Toxicity of polyelectrolyte-functionalized titania nanoparticles in zebrafish (Danio rerio) embryos

We investigated the effects of short-term exposure of bare TiO2NPs and polyelectrolyte-coated TiO2NPs in the 5–25 nm size range, at relatively high concentrations (of 500 and 1000 mg/L) under light or dark conditions, in D. rerio embryos. The biological endpoints investigated included embryo viability and mRNA transcript levels of antioxidant and membrane transport genes relative to control embryos. The presence of nanoparticles on the surface of embryos was assessed using TEM. The results confirm an accumulation of TiO2NPs on the outer surface (chorion) of the embryo, but not within the embryo. No significant difference in embryo viability was detected following each exposure regime. The expression of antioxidant biomarker, SOD2, was significantly impacted by the type of TiO2NP, with TiO2NPs/PSS/PAH coating exposure showing down regulation; the concentration of the nanoparticles, with down regulation at 500 mg/L; and dark/light condition with down regulation in the light. The expression levels of the hypoxia and membrane markers, HIF1 and Pxmp2, were not significantly impacted by any factor. The study indicates that SOD2 mRNA expression levels may be useful in the detection of apparent oxidative stress induced by the titania nanoparticle build up on the embryo chorion surface.


Introduction
One of the most widely used engineered nanoparticles (NPs) worldwide is nano-sized titanium dioxide NPs (TiO 2 NPs) that is used for its photocatalytic properties [1]. TiO 2 NPs are used in sunscreens and cosmetic creams [2] for their ability to block UV light [3]. Titania is also used as a pigment in toothpaste [4], in skin treatments [5], and paints, as well as food industry applications [6,7]. They have further wide-ranging uses from photosensitizing agents for photodynamic therapy of endobronchial and esophageal cancers [8], as disinfection agents in wastewater treatment [9], as well as in the environmental decontamination of soil, air, and water [10,11].
TiO 2 NPs may be toxic when released into the aquatic environment as they form superoxide and hydroxyl radicals on exposure to sunlight (UV) and oxygen, which could then lead to damage of the cell contents if taken up by organisms [12]. TiO 2 NPs have been categorised into different forms: rutile, anatase, and amorphous [13] whereby the anatase forms are generally found to be more photoactive than rutile TiO 2 NPs. Al-Awady et al. [14] reported a Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s4245 2-020-3137-x) contains supplementary material, which is available to authorized users. significant nanotoxicity of TiO 2 NPs against algae and yeast, which depends on the particle surface charge. Nanoparticles with cationic surfaces are generally found to be more toxic than those with anionic surfaces [15]. Several sunscreen products contain anatase TiO 2 NPs, which has been found to generate reactive oxygen species (ROS) upon illumination with UV light [16]; this potentially leads to biological damage [17]. Previous studies have shown that TiO 2 NPs (at an exposure level of 5 mg/L) induce oxidative stress in the liver of zebrafish [18] and in the brain tissue of rainbow trout (Oncorhynchus mykiss) [19]. In contrast, the potential of TiO 2 NPs to produce oxidative damage to DNA without photoactivation is still unclear [20].
TiO 2 NP toxicity has been shown to be dependent on both particle size and the degree of particle aggregation. Smaller nanoparticles have been found to have higher mobility between biological compartments. For example, the 4 nm TiO 2 NPs have been observed penetrating into the deeper layer of the epidermis (to the basal cell layer) in hairless mice [21]. In contrast though, TiO 2 NP aggregation, increasing the size factor, has been shown to have a larger effect on cell viability and gene expression of biomarkers focused on stress, inflammation, and cytotoxicity in human acute monocytic leukemia and bronchial epithelial cell lines when compared with smaller aggregates of 166 nm [22]. Encapsulating the TiO 2 NPs with coating agents (such as inert oxides of silica, alumina, or zirconium) also mediates the ROS associated impacts [23] by reducing or eliminating them. Coatings based on silicon dioxide have also been found to reduce the oxidative activity of TiO 2 NPs on pig skin [24].
Overall, the increased use of TiO 2 NPs, and their ultimate release into the environment, suggests an increasing need to evaluate their potential toxicity, including the effect of the nanoparticle coating and size range. In this study, D. rerio embryos were used to study the effect of TiO 2 NPs coated with different number of layers of anionic and cationic polyelectrolytes on embryo viability. Selected oxidative stress markers, which are conserved in all vertebrate species, were also measured including superoxide dismutase 2 (SOD2), hypoxia inducible factor 1 (HIF1) and peroxisomal membrane transporter protein (Pxmp2) gene expressions.

Preparation and characterisation of TiO 2 NPs
Three different batches of TiO 2 NPs were prepared in Milli-Q water: bare titania NPs (TiO 2 NPs), anionic NPs prepared with poly (sodium 4-styrene sulfonate) sodium salt (PSS) as TiO 2 NPs/PSS, and the cationic NPs prepared with poly (allylamine hydrochloride) (PAH) as TiO 2 NPs/PSS/PAH. The synthesis of TiO 2 NPs was conducted using the procedure described by Al-Awady et al. [14]-see Fig. 1a. Briefly, 1 M Fig. 1 a Schematic diagram of the coating of the bare TiO 2 NP with polyelectrolytes (PSS and PAH), b TEM image of the titania produced by annealing at 70 °C, c Size distribution of dispersed titanium dioxide NPs synthesised by hydrolysis and condensation of titanium isopropoxide at acidic medium for 20 h at 70 °C. d Zetapotential of the bare TiO 2 NP, TiO 2 NP/PSS and TiO 2 NP/PSS/PAH at pH 5.5 HNO 3 was added drop-wise to 250 mL of Milli-Q water to adjust the pH to 2 followed by dropwise addition of a mixture consisting of 15 mL aliquot of isopropanol and 5.0 mL of titanium isopropoxide (TTIP) to the former solution with vigorous stirring, leading to the formation of a white turbid suspension as a result of the hydrolysis of TTIP. The suspension of Ti (OH) 4 was heated to 70 °C for 20 h to form a yellow-white precipitate of titania that was filtered, washed with ethanol and further dried under vacuum (Gallenkamp vacuum oven) at 100 °C for 2 h. Aqueous dispersions of TiO 2 NPs was prepared by dispersing 4 mg of the titania sample in 10 mL aliquots of 20 mM aqueous solution of NaCl at pH 4 using a digital sonicator (Branson 450, 5 mm tip, 400 W maximal power) at 40% of the maximum power for 10 min at 1 s ON/1 s OFF pulse time and followed by filtration through a syringe filter of pore size 0.22 μm. The TiO 2 NPs were characterised in terms of size distribution and zeta potential in aqueous solutions using a Zetasizer Nano ZL (Malvern, U.K.). Transmission electron microscopy (TEM) images of the particle samples were obtained using JEM 2011 (JEOL, Japan) running at 200 kV.

Layer-by-layer polyelectrolyte-coated TiO 2 NPs
Polyelectrolyte-coated TiO 2 NPs were prepared using titania synthesised and annealed at 100 °C (anatase). 10 mL of 1500 μg mL −1 TiO 2 NPs dispersion in Milli-Q water was added drop-wise to an equal amount of 10 mg mL −1 of solution of PSS (M.W. ~ 70 kDa) dissolved in 1 mM NaCl solution. After shaking for 20 min, the particles were washed three times by centrifugation for 1 h at 8000 rpm to remove the excess of PSS and were finally re-dispersed in 10 mL of Milli-Q water. The PSS-coated TiO 2 NPs were then mixed drop-wise with 10 mL of 10 mg mL −1 PAH (M.W. 15 kDa) dissolved in 1 mM NaCl solution, shaken for 20 min and centrifuged again three times at 8000 rpm for 1 h to yield TiO 2 NPs/PSS/PAH. For further coating with PSS, the latter was mixed drop-wise with 10 mL of 10 mg mL −1 PSS whilst being sonicated. The mixture was shaken for 20 min, centrifuged and dispersed in Milli-Q water to produce TiO 2 NP/PSS/PAH/PSS. Furthermore, PSS and PAH of various molar masses (10 kDa and 70 kDa for PSS and 15 kDa and 56 kDa for PAH) were used to examine their effect on the size of the coated TiO 2 NPs. After each polyelectrolyte coating, the TiO 2 NPs were characterized by the Zetasizer Nano ZL to check their zeta potential and the particle aggregation.

Embryo exposure to TiO 2 NPs
Healthy D. rerio embryos (n = 10) at 0-72 hpf, with the chorion intact, were selected and exposed to a treatment dose of test media (bare TiO 2 NPs, TiO 2 NPs/PSS, or TiO 2 NPs/PSS/ PAH) and incubated for 3 h in either dark conditions or illuminated with visible light, at particle concentrations of 0, 500 or 1000 mg/L based on published LC 50 values to increase the likelihood of observable effects [25,26] rather than environmentally-relevant levels. Healthy D. rerio embryos (n = 10) were used as a control group in parallel.

Embryo viability following exposure to TiO 2 NPs
Embryos (at 48-72 hpf, n = 5) from each exposure regime (control, bare TiO 2 NPs, TiO 2 NPs/PSS, and TiO 2 NPs/PSS/PAH) were isolated after the exposure and washed with commercially supplied (nuclease free) molecular-biology grade water (Fisher Scientific, U.K.) three times, and re-dispersed with 1 mL molecular grade water and incubated with a drop of 98% fluorescein diacetate (FDA) (Honeywell Fluka, U.K.) in acetone (0.5 mg/L) for 15 min. This assay is based on accumulation of the fluorescent by-product (fluorescein) inside the viable embryos as a result of the hydrolysis of the diffused FDA by intracellular enzymes (esterases). The embryos were then washed again with deionised water and the cell viability examined using an Olympus BX51 fluorescence microscope attached to a DP70 digital camera and FITC fluorescence filter set. Living cells were identified as having taken up FDA and fluorescent green [27].

TEM imaging of embryos after exposure to TiO 2 NPs
The morphology of D. rerio embryos (n = 5) after a 3 h incubation with 0, 500 or 1000 mg/L of bare TiO 2 NPs TiO 2 NPs/ PSS, or TiO 2 NPs/PSS/PAH was examined with TEM using the following protocol. The embryos were washed with deionised water and fixed in 2.5% glutaraldehyde (0.5 ml 25% glutaraldehyde stock solution, 4.5 ml 0.

Target gene isolation and characterization
Total RNA was extracted from pooled samples of embryos (0-72 hpf with chorion intact, n = 10) from each treatment group, using the manufacturer's protocol (Roche Diagnostics Ltd., Burgess Hill, U.K.). The embryo pooled sample exposures consisted of bare TiO 2 NPs, TiO 2 NPs/PSS, or TiO 2 NPs/PSS/PAH at 500 or 1000 mg/L particle concentration for 3 h exposure duration, in dark or illuminated with visible light as well as the corresponding control treatment group (n = 10). To assess the integrity of total RNA, samples were analysed on a denaturing agarose gel stained with ethidium bromide (Life Technologies, Paisley, U.K.). 100 ng of pooled RNA was used to generate cDNA using SuperScript VILO cDNA Synthesis reagents and protocol (Life Technologies, Paisley, U.K.) with 14 μL (~ 100 ng) of total RNA. In a 0.2 mL tube, the following reagents were added: 4 μL of 5x VILO Reaction Mix (includes random primers, MgCl 2 , and dNTPs in a buffer formulation) 2 μL of 10x Superscript enzyme mix. Each reaction was incubated at 25 °C for 10 min, and then 60 min at 42 °C followed by 5 min at 85 °C and a holding step at 4 °C. To degrade any remaining RNA, the following reagents were added: 0.5 μL (5 units The PCR fragments were sequenced commercially by Macrogen (Amsterdam, Netherlands). Identities of PCR fragments were verified using a blastn search on the NCBI database (http://blast .ncbi.nlm.nih.gov/Blast .cgi), and aligned using a multiple sequence alignment program, Clustal Omega (http://www.ebi.ac.uk/Tools /msa/clust alo/) to determine the correct isoform.

Quantitative qPCR analysis of mRNA expression
The qPCRs analyses for each pool of embryo cDNAs from each treatment group (n = 10) were carried out using 20 μL reaction volumes consisting of 10 μL of SYBR Green Master Mix (Roche, U.K.), 7 µL of sterilised water, 1 µL of the cDNA template, and 2 µL of optimised primer concentration (EF, HIF1: 200 nM; 18S, Pxmp2: 300 nM, SOD2: 400 nM). Two reference genes (EF and 18S) were determined as the most stable across treatment groups using geNorm software. Amplifications were carried out using a CFX96 Real-time PCR system, C1000 Thermal Cycler (Bio-Rad, Hemel Hempstead, U.K.), in triplicate and with negative controls. Reactions were started with denaturation at 50 °C for 2 min, 95 °C for 10 min, followed by a three-step protocol of 40 cycles of denaturation at 95 °C for 10 s, annealing at 60 °C for 1 min, then 72 °C for 1 min. At the end, a melting/dissociation curve was conducted. A relative quantification method was used to determine changes in mRNA transcript levels of the targeted genes in the treatment group compared to untreated control samples using the geometric mean of the reference genes for normalization and the ΔΔCt method [28].

Statistical analysis
Each target gene was tested individually for significant differences among the controls and each treatment group. All data were tested for homogeneity of variances using Levene's test in SPSS. A non-parametric test (Scheirer-Ray-Hare) was used to assess the effect of anatase TiO 2 NPs coating type (factor 1), TiO 2 NP concentration (factor 2) and the exposure condition (factor 3) and to determine the interactions among them. Significance for relative gene expression, between TiO 2 NP of different coatings, concentrations, or conditions was also tested individually using the Kruskal-Wallis non-parametric test. Differences were considered significant at P < 0.05.

Characterization of the TiO 2 NPs size distribution and zeta potential
Aqueous dispersions of the titania samples were synthesized at different annealing temperatures and prepared by sonication as described by Al-Awady et al. [14]. Figure 1a illustrates schematically the process of coating of the bare TiO 2 NPs with consecutive layers of PSS and PAH respectively. Figure 1b shows a typical TEM image of the bare TiO 2 NPs used in these experiments. The titania produced was characterized as clusters of smaller crystallites of 5 nm domain size in solid state (Fig. 1b). Upon dispersing in Milli-Q water at pH 5.5, TiO 2 NPs of an average diameter 25 nm (Fig. 1c) were produced. The zeta potential of the anatase TiO 2 NPs in an aqueous solution decreased gradually from positive at low pH to negative at high pH with an isoelectric point at approximately 6.8 (see Fig. 1d).

TiO 2 NP uptake and impact on D. rerio embryo viability
Embryos displayed no significant impact on viability (Fig. 2) when exposed to all the TiO 2 NPs types (bare TiO 2 NPs, TiO 2 NPs/PSS, or TiO 2 NPs/PSS/PAH) at each exposure level (0, 500 and 1000 mg/L), in both dark and visible light conditions. D. rerio embryos from each of the treatment group were examined using EDS attached to TEM (Fig. 3, Table 1). EDS spectra confirmed the presence of TiO 2 NPs on the outer surface (chorion) of D. rerio embryo incubated with 500 mg/L of TiO 2 NPs/PSS under both dark and visible light conditions (Fig. 3b, c). In the samples examined using TEM, no TiO 2 NPs were detected inside of the embryos for any treatment or on the embryo outer surface (chorion) in the control group (Fig. 3a) or after incubation with 500 mg/L of TiO 2 NPs/PSS/PAH under both dark and visible light conditions (Fig. 3d, e). TiO 2 NPs were however detected on the outer surface (chorion) of embryos incubated with 1000 mg/L TiO 2 NPs/PSS/PAH incubated under dark conditions (Fig. 3f ). No data was available (due to human error) for embryos incubated with 1000 mg/L TiO 2 NPs/PSS/PAH incubated under light conditions.

qPCR analysis of target gene expression in D. rerio embryos following exposure to TiO 2 NPs
The expression levels of SOD2, HIF1, and Pxmp2 mRNA were analysed in control embryos and embryos pooled from each treatment group (n = 10) of TiO 2 NPs with concentrations of 0, 500 and 1000 mg/L in dark and visible light conditions using the optimised qPCR method. Firstly, an overall statistical analysis using the Scheirer-Ray-Hare test ( Table 2) showed that for the nanoparticles the TiO 2 NP type, concentration, condition, the  Table 2). The expression level of HIF1 was only affected by the condition (light or dark) used in the experiment (Table 2). Pxmp2 mRNA expression level was not significantly impacted by any of the types of TiO 2 NPs, concentration, nor the exposure regime (Table 2). A Kruskal-Wallis test highlighted further significance within the dataset as follows. Significant difference in SOD2 expression level was detected as a result of TiO 2 NP type,  Table 1 Trace amounts of titania detected on the surface (chorion) of embryo samples (n = 5) exposed to different coated titania treatments using EDS-TEM analysis Effect of concentration (Conc), Exposure condition (Cond), 1 Interaction between types of TiO 2 NPs and concentration (Int 1 ), 2 Interaction between types of TiO 2 NPs and exposure condition (Int 2 ), 3 Interaction between concentration and exposure condition (Int 3 ), 4 Interaction between types of TiO 2 NPs, concentration, and exposure condition (Int 4 ), ns not significant concentration, and condition (Fig. 4). HIF1 expression was significantly affected by condition (Fig. 5). Pxmp2 expression was not affected by any condition (Fig. 6). Separate Kruskal-Wallis tests ( Table 2) for one-way analysis of the individual data between TiO 2 NPs types, concentration, and condition, revealed that SOD2 mRNA expression was

Discussion
From this study, using two relatively high exposure levels of uncoated and coated titania NPs (500 and 1000 mg/L), it is evident that overall embryo viability is not affected (Fig. 2), even though titania can be detected on the surface (chorion) of embryos (Fig. 3b, c, f ), and significant changes in oxidative stress gene expression, particularly SOD2 was detected in selected exposure treatment groups (Fig. 4).
On the other hand, the markers adopted for hypoxia (HIF1, Fig. 5) and membrane function (Pxmp2, Fig. 6), suggest that such high exposure levels of these NPs are not having a significant impact on these specific endpoints under this exposure regime.
In terms of NPs availability and uptake, the anatase NPs employed in this study had an average diameter of 25 nm with a range of different coatings. Their impact on viability and their apparent absence within the embryos analysed, would suggest that they are not crossing the embryo and/or causing mortality under the exposure conditions used. This is consistent with studies using eggs and adult zebrafish that reported an LC 50 value of > 1600 mg/L for uncoated (< 100 nm sized) TiO 2 NPs after a 48 h exposure period [29]. Studies using significantly longer exposure times (of 23 days) have shown decreased survival of D. rerio embryos exposed to TiO 2 NPs at lower concentrations from 10 μg/L to 10 mg/L [30]. Adding UV light has also been shown to increase mortality, in larvae rather than embryos, at concentrations of 1-100 mg/L [31,32]. The D. rerio embryo is ~ 1.5 mm thick [33], is surrounded by a protective chorion, which, at 72 hpf, is considered open to the passage of materials through pores or via passive transport [34], yet these results suggest that no nanoparticle transfer across the chorion has occurred.
Two coatings were compared with each other in terms of relative uptake and toxicity. The TiO 2 NPs/PSS/PAH used are cationic and UV-photoactive, similarly to the uncoated TiO 2 NPs, i.e. their positive charge is anticipated to promote adhesion on the embryo surface. However these were only identified on the surface of the embryos exposed to the higher 1000 mg/L level (Table 1), with possible disruption of the negatively charged cell membrane [14]. Previous work using microalgae, Chlamydomonas reinhardtii, and yeast, Saccharomyces cerevisiae, has also confirmed the formation of a significant build-up of NPs on the cell surface for bare and cationic polyelectrolyte-coated TiO 2 NPs at pH 5.5 [14]. In contrast, titania was also detected on the outer surface of microtome-sectioned D. rerio embryos exposed to 500 mg/L anionic-coated TiO 2 NPs/PSS in both dark and visible light conditions (Table 1). Regarding possible mechanisms to understand how such particle build up may occur, the negative charge of the cell membrane has been suggested to facilitate internalization, and affect the toxicity of positively charged coated NPs in other studies, such as gold NPs (AuNPs), which are more toxic than negatively and/or neutrally charged AuNPs [35,36]. In terms of additional coatings and nanohybrids, the toxicity of TiO 2 NP and TiO 2 -MWCNT nanohybrid has also been assessed with and without UV light exposure using zebrafish embryos, and neither presented acute toxicity [37]. The acute effects of TiO 2 NPs in zebrafish embryos thus depend on both the type of formulation and the illumination condition.
Three biological effects markers of sub-lethal impacts were examined. The expression of SOD2 mRNA in pooled zebrafish embryos (n = 10) was affected by the type of TiO 2 NPs, concentration, and condition (Table 2, Fig. 5) specifically indicating an oxidative stress response. This finding is consistent with those reported by Bar-Ilan et al. [30] whereby exposure to 10 μg/L-10 mg/L TiO 2 NPs, illuminated with a lamp, produced toxicity through cumulative reactive oxygen species. The expression levels of the hypoxia and membrane markers, HIF1 and Pxmp2, were not significantly impacted by any factor using the same exposure conditions. The SOD enzyme catalyses the conversion of the reactive superoxide ion (O 2 − ) to yield hydrogen peroxide (H 2 O 2 ) and oxygen molecule during oxidative oxygen processes [38]. Other markers of oxidative stress, increased catalase and glutathione S-transferase expression levels have also been reported in zebrafish embryos, exposed to TiO 2 NPs for 96 h, under either visible light or a combination of visible and ultraviolet (UV) light [32]. Felix et al. [39] examined sublethal biological effects impacts of 0.1, 1 or 10 mg/L of uncoated TiO 2 NPs, poly(acrylic acid)-coated TiO 2 NPs, and the polymer coating alone, in the presence or absence of UV light, reporting that uncoated TiO 2 NPs produced hydroxyl radicals, delayed hatching, induced lipid peroxidation, increased catalase activity and total glutathione levels, and up-regulated glutathione peroxidase 1a gene expression in the presence of UV light, while polymercoated TiO 2 NP increased thiobarbituric acid reactive substances production and total glutathione levels under simulated sunlight illumination. Further experiments are needed, with an increased number of embryos in each treatment group, and a shorter defined embryo stage (from 0 to 72 hpf range), in order to reduce the variation of gene expression within treatment groups observed.

Conclusions
In summary, TiO 2 NP size, surface charge, concentration and the presence/absence of light have been shown to determine their potential toxicity measured in this study as specific gene expressions. Polyelectrolytes coatings were used in formulations to enhance dispersion stability [40]. The polyelectrolyte multilayer films (PAH/PSS) provide a stable nanocomposite thin film that interacts with the NPs [41]. The nanotoxicty of polyelectrolytecoated TiO 2 NPs have been previously been studied in yeast and microalgae and the results showed that the toxicity of the coated TiO 2 NPs changes with their surface charge where cationic polyelectrolyte coating were more toxic than the anionic polyelectrolyte coating [14]. Here, we compare the toxicity of different TiO 2 NPs coatings on D. rerio embryos and find that D. rerio embryos remain viable after exposure to 500 and 1000 mg/L of TiO 2 NPs coated with anionic and cationic polyelectrolytes for 3 h. Also, embryos exposed to TiO 2 NPs coated with cationic polyelectrolytes showed no Ti on the embryo using EDS while the higher dose of 1000 mg/L of the same coating start to show NPs residues. Importantly, the biological sub-lethal effects marker, SOD2 expression, showed significant changes related to all factors, indicative of oxidative stress. Similarly, HIF1 expression showed a significant difference in response to condition. This study focused on a relatively short-term exposure with concentrations that are not environmentally relevant. It would be interesting to understand how NPs at environmentally realistic exposure levels affect D. rerio over longer exposure periods, at defined embryo stages, and also several generations.