Development of a base set of toxicity tests using ultrane TiO2 particles as a component of nanoparticle risk management

David B. Warheit , Robert A. Hoke, Carol Finlay, E. Maria Donner, Kenneth L. Reed, Christie M. Sayes
DuPont Haskell Laboratory for Health and Environmental Sciences, Newark, DE, USA Received 16 March 2007; received in revised form 24 April 2007; accepted 24 April 2007 Available online 27 April 2007

Abstract The development of a risk management system for nanoscale or ultrane particle-types requires a base set of hazard data. Assessing risk is a function of hazard and exposure data. Previously, we have suggested parallel tracks as a strategy for conducting nanoparticle research. On the one hand, mechanistic studies on representative nanoparticles could be supported by governmental agencies. Alternatively, with regard to commercial nanoparticles, the environmental, health and safety (EHS) framework would include a minimum base set of toxicity studies which should be supported by the companies that are developing nano-based products. The minimum base set could include the following criteria: substantial particle characterization, pulmonary toxicity studies, acute dermal toxicity and sensitization studies, acute oral and ocular toxicity studies, along with screening type genotoxicity, and aquatic toxicity studies. We report here the toxicity results of a base set of hazard tests on a set of newly developed, well-characterized, ultrane TiO2 (uf-TiO2 ) particle-types. In vivo pulmonary toxicity studies in rats demonstrated low inammatory potential and lung tissue toxicity. Acute dermal irritation studies in rabbits and local lymph node assay results in mice indicated that uf-TiO2 was not a skin irritant or dermal sensitizer. Acute oral toxicity studies demonstrated very low toxicity and uf-TiO2 produced short-term and reversible ocular conjunctival redness in rabbits. Genotoxicity tests demonstrated that uf-TiO2 was negative in both the bacterial reverse mutation test and in an in vitro mammalian chromosome aberration test with Chinese hamster ovary cells. The results of aquatic toxicity screening studies demonstrated that uf-TiO2 exhibited low concern for aquatic hazard in unaerated, 48 h, static acute tests using the water ea, Daphnia magna; exhibited low concern for aquatic hazard in unaerated, 96 h, static acute tests using the rainbow trout, Oncorhynchus mykiss; and exhibited medium concern in a 72 h acute test using the green algae Pseudokirchneriella subcapitata. To summarize the ndings, the results of most of the studies demonstrated low hazard potential in mammals or aquatic species following acute exposures to the ultrane TiO2 particle-types tested in this program. 2007 Elsevier Ireland Ltd. All rights reserved.
Keywords: Ultrane particles; Nanoparticles; Nanomaterials; Nanoparticle toxicity testing; Minimum base set; Acute toxicity; Subchronic toxicity; Pulmonary toxicity; Dermal toxicity; Skin sensitization; Oral toxicity; Ocular toxicity; Genotoxicity; Aquatic toxicity

1. Introduction
Corresponding author at: DuPont Haskell Laboratory, 1090 Elkton Road, PO Box 50, Newark, DE 19714-0050, USA. Tel.: +1 302 366 5322; fax: +1 302 366 5207. E-mail address: david.b.warheit@usa.dupont.com (D.B. Warheit).

The intent of a nanoparticle risk management framework is to implement a systematic process for identifying environmental health and safety (EHS) risks related to exposures to newly developed engineered nanomaterials.

0378-4274/$ see front matter 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.toxlet.2007.04.008

The development and forthcoming commercialization of many different engineered nanomaterial-types for uses in a variety of industrial, chemical, manufacturing, consumer, medical and diagnostic applications will present a challenge for companies and regulators to ensure the safety of products for workers and ultimately, consumers. Moreover, the development of effective and safe products containing nanomaterials should be a fundamental component of the product stewardship process. This is an integral component of the broad engagement process with stakeholders on EHS issues, contributing to the public awareness and condence in products developed through nanoscale science and engineering. An EHS framework is being developed to characterize the potential risks related to exposures to nanoscale or ultrane particle-types. Of course, determination of risk is a product of both exposure and hazard assessments. However, in many cases the exposure potential cannot be ascertained, due in large part, to the current limitations of technology to measure nanoscale particle exposures in the workplace. Nonetheless, the risk management framework could include a minimum base set of toxicity (hazard) screening studies which provide a fundamental characterization of the potential hazards of the particle-types being investigated for human health and ecological effects. This base set process is modeled, in part, on the preparation of the SIDS (screening information data sets) used by OECD (Organisation for Economic Co-operation and Development) and US EPA (United States Environmental Protection Agency) for the investigation of HPV (High Production Volume) chemicals (http://www.epa.gov/chemrtk/pubs/general/sidsappb.htm http://www.epa.gov/chemrtk/pubs/general/sidsappb.pdf http://www.oecd.org/dataoecd/13/14/36045229.pdf). This base set serves as a reference point for the type of screening information that should be addressed as a new product is being developed. The generation of a base set of information is important to sufciently characterize the EHS impacts of nanoscale or ultrane materials and to build the foundational data that is necessary to develop a risk assessment framework for nanomaterials.
Table 1 Base set hazard tests Nanomaterial physiochemical characterization Size and size distribution Crystal structure Chemical composition Surface reactivity

The minimum base set is an evolving concept designed to characterize the hazards associated with exposures to nanomaterials, both in mammalian species as well as in ecological environments. Justication for these particular tests rests on the following criteria: (1) potential routes of exposures (i.e., pulmonary, dermal, oral and/or ocular); (2) screening for potential carcinogenic effects (mutation and chromosomal aberration assays); and (3) screening for potential toxic aquatic effects (exposures to rainbow trout, Daphnia, and algae). The base hazard set of tests is not meant to provide for an exhaustive assessment of toxicity, but is designed to facilitate a reasonable balance between an adequate toxicity characterization and a practical strategy for the development of new nanomaterials. Thus, the goal is to make the base set sufciently robust in order to guide adequate risk-evaluation processes, in a manner commensurate with existing regulatory and voluntary standards. This manuscript briey describes the methodology and results of ten different toxicity studies conducted with newly developed ultrane TiO2 particle-types. Prior to the commencement of the studies, substantial physicochemical characterization of the test materials was carried out. Subsequently, hazard studies were implemented which included the following tests (see Table 1): a doseresponse pulmonary bioassay study in rats with postexposure periods lasting through 3 months; a dermal irritation test in rabbits; a skin sensitization study in mice; an acute oral toxicity study in rats; and an eye irritation study in rabbits; two in vitro genotoxicity studies, i.e. a bacterial reverse mutation test and chromosomal aberration study; as well as three aquatic toxicity screening studies with rainbow trout, daphnia, and green algae. Thus, it is proposed that these studies form the basis for a minimum base set of hazard tests as a component of a nanoparticle risk management system.
2. Methods The following tests with newly developed ultrane TiO2 particle-types were conducted as components of the base set of hazard assessment tests identied in Table 1. A brief descrip-

Mammalian hazard tests Pulmonary bioassay Skin irritation Skin sensitization Acute oral toxicity Eye irritation

Genotoxicity tests Bacterial reverse mutation Chromosomal aberration

Aquatic screening battery Rainbow trout Daphnia Green algae

tion of each of the tests is provided in the methodology section below. 2.1. Animals All procedures using animals were reviewed and approved by the Institutional Animal Care and Use Committee. The DuPont Haskell Laboratory animal program is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). 2.2. Particle-types and physicochemical characterization (see Fig. 1, Table 2) An ILSI Research Foundation/Risk Science Institute Nanomaterial Toxicity Screening Working Group has previously developed the elements of a screening strategy for the hazard identication of engineered nanomaterials (Oberd orster et al., 2005). Fundamental to the conclusions of this report is that adequate characterization of nanomaterials is required. This represents one of the key aspects of toxicity screening strategies. For the pulmonary bioassay studies conducted in rats, the physicochemical characteristics of the ultrane TiO2 particletypes (identied as uf-A and uf-B) were measured in the dry state (representing the supplied material) as well as in aqueous suspension, including water and phosphate-buffered saline (PBS) solution vehicle (representing the administered material exposed to the test subject). An ultrane particle is dened as a particle of average primary size of roughly 100 nm. Because both the uf-A, uf-B and uf-C particulates have median particle sizes of 140 nm, with fractions of the size distribution that fall below 100 nm, they are termed ultrane. Often ultrane particles have different uses or have enhanced properties when compared to their bulk counterparts. For certain uses they can be more effective than their bulk counterparts because of their smaller size and higher surface area. For the pulmonary bioassay studies, two different ultrane TiO2 particles in the rutile crystal phase, designated as uf-A and uf-B, were obtained from the DuPont Company. DuPont uf-A is composed of a titanium dioxide core with an alumina surface coating (98% titanium dioxide and 2% alumina) and possesses a median particle size of 136 nm in water (aqueous suspension buffered in 0.1% tetrasodium pyrophosphate) using dynamic light scattering and an average BET surface area of 18.2 m2 /g. DuPont uf-B is composed of a titanium dioxide core with a silica and alumina surface coating (88 wt% titanium dioxide, 7 wt% amorphous silica and 5 wt% alumina,) and possesses a median particle size of 149.4 nm in water (aqueous suspension buffered in 0.1% tetrasodium pyrophosphate) using dynamic light scattering and an average BET surface area of 35.7 m2 /g (Brunauer et al., 1938) (see Fig. 1, Table 2). Controls for the pulmonary bioassay study included nesized TiO2 particle-types in the rutile form obtained from the DuPont Company; P25 ultrane TiO2 particles, consisting of an 80:20 ratio of anatase:rutile TiO2 particles, purchased from

Fig. 1. Characterization of ultrane and ne titanium dioxide (TiO2 ) particles used in the pulmonary bioassay study. High resolution scanning electron micrographs (HR-SEM) of ultrane samples A and B (labeled a and b) and ne sample (c). Micrographs depict variations in surface treatment and particle size within each of the ultrane and ne TiO2 particle-types.

Degussa Corporation; and Min-U-Sil -quartz particles purchased from US Silica Company. Fine-TiO2 is composed of a titanium dioxide core with an alumina surface coating (99% titanium dioxide and 1% alumina) and possesses a median particle size of 380 nm in water (aqueous suspension buffered in 0.1% tetrasodium pyrophosphate) using DLS and an average BET surface area of 5.8 m2 /g. All TiO2 rutile particle samples underwent a neutralization process during production, which neutralizes acidic chloride groups on the particle surface. P25

Table 2 Characterization of ultrane titanium dioxide (TiO2 ) particles used in the pulmonary (uf-A, uf-B); skin, ocular, oral, genetic, and aquatic toxicity studies (uf-C) Sample Surface area (m2 /g) Crystalline phase Chemical reactivity delta ba Particle size and distribution (nm) in uf-A uf-B uf-C 18.2 35.7 38.5 100% rutile 100% rutile 79% rutile; 21% anatase 10.1 1.2 0.9 watera (%) in PBS (%) 1990 25 2669 25 5.64 7.14 4.80 pH in water

136.0 35 149.4 50 140.0 44

Table of physical properties of the DuPont ultrane TiO2 samples used in the study, including specic surface area, crystallinity, chemical reactivity, particle size and distribution, and pH. a Indicates aqueous solution of 0.1% tetrasodium pyrophosphate.

ultrane TiO2 particles, consist of 80% in the anatase crystal phase and 20% in the rutile crystal phase and are composed of 100 wt% titanium dioxide with a median particle size of 129.4 nm in water (aqueous solution buffered in 0.1% tetrasodium pyrophosphate) using dynamic light scattering and an average BET surface area of 53.0 m2 /g. Quartz particles (crystalline silica (-quartz), Min-U-Sil 5) with reported size range of 0.22 m were obtained from US Silica Company (Berkeley Springs, WV). Min-U-Sil 5 is composed of 100 wt% silicon dioxide in the -quartz crystal phase, with a median particle size of 480 nm in water (aqueous solution buffered in 0.1% tetrasodium pyrophosphate) and an average surface area of 5.2 m2 /g. Each of the TiO2 samples were characterized to identify its crystallinity and surface area in its dry native state and its size, size distribution, pH, and chemical reactivity in water and buffered solutions. X-ray uorescence was used to measure purity/composition. X-ray diffraction (XRD) (Philips XPERT automated powder diffractometer, Model 3040) was used to determine crystal structure, and crystallite size (Otwinowski and Minor, 1997). XRD patterns do provide information on the primary crystallite size, but do not, however, accurately represent the particle size of the dispersed aggregates. Therefore, sizing data was obtained using two separate methods: dynamic light scattering (DLS) and BET surface area analysis (Brunauer et al., 1938). DLS measurements (Malvern Zetasizer NanoS, model Zen1600) were taken on each particle suspension in water (aqueous suspension buffered in 0.1% tetrasodium pyrophosphate) and phosphate-buffered saline (PBS) solution. BET (Micromeritics ASAP 2405), to measure specic surface area, was taken on each particle sample in its dry state under nitrogen. For the remaining, non-pulmonary toxicity studies (i.e., dermal, oral, ocular, genotoxicity and aquatic studies), a next generational form of ultrane TiO2 particle-type (termed ufC) was used which has a similar particle size distribution to uf-A and uf-B (Table 2). Ultrane-C TiO2 had a crystalline phase determination of 79% rutile and 21% anatase. X-ray uorescence determined a composition of approximately 90 wt% TiO2 , 7% alumina, and 1% amorphous silica. Using dynamic light scattering, the median particle size was 140 nm in water (aqueous solution buffered in 0.1% tetrasodium pyrophos-

phate). The BET surface area was 38.5 m2 /g. Similar to the 100% rutile TiO2 particle-types, uf-C TiO2 particle samples underwent a neutralization process during production.

3. Toxicity studies 3.1. Pulmonary toxicity study in rats with ultrane-TiO2 particles The aim of this study was to assess lung toxicity in rats of newly developed, well-characterized, ultrane-TiO2 particles and compare them to TiO2 samples in two different size ranges and surface modications. Groups of male rats were intratracheally instilled with doses of 1 or 5 mg/kg of either two ultrane rutile TiO2 particle-types (uf-A or uf-B); rutile ne-sized TiO2 particles; 80:20 anatase:rutile ultrane-TiO2 particles; or -quartz particles (positive control). Phosphate-buffered saline (PBS) solution instilled rats served as vehicle controls. Following exposures, the lungs of PBS and particle-exposed rats were evaluated for bronchoalveolar lavage (BAL) uid inammatory markers, cell proliferation, and by histopathology at post-instillation time points of 24 h, 1 week, 1 month, and 3 months. The fundamental features of this pulmonary bioassay are doseresponse evaluations and time-course assessments to determine the sustainability of any observed effect. Thus, the major endpoints of this study were the following: (1) time course and dose/response intensity of pulmonary inammation and cytotoxicity; (2) airway and lung parenchymal cell proliferation; and (3) histopathological evaluation of lung tissue (see Table 3 for the experimental protocol). For the bronchoalveolar lavage studies, groups of rats (5 rats/group/dose/time point) were intratracheally instilled with single doses of either 1 or 5 mg/kg ultrane-TiO2 particle-type A (uf-A); ultrane-TiO2 particle-type B (uf-B); ne-TiO2 particles; 80% anatase:

Table 3 Protocol for ultrane-TiO2 particle pulmonary bioassay study

20% rutile ultrane TiO2 particle-types; or quartz (crystalline silica) particles (see Table 3). The intratracheal instillation method of exposure can be a reliable qualitative screen for assessing the pulmonary toxicity of inhaled particles (Warheit et al., 2005). All particle samples were prepared in a volume of phosphate-buffered saline (PBS) solution and subjected to sonication for 15 min at 60 Hz. Groups of PBS-instilled rats served as controls. The lungs of PBS and particle-exposed rats were evaluated by BAL uid analyses at 24 h, 1 week, 1 month and 3 months postexposure (pe). For the lung tissue studies, additional groups of animals (4 rats/group/high dose/time period) were instilled with the particle-types listed above plus the vehicle control, i.e., PBS. These studies and corresponding groups of rats were dedicated to lung tissue analyses but only the high dose groups (5 mg/kg) and PBS controls were utilized in the morphology studies. These studies consisted of cell proliferation assessments and histopathological evaluations of the lower respiratory tract. Similar to the BAL uid studies, the intratracheal instillation exposure period was followed by 24 h, 1-week, 1-month, and 3month recovery periods (see Warheit et al., 1991, 1997, 2007 for more details). For statistical analyses, each of the experimental values was compared to their corresponding sham control values for each time point. A one-way analysis of variance (ANOVA) and Bartletts test were calculated for each sampling time. When the F-test from ANOVA was signicant, the Dunnetts test was used to compare means from the control group to each of the groups exposed to particulates. Signicance versus PBS controls was judged at the 0.05 probability level. 3.2. Acute dermal irritation study in rabbits The acute dermal irritation tests were conducted according to US EPA and OECD 404 guidelines

(USEPA, 1998; OECD, 2002). Ultrane TiO2 -C (uf-C) particles were applied as a single 0.5 g dermal dose to the shaved intact skin of three male New Zealand White rabbits. The test substance, moistened with 0.25 mL of deionized water, was applied to a 6 cm2 area of skin. The application area was covered with a 2-ply gauze square which was held in place with non-irritating tape and covered with porous tape for a semi-occlusive dressing. The rabbits were exposed to the test substance for 4 h after which the test substance was removed. Test sites were evaluated by Draize score for signs of dermal irritation approximately 60 min, and 24, 48, and 72 h after test substance removal. The rabbit that was initially treated was also examined immediately after test substance removal. 3.3. Dermal sensitization test: local lymph node assay (LLNA) in mice For the dermal sensitization component of the base set study, responses in mice using the local lymph node assay (LLNA) were utilized (OECD 429 Guideline). The objective of this study was to evaluate the potential of uf-C TiO2 particles to produce a dermal sensitization response in mice using the local lymph node assay (LLNA). Five groups of ve female CBA/JHsd mice were dosed for 3 consecutive days with 0% (vehicle control), 5%, 25%, 50%, or 100% ultrane TiO2 particle-types on both ears. N,N-Dimethyl formamide was used as the diluting vehicle. One group of ve female mice was dosed for 3 consecutive days with 25% hexylcinnamaldehyde (HCA) in 4:1 acetone:olive oil (AOO) as a positive control and one group of ve female mice was dosed for 3 consecutive days with AOO as a positive control vehicle. On test day 5 of the assay, mice received 3 H-Thymidine by tail vein injection and were sacriced approximately 5 h later. The cell proliferation in the draining auricular lymph

nodes of the ears from the test substance groups was then evaluated and compared to the vehicle control group. A stimulation index (SI) was derived for each experimental group by dividing the mean disintegrations per minute (dpm) of each experimental group by the mean dpm of the vehicle control group. The decision process in regard to a positive response includes an SI of greater than or equal to 3.0 together with consideration of doseresponse and, where appropriate, statistical signicance. Signicance was judged at p < 0.05 except for dpm data that were judged at p < 0.01. Lymph node dpm data were log transformed to obtain normality or homogenous variances. When possible, an EC3 value for the stimulation index data was derived from linear interpolation of points on the doseresponse curve immediately above and below the threefold threshold. 3.4. Acute oral toxicity study in ratsup and down procedure The acute oral toxicity test-up and down procedure was conducted according to US EPA and OECD 425 guidelines (USEPA, 2002; OECD, 2001). A single dose of uf-C TiO2 particles suspended in deionized water was administered by oral gavage to one fasted female rat each at a dose of 175, 550, or 1750 mg/kg and to three fasted female rats at a dose of 5000 mg/kg. The rats were dosed one at a time at a minimum of 48 h intervals. The rats were observed for mortality, body weight effects, and clinical signs for 14 days after dosing. All rats were necropsied to detect grossly observable evidence of organ and tissue damage or dysfunction. 3.5. Acute ocular irritation study in rabbits The acute eye irritation tests were conducted according to US EPA and OECD 405 guidelines (USEPA, 1998; OECD, 2002). Uf-C TiO2 particles were evaluated for acute eye irritation potential in three young adult New Zealand White rabbits. The study was conducted after conrming that the compound was not a severe irritant or corrosive to the skin. Approximately 57 mg of test substance was administered to one eye of each animal. The eyes remained unwashed following treatment. One rabbit was initially treated. Since no severe irritation or corrosion was observed, two additional rabbits were treated to complete the test. The conjunctiva, iris, and cornea of each treated eye were evaluated and scored according to a numer-

ical scale approximately 1, 24, 48, and 72 h following administration of the test substance. 4. Genotoxicity test methods 4.1. Bacterial reverse mutation test Uf-C TiO2 particles were evaluated for mutagenicity in the Bacterial Reverse Mutation (Ames) Test using the plate incorporation method. Salmonella typhimurium strains TA98, TA100, TA1535, and TA1537 and Escherichia coli strain WP2uvrA were tested in the absence and presence of an exogenous metabolic activation system (Aroclor-induced rat liver S9). Sterile water was chosen as the dosing vehicle based on the solubility of the test substance and compatibility with the target cells. The test substance formed a homogeneous suspension at 50 mg/mL, the highest concentration that was tested in the study. A plating aliquot of 100 L was used. This dose was achieved using a concentration of 50 mg/mL and a 100 L plating aliquot. The dose levels in the study were 100, 333, 1000, 3333, and 5000 g per plate. Appropriate positive controls were included in the study. The study was conducted according to the US EPA and OECD 471 testing guidelines (USEPA, 1998; OECD, 1998). Milli-Q water was chosen as the dosing vehicle for the Ames test based on compatibility with the target cells. The test substance formed a homogeneous suspension at 50 mg/mL, the highest concentration that was tested in the study. A plating aliquot of 100 g/mL was used. The dose levels in the study were 100, 333, 1000, 3333, and 5000 g per plate. 4.2. In vitro mammalian chromosome aberration test in Chinese hamster ovary cells Uf-C TiO2 particles were tested for their ability to induce structural chromosome aberrations in Chinese hamster ovary (CHO) cells in the absence and presence of an exogenous metabolic activation system (Aroclor-induced rat liver S9). The test substance was prepared in Milli-Q water as this vehicle was determined to be the solvent of choice based on the solubility of the test substance and compatibility with the target cells. The test substance formed a white suspension in the vehicle at approximately 50 mg/mL, the highest stock concentration prepared on the study. Osmolality, pH, and test substance precipitation were taken into account for dose level setting, in addition to cell count and mitotic activity. Cytogenetic evaluations of structural aberrations were conducted in 200 cells in

metaphase at 750, 1250, and 2500 g/mL for the 4 h nonactivated test condition; at 62.5, 125, and 250 g/mL, for the 4 h activated test condition, and at 25, 50, and 100 g/mL for the 20 h non-activated test condition. Numerical aberrations were recorded as well. Appropriate positive controls were included in the test. The study was conducted according to the US EPA and OECD 473 testing guidelines (USEPA, 1998; OECD, 1998). 5. Aquatic toxicity test methods 5.1. Static, acute, 96 h toxicity screening tests with Oncorhynchus mykiss (rainbow trout) The acute toxicity of ne and uf-C TiO2 particle-types to the rainbow trout, O. mykiss was determined in unaerated, 96 h static tests according to OECD 203 testing guidelines (OECD, 1992). The study was conducted with four concentrations each of ne-sized rutile TiO2 particles and ultrane TiO2 particles and a dilution water control at a mean temperature of 12.2 C (range of 12.112.3 C) and 12.2 C (range of 12.112.5 C), respectively. One test chamber was used per test substance concentration for each test substance with ve test organisms in each chamber. Based on visual observations, the dilution water controls, 0.1 and 1.0 mg/L test concentrations were clear and colorless with no precipitate at test start. The 10 and 100 mg/L test concentrations were cloudy with a slight amount of suspended substance present at test start. All water quality parameters were within acceptable limits during the exposure. 5.2. Static, acute, 48 h toxicity screening tests with Daphnia magna The acute toxicity of ne or uf-C TiO2 particles to the water ea, D. magna (less than 24 h old) was determined in unaerated, 48 h static tests according to OECD 202 testing guidelines (OECD, 2004). The study was conducted with four concentrations each of ne or ultrane TiO2 particles and a dilution water control at a mean temperature of 20.2 C (range of 20.120.3 C) and 20.1 C (range of 20.020.2 C), respectively. One test chamber was used per test substance concentration for each test substance with 10 test organisms in each chamber. Based on visual observations, the dilution water controls, 0.1, and 1.0 mg/L test concentrations were clear and colorless with no precipitate at test start. The 10 and 100 mg/L test concentrations were cloudy (white in color) and had

suspended substance present at test start. All water quality parameters were within acceptable limits during the exposure. 5.3. Static, acute, 72 h growth inhibition toxicity screening test to the green algae, Pseudokirchneriella subcapitata The acute toxicity of ne or uf-C TiO2 particles to the green algae, P. subcapitata, was determined in a 72 h, static toxicity test according to OECD 201 testing guidelines (OECD, 1984). The study was conducted with a synthetic algalassay procedure (AAP) nutrient medium blank control and ve concentrations of ne or ultrane TiO2 particles at a mean lighting intensity of 8938 lux (range of 82009500 lux), and a dilution water control at a mean temperature of 23.8 C (range of 23.723.8 C) and a shaking speed of 100 rpm. Two replicates were used per blank control and test concentration each with an initial cell count (density) of 10,000 cells/mL. Based on visual observations the 100 mg/L test concentration solutions were very cloudy with suspended substance present at test start. The 10 mg/L test concentration solutions were slightly cloudy with suspended substance present at test start. The blank control and remaining test concentration solutions were clear and colorless with no visible precipitate at test start. All environmental parameters were within acceptable limits during the exposure. 6. Results of the various assays (see Table 4) 6.1. Physicochemical characterization of particles The particle size distribution (PSD) results for the ultrane TiO2 particle-types were highly agglomerated following dispersion in the phosphate-buffered saline solution, the vehicle utilized for the intratracheal instillation exposures, although the uf-A sample showed a slightly smaller average particle size (see Table 2). The high degree of agglomeration in general arises from the proximity of the buffer solution pH (near-neutral) to isoelectric point of the samples. The relatively high ionic strength of the buffer solution is also a factor that promotes agglomeration. As described in an earlier section, the uf-A and uf-B samples were utilized for the pulmonary bioassay studies, and uf-C samples were used in the remaining studies. This is not surprising since the development of materials for commercialization can be an evolving process and one developmental particulate candidate could demonstrate

Table 4 Base set teststoxicity results Test Acute and short-term tests Pulmonary bioassay Acute oral toxicity test Skin irritation test Skin sensitization Eye irritation test Genetic toxicity tests Bacterial reverse mutation test Chromosomal aberration study Aquatic toxicity battery Rainbow trout Daphnia Algae Result Low toxicity Low toxicity Not a skin irritant Not a sensitizer Minor ocular conjunctival redness Negative Negative Low concern Low concern Medium concern

6.4. Dermal sensitization test: local lymph node assay (LLNA) in mice No statistically signicant differences in mean body weights and body weight gains compared to the vehicle control group were observed at any test concentration. No clinical signs of toxicity were observed in the study. Statistically signicant increases in cell proliferation measurements compared to the vehicle control group were observed at the 50% and 100% test concentrations. Stimulation indices (SIs) of less than 3.0 were observed at all test concentrations of uf-C TiO2 particle-types. Therefore, the EC3 value (the estimated concentration required to induce a threshold positive response, i.e., SI = 3) for the test substance under the conditions of this study was not calculable. A 25% concentration of the positive control, HCA, produced a dermal sensitization response in mice. Therefore, the LLNA test system was valid for this study with ultrane TiO2 particles. Under the conditions of this study, uf-C TiO2 particles did not produce a dermal sensitization response in mice. 6.5. Acute oral toxicity study in ratsup and down procedure No mortality occurred on the study. The single rat dosed at 1750 mg/kg and the three rats dosed at 5000 mg/kg exhibited grey colored feces during the study. No biologically important body weight losses occurred. No gross lesions were present in the rats at necropsy. Based on this study, the oral LD50 for uf-C TiO2 particles was greater than 5000 mg/kg for female rats. 6.6. Acute ocular irritation study in rabbits The test substance produced conjunctival redness (score 1 or 2) in the treated eye of all three rabbits. Fluorescein stain examinations did not reveal any corneal injury. The treated eyes of the rabbits were normal by 24 or 48 h after instillation of the test substance. No clinical signs were observed, and no body weight loss occurred. Based on this study, uf-C TiO2 particles produced conjunctival redness in the treated rabbit eye which was reversible. 7. Genotoxicity results

greater efcacy than another for a given commercial application. 6.2. Pulmonary toxicity study in rats with ultrane-TiO2 particles The ranking of lung inammation/cytotoxicity/cell proliferation and histopathological responses (in descending order) was quartz > 80:20 anatase:rutile uf TiO2 > ne-sized TiO2 = uf-A = uf-B. Exposures to quartz and to a lesser degree, 80:20 anatase:rutile TiO2 particles produced pulmonary inammation, cytotoxicity and adverse lung tissue effects. In contrast, exposures to ne-TiO2 particles or to rutile uf-A/uf-B TiO2 particle-types produced transient inammation and no adverse histopathological or cell proliferative effects in the lungs of exposed rats at any postexposure time period. It was concluded that differences in responses to 80:20 anatase:rutile TiO2 particles versus the rutile uf-A and uf-B TiO2 particle-types could be related to crystal structure, inherent pH of the particles, or surface chemical reactivity. Finally, the results demonstrate that exposures to ultrane-TiO2 particletypes can produce differential pulmonary effects, based upon their composition, crystal structure and/or surface reactivity. 6.3. Acute dermal irritation study in rabbits The rabbits exhibited no dermal irritation during the study. No clinical signs of toxicity were observed, and no body weight loss occurred. Under the conditions of this study, uf-C TiO2 particles were not considered to be skin irritants.

7.1. Bacterial reverse mutation test No positive mutagenic responses or compoundrelated toxicity were observed at any dose level with

any tested strain (Salmonella typhimurium tester strains TA98, TA1000, TA 1535, TA 1537, or Escherichia coli strain WP2uvrA) when tested either with or without an S9 metabolic activation system (Arocolor-induced rat liver S9). Compound precipitate was observed at the top three or four dose levels. Uf-C TiO2 particles showed no evidence of mutagenicity in this study. 7.2. In vitro mammalian chromosome aberration test in Chinese hamster ovary cells In the preliminary toxicity assay, the highest concentration tested was 5000 g/mL, the OECD 473 guideline limit dose for this test system. The osmolality or pH of the highest test substance concentration in medium was not signicantly different from the vehicle control. Substantial toxicity (at least a 50% reduction in cell growth relative to the solvent control) was not observed at any concentration level in any test condition. Based on the observed test substance precipitation (generally regarded as an overload dose) in the preliminary toxicity assay, the highest concentration initially chosen for the chromosome aberration assay was 2500 g/mL, for all three test conditions. However, for the 4 h activated and the 20 h non-activated test conditions, substantial test substancerelated inhibition of the mitotic activity was observed at >750 g/mL in the 4 h activated test condition, and all concentration levels in the 20 h non-activated test condition. Therefore, cytogenetic evaluations were conducted at 750, 1250, and 2500 g/mL for the 4 h non-activated test condition, at 62.5, 125, and 250 g/mL, for the 4 h activated test condition, and at 25, 50, and 100 g/mL for the 20 h non-activated test condition. The percentage of cells with structural or numerical aberrations in the test substance-treated groups was not signicantly increased above that of the vehicle control at any concentration (p < 0.05, Fishers exact test). Uf-C TiO2 particles did not induce structural or numerical chromosome aberrations in this study. 8. Aquatic toxicity screening results 8.1. Static, acute, 96 h toxicity screening tests with O. mykiss (rainbow trout) Exposure of rainbow trout to a dilution water control and nominal ne TiO2 and uf-C TiO2 particle concentrations of 0.1, 1.0, 10, and 100 mg/L resulted in 0, 0, 0, 10, and 10% or 0, 0, 0, 0, and 0% immobility, respectively, at the end of 96 h.

The 96 h LC50 for both types of TiO2 particles was >100 mg/L based on nominal test concentrations. The results demonstrated that ne or uf-C TiO2 particles each exhibited low concern (Smrchek et al., 1993) for aquatic hazard in unaerated, 96 h, static acute tests. 8.2. Static, acute, 48 h toxicity screening tests with D. magna (daphnids) Exposure of daphnids to the dilution water control and nominal ne or ultrane TiO2 particle concentrations of 0.1, 1.0, 10, and 100 mg/L resulted in 0, 0, 0, 10, and 10% or 0, 0, 0, 10, and 0% immobility, respectively, at the end of 48 h. The D. magna 48 h EC50 values for ne and ultrane TiO2 particles, based on nominal concentrations were >100 mg/L.The results demonstrated that ne and uf-C TiO2 particles exhibited low concern for aquatic hazard in unaerated, 48 h, static acute tests. 8.3. Static, acute, 72 h growth inhibition toxicity screening test to the green algae, P. subcapitata Exposure of algae to nominal concentrations of 0.01, 0.1, 1, 10 and 100 mg/L ne TiO2 particles resulted in 2, 3, 2, 31, and 97% inhibition, respectively, based on healthy cell count compared to the blank control at the end of 72 h; percent inhibition of growth rate was 0, 1, 0, 8, and 66%, respectively. Exposure of algae to nominal concentrations of 0.01, 0.1, 1, 10, and 100 mg/L uf-C TiO2 particles resulted in 19, 6, 11, 15, and 94% inhibition, respectively, based on healthy cell count compared to the blank control at the end of 72 h; percent inhibition of growth rate was 3, 1, 2, 3, and 54%, respectively. Healthy cell counts increased in the blank controls by at least a factor of 16 in 72 h, thereby satisfying the appropriate test acceptance criteria. The algae 72 h EC50 values (95% ducial limits) based on inhibition of growth and healthy cell counts were 16 (1222) mg/L for ne TiO2 particles and 21 (1626) mg/L for uf-C TiO2 particles. The 72 h EC50 values (95% ducial limits) for growth rate based on nominal concentrations and healthy cell counts were 61 (5272) mg/L for ne-TiO2 particles and 87 (8391) mg/L for uf-C TiO2 particles. These results demonstrated that ne and uf-C TiO2 particles exhibited medium concern under TSCA in a 72 h acute test (Smrchek et al., 1993). 9. Discussion Ten different toxicity assays were carried out to evaluate the hazard potential of the ultrane TiO2 (uf-TiO2 )

particle-types, uf-A, uf-B, or uf-C. To summarize the hazard evaluation of uf-TiO2 particle-types, most of the tests demonstrated low hazard potential in mammals or aquatic species following acute exposures to ultrane TiO2 particles. The hazard identication base set is an evolving concept developed to characterize the inherent hazards related to nanomaterial exposures, both in mammalian species, as a representation of human health effects, and in ecological environments using aquatic species. In the absence of accurate exposure determinations for nanomaterial aerosol exposures, information gained from these hazard assessment tests provides a basis for making reasonable and responsible decisions and for taking action in limiting occupational and consumer exposures. The intratracheal instillation method of exposure can be a reliable qualitative screen for assessing the pulmonary toxicity of inhaled particles. This comparison was recently made with different formulations of titanium dioxide particle-types. The results demonstrated that the intratracheal instillation-derived, pulmonary bioassay studies represent an effective preliminary screening tool for inhalation studies with the identical particle-types used in this study (Warheit et al., 2005). The justication for the base set hazard tests rests on the following criteria: (1) potential routes of exposures related to human health effects (i.e., pulmonary, dermal, oral and/or ocular); (2) early screening for potential carcinogenic effects (i.e., utilizing well established mutational and chromosomal aberration screening assays); and (3) assessments of potential toxic effects in representative aquatic organisms (i.e., exposures of rainbow trout, Daphnia, and algae to nanomaterials). It is important to note that the base hazard set of tests described herein does not account for an exhaustive evaluation or full toxicological prole for a given nanomaterial, but rather serves to provide a reasonable balance between an adequate toxicity characterization and a practical strategy for the development of new nanomaterials. Thus, the aim of this strategy is to develop hazard information sufciently robust to guide adequate risk-evaluation processes, commensurate with existing regulatory and voluntary safety standards. With regard to the development of testing strategies to gauge the toxicity of nanomaterials, two notable efforts have recently been reported. To this end, an International Life Sciences Research Foundation/Risk Sciences Institute working group developed the elements of a screening strategy for assessing hazard proles of nanomaterials. According to this working group, the three key elements of the toxicity screening strategy were (1) physicochemical characteristics; (2) in vitro

assays (cellular and non-cellular) and (3) in vivo toxicity assays. The physicochemical properties that were considered to be important for assessing toxic effects of nanomaterials included particle size and size distribution, agglomeration state, shape, crystal structure, chemical composition, surface area, surface chemistry, surface charge and porosity. Suggested in vitro assays were proposed to focus upon specic biological and mechanistic pathways in ways that are not feasible in in vivo tests. Tier I in vivo assays were proposed for pulmonary, oral, skin and injection exposures. Tier I assessments included markers of inammation, oxidant stress, and cell proliferation in portal-of-entry and selected remote organs and tissues. More intensive Tier 2 evaluations for pulmonary exposures were suggested for deposition, translocation, toxicokinetic and biopersistence studies, as well as effects of multiple exposures. Additional Tier 2 studies could include potential effects on the reproductive system, placenta, and fetus; alternative animal models; and mechanistic studies (Oberd orster et al., 2005). More recently, the European Center for Ecotoxicology and Toxicology of Chemicals (ECETOC) sponsored a workshop to develop testing strategies to evaluate the safety of nanomaterials. It brought together scientic and clinical experts from industry, academia, government agencies, research institutes, and nongovernmental organizations. The fundamental questions addressed by the workshop participants were the following: What can we do today? What do we need tomorrow? (ECETOC, 2006). Some of the summary conclusions of the workshop included the following: For assessment of nanoparticle physicochemical characterization, the working denition of nanoparticles was particle size <100 nm in one dimension or <1000 nm to include aggregates and agglomerates. For evaluations of external exposures and metrics appropriate for nanoparticulates, it was considered to be premature to select one form of dose metric (i.e., mass, surface area, or particle number) as the most effective metric. An initial testing step to gauge potency may include an in vitro screening strategy to assess the possible reactivity, inammatory potential and cellular uptake of nanoparticles, however, these assays should be validated using in vivo techniques. (ECETOC, 2006). The hazard identication base set toxicity tests that have been proposed contain only two in vitro

studiesboth of which are genotoxicity screening assays. The remaining studies in the battery of tests are in vivo studies, using mammalian and aquatic organisms. Clearly the development of validated in vitro screens to substitute for in vivo tests would be benecialin terms of costs, timing, and reductions in animal use. Unfortunately, none of the in vivo tests currently utilized in the base set can be accurately replaced by in vitro assays. We have made efforts to improve the predictability of in vitro assays to assess the in vivo pulmonary toxicity of inhaled nanomaterials. The results obtained from a previously reported study utilizing in vitro methodologies were not representative of in vivo pulmonary inammatory or cytotoxic effects for the various particle-types tested and thus were not well correlated. It was concluded that in vitro cellular systems will need to be further developed, standardized and validated in order to provide useful screening data on the relative toxicity of ne and nanoscale particulates (Sayes et al., 2007). In addition to the hazard identication base set methodology described herein, bridging information can provide additional useful information on a nanoparticletype. In this regard, when a material has few specic hazard data, one way to inform decisions about it is to bridge it to a material that has robust hazard data. The two materials may then be entered into a toxicological study, with the well-characterized material serving as a reference particle for the material of interest. This bridging strategy has been utilized in a recent publication and is reviewed in the pulmonary bioassay results herein, wherein toxicity assessments for new rutile-type ultrane TiO2 particle-types were conducted (Warheit et al., 2007). To summarize the ndings reported herein, a hazard identication base set methodology has been developed which incorporates tests for assessing the toxicity of ultrane particle-types or nanomaterials in mammalian and aquatic speciesrelated, in part, to potential routes of exposure. When considering all of the test results, the ndings of most of the toxicity studies with the ultrane TiO2 particle-types tested herein (i.e., uf-A, uf-B, and ufC) demonstrated low hazard potential in mammals and aquatic species following short-term exposures. Acknowledgments This study was supported by DuPont Titanium Technologies. We thank Drs. Brian Coleman, Gerald Kennedy, Jr., Scott Loveless, Gary Whiting, and Scott Frerichs for helpful comments on this manuscript. The following individuals provided invaluable technical

assistance in the conduct of the various studies, Denise Hoban, Elizabeth Wilkinson, Carolyn Lloyd, Lisa Lewis, John Barr, Don Hildabrandt, Susan Munley, Lynn Ford, Dr. John OConnor, Jeffrey Turner, Sr., Terry Lee Sloman, and Dr. Steven R. Frame. References
Brunauer, S., Emmett, P.H., Teller, E., 1938. Adsorption of gases in multimolecular layers. J. American Chem. Soc. 60, 309319. ECETOC, 2006. Testing Strategies to Establish the Safety of Nanomaterials78 November 2005, Barcelona. Workshop Report No. 7. European Centre for Ecotoxicology and Toxicology of Chemicals, Brussels, Belgium. Oberd orster, 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., 2005. A report from the ILSI Research Foundation/Risk Science Institute Nanomaterial Toxicity Screening Working Group. Principles for characterizing the potential human health effects from exposure to nanomaterials: Elements of a screening strategy. Part. Fibre Toxicol. 2:8. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals: (Part 201), 1984. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals: (Part 203), 1992. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals: (Part 473), 1998. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals: (Part 471), 1998. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals Section 4 (Part 425), 2001. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals Section 4 (Part 404), 2002. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals Section 4 (Part 405), 2002. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals: (Part 429), 2002. Organisation for Economic Co-operation and Development (OECD). Guideline for the Testing of Chemicals: (Part 202), 2004. Otwinowski, Z., Minor, W., 1997. Macromolecular crystallography, Part A, vol. 276. Academic Press Inc., San Diego, pp. 307326. Sayes, C.M., Reed, K.L., Warheit, D.B., 2007. Assessing toxicity of ne and nanoparticles: comparing in vitro measurements to in vivo pulmonary toxicity proles. Toxicol. Sci. 97, 163180. Smrchek, J., Clements, R., Morcock, R., Rabert, W., 1993. Assessing ecological hazard under TSCA: Methods and evaluation of data. Environmental Toxicology and Risk Assessment, ASTM STP 1179 (W.G. Landis, JS Hughes, and MA Lewis, Eds.) pp. 2239. American Society of Testing and Materials, Philadelphia. U.S. EPA Health Effects Test Guidelines OPPTS 870. 2400 (1998). U.S. EPA Health Effects Test Guidelines OPPTS 870. 2500 (1998). U.S. EPA Health Effects Test Guidelines OPPTS 870. 5100 (1998). U.S. EPA Health Effects Test Guidelines OPPTS 870. 5375 (1998). U.S. EPA Health Effects Test Guidelines OPPTS 870. 1100 (2002). Warheit, D.B., Carakostas, M.C., Hartsky, M.A., Hansen, J.F., 1991. Development of a short-term inhalation bioassay to assess pulmonary toxicity of inhaled particles: comparisons of pulmonary responses to carbonyl iron and silica. Toxicol. Appl. Pharmacol. 107, 350368.

Warheit, D.B., Hansen, J.F., Yuen, I.S., Kelly, D.P., Snajdr, S., Hartsky, M.A., 1997. Inhalation of high concentrations of low toxicity dusts in rats results in pulmonary and macrophage clearance impairments. Toxicol. Appl. Pharmacol. 145, 1022. Warheit, D.B., Brock, W.J., Lee, K.P., Webb, T.R., Reed, K.L., 2005. Comparative pulmonary toxicity inhalation and instillation and

studies with different TiO2 particle formulations: impact of surface treatments on particle toxicity. Toxicol. Sci. 88, 514524. Warheit, D.B., Webb, T.R., Reed, K.L., Frerichs, S., Sayes, C.M., 2007. Pulmonary toxicity study in rats with three forms of ultraneTiO2 particles: differential responses related to surface properties. Toxicology 230, 90104.