Toxicology Letters 225 (2014) 177–184

Contents lists available at ScienceDirect

Toxicology Letters journal homepage: www.elsevier.com/locate/toxlet

Phosphate-enhanced cytotoxicity of zinc oxide nanoparticles and agglomerates W. Neil Everett a,b , Christina Chern b , Dazhi Sun a , Rebecca E. McMahon b , Xi Zhang a , Wei-Jung A. Chen c , Mariah S. Hahn b,d,∗ , H.-J. Sue a,∗∗ a

Mechanical Engineering, Texas A&M University, College Station, TX 77843, United States Chemical Engineering, Texas A&M University, College Station, TX 77843, United States c Neuroscience and Experimental Therapeutics, Texas A&M Health Science Center College of Medicine, Bryan, TX 77807, United States d Biomedical Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180, United States b

h i g h l i g h t s • • • • •

NIH/3T3 cells were exposed to individualized and agglomerated ZnO nanoparticles. The cytotoxic response in growth media with and without phosphates was assessed. With phosphates present, zinc phosphate formation enhanced cell death. Agglomeration-dependent cell death only existed in the presence of phosphates. Nanomaterial interactions with growth media constituents can be easily overlooked.

a r t i c l e

i n f o

Article history: Received 29 October 2013 Received in revised form 5 December 2013 Accepted 6 December 2013 Available online 18 December 2013 Keywords: Zinc phosphate toxicity Zinc oxide toxicity Nanohazard Culture media byproduct Aggregation

a b s t r a c t Zinc oxide (ZnO) nanoparticles (NPs) have been found to readily react with phosphate ions to form zinc phosphate (Zn3 (PO4 )2 ) crystallites. Because phosphates are ubiquitous in physiological fluids as well as waste water streams, it is important to examine the potential effects that the formation of Zn3 (PO4 )2 crystallites may have on cell viability. Thus, the cytotoxic response of NIH/3T3 fibroblast cells was assessed following 24 h of exposure to ZnO NPs suspended in media with and without the standard phosphate salt supplement. Both particle dosage and size have been shown to impact the cytotoxic effects of ZnO NPs, so doses ranging from 5 to 50 ␮g/mL were examined and agglomerate size effects were investigated by using the bioinert amphiphilic polymer polyvinylpyrrolidone (PVP) to generate water-soluble ZnO ranging from individually dispersed 4 nm NPs up to micron-sized agglomerates. Cell metabolic activity measures indicated that the presence of phosphate in the suspension media can led to significantly reduced cell viability at all agglomerate sizes and at lower ZnO dosages. In addition, a reduction in cell viability was observed when agglomerate size was decreased, but only in the phosphate-containing media. These metabolic activity results were reflected in separate measures of cell death via the lactate dehydrogenase assay. Our results suggest that, while higher doses of water-soluble ZnO NPs are cytotoxic, the presence of phosphates in the surrounding fluid can lead to significantly elevated levels of cell death at lower ZnO NP doses. Moreover, the extent of this death can potentially be modulated or offset by tuning the agglomerate size. These findings underscore the importance of understanding how nanoscale materials can interact with the components of surrounding fluids so that potential adverse effects of such interactions can be controlled. © 2013 Elsevier Ireland Ltd. All rights reserved.

Abbreviations: NP, nanoparticle; PBS, phosphate buffered saline; PVP, polyvinylpyrrolidone; PS, penicillin streptomycin; PSG, penicillin streptomycin l-glutamine; FBS, fetal bovine serum; DMEM+phos , Dulbecco’s modified Eagle’s medium (DMEM) with phosphates; DMEM−phos , DMEM without phosphates; XRD, X-ray diffraction; HRTEM, high-resolution transmission electron microscopy; AAS, atomic absorption spectrometry; DLS, dynamic light scattering; ICP-MS, inductively coupled plasma-mass spectrometry. ∗ Corresponding author. Present address: Biomedical Engineering, Rensselaer Polytechnic Institute, Troy, NY 12180, United States. Tel.: +1 518 276 2236. ∗∗ Corresponding author. Tel.: +1 979 845 5024. E-mail addresses: [email protected] (M.S. Hahn), [email protected] (H.-J. Sue). 0378-4274/$ – see front matter © 2013 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.toxlet.2013.12.005

178

W.N. Everett et al. / Toxicology Letters 225 (2014) 177–184

1. Introduction

2. Materials and methods

The rapid expansion of nanotechnology research in recent years has led to an increased awareness of how advancements could potentially elicit both positive and negative consequences for the environment and human health. Our current level of exposure to nanomaterials, in the products we buy and industrial waste we produce, is expected to increase significantly in the near future. As such, researchers and public officials have started devoting more attention and resources toward understanding nanomaterialrelated health hazards and how to establish the proper framework from which generalizations can be drawn that relate the danger of a nanoscale material to its physical properties (e.g., size, shape, composition, surface chemistry) (Barnard, 2006; Nel et al., 2006). Nanometer-sized ZnO falls under the classification of “potential nanohazard” due to its use in consumer products such as cosmetics and sunscreens (Gulson et al., 2010; Sadrieh et al., 2010), its introduction into the environment through wastewater and other processing/disposal pathways (Adams et al., 2006; Limbach et al., 2008), and the potential for inhalation (Brown, 1988). The health and environmental safety of ZnO nanoparticles has thus far been assessed by examining its penetration into human skin (Gamer et al., 2006; Gulson et al., 2010; Nohynek et al., 2007; Sadrieh et al., 2010) and its toxicity in bacterial (Adams et al., 2006; Jin et al., 2009; Reddy et al., 2007; Sawai et al., 1995) and fungal strains (Sawai and Yoshikawa, 2004), various mammalian cell lines (Brunner et al., 2006; Deng et al., 2009; Gojova et al., 2007; Jeng and Swanson, 2006; Reddy et al., 2007; Xia et al., 2006, 2008; Yin et al., 2010), and in whole-animal models (Xia et al., 2011). These studies have primarily focused on the toxicological effects of particle size, ZnO dose dependence, and nanoparticle (NP) dissolution resulting in Zn2+ liberation into the growth media. Other work (Cho et al., 2011; Limbach et al., 2005) has looked more broadly at the role of NP agglomeration and associated sedimentation rates on the toxicological response specific to the cell culture environment. Importantly, ZnO has been demonstrated (Jung et al., 2009; Reed et al., 2012) to convert to zinc phosphate (Zn3 (PO4 )2 ) crystallites in the presence of phosphate salts. As phosphates are ubiquitous in physiological fluids as well as wastewater streams, it is important to examine the potential effects that the formation of Zn3 (PO4 )2 crystallites may have on cell viability with varying agglomerate size and dosage. Toward this goal, the cytotoxicity of individual water-soluble ZnO NPs (with a nominal diameter of ∼4 nm) and ZnO NP agglomerates (ranging from ∼8 nm to several microns in size) in cell culture media with and without phosphates was examined. ZnO NPs in agglomerated states (Limbach et al., 2005) (as opposed to studying the toxicity of ZnO NPs with different nominal particle diameters (Adams et al., 2006; Deng et al., 2009)) were examined to find potential agglomerate size effects. To control agglomerate size, NPs were sterically stabilized (Lafuma et al., 1991) using polyvinylpyrrolidone (PVP), an amphiphilic polymer with a small molecular weight that is considered to be non-toxic in vivo (even at moderately high concentrations (Bergfeld et al., 1998; Robinson, 1990)) and is used for a variety of applications in areas such as medicine, pharmaceuticals, and food and cosmetics production. By adjusting the amount of soluble PVP available to physisorb and stabilize the ZnO, agglomerate size was adjusted from the micron scale down to individually dispersed NPs. Following 24 h of exposure to various ZnO dispersions in media with or without phosphates, cell viability was assessed by quantifying cell metabolic activity and measuring the release of cytosolic lactate dehydrogenase. These relatively short-term cell culture studies were selected to avoid biasing viability due to cells being unable to replenish their internal phosphate reserves in the phosphate-free media.

2.1. Materials Zinc acetate, potassium hydroxide, sodium chloride, methanol, hexane, isopropanol, 4-(2-hydroxyethyl)piperazine1-ethanesulfonic acid (HEPES), phosphate buffered saline (PBS; 1×, pH 7.4), and 8 kDa polyvinylpyrrolidone (PVP) were purchased from Sigma Aldrich and used without further purification. Penicillin streptomycin (PS) solution, penicillin streptomycin lglutamine (PSG) solution, and DMEM growth media were obtained from Mediatech (Manassas, VA). Heat-deactivated fetal bovine serum (FBS) was acquired from Hyclone Laboratories (Logan, UT). 2.2. Media preparation Prior to media preparation, heat-inactivated FBS was dialyzed against HEPES buffered saline (10 mM HEPES, 150 mM NaCl) with a 0.5–1.0 kDa membrane (Spectrum Laboratories) to remove phosphates present within the FBS. This dialyzed FBS was used for both media preparations (with and without phosphates), so that any loss of serum constituents would have an equal effect on both groups. DMEM with l-glutamine (cat# 10-013-CM) was supplemented with 10% dialyzed FBS and 1% PS and used as the “with-phosphates” medium (hereafter termed DMEM+phos ). DMEM without phosphates or l-glutamine (cat# 17-206-C1) was prepared with 10% dialyzed FBS and 1% PSG and used as the “withoutphosphates” growth medium (hereafter termed DMEM−phos ). For preparation of ZnO NP dilutions, both fully supplemented media types were concentrated to 4× using a centrivap and then filtered using 0.22 ␮m syringe filters (Millipore). ICP-MS analyses found that the concentration of Zn2+ in both DMEM preparations was below the detectable limit of the instrument, and no phosphorous was detected in DMEM−phos . 2.3. ZnO nanoparticle preparation Colloidal ZnO NPs, ∼4 nm in diameter, were synthesized by hydrolyzing zinc acetate dihydrate in a potassium hydroxide/methanol solution in the presence of PVP (an amphiphilic polymer) at a weight ratio of 3:5 (PVP:Zn2+ ); for a more detailed description of the procedure, please refer elsewhere (Guo et al., 2000; Sun et al., 2007). Excess free ions were removed from the ZnO dispersion through repeated (i) destabilization of the NPs with hexane, (ii) removal of the supernatant, and then (iii) redispersion in methanol under sonication (Sun et al., 2007). The ZnO particles were then concentrated by evaporating methanol from the purified ZnO with a rotovap until a near gel-like state was achieved. Following the initial ZnO preparation, colloidal ZnO was further modified under ultrasonication at 25 ◦ C through physisorption of PVP dissolved in deionized water (DI water) at three pre-determined weight ratios (PVP:ZnO = 300:1, 200:1, 100:1) based on surface area considerations (∼265 m2 /g fully dispersed), yielding ZnO agglomerates varying from several microns in size down to individually dispersed NPs. The ZnO stock dispersions were then diluted with DI water and 4× media to yield a final media concentration of 1×, and once added to the cell culture wells, resulting ZnO dosages of 50, 40, 30, 20, 10, or 5 ␮g/mL. 2.4. ZnO nanoparticle characterization ZnO NPs were characterized with a variety of methods to determine particle size, structure, and spectroscopic properties and to confirm that the PVP present during synthesis and ultrasonication had simply physisorbed to the NP surface. The crystalline structure of the colloidal ZnO was analyzed by X-ray diffraction (XRD; D8

W.N. Everett et al. / Toxicology Letters 225 (2014) 177–184

Advance, Bruker, WI), which was carried out using a Cu-K␣ source ˚ on a dried ZnO sample. In addition, XRD patterns were ( = 1.54 A) used to give an indirect, rough estimate of average nanocrystallite size through the Scherrer formula, d = K/ˇ cos , where K is the Scherrer constant and assumed to be 0.89 (Langford and Wilson, 1978) and ˇ is the line broadening at full width half at maximum in radians. Aside from assuming a monodisperse size distribution, one caveat to mention when using the Scherrer formula to estimate crystalline NP diameter from XRD data is that the diffraction peaks are required to be linked to a family of crystallographic planes in a size-limited crystal, which is most likely not the case here. While more rigorous estimates can be made from XRD data (Hall et al., 2000), this rough estimate is adequate for the scope of this study and is supplemented by high-resolution transmission electron microscopy (HRTEM) measurements. HRTEM (JEOL 2010 operated at 200 keV) was conducted in order to directly measure NP size, where samples were prepared by evaporating a droplet of the ZnO dispersion onto a 400-mesh, carbon-coated copper grid. Average NP diameter was found by analyzing multiple HRTEM images and sizing ∼200 NPs serially using ImageJ. Atomic absorption spectrometry (AAS; Varian SpectrAA300) measurements were performed on the ZnO NP dispersion following each rinsing step to track the removal of free K+ (rationale for this approach described in Section 3). UV–vis spectra (USB2000-DT Spectrometer, Ocean Optics, FL) were collected for unpurified ZnO in methanol, purified ZnO in methanol, and 200:1 PVP-modified ZnO in water to observe any influence the PVP, ultrasonication, or purification steps may have had on the spectral properties of the ZnO NPs, as an independent verification of maintained structure and size. Dynamic light scattering (DLS ZetaPALS, Brookhaven Instruments Corp., NY) was used to measure particle/agglomerate zeta potentials and average ZnO agglomerate sizes ranging from hundreds of nanometers down to individually dispersed NPs, thereby providing an ex situ estimate of agglomerate size and surface charge prior to being applied to the cell culture environment. Refer to Supplementary Information for additional DLS experimental details

179

(Synergy HT, Biotek). The degree of fluorescence served as an indicator of overall cell metabolic activity and hence of cell viability in each culture well. All NP-based background controls were indistinguishable from pure DMEM+phos or DMEM−phos controls. An independent evaluation of cell death was completed by analyzing the release of cytosolic lactate dehydrogenase (LDH) from damaged or dying cells into the media using a Cytotoxicity Detection Kit per manufacturer instructions (Roche Applied Science, IN). The colorimetric assay was conducted in 96-well plates using a plate reader, and samples were run in triplicate. Control values are based on LDH concentrations from cells cultured in wells for 24 h that had not been exposed to ZnO, with the “low control” representing the baseline cell death and the “high control” value found by lysing all cells within the control wells. 2.6. Inductively coupled plasma-mass spectrometry (ICP-MS) Analyses were performed on an HP 4500 system to quantify zinc levels in both types of growth media, ZnO dispersions, and ZnO/growth media mixtures. Internal and external calibration standards were run for all samples tested. Each of the three ZnO dispersions (300:1, 200:1, 100:1, all at 0.1 mg/mL ZnO) was mixed in equal volume with fully supplemented DMEM+phos or DMEM−phos . The resulting samples were centrifuged at 10,000 g for 10 min, and then supernatant from each tube was collected and diluted with DI water by 50× for ICP-MS analyses. 2.7. Statistics Data are reported as mean ± standard deviation. Comparison of sample means was performed using two-way ANOVA. When the pvalue associated with the main effect of an independent variable fell below 0.05, pair-wise comparisons between specific formulations were then conducted using Tukey’s post hoc tests (two-tailed, SPSS software), p < 0.05. 3. Results and discussion

2.5. Cell culture, ZnO exposure, metabolic activity, and lactate dehydrogenase measurements

3.1. ZnO-PVP dispersion characteristics prior to exposure to media serum

NIH/3T3 fibroblast cells were collected and resuspended in either 1× fully supplemented DMEM+phos or DMEM−phos media. Cells were seeded at ∼25,000 cells/cm2 in 96-well plates (Costar), and NP solutions were applied when cells reached at least 90% confluence. The ZnO solutions in 1× DMEM+phos or DMEM−phos were then added to cell culture wells at final ZnO concentrations of 50, 40, 30, 20, 10, or 5 ␮g/mL. For the metabolic activity assay, two types of controls were performed: (i) “cell control” consisting of cells cultured in DMEM+phos or DMEM−phos (but no NPs) and (ii) “media background” controls consisting of either culture media or NP-media solutions only (i.e. no cells). NP background controls were run in addition to media background controls to determine potential absorbance/fluorescence quenching offsets due to the presence of NPs and interactions with assay reagents, but no discernible effect was measured (see Supplementary Information). All sample types and controls were performed in triplicate. Following 24 h of exposure to ZnO NPs, media was gently removed and fresh DMEM+phos or DMEM−phos media supplemented with C12-resazurin (Vybrant Metabolic Assay Kit, Invitrogen) was then added to each culture well, including control wells. After 3 h of incubation, a portion of the media from each culture well was collected and its fluorescence measured at an excitation/emission setting of 560 nm/590 nm on a spectrophotometric plate reader

Before conducting the cell culture studies, it was important to fully characterize our synthesized NPs and NP dispersions. X-ray diffraction (XRD) patterns given in Fig. 1(A) reveal the expected Wurtzite structure for our ZnO NPs both with and without PVP present during synthesis. The chosen PVP concentration resulted in spherical ZnO NPs, and the slightly more pronounced 0 0 2 XRD peak (see Supplementary Information Fig. S1) indicates that the average crystallite size of ZnO grown without PVP is marginally larger than ZnO NPs grown in the presence of PVP. As such, the amount of PVP present during synthesis did not significantly affect ZnO NP crystal growth. Using the Scherrer formula, the average NP size estimated from the 1 0 2 , 1 1 0 , and 1 0 3 peaks was determined to be ∼4 nm, consistent with our high-resolution transmission electron microscopy (HRTEM) observations (see inset in Fig. 1(A)) that estimated the particle size to be 4.1 ± 0.5 nm. Furthermore, modification of ZnO with PVP resulted in a similar UV/vis adsorption peak compared with unmodified ZnO dispersions, as shown in Fig. 1(B). These two results confirm that ZnO NPs can be both synthesized in the presence of PVP and dispersed in an aqueous medium under ultrasonication using PVP without degradation or loss of the spectral and structural properties commonly associated with ZnO NPs of this size. Because the aim of this study was to isolate the cytotoxicological impact of ZnO NPs and Zn3 (PO4 )2 particles/agglomerates, it was

180

W.N. Everett et al. / Toxicology Letters 225 (2014) 177–184

Fig. 2. Dynamic light scattering data showing particle/agglomerate size, d, distributions for (A) PVP only, (B) 300:1 PVP:ZnO, (C) 200:1 PVP:ZnO, (D) supernatant from 100:1 PVP:ZnO, (E) 10% FBS, (F) 10% FBS added to 50 ␮g/mL 200:1 PVP:ZnO, and (G) 10% FBS added to 50 ␮g/mL 300:1 PVP:ZnO; all dispersions analyzed in a 0.15 M NaCl background.

Fig. 1. (A) X-ray diffraction patterns of ZnO NPs with (blue) and without (black) the PVP additive present during synthesis; data offset for clarity. Inset shows an HRTEM image of a single representative ZnO NP. The white dotted line has been added as a guide to the eye, and the scale bar is 2 nm. HRTEM analysis gave an average NP size of 4.1 ± 0.5 nm. (B) UV/vis spectra of three ZnO samples, showing a small shift in the absorption peaks after purification steps and PVP modification; spectra intensity offset for clarity. (C). AAS measurements of the K+ (aq) concentration in the as-synthesized ZnO dispersion following each rinsing step. (For interpretation of references to color in this figure legend, the reader is referred to the web version of this article.)

critical that all other constituents within the prepared ZnO dispersions be diluted to reduce the concentrations of contaminant ions to negligible levels. More specifically, any free ions such as K+ (aq), OAc− (aq), or Zn2+ (aq) remaining after ZnO synthesis (from hydrolysis of zinc acetate dihydrate in a potassium hydroxide/methanol solution) could significantly exaggerate the apparent cytotoxicity of our ZnO NP dispersions. Although an essential cell nutrient, Zn2+ (aq) has been shown, for example, to be toxic to mammalian cell lines at concentrations ranging from 50 ␮M (Steinebach and Wolterbeek, 1993) to 125 ␮M (Borovansky and Riley, 1989). Here, we use [K+ (aq)] as our gauge of contaminant ion concentration, because (i) K+ (aq) is assumed to be at a much higher starting concentration than Zn2+ (aq) in our as-synthesized ZnO dispersion and (ii) K+ (aq) measures are less prone to artifact than Zn2+ (aq) in the present context. AAS data taken after each rinsing step (Fig. 1(C)) show that [K+ (aq)] within the as-synthesized ZnO NP dispersion dropped from ∼80 mM to below 0.2 mM after four rinses. The concentrations of residual free ions in the NP dispersions were further reduced during the dilution preparation steps by more than 1000×. Thus, baseline “contaminant” ions such as Zn2+ (aq)

and K+ (aq) resulting from NP synthesis itself were estimated to be present in the nanomolar range within the cell culture wells upon dosing. Presented in Fig. 2(A)–(D) are dynamic light scattering (DLS) size distributions of particle/agglomerate diameter, d, for PVP and the three ZnO-PVP dispersions used in our cell studies. The size distribution for water-solvated 8 kDa PVP shows a primary peak at 1.6 nm, corresponding closely with the radius of gyration value of 1.38 nm reported in the literature (Sato et al., 1998). The 300:1 PVP:ZnO sample gave a single size population with a most probable peak around 4.5 nm, which is both consistent with our XRD and HRTEM estimates of average NP diameter and indicative of complete ZnO NP dispersion. In the case of the 200:1 PVP:ZnO dispersion, the smallest size population was broadened and shifted to a slightly larger diameter relative to the 300:1 sample, and a measureable secondary agglomerate population appeared around 170 nm. This change in the population at the smallest size range could be due to either the formation of doublets and triplets or a sizing bias induced by the d6 scattering dependence (via the Rayleigh approximation) in DLS of the larger agglomerates (Pecora, 1985). Regardless of the cause, it is clear that some degree of agglomeration exists for the 200:1 case that is the result of fewer PVP molecules being available to adsorb and stabilize the ZnO NPs in suspension. Supernatant from the 100:1 dispersion was collected for DLS analysis following 10 min of centrifugation at 10,000 × g. The micron-sized agglomerates in the pre-centrifuged dispersion were not directly measurable with DLS due to their sedimentation rate and also the scattering from larger agglomerates considerably obscured detection of scattering from smaller agglomerates. While optical microscopy revealed that the average agglomerate size for the 100:1 dispersion was between 2 and 10 ␮m, the supernatant contained small ZnO NP clusters and possibly individual NPs (see Fig. 2(D)) at a concentration of 5 ␮m. The dispersion did not appear turbid when NPs were added to the FBS, and no precipitate formed. Thus the resulting agglomeration likely takes the form of low-density fractal structures created by (i) protein flocculation via NP bridging (Wong et al., 1992) (as opposed to NP flocculation that would result in visible turbidity), (ii) depletion flocculation of proteins by excess non-adsorbed PVP, and/or (iii) depletion flocculation of protein agglomerates by NPs (Snowden et al., 1991; Walz and Sharma, 1994). Combinations of these mechanisms are also possible and can occur in a cascading manner. Similar DLS observations of agglomeration have been reported for dispersions where NPs have been added to serum protein solutions (Deng et al., 2009; Murdock et al., 2008), although direct comparisons would require the same media composition and NPs of approximately the same size and surface functionalization. The size distribution data for the 300:1 dispersion added to a 10% FBS solution (Fig. 2(G)) was similar to that of the 200:1 dispersion, indicative of the same agglomeration mechanisms occurring. However, there was a downward shift in the size, presumably due to 1.5× more PVP being present within the mixture to aid in stabilization. This same downward shift in NP size with increased PVP levels was consistent with that observed in the absence of FBS. We were unable to obtain meaningful DLS data from ZnO dispersions mixed into DMEM+phos , because the scattering signal was unstable, with agglomeration appearing to continue indefinitely. That said, DLS DMEM+phos results qualitatively agreed with the above FBS-saline results. Specifically, the rate of increase in agglomerate size in DMEM+phos was smaller for the 300:1 dispersion relative to the 200:1 dispersion and may result from the difference in availability of free PVP in solution that can act to stabilize the agglomerating precipitate as it forms by altering particle–particle association kinetics and possibly serve as a passivation layer that hinders dissolution and re-precipitation. When the ZnO–PVP NP dispersions were each added to pure PBS, the mixtures became turbid almost instantaneously due to the formation of a waterinsoluble precipitate. The precipitate from the ZnO–PBS dispersion comprised platelet-like crystallites such as those shown in the inset in Fig. 5. These crystallites were collected and rinsed by repeated centrifugation and re-dispersion in DI water. As illustrated in Fig. 3, XRD showed the precipitate to be Zn3 (PO4 )2 , consistent with previous observations of zinc-containing compounds, such as ZnO, spontaneously forming complexes with phosphates throughout a broad range of pH values (Yamabi et al., 2005) and also when added to either DMEM or RPMI growth media (Reed et al., 2012).

Fig. 3. (Black line) XRD pattern of water-insoluble precipitate formed from the addition of 200:1 PVP-stabilized ZnO to PBS and (red line) reference XRD pattern for Zn3 (PO4 )2 ·4H2 O (RRUFF ID# R050254.1). Reference data offset for clarity. Inset is a magnified view of a portion of the overlaid data (from main plot) showing the high degree of peak alignment. (For interpretation of references to color in this figure legend, the reader is referred to the web version of this article.)

To supplement the quantitative DLS analyses, ICP-MS was used to indirectly observe how the presence of phosphates and serum proteins altered the agglomeration characteristics of the dispersion. Specifically, differences in zinc concentrations from supernatants of centrifuged media–NP mixtures were measured with ICP-MS relative to corresponding non-centrifuged media–NP mixtures. These analyses yielded measures of “dispersed” zinc—that is, Zn2+ (aq) resulting from particle dissolution or Zn2+ (s) from Zn3 (PO4 )2 and/or ZnO particles too small to be sedimented out within 10 min at 10,000 × g. As shown in Fig. 4, the drop in zinc within NP-DMEM−phos mixtures following centrifugation is approximately the same as for the NP–water control samples, for all three PVP–ZnO dispersions. Conversely, the NP–DMEM+phos mixtures were associated with a significantly increased drop in zinc levels following centrifugation relative to the corresponding NP–water controls for the 200:1 and 300:1 PVP–ZnO mixtures. These results are consistent with the formation of Zn3 (PO4 )2 , which has a lower degree of dispersion stability and was eliminated more readily from the supernatant during centrifugation. Cumulatively, the observations reported here are further substantiated by another study that found a significant increase in ZnO particle/agglomerate size when the particles were added to either standard DMEM or PBS (growth from an initial size of 10 nm up to 320 nm) (Deng et al., 2009).

Fig. 4. Percent change in [Zn2+ ] within supernatants of centrifuged DMEM/NP mixtures relative to non-centrifuged PVP–ZnO dispersions. Control values represent % change in [Zn2+ ] for centrifuged PVP–ZnO dispersions in DI water relative to non-centrifuged dispersions in DI water at neutral pH.

182

W.N. Everett et al. / Toxicology Letters 225 (2014) 177–184

3.3. Cytotoxicity of ZnO dispersions Following NP characterization, the cytotoxic response of NIH/3T3 fibroblast cells was assessed after 24 h of exposure to ZnO NPs suspended in DMEM+phos and DMEM−phos . Because both particle dosage and size have been shown previously to impact the cytotoxic effects of ZnO NPs, doses ranging from 5 to 50 ␮g/mL were examined. As with Brunner et al., we chose to use NIH/3T3 fibroblasts due to their extensive use in earlier toxicological studies, thereby providing a good basis for comparison with previous work. In addition, mouse NIH/3T3 cells have been show to demonstrate a sensitivity to ZnO NPs that is similar to human MSTO-211H cells (Brunner et al., 2006). As shown in Fig. 5A and B, cell viability and metabolic activity were found to decline markedly at ZnO concentrations ≥20 ␮g/mL for all three PVP–ZnO dispersions tested in both DMEM media preparations. The dose-dependent cytotoxic response found here for NIH/3T3 cells is consistent with results from previous studies of ZnO NP toxicity measured against bacterial strains (Adams et al., 2006; Jin et al., 2009) and mammalian cell lines (Brunner et al., 2006; Gojova et al., 2007; Jeng and Swanson, 2006; Xia et al., 2008). However, while the metabolic activity of cells exposed to PVP:ZnO

NPs at 5 and 10 ␮g/mL in DMEM+phos was reduced relative to the corresponding “cell control” (i.e., no NPs), no statistically significant decrease in overall viability was observed for cells in DMEM−phos at the same PVP:ZnO dosages (Fig. 5B). Furthermore, cell metabolic activity in DMEM+phos was dependent on initial agglomerate size within the PVP–ZnO dispersion at ZnO dosages of 5 and 10 ␮g/mL, with individually dispersed ZnO NPs (300:1 dispersion) eliciting the most toxic response and ZnO in a highly agglomerated state (100:1 dispersion) leading to significantly lower levels of toxicity. Conversely, no size-dependent trend was observed for the DMEM−phos metabolic activity data at ZnO dosages of 5 and 10 ␮g/mL. The lack of apparent size-dependent effects at NP concentrations of 20 ␮g/mL in both culture media was likely a result of what is essentially complete cell death observed at this ZnO dosage, as can be seen by comparing the metabolic activity reads to the media (no cell) control. To rule out potential effects of variable PVP levels on the observed DMEM+phos cell viability data, controls were performed with pure PVP without ZnO NPs (see Supplementary Information). These results indicated that the maximum possible PVP concentration within the cell cultures (i.e., 1.5% total PVP content, assuming complete desorption from the surface of all NPs within the 300:1 dispersion at the maximum ZnO dosage) did not

Fig. 5. (A) Low-magnification optical microscopy images of NIH/3T3 cells cultured in the two media preparations (DMEM+phos , DMEM−phos ) following 24 h of exposure to 300:1 PVP:ZnO at different ZnO dosages. At ZnO concentrations ≤10 ␮g/mL, the cells formed confluent sheets. Beyond 10 ␮g/mL, the cytotoxic effects of higher ZnO doses became apparent in both media types, with cells clustering and peeling away from the culture surface. When different PVP:ZnO ratios were examined at the lower ZnO dosages, differences in cell viability with media type were observed (B, C). (B) Metabolic activity measures of viability for cells cultured in each media type for the different PVP:ZnO dispersions (300:1, 200:1, 100:1) at three different ZnO dosages. Inset is an SEM image of the resulting zinc phosphate precipitate from our 100:1 dispersion added to DMEM+phos ; scale bar = 100 nm. (C) Cytotoxicity measures based on LDH release from apoptotic or necrotic cells for the same conditions described above. Insets show magnified data from the main plots for 5 and 10 ␮g/mL ZnO dosages. In all plots, error bars denote standard deviation of at least three independent experimental runs. ∗ Indicates significant difference with corresponding 300:1 result (p < 0.05). # Indicates significant difference with corresponding DMEM+phos result (p < 0.05)

W.N. Everett et al. / Toxicology Letters 225 (2014) 177–184

result in an ancillary cytotoxicity, which is consistent with previous reports (Bergfeld et al., 1998; Robinson, 1990; Wang et al., 2003). Fig. 5C shows an independent assessment of the cytotoxic response, as quantified by assaying levels of LDH released into the culture media from apoptotic or necrotic cells. In this assay, increased levels of measured LDH activity correspond to increasing levels of cell death. The following trends were observed: (i) a marked increase in cell death (increased LDH levels) at 20 ␮g/mL ZnO in both media types relative to lower ZnO concentrations, (ii) increased cell death in the PVP:ZnO dispersions at 5 and 10 ␮g/mL in the DMEM+phos as compared to the equivalent dispersions in DMEM−phos ; and (iii) decreased cell death with increased agglomerate size was observed for 5 ␮g/mL ZnO dosages in the DMEM+phos media but not in the DMEM−phos media. These LDH results are consistent with the metabolic assay results in showing that increased environmental phosphates lead to increased ZnO toxicity. The LDH data also indicate an agglomerate size effect at lower ZnO dosages, although this effect is more subtle and less consistent between assays. In comparing the metabolic activity and LDH data, it is important to note that the LDH assay is complicated by an inherent assay artifact: LDH released from a cell has a half-life of approximately 9 h at 37◦ C. Thus, toxicity resulting in an average time of cell death that significantly precedes the assay endpoint (24 h in this study) would be underestimated by this assay. The apparent absence of a size-dependent trend in cytotoxicity for NPs in DMEM−phos suggests that the interaction of media phosphates with the NPs may underlie at least part of the size-dependent NP toxicity observed in previous studies. Indeed, TEM images of NPs from at least one paper (Deng et al., 2009) indicate that Zn3 (PO4 )2 crystallites may have been unintentionally evaluated previously. The mechanism(s) by which the presence of phosphates impact(s) cell viability remain to be elucidated, but it seems probable that they relate either directly or indirectly to the formation of Zn3 (PO4 )2 in ZnO NP dispersions exposed to phosphates. ZnO NP uptake via a Trojan Horse-type mechanism (Brunner et al., 2006; Limbach et al., 2007) could explain our observed differences in agglomerate-size dependent toxicity as well as overall viability in both media types. Namely, conversion of ZnO to Zn3 (PO4 )2 may substantially alter the surface properties of the particles (e.g., surface charge, hydrophobicity, shape), affecting the interaction with the cell membrane (Barnard, 2006; Nel et al., 2006) and the ability of a cell to phagocytize a material, which is largely a protein-mediated active process (Slowing et al., 2006; Unfried et al., 2007). However, other mechanisms are also possible. For instance, the dissolution of ZnO NPs, resulting in liberated Zn2+ , has been demonstrated to be a primary cause of toxicity for a similar set of cell culture conditions, wherein the mechanism of Zn2+ -mediated cell death is suggested to be a complex cascade of events that may involve processes such as reactive oxygen species (ROS) production (Donaldson et al., 2003; Nel et al., 2006; Xia et al., 2008), disruption of mitochondrial processes (Brown et al., 2000), secondary Zn2+ elution by metallothioneins (Frazzini et al., 2006; Sensi et al., 1999), disruption of the glycolytic pathway (Dineley et al., 2003; Gazaryan et al., 2007), and the release of cytochrome c protein following mitochondrial membrane damage (Jiang et al., 2001; Wudarczyk et al., 1999). Measured levels of Zn2+ released into water from ZnO NPs (Derfus et al., 2004; Kirchner et al., 2005; Xia et al., 2008) larger in diameter than those used here indicate that toxic concentrations of Zn2+ could be reached in the ZnO dosing range and for the incubation period we tested. Furthermore, Xia et al. found a considerable increase in the rate and magnitude of Zn2+ (aq) release when ZnO was dispersed in DMEM containing phosphate at the same phosphate concentration (∼0.9 mM) used in our DMEM+phos media versus DI water (Xia et al., 2011). Separate measurements of Zn2+ (aq) levels have also shown that that complete dissolution of 70 nm ZnO NPs at 10 ␮g/mL occurs within minutes in DMEM

183

containing phosphate and that the extent of dissolution appears to be unaffected by the addition of serum proteins (Reed et al., 2012). Given the present data, it is therefore possible that the phosphate normally present in DMEM increases the rate and degree of Zn2+ release from ZnO nanoparticles, either via the partial dissociation of formed Zn3 (PO4 )2 or via another mechanism. This increased Zn2+ elution would increase apparent NP toxicity relative to that observed in phosphate-free media. Our agglomeration-dependent outcomes in DMEM+phos could also potentially be explained by differences in the Zn2+ elution rate and/or magnitude brought about varying the total media–NP surface area of interaction (i.e., agglomerated particles have less direct interaction with the surrounding media due to more contact points with surrounding NPs within the agglomerate). Any single mechanism or multiple mechanisms working in concert may explain our agglomeration- and phosphate-dependent cytotoxicity results. Future studies will focus on determining the mechanism(s) by which environmental phosphates impact cell viability. 4. Conclusions Our results suggest the presence of phosphates in the surrounding fluid can lead to significantly elevated levels of cell death at lower ZnO NP doses. Moreover, the extent of this death can potentially be modulated or offset by tuning the degree of agglomeration. These findings underscore the importance of understanding how nanoscale materials can interact with the various components of fluids so that potential adverse effects from such interactions can be predicted and possibly controlled. Conflict of interest statement The authors declare that there are no conflicts of interest. Acknowledgments This work was supported by a Texas Advanced Technology Program grant (No. 000512-0311-2003), Kaneka Corporation, and Texas Engineering Experimental Station funds. We would like to thank: Dan Shantz and Michael Bevan for ZetaPals DLS instrument usage; Alejandra Rivas-Cardona, Dany Munoz-Pinto, and Andrew M. Smith for their assistance with experiments; and the Texas A&M Trace Element Research Laboratory with their assistance with ICPMS measurements. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.toxlet.2013.12.005. References Adams, L.K., Lyon, D.Y., Alvarez, P.J.J., 2006. Comparative eco-toxicity of nanoscale TiO2 , SiO2 , and ZnO water suspensions. Water Research 40, 3527–3532. Barnard, A.S., 2006. Nanohazards: knowledge is our first defence. Nature Materials 5, 245–248. Bergfeld, W.F., Belsito, D.V., Carlton, W.W., Klaassen, C.D., Schroeter, A.L., Shank, R.C., Slaga, T.J., 1998. Final report on the safety assessment of polyvinylpyrrolidone (PVP). International Journal of Toxicology 17, 95–130. Borovansky, J., Riley, P.A., 1989. Cytotoxicity of zinc in vitro. Chemico-Biological Interactions 69, 279–291. Brown, A.M., Kristal, B.S., Effron, M.S., Shestopalov, A.I., Ullucci, P.A., Sheu, K.F.R., Blass, J.P., Cooper, A.J.L., 2000. Zn2+ inhibits alpha-ketoglutarate-stimulated mitochondrial respiration and the isolated alpha-ketoglutarate dehydrogenase complex. Journal of Biological Chemistry 275, 13441–13447. Brown, J.J.L., 1988. Zinc fume fever. British Journal of Radiology 61, 327–329.

184

W.N. Everett et al. / Toxicology Letters 225 (2014) 177–184

Brunner, T.J., Wick, P., Manser, P., Spohn, P., Grass, R.N., Limbach, L.K., Bruinink, A., Stark, W.J., 2006. In vitro cytotoxicity of oxide nanoparticles: comparison to asbestos, silica, and the effect of particle solubility. Environmental Science & Technology 40, 4374–4381. Cho, E.C., Zhang, Q., Xia, Y.N., 2011. The effect of sedimentation and diffusion on cellular uptake of gold nanoparticles. Nature Nanotechnology 6, 385–391. Deng, X.Y., Luan, Q.X., Chen, W.T., Wang, Y.L., Wu, M.H., Zhang, H.J., Jiao, Z., 2009. Nanosized zinc oxide particles induce neural stem cell apoptosis. Nanotechnology 20. Derfus, A.M., Chan, W.C.W., Bhatia, S.N., 2004. Probing the cytotoxicity of semiconductor quantum dots. Nano Letters 4, 11–18. Dineley, K.E., Votyakova, T.V., Reynolds, I.J., 2003. Zinc inhibition of cellular energy production: implications for mitochondria and neurodegeneration. Journal of Neurochemistry 85, 563–570. Donaldson, K., Stone, V., Borm, P.J.A., Jimenez, L.A., Gilmour, P.S., Schins, R.P.F., Knaapen, A.M., Rahman, I., Faux, S.P., Brown, D.M., MacNee, W., 2003. Oxidative stress and calcium signaling in the adverse effects of environmental particles (PM10). Free Radical Biology and Medicine 34, 1369–1382. Frazzini, V., Rockabrand, E., Mocchegiani, E., Sensi, S.L., 2006. Oxidative stress and brain aging: is zinc the link? Biogerontology 7, 307–314. Gamer, A.O., Leibold, E., van Ravenzwaay, B., 2006. The in vitro absorption of microfine zinc oxide and titanium dioxide through porcine skin. Toxicology In Vitro 20, 301–307. Gazaryan, I.G., Krasinskaya, I.P., Kristal, B.S., Brown, A.M., 2007. Zinc irreversibly damages major enzymes of energy production and antioxidant defense prior to mitochondrial permeability transition. Journal of Biological Chemistry 282, 24373–24380. Gojova, A., Guo, B., Kota, R.S., Rutledge, J.C., Kennedy, I.M., Barakat, A.I., 2007. Induction of inflammation in vascular endothelial cells by metal oxide nanoparticles: effect of particle composition. Environmental Health Perspectives 115, 403–409. Gulson, B., McCall, M., Korsch, M., Gomez, L., Casey, P., Oytam, Y., Taylor, A., McCulloch, M., Trotter, J., Kinsley, L., Greenoak, G., 2010. Small amounts of zinc from zinc oxide particles in sunscreens applied outdoors are absorbed through human skin. Toxicological Sciences 118, 140–149. Guo, L., Yang, S.H., Yang, C.L., Yu, P., Wang, J.N., Ge, W.K., Wong, G.K.L., 2000. Synthesis and characterization of poly(vinylpyrrolidone)-modified zinc oxide nanoparticles. Chemistry of Materials 12, 2268–2274. Hall, B.D., Zanchet, D., Ugarte, D., 2000. Estimating nanoparticle size from diffraction measurements. Journal of Applied Crystallography 33, 1335–1341. Jeng, H.A., Swanson, J., 2006. Toxicity of metal oxide nanoparticles in mammalian cells. Journal of Environmental Science and Health Part A: Toxic/Hazardous Substances & Environmental Engineering 41, 2699–2711. Jiang, D.M., Sullivan, P.G., Sensi, S.L., Steward, O., Weiss, J.H., 2001. Zn2+ induces permeability transition pore opening and release of pro-apoptotic peptides from neuronal mitochondria. Journal of Biological Chemistry 276, 47524–47529. Jin, T., Sun, D., Su, J.Y., Zhang, H., Sue, H.J., 2009. Antimicrobial efficacy of zinc oxide quantum dots against Listeria monocytogenes, Salmonella enteritidis, and Escherichia coli O157:H7. Journal of Food Science 74, M46–M52. Jung, S.H., Oh, E., Lim, H., Shim, D.S., Cho, S., Lee, K.H., Jeong, S.H., 2009. Shapeselective fabrication of zinc phosphate hexagonal bipyramids via a disodium phosphate-assisted sonochemical route. Crystal Growth & Design 9, 3544–3547. Kirchner, C., Liedl, T., Kudera, S., Pellegrino, T., Javier, A.M., Gaub, H.E., Stolzle, S., Fertig, N., Parak, W.J., 2005. Cytotoxicity of colloidal CdSe and CdSe/ZnS nanoparticles. Nano Letters 5, 331–338. Lafuma, F., Wong, K., Cabane, B., 1991. Bridging of colloidal particles through adsorbed polymers. Journal of Colloid and Interface Science 143, 9–21. Langford, J.I., Wilson, A.J.C., 1978. Scherrer after 60 years—survey and some new results in determination of crystallite size. Journal of Applied Crystallography 11, 102–113. Limbach, L.K., Bereiter, R., Mueller, E., Krebs, R., Gaelli, R., Stark, W.J., 2008. Removal of oxide nanoparticles in a model wastewater treatment plant: influence of agglomeration and surfactants on clearing efficiency. Environmental Science & Technology 42, 5828–5833. Limbach, L.K., Li, Y.C., Grass, R.N., Brunner, T.J., Hintermann, M.A., Muller, M., Gunther, D., Stark, W.J., 2005. Oxide nanoparticle uptake in human lung fibroblasts: effects of particle size, agglomeration, and diffusion at low concentrations. Environmental Science & Technology 39, 9370–9376. Limbach, L.K., Wick, P., Manser, P., Grass, R.N., Bruinink, A., Stark, W.J., 2007. Exposure of engineered nanoparticles to human lung epithelial cells: influence of chemical composition and catalytic activity on oxidative stress. Environmental Science & Technology 41, 4158–4163. Murdock, R.C., Braydich-Stolle, L., Schrand, A.M., Schlager, J.J., Hussain, S.M., 2008. Characterization of nanomaterial dispersion in solution prior to in vitro exposure using dynamic light scattering technique. Toxicological Sciences 101, 239–253. Nel, A., Xia, T., Madler, L., Li, N., 2006. Toxic potential of materials at the nanolevel. Science 311, 622–627. Nohynek, G.J., Lademann, J., Ribaud, C., Roberts, M.S., 2007. Grey goo on the skin? Nanotechnology, cosmetic and sunscreen safety. Critical Reviews in Toxicology 37, 251–277.

Pecora, R., 1985. Dynamic Light Scattering: Applications of Photon Correlation Spectroscopy. Springer, New York, pp. 442. Reddy, K.M., Feris, K., Bell, J., Wingett, D.G., Hanley, C., Punnoose, A., 2007. Selective toxicity of zinc oxide nanoparticles to prokaryotic and eukaryotic systems. Applied Physics Letters 90. Reed, R.B., Ladner, D.A., Higgins, C.P., Westerhoff, P., Ranville, J.F., 2012. Solubility of nano-zinc oxide in environmentally and biologically important matrices. Environmental Toxicology and Chemistry 31, 93–99. Robinson, B.V., 1990. PVP: A Critical Review of the Kinetics and Toxicology of Polyvinylprrolidone. Lewis Publishers Chelsea, MI, pp. 209. Sadrieh, N., Wokovich, A.M., Gopee, N.V., Zheng, J.W., Haines, D., Parmiter, D., Siitonen, P.H., Cozart, C.R., Patri, A.K., McNeil, S.E., Howard, P.C., Doub, W.H., Buhse, L.F., 2010. Lack of significant dermal penetration of titanium dioxide from sunscreen formulations containing nano- and submicron-size TiO2 particles. Toxicological Sciences 115, 156–166. Sato, T., Sato, A., Arai, T., 1998. Adsorption of polyvinylpyrrolidone on titanium dioxide from binary solvents (methanol/water) and its effect on dispersion stability. Colloids and Surfaces A: Physicochemical and Engineering Aspects 142, 117–120. Sawai, J., Igarashi, H., Hashimoto, A., Kokugan, T., Shimizu, M., 1995. Effect of ceramic powder slurry on spores of bacillus-subtilis. Journal of Chemical Engineering of Japan 28, 556–561. Sawai, J., Yoshikawa, T., 2004. Quantitative evaluation of antifungal activity of metallic oxide powders (MgO, CaO and ZnO) by an indirect conductimetric assay. Journal of Applied Microbiology 96, 803–809. Sensi, S.L., Yin, H.Z., Carriedo, S.G., Rao, S.S., Weiss, J.H., 1999. Preferential Zn2+ influx through Ca2+ -permeable AMPA/kainate channels triggers prolonged mitochondrial superoxide production. Proceedings of the National Academy of Sciences of the United States of America 96, 2414–2419. Slowing, I., Trewyn, B.G., Lin, V.S.Y., 2006. Effect of surface functionalization of MCM-41-type mesoporous silica nanoparticles on the endocytosis by human cancer cells. Journal of the American Chemical Society 128, 14792–14793. Snowden, M.J., Clegg, S.M., Williams, P.A., Robb, I.D., 1991. Flocculation of silica particles by adsorbing and nonadsorbing polymers. Journal of the Chemical Society: Faraday Transactions 87, 2201–2207. Steinebach, O.M., Wolterbeek, H.T., 1993. Effects of zinc on rat hepatoma HTC cells and primary cultured rat hepatocytes. Toxicology and Applied Pharmacology 118, 245–254. Sun, D.Z., Wong, M.H., Sun, L.Y., Li, Y.T., Miyatake, N., Sue, H.J., 2007. Purification and stabilization of colloidal ZnO nanoparticles in methanol. Journal of Sol-Gel Science and Technology 43, 237–243. Unfried, K., Albrecht, C., Klotz, L.O., Von Mikecz, A., Grether-Beck, S., Schins, R.P.F., 2007. Cellular responses to nanoparticles: target structures and mechanisms. Nanotoxicology 1, 52–71. Walz, J.Y., Sharma, A., 1994. Effect of long-range interactions on the depletion force between colloidal particles. Journal of Colloid and Interface Science 168, 485–496. Wang, Y.B., Lou, Y., Luo, Z.F., Zhang, D.F., Wang, Y.Z., 2003. Induction of apoptosis and cell cycle arrest by polyvinylpyrrolidone K-30 and protective effect of alpha-tocopherol. Biochemical and Biophysical Research Communications 308, 878–884. Wong, K., Lixon, P., Lafuma, F., Lindner, P., Charriol, O.A., Cabane, B., 1992. Intermediate structures in equilibrium flocculation. Journal of Colloid and Interface Science 153, 55–72. Wudarczyk, J., Debska, G., Lenartowicz, E., 1999. Zinc as an inducer of the membrane permeability transition in rat liver mitochondria. Archives of Biochemistry and Biophysics 363, 1–8. 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 of ambient and manufactured nanoparticles to induce cellular toxicity according to an oxidative stress paradigm. Nano Letters 6, 1794–1807. Xia, T., Kovochich, M., Liong, M., Madler, L., Gilbert, B., Shi, H.B., Yeh, J.I., Zink, J.I., Nel, A.E., 2008. Comparison of the mechanism of toxicity of zinc oxide and cerium oxide nanoparticles based on dissolution and oxidative stress properties. ACS Nano 2, 2121–2134. Xia, T.A., Zhao, Y., Sager, T., George, S., Pokhrel, S., Li, N., Schoenfeld, D., Meng, H.A., Lin, S.J., Wang, X., Wang, M.Y., Ji, Z.X., Zink, J.I., Madler, L., Castranova, V., Lin, S., Nel, A.E., 2011. Decreased dissolution of ZnO by iron doping yields nanoparticles with reduced toxicity in the rodent lung and zebrafish embryos. ACS Nano 5, 1223–1235. Yamabi, S., Yahiro, J., Iwai, S., Imai, H., 2005. Formation of cellular films consisting of wurtzite-type zinc oxide nanosheets by mediation of phosphate anions. Thin Solid Films 489, 23–30. Yin, H., Casey, P.S., McCall, M.J., Fenech, M., 2010. Effects of surface chemistry on cytotoxicity, genotoxicity, and the generation of reactive oxygen species induced by ZnO nanoparticles. Langmuir 26, 15399–15408.