ology 222 (2007) 141–151www.elsevier.com/locate/ytaap

Toxicology and Applied Pharmac

Endocytosis, oxidative stress and IL-8 expression in human lung epithelialcells upon treatment with fine and ultrafine TiO2: Role of the specific surface

area and of surface methylation of the particles

Seema Singh a,d, Tingming Shi a,e, Rodger Duffin a,f, Catrin Albrecht a, Damien van Berlo a,Doris Höhr a, Bice Fubini b, Gianmario Martra b, Ivana Fenoglio b,

Paul J.A. Borm a,c, Roel P.F. Schins a,⁎

a Institut für umweltmedizinische Forschung (IUF) an der Heinrich-Heine Universität Düsseldorf gGmbH, Germanyb Dipartimento di Chimica IFM, Interdepartmental Center “G. Scansetti” for Studies on Asbestos and other Toxic Particulates and

NIS Centre of Excellence of Nanostructured Interfaces and Surfaces, Università degli Studi di Torino, Italyc Hogeschool Zuyd, Heerlen, The Netherlands

d The Energy and Resources Institute (TERI), New Delhi, Indiae Hubei Provincial Centre for Disease Control and Prevention, Wuhan, Hubei, PR China

f Centre for Inflammation Research (CIR), The Queen's Medical Research Institute, University of Edinburgh, UK

Received 24 August 2006; revised 4 May 2007; accepted 9 May 2007Available online 18 May 2007

Abstract

Inhaled ultrafine particles show considerably stronger pulmonary inflammatory effects when tested at equal mass dose with their finecounterparts. However, the responsible mechanisms are not yet fully understood. We investigated the role of particle size and surface chemistry ininitiating pro-inflammatory effects in vitro in A549 human lung epithelial cells on treatment with different model TiO2 particles. Two samples ofTiO2, i.e. fine (40–300 nm) and ultrafine (20–80 nm) were tested in their native forms as well as upon surface methylation, as was confirmed byFourier transformed infrared spectroscopy. Radical generation during cell treatment was determined by electron paramagnetic resonance with 5,5-dimethyl-1-pyrroline-N-oxide or 4-hydroxy-2,2,6,6-tetramethylpiperidine-1-oxyl. Interleukin-8 mRNA expression/release was determined by RT-PCR and ELISA, whereas particle uptake was evaluated by transmission electron microscopy. TiO2 particles were rapidly taken up by the cells,generally as membrane bound aggregates and large intracellular aggregates in vesicles, vacuoles and lamellar bodies. Aggregate size tended to besmaller for the ultrafine samples and was also smaller for methylated fine TiO2 when compared to non-methylated fine TiO2. No particles wereobserved inside nuclei or any other vital organelle. Both ultrafine TiO2 samples but not their fine counterparts elicited significantly strongeroxidant generation and IL-8 release, despite their aggregation state and irrespective of their methylation. The present data indicate that ultrafineTiO2, even as aggregates/agglomerates, can trigger inflammatory responses that appear to be driven by their large surface area. Furthermore, ourresults indicate that these effects result from oxidants generated during particle–cell interactions through a yet to be elucidated mechanism(s).© 2007 Elsevier Inc. All rights reserved.

Keywords: Titanium dioxide; Oxidative stress; Ultrafine particles; Nanoparticles; Inflammation; Lung epithelial cells; Endocytosis

Introduction

Particles less than 100 nm in size, which are commonlyreferred to as ultrafine particles by toxicologists and nanopar-ticles by material scientists, have been considered to beimportant in driving the adverse health effects of particulate

⁎ Corresponding author. Fax: +49 211 3389 331.E-mail address: roel.schins@uni-duesseldorf.de (R.P.F. Schins).

0041-008X/$ - see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.taap.2007.05.001

air pollution (Peters et al., 1997; Utell and Frampton, 2000;Donaldson et al., 2005). Experimental support for this hypoth-esis has been predominantly provided from studies in rats withcommercially available model particles such as carbon black(CB) and titanium dioxide (TiO2), where such model particles inthe ultrafine size range gave markedly stronger inflammatoryand toxic effects in the lungs of experimental animals whencompared to their non-ultrafine counterparts (Ferin et al., 1992;Li et al., 1999; Oberdorster et al., 2000; Donaldson et al., 2002;

Table 1Characteristics of fine and ultrafine TiO2 samples

Sample BET-specificsurface area(m2 g−1)*

Size range(nm)

Crystal phase Source/name

Fine 10 40–300 pure anatase MerckUltrafine 50 20–80 80% anatase,

20% rutileDegussa P25

*According to Brunauer, Emmett and Teller (data provided by company).

142 S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

Oberdorster et al., 2005a). Taken together with in vitro findingswith such materials (e.g. Rahman et al., 2002; Brown et al.,2004), these studies have raised questions about the possibleadverse health implications for large scale, uncontrolledcommercial production of nanoparticles (Colvin, 2003; Bormand Kreyling, 2004; Nel et al., 2006; Borm et al., 2006; Unfriedet al., 2007). Notably, both CB and TiO2 have recently been (re)classified as possible human carcinogens (group 2B) by theInternational Agency of Research on Cancer, which was partlysupported by mechanistic considerations on the importance ofparticle-induced chronic inflammation in tumorigenesis in ratinhalation studies (Baan et al., 2006).

Concerning the mechanism of toxicity of nanoparticles, ithas been suggested that the specific surface area and/or itschemical composition are major determinants (Donaldson et al.,2002; Oberdorster et al., 2005a). Awide variety of low toxicity,poorly soluble particle types have been shown to induceinflammation proportional to their surface area exposed,suggesting that a large amount of surface area alone in thelung may be sufficient to initiate inflammation by particles(Duffin et al., 2002; Stoeger et al., 2006). One hypothesis toexplain these surface related effects is the generation of cellularoxidative stress leading to the activation of redox sensitivesignaling pathways that culminate in the transcription of pro-inflammatory cytokines and chemokines (Donaldson et al.,2002; Oberdorster et al., 2005a; Unfried et al., 2007). In supportof this hypothesis, several types of ultrafine particles have beenshown to elicit stronger oxidizing properties than their finecounterparts, if subjected to acellular assays such as plasmidDNA unwinding/breakage or fluorescence dye oxidation ifcompared on a per weight basis (Donaldson et al., 1996; Brownet al., 2001). Notably, however, apart from direct oxidantgeneration, particles can also generate cellular derived ROS andoxidative stress in an indirect manner, i.e. as a response ofmacrophages and epithelial cells to particle interaction. Indeed,several particulate materials including asbestos, quartz, coaldust or ambient particulate matter have been shown to generateROS via processes involving activation of NAD(P)H-likeenzyme systems or via modulation of mitochondrial respiration(Upadhyay et al., 2003; Li et al., 2003; Fubini and Hubbard,2003; Knaapen et al., 2004). However, the significance of thiscell-mediated ROS generation for the observed contrastingeffects between fine and ultrafine particles is currently stillpoorly understood (Xia et al., 2006).

Among the best investigated material in current nanotoxicol-ogy research is TiO2. Non-ultrafine, pigmentary grade TiO2 hasseen wide applications, e.g. as paint filler, food additive or insunscreen formulations, and has been incorporated as a negativecontrol in many particle toxicology studies. In contrast, ultrafineTiO2 has been shown to elicit toxic and inflammatory effects in avariety of in vitro and in vivo studies. Hallmark studies byOberdorster and colleagues demonstrated that ultrafine TiO2,when instilled intratracheally into rats and mice, induced a muchgreater pulmonary inflammatory response when compared tofine TiO2 at the same instilled mass dose (reviewed inOberdorster et al., 2005a). In the presence of UV light, TiO2

can be highly photoreactive and such a mechanism has been

considered to drive free radical-mediated toxicity and DNAdamage in skin cells (Dunford et al., 1997). Notably however, inthe absence of photosensitization, ultrafine TiO2 has also beenshown to elicit comparatively stronger plasmid DNA unwindingthan fine TiO2 (Donaldson et al., 1996), although this was notobserved in subsequent studies (Dick et al., 2003). Takentogether, many studies nowadays suggest that TiO2 in thenanosize range represents a potentially toxic material. However,in contrast to many engineered nanoparticles, TiO2 typicallyoccurs as aggregates of particles in biological environments(Stearns et al., 2001; Rehn et al., 2003). Furthermore, manycommercial forms of TiO2 contain coated surfaces, a procedurethat is usually applied with engineered nanoparticles to improvetheir physicochemical properties. Such aspects of aggregationand agglomeration, as well as surface coatings of nanoparticlesare very likely to affect their interactions with cells andsubsequent toxic stress responses. The effect of size and ofsurface methylation was investigated in our laboratories in aprevious study by monitoring the effect on intratrachealinstillation in a rat model of fine and ultrafine TiO2, as preparedor surface methylated. This study showed a stronger inflamma-tory toxicity for the ultrafine particles compared with their finecounterparts, while the effect of methylation on both sampleswas negligible (Höhr et al., 2002).

In order to gain further insight into the mechanisms wherebythese contrasting effects occurred, we investigated the inflam-matory properties of the same TiO2 samples in vitro, in relationto their (1) primary particle size distribution, (2) aggregation/agglomeration state, (3) hydrophobicity and (4) radicalgenerating potential. Therefore, A549 human lung epithelialcells were treated with either fine or ultrafine TiO2 in theirnative or methylated form and evaluated for the expression andrelease of the chemokine interleukin-8 by RT-PCR and ELISA,respectively. Transmission electron microscopy was used todetermine particle endocytosis and state of aggregation of theparticles for the different samples. The radical inducing capacityof the different samples in A549 cell cultures was determined bymeans of electron paramagnetic resonance (EPR) with spin-trapping techniques.

Methods

Titanium dioxide samples. Two commercial samples, representing fine TiO2

and ultrafine TiO2 products, were selected for the present study. Physicalcharacteristics of the powders are reported in Table 1. Both samples were of ahigh cationic purity. As for anions, the fine TiO2 contained traces of sulphatespecies (b0.005 wt%), while in the ultrafine TiO2 a higher amount of chlorideimpurities (b0.08 wt%) was present. Such level of Cl− ions does not affect the

143S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

Lewis surface properties of stoichiometric TiO2, as resulting from a study on asimilar Cl-free TiO2 powder (Morterra et al., 1980). Both samples weresubjected to a methylation procedure in order to change the substantiallyhydrophilic surface of these particles into a more hydrophobic one. Thus, in totalfour different TiO2 samples were applied in the present study, i.e. the nativeforms of fine titanium dioxide (F-TiO2) and ultrafine titanium dioxide (UF-TiO2), as well as the methylated forms of fine titanium dioxide (MF-TiO2) andultrafine titanium dioxide (MUF-TiO2). The same samples have previously alsobeen used in an in vivo study by our group (Höhr et al., 2002).

Methylation of titanium dioxide particles. For this procedure both sampleswere outgassed at room temperature for 45 min in a conventional vacuum line,under a residual pressure of 1.0×10−6 Torr (1 Torr=133.33 Pa); the outgassingtemperature was then raised up to 250 °C and kept at such a value for 45 min, toremove most of the water molecules adsorbed on the TiO2 surface withoutsignificant loss of surface hydroxyl groups by condensation. Subsequently,CH3OHwas added (40 Torr) and the samples were kept in contact with methanolfor 45 min at 250 °C and finally outgassed at room temperature for 45 min. Theefficacy of methylation of the TiO2 dusts was investigated by Fouriertransformed infrared spectroscopy (FTIR) (Bruker Vector 22, detector: MCT;res=4 cm−1). Therefore, the TiO2 powders were pressed in the form of self-supporting pellets (ca. 20 mg·cm−2) and placed in a conventional IR quartz cellequipped with KBr windows, permanently connected to a vacuum line, allowingall thermal adsorption–desorption experiments to be carried out in situ.

Evaluation of endotoxin contamination. The four different TiO2 samplesused in the present study were also analyzed for possible endotoxincontamination. Therefore, the four different particle preparations weresuspended in endotoxin free water and then subjected to a quantitative kineticchromogenin Limulus Amoebocyte Lysate (LAL) method (Bio Witthaker),using Escherichia coli 055:B5 endotoxin (Bio Witthaker) as standard.

Culture and treatment of A549 human lung epithelial cell line. In the pre-sent study, we used A549 human lung epithelial cells. This cell line has beenproved to be a suitable in vitro model to study endocytosis by ultrafine particlesincluding TiO2 (e.g. Stearns et al., 2001), as well as to investigate theinflammatory potential of various particles (Simeonova and Luster, 1996;Schins et al., 2000). A549 cells (American Type Culture Collection/ATTC) weregrown in Dulbecco's Modified Eagle's Medium (DMEM; Life Sciences),supplemented with 10% heat inactivated fetal calf serum (FCS; Life Sciences),L-glutamine (Life Sciences) and 30 IU/ml penicillin-streptomycin (LifeSciences) at 37 °C and 5% CO2. For experiments, cells were grown until nearconfluency (90–95%) in complete culture medium. The medium was thenreplaced with FCS-free medium for 24 h until treatment. Immediately before thestart of the incubations, particle suspensions were prepared in culture medium orHBSS and sonicated for 5 min using a water bath (Sonorex TK52 water bath;60 W, 35 kHz). A549 cells were rinsed twice with phosphate-buffered saline(PBS) and then immediately treated with the particle suspensions at theindicated concentrations and incubation time intervals. The respirable quartzstandard sample DQ12 (9.6 m2 g−1, Batch 6-IUF) and Tumour Necrosis Factoralpha (TNF-α) were used as particulate and non-particulate positive controls,respectively (Schins et al., 2000; Fiedler et al., 1998).

Analysis of particle uptake in A549 cells by transmission electron microscopy.A549 cells were grown in complete culture medium in 35 mm culture dishesuntil 90–95% confluency and then cultured for a further 24 h in serum freemedium. The cells were then washed twice with PBS and treated with 16 or80 μg/cm2 native or methylated TiO2 for 4 h in complete medium. Followingincubation, the cell monolayers were immediately rinsed three times with serumfree medium and then fixed with 2% glutaraldehyde in 0.1 M sodium cacodylatebuffer for 1 h at 4 °C. After post-fixation in 2% OsO4 in 0.1 M sodiumcacodylate buffer for 1 h at 4 °C, the samples were en bloc stained with 1.5%uranylacetate dihydrate and phosphotungstic acid, dehydrated in ethanol seriesand embedded in Epon epoxy resin (Serva, Heidelberg, Germany). Morphologiccharacteristics of the cells and the distribution and agglomeration state of theparticles within the cells were investigated using ultra thin sections (50 nm)placed on 150 mesh grids and examined by transmission electron microscopy

(STEM CM12, Philips) in combination with a digital imaging system (SIS,Münster, Germany). Energy Dispersive X-ray analysis (EDX) and elementalmapping was used for identification of the TiO2 samples.

Analysis of reactive oxygen species formation by electron paramagnetic

resonance (EPR) analysis with the spin-trapping technique. Electron para-magnetic resonance (EPR) spectroscopy in combination with spin-trappingagents is a technique used extensively to detect and identify many short-livedfree radical compounds, such as superoxide and hydroxyl radicals, in biologicalsystems. With this approach we determined the formation of reactive oxygenspecies by the different preparations of TiO2. Two different spin-trapping agentswere used, i.e. the spin-trap 5,5-dimethyl-1-pyrroline-N-oxide (DMPO), whichgives stable radical adducts with small free radicals, and the spin-probeTEMPOL (4-hydroxy-TEMPO 4-hydroxy-2, 2, 6, 6 tetramethylpiperidine-1-oxyl), a stable radical which is progressively blunted upon contact with radicals.The generation of radicals during the treatment of the A549 cells with thedifferent particle preparations was determined as follows. Cells were grown asdescribed before, starved for 24 h using serum free medium, rinsed twice withPBS and then immediately treated with the different particle samples. All TiO2

samples were freshly suspended in HBSS at 0.8 mg/ml, sonicated for 5 min(Sonorex TK52 water bath; 60 W, 35 kHz) and then 450 μl of these suspensionswere immediately added to the cell culture dishes. Following 1 h incubationeither 50 μl of 0.05 mM TEMPOL or 1 M DMPO (in PBS) was added to the cellcultures for 1 h or 3 h. Control incubations were performed for both spin-trapping agents in the absence of the various particle samples and/or A549 cells,using identical experimental settings (i.e. incubation in culture dishes at 37 °Cand 5% CO2). Supernatants were collected at the indicated time points, brieflyvortexed, immediately transferred into a 50-μl glass capillary and measured witha Miniscope MS100 EPR spectrometer (Magnettech, Berlin, Germany). TheEPR-spectra were recorded at room temperature using the followinginstrumental conditions: magnetic field: 3360 G; sweep width: 100 G; scantime: 30 s; number of scans: 3; modulation amplitude: 1.8 G; receiver gain:1000. The experiments done with DMPO in the absence of A549 cells were alsoperformed in the presence of ethanol to determine if the signal was due tohydroxyl or superoxide radicals.

Notably, the experimental conditions chosen are far from what usuallyemployed to detect free radicals in cell-free suspensions of particles (Fubini et

al., 1995; Fenoglio et al., 2001). Larger concentrations of the spin-trappingreagents and longer incubation times were used to adapt the conditions tocellular tests. Kinetic investigations using different addition times for DMPOfollowing particle treatment in the A549 cell culture (i.e. immediate addition,20 min, 60 min), confirmed that the selected protocol was optimal for signaldetection in this system. It was decided not to consider ESR measurements usingconcomitant addition of spin-trapping agents and the particles to avoid potentialartefacts that may result from cell activation by medium replacement, adsorptionof DMPO or TEMPOL onto the particle surface, or the associated facilitation ofinternalization of these reagents into the cells during initial particle endocytosis.In support of this, for DMPO the concomitant addition with particles was foundto result in lacking signal detection (data not shown). Besides, we also did notconsider to perform ESR measurements of particle treated A549 cells insuspension after detachment of cells (by trypsinization or scraping) as describedelsewhere (Zhang et al., 2001). Such approach does not reflect exposureconditions for lung epithelial cells (i.e. treatment of cells in monolayer), whichwe also used for the comparative analysis of particle uptake and IL-8 expressionanalysis.

Interleukin-8 mRNA and protein expression. Human IL-8 messenger RNA(mRNA) expression was determined by reverse transcriptase polymerase chainreaction (RT-PCR). Total RNA was extracted from the treated A549 cells withTRIZOL reagent (Invitrogen) using the recommended protocol. The PromegaAccess RT-PCR System kit was used for the amplification of human IL-8mRNA and the human housekeeping gene, glyceraldehyde-3-phosphatedehydrogenase (GAPDH), according the manufacturer's instructions.Sequences of the primers used, and details of the RT-PCR conditions arepreviously described (Schins et al., 2000). The release of IL-8 from the A549cells was determined by with an enzyme-linked immunosorbent assay (ELISA),as follows. After incubation with the different particle preparations, cellsupernatants were collected and immediately centrifuged (5 min, 15000 rpm).

144 S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

The resulting supernatants were stored at −20 °C until measurement (PeliKinecompactTM human IL-8 ELISA kit, Sanquin, The Netherlands). For eachtreatment, cytokine release is expressed as a percentage of the IL-8 release asmeasured in the control incubations.

Statistical analysis. Data are expressed as mean±SEM unless stated other-wise. Treatment-related differences were evaluated using one-way analysis ofvariances (ANOVA), with Tukey LSD or Dunett post hoc comparisons. Adifference was considered significant at Pb0.05.

Results

Methylation of the TiO2 samples

The efficiency of particle methylation was determined byFTIR spectroscopy. Representative IR spectra for ultrafine TiO2

(UF-TiO2) are shown in Fig. 1. Spectrum “a” corresponds toTiO2 outgassed at room temperature for 45 min. In thiscondition a full monolayer of hydroxyl groups, both vibration-ally “free” (i.e. not interacting through hydrogen bonding withneighbor OH) or hydrogen bonded, as well as water moleculescoordinated onto surface Ti4+ cations (coordinatively unsatu-rated) are visible. Such features represent the first hydrationlayer of the materials, which, in the experimental conditionsadopted, are left on the surface. After treatment in CH3OH andsubsequent outgassing at room temperature (spectrum b), on the

Fig. 1. Fourier-transformed infrared spectra of ultrafine titanium dioxide beforeand after methylation. The IR spectra reveals the absorbance bands which areattributed to the presence of distinctive hydroxyls, water molecules and –OCH3

groups on the particle surfaces. The graph represents UF-TiO2 after outgassingat room temperature (spectrum a), after treatment at 250 °C in CH3OH (40 Torr)for 45 min with subsequent outgassing at room temperature (spectrum b) andafter three cycles of readmission of H2O vapor (18 Torr) at room temperature/contact for 30 min/outgassing at room temperature (spectrum c).

one hand the bands due to H2O molecules disappeared, becauseof water desorption, as well as signals due to vibrationally freehydroxyl, while a fraction of the absorptions related to H-bonded surface hydroxyls is left. On the other hand, bandsappeared due to stretching (3000–2700 cm−1) and deformation(1400–1300 cm−1) modes of –CH3 and CO− stretching ofmethoxy groups (components at 1120 and 1060 cm−1 compo-nents, superimposed to the main band due to lattice modes ofTiO2). The overall pattern indicated that a significant part of thesurface –OH groups had been replaced by –OCH3 methoxygroups. Finally, water vapor (18 Torr, saturation pressure atroom temperature) was admitted on the methylated sample, inorder to form on the surface multilayers of physisorbed watercorresponding to the interface of the material when suspendedin aqueous media. The sample was thus kept in contact withwater vapor for 30 min and then outgassed at room temperature.This treatment was repeated three times, until no more changeswere observed in the OR spectrum, indicating that anequilibrium condition in the interactions between the methy-lated surfaces and H2O molecules was reached. The corre-sponding IR pattern is shown as spectrum c: the bands related tomethoxy groups appear decreased of ca. 50% in intensity withrespect to the sample immediately after methylation, but theirpresence indicates that approximately half of the organic groupssubstituting surface hydroxyls is still present and apparentlyirreversible upon contact with water. Of course, in addition thebands related to water irreversibly adsorbed on the surfaceunder the repeated contact with water appeared again.

Endocytosis and subcellular localization of TiO2 particles

The endocytosis of the different particle preparations wasinvestigated by TEM and EDX analysis. Representativemicrographs are shown in Fig. 2. The TEM analysis demon-strated a rapid internalization of the TiO2 particles in the A549cells. Confirmation of the identity of the electron dense particlesas TiO2 was achieved by EDX analysis and elemental mapping.The particles, which were identified both outside and inside theepithelial cells by their specific ultra-structure at high micro-scopic magnification, showed the characteristic X-ray emissionline of the element Ti that refers to an unequivocal identificationof the particles. Overall, the TiO2 particles were found to beassociated as aggregates and/or their agglomerates and appearedto enter the cells by phagocytosis (see Fig. 2). For the UF-TiO2

particles, occasionally small aggregates containing 3–5 ultra-fine primary TiO2 particles (size 20–30 nm) were observed tobe endocytosed by clathrin-coated vesicles.

Neither the different size of the particles nor their surfaceproperties had an influence upon the intracellular particledistribution. Aggregates of both the fine TiO2 (F-TiO2) and theUF-TiO2 particles were predominantly incorporated in membrane-bound vacuoles. Particles aggregates were also found associatedwith both loosely and highly packed lamellar bodies (see Fig. 2).Multivesiculated bodies, which are regarded as acid phosphatasecontaining lysosomes mixed with residual membranous materialand other cell debris, were also found to be enriched with TiO2

particles. Membrane-enclosed particle aggregates of different size

Fig. 2. Transmission electron micrographs of endocytosis of ultrafine titanium dioxide particles by A549 cells. Representative pictures are shown for A549 cells upontreatment with 16 μg/cm2 UF-TiO2 for 4 h. Upper left panel: lamellipodia engulfing TiO2 particle aggregates. Lower left panel: localization of particles inside alamellar body. Right panels: membrane bound aggregates of particles near mitochondria (upper right panel) and nuclear membrane (lower right panel).

145S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

were often observed next to the nucleus but never inside thenucleus. Golgi apparatus, rough endoplasmic reticulum (ER) andmitochondria were all found to be free of TiO2 particles. TiO2

particles that were located in the space between the cells were neverobserved in association with moving through the tight junctionsinto the cells. The overall electronmicroscopical analyses indicatedthat all of the above effects occurred in the absence of signs ofnecrosis or apoptosis (not shown).

Fig. 3. Comparison of aggregate size distribution for the four different preparations ofof 16 μg/cm2. Aggregate diameters (nm) in A549 cells were measured using transm

The particle clustering limited the determination of the exactparticle number that entered the cells. Therefore, we used theproportion of particle aggregates according to their maximalmean diameter as parameter to achieve some indications ofparticle uptake in relation to particle size and methylation. Theeffects of particle size as well as of methylation of TiO2 on thediameter distribution of the endocytosed aggregates is shown inFig. 3. Approximately half of the F-TiO2 particle aggregates

TiO2 following uptake in A549 cells. Each sample was incubated for 4 h at a doseission electron microscopy with software assisted analysis.

146 S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

inside the A549 cells were found to be in the size range of2000–500 nm. In contrast, for the UF-TiO2, the predominantfraction of aggregates was lower in size range (i.e. b500 nm).For F-TiO2, the methylation procedure was associated with areduction in the diameters of endocytosed aggregates. However,for the UF-TiO2 no such clear methylation-associated effect wasobserved (see Fig. 3).

Free radical release upon contact with cells

Electron paramagnetic resonance analysis was performed todetermine the radical generation properties of the four differentTiO2 samples in the cell cultures. The results of these experimentsare shown in Fig. 4.With regard toDMPO (panelA), bothUF-TiO2

and methylated UF-TiO2 (MUF-TiO2) showed significantly higherROS generation, in contrast to F-TiO2 andmethylated F-TiO2 (MF-TiO2), for which the EPR-signal intensities were not different from

Fig. 4. Reactive oxygen species formation from A549 cells upon treatment with the dsonicated and then immediately added to the A549 cells for a total of 2 or 4 h in HBSSadded. At the end of each incubation, the medium was collected and analyzed with ES(ANOVAwith Dunnett post hoc comparison). Panel A: results with DMPO. Panel BMF=methylated fine; UF=ultrafine; MUF=methylated ultrafine. The inserts show

controls. Similar observations were observed using TEMPOL (Fig.4, panel B): both UF-TiO2 and MUF-TiO2 showed a significantdecrement of the signal after 2 h as well as after 4 h. Although somereduction was seen with F-TiO2 and MF-TiO2, this did not reach astatistical significance. In the absence of A549 cells, EPRmeasurements with the different TiO2 samples with both spintraps showed no significant differences from the control. Notably,the cell-based experiments were found to yield lower signals (forboth DMPO and TEMPOL) compared to those in the absence ofcells. This can be explained by various reasons. Firstly, the cell freesystem is not subject to potential ROS scavenging effects or bindingof the spin-trapping agents with cellular structures and products(e.g. cell membrane, secreted proteins). Secondly, in the cell-basedmeasurements part of the spin-trapping agents will diffuse into thecells and thereby escape detection from analysis in the cellsupernatants collected. These results are also reflected bycomparison of the control incubations (i.e. in the absence of

ifferent dust preparations. The different TiO2 samples were suspended in HBSS,at 400 μg/cm2. One hour after the start of the treatment, DMPO or TEMPOLwasR. Data represent the mean and standard errors of three independent experiments: results with TEMPOL. *Pb0.05 and **Pb0.01 versus control (=C). F=fine;the signals as observed after 4 h in the absence of A549 cells.

147S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

particles) in presence versus absence of the cells, which showed asignal intensity reduction of 40% and 50% for DMPO andTEMPOL, respectively. Taken together, our findings suggest thatradicals were generated by the cells but not by the particles.However, it is important to notice that our ESR settings werespecifically optimized for measurements under cell cultures, andthat these setting are far form those usually employed to detectparticles derived radicals (e.g. Fenoglio et al., 2001).

Interleukin-8 mRNA expression and release from A549 cells

The effects of the different TiO2 sample preparations on theexpression and release of interleukin-8 from A549 cells areshown in Fig. 5. Both UF-TiO2 and MUF-TiO2 caused asignificant increase in interleukin-8 release from the A549 cellsfollowing 24 h treatment. This effect was not observed with F-TiO2 and MF-TiO2. Methylation of the TiO2 samples did notaffect the production of interleukin-8. When compared on equalmass basis, a significant effect of DQ12 was observed at a fivefold lower concentration than for both TiO2 sample prepara-tions. In line with the observations on interleukin-8 release, bothultrafine TiO2 samples also caused enhanced mRNA expressionof IL-8 in the A549 cells, in contrast to both fine TiO2 samples.Enhanced IL-8 mRNA expression was also observed aftertreatment with DQ12 and TNFα. Endotoxin determinations ofthe different dusts suspended in endotoxin free water showedvalues of 0.006, 0.007, 0.005 and 0.009 EU/mg dust,respectively, for F-TiO2, UF-TiO2, MF-TiO2 and MUF-TiO2.

Fig. 5. IL-8 mRNA expression and protein release from A549 cells upontreatment with the different preparations of titanium dioxide. Panel A:representative gel showing interleukin-8 mRNA expression in A549 cellsupon 4 h treatment with the different TiO2 samples (80 μg/cm2). TNF-alpha(10 ng/ml) or DQ12 quartz (80 μg/cm2) were used as positive controls, andGAPDH was used as housekeeping gene. Panel B: release of IL-8 after 24 htreatment of cells A549 with the different titanium dioxide particles. F=fine;MF=methylated fine; UF=ultrafine; MUF=methylated ultrafine. DQ=Quartzparticles (DQ12) were used as positive control. Data are mean and standarderrors of n=3 independent experiments (⁎Pb0.05 vs. untreated cells).

Thus endotoxin contamination was negligible for all samplesused (i.e. b0.003 EU/ml during cell treatment) and hence didnot affect the outcome of our cell experiments. This was alsoconfirmed in independent experiments where LPS failed toelicit increased IL-8 release from the A549 cells at concentra-tions below 0.1 μg/ml (data not shown).

Discussion

Over the past years, concern has been raised about thepotential harmful effects of nanoparticles (e.g. Colvin, 2003;Nel et al., 2006). Although particle size and more closely thetotal particle surface dose have been shown to relate both totheir in vitro and in vivo toxicity (Tran et al., 2000, Oberdorsteret al., 2005a), detailed molecular mechanisms explaining theseeffects are still incompletely understood. Apart from surfacearea, surface chemistry and particle aggregation state andgeneration of oxidative stress are considered to be importantdrivers of nanoparticle toxicity (Donaldson et al., 2002).Elucidation of the responsible mechanism(s) is of majorimportance, since this allows for potential identification ofunifying factors which may be applied in initial hazardscreening strategies of novel nanomaterials (Oberdorster et al.,2005b; Unfried et al., 2007).

In this study, we evaluated the ability of TiO2 to induce theexpression of IL-8 in A549 human lung epithelial cells, inrelation to their surface exposure dose, their surface composi-tion, their aggregation/agglomeration state and their ability toelicit ROS generation. To the best of our knowledge, our studyis the first to show that ultrafine TiO2 particles in contrast tofine TiO2 can trigger a pro-inflammatory response (1) thatappears to be driven by their large surface area per se and (2)that takes place despite the fact that their uptake andtranslocation within lung epithelial cells as stable aggregates.Our results indicate that these effects result from oxidants thatare generated during particle–cell interactions through yet to beelucidated mechanism(s).

The effect of surface area per se was determined bycomparing the effects of two samples of TiO2 with differentparticle size distribution and hence surface area. A methylationprocedure, used to change the hydrophobicity of the particlesurface, left half of the surface methylated in water. Althoughthe treatments of A549 cells were performed in culture medium,one can assume that the stability of approximately 50% of themethylation of the TiO2 surface as monitored in water by IR ismaintained in the buffered medium. In biological systems, thedegree of hydrophobicity of a surface is considered to affectcell-surface adhesion, protein denaturation at the interface andthe selective adsorption of components from the liquid phase(Van Oss, 1994). Variations of the surface can result in differenttranslocation routes in various biological compartments,different coatings of the surface by endogenous materials anddifferences in the interfacing of the solid with cells (Fubini etal., 1990). Specifically with regard to TiO2, the role of surfacecoating in its toxicity has been addressed in several in vivoinvestigations with contrasting outcomes (Oberdörster, 2001;Höhr et al., 2002; Rehn et al., 2003; Warheit et al., 2002, 2007).

148 S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

In the present in vitro study, we observed a rapid uptake inA549 cells for each of the TiO2 samples, predominantly asmembrane bound aggregates and large intracellular aggregatesin vesicles, vacuoles, multivesiculated bodies, as well as inlamellar bodies which are the main compartment of surfactantproduction. Previously, Stearns et al. (2001) investigated theuptake of ultrafine TiO2, but not fine TiO2, in A549 cells. Theparticles used in their study had a primary particle size of about50 nm, which was roughly similar to the ultrafine sample usedin our study. Although various experimental settings differed inthe two studies (e.g. A549 culturing, particle sonication,treatment concentrations), the overall findings were rathersimilar. In line with Stearns et al. (2001), particles appeared tobe phagocytosed in agglomerates or clusters and no singleultrafine particle was observed. Large aggregates of particleswere also observed in association with the plasma membrane(within filopodia or lamellipodia) irrespective of their surfacemethylation. We also found TiO2 particles (both fine andultrafine) to be predominantly associated as large aggregatesand/or their agglomerates, and these appeared to enter the cellsby phagocytosis. These observations were well in line withstudies carried out with fine and ultrafine TiO2 in trachealexplants (Churg et al., 1998). For the UF-TiO2, we occasionallyobserved endocytosis of small aggregates (b30 nm) by clathrin-coated vesicles, indicative of pinocytosis. Major organellesincluding the nucleus, mitochondria, Golgi apparatus and roughER were all found to be devoid of TiO2. As such, our presentdata are in contrast with observations by others, who foundambient ultrafine particles interacting within mitochondrialmembranes (Li et al., 2003), or engineered SiO2 nanoparticleslocated inside the nucleus (Chen and von Mikecz, 2005). Ourobservations are also in contrast to recent investigations withspark-generated ultrafine titanium particles, where singleultrafine particles were noted to locate inside cells as well astheir nuclei (Geiser et al., 2005).

Inspection of the aggregate size distribution showed atendency for the UF-TiO2 particles to have aggregates in alower size range (b500 nm), when compared to the F-TiO2

particles. Interestingly, for the fine particles, the methylation wasassociated with an apparent reduction in the size of particleaggregated within the cells. This finding also provides somesupport that the surface methylation remained indeed effectiveunder our experimental conditions. Although the observed effecton aggregation may be due to the variation in hydrophobicityupon methylation, it should be noted that this effect was far morepronounced with the fine than the ultrafine particles.

The ability of the different particle preparations to generatereactive oxygen species during treatment of the A549 cells is ofparamount importance since oxidative stress is considered torepresent a hallmark of the toxic and inflammogenic effects ofultrafine particles (Donaldson et al., 2002). Two spin-trappingagents were used in the present study. DMPO, which is knownto predominantly react with several free radicals by givingspectra having different hyperfine structure, has been used byseveral investigators to determine generation of ROS in cellsystems (Li et al., 2000; Zhang et al., 2001). TEMPOL is a lowmolecular weight, membrane permeable stable free radical with

a well-defined ESR spectrum. The decay rate of the TEMPOLsignal intensity has been considered to reflect the production ofsuperoxide and has accordingly been used in cell-culture studies(Nagakawa et al., 2000). Since the two agents show contrastingeffects upon their reaction with ROS (i.e. an increase orreduction of the signal intensity for DMPO or TEMPOL,respectively), false interpretation of our overall findings due topossible artefacts (e.g. intracellular degradation, chemicaldegradation of the spin trap) is avoided. Interestingly, withboth DMPO and TEMPOL we observed that the two differentpreparations of UF-TiO2 elicited significant increases in ROSgeneration during treatment of the A549 cells, in contrast totheir fine counterparts. Furthermore, the effects observed withthe UF-TiO2 were found to be irrespective of the methylationstate, indicating that the higher specific surface of the ultrafineversus fine particles per se, rather than hydrophobicity, drivesoxidative stress in the lung epithelial cells.

Our findings in the cell cultures contrasted to the EPR resultsas obtained in the absence of cells (i.e. control incubations).With DMPO, all samples showed a slight increase in theDMPO-OH adduct levels above controls, although these valuesdid not reach statistical significance. The observed effect maybe due to formation of artefacts and induction of free radicalgeneration during particle sonication or because of metalimpurities on the samples. Importantly, however, for DMPO nodifferences were observed between the different samples, andwith TEMPOL all samples, including the controls, showed thesame signal intensities. Taken together, these data indicate thatthe observed biological responses in our current study resultfrom ROS as generated from the interactions of TiO2 with theA549 cells, rather than as derived from the particle surfaceitself. However, it should be emphasized that the ESR settingsapplied in our current study were specifically optimized formeasurements under cell cultures, and not with the aim todetermine ROS generation from the particles directly. DMPOand TEMPOL concentrations were in fact chosen on the basis oftheir absence of cytotoxicity and stability on biological systems(Khan et al., 2003; May et al., 2005), and incubations wereperformed in particle suspensions placed in plastic dishes in acell culture incubator (i.e. in the dark). Conversely in theexperimental setting adopted, different from what usually isemployed in a cell free system, some initial free radicalgeneration from the particles may have escaped detection. Theradical generating properties of different TiO2 samples in cellfree environment and its association with current findings are akey topic of our ongoing investigations. The ability of TiO2 togenerate ROS in an acellular environment has been indicatedfrom various independent studies, i.e. upon co-treatment withsimulated sunlight/UV, with comparatively more pronouncedeffects for anatase than rutile (Wamer et al., 1997; Dunford etal., 1997; Hirakawa et al., 2004). In the absence of lightirradiation, such effects are less conclusive but tend to showstronger effects for UF-TiO2 compared to F-TiO2 (Donaldson etal., 1996; Dick et al., 2003). The ultrafine sample in our studycontained 20% rutile, whereas the fine sample was pure anatase.Contrasts in cellular responses were recently reported forequivalent surface area samples of anatase and rutile, with pure

Fig. 6. Release of interleukin-8 from A549 cells as function of particle surfacedose. Interleukin-8 expression is shown as a percentage above control asfunction of the dose expressed as particle surface area (BET, see Table 1) perA549 monolayer surface area (cm2/cm2). The graph indicates that whenconsidered at equal surface area, all TiO2 samples elicited similar potency incausing IL-8 release, whereas DQ12 has intrinsically higher IL-8 inducingpotency.

149S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

rutile samples having markedly lower effects (Sayes et al.,2006; Warheit et al., 2007). These results emphasize that oneshould be cautious about extrapolating current observationswith our specific UF-TiO2 to all types of ultrafine TiO2.

A major observation in the present study is that both theultrafine and the fine TiO2 samples appeared as aggregates/agglomerates, yet they showed different ROS responses in cellculture. Although we cannot rule out the possibility thatoccasional nanoparticles may have escaped detection in ourTEM analysis, this would be an unlikely explanation for thecontrasting ROS generation. Thus, the ability of the UF-TiO2 toactivate cells for ROS production appeared to be an aggregate/agglomerate-triggered effect. In line with our observations, UF-TiO2 (from the same source as used in our study) was found, inaggregated form, to trigger ROS generation from brain microglia(Long et al., 2006). Although membrane bound NAPDHoxidases and/or mitochondria (Li et al., 2003; Knaapen et al.,2004; Long et al., 2006) represent major candidates for sucheffects, we could not verify this, as inconclusive results wereobtained after co-treatment of the cells with diphenyleneiodi-nium (DPI) (data not shown). Hence, further investigations arerequired to determine the mechanisms whereby interactionsbetween (aggregates of) UF-TiO2 and lung epithelial cells leadto ROS generation. This will also be of importance in relation totoxicity screening as proposed for nanoparticles in general(Oberdorster et al., 2005b; Nel et al., 2006).

The effects of the different particle preparations on IL-8expression in A549 cells were determined in this study forseveral reasons. Human lung epithelial cells are an importantsource of IL-8 in the lung, and this chemokine is wellrecognized as an important mediator of pulmonary inflamma-tion in humans (Kunkel et al., 1991; Keatings et al., 1996).A549 cells have been shown to release IL-8 upon exposure tovarious particles (Stringer et al., 1996; Simeonova and Luster,1996; Schins et al., 2000). Most importantly, IL-8 is shown tobe induced in A549 cells via a mechanism involving oxidativestress and activation of the redox-sensitive transcription factornuclear factor kappa-B (NF-κB)(Kunsch and Rosen, 1993;Schins and Donaldson, 2000). Recent investigations, however,have shown that ultrafine carbon black particles elicit IL-8expression and release from human bronchial epithelial cells inan apparent NF-κB-independent manner, involving p38 mito-gen-activated protein kinase (MAPK) activation (Kim et al.,2005). In the present study, we showed that both ultrafinesamples, in contrast to their fine counterparts, caused asignificant release of IL-8 from A549 cells and this wasassociated with their ability to enhance the expression of IL-8mRNA. The observed effects appeared to be independent of themethylation state of the particle surface. Importantly, endotoxinanalysis of the samples showed that this effect was also not dueto contamination of the particles with this well-known activatorof IL-8 in A549 cells (Hansen et al., 1999). Taken together, theeffects on IL-8 release paralleled the effects on ROS generationas observed with the different TiO2 samples. Notably, whencompared on equal mass basis, a significant effect of DQ12 wasobserved at a five-fold lower concentration than for both TiO2

samples, irrespectively of their methylation status. As most of

the biological responses elicited by particles are driven by theirsurface, any comparison between particles with differentspecific surfaces has to be made both per weight and perexposed surface. In Fig. 6, IL-8 release from A549 cells,expressed as a percentage above control is shown as a functionof the dose expressed as exposed particle surface per unitculture dish surface area (cm2/cm2). The graph indicates that,when considered at equal surface area, all four TiO2 sampleselicited a similar potency in causing IL-8 release. Thus, the UF-TiO2 samples used in this study are not intrinsically more activeper unit surface area compared to the fine samples. Since ourfindings are in contrast with the recent observations by Sayes etal. (2006), further investigations are needed to determinewhether our current observations are specific for lung epithelialcells and their major chemokine products, such as IL-8.Importantly, Fig. 6 also shows that, compared to TiO2 samples,the quartz sample DQ12 has an intrinsically higher IL-8inducing potency. The comparatively strong effect of the quartzsample is most likely due to its intrinsically higher surfacereactivity (Fubini and Hubbard, 2003) and hence toxicity, whencompared to relatively inert materials such as TiO2 in animalmodels (e.g. Driscoll et al., 1990; Duffin et al., 2002). Similarly,other nanoparticles including those which are engineered andused specifically because of their “increased” reactivity on thenanoscale may also show a specific reactivity not shown bylarger particles of the same nominal composition.

At this stage, we would also like to emphasize that overall,high particle concentrations were required to observe anysignificant effect in the A549 cells. In view of its implicationsfor risk assessment, these concentrations will extrapolate toestimated lung burdens that would require unrealistically highexposures to this material (Oberdorster and Yu, 1999). Never-theless, present experiments allowed us to identify new insightson the potential mechanisms involved in the toxic effects of

150 S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

ultrafine (TiO2) particles. In this regard, our current in vitroobservations are found to be well in line with previous in vivoinvestigations in rats with the same four TiO2 samples. In bothcases, relatively independent of the methylation status UF-TiO2

showed stronger pulmonary responses after intratrachealinstillation than F-TiO2 (Höhr et al., 2002). Similarly, Rehn etal. (2003) compared the inflammatory effects of native (P25)versus silanized (P805) UF-TiO2, but found only minimal, andnot significantly differing inflammation, 3 days after singleinstillations of 0.15, 0.3, 0.6 and 1.2 mg. More interestingly, ourpresent in vitro observations on the apparent similarities of allfour TiO2 samples in eliciting IL-8 release as a marker ofinflammogenic potency upon considering dose in terms ofapplied surface area, is well in line with several in vivoobservations by other investigators with TiO2. As reviewed inOberdorster et al. (2005a), instillation studies with two samplesof TiO2, both in rat and mice, showed that neutrophilicinflammation was correlated to the total administered particlesurface area, with the UF-TiO2 and the F-TiO2 fitting the samedose-response curve. Taken together with similar in vivoobservations with various particles of different surface area(Tran et al., 2000; Oberdorster et al., 2000; Duffin et al., 2002;Stoeger et al., 2006), our in vitro data provide further supportthat, for relatively low toxicity particles of different sizes, suchas TiO2, the administered total surface area is a better dosimetricthan the administered mass or particle number.

In summary, in the present study we have demonstrated thatUF-TiO2 samples but not their fine counterparts were found toelicit oxidative stress and IL-8 release from A549 cells, (1)irrespective of their methylation and (2) despite the fact thatthese remained highly aggregated in cell culture as well asinside the cells. Our results indicate that UF-TiO2, even asaggregates/agglomerates, has inflammatory properties thatappear to be driven by their specific surface area. Furthermore,our results indicate that these effects are mediated by oxidativestress as elicited by particle–cell interactions, although theresponsible mechanism(s) herein remains to be elucidated.Further research is also warranted to determine whether ourcurrent observations with TiO2 are also applicable to othernanoparticles/ultrafine particles, including for instance dieselexhaust particles that also typically exist as aggregates.

Acknowledgments

This study was supported by a grant from the Germanministry of Environment (BMU) and partly by the GermanResearch Council (DFG-Sonderforschungsbereich 503). Sup-port was also given by the Regione Piemonte, Italy (researchproject “Biocompatible Nanostructured Materials for Biomedi-cal Applications, call 2004, project code: D33). We acknowl-edge Dr. Wolfgang Bischof (Institute of Occupational, Socialand Environmental Medicine, University of Jena, Germany) forthe endotoxin determinations and Chiara Sciolla (Department ofInorganic Physical and Material Chemistry, University ofTorino, Italy) and Dr. Ad Knaapen (Maastricht University,The Netherlands) for their helpful contributions to the initialexperiments related to this work.

References

Baan, R., Straif, K., Grosse, Y., Secretan, B., El Ghissassi, F., Cogliano, V.,2006. WHO International Agency for Research on Cancer MonographWorking Group. Carcinogenicity of carbon black, titanium dioxide, and talc.Lancet Oncol. 7, 295–296.

Borm, P.J.A., Kreyling, W., 2004. Toxicological hazards of inhaled nanopar-ticles. Potential implications for medical application. J. Nanosci. Nanotech-nol. 4, 521–531.

Borm, P.J., Robbins, D., Haubold, S., Kuhlbusch, T., Fissan, H., Donaldson, K.,Schins, R.P., Stone, V., Kreyling, W., Lademann, J., Krutmann, J., Warheit,D., Oberdorster, E., 2006. The potential risks of nanomaterials: a reviewcarried out for ECETOC. Part. Fibre 3–11.

Brown, D.M., Wilson, M.R., MacNee, W., Stone, V., Donaldson, K., 2001. Size-dependent proinflammatory effects of ultrafine polystyrene particles: a rolefor surface area and oxidative stress in the enhanced activity of ultrafines.Toxicol. Appl. Pharmacol. 175, 191–199.

Brown, D.M., Donaldson, K., Borm, P.J., Schins, R.P.F., Denhart, M., Gilmour,P., Jimenez, L.A., Stone, V., 2004. Calcium-mediated activation oftranscription factors and TNF-alpha cytokine gene expression in macro-phages exposed to ultrafine particles. Am. J. Physiol.: Lung Cell. Mol.Physiol. 286, L344–L353.

Chen, M., von Mikecz, A., 2005. Formation of nucleoplasmic proteinaggregates impairs nuclear function in response to SiO2 nanoparticles.Exp. Cell Res. 305, 51–62.

Churg, A., Stevens, B., Wright, J.L., 1998. Comparison of the uptake of fine andultrafine TiO2 in a tracheal explant system. Am. J. Physiol. 274, L81–L86.

Colvin, V.L., 2003. The potential environmental impact of engineerednanomaterials. Nat. Biotechnol. 21, 1166–1170.

Dick, C.A., Brown, D.M., Donaldson, K., Stone, V., 2003. The role of freeradicals in the toxic and inflammatory effects of four different ultrafineparticle types. Inhal. Toxicol. 15, 39–52.

Donaldson, K., Beswick, P.H., Gilmour, P.S., 1996. Free radical activityassociated with the surface of particles: a unifying factor in determiningbiological activity? Toxicol. Lett. 88 (1–3), 293–298.

Donaldson, K., Brown, D., Clouter, A., Duffin, R., MacNee, W., Renwick, L.,Tran, L., Stone, V., 2002. The pulmonary toxicology of ultrafine particles. J.Aerosol Med. 15, 213–220.

Donaldson, K., Tran, L., Jimenez, L.A., Duffin, R., Newby, D.E., Mills, N.,MacNee,W., Stone, V., 2005. Combustion-derived nanoparticles: a review oftheir toxicology following inhalation exposure. Part. Fibre Toxicol. 2–10.

Driscoll, K.E., Lindenschmidt, R.C., Maurer, J.K., Higgins, J.M., Ridder, G.,1990. Pulmonary response to silica or titanium dioxide: inflammatory cells,alveolar macrophage-derived cytokines, and histopathology. Am. J. Respir.Cell Mol. Biol. 2, 381–390.

Duffin, R., Tran, C.L., Clouter, A., Brown, D.M., MacNee, W., Stone, V.,Donaldson, K., 2002. The importance of surface area and specific reactivityin the acute pulmonary inflammatory response to particles. Ann. Occup.Hyg. 46 (Suppl. 1), 242–245.

Dunford, R., Salinaro, A., Cai, L., Serpone, N., Horikoshi, S., Hidaka, H.,Knowland, J., 1997. Chemical oxidation and DNA damage catalysed byinorganic sunscreen ingredients. FEBS Lett. 418, 87–90.

Fenoglio, I., Prandi, L., Tomatis, M., Fubini, B., 2001. Free radical generation inthe toxicity of inhaled mineral particles: the role of iron speciation at thesurface of asbestos and silica. Redox Rep. 6, 235–241.

Ferin, J., Oberdorster, G., Penney, D.P., 1992. Pulmonary retention of ultrafineand fine particles in rats. Am. J. Respir. Cell Mol. Biol. 6, 535–542.

Fiedler, M.A., Wernke-Dollries, K., Stark, J.M., 1998. Inhibition of TNF-alpha-induced NF-kappaB activation and IL-8 release in A549 cells withthe proteasome inhibitor MG-132. Am. J. Respir. Cell Mol. Biol. 19,259–268.

Fubini, B., Hubbard, A., 2003. Reactive oxygen species (ROS) and reactivenitrogen species (RNS) generation by silica in inflammation and fibrosis.Free Radical Biol. Med. 34, 1507–1516.

Fubini, B., Giamello, E., Volante, M., Bolis, V., 1990. Chemical functionalitiesat the silica surface determining its reactivity when inhaled. Formation andreactivity of surface radicals. Toxicol. Ind. Health 6, 571–598.

151S. Singh et al. / Toxicology and Applied Pharmacology 222 (2007) 141–151

Fubini, B., Mollo, L., Giamello, E., 1995. Free radical generation at the solid/liquid interface in iron containing minerals. Free Radical Res. 23,593–614.

Geiser, M., Rothen-Rutishauser, B., Kapp, N., Schurch, S., Kreyling, W.,Schulz, H., Semmler, M., Im Hof, V., Heyder, J., Gehr, P., 2005. Ultrafineparticles cross cellular membranes by nonphagocytic mechanisms in lungsand in cultured cells. Environ. Health Perspect. 113, 1555–1560.

Hansen, L.A., Poulsen, O.M., Wurtz, H., 1999. Endotoxin potency in the A549lung epithelial cell bioassay. J. Immunol. Methods 226, 49–58.

Hirakawa, K., Mori, M., Yoshida, M., Oikawa, S., Kawanishi, S., 2004. Photo-irradiated titanium dioxide catalyzes site specific DNA damage viageneration of hydrogen peroxide. Free Radical Res. 38, 439–447.

Höhr, D., Steinfartz, Y., Schins, R.P.F., Knaapen, A.M.,Martra, G., Fubini, B., Borm,P.J.A., 2002. The surface area rather than the surface coating determines the acuteinflammatory response after instillation of fine and ultrafine TiO2 in the rat. Int. J.Hyg. Environ. Health 205, 239–244.

Keatings, V.M., Collins, P.D., Scott, D.M., Barnes, P.J., 1996. Differences ininterleukin-8 and tumor necrosis factor-alpha in induced sputum frompatients with chronic obstructive pulmonary disease or asthma. Am. J.Respir. Crit. Care Med. 153, 530–534.

Khan, N., Wilmot, C.M., Rosen, G.N., Demidenko, E., Sun, J., Joseph, J.,O'Hara, J., Kalyanaraman, B., Swartz, H.M., 2003. Spin traps: in vitrotoxicity and stability of radical adducts. Free Radical Biol. Med. 34,1473–1481.

Kim, Y.M., Reed, W., Lenz, A.G., Jaspers, I., Silbajoris, R., Nick, H.S., Samet,J.M., 2005. Ultrafine carbon black particles induce interleukin-8 genetranscription and p38 MAPK activation in normal human bronchialepithelial cells. Am. J. Physiol. 288, L432–L441.

Knaapen, A.M., Borm, P.J.A., Albrecht, C., Schins, R.P.F., 2004. Inhaledparticles and lung cancer, part A. Mechanisms. Int. J. Cancer 109, 799–809.

Kunkel, S.L., Standiford, T., Kasahara, K., Strieter, R.M., 1991. Interleukin-8(IL-8): the major neutrophil chemotactic factor in the lung. Exp. Lung Res.17, 17–23.

Kunsch, C., Rosen, C.A., 1993. NF-kappa B subunit-specific regulation of theinterleukin-8 promoter. Mol. Cell. Biol. 13, 6137–6146.

Li, X.Y., Brown, D., Smith, S., MacNee, W., Donaldson, K., 1999. Short-terminflammatory responses following intratracheal instillation of fine andultrafine carbon black in rats. Inhal. Toxicol. 11, 709–731.

Li, H., Hu, J., Xin, W., Zhao, B., 2000. Production and interaction of oxygen annitrogen free radicals in PMA stimulated macrophages during the respiratoryburst. Redox Rep. 5, 353–358.

Li, N., Sioutas, C., Cho, A., Schmitz, D., Misra, C., Sempf, J., Wang, M.,Oberley, T., Froines, J., Nel, A., 2003. Ultrafine particulate pollutants induceoxidative stress and mitochondrial damage. Environ. Health Perspect. 111,455–460.

Long, T.C., Saleh, N., Tilton, R.D., Lowry, G.B., Veronesi, B., 2006. Titaniumdioxide (P25) produces reactive oxygen species in immortalised brainmicroglia (BV2): implications for nanoparticle neurotoxicity. Environ. Sci.Technol. 40, 4346–4352.

May, J.M., Qu, Z.C., Juliao, S., cobb, C.E., 2005. Ascobic acid decreasesoxidant stress in endothelial cells caused by the nitroxide TEMPOL. FreeRadical Res. 39, 195–202.

Morterra, C., Ghiotti, G., Garrone, E., Fisicaro, E., 1980. Spectroscopic study ofanatase properties. J. Chem. Soc., Faraday Trans. 1, 2102–2113.

Nagakawa, H., Moritake, T., Tsuboi, K., Ikota, N., Ozawa, T., 2000. Induction ofsuperoxide in glioma cell line U87 stimulated with lipopolysaccharide andinterferon-gamma: ESR using a new flow-type quartz cell. FEBS Lett. 471,187–190.

Nel, A.E., Xia, T., Madler, L., Li, N., 2006. Toxic potential of materials at thenanolevel. Science 311, 622–627.

Oberdorster, G., Oberdorster, E., Oberdorster, J., 2005a. Nanotoxicology: anemerging discipline evolving from studies of ultrafine particles. Environ.Health Perspect. 113, 823–839.

Oberdorster, G., Maynard, A., Donaldson, K., Castranova, V., Fitzpatrick, J.,Ausman, K., Carter, J., Karn, B., Kreyling, W., Lai, D., Olin, S., Monteiro-Riviere, N., Warheit, D., Yang, H., 2005b. Principles for characterising thepotential human health effects from exposure to nanomaterials: elements of ascreening strategy. Part. Fibre Toxicol. 2–8.

Oberdörster, G., 2001. Pulmonary effects of inhaled ultrafine particles. Int.Arch. Occup. Environ. Health 74, 1–8.

Oberdorster, G., Yu, C.P., 1999. Lung dosimetry considerations for noninhala-tion studies. Exp. Lung Res. 25, 1–6.

Oberdorster, G., Finkelstein, J.N., Johnston, C., Gelein, R., Cox, C., Baggs, R.,Elder, A.C., 2000. Acute pulmonary effects of ultrafine particles in rats andmice. Res. Rep.-Health Eff. Inst. 96, 5–74.

Peters, A., Wichmann, H.E., Tuch, T., et al., 1997. Respiratory effects areassociated with the number of ultrafine particles. Am. J. Respir. Crit. CareMed. 155, 1376–1383.

Rahman, Q., Lohani, M., Dopp, E., Pemsel, H., Jonas, L., Weiss, D.G.,Schiffmann, D., 2002. Evidence that ultrafine titanium dioxide inducedmicronuclei and apoptosis in Syrian hamster embryo cells. Environ. HealthPerspect. 110, 797–800.

Rehn, B., Seiler, F., Rehn, S., Bruch, J., Maier, M., 2003. Investigations on theinflammatory and genotoxic lung effects of two types of titanium dioxide:untreated and surface treated. Toxicol. Appl. Pharmacol. 189, 84–95.

Sayes, C.M., Whai, R., Kurian, P.A., Liu, Y., West, J.L., Ausman, K.D., Warheit,D.B., Colvin, V.L., 2006. Correlating nanoscale titanium structure withtoxicity: a cytotoxicity and inflammatory response study with human dermalfibroblasts and human lung epithelial cells. Toxicol. Sci. 92, 174–185.

Schins, R.P.F., Donaldson, K., 2000. Nuclear factor kappa-B activation byparticles and fibres. Inhal. Toxicol. 12, 317–326.

Schins, R.P.F., McAlinden, A., MacNee, W., Jimenez, L.A., Ross, J.A., Guy, K.,Faux, S., Donaldson, K., 2000. Persistent depletion of I kappa B alpha(IkBa) and interleukin-8 expression in human pulmonary epithelial cellsexposed to quartz. Toxicol. Appl. Pharmacol. 167, 107–117.

Simeonova, P.P., Luster, M.I., 1996. Asbestos induction of nuclear transcriptionfactors and interleukin 8 gene regulation. Am. J. Respir. Cell Mol. Biol. 15,787–795.

Stearns, R.C., Paulauskis, J.D., Godleski, J.J., 2001. Endocytosis of ultrafineparticles by A549 cells. Am. J. Respir. Cell Mol. Biol. 24, 108–115.

Stoeger, T., Reinhard, C., Takenaka, S., Schroeppel, A., Karg, E., Ritter, B.,Heyder, J., Schulz, H., 2006. Instillation of six different ultrafine carbonparticles indicates a surface area threshold dose for acute lung inflammationin mice. Environ. Health Perspect. 114, 328–333.

Stringer, B., Imrich, A., Kobzik, L., 1996. Lung epithelial cells (A549)interaction with unopsonized environmental particulates: quantitation ofparticle-specific binding and IL-8 production. Exp. Lung Res. 22, 295–508.

Tran, C.L., Buchanan, D., Cullen, R.T., Searl, A., Jones, A.D., Donaldson, K.,2000. Inhalation of poorly soluble particles. II. Influence of particle surfacearea on inflammation and clearance. Inhal. Toxicol. 12, 1113–1126.

Unfried, K., Albrecht, C., Klotz, L.O., von Mikecz, A., Grether-Beck, S.,Schins, R.P.F., 2007. Cellular responses to nanoparticles: target structuresand mechanisms. Nanotoxicology 1, 52–71.

Upadhyay, D., Panduri, V., Ghio, A., Kamp, D.W., 2003. Particulate matterinduces alveolar epithelial cell DNA damage and apoptosis: role of freeradicals and mitochondria. Am. J. Respir. Cell Mol. Biol. 92, 180–187.

Utell, M.J., Frampton, M.W., 2000. Acute health effects of ambient air pollution:the ultrafine hypothesis. J. Aerosol Med. 13, 355–359.

Van Oss, C.J., 1994. Interfacial Forces in Aqueous Media. Marcel Dekker Inc.,New York, p. 440.

Wamer, W.G., Yin, J.J., Wei, R.R., 1997. Oxidative damage to nucleic acidsphotosensitized by titanium dioxide. Free Radical Biol. Med. 23, 851–858.

Warheit, D.B., Reed, K.L., Webb, T.R., 2002. Utilization of pulmonary bridgingstudies: lung toxicity studies with triethoxyoctylsilane-coated TiO2 particles.Ann. Occup. Hyg. 46, 207–209.

Warheit, D.B., Webb, T.R., Reed, K.L., Frerichs, S., Sayes, C.M., 2007.Pulmonary toxicity studies in rats with three forms of ultrafine-TiO2

particles: differential responses related to surface properties. Toxicology230, 90–104.

Xia, T., Kovochich, M., Brant, J., Hotze, M., Sempf, J., Oberley, T., Sioutas, C.,Yeh, J.I., Wiesner, M.R., Nel, A.E., 2006. Comparison of the abilities ofambient andmanufactured nanoparticles to induce cellular toxicity accordingto an oxidative stress paradigm. Nano Lett. 6, 1794–1807.

Zhang, Z., Leonard, S.S., Wang, S., Vallyathan, V., Castranova, V., Shi, X.,2001. Role of reactive oxygen species and MAPKs in vanadate-induced G(2)/M phase arrest Mol. Cell Biochem. 222, 77–78.