**4. Discussion**

Engineered nanomaterials and nanoparticles (ENPs) emission are estimated to be mainly derived from landfills (63–91%), with over 260,000–309,000 metric tons of global production in 2010, followed by release to soil (8–28%), water bodies (0.4–7%), and air (0.1–0.5%) [19]. In certain environmental compartments, ENPs may pose a relatively low

risk, whereas organisms residing near ENP production plant outfalls or waste treatment plants may be at increased risk [20]. When ENPs enter terrestrial and aquatic systems, they may threaten the ecological environment and human health [21]. In general, the fate of ENPs in aquatic systems is mainly determined by three processes: heteroaggregation, dissolution, and sedimentation [22,23]. The factors that influence ENP behavior in the environment include size, surface coating materials and their changes (e.g., degradation or replacement by natural organic matter (NOM)), oxidation, dissolution, sulfidation, heteroaggregation, homoaggregation, and sedimentation/resuspension [24–26]. Other environmental factors include sunlight, the pH of the solution, inorganic salts, the interaction with surrounding metals, and dissolved NOM, which will interact with ENPs and lead to their transformation [27–31].

After AgNPs and ZnONPs were added to natural water, we observed that the color and particle size of the nanoparticles changed significantly (Figures 1 and 2). Thus, the nanoparticles may undergo chemical reactions with other components in the water environment. The dissolution of nanoparticles might be generally promoted after interacting with NOM. Two major surface transformation processes, oxidation and sulfidation, may occur on the surface of nanoparticles in the presence of NOM [32,33]. Therefore, NOM alters the toxicity of ENPs by changing suspension stabilization, the bioavailability of metal ions, electrostatic interactions and steric repulsion between nanoparticles and organisms, and induced reactive oxygen generation [33]. Bundschuh et al. noted that the phenomenon of co-occurring contaminants interacting with nanoparticles and indicated that nanoparticles serve as a sink for organic and inorganic co-contaminants in the water column [34]. Therefore, we conducted an elemental mapping analysis to investigate the composition of sediments of AgNPs or ZnONPs after their addition to natural water. Sulfur (S) and chlorine (CI) attached to AgNPs and ZnONPs, and iron (Fe) and phosphorous (P) attached on ZnONPs, which may in turn change the physicochemical properties of the nanoparticles due to the interaction between these molecules (Figure 3). AgNPs release silver only after they are oxidized by dissolved oxygen, and the released silver is readsorbed onto the surface of the nanoparticles or forms a secondary precipitate with complexing species (e.g., Cl− and SO4 <sup>2</sup>−) [35]. Sulfidation of AgNPs or ZnONPs frequently occurs under various environmental conditions and leads to the formation of core–shell Ag0–Ag2S structures or hollow Ag2S NPs. Sulfidation leads to nearly inert NP surfaces that alter their reactivity and toxicity [34]. The fate and stability of nanoparticles in both raw lake water and filtered lake water containing different NOM lead to different aggregation profiles [36]. The authors concluded that the use of pure NOM analogs may not accurately represent nanoparticles' interactions and fate in real natural systems [36]. Our experiment results suggested that the natural water may have mitigated the toxic effects of AgNPs and ZnONPs through nanoparticles aggregation and interaction with NOM, resulting in the formation of larger particles and sedimentation. Nonetheless, the underlying mechanisms of the interactions and relationships among nanoparticles and organic/inorganic substances in the ecosystem require further investigation.

With the advantages of rapid development and optical transparency, the zebrafish embryos are rapidly becoming an attractive vertebrate model species for screening ENPs [37]. Our current study showed a very high survival rate of zebrafish embryos exposed to three different original natural waters samples and their filtrates obtained after passing through different pore sizes of filter. Interestingly, all the above-mentioned natural water samples led to a longer body length of larva than the embryos exposed to MQ water (Figure S2). On the one hand, natural water samples may contain certain essential elements that enhance the development of the embryos. On the other hand, chemicals and/or ENP contamination of surface waters from rivers, lakes, and reservoirs in Taiwan may still be limited and promote the survival of zebrafish embryos. Currently, aquatic AgNP concentrations in fresh water are predicted to range from approximately a few pg/L to 10 ng/L between 2017 and 2050, which might be nontoxic to fish embryos [20]. Although the current ENP contamination may pose a relatively low risk to natural aquatic systems, the organisms

living in the ecosystem near ENP production plants or waste treatment plants may be at higher risk. Therefore, we conducted acute zebrafish embryo toxicity assays by spiking AgNPs or ZnONPs into natural water samples. As shown above (Figures 6–8), both AgNPs and ZnONPs led to significant acute toxicity toward zebrafish embryos in a dose-dependent manner. The level of acute toxicity was relatively lower in the filtered natural water samples than in the MQ water samples, indicating that the interaction and transformation of these nanoparticles with the complex components in a water environment led to a reduced toxicity.

**Figure 8.** Autophagy induced by AgNPs and ZnONPs. Zebrafish embryos were exposed to natural water after passing through a 0.45 μm filter and then spiked with (**a**) AgNPs and (**b**) ZnONPs. After 72 hpf, zebrafish embryos were stained with LysoTracker RED. The fluorescence intensity was assigned as + (weak), ++ (middle), and +++ (strong). The 10 μg/mL MQ groups exhibited higher autophagy signals than groups treated with NPs spiked in natural water samples. AgNPs, silver nanoparticles; ZnONPs, zinc oxide nanoparticles; 3,4-DCA, 3,4-dichloroaniline; hpf, hours post-fertilization.

Zebrafish embryotoxicity tests have been indicated as a suitable approach for assessing the toxicity of both traditional chemicals and ENPs [38,39]. Nonetheless, the majority of

the published studies were conducted in the laboratory with controlled standard water samples. Here, we aimed to reveal the potential toxic effects and mechanisms of AgNPs and ZnONPs on zebrafish embryos in natural water. One of the interesting findings is that ZnONPs, but not AgNPs, triggered a significant delay in embryo hatching (Figure 5). Consistent with our finding, Chen. et al. reported that exposure to ZnONPs suspensions and their respective centrifuged supernatants caused similar hatching delays, whereas the supernatants did not cause larval mortality or malformation. In addition, coexposure to N-acetylcysteine (NAC), a well-known antioxidant, did not alter the effects of ZnONPs on hatchability but rescued their behavioral effect [40]. Thus, the toxicity of ZnONPs may be due to a combination of the effects of dissolved Zn ions and particle-induced oxidative stress. Zinc is an essential transition metal in living organisms that plays an important role in the maintenance of protein structure and enzymatic function. However, excessive free Zn ions are toxic and may be bound by Zn-binding proteins, such as metalloproteins [41]. The dissolved Zn ions interfere with embryo hatching through a chelator-sensitive mechanism that involves the ligation of histidines in the metalloprotease ZHE1, which is responsible for degradation of the chorionic membrane [42,43]. The effects of ZnONPs on delaying hatching were attenuated in filtered natural water samples compared with MQ water, suggesting the slow release of Zn ions from ZnONPs and the interaction of dissolved Zn ions with the complex components in the water environment, which subsequently mitigate effects on embryonic development. Surface coating with different chemicals or NOM may influence the colloidal stability and solubility of ZnONPs or AgNPs and thereby modulate toxicity [44].

Both ENP-induced mortality and developmental toxicity seem to be related to oxidative stress. Excess ROS production may contribute to tissue damage and participate in signal transduction, the proliferative response, gene expression, and protein redox regulation [45]. ENP-induced oxidative stress was proposed as one of the initiators of the disruption of the mitochondrial membrane potential, the induction of ER stress, and cell death mediated by apoptosis and/or autophagy [9]. The mechanisms underlying ENP-induced toxicity have become one of the most frequently studied topics in toxicology during the last two decades. Our previous studies were the first to show that autophagy activated by AgNPs fails to trigger the lysosomal degradation pathway and leads to dysfunctional autophagy, which is relevant to the accelerated cellular pathogenesis of diseases [10,45,46]. More recently, we also prioritized the factors affecting the toxic potential of AgNPs, which included exposure dose/time, cell type, and the size and surface coating of AgNPs. Using an in silico decision tree-based knowledge discovery-in-databases process, the toxicity-related parameters are ranked as follows: exposure dose > cell type > particle size > exposure time ≥ surface coating [47]. AgNPs with larger particle sizes appeared to induce higher levels of autophagy during the earlier phase of both subcytotoxic and cytotoxic exposures in the in vitro cell culture models, whereas apoptosis, but not necrosis, accounted for the compromised cell survival over the same dosage range [47]. In addition, we determined the skin toxicity and the potential mechanisms of ZnONPs combined with UVB exposure and the preventive effect of a well-known antioxidant, pterostilbene. Exposure to both ZnONPs and UVB disrupts cellular autophagy, which in turn increases exosome release from cells. Application of the antioxidant pterostilbene reversed autophagy abnormalities by restoring normal autophagy flux and decreasing NLRP3 inflammasome-loaded exosome release through the attenuation of total ROS and mitochondrial ROS levels [11]. In general, autophagy is a cellular recycling pathway by which lysosomes degrade damaged organelles and/or proteins to maintain cellular homeostasis. However, ENPs have been proven to induce autophagic cell death in several cell types by interfering with autophagy flux and disrupting lysosomal function [48]. The leakage of lysosomal enzymes activates procaspases or damages the mitochondrial outer membrane to induce apoptosis. As shown in the present study, zebrafish embryos exposed to filtered natural water spiked with AgNPs or ZnONPs presented increased ROS levels, apoptosis, and lysosomal activity, an indicator of autophagy (Figures 7 and 8). To the best of our knowledge, the induction of autophagy in

zebrafish embryos triggered by ENPs in natural water has seldom or never been reported previously. As human being and ecosystem exposure to ENPs is unavoidable, an in-depth understanding of ENP-modulated autophagy is required to assess their safety [48].

The existing literature on the embryotoxicity and teratogenicity of ENPs in zebrafish has been reported in a recent review article [39]. The interaction and bioaccumulation of ENPs in zebrafish embryos are associated with several toxic effects, such as delayed hatching, yolk sac alterations, circulatory changes, and musculoskeletal disorders. In addition, the toxic effects of ENPs on innate immunity in a zebrafish model have also been reported [49]. Most of the abovementioned toxic effects are related to dysregulated autophagy. Since autophagy is considered an early indicator of ENP interactions with cells and has been recognized as an important form of cell death in ENP-induced toxicity, creating an autophagy-related transgenic zebrafish line could be a good approach to monitor the ENP pollution in an ecosystem. Overall, our study revealed that AgNPs and ZnONPs spiked in natural water increased zebrafish embryo mortality at higher concentrations, delayed the hatching rate, and induced ROS production, autophagy, and apoptosis (Figure 9). Our current study focused on AgNPs and ZnONPs, which are widely used in several industries, and described their behavior, characteristics, embryotoxicity, and underlying mechanisms in natural aquatic systems. These results will enable the development of more relevant testing methods to predict the possible long-term ecotoxicity of ENPs and can be applied in the future for regulatory decision-making and risk assessments of ENPs.

**Figure 9.** Illustration of the toxic mechanisms of AgNPs and ZnONPs spiked into natural water. The NP suspension was spiked in natural water obtained from NCKU Lake, Zengwun Reservoir, and Erren River and caused lethality and developmental toxicity in embryos. Mechanistically, AgNPs and ZnONPs spiked in natural water induced excessive ROS production, programmed cell death, and overactivated autophagy. AgNPs, silver nanoparticles; ZnONPs, zinc oxide nanoparticles.
