Zinc oxide (ZnO) nanoparticles (NPs) are widely used in various commercial and industrial products, such as batteries, cement, ceramics, fire retardants, foods, glass, paints, pigments, plastics, and rubber (Aitken et al. 2006; Chen et al. 2014). It has been estimated that between 550 and 33,400 tonnes of ZnO NPs are produced each year worldwide. The global production volume of ZnO NPs is the third highest of the NPs that contain metals, after SiO2 NPs and TiO2 NPs (Bondarenko et al. 2013). Their widespread use has led to ZnO NPs being found in environmental media, and ZnO NP concentrations in sediment, sewage sludge, soil, surface water, and wastewater of 0.49–56, 13.6–64.7, 0.026–0.66, 1–60, and 0.22–1.42 µg/L, respectively, have been predicted and/or measured (Gottschalk et al. 2009, 2010). Higher ZnO NP concentrations have been found in aquatic environments that are the final destinations of contaminants than in other environments (Bystrzejewska-Piotrowska et al. 2009).

The dose is not the only factor that is used to evaluate the toxicities of NPs unlike for conventional contaminants. The physicochemical properties of the NPs, such as the degree to which aggregations form and the chemical compositions, particle shapes, and sizes of the NPs can affect the toxicities of the NPs. For metal-based NPs, it has been found that rod-shaped iron oxide nanoparticles are more toxic than spherical nanoparticles to murine macrophage cells, that the toxicities of silver NPs to bacillus are charge-dependent, and that amorphous TiO2 NPs 30 nm in diameter induce reactive oxygen species more than do other TiO2 NPs (Jiang et al. 2008; El Badawy et al. 2010; Lee et al. 2014). However, the roles the physicochemical properties of ZnO NPs play in the aquatic toxicities of the NPs have not been fully elucidated. In recent studies, the release of Zn2+ ions was suggested to be an important factor affecting ZnO NP toxicity (Xia et al. 2011; Fukui et al. 2012; Gilbert et al. 2012). It is therefore necessary to compare the toxicities of ZnO NPs and ZnSO4 (an efficient source of soluble zinc ions) to gain an understanding of the contributions of other factors to the aquatic toxicities of ZnO NPs.

The development of omics techniques has made it possible to simultaneously assess the expression profiles of thousands of genes in response to different stresses (Ung et al. 2010). Omics techniques can provide insights into the mechanisms involved in toxic effects and allow biomarkers related to molecular level responses to be screened rapidly (Long et al. 2012; Griffitt et al. 2013). Griffitt et al. (2013) recently found that the chronic exposure of zebrafish (Danio rerio) to Ag NPs caused transcriptomic changes related to cellular restructuring, developmental processes, and the repair of DNA damage. However, despite the advantages of omics techniques, only a few attempts have been made to investigate the toxicities of ZnO NPs to aquatic organisms.

The zebrafish is a small and transparent fish that is easy to maintain. Zebrafish continually reproduce, and embryogenesis is rapid (Westerfield 1995; Hill et al. 2002; Zhao et al. 2013). The zebrafish genome has been sequenced, and genetic information on zebrafish is being accumulated quickly (Berry et al. 2007). Zebrafish are commonly used as an animal model in high-throughput acute toxicity studies, and zebrafish have been used to assess the toxicities of nanomaterials (Lin et al. 2013; Felix et al. 2013). Combining a well-established model organism and a microarray platform capable of assessing changes in a large number of genes would allow detailed and accurate insights into the molecular mechanisms underlying the adaptive responses of fish to toxic substances to be gained.

The objectives of this study were: (1) to compare the toxicities of ZnO NPs and ZnSO4 to zebrafish larvae in acute toxicity tests; (2) to identify differently expressed genes (DEGs) by performing whole gene expression profiling on larval zebrafish exposed to low and sub-lethal concentrations of ZnO NPs and ZnSO4; and (3) to use a pathway analysis tool to attempt to identify upstream biological causes and probable downstream effects of sub-lethal concentrations of ZnO NPs and ZnSO4.

Materials and Methods

The uncoated ZnO NPs used in the experiments had nominal sizes of 10–30 nm, and were purchased from SkySpring Nanomaterials (Houston, TX, USA). A transmission electron microscopy image and a histogram of the size distribution of the ZnO NPs, provided by the supplier, are shown in Fig. 1. The transmission electron microscopy image was acquired using a JEM2010 instrument (JEOL, Tokyo, Japan) at 200 kV. A 100 mg/L solution of the ZnO NPs was prepared by dispersing the NPs in deionized water and sonicating the mixture at a setting of 53 W for 1 h in a Sonics instrument supplied by Sonics & Materials (Newtown, CT, USA).

Fig. 1
figure 1

a Transmission electron microscopy image showing the morphologies of the ZnO nanoparticles and b a histogram of the ZnO nanoparticle (NP) size distribution

Full size image

The dissolved zinc in the supernatant was quantified using an inductively coupled plasma optical emission spectrometer (PerkinElmer, Waltham, MA, USA), which was equipped with an S10 autosampler (PerkinElmer). The dissolved zinc concentrations that were found are shown in Table 1.

Table 1 Characteristics of the ZnO nanoparticles and the dissolved Zn concentrations in the suspensions and solutions used in the experiments
Full size table

Zebrafish larvae at 72 hpf (i.e. after hatch) were obtained from the Gyeongnam Department of Environmental Toxicology and Chemistry, Korea Institute of Toxicology. Each experiment was performed in a 250 mL beaker containing 200 mL of zebrafish culture water (carbon-filtered dechlorinated tap water), and each experiment lasted 72 h. Preliminary tests were performed to assess the toxicities of sublethal ZnO NP concentrations (0.153, 0.307, 1.228, 2.457, 4.914, 9.827, 12.284, and 122.838 μmol/L) and sublethal ZnSO4 concentrations (0.619, 3.902, 7.741, 15.483, 31.585, and 619.310 μmol/L). The LC10 and LC30 values for the ZnO NPs and ZnSO4 were determined using a linear regression method (CETIS v1.8.7; Tidepool, McKinleyville, CA, USA). For the microarray analyses, zebrafish larvae at 72 hpf were exposed to ZnO NPs for 72 h. An aliquot of the ZnO NP suspension was added to the zebrafish culture water in each beaker to give the control concentration or a nominal concentration of 1.032 μmol/L (the LC10) or 3.955 μmol/L (the LC30). Reference treatments using ZnSO4 at concentrations of 29.838 μmol/L (the LC10) and 38.422 μmol/L (the LC30) were also performed to allow the toxic effects of the ZnO NPs to be compared with the toxic effects of zinc ions. Each treatment was performed in triplicate. After the larvae had been exposed to ZnO NPs or ZnSO4 for 72 h, they were anesthetized using tricaine methanesulfonate (MS-222; Sigma-Aldrich, St. Louis, MO, USA) at a concentration of 100 mg/L following guidelines provided by the Institutional Animal Care and Use Committee of the Korea Institute of Toxicology.

The total RNA isolated from ~70 individual zebrafish larvae cultured in the same dish was treated using an RNeasy mini kit (Qiagen, Hilden, Germany). The RNA integrity was confirmed by checking the RNA integrity number using an Agilent Bioanalyzer 2100 and an Agilent RNA 6000 Nano Kit (Agilent Technologies, Wilmington, DE, USA).

One-color microarray analyses to determine gene expression in the zebrafish larvae were performed using a NEX BiO system (Daejeon, Korea) using an Agilent zebrafish oligo microarray (4 × 44 K; Agilent Technologies) containing 43,803 zebrafish probes. Three replicates were independently collected for each treatment, and a total of 15 microarray assays were conducted. The steps between RNA labeling and scanning were performed following the “one-color microarray-based gene expression analysis (low input quick amp labeling)” protocol provided by Agilent Technologies with slight modifications.

The total RNA was amplified using a Low RNA Input Linear Amplification kit (Agilent Technologies) and labeled using 1.65 µg of Cyanine three following the manufacturer’s instructions. Labeled cRNA was purified using an RNeasy mini kit (Qiagen). Each slide was hybridized with cRNA labeled with Cyanine three using a Gene Expression Hybridization kit (Agilent Technologies) for 17 h, then each slide was washed with the stabilization and drying solution provided in a Gene Expression Wash Buffer kit (Agilent Technologies). The microarray slides were scanned using an Agilent Scanner B system (Agilent Technologies), and the signal intensities were analyzed using Feature Extraction 10.7 software (Agilent Technologies) using the default settings. The raw data were normalized using the percentile shift method.

Each microarray dataset was analyzed using the significance analysis of microarrays method described by Tusher et al. (2001). Genes that showed significant expression changes (by a factor of more than two) were selected with a q value cutoff of 5%. The genes were classified by performing gene set enrichment analysis, which assigned genes for which significant differences were found in the ZnO NP and ZnSO4 treatments to previously defined pathways. The gene set enrichment analysis results were analyzed using Ingenuity Pathway Analysis software (Ingenuity Systems, Redwood City, CA, USA).

Results and Discussion

The estimated 72 h LC10 and LC30 values for the ZnO NPs for zebrafish larvae at 72 hpf were 1.032 and 3.955 μmol/L, respectively, and the estimated 72 h LC10 and LC30 values for ZnSO4 for zebrafish larvae at 72 hpf were 29.838 and 38.422 μmol/L, respectively. This indicated that the ZnO NPs were markedly more toxic than ZnSO4 to the larval zebrafish. Differently expressed genes (DEGs) were identified from the microarray-based expression dataset, and these DEGs allowed the responses to the ZnO NPs and the cytotoxicities of the ZnO NPs to be predicted. As is shown in Fig. 2, marked transcriptional responses to exposure to ZnO NPs relative to the controls were not found, but a number of genes were upregulated or downregulated in zebrafish exposed to ZnSO4 relative to the controls.

Fig. 2
figure 2

Numbers of differently expressed genes (DEGs) in zebrafish exposed to different concentrations (the LC10 and LC30) of ZnO nanoparticles (NPs) and ZnSO4 relative to the three groups (control, LC10 ZnO NPs, and LC30 ZnO NPs)

Full size image

These results suggest that the ZnO NPs were more toxic than ZnSO4 but that the ZnO NPs did not cause significant gene expression changes relative to the controls, indicating that the physical properties of the ZnO NPs may have been responsible for the increased zebrafish larvae mortality. It has previously been found that increased mortality in fish larvae exposed to ZnO NPs was related to mouth-gaping behavior leading to increased uptake of the ZnO NPs (Zhou et al. 2015). Lin et al. (2013) found that hatched larvae take up NPs through the gills, intestines, skin, and other organs, and that, therefore, toxicological responses to NP exposure may be related to the sizes of NPs. It has been found that zebrafish mortality was more markedly increased by exposure to ZnO NPs than by bulk ZnO (Xiong et al. 2011), which could indicate that the particle size could be the most important factor controlling ZnO NP toxicity to hatched larvae (when each larva was not protected by the chorion). It has been suggested that the release of zinc ions plays an important role in the effects of ZnO NPs on gene expression (Shen et al. 2013), but we found that relatively small amounts of zinc ions were released at the ZnO NP LC10 and LC30. We found a negative correlation between gene expression changes and ZnO NP toxicity to the zebrafish larvae. In contrast, it has previously been found that the number of significantly altered transcripts increased dose-dependently as the ZnSO4 concentration increased, suggesting that an overload of cellular zinc ions can activate or inhibit molecular mechanisms by interfering with zinc ion homeostasis (Kao et al. 2012).

As is shown in Fig. 2, the number of upregulated DEGs at the ZnSO4 LC10 relative to the controls and at the ZnSO4 LC10 relative to the ZnO NP LC10 were similar, indicating that the ZnO NPs at the LC10 did not affect the number of upregulated DEGs. However, the proportion of DEGs that were upregulated decreased when the ZnSO4 LC30 results were compared with the ZnO NP LC30 results rather than the control results, indicating that more than half of the upregulated DEGs at the ZnSO4 LC30 relative to the controls were also activated by the ZnO NPs at the LC30. We analyzed the canonical pathways of the upregulated DEGs at the ZnSO4 LC30 using ingenuity pathway analysis. As is shown in Fig. 3a, b the ranked pathway order and ratio values were different, suggesting that different pathways were affected at the ZnO NP LC30. In particular, the pathways affected at the ZnO NP LC30 were associated with cell differentiation and the immune system. It has been found in recent studies that ZnO NPs induce 3T3-L1 adipocyte differentiation in mouse embryos and inflammatory responses (such as the release of pro-inflammatory cytokines) in both mammalian cells and zebrafish (Brun et al. 2014; Saptarshi et al. 2015; Pandurangan et al. 2016).

As is shown in Fig. 2, twice as many DEGs were downregulated at the ZnSO4 LC10 relative to at the ZnO NP LC10 than at the ZnSO4 LC10 relative to the controls, indicating that some of the DEGs that were downregulated at the ZnSO4 LC10 relative to at the ZnO NP LC10 were upregulated at the ZnO NP LC10. As is shown in Table 2, the canonical pathways that were identified for the DEGs that were downregulated at the ZnSO4 LC10 relative to at the ZnO NP LC10 were typically related to cancer cell differentiation, endocytic transport, and genes such as the epidermal growth factor receptor (EGFR), V-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS), and phosphoinositide-3-kinase regulatory subunit 6 (PIK3R6) genes.

Fig. 3
figure 3

Canonical pathway analysis results obtained using ingenuity pathway analysis for upregulated differently expressed genes in response to exposure to a ZnSO4 at the LC30 relative to the control and b ZnSO4 at the LC30 relative to exposure to ZnO NPs. The bars measured against the Y-axis on the left, are the negative logarithms of the p-values. The threshold was a −log(p-value) value of 1.3 (corresponding to a p-value of 0.05). The square markers and line, measured against the Y-axis on the right, are the gene number ratios for the pathways that met the cutoff criteria.

Full size image
Table 2 The top five canonical pathways results obtained using ingenuity pathway analysis for downregulated differently expressed genes at the ZnSO4 LC10 relative to the ZnO NP LC10
Full size table

KRAS and EGFR are important effector genes in tumorigenesis and tumor progress, and PIK3R6 also plays an oncogenic role in several cancers (Markman et al. 2010; Zhou et al. 2012). In particular, mutated kras and egfr genes are the most powerful predictors in several cancer cell lines (Suda et al. 2010; Misale et al. 2012). With regard to NP toxicity, Shvedova et al. (2014) found that single-walled carbon nanotubes and carbon nanofibers increase the incidence of KRAS oncogene mutations in the lung. The genotoxicities of nanotubes have been demonstrated by the presence of cellular defects, such as chromosomal aberrations, enhanced micronuclei frequencies, and increased DNA strands in cells exposed to nanotubes (Schins et al. 2012). The activation of the kras and egfr genes induced by the exposure of larval zebrafish to ZnO NPs at the LC10 may therefore indicate that ZnO NP toxicity can potentially lead to genotoxic effects. Sharma et al. (2009) found that ZnO NPs at low concentrations can be genotoxic to human epidermal cells. However, further studies will need to be performed to determine whether the induction of kras and egfr gene expression is closely linked to the pathways associated with ZnO NP toxicity and the downstream effects of ZnO NP toxicity. Unlike results found previously, most of the DEGs that were downregulated at the ZnSO4 LC30 relative to the controls were not downregulated at the ZnSO4 LC30 relative to at the ZnO NP LC30, indicating that the ZnO NPs and ZnSO4 at their LC30s inactivated similar biological processes, such as cholesterol synthesis and the cell cycle (data not shown).

In conclusion, we found that ZnO NPs are relatively toxic to larval zebrafish even though they do not, compared with ZnSO4, markedly modulate transcripts. The different DEG patterns for the ZnO NPs and ZnSO4 might be caused by the physical properties of the NPs rather than the release of zinc ions from the ZnO NPs, implying that ZnO NPs and ZnSO4 have different toxic mechanisms. Comparison analysis of the microarray data using ingenuity pathway analysis showed that the inflammation system and several genes involved in cancer cell differentiation were altered in zebrafish larvae exposed to ZnO NPs. The genes that were activated by exposure to ZnO NPs were not affected to a significant degree (the number of fold changes was not greater than two), but our results provide clues that will help in the interpretation of the molecular mechanisms that underlie changes in the gene expression regulatory networks caused by exposure to ZnO NPs. Further studies will be required to elucidate the relationships between the uptake kinetics of unequally sized or shaped ZnO NPs and cellular pathology.