VectorLaboratories, Burlingame, CA, USA) solution. To controlthe specificity of the primary antibody, lymphocytes werestained with the secondary Alexa 488 antibody, but not withthe primary �H2AX antibody. Furthermore, lymphocytesexposed to 50 �M H2O2 for 1 h at 37°C were used as astandard sample to make the data comparable amongimmunostaining experiments performed on different days.All chemicals were purchased from Sigma-Aldrich, unlessotherwise stated.

Microscopy and image analysis

For each individual, �H2AX foci immunolocalized in110�150 cells were acquired using a Zeiss LSM 510 confocallaser scanning microscope with a Plan Neofluar �63 objective(Carl Zeiss, Jena, Germany). Z stacks at random positions ofthe slide were taken for each sample in 20–25 slices/stack.Data were analyzed on a per-nucleus basis using ImageJsoftware (U.S. National Institutes of Health, Bethesda, MD,

3948 Vol. 26 October 2012 LAUBENTHAL ET AL.The FASEB Journal � www.fasebj.org

USA) with a custom software macro, FociCount, to countparticles. Results of the analysis are presented as total focinumber per image; individual foci positions were not ana-lyzed. In short, the software macro operates by evaluating themaximum intensity projection of a 3-dimensional stack ofimages to produce a single 2-dimensional intensity image,which is then processed. The images were normalized, andbackground noise was corrected. The noise suppression wasperformed with minimum reduction in focus detection sen-sitivity. Nuclear boundaries and foci were automatically iden-tified in images by a threshold algorithm. For identifyingnuclear boundaries, DAPI staining was used. Minimum focisize of 5 � 5 (0.25 �m2) pixels and maximum foci size 200pixels were selected. Adjunct or overlapping nuclei wereadequately segmented using watershed transformation. Cellsthat were partially on the edge of the image or deformed cellswere excluded by the software and not counted. Cells withpan-nuclear staining or band-like staining were also notanalyzed (33–35).

Statistical analyses

Determining statistical differences between preconceptionaland gestational exposures for DNA damage in human F1offspring is difficult because of limitations in separatinggestational from preconceptional exposure effects and con-founding variables (e.g., age or occupation) resulting inbiased estimation of exposure effects. The extreme case of abiased estimation is the detection of a causal effect wherenone exists or the fact that a true effect is unseen. Amultivariate linear regression model offers a way of separatinggestational and preconceptional exposure effects, avoidingconfounding, and comparing all independent variables avail-able (maternal, paternal, and newborn characteristics inTable 1) to identify only those that significantly predict thedependent variable (DNA damage in percentage tail DNA or�H2AX foci). The central assumption of a regression modelis that there is only one dependent variable (DNA damage),whereas all other variables are independent from each other

TABLE 1. Demographic characteristics of the newborn-mother-father triads

Variable

Mothers Fathers

n Mean � sd Range n Mean � sd Range

Age (yr) 39 29.1 � 5.4 18.0–40.0 39 32.9 � 5.4 21–43BMI (kg/m2)a 34 26.3 � 5.3 18.3–40.4 38 27.1 � 4.6 17.8–41.5Origin Crete, Greece (%) 39 61.5 39 61.5Smoking, answering yes (%)a,b 39 33.3 39 46.2No. cigarettes/d smoked known (%)a,b 39 10.3 39 46.2

Smoking 1–10 cigarettes/d (%) 39 7.7 39 12.8Smoking 11–20 cigarettes/d (%)a 39 2.6 39 10.2Smoking �20 cigarettes/d (%) 39 0 39 23.1

Cotinine concentration (ng/ml)a,b 15 15.7 � 20.7 1.5–76.5 15 88.2 � 93.4 12.2–324Smokers 7 24.0 � 25.7 3.0–76.5 7 158.1 � 110 55.2–324Nonsmokers 8 8.56 � 12.9 1.5–40.0 8 38.1 � 24.9 12.2–77.5

Environmental tobacco smoke,answering yes (%)a,b 30 43.3 24 66.7

Supplements, answering yes (%)a,b 33 72.7 39 7.7Alcohol, answering yes (%)a,b 29 20.7 39 71.8Total alcohol intake (g/d)a,b 29 1.9 � 1.2 0.6–3.4 25 14.9 � 23.5 1.1–114.3Ethnicity Caucasian (%) 39 79.5 39 79.5In occupation, answering yes (%)a,b 39 46.2 39 97.4

Plant and machine operators andassemblers (%) 39 0 39 15.4

Craft and related trades workers (%) 39 5.1 39 5.1Managers (%) 39 0 39 25.6Service and sales workers (%) 39 10.3 39 10.3Elementary occupations (%) 39 5.1 39 5.1Professionals (%) 39 5.1 39 7.7Technicians and associate

professionals (%) 39 5.1 39 15.4Clerks (%) 39 12.8 39 0Armed forces occupations (%) 39 2.56 39 2.6Clerical support workers (%) 39 0 39 2.6Skilled agricultural, forestry, and

fishery workers (%) 39 0 39 5.1Skilled agricultural, forestry, and

fishery workers and service andsales workers (%) 39 0 39 2.6

Newborn

Birth weight (g) 39 3342 � 434 2450–4250Gestational age (wk) 39 38.5 � 0.8 37–40Sex male (%) 39 64.1Delivery mode cesarean (%) 39 74.4

aAt the time of conception to delivery. b During conception.

3949SMOKING-INDUCED TRANSGENERATIONAL INSTABILITY

(characteristics in Table 1). Hence, any identified variablethat significantly predicts (known as the significant predictor)the dependent variable (DNA damage) is always independentfrom all other variables. To ensure that one or more predic-tor variables in a regression model are not highly correlated,i.e., dependent on each other, the variance inflation factor isquantified. Although exposure effects (known as exposurevariables) are of main interest for predicting DNA damage,and confounding variables are not directly interesting, theyare considered equally in a regression model (24, 27). Here,for the initial regression model, all maternal, paternal, andnewborn variables were combined and tested one by one fortheir statistical significance (set at P�0.05). The weakestpredictor variable was then removed, and the linear regres-sion was recalculated. If this significantly weakened themodel, then the variable was reentered or otherwise elimi-nated from the model. This algorithm was successively re-peated until only the statistically significant predictor variable(significant predictor) remained in the model. [To reiterate:identification of the significant predictor proved that onlythis variable determined statistically significant DNA damage(the dependent variable) and was not influenced or interact-ing with any of the other initially tested (independent)variables.] Each initial regression model was restricted toobservations with complete data for all variables and in-cluded, for the mothers, prepregnancy BMI, age, active andpassive smoking during pregnancy, occupation, use of anti-oxidant supplements, ethnicity, and alcohol intake duringpregnancy; for the fathers, preconceptional BMI, age, activesmoking, passive smoking, occupation, alcohol intake, use ofantioxidant supplements, and ethnicity; and for the newborn,sex, gestational age, birth weight, and delivery type. Associationsbetween demographic factors and DNA damage values wereevaluated using Pearson correlation analysis (24, 27). Becausethe regression model identified all possible significant predic-tors for DNA damage in this study, there was no need toperform any additional statistical analysis of the independentvariables separately (Table 1). Consequently, only for thecalculation of statistical differences of baseline DNA damagein newborns, mothers, and fathers, a paired t test (level ofsignificance, P�0.05) was applied (see Fig. 2) after a Kolm-ogorov-Smirnov goodness-of-fit test showed normal distribu-tion of the data for all samples. SPSS 16.0 (SPSS, Chicago, IL,USA) was used for all statistical analyses.

RESULTS

Study population

For nearly all families, complete demographic data setswere available (Table 1). The maternal age rangedbetween 18 and 40 yr (mean 29.1 yr), and the paternalage ranged between 21 and 43 yr (mean 32.9 yr). Themajority of the parents were Caucasian (79.5%) andlived in Crete, Greece. The majority of mothers(53.8%) were not in occupation during pregnancy.Following the International Standard Classification ofOccupations 2008 (ISCO-08), the most frequent mater-nal occupations were clerks (12.8%) and sales andservice workers (10.3%). Of the 39 fathers, all but 1(97.4%) were employed, with 25.6% working as man-agers and 15.4% as plant/machine operators and tech-nicians, followed by sales and service workers (10.3%)and professionals (7.7%). Mothers and fathers hadsimilar BMIs before conception, with a mean maternal

BMI of 26.3 kg/m2 and paternal BMI of 27.1 kg/m2.Whereas 71.8% of the fathers used alcohol at the timeof conception (mean ethanol intake 14.9 g/d), only20.7% of mothers drank alcohol during pregnancy (1.9g/d). Of mothers, 33.3% were actively smoking duringpregnancy, having a mean urine cotinine concentra-tion of 24.0 ng/ml, whereas for nonsmoking mothers, amean cotinine concentration of 8.56 ng/ml was calcu-lated. In addition, for 46.2% of the fathers who activelysmoked within the period between conception anddelivery, a mean cotinine in semen concentration of158 ng/ml compared with 38 ng/ml in nonsmokingfathers confirmed the self-reported smoking status. Themajority (50%) of these fathers smoked �20 cigarettes/d,22.2% smoked between 10 and 20 cigarettes/d, and27.8% smoked between 1 and 10 cigarettes/d. Unfor-tunately, only 4 of the 13 mothers who smoked pro-vided this information. Regarding environmental to-bacco smoke exposure, 43.3% of mothers were exposedin utero during pregnancy and 66.7% of fathers at thetime of conception. In the analysis of newborns, anaverage gestational age of 38.5 wk was found, with themajority of the mothers delivering via cesarean section(74.4%), and 64.1% of the newborns were male. Thebirth weights ranged from 2450 to 4250 g, with a meanof 3342 g. Birth weights of newborns from mothers whosmoked were decreased by 190 g and from fathers whosmoked were decreased by 111 g.

Biomarker distribution

DNA damage values for the Comet assay in percentagetail DNA, which mainly detects single-strand breaks, aswell as for laser scanning confocal microscopy (LSCM)of �H2AX foci, which detects double-strand breaksonly, were available for all 39 mother-father-child tri-ads. The measured levels of both DNA single- anddouble-strand breaks were found to be lower in lym-phocytes of newborns than in those of their respectiveparents (% tail DNA: P�0.001; �H2AX foci: P�0.001).DNA damage values in maternal and paternal lympho-cytes were similar (Figs. 1 and 2). A significant corre-lation between DNA damage assessed via �H2AX fociand percentage tail DNA was found in fathers (r�0.62,P�0.011). However, for newborns and mothers, nocorrelations of statistical significance between the twoDNA damage biomarkers were found (r�0.11 andr�0.27, respectively, both P�0.05; Fig. 3). When DNAdamage values in the blood of fathers and newbornswere compared, a highly significant association wasfound for percentage tail DNA (r�0.75; P�0.001), butnot for �H2AX foci (r�0.17; P�0.34). Similar correla-tions were found between newborns and mothers,showing a significant correlation for percentage tailDNA (r�0.49; P�0.004) but no significance for the�H2AX foci biomarker (r�0.17; P�0.33). In addition,between mothers and fathers, there was a significantcorrelation only for percentage tail DNA (r�0.51;P�0.002) but not for �H2AX foci (r�0.21; P�0.24).

3950 Vol. 26 October 2012 LAUBENTHAL ET AL.The FASEB Journal � www.fasebj.org

Multivariate analysis

Multivariate linear regression analysis offered a way toidentify out of all independent variables given (charac-teristics in Table 1) only those that were statisticallysignificant in predicting the dependent variable (DNAdamage in % tail DNA or �H2AX foci). The centralassumption of this model was that there is only onedependent variable (DNA damage), whereas all othervariables are independent from each other. Hence, anyidentified variable that significantly predicts (signifi-cant predictor) the dependent variable was alwaysindependent from all other variables. Regression anal-ysis identified active smoking of the fathers at the time

of conception as a significant predictor for percentagetail DNA in newborn blood. Both maternal smokingduring gestation and paternal smoking at the time ofconception were significant predictors for �H2AX fociin newborn blood. Maternal ETS exposure was not asignificant predictor, excluding the possibility that pa-ternal smoking during gestation might have affectedthe DNA integrity in newborn blood indirectly throughpassive smoking of the mothers. The variance inflationfactor for this important constellation was 1.3, confirm-ing that paternal smoking and maternal exposure toETS were independent and thus not interacting witheach other. When DNA damage in the respectiveparents’ blood was measured, only active smoking andhigh alcohol consumption were significant predictors(with slopes in different directions) for �H2AX foci inpaternal blood samples. No predictors of statisticalsignificance for DNA damage were found in maternalblood or in spermatozoa (Table 2).

Comparison of the mean DNA damage levels for 3constellations of parental smoking

Table 3 shows differences in mean DNA damage for 3constellations of parental cigarette smoking: case I,both parents neither actively nor passively smoked atthe time of conception to delivery (n�10); case II,mothers neither actively nor passively smoked duringgestation, but fathers actively smoked (n�4); and caseIII, both parents actively smoked during the time ofconception to delivery (n�9). Notably, maternal smok-ing and paternal smoking were already determined as

Figure 1. Representative images of �H2AX foci. Maximal projection of a Z stack shows differential interference contrast (DIC)as well as staining for DAPI and �H2AX-foci. The 16 selected confocal slices of a single lymphocyte (yellow frame) show howLSCM can detect all possible �H2AX foci within a single cell via scanning of up to 25 confocal slices, whereas conventionalepifluorescence microscopy can only visualize a single slice.

***

***

***

***

Newbornlymphocytes

Spermatoza

% Tail DNA

***p<.001Error bars: ±1SE

Maternallymphocytes

Paternallymphocytes

γH2A

X fo

ci /

% T

ail D

NA

35

30

25

20

15

10

5

0

γH2AX foci

Figure 2. Distribution of the DNA damage biomarkerspercentage tail DNA and �H2AX foci per 100 cells quantifiedin cord, maternal, and paternal lymphocytes as well as pater-nal spermatozoa.

3951SMOKING-INDUCED TRANSGENERATIONAL INSTABILITY

significant predictors for DNA damage in newborns viaregression analysis, which separated gestational andpreconceptional smoking exposures and excluded thefact that the effects of cigarette smoke on DNA damagewere influenced by any confounding variables. Hence,it would be against the assumptions of the present studyto analyze these 3 constellations of smoking again forsignificance, because such analysis would only test onevariable (smoking vs. nonsmoking) while neglectingpossible confounding effects of all other variables.When case I was compared with case II, there was a 35%increase in �H2AX foci and a 16.4% increase in per-centage tail DNA in blood of newborns whose fatherssmoked but whose mothers were not exposed to passivesmoke during pregnancy or actively smoking. It is

thought that only the mothers in case II were notexposed to passive smoke, because the fathers smokedeither outside or when mothers were away. Fathers whosmoked showed a 31.5% (�H2AX foci) and 15.7% (%tail DNA) DNA damage increase in their own bloodand a 9.5% increase in their spermatozoa. However,there were no differences in DNA damage in the bloodof the associated mothers. Comparison of case I withcase III showed that DNA damage in cord blood ofoffspring from mothers and fathers who activelysmoked during pregnancy or at the time of conceptionwas 39% (�H2AX foci) and 15.8% (% tail DNA),respectively. The respective actively smoking mothershad a 17.1% increase in DNA damage for �H2AX fociand a 2.2% increase for percentage tail DNA, whereas

% T

ail D

NA

γH2AX foci

Cord lymphocytes ( r = 0.11; p > 0.05 )

Father lymphocytes ( r = 0.62; p = 0.011 )

Mother lymphocytes ( r = 0.27; p > 0.05 )

20

18

16

14

12

10

8

6

4

2

00 5 10 15 20 25 30 35

Figure 3. Correlation between percentage tail DNA and �H2AX foci biomarkers for cord, maternal, and paternal lymphocytes.A significant correlation for biomarkers used on paternal lymphocytes but not on maternal and cord lymphocytes was found.

TABLE 2. Significant predictors after multiple regression analysis with backward selection for demographical factors of the DNA damagebiomarkers percentage tail DNA and �H2AX foci/100 cells in lymphocytes of newborns and their parents and spermatozoa of the fathers

Population Dependent Independent Partial r2 Slope R2 P

Newborn lymphocytes % Tail DNA Father smoking 0.154 0.393 0.154 0.032�H2AX foci Mother smoking 0.130 0.397 0.033

Father smoking 0.165 0.437 0.0180.411 0.001*

Maternal lymphocytes % Tail DNA No significant predictors�H2AX foci No significant predictors

Paternal lymphocytes % Tail DNA No significant predictors�H2AX foci Father smoking 0.104 0.494 0.005

Father alcohol 0.153 �0.337 0.0390.375 0.002*

Spermatozoa % Tail DNA No significant predictors�H2AX foci

*Significance level set in the model: P � 0.05.

3952 Vol. 26 October 2012 LAUBENTHAL ET AL.The FASEB Journal � www.fasebj.org

the respective actively smoking fathers showed 35.2%(�H2AX foci) and 10.4% (% tail DNA) increases inDNA damage in their blood and a 21.2% increase intheir spermatozoa.

DISCUSSION

Although the mechanisms, functions, and conse-quences of many human (somatic) toxicants are wellknown, a human germ cell toxicant or mutagen induc-ing transgenerational alterations in gene expression,DNA damage, or mutations in the unexposed offspringhas not been identified yet. This may be due to thedifficulty in separating preconceptional from gesta-tional exposures as well the choice of experimentalapproaches to detect transmissible genetic alterationsin humans. Parental cigarette smoking is considered ahighly relevant example of an exposed cohort in hu-man population studies, and there is compelling evi-dence from demographic and rodent studies (7, 23)that cigarette smoke might be the first identified hu-man germ cell mutagen (36). The present study, al-though limited in the number of individuals (n�117),profited from an unusually high number of fathers(46.2%) and mothers (33.3%) who smoked heavilyduring gestation and/or at the time of conception(Table 1). In addition, a distinctive situation was given,in which 4 mothers were neither actively nor passivelyexposed to cigarette smoke during gestation, whereasthe respective fathers were heavy smokers at the time ofconception to delivery (Table 3). Multivariate linearregression analysis identified only paternal smoking atthe time of conception and maternal smoking duringgestation as significant predictors for elevated DNAdouble- and single-strand breaks (paternal smoking) ordouble-strand breaks only (maternal smoking) in cordblood of the offspring, independent from each other(Table 2). These data were generated by using atraditional biomarker for DNA strand breaks (% tailDNA) and the most sensitive way method for thedetection of double-strand breaks, confocal laser scansof �H2AX foci followed by semiautomated image anal-

ysis of the acquired image stacks. The same DNAdamage biomarkers (% tail DNA and �H2AX foci) asapplied here combined with a mutation test (small-molecule PCR) were previously used in a restrictedlaboratory environment to analyze the nonexposed F1offspring of irradiated male mice, showing significantlyincreased expanded simple tandem repeat (ESTR)mutations along with transgenerational alterations inDNA damage (37). Notably, increased ESTR mutationfrequencies were also found in spermatozoa of miceexposed to active (38) and passive cigarette smoke(23), concluding that both are germ cell mutagens.Based on these studies, it is likely that in humans alsotransgenerational DNA alterations may be converted tomutations in the F1 offspring of fathers who wereexposed to cigarette smoke before conception. How-ever, this conclusion would have dramatic implicationsfor public health decision makers, because this hypoth-esis suggests that cigarette smoke has to be considereda human germ cell mutagen (7).

Mechanistically, transgenerational alterations ingenomic stability due to alterations in DNA damage inthe cord blood of human F1 offspring can be wellexplained by cigarette smoke metabolites inducingtransgenerational epigenetic inheritance of vulnerabil-ity for DNA damage. First, it was shown that subsets ofcytosine methylation and chromatin structure patternsin spermatozoa are transmissible throughout subse-quent generations (13, 39), suggesting that adverseDNA methylation or chromatin structure patterns caninfluence the regulation of genes responsible for main-taining genome stability and alter regular DNA dam-age, response, and repair processes. In that way, spon-taneous DNA strand breaks can remain undetectedand/or unrepaired, explaining the elevated strandbreaks in human cord blood of exposed parents here(37, 40). Because lymphocytes are considered to besurrogate cells for other tissues (41), it is likely that theelevated DNA strand breaks detected in the cord bloodof F1 offspring exposed to cigarette smoke represent tosome extent an average instability level of the entirefetal genome. Second, another mechanism to transmitmore functional information to the offspring than that

TABLE 3. Comparison of mean DNA damage levels for 3 particular constellations of parental smoking in a subset of 23 families.

Variables n

Newborn lymphocytes Maternal lymphocytes Paternal lymphocytes

Spermatozoa:% tail DNA

�H2AXfoci

% TailDNA

�H2AXfoci

% TailDNA

�H2AXfoci

% TailDNA

Case I: mother nonsmoker andnot ETS-exposed/fathernonsmoker 10 9.1 � 2.8 10.7 � 1.8 19.8 � 2.0 13.9 � 1.8 14.4 � 1.4 12.9 � 2.1 24.9 � 3.3

Case II: mother nonsmokerand not ETS-exposed/fathersmoker 4 14.0 � 6.3 12.8 � 2.3 19.2 � 2.1 14.0 � 2.0 21.0 � 2.3 15.3 � 2.4 27.5 � 4.8

Case III: mother and fathersmoker 9 14.8 � 2.1 12.7 � 1.8 23.9 � 1.7 14.2 � 2.5 22.2 � 1.9 14.4 � 2.0 31.6 � 2.5

No analysis for significance was performed here, because smoking was already identified as the significant predictor for DNA damage innewborns out of all available variables via regression analysis (data are merely shown for completeness).

3953SMOKING-INDUCED TRANSGENERATIONAL INSTABILITY

carried in the DNA sequences is the RNA content of thespermatozoa. Spermatozoal RNA molecules were shownto be transferred to the fertilized egg and to consequentlymodulate gene expression in the embryo (42, 43). In linewith this finding, our laboratories recently showed thatthe spermatozoal mRNA content is affected by paternalsmoking (44, 45) and that gene expression patternswere altered during the very early stages of cell division(1-, 2-, 4- and 8-cell) of mouse embryos from oocytesfertilized in vitro, when the spermatozoon was derivedfrom mice treated intraperitoneally with glycidamide(46) or with benzo[a]pyrene (47). Based on theseobservations, it is conceivable that spermatozoal RNAcontent transferred to the fertilized egg is affected bypaternal smoking, which consequently influenced theepigenetic programming of the zygote (43, 48). Theresult could be loss of control over DNA integrity,altered DNA damage response, and consequently amore vulnerable phenotype for DNA strand breaks inthe offspring as detected here. However, further studiesare needed to confirm the hypothesis that paternalRNAs affected by smoking can have a role in modulat-ing adverse gene expression in the offspring.

Unrepaired DNA adducts in spermatozoa, resultingfrom the father’s smoking habit relatively close toconception, could be an alternative mechanistic basisfor the genetic instability in the offspring detectedhere. This hypothesis would be in line with severalmechanistic studies of our group on benzo[a]pyrene-exposed mice, which showed that benzo[a]pyrene-induced DNA damage in spermatogonial stem cells wasrepaired, whereas DNA damage induced later in sper-matogenesis (spermiogenesis) is not repaired but existsas benzo[a]pyrene-7,8-diol-9,10-epoxide adducts, which maythen be transmitted to the offspring (8, 49, 50). More-over, this is in agreement with our data (32) and datafrom Zenzes et al. (51) on human spermatozoa samplesof smokers, showing that benzo[a]pyrene is able tocross the blood-testis barrier, induce DNA damage, andbe transmitted to the genome of the unborn offspring.

The effects of paternal cigarette smoking on humanspermatozoa and lymphocytes in the induction of oxida-tive DNA damage, strand breaks, chromosomal aberra-tions, mutations, and gene expression are well docu-mented (21, 22, 44) and consistent with the identificationof heavy cigarette smoking as a significant predictor forDNA double-strand breaks in paternal blood. In addi-tion, a 9.5% DNA strand break increase in spermatozoafrom smokers is in line with previous results from thisgroup (32, 45); however, smoking was not found to besignificantly predictive for DNA damage in the fathers’sperm here, probably because a regression model tocontrol for confounding variables, which apparentlyinfluenced the effects of smoke on sperm, was applied.Thus, the possibility that the significant differencesreported earlier (reviewed in ref. 22) were influencedby confounding variables cannot be excluded becausethese studies did not control for confounding. Inaddition, using spermatozoa samples from the samedonors, our group showed that elevated spermatozoal

DNA damage originated from oxidative stress (45).Many studies showed that newborns have an increasedsusceptibility to toxicant-induced DNA damage in com-parison with that of their mothers (3, 52, 53). Thisfinding is in agreement with the identification of ma-ternal cigarette smoke as a significant predictor forDNA damage in newborns, but not their mothers, whoare known to have lower susceptibilities to toxicants.Furthermore, evidence for the impact of maternalcigarette smoking on the health of unborn offspringwas given in many mother-child birth cohort studies(54) reporting chromosomal instability in amniocytesas well as chromosomal translocations, DNA damage,mutations, and micronucleation in the F1 offspringcord blood (17, 54–56). However, none of these mother-child birth cohort studies adjusted for the possiblepaternal impact as was done in the present study.Hence, the data presented here suggest rethinkingbasic practices in birth cohort studies, which exclusivelyfocus on the maternal impacts while neglecting possi-ble paternal impacts (6). Although the consequences ofcigarette smoke-derived metabolites on somatic cells,including a dramatically increased risk to develop can-cer, are known to the public because of campaigns andmore restrictive laws, a hardcore population of smokersremains, which in the United States, for example,accounts for 35% of reproductive age men and 30% ofreproductive age women (57). The deleterious effectsof cigarette smoke on the male and female reproduc-tive function and health of the offspring, however, areless known to the public. In epidemiological studies,active maternal smoking during gestation was identi-fied as a well-documented cause of fetal mortality andmorbidity and an average 240-g lower birth weight(here 190- and 111-g decreases in birth weight ofnewborns of smoking fathers and mothers, respectively,was found) as well as congenital malformations (18).Maternal environmental smoke exposure was shown toincrease the risk for stillbirth and congenital malforma-tions significantly (58). There is also compelling evi-dence that paternal smoking is associated with adversereproductive outcome, including birth defects (18) orincreases in spontaneous abortions (59). In addition,the International Agency for Research on Cancer al-ready stated recently that there is sufficient evidence tolink both preconceptional paternal and maternal as well asgestational tobacco smoke consumption with an increasedrisk for childhood cancer in the unexposed offspring (20).These epidemiological data are in line with a cigarettesmoke-induced transgenerational genomic instability, be-cause unrepaired lesions in critical genes (such as tumorsuppressor or DNA damage response genes) can impede theability of a cell to perform its function and considerablyincrease the likelihood of tumor formation (3, 60).

In summary, this study shows for the first time, to ourknowledge, a transgenerational role for preconcep-tional and gestational cigarette smoke exposure in theinduction of or vulnerability for DNA damage in hu-man offspring. Because DNA alterations may causepermanent mutations, these data also support the claim

3954 Vol. 26 October 2012 LAUBENTHAL ET AL.The FASEB Journal � www.fasebj.org

that cigarette smoke is a human germ cell mutagen.Hence, protection from toxicant exposures, particu-larly cigarette smoke, may not only be essential duringgestation but also before conception, as shown here forthe male.

The authors thank M. Kogevinas (Centre for Research inEnvironmental Epidemiology, Barcelona, Spain) for enablingsamples to be provided from the NewGeneris subcohort inCrete, Greece; Leda Chatzi for her considerable contributionto the cotinine data (University of Crete, Heraklion, Greece);G. Chalkiadaki and T. Roumeliwtaki (University of Crete) aswell as M. Kurzawa-Zegota and E. Cemeli (Food StandardsAgency, London, UK) for support with sample collection andpreparation; V. A. Botchkarev (University of Boston, Boston,MA, USA) for support with LSCM; Constantine Vardavas(Harvard University, Boston, MA, USA) for discussion oncotinine analysis; and M. Kogevinas and I. F. Merlo (IstitutoDi Ricovero e Cura a Carattere Scientifico Azienda Os-pedaliera Universitaria San Martino-IST-National Cancer Re-search Institute, Genoa, Italy) for critical review of the article.This work was financed by the European Union IntegratedProject NewGeneris, 6th Framework Programme, Priority 5:Food Quality and Safety, contract FOOD-CT-2005-016320(http://www.newgeneris.org).

REFERENCES

1. Anderson, L. M., Diwan, B. A., Fear, N. T., and Roman, E. (2000)Critical windows of exposure for children’s health: cancer inhuman epidemiological studies and neoplasms in experimentalanimal models. Environ. Health Perspect. 108(Suppl. 3), 573–594

2. Selevan, S. G., Kimmel, C. A., and Mendola, P. (2000) Identify-ing critical windows of exposure for children’s health. Environ.Health Perspect. 108(Suppl. 3), 451–455

3. Wild, C. P., and Kleinjans, J. (2003) Children and increasedsusceptibility to environmental carcinogens: evidence or empa-thy? Cancer Epidemiol. Biomarkers Prev. 12, 1389–1394

4. Gluckman, P. D., Hanson, M. A., Cooper, C., and Thornburg,K. L. (2008) Effect of in utero and early-life conditions on adulthealth and disease. N. Engl. J. Med. 359, 61–73

5. Merlo, D. F., Wild, C. P., Kogevinas, M., Kyrtopoulos, S., andKleinjans, J. (2009) NewGeneris: a European study on maternaldiet during pregnancy and child health. Cancer Epidemiol. Bio-markers Prev. 18, 5–10

6. Anderson, D. (2005) Male-mediated developmental toxicity.Toxicol. Appl. Pharmacol. 207, 506–513

7. Singer, T. M., and Yauk, C. L. (2010) Germ cell mutagens: riskassessment challenges in the 21st century. Environ. Mol. Mutagen51, 919–928

8. Olsen, A. K., Andreassen, A., Singh, R., Wiger, R., Duale, N.,Farmer, P. B., and Brunborg, G. (2010) Environmental expo-sure of the mouse germ line: DNA adducts in spermatozoa andformation of de novo mutations during spermatogenesis. PLoSOne 5, e11349

9. Carone, B. R., Fauquier, L., Habib, N., Shea, J. M., Hart, C. E.,Li, R., Bock, C., Li, C., Gu, H., Zamore, P. D., Meissner, A.,Weng, Z., Hofmann, H. A., Friedman, N., and Rando, O. J.(2010) Paternally induced transgenerational environmental re-programming of metabolic gene expression in mammals. Cell143, 1084–1096

10. Ng, H. K., Novakovic, B., Hiendleder, S., Craig, J. M., Roberts,C. T., and Saffery, R. (2010) Distinct patterns of gene-specificmethylation in mammalian placentas: implications for placentalevolution and function. Placenta 31, 259–268

11. Dubrova, Y. E., Plumb, M., Gutierrez, B., Boulton, E., andJeffreys, A. J. (2000) Transgenerational mutation by radiation.Nature 405, 37

12. Anway, M. D., Cupp, A. S., Uzumcu, M., and Skinner, M. K.(2005) Epigenetic transgenerational actions of endocrine dis-ruptors and male fertility. Science 308, 1466–1469

13. Youngson, N. A., and Whitelaw, E. (2008) Transgenerationalepigenetic effects. Annu. Rev. Genomics Hum. Genet. 9, 233–257

14. Ng, S. F., Lin, R. C., Laybutt, D. R., Barres, R., Owens, J. A., andMorris, M. J. (2010) Chronic high-fat diet in fathers programs-cell dysfunction in female rat offspring. Nature 467, 963–966

15. Dubrova, Y. E. (2003) Radiation-induced transgenerational in-stability. Oncogene 22, 7087–7093

16. Glen, C. D., Smith, A. G., and Dubrova, Y. E. (2008) Single-molecule PCR analysis of germ line mutation induction byanticancer drugs in mice. Cancer Res. 68, 3630–3636

17. Bennett, L. M., Wang, Y., Ramsey, M. J., Harger, G. F., Bigbee,W. L., and Tucker, J. D. (2010) Cigarette smoking duringpregnancy: chromosome translocations and phenotypic suscep-tibility in mothers and newborns. Mutat. Res. 696, 81–88

18. Soares, S. R., and Melo, M. A. (2008) Cigarette smoking andreproductive function. Curr. Opin. Obstet. Gynecol. 20, 281–291

19. ASRM (2008) Smoking and infertility. Fertil. Steril. 90, S254–25920. Secretan, B., Straif, K., Baan, R., Grosse, Y., El Ghissassi, F.,

Bouvard, V., Benbrahim-Tallaa, L., Guha, N., Freeman, C.,Galichet, L., and Cogliano, V. (2009) A review of humancarcinogens–part E: tobacco, areca nut, alcohol, coal smoke,and salted fish. Lancet Oncol. 10, 1033–1034

21. Aitken, R. J., Koopman, P., and Lewis, S. E. (2004) Seeds ofconcern. Nature 432, 48–52

22. DeMarini, D. M. (2004) Genotoxicity of tobacco smoke andtobacco smoke condensate: a review. Mutat. Res. 567, 447–474

23. Marchetti, F., Rowan-Carroll, A., Williams, A., Polyzos, A.,Berndt-Weis, M. L., and Yauk, C. L. (2011) Sidestream tobaccosmoke is a male germ cell mutagen. Proc. Natl. Acad. Sci. U. S. A.108, 12811–12814

24. Vande Loock, K., Fthenou, E., Decordier, I., Chalkiadaki, G.,Keramarou, M., Plas, G., Roelants, M., Kleinjans, J., Chatzi, L.,Merlo, F., Kogevinas, M., and Kirsch-Volders, M. (2011) Mater-nal and gestational factors and micronuclei frequencies inumbilical blood: the NewGeneris Rhea Cohort in Crete. Environ.Health Perspect. 119, 1460–1465

25. Chatzi, L., Plana, E., Daraki, V., Karakosta, P., Alegkakis, D.,Tsatsanis, C., Kafatos, A., Koutis, A., and Kogevinas, M. (2009)Metabolic syndrome in early pregnancy and risk of pretermbirth. Am. J. Epidemiol. 170, 829–836

26. World Health Organization. (1999) WHO Laboratory Manual forthe Examination of Human Semen and Sperm-Cervical Mucus Inter-action. Cambridge University Press, Cambridge, UK

27. Vardavas, C. I., Fthenou, E., Patelarou, E., Bagkeris, E., Murphy,S., Hecht, S. S., Connolly, G. N., Chatzi, L., and Kogevinas, M.(2012) Exposure to different sources of second-hand smokeduring pregnancy and its effect on urinary cotinine and tobac-co-specific nitrosamine (NNAL) concentrations. [E-pub aheadof print] Tob. Control. doi: 10.1136/tobaccocontrol-2011-050144

28. Hecht, S. S., Carmella, S. G., Le, K. A., Murphy, S. E., Boettcher,A. J., Le, C., Koopmeiners, J., An, L., and Hennrikus, D. J.(2006) 4-(Methylnitrosamino)-1-(3-pyridyl)-1-butanol and itsglucuronides in the urine of infants exposed to environmentaltobacco smoke. Cancer Epidemiol. Biomarkers Prev. 15, 988–992

29. Hecht, S. S., Carmella, S. G., Chen, M., Dor Koch, J. F., Miller,A. T., Murphy, S. E., Jensen, J. A., Zimmerman, C. L., andHatsukami, D. K. (1999) Quantitation of urinary metabolites ofa tobacco-specific lung carcinogen after smoking cessation.Cancer Res. 59, 590–596

30. Hammadeh, M. E., Hamad, M. F., Montenarh, M., and Fischer-Hammadeh, C. (2010) Protamine contents and P1/P2 ratio inhuman spermatozoa from smokers and non-smokers. Hum.Reprod. 25, 2708–2720

31. Singh, N. P., McCoy, M. T., Tice, R. R., and Schneider, E. L.(1988) A simple technique for quantitation of low levels of DNAdamage in individual cells. Exp. Cell. Res. 175, 184–191

32. Sipinen, V., Laubenthal, J., Baumgartner, A., Cemeli, E., Linschoo-ten, J. O., Godschalk, R. W., Van Schooten, F. J., Anderson, D., andBrunborg, G. (2010) In vitro evaluation of baseline and inducedDNA damage in human sperm exposed to benzo[a]pyrene or itsmetabolite benzo[a]pyrene-7,8-diol-9,10-epoxide, using the cometassay. Mutagenesis 25, 417–425

33. Schmid, T. E., Dollinger, G., Beisker, W., Hable, V., Greubel, C.,Auer, S., Mittag, A., Tarnok, A., Friedl, A. A., Molls, M., andRoper, B. (2010) Differences in the kinetics of �-H2AX fluores-cence decay after exposure to low and high LET radiation. Int.J. Radiat. Biol. 86, 682–691

3955SMOKING-INDUCED TRANSGENERATIONAL INSTABILITY

34. Shutin, D., and Zlobinskaya, O. (2010) Application of informa-tion-theoretic measures to quantitative analysis of immunofluo-rescent microscope imaging. Comput. Methods Programs Biomed.97, 114–129

35. Zlobinskaya, O., Dollinger, G., Michalski, D., Hable, V., Greu-bel, C., Du, G., Multhoff, G., Roper, B., Molls, M., and Schmid,T. E. (2012) Induction and repair of DNA double-strand breaksassessed by gamma-H2AX foci after irradiation with pulsed orcontinuous proton beams. Radiat. Environ. Biophys. 51, 23–32

36. Demarini, D. M. (2012) Declaring the existence of humangerm-cell mutagens. Environ. Mol. Mutagen 53, 166–172

37. Barber, R. C., Hickenbotham, P., Hatch, T., Kelly, D., Topchiy,N., Almeida, G. M., Jones, G. D., Johnson, G. E., Parry, J. M.,Rothkamm, K., and Dubrova, Y. E. (2006) Radiation-inducedtransgenerational alterations in genome stability and DNAdamage. Oncogene 25, 7336–7342

38. Yauk, C. L., Berndt, M. L., Williams, A., Rowan-Carroll, A.,Douglas, G. R., and Stampfli, M. R. (2007) Mainstream tobaccosmoke causes paternal germ-line DNA mutation. Cancer Res. 67,5103–5106

39. Holliday, R. (1987) The inheritance of epigenetic defects.Science 238, 163–170

40. Yauk, C., Polyzos, A., Rowan-Carroll, A., Somers, C. M., God-schalk, R. W., Van Schooten, F. J., Berndt, M. L., Pogribny, I. P.,Koturbash, I., Williams, A., Douglas, G. R., and Kovalchuk, O.(2008) Germ-line mutations, DNA damage, and global hyper-methylation in mice exposed to particulate air pollution in anurban/industrial location. Proc. Natl. Acad. Sci. U. S. A. 105,605–610

41. Waters, D. J., Shen, S., Xu, H., Kengeri, S. S., Cooley, D. M.,Chiang, E. C., Chen, Y., Schlittler, D., Oteham, C., Combs, G. F.,Jr., Glickman, L. T., Morris, J. S., and Bostwick, D. G. (2007)Noninvasive prediction of prostatic DNA damage by oxidativestress challenge of peripheral blood lymphocytes. Cancer Epide-miol. Biomarkers Prev. 16, 1906–1910

42. Chandler, V. L., and Stam, M. (2004) Chromatin conversations:mechanisms and implications of paramutation. Nat. Rev. Genet.5, 532–544

43. Rassoulzadegan, M., Grandjean, V., Gounon, P., Vincent, S.,Gillot, I., and Cuzin, F. (2006) RNA-mediated non-mendelianinheritance of an epigenetic change in the mouse. Nature 441,469–474

44. Linschooten, J. O., Van Schooten, F. J., Baumgartner, A.,Cemeli, E., Van Delft, J., Anderson, D., and Godschalk, R. W.(2009) Use of spermatozoal mRNA profiles to study gene-environment interactions in human germ cells. Mutat. Res. 667,70–76

45. Linschooten, J. O., Laubenthal, J., Cemeli, E., Baumgartner, A.,Anderson, D., Sipinen, V. E., Brunborg, G., Haenen, G. R.,Fthenou, E., Briede, J. J., van Schooten, F. J., and Godschalk,R. W. (2011) Incomplete protection of genetic integrity ofmature spermatozoa against oxidative stress. Reprod. Toxicol. 32,106–111

46. Brevik, A., Rusnakova, V., Duale, N., Slagsvold, H. H., Olsen,A. K., Storeng, R., Kubista, M., Brunborg, G., and Lindeman, B.(2011) Preconceptional paternal glycidamide exposure affects

embryonic gene expression: single embryo gene expressionstudy following in vitro fertilization. Reprod. Toxicol. 32, 463–471

47. Brevik, A., Lindeman, B., Brunborg, G., and Duale, N. (2012)Paternal benzo[a]pyrene exposure modulates microRNA ex-pression patterns in the developing mouse embryo. Int. J. CellBiol. 2012:407�431

48. Bernstein, E., and Allis, C. D. (2005) RNA meets chromatin.Genes Dev. 19, 1635–1655

49. Verhofstad, N., van Oostrom, C. T., van Benthem, J., vanSchooten, F. J., van Steeg, H., and Godschalk, R. W. (2010) DNAadduct kinetics in reproductive tissues of DNA repair proficientand deficient male mice after oral exposure to benzo(a)pyrene.Environ. Mol. Mutagen 51, 123–129

50. Verhofstad, N., van Oostrom, C. T., Zwart, E., Maas, L. M., vanBenthem, J., van Schooten, F. J., van Steeg, H., and Godschalk,R. W. (2011) Evaluation of benzo(a)pyrene-induced gene mu-tations in male germ cells. Toxicol. Sci. 119, 218–223

51. Zenzes, M. T., Puy, L. A., Bielecki, R., and Reed, T. E. (1999)Detection of benzo[a]pyrene diol epoxide-DNA adducts inembryos from smoking couples: evidence for transmission byspermatozoa. Mol. Hum. Reprod. 5, 125–131

52. Perera, F. P., Tang, D., Tu, Y. H., Cruz, L. A., Borjas, M., Bernert,T., and Whyatt, R. M. (2004) Biomarkers in maternal andnewborn blood indicate heightened fetal susceptibility to pro-carcinogenic DNA damage. Environ. Health Perspect. 112, 1133–1136

53. Wyatt, N. P., Falque-Gonzalez, C., Farrar, D., Tuffnell, D.,Whitelaw, D., Knudsen, L. E., and Anderson, D. (2007) In vitrosusceptibilities in lymphocytes from mothers and cord blood tothe monofunctional alkylating agent EMS. Mutagenesis 22, 123–127

54. Demarini, D. M., and Preston, R. J. (2005) Smoking whilepregnant: transplacental mutagenesis of the fetus by tobaccosmoke. JAMA 293, 1264–1265

55. de la Chica, R. A., Ribas, I., Giraldo, J., Egozcue, J., and Fuster,C. (2005) Chromosomal instability in amniocytes from fetusesof mothers who smoke. JAMA 293, 1212–1222

56. Keohavong, P., Xi, L., Day, R. D., Zhang, L., Grant, S. G., Day,B. W., Ness, R. B., and Bigbee, W. L. (2005) HPRT genealterations in umbilical cord blood T-lymphocytes in newbornsof mothers exposed to tobacco smoke during pregnancy. Mutat.Res. 572, 156–166

57. American Society of Reproductive Medicine (2008) Smokingand infertility. Fertil. Steril. 90, S254–S259

58. Leonardi-Bee, J., Britton, J., and Venn, A. (2011) Secondhandsmoke and adverse fetal outcomes in nonsmoking pregnantwomen: a meta-analysis. Pediatrics 127, 734–741

59. Venners, S. A., Wang, X., Chen, C., Wang, L., Chen, D., Guang,W., Huang, A., Ryan, L., O’Connor, J., Lasley, B., Overstreet, J.,Wilcox, A., and Xu, X. (2004) Paternal smoking and pregnancyloss: a prospective study using a biomarker of pregnancy. Am. J.Epidemiol. 159, 993–1001

60. Hanahan, D., and Weinberg, R. A. (2011) Hallmarks of cancer:the next generation. Cell 144, 646–674

Received for publication January 3, 2012.Accepted for publication June 5, 2012.

3956 Vol. 26 October 2012 LAUBENTHAL ET AL.The FASEB Journal � www.fasebj.org