Nanotoxicology, 2013; Early Online, 1–10© 2013 Informa UK, Ltd.ISSN: 1743-5390 print / 1743-5404 onlineDOI: 10.3109/17435390.2013.807446

Enhanced toxicity of ‘bulk’ titanium dioxide compared to ‘fresh’ and‘aged’ nano-TiO2 in marine mussels (Mytilus galloprovincialis)

Alessia D’Agata1, Salvatore Fasulo1, Lorna J. Dallas2, Andrew S. Fisher3, Maria Maisano1,James W. Readman4, & Awadhesh N. Jha2

1Department of Biological and Environmental Sciences, University of Messina, Viale F. Stagno d’Alcontres 31, S. Agata – 98166,Messina, Italy, 2School of Biomedical & Biological Sciences, Plymouth University, Plymouth, PL4 8AA, UK, 3School of Geography,Earth and Environmental Sciences, Plymouth University, Plymouth, PL4 8AA, UK and 4Plymouth Marine Laboratory, ProspectPlace, The Hoe, Plymouth PL1 3DH, UK

AbstractMarine bivalves (Mytilus galloprovincialis) were exposed totitanium dioxide (10 mg L�1) either as engineered nanoparticles(nTiO2; fresh, or aged under simulated sunlight for 7 days) or thebulk equivalent. Inductively coupled plasma-optical emissionspectrometry analyses of mussel tissues showed higher Tiaccumulation (>10-fold) in the digestive gland compared to gills.Nano-sized TiO2 showed greater accumulation than bulk,irrespective of ageing, particularly in digestive gland (>sixfoldhigher). Despite this, transcriptional expression ofmetallothionein genes, histology and histochemical analysissuggested that the bulk material was more toxic. Haemocytesshowed significantly enhanced DNA damage, determined by themodified comet assay, for all treatments compared to thecontrol, but no significant differences between the treatments.Our integrated study suggests that for this ecologically relevantorganism photocatalytic ageing of nTiO2 does not significantlyalter toxicity, and that bulk TiO2 may be less ecotoxicologicallyinert than previously assumed.

Keywords: engineered nanoparticles, histopathology, geneexpression, oxidative DNA damage, comet assay

Introduction

Despite their economic importance, the recent large-scale production and utilisation of engineered nanoparticles(ENPs) has raised concerns over their environmental impact.Consequently, to meet these concerns, there has been callfor proper environmental risk assessment of these anthro-pogenic particles (European Commission 2005; Moore 2006;Nowack & Bucheli 2007; Handy et al. 2008; Klaine et al.2008). ENPs have different properties than their bulk coun-terparts as they present a very large surface area-to-volumeratio. When particle size shrinks, there is potential for

enhanced toxicity to biota, even if the material is relativelyinert in bulk form (Farrè et al. 2009; Al-Subiai et al. 2012).This enhanced toxicity may be due to a range of mechan-isms, including the greater surface reactivity of ENPs andtheir ability to penetrate into and accumulate within cellsand organisms (Xiong et al. 2011). Direct and indirect releaseof ENPs into aquatic environments via bathing, sewageeffluent (Handy & Shaw 2007) and engineering applications(Chen et al. 2004; Nagaveni et al. 2004) has increasedpotential exposures to both natural biota and humans (Now-ack & Bucheli 2007). Furthermore, the effects of ENPs onaquatic organisms may also impact human health via thefood chain. Despite this, relatively few studies have exam-ined the toxic or genotoxic effects of ENPs on aquaticorganisms, particularly on invertebrates, which play impor-tant role in ecosystem functioning (Baun et al. 2008;Handy et al. 2008; 2012; Jha, 2008; Al-Subiai et al. 2012).

Nano-sized titanium dioxide (nTiO2) is a widely usedENP, found in a variety of consumer products (e.g. paper,paint and toothpaste) and used in the decontamination ofair, soil and water (Esterkin et al. 2005; Weir et al. 2012). TiO2

nanoparticles have been shown to induce detrimental bio-logical responses in different in vitro and in vivo systems(Reeves et al. 2008; Vevers & Jha 2008). Historically, one ofthe important applications of TiO2 particles has been insunscreens. When these particles, with apparently low cuta-neous penetration, are washed off the skin they are releasedinto the aquatic environment. Environmental contaminationby nTiO2 therefore seems inevitable and raises concerns forboth humans and aquatic organisms (Kaegi et al. 2008;Kiser et al. 2009; Weir et al. 2012; Handy et al. 2012). Giventhe dynamic nature of the marine environment, there ispotential for ageing of nTiO2 and generation of new products(Fouqueray et al. 2012), the fate and impact of which are notknown (Labille et al. 2010). Consequently, the potentialimpact of nTiO2 on aquatic ecosystems has attracted special

Correspondence: Awadhesh N. Jha, School of Biomedical & Biological Sciences, Plymouth University, Plymouth, PL4 8AA, UK. Tel: +44 1752 584633.Fax: +44 1752 584605. E-mail: a.jha@plymouth.ac.uk

(Received 7 February 2013; accepted 10 May 2013)

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

attention (Oberdörster et al. 2005; Moore 2006; Handy andShaw 2007; Handy et al. 2008). Despite this, little is knownabout its toxic potential to marine invertebrates (Moore2006; Klaine et al. 2008; Ward & Kach 2009; Scown et al.2010).

It is now well accepted that the biological responses ofENPs are dependent on the physico-chemical environmentin which they contact cells (Vevers & Jha 2008; Handy et al.2012). In addition, aggregation of particles in the aquaticenvironment can influence their interaction and/or uptakeand therefore the biological responses. Since TiO2 nanopar-ticles are photocatalytic (Tsuang et al. 2008, Dodd & Jha2009, 2011), ultraviolet (UV) radiation has been shown tosubstantially increase nTiO2 toxicity by production of reac-tive oxygen species (ROS; Nakagawa et al. 1997; Uchino et al.2002; Reeves et al. 2008) which can induce oxidative stressand cellular damage (Labille et al. 2010; Dodd & Jha 2009,2011).

Against the backdrop of the above information, this studyadopted an integrated approach to evaluate the biologicaleffects of fresh nTiO2, aged nTiO2 and the bulk equivalent ina representative bivalve, Mytilus galloprovincialis. We alsocompared the biological responses with a reference toxicmetal, copper (Cu). Several assays were used simultaneouslyin different target cell types to obtain a holistic perspective.These techniques have been thoroughly optimised andvalidated in our laboratories using this organism (Jha2005; Al-Subiai et al. 2011; Fasulo et al. 2008; 2012;Ciacci et al. 2012, Millward et al. 2012, Dallas et al. 2013).In particular, we determined (a) histomorphological altera-tions in gill and digestive gland; (b) Alcian blue and Periodicacid-Schiff (AB/PAS) staining of acid mucopolysaccharidesin gill; (c) transcriptional expression and localisation in situof two different isoforms of metallothionein (mt) genes(using RT-PCR products and fluorescent in situ hybridisa-tion); and (d) oxidative DNA damage in the haemocytes,using modified comet assay (Reeves et al. 2008; Vevers & Jha2008, Mustafa et al. 2012, Dallas et al. 2013). In addition, Tiaccumulation in tissues (i.e. gill and digestive gland) andconcentrations of different elements in different forms ofTiO2 used were determined using inductively coupledplasma-optical emission spectrometry (ICP-OES) andinductively coupled plasma mass spectrometry (ICP-MS),respectively.

Methods

Preparation and characterisation of TiO2

The nano-sized titanium dioxide (DeGussa AG, UK) usedhad a crystalline composition of 75% anatase and 25% rutile,average diameter 21 nm and purity >99% (according tomanufacturer’s information). The bulk TiO2 used here (tita-nium[IV]oxide; Acros Organics, UK) was the same as thatused by Al-Jubory & Handy (2012), who indicated a crystalstructure of 74.7% anatase and 25.3% rutile. The nTiO2

samples have been characterised under our laboratory con-ditions using a range of techniques (Federici et al. 2007;Ramsden et al. 2009; Windeatt & Handy 2012). Nano-TiO2

was aged by exposing a 200 mg L�1 solution of the powder

(in seawater) to simulated sunlight (16 h light: 8 h dark) for7 days. Natural sunlight was simulated using Edison fluo-rescent lamps, each 230 V, 70 W, 35–535 colour. Lightintensity was 4.07 klx, measured using an RS-01 Light Meter(RS components, UK).

Stock solutions were dispersed using a bath sonicator(35 kHz, Fisherbrand FB1010, Germany) as described else-where (Federici et al. 2007; Windeatt & Handy 2012). A well-mixed subsample (20 mL) of each stock was also sonicatedfor a further 1 h immediately prior to each application.Subsamples were then further characterised by observingdirectly on a plastic film with a copper grid using transmis-sion electron microscopy (TEM; JEM-1200 EX II; JEOL Ltd,USA). The diameter of 100 ENPs wasmeasured for each formusing an image analysis program (Image J, v1.44; NationalInstitute of Health, USA).

Animal collection and maintenanceMussels (M. galloprovincialis; 45–50 mm) were collected atlow tide from Trebarwith Strand (Cornwall, UK), a relativelyclean site, used for reference in previous studies(Al-Subiai et al. 2009, 2011, 2012). After collection musselswere immediately transported to the laboratory and placedin an aerated tank (1 mussel L�1) with filtered seawater(<10 mm). Mussels were maintained at 15�C, fed daily withmicro algae (Isochrysis galbana, Interpet, UK) and waterchanged daily. At least two weeks were allowed for musselsto acclimatise before exposure.

Exposure conditionsMussels were exposed to a nominal concentration of10 mg L�1 of either bulk, fresh or aged nTiO2 in 2-L glassbeakers (4 animals beaker�1). Mussels were also exposed to40 mg L�1 CuSO4.5H2O (99% purity) as a positive control(Al-Subiai, et al. 2011). Each exposure was carried out intriplicate. The concentration of Cu was based on earlierin vivo studies using bivalve molluscs (Bolognesi et al.1999; Al Subiai et al. 2009; 2011) while the concentrationof TiO2 was based on several studies determining biologicalresponses in various organisms (Canesi et al. 2010;Lapied et al. 2011; Xiong et al. 2011). The 4-day exposureof mussels to Cu and TiO2 was based on earlier studies in ourlaboratory using a range of toxicants (Jha 2005; Canty et al.2009; Al-Subiai et al. 2011). Each treatment was reneweddaily and the water quality parameters were maintainedwithin acceptable ranges (temperature: 15 ± 1�C; dissolvedoxygen: 96.1 ± 0.3%; total ammonia: 0.04 ± 0.02 mg L�1; pH7.8 ± 0.02; salinity: 31.5 ± 0.15; Hach HQ40D Multi-meter [Hach-Lange, Germany]). Animals were not fed anddid not spawn during the experiment.

Collection of haemolymph samples for thedetermination of DNA strand breaksHaemolymph was withdrawn via the posterior adductormuscle using a 1-mL syringe and 21-gauge needle(Al-Subiai et al. 2009). Samples were diluted with an equalvolume of physiological saline (0.02 M HEPES, 0.4 M NaCl,0.1 M MgSO4, 0.01 M KCl and 0.01 M CaCl2, pH 7.4) andcentrifuged at 350� g for 2 min at 4�C (Al-Subiai et al. 2009).

A. D’Agata et al.

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

Samples were then placed on ice until processing for thecomet assay.

Determination of oxidative DNA damage using themodified comet assayInduction of DNA strand breaks in the haemocytes wasdetermined as described elsewhere (Jha et al. 2005; AlSubiai et al. 2011, 2012). The comet assay can be modifiedby addition of bacterial enzymes which specifically targetoxidised purine or pyrimidine bases (Collins et al. 1995).Measurement of oxidative DNA damage in mussel haemo-cytes using this technique has been validated in our labo-ratory (Dallas et al. 2013). In common with previous studiesin our laboratory (Reeves et al. 2008; Vevers & Jha 2008;Mustafa et al. 2012, Dallas et al. 2013), the enzymes usedwere formamidopyrimidine DNA glycosylase (Fpg) andendonuclease III (Endo III). Briefly, the final haemocyte-agarose suspension was added to a slide pre-coated with 1%normal melting point agarose as two replicate 75-ml micro-gels. Three slides were prepared per sample, one enzymebuffer control, one with Fpg and one with Endo III. The slideswere covered with chilled lysis solution (pH 10.0) and kept at4�C for 1 h. After lysis, slides were washed three times(5 min) in enzyme buffer (40 mM HEPES, 0.1 M KCl,0.5 mM EDTA, 0.20 mg mL�1 BSA, pH 8.0; Sigma-Aldrich,UK), drained and 50 ml of Fpg, Endo III or enzyme bufferadded to each microgel. Slides were transferred to a humid-ity chamber and incubated at 37�C for 45min. To unwind theDNA, slides were placed in a horizontal electrophoresis boxfilled with freshly prepared alkaline buffer (0.3 M NaOH,1 mM EDTA, pH 13) for 30 min at 4�C. After electrophoresis(300 mA, 25 V, 30 min) the slides were washed three timeswith neutralisation buffer and rinsed with refrigerated dis-tilled water. Each microgel was stained with 50 ml of a1:1 solution of Sybr Safe (10�; Invitrogen) and anti-fadingagent (1:1) and examined with a fluorescence microscope(excitation: 488 nm excitation, emission: 520 nm). Thepercentage of DNA in the comet tail (% tail DNA) wasquantified for 100 cells per slide using Komet 5.0 software(Kinetic Imaging, UK). Comet analysis was carried out onsamples of haemocytes from individual mussels. These datawere then combined by taking the mean for two musselsfrom three replicate beakers (n = 6).

There is the potential for cytotoxicity to cause a positiveresponse in the comet assay (i.e. DNA migration); however,there are standard procedures in place in our laboratory tominimise this effect. Cell viability was determined prior tothe comet assay (by eosin Y dye exclusion assay) and onlysamples with >90% viability were used (Henderson et al.1998). In addition, during scoring, so-called ‘hedgehog cells’were excluded (Brendler-Schwaab et al. 2005).

Histology and histochemical analysesFollowing haemolymph sampling, mussels were dissectedto obtain gill and digestive gland tissues for histologicaland histochemical assessments, using standard proceduresroutinely used in our laboratory (Fasulo et al. 2008; 2010a,Al Subiai et al. 2011, 2012, Cappello et al. 2013). Briefly,dissected tissues were fixed in 10% buffered formal

saline, dehydrated in ethanol and embedded in paraffin(Bio-Optica, Italy). Histological sections (4 mm) were cutwith a rotary automatic microtome (Leica Microsystems,Germany), mounted on glass slides and stained with hae-matoxylin/eosin (Bio-Optica, Italy) to visualise typical mor-phological features. A combined method, using AB/PASstaining (pH 2.5), was used to detect neutral and acidmucocytes in gill epithelium. Micrographs were obtainedusing an Axio Image Z1 microscope (Carl Zeiss AG, Ger-many) with Axio Vision software (v 4.5; Carl Zeiss AG,Germany) for acid mucocyte quantification. These datawere then combined by taking the mean for two musselsfrom the three replicate beakers (n = 6).

Determination of transcriptional expression of mt genesFor the transcriptional expression of mt genes, gill anddigestive gland tissues were collected and stored at �80�Cuntil use (Fasulo et al. 2008, 2010a, 2010b). Briefly, total RNAwas extracted using TRIzol LS reagent (Invitrogen,USA; Chomczynski & Sacchi 1987). The RNA content wasquantified using a UV spectrophotometer (NanodropTM

2000, Thermo Scientific, UK). cDNA was synthesised using4 mg of total RNA and oligo (dt)20 primers (150 pmol/mL;Invitrogen, UK), with M-MLV reverse transcriptase (Invitro-gen, UK) according to the manufacturer’s instructions. Onemicrolitre of the resulting cDNA was amplified in a totalreaction volume of 25 mL (2.5 mL of 10� buffer, 0.65 UEuroTaq polymerase [Euroclone s.p.a., Italy], 1.5 mL of50 mM MgCl2, 2 mL of 10 mM mixed primers and 2 mL of10 mM dNTPs, in Milli-Q water). The thermocycling pro-gram used to amplify fragments was 95�C for 2 min, then35 cycles of 95�C for 30 s, 55�C for 30 s, 72�C for 1 min andfinal extension at 72�C for 5 min (Mastercycler EP-Gradient,Eppendorf, Italy).

The sequences of the primers were based on those ofmt10-III and mt20-II of M. galloprovincialis (GeneBankaccession ns. AY566248 and AY566247) and were expectedto amplify sequences 193 and 161 bp long, respectively.

mt10 sense primer: 5¢-TGCCTGCACCTTGTAACTGT-3¢mt10 antisense primer: 5¢-CTACACGTTGAAGGCCCT

GT-3¢mt20 sense primer: 5¢-AGGATGCAGCGAGAAATGTT-3¢mt20 antisense primer: 5¢-AGGAGCACCCAGATTCAC

AT-3¢RT-PCR products were characterised by electrophoresis

on SYBR-safe-stained agarose gels. Bands of ~193 and~161 bp for mt10 and mt20 were obtained. The resultswere normalised to the expression of actin, which wasexpressed at basal levels in both control and treated animals(data not shown). The actin gene was amplified usingsequence primers based on the cDNA sequence of M.galloprovincialis (accession number AF157491) to amplify200 bp.

Localisation of mt mRNA transcript using fluorescencein situ hybridisation techniqueThe mRNA expression and morphological localisation ofmt10 and mt20 genes in gill and digestive gland wereinvestigated by means of fluorescence in situ hybridisation

Biological effects of different forms of titanium dioxide in mussels

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

(FISH) on histological sections (Fasulo et al. 2008;Minniti et al. 2009). Briefly, antisense 40-bp DNA probes(based on the mt10-III and mt20-IV mt cDNA sequences ofM. galloprovincialis) were synthesised, 3¢ marked with fluo-rescein (mt10) or Rhodamine Red (mt20), and purified byMWG-Biotech AG. Tissue sections (7 mm) were deparaffi-nised, rehydrated and washed twice in Diethylpyrocarbo-nate (DEPC)-treated water and in DEPC-PBS. Sectionswere then post-fixed in 4% paraformaldehyde, 0.1 M PBSand permeabilised in Triton X-100 (0.3% in PBS). Afterwashing in PBS, pre-hybridisation buffer was added to eachtissue section and the slides incubated in a humidifiedchamber for 2 h at 37�C. Slides were then washed in saline-sodium citrate (SSC) buffer. Hybridisation buffer with fluo-rescent probe (final concentration 2 mg mL�1) was added toeach tissue section before incubation in a dark humidifiedchamber overnight at 37�C. After hybridisation, the slideswere washed with a low-stringency washing buffer. Sec-tions were rinsed in distilled water to remove salts andmounted using a medium containing an anti-fading agent.An RNA-negative control was prepared by digesting thetissue with RNases prior to hybridisation with the oligo-nucleotide probes (Fasulo et al. 2008; Minniti et al. 2009).Staining was evaluated using an Axio Imager Z1 (Zeiss)epifluorescence microscope and analysed with Axio Visionsoftware. Statistical significance was determined by one-way analysis of variance (ANOVA).

Determination of titanium concentrations in water andtissue samplesWater samples (1 mL) were collected from each exposurebeaker daily before and after the water change. Samples werediluted 1:10 with 2% HNO3 and were stored until analysis(Federici et al. 2007; Ramsden et al. 2009). Tissue sampleswere analysed using the protocols modified from Al-Juboryand Handy (2012). Briefly, tissues were oven dried at 60�C toconstant weight, digested in 1 mL of concentrated nitric acid(trace analysis grade) for 3 h at 70�C. After cooling, the

samples were diluted with Milli-Q water. Samples werethoroughly vortex mixed immediately prior to analysisusing ICP-OES (Varian 725 ES). To aid in comparisonsbetween nTiO2 and its bulk equivalent, 200 mg L�1 solutionsof both forms were sonicated for >1 h, and then analyseddirectly for a range of additional elements using ICP-MS(Thermo Scientific, X Series 2). The results of these analysesare presented in Table S1.

Statistical analysesStatistical analyses of data were conducted with GraphPadInstat software (GraphPad Software Inc., USA). All results arepresented as mean ± SD. Significant differences were deter-mined using one-way ANOVA, followed by multiple-rangetests to differentiate between groups of data. Significance forall tests was set at p < 0.05.

Results

Characterisation of TiO2 ENPsTEMmeasurements indicated average diameters (n = 100) of151.40 ± 6.7 nm for bulk TiO2 and 24 ± 4.6 nm for fresh nTiO2

(Figure S1). In both cases, these measurements were similarto the manufacturer’s specifications and are in line withprevious results from our laboratory (e.g. Windeatt andHandy 2012). In contrast, the aged nTiO2 formed aggregatesof particles with average diameters ranging from 27.60 ±6.9 to 108.40 ± 5.2 nm (Figure S1).

Determination of oxidative DNA damage using themodified comet assayControl animals showed low levels of DNA damage (9.38 ±3.10 % tail DNA; Figure 1) for buffer controls and enzymetreatments. All treatments showed significantly higher DNAdamage than controls (p < 0.0001). Interestingly all TiO2

treatments resulted in approximately 40% tail DNA and therewere no significant differences between the treatments. Theuse of Fpg and Endo III showed no significant differencescompared to the buffer controls.

Histological and histochemical observationsIn all control organisms, the gills had no pathological tissuealterations (Figure S2). AB/PAS staining indicated that twoout of six control specimens had acidophil cells (Figure 2A),but gill filaments in control mussels had proportionallymore neutrophil cells than all forms of TiO2 (Figure 2F).Histological and histochemical observations in the gillsshowed substantial influx of haemocytes along the fila-ments and a large number of acidophilic mucous cellsalong the apex of the filament in all Cu-exposed mussels(Figure S2; 2b; Table I). Exposure to both forms of nTiO2

resulted in moderate influx of haemocytes along the gillfilaments, but no particular alterations of the tissue (FigureS2; Table I). In all mussels exposed to bulk TiO2, a loss ofstructural definition of the ciliated epithelium and hypo-plasia was observed (Figure S2; Table I) and acidophil cellswere distributed throughout the branchial filament epithe-lium (Figure 2C). Mussels treated with nTiO2 and its aged

0Ctrl Cu Bulk nTiO2 nTiO2 aged

10

20

30

% t

ail D

NA

40

50

60

70

Control Endo Fpg

Figure 1. Induction of DNA strand breaks, represented as % tailDNA, in mussel haemocytes following 4 days in vivo exposure tocopper, nTiO2 aged nTiO2 and bulk TiO2. Values are mean ± SD.All treatments showed significantly higher DNA damage than thecontrols (p < 0.0001).

A. D’Agata et al.

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

form showed intermediate numbers of acidophilic cellsaround the apex of filaments (Figure 2D and 2E; TableI). Numbers of acidophilic cells varied between the forms ofnTiO2 in the order bulk > Cu > nTiO2 > aged nTiO2

(Figure 2F, p < 0.0001). There was a statistically significantdifference from the control for all treatments except foraged nTiO2.

In common with gills, none of the digestive glands fromcontrol mussels showed tissue alterations (Figure S2f). Allorganisms treated with Cu showed damaged tissues withvacuolation of the digestive tubules (Figure S2g). The freshand aged nTiO2 caused vacuolation of digestive tubules(Figure S2i and l), in the same percentage of specimensseen previously for gill alterations (i.e. 83.3% for nTiO2; 66.6%for its aged form; Table I). In all samples treated with the bulkform there was increased vacuolisation of the digestivetubules and substantial haemocyte infiltration (Figure S2h;Table I).

Expression of mt-specific complementary DNAIn gills the expression of the mt10 gene was significantlyhigher for bulk and nTiO2 than for controls, in contrast toaged nTiO2 which had lower band intensity (Figure 3A;p < 0.05). In the digestive gland, expression of the mt10gene was similar for all treatments including the control. Themt20 gene was over-expressed in mussels treated with Cuand with bulk TiO2 in both gills and digestive gland(Figure 3A and 3B, p < 0.0001). In digestive glands, thisinduced mt20 transcriptional expression was significantlyhigher (for Cu and bulk) than for both nTiO2 forms(p < 0.001).

FISH analysesThe band intensity data were confirmed by the localisation ofthe respective mRNAs. In gill samples of control mussels,numerous mt10-positive cells were present along the gillfilaments (Figure S3a), but with only a few mt20-positive

A

Ctrl Cu

Bulk

nTiO2 aged

nTiO2

B

C D

E F

0Ctrl Cu

∗∗

∗

Bulk nTiO2 nTiO2 aged

NeutralAcids

100

200

Ave

rag

e p

osi

tive

cel

ls

300

400

500

600

Mucopolysaccharides M. edulis

Figure 2. AB/PAS staining of 4-mm sections ofM. galloprovincialis gills showing neutral and acid mucous cells following 4 days in vivo exposure tocopper, nTiO2 aged nTiO2 and bulk TiO2. Scale bar: 20 mm. Values are mean ± SD (n = 6). Significance at p < 0.0001 is indicated by *.

Biological effects of different forms of titanium dioxide in mussels

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

cells (Figure S3f). The mussels exposed to Cu (Figure S3band g) and bulk nTiO2 (Figure S3c and h) showed numerouspositive cells for both probes, along the branchial filamentand in particular in the apices of the gill. The gills frommussels treated with nTiO2 (Figure S3d and i) and its agedform (Figure S3e and l) showed numerous haemocytespositive to both probes, but less compared to Cu- andbulk TiO2-exposed animals. In the digestive gland, all treatedand control mussels showed numerous mt10-positive cells(Figure S3). Formt20, control, nTiO2 and aged nTiO2 showedfewer positive cells than those treated with Cu and bulk TiO2

(Figure S4). Figure 4 summarises this informationgraphically.

Determination of titanium concentrations in water andtissue samplesWater titanium concentrations did not show any cleartrends. Before renewal, mean TiO2 concentrations were82.32 ± 8.73 mg L�1 for fresh nTiO2, 73.92 ± 3.96 for agedand 6.72 ± 0.67 mg L�1 for bulk. After changing water, themean concentrations were approximately 77.28 ± 6.05 mg L�1

for fresh, 99.12 ± 11.76 mg L�1 for aged and 26.88 ± 1.84 mg L�1

for bulk. This decreased Ti concentration (from nominalvalues) was sample-specific (i.e. not due to instrumental

drift) as all calibration check standards measured duringthe ICP-OES run were within 10% of their initial value.

Comparison of nTiO2 concentrations in tissue samplesrevealed that the metal accumulated approximately 10 timeslower in the gills compared with digestive gland (Figure 5).The form of TiO2 also had a major impact on its tissue-specific accumulation. Interestingly, nano-sized TiO2,whether fresh or aged, was present at significantly greaterconcentrations then the bulk form in the digestive gland(p < 0.05; Figure 5B).

Discussion

Whilst there are several studies pertaining to size, aggrega-tion and properties of nTiO2 (e.g. Lin et al. 2006; Labille et al.2010; Jassby et al. 2012), to our knowledge this study is thefirst to compare the biological effects of different forms ofTiO2 (including fresh or aged) on a representative marineorganism. With respect to the Ti concentrations measured inthe water samples, these were very low probably due to thefact that ENPs tend to form aggregates (Nowack & Bucheli2007) and thus sink to the bottom quickly despite vigorousstirring before sampling. Previous studies have observedsuch precipitation of nTiO2 onto the bottom of tanks during

0Ctrl

0.5

1.5

Ban

d in

ten

sity

/act

in

2

3

3.5

4

Cu Bulk

Gills

nTiO2 nTiO2 aged

1

2.5

MT10 MT20

A

0Ctrl

0.5

1.5

Ban

d in

ten

sity

/act

in

2

3

3.5

4

Cu Bulk

Digestive glands

nTiO2 nTiO2 aged

1

2.5

MT10 MT20

B

∗∗

∗

∗

∗

∗

∗

Figure 3. Relative band intensity of mt10 and mt20 (compared to actin) RT-PCR products in control and treated mussels. Values are mean ± SD(n = 6). Significant differences are indicated by * (p < 0.001).

Table I. Summary of histomorphological and histochemical alterations in mussels exposed to Cu or TiO2.

Tissue TreatmentHistologicaltest/Reaction

Animals showingabnormality (% a) Main observed effects

Gills Control H/E 0 All normal

Cu 100 Influx of haemocytes along the filament

Bulk 100 Hypoplasia, loss of cilia

nTiO2 83.3 Influx of haemocytes along the filament

nTiO2 aged 66.6 Moderate influx of haemocytes along the filament

Digestive gland Control H/E 0 All normal

Cu 100 Vacuolation of the digestive tubules

Bulk 100 Increased vacuolation of the digestive tubuleswith substantial influx of haemocytes

nTiO2 83.3 Vacuolation of the digestive tubules

nTiO2 aged 66.6 Vacuolation of the digestive tubules

Gills Control AB/PAS 33.2 Few acidophilic mucous cells

Cu 100 Acidophilic mucous cells in apex of the filament

Bulk 100 Acidophilic mucous cells in the whole filament

nTiO2 83.3 Acidophilic mucous cells in apex of the filament

nTiO2 aged 83.3 Acidophilic mucous cells in apex of the filamentaPercentages were calculated from an n of 6.

A. D’Agata et al.

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

mussel exposures (e.g. Canesi et al. 2010). Also, the presenceof organic matter in the natural seawater used in our studycould also cause complexation and influence the stabilityand/or bioavailability of these materials (as suggested byother workers, e.g. Labille et al. 2010).

The ICP-OES measurements confirmed that musselsaccumulated TiO2 in a tissue-specific manner, with a signif-icant difference between the behaviour of bulk and nanoforms (aged or fresh). In common with other studies, ourresults suggest differential accumulation of metallic contami-nants in mussel tissues, in particular the digestive glandaccumulating more compared with gills (e.g. Millward et al.2012). It is possible to envisage a scenario where a highproportion of larger (bulk) particles are rejected and excretedas pseudofaeces at the gills, but that smaller (i.e. nano)particles are transported to the digestive gland. Our resultssupport this idea and an earlier study on NP agglomeratestaken up by the digestive gland (Canesi et al. 2012). However,as the TEM measurements of aged nTiO2 revealed largeraggregates, it is interesting that there appears to be no‘trapping’ effect in the gills for this form. It is, however,possible that – as a result of preparation – particle sizesmeasured using electron microscopy are different from thereal aggregation pattern in seawater. Furthermore, there islittle information on the effects of ageing on the bioavail-ability and uptake of ENPs (e.g. Labille et al. 2010) withwhich to compare our data. Accumulation of aged nTiO2 ingills and digestive glands of mussels in our study is incontrast to observations made in earthworms (Lumbricus

terrestris) exposed to an aqueous solution of aged TiO2

nanocomposite (Lapied et al. 2011). It appeared thatnTiO2 was unable to cross the intestinal epithelium/chloragogenous matrix barrier to enter the coelomic liquid,or the cuticle barrier to reach the muscular layers. Sincedifferent nanoparticles (T-lite) and an alternative ageingtechnique (48 h under white light) were used, these mayhave resulted in altered physico-chemical properties of thenanoparticles. Apart from the surface chemistry of the nTiO2,routes of exposure (mussels being filter feeders with an openvascular system) and complexation with organic matter(especially for earthworms) could also account for theseapparently contrasting results (van Herwijnen et al. 2007).

Histopathologically, the tissues treated with bulk TiO2

showed enhanced damage compared to other treatments.In both the gills and digestive gland more vacuolation andinflux of haemocytes was observed in the bulk-treated groupdespite the concentration of titanium being much lowerfollowing exposure to bulk compared to nTiO2 (aged orfresh). This indicates that, for these forms of damage, bulkTiO2 is considerably more toxic than its nano equivalents.Further support for this hypothesis comes from the histo-chemical analysis, where the gills of mussels exposed to bulkTiO2 had the highest levels of acidophilic mucous cells,although this was not significantly higher than for nTiO2

forms. Similar histochemical responses have also beenobserved for several marine organisms following exposureto different environmental stresses (Bettercourt et al. 2008,Fasulo et al. 2008, Brunelli et al. 2011).

0Ctrl

100

300

Ave

rag

e p

osi

tive

cel

l

400

600

700

Cu Bulk

FISH in gills

nTiO2 nTiO2 aged

200

500

MT10 MT20

A

∗

∗

0Ctrl

100

300

Ave

rag

e p

osi

tive

cel

l

400

600

700

Cu Bulk

FISH in digestive glands

nTiO2 nTiO2 aged

200

500

MT10 MT20

B

∗

∗

Figure 4. Quantification of FISH-positive cells in mussel gill (A) and digestive gland (B) following 4 days in vivo exposure to copper, nTiO2 agednTiO2 and bulk TiO2. Values are mean ± SD (n = 6). Significant differences are indicated by * (p < 0.001).

0Ctrl Cu Bulk

Ti in gills

nTiO2 nTiO2 aged

0.10.20.30.40.5

mg

/g

0.60.70.80.9

1

0Ctrl Cu Bulk

Ti in digestive glands

nTiO2

∗∗

nTiO2 aged

0.10.20.30.40.5

mg

/g

0.60.70.80.9

1A B

Figure 5. Accumulation of titanium in gills and digestive glands. Values are mean ± SD (n = 6). Significant differences from the controls areindicated by * (p < 0.05).

Biological effects of different forms of titanium dioxide in mussels

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

Results for transcriptional expression of mt10 and mt20genes and for localisation of the respective mRNAs suggestthat in digestive gland, mt10 is the basally expressed form,while mt20 is an inducible mt gene. This observation is inline with other studies in bivalve molluscs (Lemoine et al.2000, Dondero et al. 2005, Fasulo et al. 2008, Dallas et al.2013). Significant overexpression of the mt20 gene in thedigestive gland of mussels exposed to bulk TiO2, comparedto those of nTiO2 (fresh and aged), provides further evidencethat the bulk analogue induces stress in mussels. Our resultsindicate a tissue- or concentration-specific difference in mtgene induction as – in contrast to the digestive gland – bothmt genes were induced in the gills despite much lowertitanium accumulation. Although nTiO2 showed highermt10 induction in gills, this result was not supported byFISH or histochemical data. However, analysis of geneexpression at the transcriptional level does not necessarilyreflect the true stress scenario. Concurrent measurement ofmt proteins could potentially elucidate the mechanismsinvolved.

The relative histological and histochemical responsesfollowing exposure to different forms of TiO2 (i.e. bulk,aged and fresh nano) are in contrast to acute toxicity(96-h LC50) observations in zebra fish, where nano formsshowed higher toxicity compared to bulk form, which wasessentially non-toxic. This differential response was attrib-uted to enhanced production of hydroxyl radicals by nTiO2

(Xiong et al. 2011). Similar acute toxicities for nano com-pared to bulk form have been reported for soil nematode(Caenorhabditis elegans) and water flea (Daphnia magna)(Zhu et al. 2008; Wang et al. 2009). In our study, we did notevaluate the acute toxicity in terms of lethality/survival of themussels and used only sublethal concentrations, based onprevious studies. Phylogenetic differences may explain dif-ferential biological responses in marine mussels comparedto zebra fish, nematodes or water fleas.

As a naturally occurring mineral, TiO2 is present in twomain forms, rutile and anatase. The relative content of theseforms in the bulk and nano materials used in these experi-ments was very similar. In the environment both forms arebound to impurities and the raw material is chemicallyprocessed to remove these, leaving the commercial pro-ducts. The treatments to produce the bulk and nano materi-als differ and probably the associated impurities are removedto a different extent. This could potentially explain theenhanced toxicity of the bulk material. Our initial ICP-MS results (Table S1) have identified some differences inthe elemental composition of each form of TiO2 (e.g. withregard to Al, K and As) but more research is needed toinvestigate this issue.

Analysis of induction of oxidative DNA damage in thehaemocytes of mussels, as determined by modified cometassay, did not highlight significant differences among thethree forms of TiO2 used here. This is in contrast to in vitrostudies carried out in our laboratory conditions using fishcells which have shown enhanced oxidative damage to DNAfollowing exposure to nTiO2 (Reeves et al. 2008; Vevers & Jha2008). Using electron spin resonance techniques, it has beenfurther suggested that nTiO2-induced oxidative damage

results from hydroxyl radicals (.OH), carbonyl radical anions(CO2�) and superoxide radical anions (O2�; Dodd & Jha2009, 2011). It has been suggested that aggregation of metal-lic nanoparticles (such as that caused by electrolytes [e.g.NaCl in seawater; Jassby et al. 2012]) can decrease theirphotocatalytic properties (Lin et al. 2006; Tseng et al. 2006).Roles of aggregate size in decreased production of ROS have,however, not been demonstrated as yet (Jassby et al. 2012).Apart from generation of ROS, it is known that metalliccontaminants exert their genotoxic effects by directlyinteracting with cellular macromolecules (e.g. formation ofmetal–DNA adducts (Singh et al. 1998) or inhibition of DNArepair processes (Hartwig et al. 2002)). Therefore, inductionof DNA damage in haemocytes could be independent ofROS, as treatment with bacterial enzymes did not showsignificantly enhanced levels of DNA damage. It is alsoimportant to remember that DNA damage in the presentstudy was determined only in the haemocytes of themussels.These cells play a varied role in the organism and could havethe capacity to repair induced-DNA damage (Jha et al. 2005).Furthermore, in common with mammalian blood cells,haemocytes have a well-known capacity to undergo apopto-tic processes (Jha et al. 2005). It has been demonstrated thatnTiO2 induces tissue-specific apoptosis in earthworms fol-lowing exposure to concentrations which would normallydecrease survival, growth and fecundity (Lapied et al. 2011).Assessment of other markers in different tissues could beused to elucidate the role of oxidative stress in the observedbiological responses in mussels exposed to nano- and bulkTiO2.

Conclusions

Overall, our results show that the exposures of mussels tothree different forms of TiO2 induce responses at the cellularand subcellular level both in gills and digestive gland andDNA damage in haemocytes. The response of these organ-isms confirms the potential risk of ENPs to aquatic life. Itappears that, for these organisms, photocatalytic ageing ofthe nTiO2 does not cause any appreciable difference incellular/subcellular toxicity, and the bulk form may bemore toxic with respect to histopathological and histochem-ical changes. The use of Cu as a positive control has validatedthese findings, which are in contrast to many earlier authorswho have reported that bulk is more ecotoxicologically inertthan nanoparticulate forms. This study has clear implica-tions for the field of nano-ecotoxicology and environmentalprotection. Further research is required to investigate thecause of elevated bulk toxicity and to investigate this effect inother species and with other ENPs and their bulkequivalents.

Acknowledgements

Partial financial support from the European Regional Devel-opment Fund, INTERREG IVA (Grant No. 4059), is gratefullyacknowledged. We thank Richard Handy (Plymouth) andAndrew Collins (Oslo) for providing the samples of nTiO2

and bacterial enzymes, respectively.

A. D’Agata et al.

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

Declaration of interest

The authors report no conflicts of interest. The authors aloneare responsible for the content and writing of the paper.

ReferencesAl-Jubory AR, Handy RD. 2012. Uptake of titanium from TiO2 nano-

particle exposure in the isolated perfused intestine of rainbow trout:vanadate and novel CO2- sensitive components. Nanotoxicology inpress; DOI: 10.3109/17435390.2012.735268.

Al-Subiai SN, Arlt VM, Frickers PE, Readman JW, Stolpe B, Lead JR, etal. 2012. Merging nano-genotoxicology with eco-genotoxicology: Anintegrated approach to determine interactive genotoxic and sub-lethal toxic effects of C60 fullerenes and fluoranthene in marinemussels, Mytilus sp. Mutat Res Gen Tox En 745:92–103.

Al-Subiai SN, Jha AN, Moody AJ. 2009. Contamination of bivalvehaemolymph samples by adductor muscle components: implica-tions for biomarker studies. Ecotoxicology 18:334–342.

Al-Subiai SN, Moody AJ, Mustafa SA, Jha AN. 2011. A multiple bio-marker approach to investigate the effects of copper on the marinebivalve mollusc, Mytilus edulis. Ecotox Environ Safe 74:1913–1920.

Baun A, Hartmann N, Grieger K, Kusk K. 2008. Ecotoxicity of engi-neered nanoparticles to aquatic invertebrates: a brief review andrecommendations for future toxicity testing. Ecotoxicology 17:387–395.

Bettercourt R, Dando P, Rosa D, Riou V, Colaco A, Sarrazin J, et al.2008. Changes of gill and hemocyte-related bio-indicators duringlong term maintenance of the vent mussel Bathymodiolus azoricusheld in aquaria at atmospheric pressure. Comp Biochem PhysiolPart A 150:1–7.

Bolognesi C, Landini E, Roggieri P, Fabbri R, Viarengo A. 1999. Geno-toxicity biomarkers in the assessment of heavy metal effects inmussels: Experimental studies. Environ Mol Mutagen 33:287–292.

Brendler-Schwaab S, Hartmann A, Pfuhler S, Speit G. 2005. The in vivocomet assay: use and status in genotoxicity testing. Mutagenesis20:245–254.

Brunelli E, Mauceri A, Maisano M, Bernabò I, Giannetto A, DeDomenico E, et al. 2011. Ultrastructural and immunohistochemicalinvestigation on the gills of the teleost, Thalassoma pavo L., exposedto cadmium. Acta Histochem 113(2):201–213.

Canesi L, Ciacci C, Fabbri R, Marcomini A, Pojana G, Gallo G. 2012.Bivalve molluscs as a unique target group for nanoparticle toxicity.Mar Environ Res 76:16–21.

Canesi L, Fabbri R, Gallo G, Vallotto D, Marcomini A, Pojana G. 2010.Biomarkers in Mytilus galloprovincialis exposed to suspensions ofselected nanoparticles (Nano carbon black, C60 fullerene, Nano-TiO2, Nano-SiO2). Aquat Toxicol 100:168–177.

Canty MN, Hutchinson TH, Brown RJ, Jones MB, Jha AN. 2009. Linkinggenotoxic responses with cytotoxic and behavioural or physiologicalconsequences: Differential sensitivity of echinoderms (Asteriasrubens) and marine molluscs (Mytilus edulis). Aquat Toxicol94:68–76.

Cappello T, Maisano M, D’Agata A, Natalotto A, Mauceri A,Fasulo S. 2013. Effects of environmental pollution in caged mussels(Mytilus galloprovincialis). Mar Environ Res doi: 10.1016/j.marenvres.2012.12.010.

Chen J, Liu M, Zhang J, Ying X, Jin L. 2004. Photocatalytic degradationof organic wastes by electrochemically assisted TiO2 photocatalyticsystem. J Environ Manage 70:43–47.

Chomczynski P, Sacchi N. 1987. Single-step method of RNA isolationby acid guanidinium thiocyanate–phenol–chloroform extraction.Anal Biochem 162:156–159.

Ciacci C, Barmo C, Gallo G, Maisano M, Cappello T, D’Agata A, et al.2012. Effects of sub-lethal, environmentally relevant concentrationsof hexavalent Chromium in the gills of Mytilus galloprovincialis.Aquat Toxicol 120–121:109–118.

Collins AR, Ma AG, Duthie SJ. 1995. The kinetics of repair of oxidativeDNA damage (strand breaks and oxidised pyrimidines) in humancells. Mutat Res 336:69–77.

Dallas LJ, Bean TP, Turner A, Lyons BP, Jha A. 2013. OxidativeDNA damage may not mediate Ni-induced genotoxicity in marinemussels: Assessment of genotoxic biomarkers and transcriptionalresponses of key stress genes. Mutat Res 754:22–31.

Dodd NJF, Jha AN. 2009. Titanium dioxide induced cell damage:a proposed role of the carboxyl radical. Mutat Res 660:79–82.

Dodd NJF, Jha AN. 2011. Photoexcitation of aqueous suspensions oftitanium dioxide nanoparticles: an electron spin resonance spintrapping study of potentially oxidative reactions. Photochem Photo-biol 87:632–640.

Dondero F, Piacentini L, Banni M, Rebelo M, Burlando B,Viarengo A. 2005. Quantitative PCR analysis of two molluscanmetallothionein genes unveils differential expression and regula-tion. Gene 345:259–270.

Esterkin CR, Negro AC, Alfano OM, Cassano AE. 2005. Air pollutionremediation in a fixed bed photocatalytic reactor coated with TiO2.AIChE J 51(8):2298–2310.

European Commission. 2005. Scientific Committee on Emerging andNewly Identified Health Risks (SCENIHR). Modified Opinion (afterpublic consultation) on the appropriateness of existing methodol-ogies to assess the potential risks associated with engineered andadventitious products of nanotechnologies. http://ec.europa.eu/health/ph_risk/committees/04_scenihr/docs/scenihr_o_003b.pdf;Luxembourg, European Commission: 1-79.

Farré M, Gajda-Schrantz K, Kantiani L, Barceló D. 2009. Ecotoxicityand analysis of nanomaterials in the aquatic environment. AnalBioanal Chem 393:81–95.

Fasulo S, Iacono F, Cappello T, Corsaro C, MaisanoM, D’Agata A, et al.2012. Metabolomic investigation of Mytilus galloprovincialis(Lamarck 1819) caged in aquatic environments. Ecotox Envir Saf84:139–146.

Fasulo S, Marino S, Mauceri A, Maisano M, Giannetto A, D’Agata A, etal. 2010b. A multi biomarker approach in Coris julis living in anatural Environment. Ecotox Environ Safe 73:1565–1573.

Fasulo S, Mauceri A, Giannetto A, Maisano M, Bianchi N,Parrino V. 2008. Expression of metallothionein mRNAs by in situhybridization in the gills of Mytilus galloprovincialis, from naturalpolluted environments. Aquat Toxicol 88:62–68.

Fasulo S, Mauceri A, MaisanoM, Giannetto A, Parrino V, Gennuso F, etal. 2010a. Immunohistochemical and molecular biomarkers in Corisjulis exposed to environmental contaminants. Ecotox Environ Safe73:873–882.

Federici G, Shaw BJ, Handy RD. 2007. Toxicity of titanium dioxidenanoparticles to rainbow trout (Oncorhynchus mykiss): Gill injury,oxidative stress, and other physiological effects. Aquat Toxicol84:415–430.

Fouqueray M, Dufils B, Vollat B, Chaurand P, Botta C, Abacci K, et al.2012. Effects of aged TiO2 nanomaterial from sunscreen on Daphniamagna exposed by dietary route. Environ Pollut 163:55–61.

Handy R, Shaw B. 2007. Toxic effects of nanoparticles and nanoma-terials: implications for public health, risk assessment and the publicperception of nanotechnology. Health Risk Soc 9:125–144.

Handy R, von der Kammer F, Lead J, HasselloÌM, Owen R, et al. 2008.The ecotoxicology and chemistry of manufactured nanoparticles.Ecotoxicology 17:287–314.

Handy RD, van den Brink N, Chappell M, Mühling M, Behra R,Dusinskà M, et al. 2012. Practical considerations for conductingecotoxicity test methods with manufactured nanomaterials: whathave we learnt so far? Ecotoxicology 21:933–972.

Hartwig A, Asmuss M, Ehleben I, Herzer U, Kostelac D, Pelzer A, et al.2002. Interference by toxic metal ions with DNA repair processesand cell cycle control: molecular mechanisms. Environ HealthPerspect 110 S5:797–799.

Henderson L, Wolfreys A, Fedyk J, Bourner C, Windebank S. 1998. Theability of the Comet assay to discriminate between genotoxins andcytotoxins. Mutagenesis 13:89–94.

Jassby D, Budarz JF, Wiesner M. 2012. Impact of aggregate size andstructure on the photocatalytic properties of TiO2 and NO nano-particles. Environ Sci Technol 46:6934–6941.

Jha AN, Dogra Y, Turner A, Millward GE. 2005. Impact of low doses oftritium on the marine mussel, Mytilus edulis: genotoxic effects andtissue-specific bioconcentration. Mutat Res 586:47–57.

Jha AN. 2008. Ecotoxicological applications and significance of thecomet assay. Mutagenesis 23:207–221.

Kaegi R, Ulrich A, Sinnet B, Vonbank R, Wichser A, Zuleeg S, et al. 2008.Synthetic TiO2 nanoparticle emission from exterior facades into theaquatic environment. Environ Pollut 156(2):233–239.

Kiser MA, Westerhoff P, Benn T, Wang Y, Perez-Rivera J,Hristovski K. 2009. Titanium nanomaterial removal and releasefrom wastewater treatment plants. Environ Sci Technol 43(17):6757–6763.

Klaine SJ, Alvarez PJ, Batley GE, Fernandes TF, Handy RD, Lyon DY, etal. 2008. Nanomaterials in the environment: Behavior, fate, bio-availability, and effects. Environ Toxicol Chem 27(9):1825–1851.

Biological effects of different forms of titanium dioxide in mussels

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

Labille J, Feng J, Botta C, Borschneck D, Sammut M, Cabie M, et al.2010. Aging of TiO2 nanocomposite used in sunscreen. Dispersionand fate of the degradation products in aqueous environment.Environ Pollut 158. 3482:3489.

Lapied E, Nahmani JY, Moudilou E, Chaurand P, Labille J, Rose J, et al.2011. Ecotoxicological effects of an aged TiO2 nanocomposite mea-sured as apoptosis in the anecic earthworm Lumbricus terrestris afterexposure through water, food and soil. Environ Int 37:1105–1110.

Lemoine S, Bigot Y, Sellos D, Cosson RP, Laulier M. 2000. Metallothio-nein isoforms in Mytilus edulis (Mollusca, Bivalvia): complementaryDNA characterization and quantification of expression in differentorgans after exposure to cadmium, zinc, and copper. Mar Biotech-nol 2:195–203.

Lin H, Huang CP, Li W, Ni C, Shah SI, Tseng YH. 2006. Size dependencyof nanocrystalline TiO2 on its optical property and photocatalyticreactivity exemplified by 2-chlorophenol. Appl Catal B 68:1–11.

Millward GE, Kadam S, Jha AN. 2012. Tissue-specific assimilation,depuration and toxicity of nickel in Mytilus edulis. Environ Pollut162:406–412.

Minniti F, Maisano M, Mauceri A, Giannetto A, Fasulo S. 2009. GTH Iand GTH II in the pituitary gland of swordfish (Xiphias gladius). Ital JZool 76(3):269–278.

Moore MN. 2006. Do nanoparticles present ecotoxicological risks forthe health of the aquatic environment? Environ Int 32:967–976.

Mustafa SA, Davies SJ, Jha AJ. 2012. Determination of hypoxia anddietary copper mediated sub-lethal toxicity in carp, Cyprinus carpio,at different levels of biological organisation. Chemosphere 87:413–422.

Nagaveni K, Sivalingam G, Hegde MS, Madras G. 2004. Photocatalyticdegradation of organic compounds over combustion-synthesizednano-TiO2. Environ Sci Technol 38:1600–1604.

Nakagawa Y, Wakuri S, Sakamoto K, Tanaka N. 1997. The photoge-notoxicity of titanium dioxide particles. Mutat Res Genet ToxicolEnviron Mutagen 394:125–132.

Nowack B, Bucheli TD. 2007. Occurrence, behavior and effects ofnanoparticles in the environment. Environ Pollut 150:5–22.

Oberdörster G, Oberdörster E, Oberdörster J. 2005. Nanotoxicology:an emerging discipline evolving from studies of ultrafine particles.Environ Health Persp 113:823–839.

Ramsden CS, Smith TJ, Shaw BJ, Handy RD. 2009. Dietary exposure totitanium dioxide nanoparticles in rainbow trout (Oncorhynchusmykiss): no effect on growth, but subtle biochemical disturbancesin the brain. Ecotoxicology 18:939–951.

Reeves JF, Davies SJ, Dodd NJF, Jha AN. 2008. Hydroxyl radicals (OH)are associated with titanium dioxide (TiO2) nanoparticle-induced

cytotoxicity and oxidative DNA damage in fish cells. Mutat Res640:113–122.

Scown TM, van Aerle R, Tyler CR. 2010. Review: Do engineerednanoparticles pose a significant threat to the aquatic environment?Crit Rev Toxicol 40:653–670.

Singh J, Mclean JA, Pritchard DE, Montaser A, Patierno SR. 1998.Sensitive quantitation of chromium-DNA adducts by inductivelycoupled plasma mass spectrometry with a direct injection high-efficiency nebulizer. Toxicol Sci 46:260–265.

Tseng YH, Lin HY, Kuo CS, Li YY, Huang CP. 2006. Thermostabilityof nano-TiO2 and its photolytic activity. React Kine Catal Lett 89:63–69.

Tsuang Y, Sun J, Huang Y, Lu C, Chang W, Wang C. 2008. Studies ofphotokilling of bacteria using titanium dioxide nanoparticles. ArtifOrgans 32:167–174.

Uchino T, Tokunaga H, Ando M, Utsumi H. 2002. Quantitative deter-mination of OH radical generation and its cytotoxicity induced byTiO2-UVA treatment. Toxicol In Vitro 16:629–635.

Van Herwijnen R, Laverye T, Poole J, Hodson ME, Hutchings TR. 2007.The effect of organic materials on the mobility and toxicity of metalsin contaminated soils. Appl Geochem 22:2422–2434.

Vevers WF, Jha AN. 2008. Genotoxic and cytotoxic potential of titaniumdioxide (TiO2) nanoparticles on fish cells in vitro. Ecotoxicology17:410–420.

Wang H, Wick R, Xing B. 2009. Toxicity of nanoparticulate and bulkZnO, Al2O3 and TiO2 to the nematode Caenorhabditis elegans.Environ Pollut 157:1171–1177.

Ward JE, Kach DJ. 2009. Marine aggregates facilitate ingestion ofnanoparticles by suspension-feeding bivalves. Mar Environ Res 68(3):137–142.

Weir A, Westerhoff P, Fabricius L, Hristovski K and von Goetz N. 2012.Titanium dioxide nanoparticles in food and personal care products.Environ Sci Technol 46(4):2242–2250.

Windeatt KM, Handy RD. 2012. Effect of nanomaterials on the com-pound action of the shore crab, Carcinus maenas. NanotoxicologyDOI: 10.3109/17435390.2012.663809.

Xiong D, Fang T, Yu J, Sima X, Zhu W. 2011. Effects of nano-scale TiO2,ZnO and their bulk counterparts on zebra fish: Acute toxicity,oxidative stress and oxidative damage. Sci Total Environ 409:1444–1452.

Zhu XS, Zhu I, Duan ZH, Qi RQ, Li Y, Lang YP. 2008. Comparativetoxicity of several metal oxide nanoparticle aqueous suspensions toZebrafish (Danio rerio) early developmental stage. J Environ SciHealth A Tox Hazard Subst Environ Eng 43:278–284.

Supplementary materials available online

Figures S1, S2, S3, S4 and S5

Table S1

A. D’Agata et al.

Nan

otox

icol

ogy

Dow

nloa

ded

from

info

rmah

ealth

care

.com

by

Rac

hel A

ndre

ws

on 0

7/10

/13

For

pers

onal

use

onl

y.

D’Agata et al.: Manuscript ID: TNAN-2013-0049

Table S1. Concentration of elements in different forms of TiO2 used in this study, as measured by ICP-MS (µg g-1 TiO2)

TiO2 Form Be B Na Mg Al K Ca V Cr Mn Fe Co nTiO2 bd 86.45 ± 28.875 680.5 ± 19.845 313.45 ±

56.8 ± 3.9 543.5 ± 27.29 166.25 ±

4.905 ± 0.28 5.205 ±

0.305 ±

49.03 ± 9.88 0.075 ± 0.03

Bulk bd 10.17 ± 4.405 756 ± 19.2 357.1 ±

171.3 ±

1970.5 ±

151.75 ± 15.1 7.005 ±

0.95 ± 0.105 bd 10.235 ±

0.04 ± 0.02 Ni Cu Zn As Se Mo Ag Cd Sb Ba Tl Pb

nTiO2 1.58 ± 0.265 36.225 ± 1.51 67.05 ± 0.945 2.335 ±

0.55 ± 0.235 3.065 ± 0.045 0.115 ± 0.015 0.185 ±

0.79 ± 0.11 6.08 ± 0.15 0.135 ± 0.01 11.15 ± Bulk 1.215 ± 0.305 37.765 ± 0.49 73.5 ± 1.615 7.475 ± 0.51 bd 6.84 ± 0.37 0.15 ± 0.02 0.165 ±

1.58 ± 0.035 0.715 ±

0.13 ± 0.015 5.22 ± 0.075

a Values are presented as mean ± one SD; bd Element was below the limit of detection for ICP-MS.

D’Agata et al.: Manuscript ID: TNAN-2013-0049