A review of the effects of metallic nanoparticles on fish

Many important discoveries have been made in the field of nanotechnology in the last 40 years. Since then, nanoparticles became nearly ubiquitous. With their spreading use, safety concerns have warranted extensive research of nanotoxicity. This paper offers information about the occurrence, transport, and behaviour of metallic nanoparticles in the aquatic environment. It further summarizes details about parameters that dictate the toxicity of nanoparticles and discusses the general/common mechanisms of their toxicity. This review also focuses on fish exposure to nanoparticles, including the possibility of trophic transport through the food chain. Information on some of the most frequently used metallic nanoparticles, such as silver, gold, and titanium dioxide, is further elaborated on.

have to take into consideration the formation of stabilized suspensions. Most metallic NPs are hydrophilic and often have low solubility, which can be perceived positively because metal ions tend to be their most toxic form ). An example is the silver ion Ag + , which is more toxic than silver bound in the form of compounds and Ag 0 within NPs (Hogstrand and Wood 1998;Ribeiro et al. 2014). The fate of dissolved ions is influenced by the composition of the environment, Ag + in natural waters rapidly forms sulphides and chlorides, which are insoluble, have lower toxicity, and are deposited in sediments (Kaegi et al. 2013). Sedimentation is another important way of loss of NPs from the water column. It is usually preceded by particle aggregation, the extent of which depends on pH, the presence of cations, ionic strength, and the size, shape, and charge of particles Schaumann et al. 2015). The sea and brackish regions have high ionic strength, so lower concentrations of NPs can be expected due to rapid aggregation in those environments (Pamies et al. 2014;Li et al. 2016).
Both homoaggregation (NPs form aggregates with each other) can occur, or more often heteroaggregation, when they bind to naturally occurring colloids, which are present in waters in significant concentrations, namely 1-10 mg·l -1 . These colloids consist of inorganic (e.g. metal oxides and sulphides, amorphous silicon), organic (formed primarily by humic substances and fibrous material, which may be polysaccharides and proteins formed by microorganisms), and biological components (viruses, bacteria). The fibrous material facilitates the aggregation of NPs, while humic substances can stabilize the dispersion of NPs by creating a coating that stabilizes the charge of the particle Kaegi et al. 2013). Nanoparticles can also undergo redox reactions that change the chemical properties of the particle surface and can affect its environmental behaviour.This is typical for Fe 0 which is used for water remediation. Iron particles oxidize rapidly in the environment, reducing the pollutant, such as trichloroethylene (Liu et al. 2005).

Determining properties of metallic nanoparticles
The toxicity of metallic NPs is dictated by multiple parameters, which must be taken into consideration as a complex of information, that includes their chemical composition, crystal structure, and purity of the sample, method of synthesis, size, and surface area of the nanoparticle (Warheit 2008;Wijnhoven et al. 2009) Larger particles are usually found to be less toxic, due to a smaller surface area. At the same volume, they have fewer atoms in the surface layer available for interaction with other compounds or ionization (Angel et al. 2013). At the same time, the nanosize provides opportunity to cross biological barriers and cell membranes (Nel et al. 2006). A study by Lee et al. (2007) demonstrates the passage of Ag NPs with Brownian diffusion through the chorion pore canals into the chorionic space and inner mass of zebrafish (Danio rerio) embryos in vivo. Ivask et al. (2014) show the size-and concentration-dependent effects of citrate-coated Ag NPs (10 nm, 20 nm, 60 nm, and 80 nm) on bacteria, yeast, algae, crustaceans, and mammalian cells in vitro. The shape of the particle plays an important role, as well. Hua et al. (2014) found that exposure of zebrafish embryos to ZnO nanosticks had more serious effects on the survival rate and hatching of the embryos than ZnO nanospheres or cuboidal submicron particles. Another aspect is surface chemistry, such as their ability to form reactive species, catalytic and photocatalytic activity, leading to reactive oxygen species (ROS) production when irradiated by ultraviolet rays, as has been described for TiO 2 and ZnO particles (Clemente et al. 2013;Sharma et al. 2020;Valério et al. 2020). The presence of coatings and addition of functional groups to the NPs surface alter their surface properties, improve their overall function, biocompatibility, or dispersion in aquatic media or reduce photocatalytic ROS production (Warheit 2008). For example, polyvinylpyrrolidone (PVP) coating of Au NPs is considered safer and more biocompatible than citrate coating (Zhou et al. 2009;Iswarya et al. 2016).
This information must be considered in relation to the characteristics of the exposure media (pH, presence of ions, and ligands, as mentioned above) which in reality differ greatly for in vitro and in vivo settings. The precise particle and exposure characterization is a key factor of each experiment (Nel et al. 2006;Warheit 2008;Wijnhoven et al. 2009).

Toxicodynamics of metallic nanoparticles in fish
There are several distinct toxic mechanisms by which NPs can impact aquatic organisms, which often act together and lead to similar results.
The most significant mechanism is the excessive production of ROS, overwhelming the cell antioxidant mechanisms, which can cause oxidative stress, leading to protein destabilization, lipoperoxidation of the cell membrane, DNA damage, causing genotoxic effects and cell death (Nel et al. 2006;Manke et al. 2013;Fu et al. 2014;Fard et al. 2015).
In many of the metallic NPs, the release of ions is a significant cause of adverse effects. The cytotoxic action of metal ions is caused by the complexation of metal cations with thiol groups of proteins and enzymes, which causes their inactivation (Sevcikova et al. 2011). Thiol groups are present in enzymes responsible for dealing with oxidative stress, such as lactate dehydrogenase and glutathione, which complicates the distinction of the two above mentioned toxicological mechanisms, when the enzyme activity is measured (Ulrich and Jakob 2019). In the case of silver ions, there is also a more pronounced specific toxicological mechanism in fish. Specific inhibition of Na + /K + -ATPase leading to blockage of active uptake of Na + and Clon gill basolateral cells, leading to disruption of osmoregulation and death (Morgan et al. 1997;Hogstrand and Wood 1998). However, Yue et al. (2016) show, that the inhibition of Na + /K + -ATPase was caused by both AgNO 3 and particulate Ag NPs (dissolution rate of Ag + was 2.37%, thus ions can be only accounted for 16% of observed effect).
The mechanical effect is often attributed to particles with a greater aggregation rate. In fish, the adhesion of NPs to the gills and a subsequential mechanical restriction of gas exchange was described by Ma and Lin (2013). In fish embryos, aggregation of NPs on a surface of the chorion can block the chorion pore canals and cause hypoxia, which manifests with a decreased hatching rate (Shih et al. 2016;Caloudova et al. 2018).
Due to the larger surface area, small dimensions, and high sorption capacity, NPs can also act as a Trojan horse, binding contaminants to their surface, transporting them into organisms, where they can be released (Deng et al. 2017). This effect was reported by Zhu et al. (2011) in Haliotis sp. embryos exposed to tributyltin, and coexposed to TiO 2 NPs and tributyltin. While the TiO 2 NPs alone had no negative developmental effect at 2 mg·l -1 , the same concentration increased the toxicity of tributyltin 20-fold, causing a drop in the hatching rate and an increase in the presence of malformations compared to exposition of embryos to tributyltin only.
In the case of metallic nanoparticles with antimicrobial properties, such as ZnO NPs, TiO 2 NPs or Ag NPs, the adverse effect on the diversity and composition of the microbiome of the fish should be taken into consideration (Azizi-Lalabadi et al. 2019;Chupani et al. 2019).

Toxicokinetics of metallic nanoparticles in fish
It has been shown that NPs can be taken up by aquatic organisms and can adversely affect them at the level of algae, invertebrates, and fish (Gaiser et al. 2011;Skjolding et al. 2016). Possible routes of uptake of NPs in fish include absorption through the gill epithelia, the gut epithelia during dietary exposure and drinking, as well as skin epithelia Nam et al. 2014). From the above mentioned, the skin is thought to have the smallest impact, due to the secretion of protective mucus which can chelate charged NPs, as well as the lack of metal transporters compared to gill (Coello and Khan 1996;Handy et al. 2008). Yue et al. (2016) show endocytic uptake of Ag NPs and their storage in endosomes and lysosomes of cells of RTgill-W1 cell line. Nanoparticle localization was different from Ag + , ions were found predominantly in cytosol, associated with metallothionein-like protein fractions.
The theory of the gut as a viable route of uptake is supported by Gaiser et al. (2011) who exposed common carp for 21 days to nano-and microsized particles of silver and subsequently analyzed silver concentrations in the gills, intestine, blood, liver, gallbladder, kidney, and brain. Significant uptake was detected in the intestine, liver, and gallbladder. Higher concentrations of silver were observed in groups exposed to smaller particles. Chronic dietary intake of ZnO NPs did not cause accumulation of zinc in common carp (Cyprinus carpio) tissues, however, adverse effects on immune system and homeostasis, as well as nephrotoxic and hepatotoxic effects were connected to the ZnO NPs exposure (Chupani et al. 2017;2018a). Whether the fish digests food during the exposure plays an important role in the bioavailability and toxicity of ingested substances. Proteins and amino acids have been shown to complexate the Ag + and prevent adverse effects on rainbow trout gut cells (RTGC) in the case of Ag NPs, they were only partially effective in sustaining the metabolic activity and integrity of the cell and lysosomal membranes (Minghetti and Schirmer 2016). It is to be noted that any damage, such as erosion or inflammation of epithelia, disables natural barriers and enables facilitated transport of NPs to the blood circulation and internal organs. The said damage can be caused by the NPs themselves. Chupani et al. (2018b) found an increased rate of apoptosis in the intestinal epithelium as well as increased levels of proteins associated with cancerous cell survival in the muscular layer of the intestine of the common carp (Cyprinus carpio) exposed to a diet containing ZnO NPs for six weeks.

Trophic transport of metallic nanoparticles
Bioaccumulation (accumulation of the NPs in the organism) and trophic transport have been shown to occur within the food chain, raising concerns about biomagnification (increased concentration of NPs in the food chain due to predation) within the ecosystems (Nam et al. 2014;Uddin et al. 2020). Ates et al. (2015) reported trophic transport between both CuO NPs and ZnO NPs contaminated crustacean (Artemia salina) and goldfish (Carassius auratus), resulting in accumulation of both CuO NPs and ZnO NPs in the intestine, liver, and gills of the goldfish. Similarly, Zhu et al. (2010) proved that TiO 2 NPs can be transferred from daphnids to zebrafish by dietary exposure, but biomagnification did not occur. On the other hand, Chen et al. (2015) provided the evidence of TiO 2 NPs transfer from an alga (Scenedesmus obliquus) to the water flea (Daphnia magna) with a biomagnification effect. Yoo-iam et al. (2014) evaluated the bioaccumulation and biomagnification of Ag NPs in a model food consisting of green algae (Chlorella spp.), water flea (Moina macrocopa), blood worm (Chironomus spp.), and silver barb (Barbonymus gonionotus). The highest bioaccumulation factor was found in the algae, the lowest bioaccumulation factor was determined in the fish. The trophic transfer occurred only between the algae and the water flea and there was no evidence of biomagnification.

Silver
Silver NPs are among the most frequently used and researched NPs, thanks to their excellent thermal and electrical conductivity, catalytic properties, nonlinear optical properties, as well as antibacterial, antivirotic, antimycotic, antiinflammatory, and anticancer activity (Tian et al. 2007;Nowack et al. 2011;Gurunathan 2015;Shaalan et al. 2016). Silver NPs have a wide range of industrial and consumer applications in the production of electronics, construction, agriculture, food industry, healthcare, and medicine (in both treatment and diagnostics), as well as in household products, appliances, cosmetics, and textiles (Park et al. 2013;European Commission 2014;Geertsma et al. 2015;Zhang et al. 2016). The annual worldwide production of Ag NPs is estimated to be 320 tonnes (Nowack et al. 2011).
Silver ions belong amongst the most toxic metals to fish (Morgan et al. 1997;Hogstrand and Wood 1998), Ag + is attributed to a significant part of the toxic effect of Ag NPs, but not the whole, there is intrinsic toxicity of NPs (Bilberg et al. 2012;Lapresta-Fernández et al. 2012;Khan et al. 2015;Speshock 2018). Van Aerle et al. (2013) applied sequencing-based transcription-profiling on zebrafish embryos exposed to Ag NPs, bulk silver, and AgNO 3 for 48 h. Significant changes in gene expression were found in all exposition groups, with significant overlaps between the groups, particularly pathways associated with oxidative phosphorylation and protein synthesis. Ag NPs specific changes in gene expression were found, suggesting that Ag + is not the only originator of the adverse effect. This claim is supported by a study performed on adult zebrafish exposed to Ag NPs, and AgNO 3 for 48 h. Total silver body burden was significantly higher in groups exposed to Ag NPs and subsequent gene expression analysis revealed a variety of genes, that were differentially expressed only in Ag NPs expositions (Griffitt et al. 2009). The effect of Ag NPs exposition on gills has been studied thoroughly because gills are one of the main spots of Ag + toxicity. Garcia-Reyero et al. (2015) compared the effects of Ag NPs and AgNO 3 on gills of fathead minnow (Pimephales promelas). During the first 4 h, mucus production in all groups increased, which was followed by a decrease in mucus production after 24 h of exposure. Microarray analysis detected 109 differentially expressed genes shared by both AgNO 3 and Ag NPs, but there as many as 615 differentially expressed genes specific to Ag NPs. In a study by Hawkins et al. (2015), AgNO 3 and Ag NPs caused similar disruptions in gill structure and ionic regulation, including circulatory disturbances, and decreased activity of the Na + /K + -ATPase.
The developmental toxicity of Ag NPs to fish is well documented (Yeo and Kang 2008;Massarsky et al. 2013;Park et al. 2013;Xia et al. 2016). A dose-dependent increase in mortality, amount of non-hatched embryos, and presence of sublethal malformations, such as the presence of notochord deformations or pericardial oedema, bradycardia, and cardiac arrhythmia were found after 72 h exposition to Ag NPs. The distribution of Ag NPs was visualized by transmission electron microscopy in the brain, heart, yolk, and blood of the embryos (Asharani et al. 2008). Furthermore, the possibility of transgenerational transfer of silver has been shown in female zebrafish injected with 1 mg·l -1 10 nm PVP capped, in which the passage to eggs has been proven. The highest load of Ag NPs was found in the liver of the fish, in which the upregulation of inflammatory gene interleukin-1β, tumour necrosis factor alpha (TNF-α), and pro-apoptotic caspase genes was measured (Speshock et al. 2016).
Impairment of the neurological development, manifesting morphologically as a hypoplastic hindbrain, reduction of the eye and overall head size, plus cardiac defects, accompanied with alteration of the neural development-related genes expression was observed in zebrafish embryos exposed to Ag NPs up to 96 h post fertilization; no free Ag + was detected in exposure solutions. Also, higher accumulation of Ag was observed in the head area of the embryo (Xin et al. 2015). Another study brought evidence of reduction of the acetylcholinesterase (AChE) activity in the brain of adult zebrafish, exposed to 0.1 ppm of Ag NPs for 15 days. This leads to continuous muscle stimulation, manifested in altered responses and behaviour of the fish. Neurological changes were accompanied by necrotic and degenerative changes in the liver, as well as decreased activity of liver antioxidant enzymes, catalase, and superoxide dismutase, as a result of oxidative stress (Devi et al. 2015). Induction of oxidative stress was measured in zebrafish liver cells exposed to 120 nm Ag NPs for 24 h. This was accompanied by upregulation of the expression of a variety of endoplasmatic reticulum specific stress marker genes, pro-apoptotic gene Bax, and TNF-α. On the other hand, the expression of the tumour suppressor gene p53 was downregulated (Christen et al. 2013). The summary of studies carried out into the toxicity of Ag NPs to fish is listed in Table 1.

Gold
Gold NPs are popular due to their relatively easy synthesis, stability, catalytic and optoelectronic properties, low toxicity, and high biocompatibility. They are used in technical applications such as electronics, photovoltaics, catalysts, probes, and sensors. Gold NPs also play an important role in medicine, both in diagnostics (for example in tumour detection and imaging) and therapeutic applications -in drug delivery, or as photothermal agents (Chandra et al. 2013;Mahapatra et al. 2015;Patibandla et al. 2018). The estimated worldwide production of Au NPs ranges from 1-3 tonnes per year (Pulit-Prociak and Banach 2016).
Although the toxicity of Au NP is significantly lower than Ag NPs (Bar-Ilan et al. 2009;Lapresta-Fernández et al. 2012;García-Cambero et al. 2013), major adverse effects have been described as well. In a study by Patibandla et al. (2018), exposure of zebrafish embryos to gold nanorods lead to increased mortality, decreased hatching, and a decreased heart rate. A significant increase in the expression of oxidative stress genes as well as increased apoptosis occurred, suggesting an important role of ROS in the mechanism of Au NPs toxicity.
Similar adverse effect on the circulatory system manifested by the presence of heart oedemas and decreased heart rate was observed in zebrafish embryos exposed to Au NPs stabilized with PVP (Hlavkova et al. 2020). Upregulation of oxidative stress gene expression, as well increased levels of protein biomarkers, such as catalase, superoxide dismutase, and metallothioneins were found in adult zebrafish exposed to Au NPs for 96 h. Furthermore, decreased swimming speed and abnormalities in the swimming behaviour were observed (Botha et al. 2019). In addition, Dedeh et al. (2014) reported modulation of the expression of genes involved in oxidative stress, mitochondrial metabolism, and DNA repair, as well as altered neurotransmission due to increased brain and muscle AChE activity in adult zebrafish, exposed to Au NPs contaminated sediment for 20 days.
Genotoxic effects of Au NPs have been observed even at concentrations as low as 4 μg·l -1 in gilthead seabream (Sparus aurata) exposed to citrate and PVP coated Au NPs for 96 h. Nanoparticles have induced DNA damage (erythrocyte DNA strand breaks), the frequency of erythrocytic nuclear abnormalities was also increased (Barreto et al. 2019). Alteration of gene expression levels, mutations, and mitochondrial dysfunction was observed in zebrafish exposed to citrate capped Au NPs (sized 12 nm and 50 nm) in their diet, over the period of 36 and 60 days. More severe effects were observed in groups fed the smaller particles (Geffroy et al. 2011). The summary of studies carried out into the toxicity of Au NPs to fish is listed in Table 2.

Titanium dioxide
Of the group of metal oxides, titanium dioxide is the most commonly used. Piccinno et al. (2012) estimate its annual global production volume to be 3 000 tonnes. Titanium dioxide NPs -nanoparticles; PSS -polystyrenesulphate, PVP -polyvinylpyrrolidone has good electrical properties and catalytic activity, high light reflectance, and a refractive index. It's stable and insoluble in water (Sungur 2020). There are three crystal structures of TiO 2 , brookite, anatase, and rutile. Anatase is considered to be the most chemically reactive (Shi et al. 2013). TiO 2 is used as a photocatalyst in solar panels, in plastics, paints, pharmaceuticals, and even as a food colouring. Since TiO 2 is transparent and blocks UV radiation, it is frequently used in cosmetics, especially sunscreens (Klaine et al. 2008;Ray et al. 2009;Stark et al. 2015). TiO 2 is generally thought to be a safe, non-toxic material. Griffitt et al. (2008) performed acute, 48-h tests with 30 nm TiO 2 NPs at concentrations of up to 10 mg·l -1 on both adult and larval zebrafish, alongside with adult crustacean Daphnia pulex, juvenile water fleas Ceriodaphnia dubia, and algae (Pseudokirchneriella subcapitata, 96-h test). TiO 2 did not cause adverse effects in any of these assays. Similarly, Jemec Kokalj et al. (2019) found no adverse effects on survival, hatching, and sublethal endpoints in zebrafish embryos during an acute 96-h assay, using three anatase TiO 2 NPs (sized 4.9 nm, 30 nm, and 5.1 nm) and three titania Magnéli suboxides (sized 192 nm, 507 nm, and 795 nm) at concentrations of up to 100 mg·l -1 . Exposure of zebrafish embryos to TiO 2 NPs for 120 h showed no effect on the survival, hatching, and malformation rates, however, the average and maximum velocity and activity level of the larvae were affected already at 0.1 mg·l -1 . However, assays with chronic and subchronic duration bring evidence of adverse effects of TiO 2 , such as reproductive toxicity, namely a 29.5% loss of eggs, which was found after a 13-week exposure of zebrafish to 0.1 mg·l -1 anatase TiO 2 NPs (Wang et al. 2011). Hepatotoxic effects, likely as an effect of oxidative stress, were found in goldfish (Carassius auratus) and zebrafish exposed to solutions TiO 2 NPs as low as 0.01 mg·l -1 for 21 days, with a subsequent 14 day depuration period. Results of this study show degeneration of hepatocytes and concentration-dependent increase of malondialdehyde, indicating lipid peroxidation and significant glutathione-S-transferase (GST) activity increase at day 7 of exposure, however, at 21 days the GST activity decreased in both species as a result of prolonged stress (Diniz et al. 2013). Similarly, acute (2 days) and subchronic (14 days) exposures of juvenile benthic fish Prochilodus lineatus to TiO 2 NPs lead to the presence of degenerative alterations of hepatocytes and increase of glutathione after both exposures and GST activity after subchronic exposure. Furthermore, a decrease of the red blood cell count and monocytosis after acute exposure was noted. In the case of subchronic exposure, both red blood cell and white blood cell counts including lymphocytes were decreased, which points to immunotoxic effects of TiO 2 NPs (Carmo et al. 2019). This claim is supported by another study, in which fathead minnow was exposed to environmentally relevant concentrations of TiO 2 NPs (2 ng·g -1 or 10 μg·g -1 of body weight), and subsequently challenged with bacterial pathogens Aeromonas hydrophila or Edwardsiella ictaluri. TiO 2 NPs were found to be accumulated in the kidney and spleen significantly more compared to the liver and rest of the body, resulting in histopathological damage of these organs. Neutrophil phagocytosis was significantly affected, increased mortality and morbidity during bacterial infection was observed (Jovanović et al. 2015). After subchronic (14 days) semi-static exposure of rainbow trout, histopathological changes in gills (oedema and thickening of the gill lamellae) and erosions of enterocytes, as well as decreased Na + /K + -ATPase activity in the gills and intestine were observed. Thiobarbituric acid reactive substances (TBARS) were increased in a concentration-dependent manner in the gills, intestine, and brain; the glutathione level was also increased in the gill, suggesting the effect of oxidative stress (Federici et al. 2007). The neurotoxic potential of TiO 2 has been shown by the decrease of muscular AChE activity observed in juvenile Prochilodus lineatus in two separate studies (Miranda et al. 2016;Carmo et al. 2019). The summary of studies carried out into the toxicity of TiO 2 NPs to fish is listed in Table 3.

Conclusion
With the rise of nanotechnology, the importance of thorough nanotoxicological research is growing, because the properties and behaviour of NPs differ from their bulk and ionic counterparts both in vitro and in vivo. A proper characterization of the determining properties of NPs, such as their chemical composition, crystal structure, size, shape, presence of coatings, surface chemistry, charge, and solubility, is a key step of every study. When assessing the overall safety of NPs, an important aspect to consider is the different behaviour of NPs in standardized test media and in the environment, in which NPs interact with the substances present -the natural organic matter and inorganic materials. The toxicity of NPs is thought to be generally lower in the environment than in the laboratory settings, but a cautious approach to nanotechnologies is warranted.