Acta Biomaterialia 3 (2007) 351–358

www.elsevier.com/locate/actabiomat

Cytotoxic effects of aggregated nanomaterials q

Karla Soto *, K.M. Garza, L.E. Murr

The University of Texas at El Paso, Material Science and Engineering, 500 W University, El Paso, TX 79986, USA

Received 2 May 2006; received in revised form 14 November 2006; accepted 20 November 2006

Abstract

This study deals with cytotoxicity assays performed on an array of commercially manufactured inorganic nanoparticulate materials,including Ag, TiO2, Fe2O3, Al2O3, ZrO2, Si3N4, naturally occurring mineral chrysotile asbestos and carbonaceous nanoparticulate mate-rials such as multiwall carbon nanotube aggregates and black carbon aggregates. The nanomaterials were characterized by TEM, as theprimary particles, aggregates or long fiber dimensions ranged from 2 nm to 20 lm. Cytotoxicological assays of these nanomaterials wereperformed utilizing a murine alveolar macrophage cell line and human macrophage and epithelial lung cell lines as comparators. Thenanoparticulate materials exhibited varying degrees of cytoxicity for all cell lines and the general trends were similar for both the murineand human macrophage cell lines. These findings suggest that representative cytotoxic responses for humans might be obtained bynanoparticulate exposures to simple murine macrophage cell line assays. Moreover, these results illustrate the utility in performing rapidin vitro assays for cytotoxicity assessments of nanoparticulate materials as a general inquiry of potential respiratory health risks inhumans.� 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Cytotoxicity assessment; Nanoparticulates; TEM analysis

1. Introduction

There is now compelling evidence that ultrafine ornanoparticulate matter (with mean or geometric diameters<100 nm) is associated with an increased prevalence ofrespiratory and cardiovascular disease and mortality[1–10]. This includes TiO2, black carbon (BC) and silica[2,11], as well as a wide range of natural mineral nanopar-ticles [12]. Recent cytotoxicological assays of a wide rangeof manufactured nanoparticulate materials, utilizing amurine lung macrophage cell line, have illustrated varyingdegrees of cytotoxicity, with nanoparticulate silver, chryso-tile asbestos, multiwall carbon nanotubes and BC exhibit-ing particularly acute cytotoxicity [13,14]. Similarly,recent in vivo studies in rats have shown lung lining inflam-

1742-7061/$ - see front matter � 2007 Acta Materialia Inc. Published by Else

doi:10.1016/j.actbio.2006.11.004

q Research presented at the TMS 2006 Biological Materials ScienceSymposium.

* Corresponding author. Tel.: +1 915 253 6880.E-mail address: karlafs@utep.edu (K. Soto).

mation, dermal inflammation and even death in responseto such materials [15–17]. While the toxicological mecha-nisms of in vitro and in vivo responses are poorly under-stood [2,18], there seems to be mounting evidence thatnanoparticles in particular exert their toxic effects throughthe formation of reactive oxygen species (ROS) whichcause oxidative stress [18–20].

Although chrysotile asbestos has been demonstrated tobe morphologically identical to many forms of multiwallcarbon nanotubes [21], and its short-term cytotoxicresponse for murine macrophage exposure has been dem-onstrated to be identical to multiwall carbon nanotubes[13,14], there is no long-term in vivo evidence that multi-wall carbon nanotubes would pose the same health risksas asbestos. Nonetheless, short-term cytotoxic responsesshould be regarded as a first alert. Indeed, since the incep-tion of the US National Nanotechnology Initiative in 2000,cautions of nanoparticulate risks in particular have per-sisted and the implementation of nanotechnology innova-tions seem linked to biological issues, especially human

vier Ltd. All rights reserved.

352 K. Soto et al. / Acta Biomaterialia 3 (2007) 351–358

health effects. It would seem unrealistic to repeat the fail-ures of the asbestos industry, which largely ignored theproduct dangers for more than 2000 years.

Although simple, short-term cytotoxicity assessmentshave been utilized to evaluate a wide range of nanopartic-ulate materials [13,14], the implications for humans maynot be convincing unless a strong correlation is establishedbetween animal cell assays and human cell assays. In thisstudy we compare cytotoxicity assays for a range of manu-factured nanoparticulate materials utilizing both a murinelung macrophage cell line and a human lung macrophagecell line. In addition, we compare the human macrophagecell line assay results with a series of human lung epithelialcell line assays to provide a more comprehensive, short-term lung function cytotoxicity assessment.

2. Materials and methods

2.1. Characterization of nanoparticulate materials

It has been evident that ultrafine and particulate materi-als (mean diameter <100 nm) in the atmosphere are toxicand pose considerable health risks, such as asthma compli-cations, chronic bronchitis and respiratory tract infections[18,27,28]. The assessment of health effects, especially thepulmonary toxicity of particulates, is often a complex issue,in particular for nanoparticulate materials. Although many,like Lam et al. [15,29] and Warheit et al. [16], have demon-strated single-wall carbon nanotubes (SWCNTs) to be toxic,there have been no detailed microscopic examinations todetermine the particle morphologies or aggregation. Theshape and size of micron-sized particulates is a complicatingissue in assessing pulmonary toxicity and especially their air-way deposition. Fibers, fiber bundles or aggregates, or othernanoparticulate composites have a variety of airstreamresponse and deposition behaviors. Correspondingly, bio-logical assays to evaluate the function of alveolar macro-phages upon exposure to nanoparticulate materials mustinclude their characterization at the nano-scale.

A wide spectrum of commercially manufactured nanop-articulate materials and carbonaceous nanoparticulateswere examined as an extension of previous work [13]. Table

Table 1Description of manufactured nanoparticulate materials

Nanoparticulate material Primary particle size range (nm)

Chrysotile asbestos 15–40

Black carbon (BC) 2–50Multi-wall carbon nanotube-R 10–30Multi-wall carbon nanotube-N 5–30(Ag-1) 3–100Al2O3 4–115Fe2O3 5–140ZrO2 7–120TiO2-anatase 5–40TiO2-rutile 2–60

1 provides a very general description of the nanomaterialsexamined in this study. In addition, nanoparticulate mate-rials were characterized by transmission electron micros-copy, as described in detail previously [13].

2.2. Viability assays

Cytotoxicity assessments of the manufactured nanopar-ticulates in Table 1 were performed using a murine alveolarmacrophage cell line (RAW 264.7) (courtesy of KennethS.K. Tung at the University of Virginia Health ScienceCenter), which was used as a standard against a humanalveolar macrophage cell line (THB-1) (The AmericanType Culture Collection ATTC, Manassas, VA) as wellas a human epithelial cell line (A549) (ATCC). Viabilityassessments and the culturing of the murine alveolar mac-rophages are described in detail elsewhere [14].

All the nanoparticulate materials were suspended in astock solution at 5 lg ml�1 in dimethyl sulfoxide (DMSO),a solvent which assures suspension of even hydrophobicsubstances. Pourahmad and O’Brien [26] have demon-strated that DMSO is an effective antioxidant and is capa-ble of inhibiting cellular death at concentrations rangingfrom 140 to 280 mM. In this investigation the DMSO con-centration ranged from 0.0344 mM to 35.25 nM, or theconcentration in which DMSO did not function as a scav-enger of ROS. The effect of DMSO was determined onbackground production of ROS and hydrogen peroxideinduction of ROS. In both cases DMSO concentrationsup to 35.25 nM did not inhibit ROS formation [13]. Themurine and human macrophages, as well as the human epi-thelial cells, were cultured in a 96-well flat-bottom plate(50,000 cells well�1) starting with a concentration of10 lg ml�1 followed by 11 doubling dilutions. Controlswere incubated with equivalent dilutions of vehicle(DMSO) and with neither vehicle nor compound. Thehuman alveolar macrophage cell line THB-1 was culturedin RPMI 1640, supplemented with 10% FCS, 5 · 105 M2-Me penicillin/streptomycin, 2 mM L-glutamine adjustedto contain 1.5 g l�1 sodium bicarbonate, 4.5 g l�1 glucose,10 mM HEPES and 1.0 mM sodium pyruvate, and supple-mented with 0.05 mM 2-mercaptoethanol. The human

Aggregate size range Specific surface area BET (m2/g)

Fiber bundles 1.30.5 lm–15 lm0.1 lm–1 lm 2390.1 lm–3 lm 160.1 lm–3 lm 21825 nm–1 lm 150.5 lm–1 lm 540.5 lm–0.9 lm 390.5 lm–1 lm 881 lm–2 lm 550.5 lm–1.5 lm 125

K. Soto et al. / Acta Biomaterialia 3 (2007) 351–358 353

epithelial cell line (A549) was cultured in Kaighn’s modifi-cation of HAM’s F-12 medium (F12K) with 2 mM L-gluta-mine containing 1.5 g l�1 sodium bicarbonate, 10% FCS, 2-Me penicillin/streptomycin formulated for use with a 5%CO2 at 37 �C. After 48 h of incubation, 20 ll of3-(4-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium (MTT)(Sigma–Aldrich Co., St Louis, MO) was added and thecells were incubated for an additional 4 h; afterwards180 ll of supernatant was removed and 50 ll of lysis buffer,containing 10 N HCl in isopropanol and Triton-X, wasadded. After several minutes, the MTT crystals formedand the content of dissolved MTT crystals were measuredwith a Molecular Devices VersaMax Tuneable microplatereader set at 570 nm. Cell viability assessment or mitochon-drial activity of living cells in all the cultures (murine andhuman) were made by measuring the relative absorbance,or optical density (OD), for mitochondrial dehydroge-nase-transformed formazan (or color product). The exper-iments were replicated three times and the data weregraphically presented as mean SEM. The data obtainedfrom the MTT results (cell viability) were also representedin EC50 graphs showing the nanoparticulate concentrationat which 50% cell death was induced.

2.3. Statistical analysis

One-way analysis of variance and Tukey tests were usedto evaluate the efficacy of the MTT test for the assessmentof cell viability on the three different cell lines and compar-ison among compounds used in this study with the negative

Fig. 1. (a) TEM bright-field image for BC: the Vulcan XC-72 SAED patterndioxide, illustrating aggregated spherule. The SAED pattern inset indicates cunanotube aggregates (MWCNT-R) and (MWCNT-N). The SAED pattern ins

control or media. Data were analyzed using Graph PadPrism 3.0 and p < 0.05 was considered statistically signifi-cant. The Pearson correlation for the negative controland the tested materials was p < 0.001 for all three celllines. The correlation coefficients were r = 0.9069,r = 0.9759 and r = 0.9836 for the murine macrophage(RAW 264.7), human macrophage (THB-1) and humanepithelial (A549) cell lines, respectively.

2.4. Surface area measurements of nanoparticulate material

Surface area measurements of the raw nanoparticulatematerials were conducted using the BET (Brunauer,Emmett and Teller) method. A Micrometrics ASAP 2020Accelerated Surface Area and Porosity system was utilizedas described generally elsewhere [25]. The results are shownin Table 1.

3. Results and discussion

3.1. TEM characterization of the nanoparticulate materials

Fig. 1a, c and d illustrates common TEM bright-fieldimages of the BC and the two different types of multiwallcarbon nanotubes Rossetter (MWCNT-R) and (MWC-NT-N). Fig. 1b shows an aggregate of ZrO2. The rest ofthe metal oxides exhibit a very similar morphology.Detailed examples of TEM preparation and images of theother nanoparticulates can be found in previous work[13,14]. The BC aggregate in Fig. 1a illustrates agglomerated

inset illustrates graphite diffraction rings. (b) TEM image for zirconiumbic crystallinity. (c, d) TEM bright-field image for manufactured carbonet demonstrates graphite/graphitic structure similar to (a).

354 K. Soto et al. / Acta Biomaterialia 3 (2007) 351–358

carbonaceous spherules ranging in size (diameter) from 2to 50 nm. Fig. 1b illustrates similar agglomerated single-crystal spherules of ZrO2 which range in size from 7 to120 nm. The MWCNT-R aggregate shown in Fig. 1c illus-trates both agglomerated (attached) and aggregated (unat-tached) mixtures of graphitic fragments, various lengths ofmultiwall carbon nanotubes and assorted concentric ful-

Fig. 2. Comparative cytotoxicities of manufactured nanoparticulate materialsconcentrations of 5 lg ml–1.

lerenic particles. Fig. 1d shows the MWCNT-N aggregateranging in size (diameter) from 5 to 30 nm, with a meandiameter of �15 nm. In contrast to Fig. 1c, this aggregateexhibits more irregular, kinked tubes. The state of aggrega-tion and the surface area of some nanoparticulate materialscan change once they are introduced into the biologicalmedia as a result of surface-tension-mediated disaggrega-

to murine and human alveolar macrophages and human epithelial cells at

K. Soto et al. / Acta Biomaterialia 3 (2007) 351–358 355

tion of electrostatically or loosely agglomerated particu-lates. Consequently, the TEM images of the nanoparticu-late materials utilized in this investigation may not becompletely representative of the appearance of the sus-pended materials.

3.2. Cell viability assays

This investigation began by utilizing a murine lung mac-rophage cell line [13,14]. The main reasons for using this

Fig. 3. EC50 values for the nanoparticulate test materials for t

cell line was the consistency of the assay and its simplicity.The biochemistry of single cells, especially cell lines, can ofcourse differ from the response occurring in specific animaltissue, especially in the normal human body. Consequently,simple cell line assays avoid the variability between thesmall animal lung response and the human response, espe-cially among disparate age groups and people with differentpredispositions to adverse effects, not to mention theinability to perform human trials involving nanoparticulateinstillation. In addition, cell culture studies do not include

he three cell lines. NT represents no cytotoxicity response.

356 K. Soto et al. / Acta Biomaterialia 3 (2007) 351–358

any effects related to the interactions of the specific partic-ulates being studied with the respiratory tract and associ-ated systems and organs [27].

Relative cell viabilities were measured at a constantnanoparticulate material concentration of 5 lg m l�1 forthe three different cell lines: murine lung macrophage,human lung macrophage and human lung epithelial. Acomparison of cytotoxic response for these different celllines is shown in Fig. 2. Using chrysotile asbestos as a posi-tive control, it can be observed that almost all of thenanoparticulate materials, including the carbonaceousmaterials, exhibit toxicities with reference to the controlmedia and DMSO, in the case of the murine alveolar mac-rophage cell line. However, TiO2, in both the anatase andrutile forms, does not show a cytotoxic response. It shouldbe noted that silver (Ag-1) is particularly cytotoxic to themurine lung macrophage cell line. The human alveolarmacrophage (THB-1) cell line had very similar results tothe murine macrophage cell line (Fig. 2). This is also dem-onstrated by the statistical analysis. The nanoparticulatematerials had a greater toxic effect in the human epithelialcell line (A549), where the BC and the two types ofMWCNTs had an equivalent effect in comparison withthe chrysotile asbestos. Both anatase and rutile TiO2 alsoshowed a slight cytotoxic effect on the human epithelial cellline (Fig. 2). The human epithelial cells were demonstratedto be more sensitive than the murine and human macro-phages because the lung epithelial cells are the first lineof defense against a variety of insults.

Fig. 3 shows the nanoparticulate material concentra-tions required to cause a 50% reduction in cell viability(or 50% cell death, EC50) for all three cell lines. It was dem-onstrated that for the murine alveolar macrophages silver

Fig. 4. Comparison between cytotoxicity index and s

was particularly cytotoxic, and the Al2O3, ZrO2, Si3N4,BC and carbon nanotube aggregate materials were as cyto-toxic as the asbestos. The TiO2 was more cytotoxic in theanatase form. The rutile form showed no cytotoxicresponse in all three cell lines, as has been previouslyobserved [14]. Similar results were shown for the humanmacrophages. For the human epithelial cell line, thecarbonaceous materials (BC, MWCNT-R and MWCNT-N) induced cell death at lower concentrations (at�2.5 lg m l�1) than the murine and human macrophages(Fig. 3).

3.3. Specific surface area of the experimental nanomaterial

It is generally shown in animal studies that ultrafine ornanoparticulate materials cause markedly greater inflam-mation than that caused by fine particulates at equal mass[22]. Driscoll [23] has shown a correlation between lungtumor incidence in rats in chronic inhalation studies withdifferent particle types and retained particle surface areas(including TiO2, BC and diesel particulates). MacNee andDonaldson [24] have also shown a direct correlationbetween inflammation expressed as total polymorphonu-clear neutrophil leukocyte in lavage of rats and the totalsurface area of particles instilled, including BC and TiO2

fine and nanoparticulates. The implications of these find-ings are that general toxicity of nanoparticulates could berelated to the specific surface area. In Fig. 4 we have com-pared the specific surface area measurements in (Table 1)with corresponding and declining (relative) cytotoxicityindex rankings computed in Ref. [13], based on chrysotileasbestos as unity. It is apparent that there is no correlation,although the cytotoxicity index must be regarded as a rel-

pecific surface area of nanoparticulate materials.

K. Soto et al. / Acta Biomaterialia 3 (2007) 351–358 357

ative value depending upon the basis chosen. Warheit et al.[30] have also recently concluded that specific surface areais not correlated with particle toxicity for pulmonary instil-lation in rats. It should also be noted that, as demonstratedby Renwick et al. [1], that the ultrafine versus fine particleeffect is observed for aggregated particles with primary par-ticle sizes for ultrafines of �20 nm versus fines of �200 nm.

On considering Table 1, it can also be observed that theparticle morphology or aggregate morphology is also notcorrelated with the cytotoxicity response for either themurine or the human cell line exposure since a variety ofmorphologies within the nanorange exhibit equivalent orsimilar cytotoxicities.

4. Conclusions

In vitro studies done in a murine alveolar macrophagecell line have demonstrated degrees of cytotoxicity for avariety of manufactured nanoparticulate materials. Similarcytotoxicity profiles were obtained for a human lung mac-rophage and a human lung epithelial cell line. These results,although very short term and performed in vitro, begin toprovide a preliminary understanding of the human respira-tory overview for potential health effects of airbornenanoparticulates.

Although we have shown a convincing correlationbetween murine lung macrophage cell line assay resultsand corresponding results for human lung macrophageand epithelial cell lines, there are no implications of thehealth effects in humans except to caution against ingestionof large concentrations of nanoparticulates. Even if nano-particulates are toxic to humans, this does not necessarilynegate their utility.

Acknowledgements

This research was supported by the University of Texasat El Paso Alliance for Graduate Education and the Pro-fessoriate Program (AGEP) (K.F.S.), an EPA StudentScholarship (K.F.S.), an NIH Research Centers at Minor-ity Institution (RCMI) Grant (K.M.G.), the Mr. and Mrs.MacIntosh Murchison Endowed Chair (L.E.M.), South-west Consortium for Environmental Research & Policy(SCERP): Project A-05-1 (L.E.M. and K.M.G.), andgrant number 1S11ES013339-01A1 from the NationalInstitute of Environmental Health Sciences (NIEHS),NIH. Its contents are solely the responsibility of theauthors and do not necessarily represent the official viewsof NIEHS, NIH. We also acknowledge the help of Profes-sor Geoff Saupe, UTEP Chemistry Department, for hishelp in the BET analysis.

References

[1] Renwick LC, Donaldson K, Clouter A. Impairment of alveolarmacrophage/phagocytosis by ultrafine particles. Toxicol Appl Phar-macol 2001;172:119–27.

[2] Warheit DB. Nanoparticles: health impacts? Mater Today2004:32–5.

[3] Oberdorster G, Gelein R-M, Ferin J, Loeiss B. Association of particleair pollution and acute mortality: involvement of ultrafine particles?Inhal Toxicol 1995;7:111–24.

[4] Oberdorster G. Pulmonary effects of inhaled ultrafine particles. IntArch Occup Environ Health 2001;74:1–8.

[5] Lighty JS, Veranth JM, Sarofim AF. Combustion aerosols:factors governing their size and composition and implicationsto human health. J Air Waste Manage Assoc 2000;50:1565–618.

[6] Salvi S, Holgate ST. Mechanisms of particulate matter toxicity. ClinExp Allergy 1999;29:1187–94.

[7] Seaton A, Macnee W, Donaldson K, Godden D. Particulateair pollution and acute health effects. Lancet 1995;345:176–8.

[8] Hoek G, Bronckreef B, Goldbohm S, Fischer P, van den Brandt PA.Association between mortality and indications of traffic-related airpollution in The Netherlands: a cohort study. Lancet2002;360:1203–9.

[9] Donaldson K, Stone V, Clouter A, Renwick L, MacNee W. Ultrafineparticles. Occup Environ Med 2001;58:211–5.

[10] Pope CA, Burnett RT, Thon MJ, Calle EE, Krewski Ito K,Thurston GD. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA 2002;287 :1132–41.

[11] Nikula KJ, Snipes MB, Barr EB, Griffith WC, Henderson RS,Mouderly JJ. Comparative pulmonary toxicities and carcinogenicitiesof chronically inhaled diesel exhaust and black carbon in F344 rats.Fund Appl Toxicol 1995;25(1):80–94.

[12] Ovrevik J, Myran T, Refunes M, Lag M, Becher R, Hetland RB,et al. Mineral particles of varying composition induce differentialchemokine release from epithelial lung cells: importance ofphysico-chemical characteristics. Ann Occup Hygiene 2005;49(3):219–31.

[13] Soto KF, Carrasco A, Powell TG, Garza KM, Murr LE.Comparative in vitro cytotoxicity assessment of somemanufactured nanoparticulate materials characterized bytransmission electron microscopy. J Nanoparticle Res2005;7:145–69.

[14] Soto KF, Carrasco A, Powell TG, Murr LE, Garza KM. Biologicaleffects of nanoparticulate materials. Mater Sci Eng 2006;C26:1421–7.

[15] Lam C, James JT, McCluskey R, Hunter RL. Pulmonary toxicity ofcarbon nanotubes in mice 7 and 90 days after intratrachaelinstallation. Toxicologist 2004;72(5–1):44–8.

[16] Warheit DB, Lawrence BR, Reed KL, Roach DH, ReynoldsGAM, Webb TR. Comparative pulmonary toxicity assessmentof single-wall carbon nanotubes in rats. Toxicol Sci 2004;77:117–25.

[17] Schsedova AA et al. Exposure to nanotube material: assessment ofnanotube cytotoxicity using human keratinocyte cells. J ToxicolEnviron Health 2003;66:1901–18.

[18] Salvi S, Holgate ST. Mechanisms of particulate matter toxicity. ClinExp Allergy 1999;29:1187–94.

[19] Dellinger B, Pryor WA, Cueto R, Squadrito GL, HedgeV, Deutsch WA. The role of combustion-generated radicalsin the toxicity of PM 2.5. Proc Combust Inst 2000;28:2675–81.

[20] Shi T, Schims RPF, Knaapen AM, Kuhbusch T, Pitz M,Heinrich J, et al. Hydroxy/radical generation by electronparamagnetic resonance as a new method to monitor ambientparticulate matter composition. J Environ Monitor 2003;5:550–6.

[21] Murr LE, Soto KF. TEM comparison of chrysotile (asbestos)nanotube and carbon nanotubes. J Mater Sci Lett 2004;39:4941–7.

[22] Donaldson K, Stove V, MacNee W. The toxicology of ultrafineparticles. In: Maynard RL, Howards CV, editors. Particulate

358 K. Soto et al. / Acta Biomaterialia 3 (2007) 351–358

matter: properties and health effects upon health. Oxford: BIOS. p.115–27.

[23] Driscoll KE. Role of inflammation in the development of rat lungtumors in response to chronic particle exposure. Inhal Toxicol1996;8(Suppl):139–53.

[24] MacNee W, Donaldson K. Mechanism of lung injury caused by PM10

and ultrafine particles with special reference to COPD. Evn Resp J2003;21:475–515.

[25] Ding B, Kim J, Kimura E, Shiratori S. Layer-by layer structured filmsof TiO2 nanoparticles and poly(acrylic acid) on electrospun nanofi-bres. Nanotechnology 2004:913–7.

[26] Pourahmad J, O’Brien PM. A comparison of hepatocyte cyto-toxic mechanisms for Cu2+ and Cd2+. Toxicology 2000;143:263–73.

[27] Oberdorster G, Ferin J, Lehnert BE. Correlation between particlesize, in vivo particle persistence and lung injury. Environ HealthPerspective 1994;102(Suppl.5):173–9.

[28] Donaldson KL, Borm PJA. The quartz hazard: a variable entity. AnnOccup Hyg 1998;42:287–96.

[29] Lam C-W, James JT, McCluskey R, Arepalli S, Hunter RC. A reviewof carbon nanotubes toxicity and assessment of potential occupa-tional and environmental health risks. Crit Rev Toxicol 2006;36:189–217.

[30] Warheit DB, Webb TR, Sayes CM, Colvin VL, Reed KL. Pulmonaryinstillation studies with nanoscale TiO2 rods and dots in rats: toxicityis not dependent upon particle size and surface area. Toxicol Sci2006;91(1):227–36.