Toxicity of titanium dioxide nanoparticles to rainbow trout
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Aquatic Toxicology 84 (2007) 415–430
Toxicity of titanium dioxide nanoparticles to rainbow trout
(Oncorhynchus mykiss): Gill injury, oxidative stress,
and other physiological effects
Gillian Federici, Benjamin J. Shaw, Richard D. Handy ∗
Ecotoxicology and Stress Biology Research Group, School of Biological Sciences,
University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK
Received 19 May 2007; received in revised form 10 July 2007; accepted 10 July 2007
Abstract
Mammalian and in vitro studies have raised concerns about the toxicity of titanium dioxide nanoparticles (TiO2 NPs), but there are very limited
data on ecotoxicity to aquatic life. This paper is an observational study where we aim to describe the toxicity of TiO2 NPs to the main body systems
of rainbow trout. Stock solutions of dispersed TiO2 NPs were prepared by sonication without using solvents. A semi-static test system was used
to expose rainbow trout to either a freshwater control, 0.1, 0.5, or 1.0 mg l−1 TiO2 NPs for up to 14 days. Exposure to TiO2 NPs caused some gill
pathologies including oedema and thickening of the lamellae. No major haematological or blood disturbances were observed in terms of red and
white blood cell counts, haematocrit values, whole blood haemoglobin, and plasma Na+ or K+ concentrations. Tissue metal levels (Na+ , K+ , Ca2+
and Mn) were generally unaffected. However, some exposure concentration-dependent changes in tissue Cu and Zn levels were observed, especially
in the brain. Exposure to TiO2 NPs caused statistically significant decreases in Na+ K+ -ATPase activity (ANOVA, P < 0.05) in the gills and intestine,
and a trend of decreasing enzyme activity in the brain (the latter was not statistically significant). Thiobarbituric acid reactive substances (TBARS)
showed exposure concentration-dependent and statistically significant (ANOVA or Kruskal–Wallis test, P < 0.05) increases (two-fold or more) in
the gill, intestine and brain, but not the liver during exposure to TiO2 NPs compared to controls. TiO2 NP exposure caused statistically significant
(ANOVA, P < 0.05) increases in the total glutathione levels in the gills, but depletion of hepatic glutathione compared to controls. Total glutathione
levels in the brain and intestine were unaffected. Liver cells exposed to TiO2 NPs showed minor fatty change and lipidosis, and some hepatocytes
showed condensed nuclear bodies (apoptotic bodies). Fish probably ingested water containing TiO2 NPs during exposure (stress-induced drinking)
which may have resulted in some areas of erosion on the intestinal epithelium. Overall we conclude that titanium dioxide nanoparticles are not a
major ionoregulatory toxicant, or haemolytic, at the concentration and exposure times used. Respiratory distress is a concern and sub-lethal toxicity
involves oxidative stress, organ pathologies, and the induction of anti-oxidant defences, such as glutathione.
© 2007 Published by Elsevier B.V.
Keywords: Titanium dioxide nanoparticles; Gill; Intestine; Haematology; Na+ K+ -ATPase; TBARS; Glutathione; Rainbow trout; Copper; Zinc
1. Introduction small size, shape, surface area, conductivity or surface chemistry
(Masciangioli and Zhang, 2003; Oberdörster et al., 2005) and
Manufactured nanomaterials have been defined as new mate- have found numerous applications in textiles, electronics, engi-
rials with at least one dimension between 1 and 100 nm neering and medicine (Masciangioli and Zhang, 2003; Aitken et
(Masciangioli and Zhang, 2003). This definition is not absolute al., 2006). Titanium dioxide nanoparticles (TiO2 NPs) are used
and toxicological studies have included work on aggregates of in a range of products including sun screen, cosmetics, paint,
particles with dimensions of a few hundred nanometers (Handy and building materials (Aitken et al., 2006).
and Shaw, 2007a). These new materials are manufactured to have Most of the current literature on the toxicity on NPs comes
unique physical or chemical properties which arise from their from mammalian studies on respiratory exposure, or from in
vitro assays with mammalian cells (see Handy and Shaw, 2007a
for review). Studies with fine and ultrafine (
416 G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430
1985; Ferin et al., 1991; Oberdörster et al., 1992; Bermudez Na+ K+ -ATPase activity) or oxidative stress (TBARS, gluta-
et al., 2002, 2004; Warheit et al., 2005, 2006). In particular, thione content).
Ferin and Oberdörster (1985) demonstrated that both anastase
and rutile forms of TiO2 were toxic, and that the retention time 2. Materials and methods
was long (half times of 51–53 days in the rat lung at low mil-
ligram doses). Oberdörster et al. (1992) also showed that the 2.1. Experimental design
level of lung inflammation in rats was associated with parti-
cle size, with the smaller ultrafine TiO2 causing more adverse Juvenile rainbow trout (n = 189) were obtained from Hatch-
effects. Others have since confirmed these general observations lands Trout Farm, Rattery, Devon, and held for 4 weeks in stock
in rodents. For example, Bermudez et al. (2004) exposed rats aquaria with flowing, aerated, dechlorinated Plymouth tap water
to a 10 mg m3 aerosol of TiO2 NPs (21 nm particle size) for (see below). Stock animals were fed to satiation on a com-
6 weeks and found pulmonary inflammation as evidenced by mercial trout food. Fish weighing 28.1 ± 0.4 g (mean ± S.E.M.,
increased numbers of macrophages and neutrophils in the lung, n = 189) were then graded into twelve experimental glass aquaria
with progressive epithelial and fibroproliferative change. Pul- (14 fish/tank), in a triplicate design (three tanks/treatment), and
monary particle overload was also seen, with 45–57% of the allowed to rest for 24 h prior to the commencement of the exper-
original doses retained in the tissue 52 weeks after the exposure. iment. Fish were exposed in triplicate to one of the following
In vitro studies also indicate some toxic effects of TiO2 NPs. Gurr treatments for 14 days using a semi-static exposure regime (80%
et al. (2005) exposed human bronchial epithelial cells (BEAS water change every 12 h with re-dosing after each change): con-
2B cells) to different crystal structures of TiO2 (10 or 20 nm trol (freshwater only), 0.1, 0.5 or 1.0 mg l−1 titanium dioxide
anatase, or 200 nm rutile forms). The anatase form of TiO2 was nanoparticles (TiO2 NPs, see below for stock solutions). These
much more toxic, inducing DNA damage, lipid peroxidation, concentrations of TiO2 NPs were selected after considering the
and micronuclei formation. TiO2 NPs may also be neurotoxic concentrations used to produce epithelial injury and oxidative
(Long et al., 2006), and can enter red blood cells in vitro (Rothen- stress in rodents (Bermudez et al., 2004; Warheit et al., 2005,
Rutishauser et al., 2006), although the precise details of the body 2006), and the experiment was designed to allow for sub-lethal
distribution and all the target organs for TiO2 NPs in mammals physiological effects over the exposure period rather than acute
remains uncertain. toxicity. The exposure time of 14 days was chosen to reflect
Ecotoxicological studies with TiO2 NPs are much more this dosimetry and enable some physiological or biochemical
limited with a few reports on invertebrates, but almost no infor- responses to the exposure, but also considering the ethical con-
mation on the toxic effects of TiO2 NPs to fish. Lovern and straint of using the minimum exposure period likely to achieve
Klaper (2006) exposed Daphnia magna to either filtered (sin- the scientific objectives. In mammalian respiratory toxicology
gle particles of 30 nm mean diameter) or unfiltered (‘clumped’ reference materials are available to perform positive controls
particles, between 100 and 500 nm) of TiO2 NPs in a 48 h acute with particles of known toxicity to rodent (carbon black or quartz
toxicity test. D. magna exposed to 0–10 mg l−1 filtered TiO2 particles are often used, e.g. Oberdörster et al., 1992). Reference
NPs showed increasing mortalities with increasing concentra- particle controls for fish are not available and we therefore did
tion (100% mortality at 10 mg l−1 ); whereas 0–500 mg l−1 of the not included these in the experimental design. The use of stan-
unfiltered TiO2 caused only 9% mortality. This study highlights dard reference particles for fish is currently under debate along
that the type of dispersion of the TiO2 NPs may influence toxic with other aspects of the aquatic test method (Handy and Shaw,
effect. Zhang et al. (2007) have also shown that TiO2 NPs can 2007b). It is unclear which reference materials (if any) should be
alter the uptake of other pollutants, and found that carp exposed selected for fish, especially since historic data on reference parti-
to cadmium in the presence of TiO2 NPs accumulated 146% cles are not available for aquatic species, and that some materials
more Cd than controls. However, there are no detailed studies used in mammalian studies (e.g. quartz or sand particles) may
on the toxic effects of TiO2 NPs on the different body systems not be good positive controls given that fish species are proba-
of fish. bly adapted to living on substrates containing large quantities of
In the current study, we aimed to make one of the these materials.
first toxicological assessments of TiO2 NPs in rainbow trout Fish were not fed 24 h prior to, or during the experiment
(Oncorhynchus mykiss). Similar to our recent work with car- in order to minimise the risk of the TiO2 NPs absorbing
bon nanotubes in trout (Smith et al., 2007), we adopted a to food or faecal material, and to help maintain water qual-
body systems approach relating structure and function in the ity. However, there was one exception. For ethical reasons
major organs to overview the toxic effects. Our aim was and to minimise aggressive behaviour associated with hunger,
to measure functional responses in key areas of physiology fish were fed once on day 10 just after the water change
(e.g. osmoregulation, haematology) as well as documenting and prior to re-dosing with TiO2 NPs. The food was eaten
organ pathologies and biochemical responses during aqueous immediately with no waste. Water samples were collected imme-
exposure. We therefore measured a range of end points includ- diately before and after each water change for pH (YSI 63
ing behaviours and mortality, gill injury, haematology and pH meter), total ammonia (HI 95715, Hanna Instruments), and
plasma ion concentrations, trace element profiles in the major oxygen saturation (YSI 85 D.O. meter). There were no treat-
organs, a suite of histopathological observations, as well as bio- ment differences in water quality between tanks (ANOVA,
chemical measurements relating to physiological function (e.g. P > 0.05). Values were (means ± S.E.M., n = 122 samples); total
G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 417
ammonia, 0.7 ± 0.2 mg l−1 ; pH, 7.11 ± 0.01; oxygen saturation, solutions were confirmed (Fig. 1), and renewing the test media
87.0 ± 0.25%; temperature, 14 ± 1 ◦ C. Photoperiod was 12 h every day would maintain the exposure, we also attempted to
light:12 h dark. The electrolyte composition of the dechlorinated measure TiO2 NP levels in each fish tank. Water samples were
Plymouth tap water used was 0.5, 0.1, 0.4, and 0.1 mmol l−1 analysed by measuring the absorbance peak associated with dis-
for Na+ , K+ Ca2+ and Mg2+ , respectively. Fish were randomly persed TiO2 NPs at 329 nm (Philips PU 8720 UV/VIS Scanning
sampled on days 0 (initial fish from the stock), 7, and 14 for Spectrophotometer, 2 nm band width) against 0–1000 !g l−1
haematology, plasma ions, tissue electrolytes, histopathology, TiO2 NP standards. Measured concentrations of dispersed
and biochemistry (see below). The entire experiment was sub- TiO2 NPs 10 min after dosing the tanks were 0.095 ± 0.006,
ject to ethical approval, and was independently monitored by a 0.490 ± 0.012, and 0.965 ± 0.021 mg l−1 TiO2 (mean ± S.E.M.,
fish health expert. n = 36/treatment) for the 0.1, 0.5, and 1.0 mg l−1 TiO2 treat-
ments, respectively. This represents 95, 98, and 97% of the
2.2. Titanium dioxide nanoparticle stock solution and target values. Water samples were measured again just before
dosing the 12 h water change to confirm that the TiO2 NPs had
remained in solution. Pooled data per treatment gave mean mea-
Powder form of ultrafine titanium dioxide nanoparticles were sured TiO2 concentrations of 0.089 ± 0.006, 0.431 ± 0.008, and
donated by DeGussa AG, Lawrence Industries, Tamworth, UK, 0.853 ± 0.014 mg l−1 (mean ± S.E.M., n = 126/treatment) over
and were the “Aeroxide” P25 TiO2 type (manufacturer’s infor- the 12 h period, equating to 89, 85, and 86% of the expected
mation); crystal structure of 75 % rutile and 25% anatase TiO2 , concentrations, respectively.
purity was at least 99% TiO2 (maximum impurity was 1% Si),
and average particle size was 21 nm with a specific surface area 2.3. Haematology and blood plasma analysis
of 50 ± 15 m2 g−1 . This product was chosen because exactly the
same nanoparticles had been used in many of the previous mam- Haematology and blood plasma analysis was performed
malian studies (e.g. Bermudez et al., 2004). Chemical analysis exactly as described in Smith et al. (2007). Briefly, two fish
of stock solutions revealed no metal impurities (data not shown), were randomly collected from each tank (six fish/treatment, and
and the batch purity was high. Stock solutions of dispersed TiO2 n = 7 for initial fish) at days 0, 7 and 14 and carefully anaes-
NPs were prepared by sonication after considering the recom- thetised with buffered MS222. Whole blood was collected via
mendations of the manufacturer and the findings of Matthews the caudal vein into heparinised syringes, and the fish weighed
(1990). Solvents were not needed. A series of trials exploring and total length was recorded. Haematological measurements
the optimum pH, ionic strength, and sonication times required included haematocrit (Hct), haemoglobin concentration (Hb),
to disperse the nanoparticles were conducted (data not shown). and calculated mean erythrocyte cell volume (MEV) and mean
A stock solution of 10 g l−1 TiO2 NPs was prepared by dis- erythrocyte haemoglobin content (MEH) according to Handy
persing the nanoparticles in ultrapure water (Millipore, ion free and Depledge (1999). Whole blood (20 !l) was fixed in Dacie’s
and unbuffered) with sonication for 6 h in a bath-type sonicator fluid for red and white blood cell counts. The remaining blood
(35 kHz frequency, Fisherbrand FB 11010, Germany), and sub- was centrifuged (13,000 rpm for 2 min, Micro Centaur MSE),
sequently for a further 30 min sonication immediately prior to and serum collected and stored at −20 ◦ C until subsequent anal-
dosing each day. ysis of plasma ions and osmometry as described in Smith et al.
Dispersion was confirmed by transmission electron (2007).
microscopy (TEM, JEOL 1200EXII). The dispersion was very
good at the final working concentrations (0.1, 0.5 and 1.0 mg l−1 2.4. Tissue ion analysis
TiO2 NPs) and the measured particle size was close to the
manufacturers information (mean ± S.E.M., n = 100 images; Tissue ion analysis was performed as described in Smith et al.
24.1 ± 2.8 nm). In addition, and despite extensive sonication, (2007) with minor modifications. Briefly, after blood sampling,
a few aggregates of nanoparticles were also observed in stock fish were terminally anaesthetised with MS222 and dissected
solutions (mean ± S.E.M., n = 8, 160 ± 24 nm), although the for tissue ion analysis. Gill, liver, skinned muscle from the
majority were individual particles. Dispersion was also con- flank, and whole brain were oven dried to a constant weight,
firmed by optical methods. Spectral scans of the sonicated TiO2 digested in concentrated nitric acid, then diluted to 16 ml with
solutions (250–700 nm, Perkin-Elmer UV/VIS Spectrometer, ultra pure (ion free) water. Samples were analysed for Ti, Zn,
Lambda Bio 20), gave the typical profile expected with a Cu, and Mn by inductively coupled plasma mass spectrometry
distinct peak at about 329 nm (Fig. 1), and was similar to (ICP-MS, Fisons Instruments, VG Plasma Quad PQ2 ICP-MS),
previous reports for TiO2 NPs (e.g. Lovern and Klaper, 2006). and for Ca (Varian SpectrAA50 ), Na and K (F-AA/AE Spec-
In order to achieve working concentrations of 0.1, 0.5, and trometer, Double Beam GBC902) by flame atomic absorption
1.0 mg l−1 TiO2 NPs in the fish tanks, each tank was dosed with spectroscopy. Analytical grade standards and reference materi-
0.2, 1.0 or 2.0 ml of the 10 g l−1 stock solution, respectively. The als were used, and mass spectrometry samples also included
aeration and water flow in the tank dispersed each dose around internal standards (1% cobalt and indium). For the Ti mea-
the tank within seconds in all experiments. Volumes of TiO2 NP surements by ICP-MS, we also did a series of pilot analytical
stock were adjusted on re-dosing to reflect the 80% water change chemistry studies to confirm the ICP-MS response to TiO2 NPs
every 12 h. Although the contents and composition of the stock compared to Ti metal standards (data not shown). We found that418 G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430
Fig. 1. Electron micrographs showing aggregated TiO2 NPs in a 10 mg l−1 stock solution prior to sonication (A) and dispersion to much smaller particle sizes after
sonication (B). Absorbance spectra from dispersed TiO2 NPs, (C) showing examples of 1.0 and 10.0 mg l−1 TiO2 solutions after sonicating for 2 h compared to clean
water from a fish tank, and Millipore water.
TiO2 NP standards, sonicated, and prepared using the Aeroxide domly collected from each tank (six fish/treatment, and n = 7
P25 particles described above gave good linear and repro- for initial fish) at days 0, 7 and 14 for biochemistry. Gill,
ducible calibrations using TiO2 NP concentrations ranging from liver, intestine, and whole brain were removed and immedi-
0 to 1000 !g l−1 (typical calibration, measured concentration in ately snap frozen in liquid nitrogen and stored at −80 ◦ C until
!g l−1 = 0.418x + 15.06, r2 = 0.98, where x is the expected con- required. Tissues (about 0.5 g or whole brain) were homogenised
centration) with a detection limit of 0.1 !g l−1 (or 2 nmol l−1 ) of (Cat X520D with a T6 shaft, medium speed, Bennett & Co.,
TiO2 . The calibrations performed with TiO2 NP solutions had a Weston-super-Mare) in five volumes (2.5 ml) of ice-cold iso-
shallower slope than calibrations performed with Ti metal stan- tonic buffer (in mmol l−1 ; 300 sucrose, 0.1 ethylenediamine tetra
dards, and the most reliable results were with isotopes 48 Ti and acetic acid (EDTA), 20 (4-(2-hydroxyethyl)piperazine-1-ethane
49 Ti for TiO NPs, indicating that ICP-MS responds differently sulfonic acid (HEPES)), adjusted to pH 7.8 with a few drops
2
to nanoparticles compared to pure metal solutions. Spike recov- of Tris (2-amino-2-hydroxylmethyl-1,3-propanediol)). Crude
ery tests for metal analysis gave good recoveries as previously homogenates were stored in 0.5 ml aliquots at −80 ◦ C until
reported (e.g. Hoyle et al., 2007). required. Tissue homogenates were analysed at the end of the
experiment (day 14) for Na+ K+ -ATPase activity to determine
2.5. Biochemical analyses possible effects on osmoregulation, thiobarbituric acid reac-
tive substances (TBARS) and total glutathione content to assess
Biochemical analyses were performed exactly as described oxidative stress. Assays were performed in triplicate exactly as
in Smith et al. (2007). Briefly, an additional two fish were ran- described in Smith et al. (2007) using 15, 40, and 20–40 !lG. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 419
of homogenate for Na+ K+ -ATPase activity, TBARS and glu- 2.7. Statistical analysis
tathione content, respectively.
Briefly, for Na+ K+ -ATPase activity each sample (15 !l, in All data were analysed using StatGraphics Plus Version
triplicate) was dispensed into 400 !l of both a K+ -containing 5.1. No tank effects were observed within treatments in any
buffer and a K+ -free buffer (plus 1.0 mmol l−1 ouabain), then experiments, so data were pooled by treatment for statistical
incubated at 37 ◦ C for 10 min. The reaction was stopped by analysis. After checking for kurtosis, skewedness and unequal
adding 1 ml of ice-cold trichloroacetic acid and 1 ml of colour variance (Bartlett’s test), data were tested for treatment or time
reagent was added to each tube (9.6%, w/v FeSO4 ·6H2 O, effects by ANOVA followed by Fisher’s 95% least-squares dif-
1.15%, w/v ammonium heptamolybdate dissolved in 0.66 M ference, at 95% confidence limits. For non-parametric data the
H2 SO4 ), and colour allowed to develop for 20–30 min at Kruskal–Wallis test was used and differences located by notched
room temperature. Absorbances were measured at 630 nm box and whisker plots. The Student’s t-test was sometimes used
(Dynex MRX microplate reader) against 0–0.5 mmol l−1 phos- to explore differences between pairs of data sets at the end of
phate standards. The TBARS assay was performed using the experiment.
40 !l of homogenate (in triplicate), added to a well of a
96-well microtitre plate (in triplicate), containing 1 mol l−1 buty- 3. Results
lated hydroxytoluene (2,6-di-O-tert-butyl-4-methylphenol), and
the final volume was made up to 200 !l with 1 mmol l−1 3.1. Aqueous exposure to titanium dioxide
phosphate buffered saline (pH 7.4). Following this, 50% nanoparticles
(w/v) trichloroacetic acid (TCA) and 1.3% (w/v) thiobar-
bituric acid (TBA) (dissolved in 0.3% (w/v) NaOH), were Aqueous exposure to TiO2 NPs did not cause mortality. Two
added, and the plate incubated at 60 ◦ C (60 min) and then fish died during the experimental period: one control fish was lost
cooled on ice. Absorbances were read at 530 and 630 nm due to fin nipping/aggression, and one fish exposed to 1.0 mg l−1
(Dynex MRX microplate reader), corrected for turbidity, and TiO2 at day 6 of exposure. The latter fish had signs of mucus
read against standards (0.5–25 nmol ml−1 1,1,3,3-tetraethoxy- secretion on the gills, consistent with gill pathologies (Fig. 2).
propane). In addition, some fish at the highest TiO2 concentration showed
Briefly, for total glutathione content (GSH), 20–40 !l of tis- loss of position holding in the water column for brief periods
sue homogenate, 20 !l of blank or standards (serial dilution towards the end of the experiment (hanging vertically in the
of a 200 !mol l−1 reduced glutathione solution), was added in water column for a few seconds–minute) which was indica-
triplicate to a microplate well containing 20 !l of 10 mmol l−1 tive of fatigue or abnormal buoyancy control. No other unusual
DTNB (5,5$ -dithiobis-(2-nitrobenzoic acid)), 260 !l of assay behaviours were observed.
buffer (100 mmol l−1 K2 HPO4 , 5 mmol l−1 EDTA, pH 7.5), Histological examination of the gills at the end of the experi-
and 20 !l of 2 U ml−1 glutathione reductase (Sigma Chemicals, ment (Fig. 2) showed normal anatomy in the freshwater controls,
Poole, UK). The reaction was commenced by the addition of with a normal background incidence of injuries on420 G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 Fig. 2. Gill morphology in trout after 14 days of exposure to (A) 0, (B) 0.1, (C) 0.5 and (D) 1.0 mg l−1 TiO2 NPs, using a semi-static exposure method. The gills of control fish show normal histology, while some oedema (osmotic swelling) of some secondary lamellae are present in other treatments (white arrows), and swollen mucocytes (black arrows). Fish exposed to TiO2 NPs showed some thickening of the primary lamellae that was absent in controls. This was more frequent at the 0.5 mg l−1 TiO2 concentration and with oedema led to some areas of fused and disrupted lamellae on some filaments (panel C). Scale bar 80 !m, sections were 8 !m thick and stained with Mallory’s trichrome. had evidence of hyperplasia on all gill filaments. This was not plasma Na+ or K+ , and values remained in the normal range for observed in any of the control fish. trout. However, there was a statistically significant increase in Total Ti concentrations in the tissues of the fish are shown in plasma Na+ in the highest TiO2 treatment at the end of the exper- Table 1. Overall, there was no clear treatment or time-dependent iment compared to controls (ANOVA, P < 0.05), and a decrease effects on Ti levels in the gill, liver or muscle, and only a transient in K+ in the 0.5 mg l−1 TiO2 NP treatment compared to the but statistically significant decrease (Kruskal–Wallis, P < 0.05) control (Table 2). Decreases in plasma Na+ were also generally in brain Ti concentrations compared to initial fish (but no expo- noted compared to the initial fish (Table 2). There were no effects sure concentration-effect). on plasma osmolarity (data not shown). 3.2. Haematology and plasma ions 3.3. Tissue electrolytes and trace metals TiO2 NP exposure did not cause any major haematolog- Fish tissues (gill, muscle, liver, whole brain) were analysed ical disturbances and values remained in the normal range for major tissue electrolytes (Na+ , K+ , Ca2+ ) and trace elements for trout (Table 2). There were no statistically significant (Cu, Zn, Mn). There were no statistically significant treatment- treatment-dependent effects on haematology compared to con- dependent effects on tissue Na+ (data not shown, ANOVA or trols, although there were some small but statistically significant Kruskal–Wallis test, P > 0.05), except for a depletion of muscle (ANOVA, P < 0.05) decreases in red and white cell counts, Hb, Na+ at day 7 at the highest TiO2 NP treatment compared to the and Hct compared to the initial stock fish. MEH showed a tran- control (!mol g−1 dry weight, mean ± S.E.M., n = 5–7: control, sient elevation at day 7 compared to values on day 14 in all the 448 ± 29; 1.0 mg l−1 TiO2 treatment, 317 ± 23), but this effect TiO2 treatments, but these were not significantly different from was lost by the end of the experiment. Tissue K+ was unaf- the within time point controls (Table 2). Mean erythrocyte cell fected by TiO2 exposure, except in the brain (Kruskal–Wallis volume (MEV) showed no time or treatment-dependent effects, test, P = 0.029), where K+ levels in the tissue of fish exposed to apart from a transient rise at day 7 compared to day 14 in the 0.1 1.0 mg l−1 TiO2 NP showed a transient rise compared to controls and 0.5 mg l−1 TiO2 treatments (Table 2). Similarly for plasma at day 7 (!mol g−1 dry weight, mean ± S.E.M., n = 5–7: control, electrolytes, there were no overall treatment-dependent trends in 225 ± 6; 1.0 mg l−1 TiO2 treatment, 303 ± 10). There were no
G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 421
Table 1
Tissue titanium concentrations in rainbow trout exposed to control (no TiO2 ), 0.1, 0.5, or 1.0 mg l−1 TiO2 NPs for up to 14 days
Tissue Time (days) Treatment
Control 0.1 mg l−1 TiO2 0.5 mg l−1 TiO2 1.0 mg l−1 TiO2
Gill 0 0.28 ± 0.01 (7)
7 0.37 ± 0.02 (6) 0.35 ± 0.02 (6) 0.32 ± 0.02 (6) 0.34 ± 0.02 (6)
14 0.29 ± 0.02 (6) 0.25 ± 0.05 (6) 0.32 ± 0.03 (6) 0.25 ± 0.02 (6)
Liver 0 0.20 ± 0.01 (7)
7 0.18 ± 0.01 (6) 0.18 ± 0.01 (6) 0.19 ± 0.01 (6) 0.19 ± 0.01 (6)
14 0.26 ± 0.01 (6)#,* 0.26 ± 0.01 (6)#,* 0.27 ± 0.01 (6)#,* 0.24 ± 0.01 (6)#,*,D
Muscle 0 0.20 ± 0.01 (7)
7 0.40 ± 0.05 (6) 0.34 ± 0.08 (6) 0.36 ± 0.05 (6) 0.30 ± 0.03 (6)
14 0.39 ± 0.06 (6) 0.34 ± 0.05 (6) 0.36 ± 0.06 (6) 0.33 ± 0.05 (6)
Brain 0 0.36 ± 0.02 (7)
7 0.14 ± 0.03 (6)# 0.20 ± 0.03 (6)# 0.29 ± 0.10 (6)# 0.17 ± 0.03 (6)#
14 0.47 ± 0.09 (6)* 0.44 ± 0.08 (6)* 0.42 ± 0.01 (6)* 0.42 ± 0.04 (6)*
Data are mean ± S.E.M. (n fish/treatment) expressed as !mol Ti metal g−1 dry weight of tissue.
# Significantly different from initial fish (ANOVA or Kruskal–Wallis, P < 0.05).
* Significantly different between day 7 and 14 within treatment (ANOVA or Kruskal–Wallis, P < 0.05).
D Significantly different from the previous TiO concentration within row (ANOVA or Kruskal–Wallis, P < 0.05).
2
treatment-dependent differences in tissue Ca2+ levels, except in 14, brain Cu and Zn levels had recovered with an overshoot,
the gill which showed a transient reduction in Ca2+ at day 7 in so that brain Cu/Zn concentrations in all the exposed fish were
the 0.5 mg l−1 treatment compared to controls (!mol g−1 dry higher than in the controls (Fig. 3). These effects on brain Zn
weight, mean ± S.E.M., n = 5–6: control, 767 ± 26; 0.5 mg l−1 and Cu occurred without similar changes in bulk electrolytes,
TiO2 treatment, 616 ± 16). There were no statistically signifi- such as Na+ in the brain, and were accompanied by a small
cant treatment-dependent effects in tissue Mn levels, except in but statistically significant increase in brain water content in
the muscle and brain. In the muscle, Mn concentrations were the fish exposed to 1 mg l−1 TiO2 NP at the end of the exper-
lower in fish from the 1.0 mg l−1 TiO2 NP treatment com- iment compared to controls (Kruskal–Wallis test, P = 0.0008).
pared to the control at the end of the experiment (ANOVA, Brain water content at the end of the experiment was (% water,
P = 0.001, !mol g−1 dry weight, mean ± S.E.M., n = 6: con- mean ± S.E.M., n = 5–6); 73.9 ± 0.6, 75.9 ± 2.4, 77.3 ± 1.0,
trol, 0.41 ± 0.02; 1.0 mg l−1 TiO2 treatment, 0.31 ± 0.03). In the 79.2 ± 0.4 for control, 0.1, 0.5 and 1.0 mg l−1 TiO2 treatments,
brain, only the 0.1 mg l−1 TiO2 NP treatment showed a statisti- respectively. There were also small changes in the water con-
cally significant increase compared to controls at the end of the tent of other tissues, although these did not follow any consistent
experiment (Kruskal–Wallis, P = 0.009, !mol g−1 dry weight, time or treatment-dependent pattern (ranges; gill, 75–79%; liver,
mean ± S.E.M., n = 4–6: control, 2.11 ± 0.04; 0.1 mg l−1 TiO2 73–77%; muscle, 76–81%). However, there was a statistically
treatment, 3.59 ± 0.70). significant increase (Kruskal–Wallis test, P = 0.028) in muscle
There were some interesting changes in tissue Zn and Cu water content at the highest TiO2 concentration compared to
levels (Fig. 3). The gills of fish exposed to 0.1 and 1.0 mg l−1 controls at the end of the experiment (% water, mean ± S.E.M.,
TiO2 NP showed statistically significant decreases in Cu levels n = 5–6); 78.8 ± 0.4, 79.3 ± 4.5, 79.6 ± 0.4, 80.3 ± 0.2 for con-
compared to controls at the end of the experiment (day 14), trol, 0.1, 0.5 and 1.0 mg l−1 TiO2 treatments, respectively.
and all TiO2 NP treatments showed a transient depletion of
gill Zn on day 7 (statistically significant), which had recovered 3.4. Na+ K+ -ATPase activity
by the end of the experiment (Fig. 3). This transient depletion
of zinc at day 7 also occurred in the muscle and brain, with Exposure to TiO2 NPs caused some decreases in Na+ K+ -
both these tissues showing an overshoot (statistically significant ATPase activity (Fig. 4). There was a concentration-dependent
increase above the controls, Kruskal–Wallis, P < 0.05) by the trend of decreasing Na+ K+ -ATPase activity in the gills and the
end of the experiment. Cu and Zn levels in the liver were not brain by the end of the experiment. Overall this trend was not
affected, apart from a statistically significant decrease in Zn lev- significantly different from the controls (ANOVA, P > 0.05),
els at the two highest TiO2 NP concentrations at day 7 (Fig. 3). although there was a significant difference between the 0.1
The most significant treatment-dependent changes in Cu and and 1.0 mg l−1 concentrations of TiO2 in the gill (Student’s
Zn levels were seen in the brain of exposed fish (Fig. 3). By t-test, P = 0.04). In the intestine, exposure to the lowest con-
day 7 fish from all TiO2 treatments showed about a two-fold centration of TiO2 caused a statistically significant decrease
decrease in brain Cu and Zn levels compared to initial fish or in Na+ K+ -ATPase activity compared to the control (Student’s
control fish (Kruskal–Wallis test, P < 0.0001). However, by day t-test, P = 0.027), but not at higher concentrations of TiO2 .422
Table 2
Haematological parameters and plasma ion concentrations in rainbow trout exposed to control (no TiO2 ), 0.1, 0.5, or 1.0 mg l−1 TiO2 NPs for up to 14 days
Parameter Time (days) Treatment
Control 0.1 mg l−1 TiO2 0.5 mg l−1 TiO2 1.0 mg l−1 TiO2
Haemoglobin (g dl−1 ) 0 6.46 ± 0.41 (7)
7 5.46 ± 0.18 (6) 5.89 ± 0.48 (6) 6.14 ± 0.48 (6) 6.78 ± 0.77 (6)
14 5.39 ± 0.51 (6) 5.09 ± 0.61 (6)# 4.26 ± 0.21 (6)*,# 5.44 ± 0.42 (6)
Haematocrit (%) 0 29.0 ± 0.64 (7)
7 24.6 ± 1.78 (5) 23.4 ± 2.31 (6)# 26.0 ± 1.70 (6) 24.0 ± 1.64 (6)#
G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430
14 23.3 ± 1.20 (6)# 19.2 ± 2.27 (6)# 22.8 ± 1.15 (5)# 24.7 ± 1.10 (5)
Red blood cell count (cells ×106 mm3 ) 0 0.72 ± 0.03 (7)
7 0.51 ± 0.03 (6)# 0.38 ± 0.06 (6)# 0.55 ± 0.03 (6)D,# 0.51 ± 0.04 (6)#
14 0.57 ± 0.05 (6)# 0.58 ± 0.03 (6)*,# 0.65 ± 0.08 (6) 0.55 ± 0.06 (6)#
White blood cell count (cells ×103 mm3 ) 0 15.97 ± 1.94 (7)
7 10.28 ± 0.98 (6)# 8.95 ± 1.65 (6)# 9.12 ± 0.58 (6)# 11.10 ± 1.80 (6)#
14 10.30 ± 1.32 (6)# 11.73 ± 1.73 (6)# 11.47 ± 1.09 (6)# 8.64 ± 1.25 (5)#
Mean erythrocyte haemoglobin content (MEH, !g cell−1 ) 0 9.08 ± 0.66 (7)
7 11.69 ± 0.90 (6) 12.41 ± 1.08 (4)# 11.07 ± 0.56 (6) 13.70 ± 1.68 (6)#
14 9.92 ± 1.35 (6) 9.05 ± 1.26 (6)* 7.02 ± 0.91 (6)* 10.25 ± 0.69 (6)*,D
Mean erythrocyte volume (MEV, nm3 ) 0 408.0 ± 17.5 (7)
7 512.7 ± 46.8 (6)# 592.3 ± 92.9 (6) 470.6 ± 22.5 (6) 489.9 ± 57.5 (6)
14 433.6 ± 55.2 (6) 337.8 ± 41.4 (6)* 327.5 ± 25.1 (6)* 508.1 ± 76.6 (6)D
Plasma Na+ (mmol l−1 ) 0 142.3 ± 2.84 (7)
7 121.6 ± 7.99 (5)# 115.8 ± 4.11 (6)# 130.3 ± 1.31 (6)D 128.8 ± 4.68 (5)
14 117.6 ± 6.46 (5)# 122.2 ± 6.20 (6)# 126.8 ± 5.30 (5)# 135.7 ± 4.48 (6)+
Plasma K+ (mmol l−1 ) 0 3.6 ± 0.20 (7)
7 3.9 ± 0.38 (5) 3.4 ± 0.17 (6) 3.1 ± 0.13 (6)+ 3.3 ± 0.33 (5)
14 3.2 ± 0.32 (4) 3.4 ± 0.22 (6) 3.3 ± 0.10 (5) 3.7 ± 0.27 (6)
Data are mean ± S.E.M. (n fish/treatment).
+ Significant difference from control within rows (ANOVA, P < 0.05).
* Significant difference between day 7 and day 14 within treatment (time-effect, ANOVA, P < 0.05).
D Significantly different from the previous TiO concentration within row (exposure concentration-effect within time point, ANOVA, P < 0.05).
2
# Significantly different from initial fish (stock fish at time zero, ANOVA, P < 0.05).G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 423
Fig. 3. Copper and zinc levels in the gill, liver, muscle and brain of trout after 7 and 14 days exposure to 0 (clear bar), 0.1 (grey bar), 0.5 (diagonal bars) or 1.0
(black bars) mg l−1 TiO2 NPs. Horizontal bars are initial (day 0) fish. Data are means ± S.E.M., !mol Cu or Zn g−1 dry weight of tissue, n = 6–7 fish, for Cu (panels
A–D), and Zn (panels E–H). Different letters within a time point indicate significant differences between treatments within each tissue (ANOVA or Kruskal–Wallis,
P < 0.05). # Significant time effect compared to initial fish (ANOVA or Kruskal–Wallis, P < 0.05). * Significant time effect within treatment compared to day 7 (ANOVA
or Kruskal–Wallis, P < 0.05).
3.5. TBARS and total glutathione centrations (Kruskal–Wallis, P < 0.05). However, TBARS in the
liver were unaffected by TiO2 NP exposure (ANOVA, P > 0.05).
Fish exposed to TiO2 NPs generally showed an increase Total glutathione levels were measured in the gill, intes-
in TBARS compared to controls at the end of the experiment tine, whole brain and liver homogenates. Only the gill and
(Fig. 5). In the gills, exposure to 1.0 mg l−1 TiO2 caused the liver showed concentration-dependent changes in total glu-
greatest increase in TBARS (112% increase, statistically signif- tathione content (ANOVA, P < 0.05), while the intestine and
icant compared to the control, ANOVA P < 0.05). The intestine brain showed no effect (Fig. 6). The gills showed a statistically
also showed a clear concentration-dependent increase in TBARS significant rise in total glutathione content at the highest TiO2
(Kruskal–Wallis test, P < 0.01), and the brain showed elevations concentration only at the end of the experiment. However, in the
in TBARS above that of the control at all TiO2 exposure con- liver progressive, concentration-dependent glutathione deple-424 G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430
sinusoid space, a few foci of lipidosis with minor fatty change
(Fig. 7). The occasional necrotic cell, cells with condensed
nuclear bodies that have the appearance of apoptotic bodies, and
cells showing nuclear division with condensed nuclear material
were noted, mainly at the highest TiO2 concentration (Fig. 7).
Histology of the intestine is shown in Fig. 8. Intestines from
control fish showed normal gross histology, but some injuries
were noted in fish exposed to titanium dioxide nanoparticles.
Two of the six fish examined that had been exposed to 0.1 mg l−1
TiO2 NPs showed some relatively minor histological change
with more diffuse definition of epithelial cells in places, and
fewer absorptive vacuoles. One of the fish showed occasional
vacuolation on the villi tips. Injuries were more progressive
in fish exposed to 0.5 mg l−1 TiO2 NPs. All six fish examined
showed occasional areas of erosion of the tips of the villi and
some fusion of villi (Fig. 8). One fish showed severe erosion of
the intestinal epithelium. Similar observations were made at the
highest TiO2 concentration (1.0 mg l−1 ) with all fish showing
occasional erosion of the villi tips, some fusion of the villi in
places, and with vacuolation present in the epithelium. Large
aggregates of nanoparticles were not seen in the gut sections
from any of the fish, but some of the sections with the most
severe injuries showed mucous residue in the lumen that also
had a milky colouration (presumably titanium dioxide mixed
with mucus secretions).
The brains of fish were removed whole, and the gross propor-
tions of the fore, mid and hind brain appeared normal. External
examination showed no evidence of gross inflammation (e.g.
no blistering of the dura or swelling of the bulk tissue) and no
evidence of cranial bleeding or blood vessel abnormality. Histo-
logical examination (not shown) confirmed that gross anatomy
was normal in all the brains examined from all treatments. Apart
from one fish in the 0.1 mg l−1 TiO2 NP treatment showing a few,
very localised individual necrotic cell bodies in one part of the
cerebrum, pathologies were absent. There was no evidence of
vacuolation, odema, cellular atrophy, or necrosis in brain tissue.
When blood vessels were evident in the sections, there was no
indication of swelling or bleeding. In sections where pituitary
Fig. 4. Na+ K+ -ATPase activity in crude homogenates from (A) gill, (B) intes-
gland was visible, the tissue appeared intact.
tine, and (C) whole brain of rainbow trout after 14 days exposure to 0 (clear
bar), 0.1 (grey bar), 0.5 (diagonal bars) or 1.0 mg l−1 (black bar) TiO2 NPs. 4. Discussion
Data are means ± S.E.M., n = 5–6 fish/treatment. Different letters indicate sig-
nificant differences between treatments within tissues (ANOVA or Student’s 4.1. Aqueous exposure to titanium dioxide nanoparticles
t-test to compare some individual data points, P < 0.05). Note the larger y-axis
scale in (panel C).
and gill injury
In the absence of suitable reference particle controls for fish,
tion was observed. The lowest concentration of TiO2 caused a we interpret toxic effects in terms of the presence or absence of
statistically significant rise in total glutathione (27% increase), TiO2 NPs (i.e. combined effect of TiO2 chemistry and nanome-
but the 0.5 and 1.0 mg l−1 TiO2 concentrations caused hepatic ter scale particle size), and only differentiate possible effects
total glutathione levels to decrease below that of the controls (37 due to particle size alone where data are available from pre-
and 65 % decreases, respectively). viously published work on TiO2 powder. Only two mortalities
were observed in the experiment (one control fish due to fin nip-
3.6. Histological observations on the liver, intestine and ping, and one fish from the highest TiO2 NP concentration). We
brain therefore argue that a 14 day exposure to TiO2 NP concentra-
tions between 0.1 and 1.0 mg l−1 is a sub-lethal experiment with
The livers of fish from the freshwater control showed normal trout. The lethal concentration of TiO2 NP for trout is unknown,
histology, and fish exposed to TiO2 NP showed some loss of but our observations would suggest the lethal concentrationG. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 425 Fig. 5. Thiobarbituric acid reactive substances (TBARS) in crude homogenates from (A) gill, (B) intestine, (C) whole brain and (D) liver of rainbow trout after 14 days exposure to 0 (clear bar), 0.1 (grey bar), 0.5 (diagonal bars) or 1.0 mg l−1 (black bar) TiO2 NPs. Different letters indicate significant differences between treatments within each tissue (ANOVA or Kruskal–Wallis, P < 0.05). Data are means ± S.E.M., n = 5–6 fish/treatment. Note the different y-axis scale for the liver data (panel D). is probably about an order of magnitude higher (10 mg l−1 or carbon nanotubes (SWCNTs) in trout. Smith et al. (2007) from more) over this time scale. Similarly, dispersed TiO2 at con- our laboratory exposed trout to 0–0.5 mg l−1 SWCNTs in iden- centrations around 0.5 mg l−1 cause only 9% mortality in D. tical conditions and found cumulative mortalities of five fish at magna over 48 h (Lovern and Klaper, 2006). In mice, dietary the 0.5 mg l−1 SWCNT concentration during a 10-day exposure doses of 5 g kg−1 of TiO2 NP also did not cause dose-dependent (compared to none over a longer time in this current study). lethality over 2 weeks (Wang et al., 2007). Together these obser- The apparently low acute toxicity of dispersed TiO2 NPs does vations suggest that large milligram or gram doses (per kg animal not mean there are no toxicological concerns, and we report a weight) may be needed to cause acute toxicity of dispersed TiO2 range of important sub-lethal effects in trout including organ NPs to animals. TiO2 NPs are also less toxic than single walled pathologies, biochemical disturbances, and respiratory distress. Fig. 6. Total glutathione content in crude homogenates from (A) gill, (B) intestine, (C) whole brain and (D) liver of rainbow trout after 14 days exposure to 0 (clear bar), 0.1 (grey bar), 0.5 (diagonal bars) or 1.0 mg l−1 (black bar) TiO2 NPs. Different letters indicate significant differences between treatments within each tissue (ANOVA, P < 0.05). Data are means ± S.E.M., n = 5–6 fish/treatment. Data are expressed as !mol of total glutathione g−1 wet weight of tissue. Note the different y-axis scales.
426 G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 Fig. 7. Liver morphology in trout after 14 days of exposure to (A) 0, (B) 0.1 (C) 0.5, and (D) 1.0 mg l−1 TiO2 NPs. The livers of control fish show normal histology with sinusoid space (S) present. Livers from TiO2 -exposed fish showed some loss of sinusoid space and foci of liposis (black arrow). Some cells showed nuclear fragments which appear to be apoptotic bodies (white arrows) or were in the early stages of necrosis/cellular atrophy. Note the area of necrotic cells (N) in (panel C). Scale bar = 50 !m, sections were 8 !m thickness and stained with Mallory’s trichrome. The gills showed injuries during exposure to TiO2 NPs (Fig. 2) et al., 2006), and this aspect may be worth future investigation that were accompanied by behavioural changes. The loss of posi- in fish. tion holding and swimming fatigue in the TiO2 NP exposed fish, mucus secretion, and gill pathology are all consistent with res- 4.2. Titanium accumulation during aqueous exposure piratory toxicity. This is similar to studies in mammals where epithelial injury and respiratory toxicity are reported during sim- Fish in this study did not accumulate TiO2 NPs in the internal ilar milligram instillations of TiO2 NPs (Warheit et al., 2005, organs during exposure (Table 1). In fish nutrition, TiO2 powder 2006). Increased mucus secretion by the gills is a common has been used as a digestibility marker because it is not appre- response to aqueous pollutants (Mallat, 1985), and we have ciably absorbed into the internal organs of fish (only about 1% shown gill injury and mucus secretion in trout exposed to SWC- of ingested dose, Vandenberg and De La Noüe, 2001; Richter et NTs (Smith et al., 2007). In the latter study, the respiratory al., 2003). The absence of accumulation in this study (Table 1) distress was more severe with SWCNT than at the same con- suggests the situation may be similar with TiO2 NPs in trout. The centrations of TiO2 used here. In the SWCNT experiment, gill tissue concentrations of Ti metal we report (0.1–0.5 !mol g−1 mucus precipitated SWCNT to help prevent direct exposure of dry weight, Table 1) are broadly similar to background lev- the sensitive gill epithelium (Smith et al., 2007). Although we els in shellfish (0.04–1.34 !mol g−1 dry weight, Bustamante did not see obvious large (mm) sized aggregates of TiO2 on gill and Miramand, 2005) and laboratory mice (1–4 !mol g−1 wet mucus in this study, the swollen mucocytes (Fig. 2) and some weight depending on the organ, Wang et al., 2007). Zhang et al. mucus secretion into the water indicate that mucous defences (2007) found Ti background levels in carp at the start of a TiO2 were active. This did not fully protect the gill as foci of oedema exposure of about 0.1 mg g−1 (about 2 !mol g−1 dry weight as and areas of branchial hyperplasia were observed (Fig. 2). These TiO2 ). Zhang et al. (2007) also reported that 6 g carp exposed gill pathologies are similar, but less severe, as those reported for to 10 mg l−1 TiO2 NP for 25 days accumulated a whole body SWCNT in trout (Smith et al., 2007). However, unlike SWCNT concentration of about 3.39 mg g−1 as TiO2 (or 70 !mol g−1 exposure, TiO2 NP treatment also caused branchial aneurisms, as TiO2 ). However, in the same study, TiO2 levels in washed suggesting either some local vascular wall injury in the branchial samples of muscle, gill and skin were much lower, suggest- capillary bed or interruptions of capillary flow. TiO2 NPs are ing the whole carp body measurements included large amounts known to cause microvascular dysfunction in rats (Nurkiewicz of surface bound TiO2 (surface adsorption rather than internal
G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 427
Fig. 8. Histology of the intestine at the end of the experiment. (A) Fresh water control showing normal intestine, (B) intestine from a fish exposed to 0.1 mg l−1 TiO2
NPs showing normal gross morphology but with more diffuse definition of epithelial cells, (C) intestine from a fish exposed to 0.5 mg l−1 TiO2 NPs showing erosion
and fusion (F) of intestinal villi, (D) intestine from a fish exposed to 1.0 mg l−1 SWCNT showing fusion (F) of intestinal villi and areas of vacuolation (V). Scale
bar = 20 !m, sections were 8 !m thickness and stained with Mallory’s trichrome.
uptake into tissues). Such phenomena have been reported before remained within the normal range for trout. Haematologi-
in metal ecotoxicology (review, Handy and Eddy, 2004). For cal values in the current study (haemoglobin concentration,
example, aluminium (Al) is a well-known surface acting toxi- 4–7 g dl−1 ; haematocrit, 19–29%, red cell counts 0.4–0.7 ×
cant that can produce indirect systemic toxic effects, but without 106 cells mm3 , white cell counts 9–16 × 103 cells mm3 ) were
appreciable internal accumulation of the metal (Handy and Eddy, around the normal range expected in a juvenile rainbow trout and
1989, 1991). A similar situation may apply to TiO2 NPs in our are similar to other reports from our laboratory (haemoglobin
2-week aqueous exposure. Further experiments on the effects concentration, 4–6 g dl−1 ; haematocrit, 21–29%, red cell
of TiO2 in the gill microenvironment, effects on gill injury, and counts, 0.4–0.7 ×106 cells mm3 , white cell counts 6–11 ×
the systemic signalling of such injury is needed to verify this 103 cells mm3 , Smith et al., 2007). The general trend of small
hypothesis. decreases in cell counts and haematological parameters over
time compared to the initial fish is expected, and is associated
4.3. Haematological and ionoregulatory disturbances with not feeding the animals during the experiment (e.g. Rios et
al., 2005).
Despite some pathological effects on gill (Fig. 2) and some Although the gills showed a clear, concentration-dependent
inhibition of branchial Na+ K+ -ATPase activity (Fig. 4), there decline in branchial Na+ K+ -ATPase activity at the end of the
were no major disturbances to salt and water balance or experiment (Fig. 4), this did not result in depletion of plasma Na+
haematology (Table 2). Although there were some treatment- or K+ . This is partly explained by the intestinal Na+ K+ -ATPase
dependent effects, for example, a small increase in plasma activity which tended to compensate for effects at the gill by
Na+ at the highest TiO2 exposure concentration compared to showing normal activity when branchial Na+ K+ -ATPase activity
controls (Table 2), these effects were small and the values was low (Fig. 4). Thus, fish could maintain NaCl absorption428 G. Federici et al. / Aquatic Toxicology 84 (2007) 415–430
from the gut lumen, even if the gill was partly compromised. Since these substrates are not normally present in the aquar-
Furthermore, plasma electrolytes are easily buffered by small ium water, then this reaction is most likely to occur on contact
changes in tissue Na+ permeability (Wood, 1992), and a small with the tissue. For example, by using the normal trace endoge-
loss of Na+ from the muscle (observed at day 7) would help neous H2 O2 production from cells, or in the case of the gut any
maintain plasma Na+ concentrations. Overall, the blood data residue organic matter in the gut lumen. Sonication is known
suggest that TiO2 NPs are not a major ionoregulatory toxicant to greatly increase the rates of these reactions in the presence
to trout, at least at the concentrations and exposure duration used of TiO2 particles (Hirano et al., 2005), and there is a concern
here. The general lack of clear persistent effects of TiO2 NPs on that sonication of the test solutions will inadvertently increase
the bulk electrolytes (Na+ , K+ and Ca2+ ) and tissue moisture oxidative damage (we did not sonicate the control solutions that
content of tissues (with the possible exception of the muscle had no added TiO2 NPs). However, this effect is unlikely in
moisture) also reflects this notion. our experiment because the TiO2 NPs were sonicated in ultra-
However, there were some effects on Cu and Zn homeosta- pure water before the TiO2 NPs had access to organic substrates
sis (Fig. 3). The gill, muscle, and brain in particular, showed (the surface of the fish or natural water). Nonetheless, it may
decreases in both Cu and Zn at day 7 which was followed by be a useful precaution to sonicate the control solutions as well
a net increase in these metals compared to controls by the end in future experiments. It is also worth considering that natural
of the experiment. This suggests that TiO2 exposure is caus- waters with large quantities of organic matter, or studies with
ing some initial disturbances to Cu and Zn regulation, but the mixtures of pollutants (especially chlorinated organic chemi-
fish are able to recover. The biggest effects were seen in the cals) could generate significant micromolar quantities of ROS
brain (Fig. 3). Similar effects were seen after aqueous expo- within seconds of sonication (Hirano et al., 2005). These risks
sure to SWCNT in trout (Smith et al., 2007), where there were of ROS generation during sonication need to balanced against
some exposure concentration-dependent losses of Zn and Cu the logic for dispersing the TiO2 NPs in every experiment.
from the gills, although not from any other organ. Clearly, Zn However, elevation of TBARS was also found in internal
and Cu homeostasis after TiO2 NP exposure requires further organs that had no direct access to the aquarium water. TBARS
investigation, especially in the brain. increased in the brain (Fig. 5) without measurable increases in
brain TiO2 levels, or TiO2 in other internal organs (Table 1),
4.4. Oxidative stress suggests some indirect mediation of oxidative stress to the inter-
nal organs. This could simply be diffusion of the highly mobile
TiO2 NP exposure caused an increase in TBARS in the hydroxy radical from sites of initial injury on the gills, and
gill, intestine and brain suggesting that the fish suffered from with blood circulation in a small trout of less than 30 s (Handy
oxidative stress (Fig. 5). In vitro studies have also demonstrated and Eddy, 2004) any ROS would be rapidly distributed around
that TiO2 NPs can cause lipid peroxidation (hamster embryo the body. A systemic hypoxia resulting in ROS also cannot be
fibroblast cells; Gurr et al., 2005) and this characteristic has excluded given the pathology in the gills (Fig. 2, and discussion
been exploited by using TiO2 NPs as a bactericide (Maness et in Smith et al., 2007). Elevation of tissue Cu levels (Fig. 3) could
al., 1999). The TBARS values we report are consistent with also contribute to oxidative stress (Hoyle et al., 2007). Alterna-
previous measurements in trout (e.g. 1–10 nmol mg−1 protein, tively, a cascade of inflammatory responses could be initiated
Carriquiriborde et al., 2004), and with our recent study on SWC- at the site of injury (the gill) that are rapidly mediated around
NTs (e.g. 2–8 nmol mg−1 protein in gill/intestine, Smith et al., the body by specific signalling pathways (e.g. by cytokines) as
2007). The elevation of TBARS in the gill and intestine does shown during respiratory TiO2 NP exposure in rodents (Driscoll,
not necessarily suggest that oxidative stress is the cause of gill 2000; Sayes et al., 2006). Clearly, nanoparticles do not have to
(Fig. 2) or intestinal pathology (Fig. 8). The elevation of TBARS be internalised to generate systemic oxidative stress.
(Fig. 5) was not sufficient to deplete tissue glutathione in the
gill, intestine or brain (Fig. 6), suggesting that other anti-oxidant 4.5. Histology of the liver and brain
defences would be available to buffer some of the reactive oxy-
gen species (ROS) generated by TiO2 NP exposure. Interestingly Titanium dioxide nanoparticles have previously been shown
in the liver, a concentration-dependent depletion (utilisation) of to cause apoptosis in mammalian cells (Rahman et al., 2002), and
total glutathione (Fig. 6) was associated with no increases in in the present study the livers of some fish exposed to TiO2 NPs
TBARS (Fig. 5), suggesting that the liver was using up anti- showed condensed nuclear bodies (probably apoptotic bodies,
oxidant defences to prevent oxidative stress (no rise in TBARS). Fig. 7). We made similar observations with SWCNT (Smith et
This is also consistent with the liver histology where only minor al., 2007) and therefore the concern raised by Smith et al. (2007)
fatty change was observed (Fig. 7). about cell cycle defects and the risk of tumour formation during
The effects of TiO2 NPs on TBARS in the gills, and intestine longer exposures to carbon nanotubes may also apply to TiO2
(see below), could be partly explained by direct contact of the NPs. The fatty change observed in the liver (Fig. 7) was relatively
tissue with TiO2 NPs. We assume that most of the ROS genera- minor, and probably would not impact upon liver function in the
tion is from the catalytic chemical properties of TiO2 NPs, which short term.
in the presence of light, can transfer electrons from substrates, Histological examination of the main regions of the brain
such as hydrogen peroxide (H2 O2 ) or chlorinated organic com- showed no overall gross pathology, although a few necrotic cells
pounds to generate the hydroxyl radical (Hirano et al., 2005). were observed. However, the brain did show an increase in waterG. Federici et al. / Aquatic Toxicology 84 (2007) 415–430 429
content during TiO2 NP exposure, elevation of TBARS (Fig. 5), a carried out while Gillian Federici was studying for an MRes
downward trend in whole brain Na+ K+ -ATPase activity (Fig. 4), degree in Aquatic Ecotoxicology. Technical assistance provided
and transient changes in K+ content. This suggests the brain by Andrew Atfield, Mike Hockings, Trevor Worsey, and espe-
is showing the onset of biochemical disturbances that had not cially Catherine Smith also on an MRes, is acknowledged. Dr.
yet manifested as organ pathology. TiO2 NPs have also been Andrew Fisher is thanked for help with ICP-MS.
shown to cause oxidative stress and injury to mouse microglial
cells in vitro (Long et al., 2006). We therefore cannot exclude
brain injury, behavioural, or neurological deficits during longer References
exposures to TiO2 NPs.
Aitken, R.J., Chaudhry, M.Q., Boxall, A.B.A., Hull, M., 2006. Manufacture and
use of nanomaterials: current status in the UK and global trends. Occup.
4.6. Intestine and risk of exposure via the gut Med. (Oxford) 56, 300–306.
Bermudez, E., Mangum, J.B., Asgharian, B., Wong, B.A., Reverdy, E.E.,
Our previous work with SWCNT showed that nanomateri- Janszen, D.B., Hext, P.M., Warheit, D.B., Everitt, J.I., 2002. Long-term pul-
als can cause a stress-induced drinking response that resulted monary responses of three laboratory rodent species to subchronic inhalation
of pigmentary titanium dioxide particles. Toxicol. Sci. 70, 86–97.
in SWCNT precipitation in the gut lumen and pathology of the Bermudez, E., Mangum, J.B., Wong, B.A., Asgharian, B., Hext, P.M., Warheit,
mucosa (Smith et al., 2007). This also appears to be the case with D.B., Everitt, J.I., 2004. Pulmonary responses of mice, rats, and hamsters
TiO2 NPs in trout. We observed an unusual milky colouration to subchronic inhalation of ultrafine titanium dioxide particles. Toxicol. Sci.
of the luminal fluid (presumably ingested TiO2 ) and intestinal 77, 347–357.
Bustamante, P., Miramand, P., 2005. Subcellular and body distributions of 17
pathology (Fig. 8). The pathology included erosion of the villi,
trace elements in the variegated scallop Chlamys varia from the French coast
fusion and vacuolation of the mucosa. Inflammation of the gas- of the Bay of Biscay. Sci. Total Environ. 337, 59–73.
tric mucosa has also been noted in mice during dietary TiO2 NP Carriquiriborde, P., Handy, R.D., Davies, S.J., 2004. Physiological modulation
exposure (Wang et al., 2007). When these pathologies in the trout of iron metabolism in rainbow trout (Oncorhynchus mykiss) fed low and
gut are considered, along with the elevation of TBARS in the high iron diets. J. Exp. Biol. 207, 75–86.
Driscoll, K.E., 2000. TNF" and MIP-2: role in particle-induced inflammation
intestine (Fig. 5), it seems probable that nutritional performance
and regulation by oxidative stress. Toxicol. Lett. 112/113, 177–184.
will be affected. These observations have broader implications, Ferin, J., Oberdörster, G., 1985. Biological effects and toxicity assessment of
not only for the future safe use of nanomaterials in aquafeeds, titanium dioxides: anastase and rutile. Am. Ind. Hyg. Assoc. J. 46, 69–72.
but also highlights the need for environmental risk assessments Ferin, J., Oberdörster, G., Soderholm, S.C., Gelein, R., 1991. Pulmonary tissue
using dietary exposures. access of ultrafine particles. J. Aerosol Med. 4, 57–68.
Gurr, J.R., Wang, A.S.S., Chen, C.H., Jan, K.Y., 2005. Ultrafine titanium dioxide
particles in the absence of photoactivation can induce oxidative damage to
4.7. Conclusions human bronchial epithelial cells. Toxicology 213, 66–73.
Handy, R.D., Depledge, M.H., 1999. Physiological responses: their measure-
This study provides one of the first detailed overviews of ment and use as environmental biomarkers in ecotoxicology. Ecotoxicology
organ integrity and the physiological effects of TiO2 NPs in rain- 8, 329–349.
Handy, R.D., Eddy, F.B., 1989. Surface absorption of aluminium by gill tissue
bow trout. Overall our findings are broadly similar to those we
and body mucus of rainbow trout, Salmo gairdneri, at the onset of episodic
recently reported for SWCNT (Smith et al., 2007) in that a num- exposure. J. Fish Biol. 34, 865–874.
ber of body systems are affected, although TiO2 may be a little Handy, R.D., Eddy, F.B., 1991. Effects of inorganic cations on sodium adsorption
less toxic than SWCNT. Our data suggest that TiO2 NPs cause to the gill and body surface of rainbow trout, Oncorhynchus mykiss, in dilute
respiratory toxicity, and disturbances to the metabolism of some solutions. Can. J. Fish. Aquat. Sci. 48, 1829–1837.
Handy, R.D., Eddy, F.B., 2004. Transport of solutes across biological membranes
trace elements like Zn and Cu. Similar to the findings in mam-
in eukaryotes: an environmental perspective. In: van Leeuwen, H.P., Köster,
mals, oxidative stress is also a main concern for trout during TiO2 W. (Eds.), Physicochemical Kinetics and Transport at Chemical-Biological
NP exposure. Notably, in both trout and rodents, these effects Interphases IUPAC Series. John Wiley, Chichester, pp. 337–356.
can occur without appreciable Ti accumulation in the internal Handy, R.D., Shaw, B.J., 2007a. Toxic effects of nanoparticles and nanomateri-
organs in the short term. We should therefore be cautious when als: implications for public health, risk assessment and the public perception
of nanotechnology. Health Risk Soc. 9, 125–144.
using body burden data for TiO2 NPs in exposures lasting only
Handy, R.D., Shaw, B.J., 2007b. Ecotoxicity of nanomaterials to fish: chal-
a few days or a couple of weeks, since the fundamental assump- lenges for ecotoxicity testing. A learned discourse in Integr. Environ. Assess.
tion that internal accumulation causes toxic effect may not apply Manag. 3 (3), 458–460.
over these shorter timescales. However, this is not a new eco- Hoyle, I., Shaw, B.J., Handy, R.D., 2007. Dietary copper exposure in African
toxicological problem, and there is much we can apply from walking catfish, Clarias gariepinus: transient osmoregulatory disturbances
and oxidative stress. Aquat. Toxicol. 83, 62–72.
our knowledge of surface acting metal toxins like Al (Handy
Hirano, K., Nitta, H., Sawada, K., 2005. Effect of sonication on the photo-
and Eddy, 1991, 2004). For example, Zhang et al. (2007) have catalytic mineralization of some chlorinated organic compounds. Ultrason.
already used our adsorption theory to partly explain unpredicted Sonochem. 12, 271–276.
interactions between Cd accumulation and TiO2 NPs in carp. Long, T.C., Saleh, N., Tilton, R.D., Lowry, G.V., Veronesi, B., 2006. Tita-
nium dioxide (P25) produces reactive oxygen species in immortalized brain
microglia (BV2): implications for nanoparticle neurotoxicity. Environ. Sci.
Acknowledgements
Technol. 40, 4346–4352.
Lovern, S.B., Klaper, R., 2006. Daphnia magna mortality when exposed to tita-
This research was funded by a grant to R. Handy from the Nat- nium dioxide and fullerene (C-60) nanoparticles. Environ. Toxicol. Chem.
ural Environment Research Council UK (NE/D007267/1) and 25, 1132–1137.You can also read
