Dn

NSD

a

ARRAA

AGGGnGnSnFnDM

KGDGDD

1

tspfedpma

h1

Mutation Research 798 (2016) 1–10

Contents lists available at ScienceDirect

Mutation Research/Genetic Toxicology andEnvironmental Mutagenesis

jo ur nal home page: www.elsev ier .com/ locate /gentoxComm uni t y ad dress : www.elsev ier .com/ locate /mutres

ifferential genotoxic and epigenotoxic effects of graphene familyanomaterials (GFNs) in human bronchial epithelial cells

ivedita Chatterjee, JiSu Yang, Jinhee Choi ∗

chool of Environmental Engineering, Graduate School of Energy and Environmental System Engineering, University of Seoul, 163 Siripdaero,ongdaemun-gu, Seoul 130-743, Republic of Korea

r t i c l e i n f o

rticle history:eceived 21 September 2015eceived in revised form 8 January 2016ccepted 27 January 2016vailable online 1 February 2016

bbreviations:FNs, graphene family nanomaterialsNP-Prist, graphene nanoplatelets-pristineNP-COOH, grapheneanoplatelets-carboxylatedNP-NH2, grapheneanoplatelets-aminatedLGO, single layer graphene oxideanomaterialsLGO, few layer graphene oxideanomaterialsNMTs, DNA methyltransferaseBD, methyl-CpG binding domain protein

a b s t r a c t

The widespread applications of graphene family nanomaterials (GFNs) raised the considerable concernover human health and environment. The cyto-genotoxic potentiality of GFNs has attracted much moreattention, albeit the potential effects on the cellular epigenome remain largely unknown. The effects ofGFNs on cellular genome were evaluated with single and double stranded DNA damage and DNA repairgene expressions while the effects on epigenome was accomplished by addressing the global DNA methy-lation and expression of DNA methylation machineries at non-cytotoxic to moderately cytotoxic doses inin vitro system. We used five different representatives of GFNs-pristine (GNP-Prist), carboxylated (GNP-COOH) and aminated (GNP-NH2) graphene nanoplatelets as well as single layer (SLGO) and few layer(FLGO) graphene oxide. The order of single stranded DNA damage was observed as GNP-Prist ≥ GNP-COOH > GNP-NH2 ≥ FLGO > SLGO at 10 mg/L and marked dose dependency was found in SLGO. The GFNspossibly caused genotoxicity by affecting nucleotide excision repair and non-homologus end joiningrepair systems. Besides, dose dependent increase in global DNA methylation (hypermethylation) wereobserved in SLGO/FLGO exposure and conversely, GNPs treatment caused hypomethylation following theorder as GNP-COOH > GNP-NH2 ≥ GNP-Prist. The decrements of DNA methyltransferase (DNMT3B gene)and methyl-CpG binding domain protein (MBD1) genes were probably the cause of global hypomethy-lation induced by GNPs. Conversely, the de novo methylation through the up-regulation of DNMT3B andMBD1 genes gave rise to the global DNA hypermethylation in SLGO/FLGO treated cells. In general, the

eywords:raphene family nanomaterials (GFNs)NA damage-repairlobal DNA methylationNA methyltransferases (DNMTs)

GFNs induced genotoxicity and alterations of global DNA methylation exhibited compounds type speci-ficity with differential physico-chemical properties. Taken together, our study suggests that the GFNscould cause more subtle changes in gene expression programming by modulating DNA methylationstatus and this information would be helpful for their prospective use in biomedical field.

© 2016 Elsevier B.V. All rights reserved.

NA demethylases (TETs)

. Introduction

Graphene, the emerging allotrope of carbon materials, isypically the single-atom-layer monocrystalline structure withp2-hybridized hexagonally arranged carbon atoms. Grapheneossess very unique and unusual properties such as, high sur-

ace area, excellent conductivity, outstanding mechanical strength,xtraordinary electrocatalytic activities etc., which aiming of futureevelopment in the area of fields of electronics, photonics, com-

osite materials, energy generation and storage, sensors andetrology, and biomedicine, specifically, biosensors, bio-imaging

nd theraputics, tissue engineering, and drug delivery [1–5].

∗ Corresponding author. Fax: +82 2 6490 2859.E-mail address: jinhchoi@uos.ac.kr (J. Choi).

ttp://dx.doi.org/10.1016/j.mrgentox.2016.01.006383-5718/© 2016 Elsevier B.V. All rights reserved.

Graphene materials vary in layer number, lateral dimension, sur-face chemistry, defect density or purity which give rise to variousrelated forms of graphenes, such as, few-layer-graphene (FLG),ultrathin graphite, graphene oxide (GO), reduced graphene oxide(rGO), graphene nanoplatelets (GNP) etc. [1,2,6]. The field of GNPs’applications include thermoplastic and thermoset composites, nat-ural or synthetic rubber, thermoplastic elastomers, adhesives,paints, coatings etc., while applications of GOs mainly are relatedto biomedical field. The GOs, in particular with a differential layernumber, has shown the potentiality of controlling the timing ororder of release of various therapeutic drugs in a layer numberdependent manner [7]. The concerns rise on the increasing unin-

tentional occupational or environmental exposure or intentionalbiomedical exposure of graphene family nanomaterials (GFNs)[2,3].

2 tion R

abadikt

otesdabscntttDtpcmmtmlm

etmaWn(ln

2

2

wGV1bfreRMwio5[ms

N. Chatterjee et al. / Muta

The GFNs show several unique modes of interaction with nucleiccids (DNA/RNA) [2]. in particular, DNA cleavage ability exhibitedy the GO [8]. Previous reports, including ours, observed DNA dam-ge potentiality of graphene nanomaterials which differentiallyepends on size [9], surface charge and functionality [10,11], chem-

cal modifications and multiform derivatives [12]. To best of ournowledge, however, the epigenetic effects as a result of exposureo GFNs using in vitro systems have yet to be characterized.

While numerous studies investigated the cyto-genotoxic effectsf GFNs, the potential of engineered nanomaterials, including GFNs,o target the cellular epigenome remains largely unexplored. Nev-rtheless, accumulating evidences recognized that environmentaltressors can modulate epigenome and could play pivotal role inisease progressions, such as, cancer [13,14]. In addition, it haslso been suggested that epigenetic alterations could serve asiomarker in risk assessment of chemical induced carcinogene-is [14]. A recent review recommended the urgent necessities toharacterize the potential epigenetic alterations induced by engi-eered nanomaterials, in particular, as these compounds possesshe potentiality to modulate the normal epigenetic landscape withheir unique physic-chemical properties [15]. Epigenetic definehe changes in gene expressions without permanently altering theNA sequence and mechanisms include DNA methylation, his-

one modifications, non-coding RNAs expressions and nucleosomeositioning. In particular, the DNA methylation, a covalent modifi-ation, plays a critical role in the maintenance of genome integrityainly by transcriptional silencing [13]. DNA methylation analysisight be an invaluable parameter when used in combination with

he basic in vitro tests, i.e., cytotoxicity and genotoxicity assess-ents and could strengthen the toxicity outcome with additional

ayer of information towards better understanding of possibleechanisms [16,17].

The objective of the current study was to evaluate GFNs inducedpigenetic modification by addressing the global DNA methyla-ion status in conjunction with genotoxicity and expression of DNA

ethylation and DNA damage repair machinery at low to moder-tely cytotoxic doses in human bronchial epithelial (BEAS-2B) cells.

e used five different representatives of GFNs-pristine grapheneanoplatelets (GNP-Prist), carboxylated graphene nanoplateletsGNP-COOH), aminated graphene nanoplatelets (GNP-NH2), singleayer graphene oxide (SLGO) and few layer graphene oxide (FLGO)anomaterials.

. Materials and methods

.1. Graphene nanomaterials (GFNs)

The commercially available SLGO, FLGO (4–8 layers) and GNPsith different surface functionalization [GNP-Prist, GNP-NH2 andNP-COOH] were purchased from Cheap Tubes Inc. (Brattleboro,T, USA) in powdered and prepared stock in distilled water at000 mg/L with sonication. In addition the information suppliedy the manufacturer, the GFNs were characterized by using atomicorce microscopy (AFM) and Raman spectroscopy. Surface topog-aphy, height profile and lateral size distribution of the GFNs werexamined by AFM (Park Systems XE-BiO) in non-contact mode.aman spectroscopy was performed at room temperature with aicro Raman system (UniRAM3500, UniNanoTech Co., Ltd., Korea)ith a 532 nm laser. The calibration was initially made using an

nternal silicon reference at 500 cm−1 and gave a peak-position res-lution of less than 1 cm−1. The spectra were measured from 0 to

000 cm−1. The size distribution and �-potential of the MWCNTs30 mg/L in Dulbecco modified eagle medium (DMEM/F12) culture

edia] were evaluated by using a photal dynamic light scatteringpectrometer (DLS) (ELSZ-1000, Otsuka Electronics Co., Inc.).

esearch 798 (2016) 1–10

2.2. Cell culture and GFNs treatment

BEAS-2B cells (human bronchial epithelial cells) were culturedin DMEM/F12 (GIBCO), supplemented with 10% (v/v) fetal bovineserum and 1% (v/v) antibiotics, at 37 ◦C in a 5% CO2 atmosphere.

The GFNs were freshly prepared in cell culture medium(DMEM/F12) at the desired concentrations with appropriateamount from the stock (1000 mg/L in distilled water) and was son-icated for 10 min before biological exposure.

2.3. Comet assay

BEAS-2B cells (2.5 × 104 cells/mL) were seeded in 6 well platesbefore 24 h of treatment and treated with GFNs at 10 and 50 mg/Lfor next 24 h. After treatments alkaline (pH > 13) comet assay wasperformed as following the standard method as described else-where [10]. Briefly, after the exposure the cells were trypsinisedand centrifuged at 1500 rpm for 3 min. 1–3 × 104 cells wereresuspended in 0.5% low-melting-point agarose (LMPA, Bio-RadLaboratories, Hercules) at 1:2 ratio. The resuspended cells inagarose were put onto slides pre-coated with 1% normal-meltingagarose. After solidification, the slides were immersed in coldlysing solution (2.5 M NaCl, 100 mM EDTA, 10 mM Tris-base, 1%N-laurolsarcosinate, 1% Triton X–100) for 1.5 h at 4 ◦C, after whichthey were transferred to an electrophoresis tank containing freshlymade electrophoresis buffer (1 mM EDTA, 300 mM NaOH; pH > 13)and were kept for 20 min at room temperature to allow DNAunwinding. Electrophoresis was performed in the same bufferat room temperature for 20 min at 25 V and 300 mA. The slideswere then neutralized three times with 0.4 M Tris-chloride buffer(pH 7.5), air-dried, and fixed in 70% ethanol. The slides werestained with 50 �L ethidiumbromide (5 �g/ mL) and analyzed usinga fluorescence microscope (Leica DM IL) at 40× magnification.Approximately, 50 cells per slide (3 slides per treatment) wereexamined. DNA damage was expressed as the tail extent momentusing an image analysis by the Komet 5.5 software (Kinetic ImagingLtd.).

2.4. Total RNA extraction and quantitative real-time PCR(qRT-PCR)

Total RNA from GFNs treated samples (10 and 50 mg/L 24 h)were extracted by using RNA extraction kit (NucleoSpin, Macherey-Nagel) and the quantity and quality of RNAs were detected in Nanodrop as well as with agarose gel separation.

Synthesis of cDNAs was performed by a reverse transcriptase(RT) reaction and PCR amplification was carried out with a ther-mal cycler (Bio-Rad). Real time RT-PCR analysis was accomplishedwith CFX manager (Bio-Rad) using the IQTM SYBR Green SuperMix(Bio-Rad). The primers were constructed (by Primer3plus) based onsequences available in NCBI and the qRT-PCR conditions were opti-mized (efficiency and sensitivity tests) for each primer prior to theexperiment (Supplementary Table S1). Three biological replicateseach with triplicates were used for each qRT-PCR analysis. Analy-sis of negative control reactions (minus RT and all reagents minustemplate) confirmed no DNA contamination. The gene expressionswere normalized by using GAPDH and ACTB as housekeeping genes.

2.5. Western blotting

BEAS-2B cells were exposed to 10 mg/L of GFNs for 24 h and afterharvesting cell extract were prepared in NP40 protein extraction

buffer and the protein concentrations were measured by Brad-ford method. Equal amounts (50 �g) of proteins were separatedon 15% SDS-PAGE gels and transferred to nitrocellulose mem-branes. The membranes were blocked with 3% BSA in TBST at room

tion R

tawtwi

2

wNdiRmfaiO%p

2

m(sa

3

3

anm(siobiafeoefTSi

3

tesEoc

N. Chatterjee et al. / Muta

emperature. The primary antibody was used at 1:1000 dilutionsnd the secondary antibody was used at 1:2000 dilutions. Blotsere developed using enhanced chemiluminescence western blot-

ing detection kit (Amersham, Little Chalfont). The tested proteinsere �-H2AX and �-Actin (antibodies purchased from Cell Signal-

ng, Beverly).

.6. Global DNA methylation assay

Total DNA from GFNs treated samples (10 and 50 mg/L 24 h)ere extracted by using DNA extraction kit (NucleoSpin, Macherey-agel) and the quantity and quality of DNAs were detected in Nanorop. Next, global DNA methylation assays were carried out accord-

ng to manufacturer’s instructions (5-mC DNA ELISA kit, Zymoesearch Corp.). Briefly, 100 ng of denatured DNA was added in 5-ethylcytosine (5-mC) coated well and incubated at 37 ◦C at dark

or 30 min. Next, the plate was incubated for another 1 h at 37 ◦Cfter adding the antibody mix (anti-5mC and secondary antibodyn provided ELISA buffer) to each well. Thereafter, absorbance atD 450 nm was measured after adding HRP color developer and

5-mC calculated from standard curve. Appropriate negative andositive controls were used to prepare the standard curve.

.7. Statistical analysis

The statistical significance of differences among/between treat-ents was determined using one way analysis of variance

ANOVA). This was followed by a post-hoc test (Tukey, P < 0.05). Alltatistical analyses were carried out using SPSS 12.0KO (SPSS Inc.)nd graphs were prepared in SigmaPlot (Version 12.0).

. Results and discussions

.1. GFNs characterization

The commercially purchased GFNs (>99% purity) were, further,nalyzed with AFM, and Raman spectroscopy. The details of layerumber(s), thickness, dimensions, functionalization, etc. are sum-arized in Table 1. We used pristine and two other functionalized

COOH and NH2) GNPs of similar layer numbers, lateral dimen-ions, average thickness and two kind of GOs which only differn layer number—SLGO (single layer) and FLGO (∼4–8 layers) butf similar dimensional size. The characteristic graphene bands (Gand ∼1580 cm−1 and D band ∼1350 cm−1) [9,10] were exhibited

n all types of GFNs attested by Raman spectroscopy (Table 1). Theggregation behavior and colloidal stability of GFNs under two dif-erent time point (0 and 24 h) was measured simply with visualvidence of settling in DMEM culture media at the concentrationf 20 mg/L. Excellent colloidal dispersion ability and stability werevident for SLGO & FLGO even after 24 h. On the contrary, GNPsormed observable aggregates and precipitation after 24 h (Table 1).he order of measured �-potential in culture media after 24 h wasLGO > FLGO > GNP-COOH > GNP-NH2 > GNP-Prist which was alson agreement with visual dispersion stability (Table 1).

.2. Cytotoxicity and dosimetry

In our previous study it has shown that GNPs exhibited higheroxicity than GOs (SLGO & FLGO) in BEAS-2B cells [18]. Based onffective concentrations (ECs) of the GFNs (Table S2), we have cho-

en 10 mg/L (less than EC10 of each compound) and 50 mg/L (aroundC20 of the GOs and less the EC50 of GNPs) doses for the evaluationf geno-epigenotoxic potentialities of the GFNs at low to moderateytotoxicity levels.

esearch 798 (2016) 1–10 3

3.3. DNA damage potentiality of GFNs

The possible genotoxic abilities of GFNs were evaluated at twolevels—the alkaline comet assay for the combinational detectionof single stranded as well as double stranded DNA damage whilethe activation of phosphorylated �-H2AX for double stranded DNAdamage potentialities. The Fig. 1 clearly exhibited that the GFNspossess genotoxic potentiality. The GFNs induced increase in DNAdamages were observed in both treated doses (10 mg/L and 50 mg/Lfor 24 h) and significant dose dependency was found in SLGO(p < 0.034) (Fig. 1A) and the order of DNA damage was observedin comet assay (the ‘olive tail moment’ numerical value) as GNP-Prist (5.0675) ≥ GNP-COOH (4.8916) > GNP-NH2 (4.2512) ≥ FLGO(2.9748) > SLGO (2.5338) at the dose of 10 mg/L. On the other hand,the GNPs, particularly GNP-Prist, were found as more potent dou-ble strand break inducer than SLGO/FLGO (Fig. 1B). The orderof double strand break potency between GFNs were found asGNP-Prist > GNP-NH2 > GNP-COOH > FLGO ≥ SLGO (Fig. 1B). In gen-eral, GNPs were found to possess more genotoxic ability thanSLGO/FLGO.

The GFNs were reported as genotoxic in BEAS-2B cells [19],flake-size dependent genotoxicity of GOs were observed in A549,CaCo2 and vero cell lines [20], rGONPs showed genotoxic effectson the human mesenchymal stem cells (hMSCs) through DNAfragmentations and chromosomal aberrations [9]. In particular,Hinzmann et al. demonstrated the differential modification depen-dent genotoxicity of GFNs, i.e., pristine graphene, rGO, graphite,caused DNA damage in U87 cells, whereas GO was found as nongenotoxic in the same system [12].

Possibly the platelet like structural shape [3] and hydrophobicagglomeration [2] of the GNPs caused physical damage in the cellmembrane and in turn direct/indirect DNA strand break. In addi-tion, layer number dependent effect between SLGO and FLGO (fore.g., higher dose dependency of SLGO), was observed as becauseof the stiffness increased with increase in layer numbers and thuspossible lower biological interactions in FLGO than that of SLGO [2].

Moreover, the hydrophilicity/hydrophobicity of GFNs, mainlygoverned by the surface functionalizations, and their agglomera-tion behavior in the dispersion medium possibly play the pivotalrole in differential mode of genotoxicity of GFNs [10,21]. The oxida-tive functionalization and in turn the hydrophilicity and higherbiocompatibility makes GOs less toxic than GNPs and among theGNPs, GNP-Prist was found as most genotoxic one (particularly atthe lower dose—10 mg/L), possibly because of higher agglomera-tion in exposure medium.

Interestingly, more profound DNA single break was observedthan the double strand break in almost all GFNs treatment, exceptGNP-Prisitne at the same dose (10 mg/L) (Fig. 1A and B). GFNs werereported to not only interact preferentially with single strandedDNA/RNA over double stranded (ds) forms but also give steric pro-tection of adsorbed nucleotides from attack by nuclease enzymes[2,22]. The fundamental reason for the preferential binding ofssDNA to GFN, particularly GOs, in respect to dsDNA was explainedas the role of GFN-base interactions in ssDNA adsorption. DNA is apolyanion and is electrostatically repelled from GFN, but the DNAbases can bind to graphenic surfaces through hydrophobic forcesand �–� stacking [18].

Next, we sought the effects of GFNs on DNA damage repairsystem not only because it is well known biomarker of geneticinstability [23] but also to evaluate the whether GFNs’ possessdirect hindering effects on nuclease enzymes [22].

3.4. Effects of GFNs on the DNA-damage repair systems

To maintain genomic stability diverse types of DNA repair path-ways have evolved which remove various types of DNA lesions

4 N. Chatterjee et al. / Mutation Research 798 (2016) 1–10

Table 1Characterization of graphene family nanomaterials.

Properties SLGO FLGO GNP-Pristine GNP-COOH GNP-NH2

Thickness (nm) 21 122 877.2 735.9 945.5Lateral size distribution (�m) 10 10 10 9.98 10Layer number(s)a 1 4–8 <4 <3 <4Functionalization Surface oxygen Surface oxygen None COOH, C OH, C O NH2, N H, O C N H2, C N

Graphene peaks with Raman spectroscopy

Zeta potential (mV) measured with DLS −8.98 ± 0.55 −9.33 ± 0.45 −14.28 ± 0.66 −9.86 ± 0.7 −10.55 ± 1.21

Dispersibility and agglomeration potentiality

a Information supplied by the manufacturer.

F amage5 xpress

wt(re

ig. 1. DNA damage potentiality of GFNs in Beas2B cells. (A) Single stranded DNA d0 mg/L doses. (B) Double stranded DNA damage measured with �-H2AX protein e

ith high substrate specificity, such as, mismatch repair system

o correct the mispaired DNA bases from newly synthesized DNAreplication error repair), base excision repair (BER) pathway toemove oxidative DNA lesions, nucleotide excision repair (NER) toliminate bulky base adducts and homologous recombination (HR)

measured with alkaline comet assay (presented with olive tail moment) at 10 andion at 10 mg/L. Data presented as mean ± SEM.

or non-homologous end joining (NHEJ) processes specifically for

DNA double strand breaks repair [24]. Therefore, the alterations inDNA repair genes expressions could shed light on the particularthe type of DNA lesion as well as the respective repair pathwayinduced by the external stimuli. We have chosen representative

tion R

(((d(a

Frg

N. Chatterjee et al. / Muta

one/two) DNA repair genes from each group, such as, MSH2 & MLH1replication error repair), OGG1, NTHL1, MTH1 (BER), ERCC1 & XPANER), RAD51 (HR), LIG4 & XRCC4 (NHEJ), most of which displayed

eregulation due to GFNs exposure, specifically at higher dose50 mg/L) (Fig. 2). Mismatch repair genes were not significantlyffected, except MSH2 at 50 mg/L of SLGO-FLGO exposed conditions

ig. 2. Alterations of DNA repair genes expressions due to GFNs exposures at 10 and 50 mgepair (BER) genes (OGG1, MTH1 & NTHL1) expressions. (C) Nucleotide excision repair (enes (LIG4 & XRCC4) expression. (E) Homologues recombination repair gene (RAD51) ex

esearch 798 (2016) 1–10 5

(Fig. 2A). Among the BER genes tested, the MTH1 genes showedmarked up-regulation in all GFNs, particularly with GNP-Prist treat-ment at the dose of 50 mg/L (Fig. 2B). Up regulation of MTH1 gene

was possibly due to the need of elimination of high amount oxi-dized guanosine (8-oxo-dGTP) in GFNs exposed BEAS-2B cells. TheNTHL1 enzyme possess specificity to these substrates such as, 5-

/L doses. (A) Mismatch Repair genes (MLH1 & MSH2) expressions. (B) Base excisionNER) genes (ERCC1 & XPA) expressions. (D) Non-homologous end-joining (NHEJ)pression. Data presented as mean ± SEM.

6 N. Chatterjee et al. / Mutation Research 798 (2016) 1–10

it Con

Oau

Fig. 2. (

HU, 5-hydroxycytosine, formamidopyrimidine derivatives of And G (FapyA and FapyG) and thymine glycol [25]; hence, thep-regulation of NTHL1 by GNP-Prist and GNP-NH2 treatment at

tinued).

50 mg/L (Fig. 2B) shed light on the possibility of formation of variousoxidized DNA lesions (5-OHU, 5-hydroxycytosine, FapyA/FapyG orthymine glycol other than 8-oxo-dGTP. The NER, specifically XPA

N. Chatterjee et al. / Mutation R

F5

g(sis

FDD

ig. 3. The changes in % Global DNA methylation due to GFNs exposures at 10 and0 mg/L to Beas2B cells. Data presented as mean ± SEM.

ene, pathway showed marked alteration due to GFNs treatmentFig. 2C). Interestingly, XPA displayed opposite mode of changes,

uch as suppression in SLGO-FLGO exposure while up-regulationn GNPs. This data is actually in agreement with our previoustudy conducted with GO and rGO exposed HepG2 cells which also

ig. 4. Alterations of DNA methylation specific genes expressions due to GFNs exposureNMT3B) expressions. (B) Genes expressions of DNA methyl-CpG binding domain proteinata presented as mean ± SEM.

esearch 798 (2016) 1–10 7

showed high repression of XPA gene [10]. The modulations of NERpathway shed light on the fact of possible formation of bulky DNAadduct by GFNs treatment. Several studies documented that thebase excision or nucleotide excision ultimately raise double strandbreak if not repaired. So, in our next step we sought for doublestrand break repair with recombination processes. We found highlysignificant changes in tested NHEJ pathway related genes, LIG4 andERCC4 (Fig. 2D), while we did not find any alteration in RAD51gene expressions, the marker of HR (Fig. 2E). The mode of changesof LIG4 followed a similar dose-dependent suppression pattern inall GFNs treatment (Fig. 2D); conversely, SLGO-FLGO displayed adose-dependent down-regulation in XRCC4 but up- regulation byGNPS, particularly at higher dose (Fig. 2D). Hence, error prone NHEJrepair processes possibly play the main role GFNs induced dou-ble strand DNA damage. In general, suppression of the DNA-repairgene expressions could be because of direct inhibitory effects ofGNFs (particularly by GOs), whereas up-regulation of DNA-repairgene expressions was possibly due to indirect activation effect forremoving the GFNs’ induced DNA lesions.

3.5. Changes in global DNA methylation status and its regulatorgenes

DNA methylation, one of the primary epigenetic modification,regulates the chromatin organization and gene expressions, thusplays intricate role not only in embryonic development, genomic

s at 10 and 50 mg/L doses. (A) DNA methyltransferase genes (DNMT1, DNMT3A &s (MBD1, MBD2 & MECP2). (C) DNA demethylases genes (TET1 & TET2) expression.

8 N. Chatterjee et al. / Mutation Research 798 (2016) 1–10

(Cont

iedtr(5ot

Fig. 4.

mprinting, X chromosome inactivation etc. but also in various dis-ases, including carcinogenesis, neurodegerations, cardiovasculariseases etc. [15,26,27]. DNA methylation, the process of addi-ion of a methyl group to carbon five in the cytosine pyridineing resulting in the formation of 5-methylcytosine (5mC) in DNAwriters) is regulated by number of enzymes, such as, binding to

mC (readers), and removing 5mC (erasers) [26,28]. The ‘writers’f 5mC which establish and maintain the DNA methylation pat-erns of the genome are DNA methyltransferases (DNMT1, DNMT3A

inued)

and DNMT3B). In particular, DNMT1 is the maintenance methylasespecifically works on hemimethylated DNA strands, while de novomethylation, the establishment of new methylation marks, are car-ried out by DNMT3A and DNMT3B [27,28]. The ‘readers’ of 5mCare methyl-CpG binding protein 2 (MECP2) and methyl-CpG bind-ing domain proteins (MBD1-4), bind to methylated DNA and cause

transcriptional repressions [28]. The demethylation (the ‘erasers’of 5mC) could occur either passively by triggering the inhibition ofDNMTs, or as actively through the process of hydroxylation of 5mC

N. Chatterjee et al. / Mutation Research 798 (2016) 1–10 9

Table 2Summarization of differential effects of GO vs GNPs.

Toxicity end points Graphene oxide (SLGO and FLGO) Graphene nanoplatelets (GNP-Pristine,GNP-COOH & GNP-NH2)

Reference

Cytotoxicity Less than GNPs Higher than GOs Table S2 (supplementarymaterials)

DNA damage Less than GNPs, but marked dosedependency in SLGO

Higher than GOs Fig. 1A

DNA repair i) SLGO showed OOG1 activationii) XPA down regulationiii) ERCC1 no changeiv) XRCC4 suppressed

i) NTHL1 activation in GNP-Pristineand GNP-NH2

ii) XPA up regulationiii) ERCC1 up regulationiv) XRCC4 up regulated

Fig. 2B–D

% Global DNA methylation Hyper-methylation Hypo-methylation Fig. 3DNA methylation machinery i) DNMT1 up regulation trend (only

significant in SLGO)ii) Dose dependent up regulation of

DNMT3Biii) MBD1 up regulated

i) DNMT1 down regulation trend (onlysignificant in GNP-NH2)

ii) Dose dependent down-regulationof DNMT3B

iii) MBD1 down-regulation

Fig. 4A and B

tt

lcplmcNGodpttwGgradmos5wMoeg(a(icmDpboiiD

iv) MECP2 up regulation trend (onlysignificant in SLGO)

o 5-hydroxymethylcytosine (5-hmC), which are catalyzed by theen-eleven translocation (TET1-3) proteins [27,28].

In our study, the GFNs induced alterations of global DNA methy-ation levels showed very unique pattern which markedly exhibitedompounds type specificity with differential physical–chemicalroperties (Fig. 3). Dose dependent increases in global DNA methy-

ation, hypermethylation, were observed in SLGO and FLGO butore strongly in SLGO (p < 0.031). Conversely, GNPs treatment

aused demethylation following the order as GNP-COOH > GNP-H2 ≥ GNP-Prist and the dose dependency were most significant inNP-NH2 (p < 0.039) (Fig. 3). It was documented that hypoxia [29],xidative stress [30], persistent organic pollutants [31], endocrineisruptors [32], smoking [33], air pollutants [15], metallic com-ounds [28] can manipulate DNA methylation status. To delineatehe differential response of GFNs on global DNA methylation sta-us, we next sought for the expressions of regulator genes, theriters, readers and erasers of 5mC (Fig. 4A–C) in two doses ofFNs treatments. Among the ‘writers’ of 5mC genes, the DNMT3Bene exhibited significant alterations by GFNs which is truly cor-elated with the global DNA methylation alteration pattern, suchs increased by SLGO-FLGO and suppressed in GNPs exposed con-itions. In addition, a dose dependency was also evident in theode of DNMT3B gene expressions (Fig. 4A). Besides DNMT3B,

nly SLGO showed increase in DNMT1, DNMT3A gene expres-ions, particularly at higher dose (50 mg/L). In case of ‘readers’ ofmC genes, only MBD1 displayed marked changes in expressionshich was in accordance with DNMT3B expressions (Fig. 4B). TheECP2 gene was found significantly modified only in higher doses

f SLGO (up-regulated), GNP-COOH, GNP-NH2 (down-regulated)xposures, while, no notable changes were observed in MBD2ene expressions, except SLGO-FLGO treated conditions at 50 mg/LFig. 4B). Unlike ‘writers’ and ‘readers’ of 5mC, no remarkable alter-tions were found in ‘erasers’ (TET1 and TET2) gene expressionsFig. 4C). On the whole, the passive mode of DNA demethylation,.e., the decrements of DNMT3B and MBD1 genes are probably theause of global hypomethylation induced by GNPs, while de novoethylation mediated by up-regulation of DNMTs, particularlyNMT3B, and in coordination of increased MBD1/2 genes are theossible reason behind the global DNA hypermethylation inducedy SLGO and FLGO. Though several studies, including epidemi-

logical one, reported modifications of DNA methylation status

nduced by environmental pollutants [34], nonetheless, only lim-ted reports published regarding engineered nanoparticles inducedNA methylation [35–37]. In particular, the SiO2 nanoparticles

iv) MECP2 down regulation trend

induce global DNA hypomethylation in accordance with DNMT1,DNMT3a and MBD2 gene expression decrements in HaCaT cells [35].Though the gene specific DNA methylation alterations were notperformed in the present study but it is possible that global DNAmethylation changes could possess long-term effects on DNA repairgene expression programming which in turn may affects the DNAdamage repair systems.

On the whole, our present work supports the hypothesis [2,38]that multiple graphene forms possess unique physicochemicalproperties and structural specificity; hence, distinct effects wereevident between graphene oxide (SLGO and FLGO) and GNPs ongenome as well as on epigenome. The differential effects of GFNs(GO vs GNPs) are summarized in Table 2.

4. Conclusion

In summary, the genotoxicity and alterations of globalDNA methylation potentiality of GFNs depends on compoundstypes’ specificity with differential physicochemical properties.The assessment of DNA methylation status, particularly withdose–response relationships, at early screening stages couldstrengthen the outcome of cytotoxicity and genotoxicity assess-ments. Moreover, this additional layer of information couldcontribute towards the better understanding of possible mech-anisms of toxicity and ranking of compounds based on theirlong-range damage potentiality without changing the DNAsequence. We believe that this information would eventually fulfillknowledge gap between the GFNs production and their prospectiveuse in biomedical field as well as in environmental human healthexposure and risk assessment context.

Conflict of interest

None.

Acknowledgements

This work was supported by Mid-career Researcher Program

(2013R1A2A2A03010980) and Basic Science Research Program(2013R1A1A2057534) through the National Research Foundationof Korea (NRF) funded by the Ministry of Science, ICT and FuturePlanning and Ministry of Education, respectively.

1 tion R

A

t0

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

0 N. Chatterjee et al. / Muta

ppendix A. Supplementary data

Supplementary data associated with this article can be found, inhe online version, at http://dx.doi.org/10.1016/j.mrgentox.2016.1.006.

eferences

[1] A. Bianco, Graphene: safe or toxic? The two faces of the medal, Angew. Chem.52 (2013) 4986–4997 (International ed. in English).

[2] V.C. Sanchez, A. Jachak, R.H. Hurt, A.B. Kane, Biological interactions ofgraphene-family nanomaterials: an interdisciplinary review, Chem. Res.Toxicol. 25 (2012) 15–34.

[3] A. Schinwald, F.A. Murphy, A. Jones, W. MacNee, K. Donaldson,Graphene-based nanoplatelets: a new risk to the respiratory system as aconsequence of their unusual aerodynamic properties, ACS Nano 6 (2011)736–746.

[4] A.M. Jastrzebska, P. Kurtycz, A.R. Olszyna, Recent advances in graphene familymaterials toxicity investigations, J. Nanopart. Res. 14 (2012) 1320.

[5] C. Chung, Y.K. Kim, D. Shin, S.R. Ryoo, B.H. Hong, D.H. Min, Biomedicalapplications of graphene and graphene oxide, Acc. Chem. Res. 46 (10) (2013)2211–2224.

[6] K.S. Novoselov, V.I. Fal’ko, L. Colombo, P.R. Gellert, M.G. Schwab, K. Kim, Aroadmap for graphene, Nature 490 (2012) 192–200.

[7] J. Hong, N.J. Shah, A.C. Drake, P.C. DeMuth, J.B. Lee, J. Chen, P.T. Hammond,Graphene multilayers as gates for multi-week sequential release of proteinsfrom surfaces, ACS Nano 6 (2012) 81–88.

[8] H. Ren, C. Wang, J. Zhang, X. Zhou, D. Xu, J. Zheng, S. Guo, J. Zhang, DNAcleavage system of nanosized graphene oxide sheets and copper ions, ACSNano 4 (2010) 7169–7174.

[9] O. Akhavan, E. Ghaderi, A. Akhavan, Size-dependent genotoxicity of graphenenanoplatelets in human stem cells, Biomaterials 33 (2012) 8017–8025.

10] N. Chatterjee, H.J. Eom, J. Choi, A systems toxicology approach to the surfacefunctionality control of graphene-cell interactions, Biomaterials 35 (2014)1109–1127.

11] A. Wang, K. Pu, B. Dong, Y. Liu, L. Zhang, W. Zhang, W. Duan, Y. Duan, Role ofsurface charge and oxidative stress in cytotoxicity and genotoxicity ofgraphene oxide towards human lung fibroblast cells, J. Appl. Toxicol. JAT 33(2013) 1156–1164.

12] M. Hinzmann, S. Jaworski, M. Kutwin, J. Jagiello, R. Kozinski, M. Wierzbicki, M.Grodzik, L. Lipinska, E. Sawosz, A. Chwalibog, Nanoparticles containingallotropes of carbon have genotoxic effects on glioblastoma multiforme cells,Int. J. Nanomed. 9 (2014) 2409–2417.

13] V. Bollati, A. Baccarelli, Environmental epigenetics, Heredity 105 (2010)105–112.

14] I. Koturbash, F.A. Beland, I.P. Pogribny, Role of epigenetic events in chemicalcarcinogenesis—ajustification for incorporating epigenetic evaluations incancer risk assessment, Toxicol. Mech. Methods 21 (2011) 289–297.

15] A. Stoccoro, H.L. Karlsson, F. Coppede, L. Migliore, Epigenetic effects ofnano-sized materials, Toxicology 313 (2013) 3–14.

16] R.E. Watson, J.M. McKim, G.L. Cockerell, J.I. Goodman, The value of DNAmethylation analysis in basic, initial toxicity assessments, Toxicol. Sci. J. Soc.Toxicol. 79 (2004) 178–188.

17] M. Szyf, The implications of DNA methylation for toxicology: towardtoxicomethylomics the toxicology of DNA methylation, Toxicol. Sci. J. Soc.Toxicol. 120 (2011) 235–255.

18] N. Chatterjee, J.S. Yang, K. Park, S.M. Oh, J. Park, J. Choi, Screening of toxicpotential of graphene family nanomaterials using in vitro and alternative

in vivo toxicity testing systems, Environ. Health Toxicol. 30 (2015) e2015007.

19] H.K. Lindberg, G.C. Falck, S. Suhonen, M. Vippola, E. Vanhala, J. Catalan, K.Savolainen, H. Norppa, Genotoxicity of nanomaterials: DNA damage andmicronuclei induced by carbon nanotubes and graphite nanofibres in humanbronchial epithelial cells in vitro, Toxicol. Lett. 186 (2009) 166–173.

[

esearch 798 (2016) 1–10

20] L. De Marzi, L. Ottaviano, F. Perrozzi, M. Nardone, S. Santucci, J. De Lapuente,M. Borras, E. Treossi, V. Palermo, A. Poma, Flake size-dependent cyto andgenotoxic evaluation of graphene oxide on in vitro A549, CaCo2 and vero celllines, J. Biol. Regul. Homeost. Agents 28 (2014) 281–289.

21] K.H. Liao, Y.S. Lin, C.W. Macosko, C.L. Haynes, Cytotoxicity of graphene oxideand graphene in human erythrocytes and skin fibroblasts, ACS Appl. Mater.Interfaces 3 (2011) 2607–2615.

22] C.H. Lu, C.L. Zhu, J. Li, J.J. Liu, X. Chen, H.H. Yang, Using graphene to protectDNA from cleavage during cellular delivery, Chem. Commun. (Camb.) 46(2010) 3116–3118.

23] A.R. Collins, L.R. Ferguson, DNA repair as a biomarker, Mutat. Res. 736 (2012)2–4.

24] A. Hartwig, T. Schwerdtle, Interactions by carcinogenic metal compoundswith DNA repair processes: toxicological implications, Toxicol. Lett. 127(2002) 47–54.

25] J.L. Parsons, D.O. Zharkov, G.L. Dianov, NEIL1 excises 3′ end proximal oxidativeDNA lesions resistant to cleavage by NTH1 and OGG1, Nucleic Acids Res. 33(2005) 4849–4856.

26] I.P. Pogribny, F.A. Beland, DNA methylome alterations in chemicalcarcinogenesis, Cancer Lett. 334 (1) (2013) 39–45.

27] N. Okashita, Y. Kumaki, K. Ebi, M. Nishi, Y. Okamoto, M. Nakayama, S.Hashimoto, T. Nakamura, K. Sugasawa, N. Kojima, T. Takada, M. Okano, Y. Seki,PRDM14 promotes active DNA demethylation through the ten–eleventranslocation (TET)-mediated base excision repair pathway in embryonicstem cells, Development 141 (2014) 269–280.

28] J. Brocato, M. Costa, Basic mechanics of DNA methylation and the uniquelandscape of the DNA methylome in metal-induced carcinogenesis, Crit. Rev.Toxicol. 43 (2013) 493–514.

29] C.M. Robinson, R. Neary, A. Levendale, C.J. Watson, J.A. Baugh,Hypoxia-induced DNA hypermethylation in human pulmonary fibroblasts isassociated with Thy-1 promoter methylation and the development of apro-fibrotic phenotype, Respir. Res. 13 (2012) 74.

30] X. Gu, J. Sun, S. Li, X. Wu, L. Li, Oxidative stress induces DNA demethylationand histone acetylation in SH-SY5Y cells: potential epigenetic mechanisms ingene transcription in Abeta production, Neurobiol. Aging 34 (2013)1069–1079.

31] L. Lind, J. Penell, K. Luttropp, L. Nordfors, A.C. Syvanen, T. Axelsson, S.Salihovic, B. van Bavel, T. Fall, E. Ingelsson, P.M. Lind, Global DNAhypermethylation is associated with high serum levels of persistent organicpollutants in an elderly population, Environ. Int. 59 (2013) 456–461.

32] S.V. Fernandez, Y. Huang, K.E. Snider, Y. Zhou, T.J. Pogash, J. Russo, Expressionand DNA methylation changes in human breast epithelial cells after bisphenolA exposure, Int. J. Oncol. 41 (2012) 369–377.

33] A. Semmler, P. Heese, B. Stoffel-Wagner, M. Muschler, A. Heberlein, L. Bigler,J.C. Prost, H. Frieling, J. Kornhuber, M. Banger, S. Bleich, T. Hillemacher, M.Linnebank, Alcohol abuse and cigarette smoking are associated with globalDNA hypermethylation: results from the German investigation onneurobiology in alcoholism (GINA), Alcohol 49 (2015) 97–101.

34] M. Kim, M. Bae, H. Na, M. Yang, Environmental toxicants-induced epigeneticalterations and their reversers, J. Environ. Sci. Health C Environ. Carcinog.Ecotoxicol. Rev. 30 (2012) 323–367.

35] C. Gong, G. Tao, L. Yang, J. Liu, Q. Liu, Z. Zhuang, SiO(2) nanoparticles induceglobal genomic hypomethylation in HaCaT cells, Biochem. Biophys. Res.Commun. 397 (2010) 397–400.

36] X. Lu, I.R. Miousse, S.V. Pirela, S. Melnyk, I. Koturbash, P. Demokritou,Short-term exposure to engineered nanomaterials affects cellular epigenome,Nanotoxicology (2015) 1–11.

37] S.V. Pirela, I.R. Miousse, X. Lu, V. Castranova, T. Thomas, Y. Qian, D. Bello, L.Kobzik, I. Koturbash, P. Demokritou, Effects of laser printer-emittedengineered nanoparticles on cytotoxicity, chemokine expression, reactive

oxygen species, DNA methylation, and DNA damage: a comprehensiveanalysis in human small airway epithelial cell. macrophages, andlymphoblasts, Environ. Health Perspect. 124 (2) (2016) 210–219.

38] A.B. Seabra, A.J. Paula, R. de Lima, O.L. Alves, N. Duran, Nanotoxicity ofgraphene and graphene oxide, Chem. Res. Toxicol. 27 (2014) 159–168.