Dietary exposure to titanium dioxide nanoparticles in rainbow trout, (Oncorhynchus mykiss): no effect on growth, but subtle biochemical disturbances in the brain
- First Online:
- Cite this article as:
- Ramsden, C.S., Smith, T.J., Shaw, B.J. et al. Ecotoxicology (2009) 18: 939. doi:10.1007/s10646-009-0357-7
- 763 Views
Our laboratory recently reported gut pathology following incidental ingestion of titanium dioxide nanoparticles (TiO2 NPs) during aqueous exposures in trout, but there are almost no data on dietary exposure to TiO2 NPs in fish. The aim of this experiment was to observe the sub-lethal effects of dietary exposure to TiO2 NPs in juvenile rainbow trout (Oncorhynchus mykiss). Stock solutions of dispersed TiO2 NPs were prepared by sonication without the use of solvents and applied to a commercial trout diet. Fish were exposed in triplicate to either, control (no added TiO2), 10, or 100 mg kg−1 TiO2 NPs diets for 8 weeks followed by a 2 week recovery period where all fish were fed the control diet. TiO2 NPs had no impact on growth or nutritional performance, and no major disturbances were observed in red or white blood cell counts, haematocrits, whole blood haemoglobin, or plasma Na+. Ti accumulation occurred in the gill, gut, liver, brain and spleen during dietary TiO2 exposure. Notably, some of these organs, especially the brain, did not clear Ti after exposure. The brain also showed disturbances to Cu and Zn levels (statistically significant at weeks 4 and 6; ANOVA or Kruskal–Wallis, P < 0.05) and a 50% inhibition of Na+K+-ATPase activity during TiO2 NP exposure. Na+K+-ATPase activity was unaffected in the gills and intestine. Total glutathione in the gills, intestine, liver and brain were not affected by dietary TiO2 NPs, but thiobarbituric acid reactive substances (TBARS) showed up to 50% decreases in the gill and intestine. We conclude that TiO2 NPs behave like other toxic dietary metals where growth rate and haematology can be protected during sub-lethal exposures, but in the case of TiO2 NPs this may be at the expense of critical organs such as the brain and the spleen.
KeywordsTitanium dioxide nanoparticlesRainbow troutDietary exposure, Na+K+-ATPaseTBARSGlutathione
Engineered nanoparticles (NPs) are novel materials with at least one dimension <100 nm and there are concerns about the fate and ecotoxicity of these materials in the aquatic environment (reviews, Moore 2006; Nowack and Bucheli 2007; Handy et al. 2008a). There are many different types of manufactured nanomaterials (e.g., nanometals, carbon nanotubes, C60 fullerenes, composites, polymers) and the potential benefits to society are vast including applications in environmental remediation, drug delivery, electronics, building materials, textiles, and cosmetics (Aitken et al. 2006). However, data are needed for environmental risk assessment of these new materials; especially on uptake and biological effects (Owen and Handy 2007; Crane et al. 2008; Handy et al. 2008a, b).
TiO2 NPs are used in a range of commercially available products such as cosmetics, sunscreens, paint, and building materials (Aitken et al. 2006). The bulk form of ordinary TiO2 powder is not considered toxic and has been used as a negative control in respiratory toxicity studies (e.g., Warheit et al. 1997), and as an inert dietary marker for fish nutrition studies (e.g., Lied et al. 1982). However, several studies with fine and ultrafine (<100 nm) TiO2 have demonstrated respiratory toxicity in rodents (e.g., Ferin and Oberdörster 1985; Ferin et al. 1991; Oberdörster et al. 1992; Bermudez et al. 2004; Warheit et al. 2005).
Only a few ecotoxicological studies have been carried out using TiO2 NPs, and most of these have used waterborne rather than dietary exposures. Daphnia magna showed mortality during exposure to TiO2 NPs, depending on the method of NP preparation (Lovern and Klaper 2006). Zhang et al. (2007) showed that exposure to TiO2 NPs can also influence the uptake of other pollutants, with carp experiencing 146% more Cd uptake in the presence of TiO2 NPs compared to Cd only controls. Recently, our laboratory exposed juvenile rainbow trout to 0–1.0 mg l−1 TiO2 NPs for up to 14 days and found a range of toxic effects and organ pathologies including evidence of oxidative stress, respiratory toxicity, and erosion of the gut epithelium (Federici et al. 2007). The latter was probably caused by stress-induced drinking, and raised the possibility that TiO2 NPs might be toxic via the oral exposure route (see Federici et al. 2007).
To our knowledge, there are no detailed reports of dietary TiO2 NP exposure in fish and only a few reports on oral toxicity in mammals. Wang et al. (2007) exposed mice to either nano-sized (25 or 80 nm) or fine (155 nm) TiO2 particles by single oral gavage. No acute mortality occurred, but changes to serum biochemistry and liver pathology were observed. Toxic effects have also been seen in mice exposed to oral nano copper or zinc. Chen et al. (2006) used a single oral gavage of Cu–NPs (108–1,080 mg kg−1, 23.5 nm NPs) in mice and reported pathologies in the kidney, spleen and liver. Wang et al. (2006) observed lethargy, vomiting, diarrhoea, and some mortality in mice exposed to nano zinc via the oral route (5 g kg−1 body weight, 58 nm NPs). Despite the fact that these oral studies on rodents have used large doses of NPs, there are some concerns about the dietary hazard. Notably, in both rodents and fish metal accumulation in the internal organs raises the possibility that NPs may cross the gastrointestinal barrier. However, the effect of NPs on the nutritional performance of animals is currently unknown.
The aim of the current study was to provide some of the first toxicological observations on sub-lethal dietary exposure to TiO2 NPs in rainbow trout, and to enable some comparison with our previous experiments on aqueous exposure (Federici et al. 2007). Our goal was simply to establish whether or not this material was toxic via the dietary route compared to an unexposed control, and used a well established nutrition trial experimental design that is identical to our previous work on dietary metals (e.g., Cu, Handy et al. 1999; Shaw and Handy 2006). We adopted a body systems approach similar to Federici et al. (2007) and measured key areas of physiology such as growth, osmoregulation, haematology, biochemical responses of organs, and a range of nutritional parameters. In addition, because we have observed latent toxic effects of dietary metals in fish (e.g., Shaw and Handy 2006), this experiment also includes a period on a control diet at the end of the exposure to look for post-exposure effects.
Juvenile rainbow trout (n = 400) were obtained from Exmoor Fisheries, Somerset, UK, and held for 10 days in a stock aquaria with flowing, aerated, dechlorinated Plymouth tap water (see below). Fish were then transferred into a recirculation system (with 10% water renewal per day) consisting of nine 120 l experimental fibre glass aquaria (40 fish/tank; identical water conditions) and acclimated for 14 days prior to experiments. Fish were individually weighed (mean ± SEM, n = 360, 21.63 ± 0.15 g) and three tanks per treatment were randomly allocated. Fish were exposed in triplicate to one of the following treatments for 8 weeks: control diet (no added TiO2 NPs), 10 or 100 mg kg−1 dry weight feed TiO2 NPs (see below for diet formulation). This was followed by a 2 week recovery period where all fish were fed the control diet. The TiO2 NP concentration in the feed was selected after considering TiO2 NP toxicity in our previous waterborne exposure experiments (Federici et al. 2007) and our previous experience of dietary metal toxicity in fish (e.g., Shaw and Handy 2006). Fish were fed to satiation twice each day (1000 and 1600 hours), and behaviour was monitored during each feeding event. Care was also taken to ensure that all the feed added to the tanks was eaten. The self-cleaning design of the aquarium system also ensured that faecal waste was quickly removed from the tanks. Ti analysis of the water before and after feeding also confirmed that no TiO2 leached from the food. Background levels of Ti remained low in the water (<25 ng l−1). Separate leaching experiments with food pellets showed no release of Ti from the food (data not shown).
Water samples were taken each day for pH, temperature, and dissolved oxygen (all measured with a HACH HZ40d multi meter). Water samples were also collected three times each week (prior to feeding) for total ammonia, nitrite and nitrate (HI95715, HI 93707, and HI 93728 Hanna Instruments respectively). There were no treatment differences in water quality between tanks (ANOVA, P > 0.05). Values were (mean ± SEM, n = 68 or 30 samples) pH, 7.09 ± 0.04; temperature, 15.46 ± 0.07°C; oxygen saturation, 90.6 ± 0.48%; total ammonia, 0.29 ± 0.07 mg l−1, nitrite, 0.35 ± 0.06 mg l−1, and nitrate, 1.96 ± 0.18 mg l−1. The photoperiod was set to a 12 h light:12 h dark cycle. The electrolyte composition of the dechlorinated tap water used for the experiments was 0.3, 0.1, and 0.4 mmol l−1 for Na+, K+ and Ca2+ respectively. Fish were randomly sampled at the start of the experiment (initial fish), and then every 2 weeks during the experiment for haematology, plasma ions, tissue electrolytes, histopathology, biochemistry, nutritional performance and growth.
Titanium dioxide NP stock solution
The titanium dioxide NPs used here was from the same batch that has been previously characterised by our laboratory and are reported in Federici et al. (2007). The preparation of stock solutions, and confirmation of the level of dispersion, was carried out according to Federici et al. (2007). Briefly, dry powder of TiO2 NPs (“Aeroxide” P25 TiO2, DeGussa AG, supplied via Lawrence Industries, Tamworth, UK) made of (revised manufacturer’s information); crystal structure of approximately 25% rutile and 75% anatase TiO2, purity was at least 99% TiO2 (maximum impurity stated was 1% Si), and an average particle size of 21 nm with a specific surface area of 50 ± 15 m2 g−1. Chemical analysis of stock solutions revealed no metal impurities and the batch purity was high (data not shown), with a measured mean primary particle size of 24.1 ± 2.8 nm (mean ± SEM, n = 100 electron microscope images, see Federici et al. 2007). A 10 g l−1 stock solution of TiO2 NPs was made (no solvents) by dispersing the NPs in ultrapure (Millipore) water with sonication (bath type sonicator, 35 kHz frequency, Fisherbrand FB 11010, Germany) for 6 h.
The control diet was a commercial fish food; Advanced Fish Feed Trout Excel 18 (2 mm pellets), with fish progressing onto a mixture of this feed and Trout Excel 30 (3 mm pellets) at week 5 as their body size increased. Proximate composition of the diets was (% of dry diet from manufacturer’s guidelines, Trout Excel 30 diet in brackets): lipid 18 (21); protein 50 (46); ash 8 (8); fibre 1 (1); phosphorus 1.2 (1.2). In order to obtain experimental diets, the NP stock solutions (above) were sonicated for 8 h, and then either 1 or 10 ml of stock solution was added to 49 or 40 ml of ultrapure water to make a 0.2 or 2.0 g l−1 TiO2 NP dilution that could be sprayed on to the food for the 10 and 100 mg kg−1 treatments respectively. The approach of spraying diets with metal solutions is a well established method in nutritional ecotoxicology (e.g., Handy et al. 2005; Shaw and Handy 2006) and is highly relevant to mimic the effects of materials known to adsorb onto the surfaces of prey organisms. The diluted TiO2 solutions were sonicated for a further 15 min just before spraying to ensure even delivery of the material through the spray nozzle. One kg of commercial feed was place in a commercial food mixer (Kenwood Catering Professional food mixer XKM810) and gradually sprayed with the appropriate TiO2 NP solution. The TiO2 NP immediately coated the feed, and was then sealed in by spraying the food with a 10% bovine gelatine (BDH, Poole, UK) solution. The gelatine coat was allowed to dry, after which the feed was transferred into airtight containers for storage. The control diet was prepared in exactly the same way, except that the TiO2 solution was replaced by an equal volume of ultrapure water. Titanium metal concentrations in the diets were confirmed by ICP-OES following nitric acid digestion (as in tissue ion analysis below) and were 5.4 and 53.6 mg kg−1 feed weight of Ti metal respectively. Our calibrations showed that Ti metal forms 59.9% of the TiO2 (data not shown) and equates to recoveries of 90 and 89% of the nominal TiO2 NP concentrations in the 10 and 100 mg kg−1 TiO2 NP diets respectively.
Growth and nutritional performance
Growth and nutritional performance were measured according to Handy et al. (1999) with minor modifications. Briefly, food intake was calculated for each tank by weighing food containers before and after feeding. All fish were individually weighed at the start of the experiment, and every 2 weeks thereafter. Individual fish weights were used in growth rate calculations. Specific growth rate (SGR), feed conversion ratio (FCR), and feed conversion efficiency (FCE), were calculated as previously described (Handy et al. 1999) for: (1) the TiO2 NP exposure phase (weeks 0–8), (2) the recovery phase (weeks 8–10), and (3) the entire experiment (weeks 0–10). Condition factor and hepatosomatic index (HSI) for each fish was also determined (Handy et al. 1999). The spleen index (SI% = spleen weight (g)/body weight (g) × 100) was also measured. Two fish per tank (6/treatment, n = 5 for initial fish) were also terminally anaesthetised with MS222 at weeks 0 and 8, and stored at −20°C for proximate composition of the whole fish (ash, lipid, protein and moisture) according to Handy et al. (1999).
Haematology and analysis of blood plasma
Haematology, plasma ions and osmometry were performed exactly as described in Smith et al. (2007). Briefly, two fish were randomly collected from each tank (six fish/treatment and initial fish) at weeks 0, 2, 4, 6, 8, and 10 and carefully anaesthetised with buffered MS222. Whole blood was collected via the caudal vein into heparinised syringes, and the fish weight and total length was recorded. Haematological measurements included haematocrit (Hct), haemoglobin concentration (Hb), and calculated mean erythrocyte cell volume (MEV), and mean erythrocyte haemoglobin content (MEH) according to Handy and Depledge (1999). Whole blood (20 μl) was fixed in Dacie’s fluid for red and white blood cell counts. The remaining blood was centrifuged (13,000 rpm for 2 min, Micro Centaur MSE), and the serum collected. The serum was stored at −20°C until subsequent analysis of plasma ions and osmometry as described in Smith et al. (2007). Plasma protein was determined using the Bio-Rad (Bradford) protein assay kit II, and plasma glucose was measured according to Sigma Diagnostics procedure No. 315 glucose (Trinder).
Tissue ion analysis
Following blood sampling, fish were terminally anaesthetised with MS222 and dissected for tissue ion analysis. Gill, liver, intestine, spleen, skinned muscle from the flank, and whole brain were harvested, then processed for ion analysis as described in Smith et al. (2007) with minor modifications. Briefly, tissues were oven dried to a constant weight, then digested in 1 or 4 ml of concentrated nitric acid. In order to disperse TiO2 NPs in the tissue digests, a few millilitres of an appropriate Triton x-100 stock solution (prepared in ion-free ultra pure water) was slowly added to each of the digested tissue samples to achieve a final concentration of 2% Triton x-100 in each tube. Each sample was then diluted to a final volume of either 5 or 20 ml with ultra pure water and analysed for Ti, Cu, Zn, Mn, Ca, Na and K by inductively coupled plasma optical emission spectrometry (ICP-OES, Varian 725 ES). Analytical grade standards and reference materials were used throughout.
Biochemistry was performed exactly as described in Smith et al. (2007). Briefly an additional two fish were randomly collected from each tank (6 fish/treatment, and initial fish) at weeks 0, 2, 4, 6, 8, and 10 for biochemistry. Gill, liver, intestine, and whole brain were removed and immediately snap frozen in liquid nitrogen, then stored at −80°C until required. Tissues (about 0.5 g or the whole brain) were homogenised (Cat X520D with a T6 shaft, medium speed, Bennett & Co., Weston-super-Mare) in five volumes (2.5 ml) of ice-cold isotonic buffer [in mmol l−1; 300 sucrose, 0.1 ethylenediamine tetra acetic acid (EDTA), 20 (4-(2-hydroxyethyl)piperazine-1-ethane sulfonic acid (HEPES)), adjusted to pH 7.8 with a few drops of Tris (2-amino-2-hydroxylmethyl-1,3-propanediol)]. Crude homogenates were stored in 0.5 ml aliquots at −80°C until required. Tissue homogenates were analysed (in triplicate) for Na+K+-ATPase activity (15 μl of homogenate), and thiobarbituric acid reactive substances (TBARS, 40 μl of homogenate) and total glutathione (GSH, 20–40 μl of homogenate) exactly as described in Smith et al. (2007).
All data were analysed by StatGraphics Plus version 5.1. No tank effects were observed throughout the experiment, so data was pooled by treatment for statistical analysis. After checking for kurtosis, skewness and unequal variance (Bartlett’s test), data were tested for treatment or time effects by ANOVA followed by Fisher’s 95% least-squares difference, at 95% confidence limits. For non-parametric data, where data transformation was not effective, the Kruskal–Wallis test was used and differences located by notched box and whisker plots. Results are presented as mean ± SEM unless otherwise specified.
Dietary exposure to titanium dioxide nanoparticles
There was a small background incidence of mortality (2% in total), that is typical of juvenile trout in recirculation sysytems, and not associated with any one treatment. A total of nine mortalities were recorded during the experiment; 2, 4 and 3 fish from the control, 10 and 100 mg kg−1 TiO2 NP treatments respectively. Most of these fish were small and were probably subordinate fish that died as a result of aggression. The remaining fish did not show any visible signs of ill health and retained normal swimming behaviours throughout.
Growth and nutritional performance
Haematology and blood plasma analysis
Dietary exposure to TiO2 NPs did not cause any major haematological disturbances, with values remaining within the normal range for trout. For example, whole blood haemoglobin levels remained between 5 and 7 g dl−1, haematocrits between 30 and 35%, and red cell counts between 0.4 and 0.6 cells × 103 mm3 throughout the experiment. Total white blood cell counts were more variable, and ranged between 13 and 30 cells × 103 mm3 with no statistically significant treatment effect; apart from a transient increase in white blood cells at week 2 in the fish exposed to 10 mg kg−1 TiO2 NPs (ANOVA, P < 0.05). In the blood plasma, there were no treatment-dependent differences in osmolarity, glucose or Na+ concentrations (ANOVA, P > 0.05) with values remained between 252 and 342 mOsm kg−1, 2.9 and 4.9 mmol l−1, and 128 and 169 mmol l−1 respectively. However plasma K+ showed a small, but statistically significant increases at week 8 in both TiO2 treatments compared to controls (means, n = 5–6 fish); 4.3 ± 0.1, 4.5 ± 0.1, 4.9 ± 0.2 mmol l−1 for controls, 10 and 100 mg kg−1 TiO2 NP treatments respectively. This effect on K+ was lost by week 10.
Tissue electrolytes, trace metals and moisture content
Fish tissues (gill, intestine, liver, spleen, muscle, whole brain) were analysed for the major tissue electrolytes (Na+, K+, Ca2+) and some trace elements (Cu, Zn, Mn). There were no time or treatment effects on tissue K+ or Ca2+ (data not shown; ANOVA or Kruskal–Wallis, P > 0.05). Tissue Na+ did exhibit some transient changes which were statistically significant in the gill, liver and spleen (ANOVA or Kruskal–Wallis, P < 0.05). However no clear treatment-dependent trends were observed overall, with the Na+ data remaining within the normal range for rainbow trout. For example, Na+ concentrations for controls and 100 mg kg−1 TiO2 NP treatment were (means, n = 6, μmol g−1 dry weight); 206.4 ± 38.0 and 361.2 ± 117.6 (gill); 93.2 ± 4.0 and 152.1 ± 47.8 (liver); 66.4 ± 5.9 and 141.1 ± 67.1 (spleen).
Na+K+-ATPase, TBARS and total glutathione
Total glutathione (GSH) levels were measured in the gill, intestine, liver and brain homogenates with only the gill showing statistically significant changes in GSH. Levels of total glutathione in the intestine, liver and brain remained stable throughout with no treatment-dependent effects (ANOVA, P > 0.05). Values ranged between 0.49 and 2.13; 1.68 and 3.73 and 0.68 and 1.92 μmol g−1 wet weight tissue for intestine, liver and brain respectively. Following exposure to 100 mg TiO2 NPs for 8 weeks, the gills displayed a statistically significant decrease in GSH compared to all other treatments and the initial fish (ANOVA, P < 0.05). Glutathione levels in the gills at week 8 were (means, n = 6): 1.28 ± 0.13, 1.46 ± 0.13, 1.24 ± 0.12 and 0.92 ± 0.08 μmol g−1 wet weight tissue for initial fish, control, 10 and 100 mg kg−1 TiO2 NP treatments respectively. This treatment-effect was lost after fish were returned to the control diet, and by week 10 no statistical differences were observed between treatments (ANOVA, P > 0.05; means, n = 6): 1.30 ± 0.12, 1.24 ± 0.16, and 1.20 ± 0.10 μmol g−1 wet weight tissue for control, 10 and 100 mg kg−1 TiO2 NP treatments respectively.
This study is one of the first reports of dietary exposure to NPs in fish, and we show that juvenile rainbow trout will eat diets containing TiO2 NPs, and can accumulate the Ti in the gut and other internal organs. Despite Ti accumulation, the fish had relatively normal growth rates, suggesting that nutritional performance was protected even though biochemical disturbances occurred in other organs such as the brain.
Dietary exposure protocols for TiO2 NPs
In our experiment we used a commercially available NP (“Aeroxide” P25 TiO2 NPs) because of the practical value to hazard assessment of using a material that is found in many commercial products. We decided not to include a bulk TiO2 (ordinary TiO2 powder) as a “particle size” control (see discussions in Federici et al. 2007; Crane et al. 2008) because the proportions of different crystal structures, stoichiometry, zeta potential, surface area, aggregation kinetics, chemical reactivity, the presence of different impurities associated with manufacturing processes, and many other properties between the P25 TiO2 particles we used and ordinary TiO2 powders are different (e.g., Štengl et al. 2007; Warheit et al. 2007; Behnajady et al. 2008; Uzunova-Bujnova et al. 2008). It would therefore be extremely difficult to identify a particle size effect from all these other differences between commercially available bulk TiO2 powders and the commercial TiO2 NP used in our experiment. In addition, animal feeds already contain a myriad of different particulate matter (the food matrix) and it is not possible to control the particle size of the food itself.
There are also special ethical considerations for dietary TiO2 experiments, which are different from aqueous studies. Ordinary TiO2 powders have been used for many years as an inert marker in fish nutrition studies (e.g., Lied et al. 1982; Weatherup and McCracken 1998; Mamun et al. 2007), and are considered to be non-toxic in the food. A typical TiO2 powder inclusions of 1% of dry matter (i.e., 10 g kg−1 food) is used in fish foods; this is orders of magnitude higher than the TiO2 NP inclusion in our experiment. It would therefore seem to be ethically questionable to repeat experiments with bulk TiO2 powders (and use more live fish), when the nutritional safety of the bulk material is already established.
Furthermore, a bulk TiO2 control may also be uninformative in a dietary study because of the background of natural titania already in the animal feed ingredients. This type of problem is well known for studies on dietary iron where the metal is so abundant in the earth’s crust and in all the feed ingredients (e.g., fish meal) that it is technically impossible to make an iron-free basal diet (e.g., Carriquiriborde et al. 2004). The same situation applies to TiO2, and we must accept that the experimenter has limited control over the titania levels in the basal diet. Attempts to dialyse or otherwise remove the background Ti are also problematic, as any harsh treatment would compromise the nutritional quality of the food (e.g., accidental removal of other trace metals and vitamins).
Dietary TiO2 NP exposure and titanium accumulation in the tissues
The exposure can be regarded as sub-lethal, with a background 2% mortality (9 out of 400 fish died, no treatment-effect) typical of spontaneous losses of juvenile trout in aquaria (e.g., Handy et al. 1999). Dietary Ti exposure was verified by the measured Ti in the feed, the fact that the fish ate the food (Fig. 2), and that measurable increases in Ti metal were found in the tissues of the fish (Fig. 1), but not in the water.
There are almost no reports of background Ti levels in juvenile rainbow trout. In this study values in the control fish ranged between about 0.8–7 nmol g−1 dw, depending on the tissue examined (Fig. 1), and are broadly within the wide range reported for fish and shellfish (nmol–μmol g−1 levels, Bustamante and Miramand 2005; about 0.2 nmol g−1 in Atlantic salmon, Salmo salar, Dubé et al. 2005). The values in controls (Fig. 1) are lower than those we previously reported for trout (Federici et al. 2007), but this is easily attributed to differences in the supply of stock trout and the natural background of titania. In the latter study, the fish were obtained from a farm with higher natural Ti levels in the environment, and the fish were eating a commercial farm food containing more natural Ti. Nonetheless, trout showed measurable Ti accumulation in the gill, gut, liver, brain and spleen during dietary TiO2 exposure compared to the controls in this study (Fig. 1). The levels of accumulation remained in the nmol g−1 range, despite the large mg levels in the food, suggesting that only a small fraction of the dietary dose was absorbed. This is consistent with other dietary metal studies (reviews, Clearwater et al. 2002; Handy et al. 2005). For Ti this is perhaps no surprise, given the use of bulk TiO2 powder as an inert digestibility marker where only 1% or less of the ingested dose is accumulated (Vandenberg and De La Noüe 2001; Richter et al. 2003). The behaviour of the NPs during the diet preparation also suggests low bioavailability. The solutions used to spray the NPs onto the food were initially dispersed (as reported in Federici et al. 2007), but the material (not surprisingly) quickly aggregated onto the surface of the food matrix.
Dietary metals often give a characteristic accumulation pattern, with metal uptake into the gut mucosa (i.e., the route of entry), and then transfer via the hepatic portal vein to the liver, and finally to other internal organs (review, Handy et al. 2005; e.g., dietary Cu, Shaw and Handy 2006; Hoyle et al. 2007). Dietary TiO2 NP exposure also seems to fit this general pattern (Fig. 1). However, Ti did not clear quickly from all of the tissues after exposure (Fig. 1), with the brain, liver, intestine and gill also showing elevated Ti concentrations in treated animals compared to controls at the end of the experiment. This is similar to the findings in mice which do not clear Ti from the tissues 2 weeks after a single oral exposure to TiO2 NPs (Wang et al. 2007).
Growth and nutritional performance
Fish from all treatments showed a steady weight gain and cumulative food intake, with no adverse effects of either TiO2 NP inclusion (Fig. 2). There were also no treatment effects on mean ration size, SGR, FCR, FCE, condition factor or HSI throughout the experiment. These observations indicate that dietary TiO2 NPs do not adversely affect growth or nutritional performance in rainbow trout at the inclusion levels and exposure times used here. This is similar to our previous studies on dietary metals where many sub lethal toxic effects can occur, but are not necessarily reflected in a loss of growth (e.g., Cu, Handy et al. 1999; Shaw and Handy 2006; Hoyle et al. 2007). Indeed, Handy et al. (1999) and Clearwater et al. (2002) argue that fish will often adopt a strategy where growth is protected, with consequent sub-lethal effects on other physiological processes. The loss of carcass lipid from the fish fed 100 mg kg−1 TiO2 NPs is consistent with this hypothesis (e.g., utilised more lipid than controls as part of a metabolic strategy to maintain growth, Handy et al. 1999), rather than as a result of lipid peroxidation (not observed) .
Several studies have reported the potential for TiO2 NPs to cause oxidative stress (e.g., fish cells, Reeves et al. 2008; trout in vivo, Federici et al. 2007). However, TBARS did not increase, but instead showed statistically significant decreases in the gill and intestine during TiO2 NP exposure (Fig. 5). This phenomenon has also been observed during waterborne SWCNT exposure in trout (Smith et al. 2007), and in the absence of changes in the total glutathione pool, suggested that the fish were probably up regulating other anti-oxidant defences to cause a fall in TBARS. For fish, much of this anti-oxidant capacity comes from the food (Baker et al. 1998) and the continuation of food intake during this study probably enabled some critical protection from the oxidising effects of TiO2.
Haematology and ionic regulation
Dietary TiO2 NP exposure had no effect on haematology, or plasma ions (apart from a small transient increase in plasma K+), and the tissue electrolytes (Na+, K+, Ca2+) were normal. Na+K+-ATPase activity was also normal in the gill and intestine (Fig. 4). Taken together these findings indicate that TiO2 NPs are not potent ionoregulatory toxicants via the dietary route, and this is consistent with the findings for waterborne exposure to the same NPs (Federici et al. 2007).
Effects on the brain
In this study, Ti accumulated in the brain of trout (Fig. 1), and this has also been reported in the brain of mice following gut gavage (Wang et al. 2007). Notably in this study (Fig. 1), and in Wang et al. (2007), the Ti accumulation in the brain persisted 2 weeks after the exposure; suggesting that the brain does not clear Ti or does so very slowly. In terms of brain as a target organ, this experiment on Ti is similar to dietary Hg exposure in fish where the brain tissue also accumulates metal during the exposure (Berntssen et al. 2003).
Brain tissue showed transient, but statistically significant depletions of Cu at week 6 in fish from both treatments compared to controls (Fig. 3). Federici et al. (2007) also noted transient depletion of tissue Cu during TiO2 NP exposure, especially in the brain. The cause of the Cu depletion remains uncertain, but effects of TiO2 NPs on Cu transporters in the brain cannot be excluded (e.g., inhibition of Cu-ATPases) given that the closely related Na+-K+-ATPase is also inhibited (Fig. 4). Notably, the Na+-K+-ATPase activity in the brain showed large decreases (around 50% inhibition, Fig. 4), and this effect was much greater than that observed with aqueous exposures to NPs in trout (TiO2, Federici et al. 2007; carbon nanotubes, Smith et al. 2007).
Zinc levels in the brain also showed some small, but statistically significant, transient increases. This was also noted in the brain after 14 days of waterborne exposure to TiO2 NPs (Federici et al. 2007). Elevations of brain zinc levels are implicated in many processes in the brain, including neuro-endocrine functions (e.g., Su et al. 1997) and memory formation (Takeda et al. 2008). We did not measure these neurological processes, but it is clear that Ti exposure may cause neurological effects via interference with Zn homeostasis. The mechanism of Ti effect on tissue Zn (and Cu) remains unclear, but the fact that two different routes of exposure (via the water or the food) can produce similar trace element disturbances in the brain requires further investigation.
Does the spleen protect the internal organs from TiO2 exposure?
The spleen showed a rapid rise in Ti in week 2, prior to Ti increasing in the other internal organs (Fig. 1), and spleens from exposed animals also increased in size during the exposure. One of the main functions of the spleen is to filter damaged cells and foreign material from the blood. The normal haematology in this study suggests the spleen continued to function. However, Ti levels in the spleens of treated animals returned to control levels by week 4, and this was coincident with Ti elevations in other internal organs (Fig. 1); suggesting the spleen was not able to protect the other internal organs from Ti exposure. Wang et al. (2007) also noted increased Ti levels in the spleen of mice following a single dose (gut gavage 5 g kg−1 of either 25 or 80 nm diameter TiO2 NPs).
There was also some transient depletion of Cu in the spleen (Fig. 3). Copper depletion has long been implicated in the alteration of cell surface markers on splenocytes, and therefore the modulation of spleen function (Flynn 1984). The effect of NPs on the function of the spleen and immunity in fish clearly requires further investigation.
Hazard assessment implications
This study demonstrates that fish can accumulate Ti from a dietary TiO2 NP exposure, and that a number of subtle physiological and biochemical disturbances occur. This information is collected against a long historic use of ordinary TiO2 powder (bulk TiO2) in fish nutrition (and the food industry generally) where no such toxic effects have been observed. This at least provides circumstantial evidence that there may be a different hazard from the commercially available nano TiO2 product used in this study compared to ordinary TiO2 powders used in foods.
Perhaps a more important question for risk assessors interested in ecological food chains, is whether or not the hazard presented by TiO2 NPs is more or less than other dietary metals (review, Handy et al. 2005). If this experiment is compared against dietary metal concentrations in other studies on salmonids where growth rate was maintained, but subtle biochemical disturbances occurred (e.g., Cu, 500 mg kg−1, Handy et al. 1999; Zn, 590–1,520 mg kg−1, Clearwater et al. 2002; Inorganic Hg, 100 mg kg−1, Berntssen et al. 2003); then TiO2 NPs might be considered more toxic than dietary Cu and Zn, and at least as toxic as Hg. The human health hazard from eating contaminated fish is also a concern, but the risk from accidental ingestion of TiO2 NP contaminated trout may be limited, because the Ti does not appear to accumulate in the edible muscle at the concentrations and time scales used here.
This research was funded by a grant to R. Handy from the Natural Environment Research Council UK (NE/E014348/1) and carried out whilst C. Ramsden and T. Smith were studying the masters in Applied Fish Biology at Plymouth University. C. Ramsden was partially supported by a grant from the States of Jersey. Technical assistance provided by Andrew Atfield and Mike Hockings is greatly appreciated. Dr. Andrew Fisher is thanked for help with ICP-OES.