Ecotoxicity of pristine graphene to marine organisms

Carlo Pretti a,n, Matteo Oliva a, Roberta Di Pietro a, Gianfranca Monni a, Giorgio Cevasco b,Federica Chiellini c, Christian Pomelli d, Cinzia Chiappe d

a Department of Veterinary Sciences, University of Pisa, San Piero a Grado (PI) 56122, Italyb Department of Chemistry and Industrial Chemistry, University of Genoa, Genoa 16146, Italyc Department of Chemistry and Industrial Chemistry, University of Pisa, Pisa 56126, Italyd Department of Pharmacy, University of Pisa, Pisa 56126, Italy

a r t i c l e i n f o

Article history:Received 23 April 2013Received in revised form25 October 2013Accepted 12 November 2013Available online 15 January 2014

Keywords:GrapheneToxicityVibrio fischeriDunaliella tertiolectaArtemia SalinaOxidative stress

a b s t r a c t

The ecotoxicity of pristine graphene nanoparticles (GNC1, PGMF) in model marine organisms wasinvestigated. PGMF resulted more toxic than GNC1 to the bioluminescent bacterium Vibrio fischeri andthe unicellular alga Dunaliella tertiolecta on the basis of EC50 values (end-points: inhibition ofbioluminescence and growth, respectively). No acute toxicity was demonstrated with respect to thecrustacean Artemia salina although light microscope images showed the presence of PGMF and GNC1aggregates into the gut; a 48-h exposure experiment revealed an altered pattern of oxidative stressbiomarkers, resulting in a significant increase of catalase activities in both PGMF and GNC1 1 mg/Ltreated A. salina and a significant increase of glutathione peroxidase activities in PGMF (0.1 and 1 mg/L)treated A. salina. Increased levels of lipid peroxidation of membranes was also observed in PGMF 1 mg/Lexposed A. salina.

& 2013 Elsevier Inc. All rights reserved.

1. Introduction

Graphene, a two dimensional crystalline material constitutedby a single layer of sp2 hybridized carbon atoms arranged in anhoneycomb-like lattice structure, has sparked a considerablescientific interest starting from its discovery in 2004 (Novoselovet al., 2004). Fascinated by the exceptional physical properties(high electronic conductivity, good thermal stability, and excellentmechanical strength), the scientific community has immediatelyenvisaged the possibility to apply this material in strategic fields,such as nanoelectronic (Westervelt, 2008), polymers (Stankovichet al., 2006), supercapacitors (Vivekchand et al., 2008), batteryelectrodes (Paek et al., 2009), printable inks (Wang et al., 2010),antibacterial paper (Dikin et al., 2007) and biomedical technolo-gies (Feng and Liu, 2011). In few years, the number of publicationson graphene is increased exponentially and other related materi-als, including few layer-graphene and ultra thin graphite,graphene oxide, reduced graphene oxide and graphene nanosheets,have been subjected to an intensive investigation. All these materialsconstituted the composite class of graphene-family nanomaterials(GFN) which, analogously to carbon nanotubes, vary in the layersnumber, lateral dimensions, surface chemistry, defect density orquality of the individual graphene sheets and purity. Undoubtedly,

the structural features of GFN significantly affect the physico-chemical properties of the resulting materials and, although thisaspect has not been systematically investigated, probably theydetermine also the main biological effects. The recent biomedicalapplications of graphene and derivatives have determined a rapidincrease of the studies related to the biological interactions of thesematerials (Sanchez et al., 2012), however, as for many other nano-materials, the issue of potential toxicity is related not only to thisspecific application. If researchers and companies believe in thepossibility to apply these chemicals in multiple fields, going fromelectronics to optics, including mechanics and sensors (Soldano et al.,2010), it is reasonable to suppose a future large-scale production ofgraphene and graphene-derived materials. GFN dispersed in airmight represent a danger to people handling these materials ondaily basis, either by contact or inhalation, and studies on this topicare therefore necessary. Nevertheless, accidental spills and effluentdischarges can determine an increased risk of release of theseexogenous nanoparticles (NPs) into aquatic environments and eventhough emissions of GFN to aquatic environment should be low (ifany), their expected low degradability requires adequate investiga-tion. Investigations on the environmental impact of graphene aretherefore mandatory before any large scale application, in agree-ment to the European regulatory framework for the Registration,Evaluation and Authorization of Chemicals (REACH). In particular,the registration process REACH indicates the requirement of (eco)toxicological assessment for all chemicals produced in or importedinto the European Union (today, above one metric tonne per year,

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/ecoenv

Ecotoxicology and Environmental Safety

0147-6513/$ - see front matter & 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.ecoenv.2013.11.008

n Corresponding author. Fax: þ39 0502210182.E-mail addresses: cpretti@vet.unipi.it, carlopretti@gmail.com (C. Pretti).

Ecotoxicology and Environmental Safety 101 (2014) 138–145

but in the future the conditions will be more restrictive). Thepredicted no-effect concentration (PNEC) represents one of themajor indicators required for the environmental hazard assessmentof chemicals starting from the available chemical and ecotoxicitydata, derived from test on organisms, such as bacteria, algae,crustaceans and fish in both acute and chronic toxicity experiments(Pretti et al., 2011).

In this paper we report results about the ecotoxicity of pristine(not-functionalized) graphene. In particular, the effects of differentnanometric particles of two commercial formulation of pristinegraphene, pristine graphene monolayer flakes (PGMF) andgraphene nanopowder grade C1 (GNC1), were studied on modelmarine organisms, in comparison with the corresponding micro-metric bulk material (graphite, GRP).

2. Materials and methods

2.1. Chemicals

Graphite (GRP), the bulk material from which graphene derives and twodifferent commercial forms of graphene, pristine graphene monolayer flakes(PGMF) and graphene nanopowder grade C1 (GNC1), were employed in this study.Graphene was purchased from Graphene Laboratories Inc., Calverton, NY (USA);graphite was purchased from Sigma-Aldrich (St. Louis, MO, USA).

The specifications of these substances, as given by manufacturers, are shown inTable 1. PGMF was ultrapure (no oxidation, no surfactants);499.99% carboncontent.

GNC1 and GRP were suspended in natural seawater (NSW), milli-Q water,buffers or media, as indicated in specific Materials and methods section; PGMF wassuspended in the same manner after ethanol evaporation. NSW (salinity 35 g/L)was oxygen saturated and filtered (0.21 μm). All other chemicals were obtainedfrom Sigma-Aldrich (St.Louis, MO, USA). Freshly opened commercial materials werealways employed.

As the hydrophobic nature of pristine graphene derivatives could lead to theformation of aggregates in water, all exposure experiments were carried out understirring and the use of dispersing agents was avoided (in the case of PGMF ethanolwas removed before dispersion in water) in order to reduce eventual interferingeffects arising from the presence of co-solvents.

2.2. Vibrio fischeri: inhibition of bioluminescence

The inhibition of bioluminescence test was performed according to standardoperating procedure using the Basic protocol (Azur Environmental, 1995), based onthe ISO (2007). The same bacterial (V. fischeri) lot no. 10J1010, exp. date 09/2012,Ecotox LDS, Pregnana Milanese, MI, Italy) was used for all experiments that werecarried out well within the expiration date. Bacteria were obtained as freeze-lyophilized cells in separated vials and always resuspended in NSW.

In order to investigate whether PGMF, GNC1 and GRP (suspended in NSW,maximum concentration of 5 mg/L) could interfere in light emission, several lightemission readings with activated bacteria at zero time were performed. Interfer-ences of graphene nanoparticles on light emission readings were evaluated byadding 100 μl of bacteria to each cuvette containing 1 ml of suspension of theconsidered substance, and reading the light emission every 5 min for 30 min.Relative bioluminescence intensity was expressed in relative bioluminescenceunits (RBU).

Prior to submit the samples to a full test (identification of ecotoxicologicalparameters such as EC20/50), a screening test (maximum % of effect, I%) at maximumconcentration of 5 mg/L was performed. Only substances that showed I%420%were submitted to the full test. Bacteria were exposed to a dilution series of thesample and their light emission was determined after incubation. The lightemission of the bacteria in the samples was measured after 5, 15 and 30 min and

compared to an aqueous control. The tests were performed at 15 1C within the pHoperative range (6–8,) by the use of three replicates and four controls. Allmeasurements were performed by using the M500 luminometer equipped withthe appropriate cells. The instrument was PC interfaced and acquisition and datahandling were performed with the Microtoxs Omni 1.16 software. Zinc sulfateeptahydrate was used as the reference toxicant. Mean EC50 values (three replicatedeterminations) were expressed as mg/L together with confidence limits (95%).

2.3. Dunaliella tertiolecta: inhibition of growth

The inhibition of growth of D. tertiolecta was evaluated according to theprotocol described in ISO procedures (ISO, 1995), with slight modifications. D.tertiolecta strain CCAP 19/27 was purchased from the reference center CCAP(Culture Collection of Algae and Protozoa—Scottish Association for MarineScience/SAMS Research Services Ltd). D. tertiolecta was cultured in F2-medium(NSW supplemented with a salt mix and a vitamin mix, according with Guillardand Ryther, 1962). Late logarithmic phase algae were inoculated in 25 mL freshmedium (50 mL conical flasks) to an initial concentration of 104 cells/mL and weregrown at 2072 1C under cool white fluorescent continuous light of 7000 lx underslow shaking (80 rpm) for 72 h. All cultures were aseptic and bacteria free.Experiments were performed in triplicate. F2-medium acted as control. PGMFand GNC1 were suspended in F2-medium at serial concentrations ranging from10 mg/L to 0.675 mg/L before inoculum. GRP, used as reference bulk material, wassuspended at the concentration of 10 mg/L. Potassium dichromate was used asreference toxicant. The endpoint was the inhibition of growth (n cells/mL) at theend of 72 h. Cells were counted by the use of Scepter 2.0 Handheld Automated CellCounter (Millipore Corporation, Billerica MA USA). EC20/50 values of PGMF andGNC1 were calculated with the Linear Interpolation Method for Sublethal Toxicitysoftware (U.S. EPA, 1993).

2.4. Artemia salina acute toxicity test (24 h)

The hatching of A. salina cysts (Artemia Gold Argentemia) followed theprocedure described in standardised short-term toxicity test (ARC-test) withnauplii (Vanhaecke and Persoone, 1981). The newly hatched nauplii were collectedand 5 nauplii were directly transferred into Petri dishes containing 5 ml of NSWthat acted as control, and NSW plus PGMF, GNC1 or GRP at different concentration,ranging from 10 mg/L to 0.625 mg/L. The plates were sealed, incubated at 25 1C inthe darkness for 24 h under a gentle shaking (80 rpm). The endpoint (immobility/death) was assessed at the end of the test a Zeiss stereomicroscope: a naupliumwas considered to be immobile or dead, if it could not move its antennae afterslight agitation of the water. Potassium dichromate was used as reference toxicant.Three independent experiments were performed.

2.5. Artemia salina 48 h-exposure: oxidative stress

Nauplii from the parental stock cultures were immersed in the testingsolutions. Tests were performed in 250 ml glass flasks covered with parafilm in avolume of 100 ml of testing solution as follow: control (NSW), PGMF (0.1 and 1 mg/L in NSW), GNC1 (0.1 and 1 mg/L in NSW), GRP (0.1 and 1 mg/L in NSW). Tests wereperformed at 2071 1C under a photoperiod of 16 h light/8 h darkness, as reportedby Nunes et al. (2006). Three replicates were used for each concentration. For alldeterminations about 2000 nauplii were used for each replicate. Values of pH,temperature and percentage of dissolved oxygen were measured every 24 h, fortest validation purposes.

After the exposure period, animals were collected on a mesh and homogenizedin ice-cold phosphate buffer (50 mM, pH 7) by sonication (Julabo bath sonicator,10 s) and homogenization with a potter Elvejem. Aliquots of crude homogenateswere stored at �80 1C until the determination of the extent of lipid peroxidation.Crude homogenates were also centrifuged at 15000� g for 10 min and supernatantswere divided into aliquots and stored at �80 1C until used for the enzymaticdeterminations (glutathione peroxidase and catalase). On crude homogenates andsupernatants the protein content was measured by the method of Lowry et al. (1951).

Table 1Commercial graphite and graphene derived materials properties.

Pristine graphene monolayer flakes Graphene nanopowder grade C1 GraphitePGMF GNC1 GRF

Purity (% carbon) 99.9 97 99.9Form Flakes, dispersed in ethanol. powder, flakes powderAverage particle (lateral) size �550 nm 5–25 μm 4150 μmAverage flake thickness 0.35 nm (1 monostrate) 5–30 nm n.a.

C. Pretti et al. / Ecotoxicology and Environmental Safety 101 (2014) 138–145 139

2.6. Lipid peroxidation (LP)

LPO was determined in tissues homogenates using the colorimetric method ofMihara and Uchiyama (1978) by measuring the formation of thiobarbituric acid-reactive substances (TBA-RS). Malondialdehyde (MDA) was used as standard. MDAis a product of LPO that reacting with TBA (thiobarbituric acid) give an adductwhose spectrum is identical with that obtained from MDA standard with TBA in a1:2 M ratio on heating to give a red adduct whose concentration is related to theextent of LPO. The TBARS concentrations were determined spectrophotometrically,utilizing the absorbance at 535 and 520 nm and the molar extinction coefficient of1.56�105 M�1 cm�1. The extent of LPO was expressed as nmol TBA-RS/mg.

2.7. Total glutathione peroxidase (GPx) assay

GPx activity was determined according to Carmagnol et al. (1983) by anindirect spectrophotometric method based on the oxidation of glutathione (GSH)to disulfide (GSSG) catalyzed by GPx, which is then coupled to the reduction ofGSSG back to GSH utilizing glutathione reductase (GR) and NADPH (b-NicotinamideAdenine Dinucleotide Phosphate, reduced). The decrease in NADPH absorbancemeasured at 340 nm during the oxidation of NADPH to NADPþ is indicative of GPxactivity, since GPx is the rate limiting factor of the coupled reactions. The reaction isperformed at 25 1C and pH 8.0, and is started by adding an organic peroxide, tert-butyl hydroperoxide (t-Bu-OOH). This substrate is suitable for the assay since itsspontaneous reaction with GSH is low and it is not metabolized by catalase. Thereaction with tert-butyl hydroperoxide measures the selenium-containingglutathione peroxidase-activity present in the sample. Enzymatic activity isexpressed as nmol NADPH/min/mg protein.

2.8. Catalase (CAT) activity

Catalase activity was determined by measuring the decrease in absorbance at240 nm (Aebi, 1983). The reaction mixture contained 200 μL of supernatant and500 μl of 50 mM phosphate buffer (pH 7) and the reaction was started by theaddition of 300 μl of hydrogen peroxide (30 mM). The decrease of absorbance was

recorded every 15 s up to 3 min. Catalase activity was expressed as μmol H2O2/min/mg protein.

2.9. Average particle size determination

The mean particle size of PGMF and GNC-1 suspensions in NSW at aconcentration of 1 mg/L, was measured by dynamic light scattering (DLS) with aDelsaTM NanoC Particle Size Analyzer (Beckman Coulter). Each analysis wasrepeated four times on freshly prepared suspensions.

2.10. Optical microscopy

D. tertiolecta cells (in untreated NSW or exposed to PGMF and GNC1, both 1 mg/L in NSW) were observed by an Olympus CH2 optical microscope.

2.11. Statistical analysis

All data are represented as mean7standard error (or confidence limits).Oneway ANOVA followed by Dunnett0s test for comparisons against control wereperformed for data analyses (PRISM software, Graphpad Software). Magnitudevalues with pr0.05 are considered statistically significant.

3. Results

3.1. Average particle size determination

Regarding the behavior of graphene NPs in NSW, the mean ofparticle size of PGMF and GNC1 suspensions in NSW was mea-sured by dynamic light scattering (DLS). The hydrodynamicdiameter of the particles was taken from the average of fourindependent measurements. Although DLS characterization can-not reveal the exact size of the investigated graphene particles inaqueous solution because of their anisotropic morphology (Liaoet al., 2011) it can be used to predict their ability to give nano-sizedaggregates in aqueous media under the adopted experimentalconditions. Results of particle size distribution by number %,revealed that the mean hydrodynamic diameter size of PGMFsample is significantly smaller (t-test, po0.05) in comparison tothe value observed for GNC1 sample (Table 2, Fig. 1).

Table 2Values of mean diameter size distribution of PGMF andGNC-1 samples (n¼4, freshly prepared solutions).

Sample Diameter (nm7SD)

PGMF 86.75720.43GNC-1 153.13737.80

Fig. 1. Size distribution of PGMF (A) and GNC1 (B) samples in natural seawater.

C. Pretti et al. / Ecotoxicology and Environmental Safety 101 (2014) 138–145140

3.2. Vibrio fischeri: inhibition of bioluminescence

Preliminary light emission measurements with activated bac-teria at zero time were carried out in order to study the possible

Table 3Acute toxicity assay of graphene-derived materials on Vibrio fischeri.

PGMF GNC1 GRF

I% (CL 95%) 69.86 (65.26–74.46) 33.19 (29.84–36.55) 16.58 (15.20–18.85)I% CV 2.65 4.06 4.32

I% values (endpoint: decrease of bioluminescence) expressed as mean % of effect(inhibition of bioluminescence) in PGMF/GNC1-exposed cells of V. fischeri atmaximum concentration of 5 mg/L (n¼3; exposure duration¼15 min). CL, con-fidence limits; CV, coefficient of variation.

Table 4Bioluminescence assay. EC20/50 values expressed as mg/L of PGMF/GNC1-exposedcells of V. fischeri.

Time(min)

PGMF GNC1

n Mean EC50 (mg/L)(95% CL)

EC50

CVn Mean EC20 (mg/L)

(95% CL)EC20CV

5 3 1.95 (1.92–1.97) 0.49 3 45 n.a.15 3 1.75 (1.6–1.94) 4.41 3 1.57 (1.49–1.64) 2.0530 3 1.92 (1.82–2.03) 2.2 3 1.42 (1.38–1.48) 1.39

Tests carried out at six concentrations starting from 5 mg/L (serial dilutionfactor¼2).Time, exposure duration; n, number of replicates; CL, confidence interval; CV,coefficient of variation; n.a. not available

0

100000

200000

300000

400000

500000

600000

700000

800000

CTRL 0.675mg/L

1.25mg/L

2.5mg/L

5mg/L

10mg/L

CTRL 0.675mg/L

1.25mg/L

2.5mg/L

5mg/L

10mg/L

PGMFGNC1

**

**

****

** **

****

*

n. c

ells

/mL

Fig. 2. Growth inhibition of Dunaliella tertiolecta cells exposed to serial dilution ofPGMF, GNC1 and GRP (72 h incubation). Results are expressed as mean of3 separated experiments (n¼3/each experiment)7SD of number of cells/mL.Means were compared by one-way ANOVA (Dunnett0s test). nsignificantly differentfrom control pr0.05;nnsignificantly different from control pr0.001. Dotted linerepresents the mean (681667720502 cells/mL, n¼3) of GRP at the concentrationof 10 mg/L.

Table 5EC50 values (endpoint: inhibition of growth) expressed as mg/L of PGMF/GNC1-exposed cells of D. tertiolecta.

Time (h) PGMF GNC1 EC50 CV

n Mean EC50 mg/L (95% CL) EC50 CV n Mean EC50 mg/L (95% CL)

72 3 1.14 (0.99–1.58) 5.38 3 2.25 (1.25–3.26) 12.96

Tests carried out at five concentrations starting from 10 mg/L (serial dilution factor¼2). Time, exposure duration; n, number of replicas; CL, confidence limits; CV, coefficientof variation.

Fig. 3. Optical microscope images of D. tertiolecta cells. Panel A: untreated cell.Panel B: GNC1 (1 mg/L)-exposed cell (cell swelling). Panel C: GNC1-exposed cells(cell disruption). Fl: flagella.

C. Pretti et al. / Ecotoxicology and Environmental Safety 101 (2014) 138–145 141

interference of NPs with light emission; results (not shown)demonstrated that there was no significant difference betweencontrol and graphene/graphite suspensions (Dunnet0s test, pr0.05). Results ranged from 93.9971.54 RBU (control) to 95.5571.66 RBU; for this reason in this study we successively applied thestandardized ISO (2007) by the use of the Microtox system, asreported also by Strigul et al. (2009) and García et al., (2011). Theresults of bioluminescence inhibition screening tests (Table 3, max.% effect at maximum concentration) showed that PGMF and GNC1exhibited an I% higher than 20% (69.86% and 33.19%) and werethen submitted to a full test (Table 4). PGMF EC50 values ranged1.75–1.95 mg/L while GNC1 showed EC20 values ranging 1.42-45 mg/L, both in 50, 150 and 300 incubations. These resultsindicated a particle dimension-dependent increasing toxicity indirection PGMF4GNC14GRP.

3.3. Dunaliella tertiolecta: inhibition of growth and opticalmicroscope analysis

A particle dimension-dependent increasing toxicity (PGMF4GNC14GRP) characterized also the inhibition tests carried outon the unicellular marine alga D. tertiolecta. Whereas the bulkmaterial, GRP, showed no significant growth inhibition withrespect to control at highest concentration (10 mg/L), PGMF andGNC1 had a quite similar pattern of inhibition; the LOEC (lowestobserved effect concentration) of PGMF was 0.675 mg/L whileGNC1 showed a LOEC of 1.25 mg/L (Fig. 2). Estimated EC50 valuesare reported in Table 5.

The optical microscope images (Fig. 3) support the hypothesisthat the exposure of D. tertiolecta to pristine graphene causes cellwall disruption. Fig. 3 (panel A) showed control D. tertiolecta. Fig. 3(panel B) showed PGMF-treated cell; a loss of flagella and a cellswelling (probably due to cell wall disruption, panel C) wasobserved. Analogous irreversible cell damages have been previouslyreported (Long et al., 2012) in a recent study referring to the toxicityof carbon nanotubes to Chlorella sp.

3.4. Artemia salina acute toxicity test (24 h)

The acute toxicity tests (24 h) on Artemia salina showed theabsence of toxicity at the maximum concentration of 10mg/L (datanot shown), although the observation of crustaceans by opticalmicroscopy (40� ) at the end of the test showed the presence ofPGMF and GNC1 into the gut of crustaceans (Fig. 4).

3.5. Artemia salina 48 h-exposure: catalase (CAT) activity

Catalase activity (CAT) results for 48 h exposure of A. salina toGRP, PGMF and GNC1 (0.1 and 1 mg/L) are plotted in Fig. 5 (leftpanel). Control group showed an activity of 8.6270.55 μmol/min/mg protein. No statistically difference was observed betweencontrol and GRP, PGMF and GNC1 in 0.1 mg/L-treated groupswhereas, in 1 mg/L-treated groups, PGMF and GNC1 were signifi-cantly different (pr0.01) from control showing a CAT activity of17.5971.73 μmol/min/mg protein and 17.4174.19 μmol/min/mgprotein, respectively.

Fig. 4. Light microscope images of CTRL, PGMF- and GNC1-treated A. salina (24 hexposure, 1.25 mg/L). Arrows indicate the presence of PGMF or GNC1 into the gut.

0

5

10

15

20

25

CRTL

0.1 mg/L 1 mg/L

**

GRF PGMF GNC1 GRF PGMF GNC1

CATALASE

μmοl

es/m

in/m

g pr

ot.

0

50

100

150

200

CRTL

0.1 mg/L 1 mg/L

**

GRF PGMF GNC1 GRF PGMF GNC1

GPx

0

5

10

15

CRTL

0.1 mg/L 1 mg/L

*

GRF PGMF GNC1 GRF PGMF GNC1

LPO

οnm

οles

/min

/mg

prot

.

οnm

οles

/min

/mg

prot

.

Fig. 5. Catalase and total gluthatione peroxidase activities in SN (15000� g) of Artemia salina exposed to 0.1 and 1 mg/L (left and center panel); TBARS content in crudehomogenate of A. salina exposed to 0.1 and 1 mg/L (right panel). Results are expressed as mean7SD of three independent experiments (mmol/min/mg protein). Means werecompared by one-way ANOVA (Dunnett0s test). npr0.05

C. Pretti et al. / Ecotoxicology and Environmental Safety 101 (2014) 138–145142

3.6. Artemia salina 48 h-exposure: gluthatione peroxidase (GPx)activity

Fig. 5 (center panel) showed gluthatione peroxidase (GPx)activity results for 48 h exposure of A. salina to GRP, PGMF andGNC1 (0.1 and 1 mg/L). In this case, no statistically differenceswere observed for GRP and GNC1 with respect to control, whereasPGMF showed a GPx activity significantly different from controlgroup (pr0.01) both at 0.1 and 1 mg/L. The values resulted163.1477.21 nmol/min/mg protein and 174.93715.26 nmol/min/mg protein, respectively.

3.7. Artemia salina 48 h-exposure: lipid peroxidation (LPO)

Fig. 5 (right panel) showed the lipid peroxidation (LPO) levelsdetermined in tissues homogenates using the colorimetric methodof Mihara and Uchiyama (1978) by measuring the formation of(TBARS) in crude homogenate of A. salina exposed to GRP, PGMFand GNC1 at 0.1 mg/L and 1 mg/L. It is noteworthy that nosignificant differences were observed in all treated groups, exceptfor PGMF 1 mg/L that exhibited a TBARS content of 12.3971.5 nmol/mg protein. That significantly differed (pr0.05) fromcontrol values (9.28 nmol/mg protein).

4. Discussion

Several studies have been performed on graphene-familytoxicity (Jastrzębska et al., 2012; Akhavan and Ghaderi, 2012;Feng and Liu, 2011; Liao et al., 2011; Liu et al., 2013), but untilnow the ecotoxicity of pristine graphene has not been the object ofdetailed and thorough investigations and this study was under-taken in order to shed light onto the environmental impact ofthese materials on the aquatic compartment. D. tertiolecta andA. salina were used as model organisms since they also play animportant role in aquatic ecosystems; together with V. fischeri,since bioluminescent bacteria represent a rapid, versatile andeconomical method to explore water contamination. The specificobjectives of this study were (1) to determine the acute effects ofnon-functionalized graphene on aquatic organisms (2) to under-stand if there is a correlation between particles size and toxicity,and (3) eventually, obtain information about toxicity mechanism.

The significance to evaluate the differences between nano- andbulk materials having same composition (e.g., fullerenes comparedwith graphite) has been recently underlined by Handy et al.,2008). The environmental fate and ecotoxicity of engineerednanoparticles may be influenced by a number of structuralfeatures, including particle size and size distribution, solubilityand state of aggregation. After release in the environmentalsystems, most engineered particles are able to aggregate to somedegree. The degree and kinetics of aggregation, as well as the sizerange of the aggregates, depend from the characteristics of theparticles, the characteristics of the environmental system, and theconcentration of the NPs (Tiede et al., 2009).

The toxicity of graphene derivatives versus aquatic species wasinitially evaluated using the bioluminescent bacterium V. fischeri.The inhibition of bioluminescence of V. fischeri it is a widespread,standardized method used to assess the toxicity of chemicals andit has been applied with different types of nanoparticles (NPs):CuO, ZnO and dendrimers; (Mortimer et al., 2008) aluminum NPs;(Strigul et al., 2009) CuO, ZnO and TiO2; (Heinlaan et al., 2008)CeO2, TiO2, Fe3O4 (García et al., 2011). However, since particlesuspensions, color or turbidity may interfere in light emission, theFlash Assay, a modification of ISO (2007) where light emission iscontinuously recorded and each sample act as reference by itself,has been used by some authors (Mortimer et al., 2008; Heinlaan

et al., 2008). Our preliminary determinations on V. fischeri lightemission in presence of graphene demonstrated that there was notany significant interference.

The results of bioluminescence inhibition test indicated aparticle size-dependent increasing toxicity in direction PGMF4GNC14GRP.

It is worth of note that, on the basis of the obtained EC50 valuesfor PGMF taking into account the acute toxicity ranking scaleshowed in Table 6 (IMO, 2002; GHS, 2007), it is possible to classifythis material as moderately toxic.

Since pristine graphene has not any substituent or functionalgroup that may be involved in toxicity mechanisms of V. fischeri (interms of inhibition of bioluminescence), it is reasonable to hypothe-size that the effects can be related to membrane mechanical damages(probably affected by particles dimension) and/or electrostatic inter-actions. Even if no experimental results about effects of graphene onV. fischeri have been reported, some authors have described effects ofother carbon-based NPs on the metabolic activity and survival ofother bacterial species (Akhavan and Ghaderi, 2012) showing abactericidal effect of graphene oxide nanowalls. Measuring the effluxof cytoplasmic materials it was found that the cell membranedamage, caused by direct contact of bacteria with the extremelysharp edges of the nanowalls, was the effective mechanism ofinactivation. In this regard, the Gram-negative Escherichia coli bac-teria with an outer membrane were more resistant to cell membranedamage caused by nanowalls than the Gram-positive Staphylococcusaureus lacking the outer membrane.

In the growth inhibition test of the unicellular marine algaD. tertiolecta, both pristine PGMF and GNC1 showed an increasedeffect in term of inhibition of growth in comparison with the bulkmaterial GRP. Like observed in V. fischeri acute toxicity test, EC50values indicated a particle size-dependent increasing toxicity indirection PGMF4GNC14GRP.

As reported in the toxicity ranking scale (Table 6) PGMF andGNC1 appeared to be moderately toxic for D. tertiolecta, althoughthe mechanism of toxicity is still uncertain. Even if pristinegraphene is largerly different from other carbon-based nanoma-terials, we have considered useful to report the results and thehypothesis of Wei et al. (2010) that observed in multiwalledcarbon nanotubes-D. tertiolecta-treated cells an inhibition of algalgrowth and alterated functions in PSII photochemical process andcellular glutathione redox status; authors hypothesized a toxiceffect due to both abiotic and biotic factors. Abiotic factors such ashigh concentrations of cationic ions in seawater, ionic strength andpH, and biotic factors such as organic matter and secretion of cellmetabolites that bridging the nanoparticles, could affect the stateof aggregation of nanoparticles exhibiting toxic effects of differentintensities in different cellular compartments.

In absence of available toxicity data on D. tertiolecta exposed topristine graphene, we reported some informations about thedegree of algal growth inhibition caused by the exposure todifferent NPs. The observed EC50 of 1.14 mg/L (PGMF) and2.25 mg/L (GNC1) in D. tertiolecta 72-h growth inhibition test

Table 6Adapted from “Revised GESAMP rating scheme foracute aquatic toxicity (IMO, 2002)”.

Description EC/IC50 (mg/L)

Non-toxic 41000Practically non-toxic 4100–r1000Slightly toxic 410–r100Moderately toxic 41–r10Highly toxic 40.1–r1Very highly toxic 40.01–r0.1Extremely toxic r0.01

C. Pretti et al. / Ecotoxicology and Environmental Safety 101 (2014) 138–145 143

was similar to those reported (Wei et al., 2010) for multiwalledcarbon nanotubes (EC50 values of 0.82 mg/L), although in this casethe used nanotubes were carboxylated to favor their waterdispersion. On the other hand, Blaise et al. (2008) showed thatin the freshwater green alga Pseudokirchneriella subcapitata singlewalled carbon nanotubes exhibited an IC25 of 1.04 mg/L. Whenthese values are compared with the few reported for inorganicnanoparticles, such as TiO2 with mean size 21 nm and watersoluble CdTe quantum dots (size 3.5–4.5 nm), for which the 72-hgrowth inhibition EC50 values were 10 mg/L and 5 mg/L for thefreshwater green alga Chlamydomonas reinhardtii, (Wang et al.,2008) pristine graphene appeared more toxic than TiO2 andquantum dots nanoparticles on algal growth. Since these experi-ments were carried out under different conditions, we cannotexclude that the continuous shaking used in our investigation canfacilitate the direct contact between algae cells and grapheneaggregates, thus increasing the toxicity.

In the acute toxicity test (24 h) on A. salina pristine graphenerevealed absence of toxicity at maximum concentration of 10 mg/L, even if the presence of swallowed graphene was detected intothe gut. Other nanoparticles in other crustaceans species revealedthe same toxicity trend. Heinlaan et al. (2008) observed that asuspensions of nano and bulk TiO2 were not toxic even at 20 g/Ltoward two freshwater crustaceans (Daphnia magna and Thamno-cephalus platyurus); in A. salina survival-larvae rates in suspen-sions bearing Pb(II), PbO2, and expandable-clay minerals, werefound to be high. In the same experiment, microscope observa-tions for A. salina before and after exposure to small-sized leadoxides revealed that Pb particles were swallowed, but caused nomortality (Cornejo-Garrido et al., 2011).

The uptake of aggregates of nanoparticles via filtration, whichleads to a much higher body concentration than the surroundingwaters, may represent the first step for the introduction ofnanoparticles in the food chain. Due to their chemical natureand large surface areas graphene NPs have the potential to carrytoxic compounds, such as lipophilic pollutants and heavy metals,causing possible toxic effects in humans and animals.

Also in the case of single- and multiwalled carbon nanotubes itwas observed that they accumulate into the gut of Daphnia magna(Edgington et al., 2010; Petersen et al., 2009). In the specific case ofgraphene, the aggregation in a living system (Caenorhabditiselegans) has been recently reported by Zanni et al. (2012). Otherauthors (Liu et al., 2013), in a drug delivery experiment, showedthe presence of graphene oxide complexes integrating anti-cancerdrugs in zebrafish embryos, without presence of sublethal toxicity.

On the other hand, since the presence of PGMF and GNC1 intothe gut could exert long time effects, we also performed also 48 hexposure experiments to verify possible oxidative stress effects.Although no information about such effects for graphene onaquatic organisms are available, the ability of nanoparticles,including carbon materials (Fang et al., 2007; Kang et al., 2008;Li et al., 2008; Lyon and Alvarez, 2008) to contribute to theoxidative stress of cells has been reported (Unfried et al., 2007).The origin of the oxidative stress is, however, debated. Reactiveoxygen species (ROS) production and subsequent damage ofcellular components have often been considered (Kang et al.,2008) as the cause of oxidative stress although it has been alsoevidenced that oxidative stress and cytotoxicity can originate fromROS independent mechanisms. The direct contact between nano-particles and cells has been indeed postulated to be the sole causeof these phenomena (Fang et al., 2007; Li et al., 2008).

The oxidative stress experiments revealed therefore that asignificant increase of CAT characterize both PGMF and GNC11mg/L treated A. salina; a slight but significant induction of GPxis present in PGMF (0.1 and 1 mg/L) treated A. salina and increasedlevels of lipid peroxidation of membranes, expressed as TBARS, can

be evidenced in PGMF 1 mg/L exposed A. salina. An inverse relationbetween graphene nano-dimension and levels of oxidative stresscan be extrapolated from these data. Furthermore, since all thesematerials are not functionalized, it is possible to hypothesize thatany toxicity mechanism should be related to the mechanicalinteraction of nanoparticles with cellular structures.

Other papers evidenced an increase of toxicity of nanostruc-tures respect to their bulk material, but the mechanisms of toxicitywas probably due to chemical-biochemical interactions, in termsof increased bioavailability due to the presence of hydrophilicgroups (hydroxylic, carboxylic) and ions. As example, Almeidaet al. (2007) and Buffet et al. (2011) studied oxidative stressbiomarkers in bivalves and other invertebrates exposed to Cueither as CuO NPs or soluble Cu. Independently of any nanotoxi-city, the Cu toxicity may be caused at least partly by the generationof reactive- oxygen species. CAT and GST activities increasedsignificantly in both species (S. plana, H. diversicolor) and SOD inS. plana exposed to CuO NP, suggesting an oxidative stress enduredby animals.

5. Conclusions

In conclusion, from this study it appears that the pristinegraphene NPs are moderately toxic (ranking scale IMO, 2002) toV. fischeri and D. tertiolecta, also in absence of bounded functionalgroups. PGMF is more toxic than GNC1, thus showing that toxicityincreases as nanoparticle sizes decrease, as also confirmed by theDLS measurements on NSW particles aggregates. With respect toaquatic crustacean A. salina, no acute toxicity was registered, evenif biomarkers of oxidative stress revealed an altered pattern and acertain level of oxidative damage that can be interpreted as anearly warning signal of damage at higher levels of biologicalorganization.

Acknowledgment

The authors thank the University of Pisa for financial support.

References

Aebi, H.E., 1983. Catalase. In Bergmeyer, H.U., Bergmeyer, J., Grassl, M.V.C.H.,Proceedings of the Method of Enzymatic analysis. Weinheim, Germany-Deer-field, FL. 3, pp. 273–286.

Akhavan, O., Ghaderi, E., 2012. Escherichia coli bacteria reduce graphene oxide tobactericidal graphene in a self-limiting manner. Carbon 50, 1853–1860.

Almeida, E.A., Bainy, A.C.D., Loureiro, A.P.M., Martinez, G.R., Miyamoto, S., Onuki, J.,Barbosa, L.F., Machado Garcia, C.C., Prado, F.M., Ronsein, G.E., Sigolo, C.A.,Brochini, C.B., Martins, A.M.G., Medeiros, M.H.G., DiMascio, P., 2007. Oxidativestress in Perna perna and other bivalves as indicators of environmental stress inthe Brazilian marine environment: antioxidants, lipid peroxidation and DNAdamage. Comp. Biochem. Physiol. A 146, 588–600.

Blaise, C., Gagne, F., Ferard, J.F., Eullaffroy, P., 2008. Ecotoxicity of selected nano-materials to aquatic organisms. Environ. Toxicol. 23, 591–598.

Buffet, P.E., Fossi Tankoua, O., Pan, J., Berhanu, D., Herrenknecht, C., Poirier, L.,Amiard-Triquet, C., Amiard, J., Bérard, J., Risso, C., Guibbolini, M., Roméo, M.,Reip, P., Valsami-Jones, E., Mouneyrac, C., 2011. Behavioural and biochemicalresponses of two marine invertebrates Scrobicularia plana and Hediste diver-sicolor to copper oxide nanoparticles. Chemosphere 84, 166–174.

Carmagnol, F., Sinet, P.M., Jerome, H., 1983. Selenium-dependent and non-selenium-dependent glutathione peroxidases in human tissue extracts. Bio-chim. Biophys. Acta 759, 49–57.

Cornejo-Garrido, H., Kibanova, D., Nieto-Camacho, A., Guzmán, J., Ramírez-Apan, T.,Fernández-Lomelín, P., Garduño, M.L., Cervini-Silva, J., 2011. Oxidative stress,cytoxicity, and cell mortality induced by nano-sized lead in aqueous suspen-sions. Chemosphere 84, 1329–1335.

Dikin, D.A., Stankovich, S., Zimney, E.J., Piner, R.D., Dommett, G.H.B., Evmenenko, G.,Nguyen, S.T., Ruoff, R.S., 2007. Preparation and characterization of grapheneoxide paper. Nature 448, 457–460.

Edgington, A.J., Roberts, A.P., Taylor, L.M., Alloy, M.M., Reppert, J., Rao, A.M., Mao, J.,Klaine, S.J., 2010. The influence of natural organic matter on the toxicity ofmultiwalled carbon nanotubes. Environ. Toxicol. Chem. 29, 2511–2518.

C. Pretti et al. / Ecotoxicology and Environmental Safety 101 (2014) 138–145144

Fang, J., Lyon, D.Y., Wiesner, M.R., Dong, J., Alvarez, P.J.J., 2007. Effect of a fullerenewater suspension on bacterial phospholipids and membrane phase behavior.Environ. Sci. Technol. 41, 2636–2642.

Feng, L., Liu, Z., 2011. Graphene in biomedicine: opportunities and challenges.Nanomed. (London, UK) 6, 317–324.

García, A., Espinosa, R., Delgado, L., Casals, E., González, E., Puntes, V., Barata, C.,Font, X., Sánchez, A., 2011. Acute toxicity of cerium oxide, titanium oxide andiron oxide nanoparticles using standardized tests. Desalination 269, 136–141.

GHS, 2007. Globally Harmonized System for the Classification and Labelling ofChemicals. ⟨http://www.unece.org/trans/danger/publi/ghs/ghs_welcome_e.html⟩.

Guillard, R.R.L., Ryther, J.H., 1962. Studies of marine planktonic diatoms. I. Cyclotellanana Hustedt and Detonula confervacea Cleve. Can. J. Microbiol. 8, 229–239.

Handy, R.D., Owen, R., Valsami-Jones, E., 2008. The ecotoxicology of nanoparticlesand nanomaterials: current status, knowledge gaps, challenges, and futureneeds. Ecotoxicology 17, 315–325.

Heinlaan, M., Ivask, A., Blinova, I., Dubourguier, H.C., Kahru, A., 2008. Toxicity ofnanosized and bulk ZnO, CuO and TiO2 to bacteria Vibrio fischeri and crusta-ceans Daphnia magna and Thamnocephalus platyurus. Chemosphere 71,1308–1316.

International Maritime Organization (IMO), 2002. The Revised GESAMP HazardEvaluation Procedure for Chemical Substances Carried by Ships. Reports andStudies no. 64.

International Organization for Standardization (ISO), 2007. ISO 11348-3: waterquality. determination of the inhibitory effect of water samples on the lightemission of Vibrio fischeri (luminescent bacteria test). Part 3: method usingfreeze-dried bacteria.

International Organization for Standardization (ISO), 1995. ISO 10253: waterquality. Marine Algal Growth Inhibition Test with Skeletonema costatum andPhaeodactylum tricornutum.

Jastrzębska, A.M., Kurtycz, P., Olszyna, A.R., 2012. Recent advances in graphenefamily materials toxicity investigations. J. Nanopart. Res. 14, 1320.

Kang, S., Mauter, M.S., Elimelech, M., 2008. Microbial cytotoxicity of carbon-basednanomaterials: implications for river water and wastewater effluent. Environ.Sci. Technol. 42, 7528–7534.

Li, Q., Mahendra, S., Lyon, D.Y., Brunet, L., Liga, M.V., Li, D., Alvarez, P.J.J., 2008.Antimicrobial nanomaterials for water disinfection and microbial control:potential applications and implications. Water Res. 42, 4591–4602.

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

Liu, C.-W., Xiong, F., Jia, H.-Z., Wang, X.-L., Cheng, H., Sun, Y.-H., Zhang, X.-Z., Zhuo,R.-X., Feng, J., 2013. Graphene-based anticancer nanosystem and its biosafetyevaluation using a zebrafish model. Biomacromolecules 14 (2), 358–366.

Long, Z., Ji, J., Yang, K., Lin, D., Wu, F., 2012. Systematic and quantitative investiga-tion of the mechanism of carbon nanotubes toxicity toward Algae. Environ. Sci.Technol. 46, 8458–8466.

Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R.J., 1951. Protein measurementwith the Folin phenol reagent. J. Biol. Chem. 193, 265–275.

Lyon, D.Y., Alvarez, P.J.J., 2008. Fullerene water suspension (nC60) exerts antibac-terial effects via ROS-independent protein oxidation. Environ. Sci. Technol. 42,8127–8132.

Mihara, M., Uchiyama, M., 1978. Determination of malonaldehyde precursor intissues by thiobarbituric acid test. Anal. Biochem. 86, 271–278.

Mortimer, M., Kasemets, K., Heinlaan, M., Kurvet, I., Kahru, A, 2008. Highthroughput kinetic Vibrio fischeri bioluminescence inhibition assay for studyof toxic effects of nanoparticles. Toxicol. in Vitro 22, 1412–1417.

Novoselov, K.S., Geim, A.K., Morozov, S.V., Jjiang, D., Zhang, Y., Dubonos, S.V.,Grogorieva, I.V., Firsov, A.A., 2004. Electric field effect in atomically thin carbonfilms. Science 306, 666–669.

Nunes, B., Carvalho, F., Guilhermino, L., 2006. Effects of widely used pharmaceu-ticals and a detergent on oxidative stress biomarkers of the crustacean Artemiaparthenogenetica. Chemosphere 62, 581–594.

Paek, S.M., Yoo, E., Honma, I., 2009. Enhanced cyclic performance and lithiumstorage capacity of SnO2/graphene nanoporous electrodes with three-dimensionally delaminated flexible structure. Nano Lett. 9, 72–75.

Petersen, S., Soller, J.T., Wagner, S., Richter, A., Bullerdiek, J., Nolte, I., Barcikowski, S.,Escobar, H.M., 2009. Co-transfection of plasmid DNA and laser-generated goldnanoparticles does not disturb the bioactivity of GFP-HMGB1 fusion protein.J. Nanobiotechnol. 7, 6.

Pretti, C., Renzi, M., Focardi, E.S., Giovani, A., Monni, G., Melai, B., Rajamani, S.,Chiappe, C., 2011. Acute toxicity and biodegradability of N-alkyl-N-methylmorpholinium and N-alkyl-DABCO based ionic liquids. Ecotoxicol.Environmen. Saf. 74, 748–753.

Sanchez, V.C., Jachak, A., Hurt, R.H., Kane, A.B., 2012. Biological interactions ofgraphene-family nanomaterials: an interdisciplinary review. Chem. Res. Tox-icol. 25, 15–34.

Soldano, C., Mahmood, A., Dujardin, E., 2010. Production, properties and potentialof graphene. Carbon 48, 2127–2150.

Stankovich, S., Dikin, D.A., Dommett, G.H.B., Kohlhaas, K.M., Zimney, E.J., Stach, E.A.,Piner, R.D., Nguyen, S.B.T., Ruoff, R.S., 2006. Graphene-based composite materi-als. Nature 442, 282–286.

Strigul, N., Vaccari, L., Galdun, C., Wazne, M., Liu, X., Christodoulatos, C., Jasinkie-wicz, K., 2009. Acute toxicity of boron, titanium dioxide, and aluminumnanoparticles to Daphnia magna and Vibrio fischeri. Desalination 248, 771–782.

Tiede, K., Hassellöv, M., Breitbarth, E., Chaudhry, Q., Boxall, A.B.A., 2009. Considera-tions for environmental fate and ecotoxicity testing to support environmentalrisk assessments for engineered nanoparticles. J. Chromatogr. A 1216, 503–509.

Unfried, K., Albrecht, C., Klotz, L.O., Von Mikecz, A., Grether-Beck, S., Schins, R.P.F.,2007. Cellular responses to nanoparticles: Target structures and mechanisms.Nanotoxocology 1, 52–71.

U.S. EPA, 1993. ⟨http://water.epa.gov/scitech/methods/cwa/wet/upload/2007_07_10_methods_wet_disk1_ctmapg-l.pdf⟩.

Vanhaecke, P., Persoone, G., 1981. Report on an intercalibration exercise on a short-term standard toxicity test with Artemia nauplii (ARC-test). INSERM 106,370–376.

Vivekchand, S.R.C., Rout, C.S., Subrahmanyam, K.S., Govindaraj, A., Rao, C.N.R., 2008.Graphene-based electrochemical supercapacitors. J. Chem. Sci. 120, 9–13.

Wang, J., Zhang, X., Chen, Y., Sommerfeld, M., Hu, Q., 2008. Toxicity assessment ofmanufactured nanomaterials using the unicellular green alga Chlamydomonasreinhardtii. Chemosphere 73, 1121–1128.

Wang, S., Ang, P.K., Wang, Z., Tang, A.L., Thong, J.T., Loh, K.P., 2010. High mobility,printable, and solution-processed graphene electronics. Nano Lett. 10, 92–98.

Wei, L., Thakkar, M., Chen, Y., Ntim, S.A., Mitra, S., Zhang, X., 2010. Cytotoxicityeffects of water dispersible oxidized multiwalled carbon nanotubes on marinealga, Dunaliella tertiolecta. Aquat. Toxicol. 100, 194–201.

Westervelt, R.M., 2008. Graphene nanoelectronics. Science 320, 324–325.Zanni, E., De Bellis, G., Bracciale, M.P., Broggi, A., Santarelli, M.L., Sarto, M.S.,

Palleschi, C., Uccelletti, D., 2012. Graphite nanoplatelets and Caenorhabditiselegans: insights from an in vivo model. Nano Lett. 12, 2740–2744.

C. Pretti et al. / Ecotoxicology and Environmental Safety 101 (2014) 138–145 145