Article pubs.acs.org/est
Mercury in Molar Excess of Selenium Interferes with Thyroid Hormone Function in Free-Ranging Freshwater Fish Paulien J. Mulder,†,⊥ Elisabeth Lie,‡ Grethe S. Eggen,† Tomasz M. Ciesielski,† Torunn Berg,§ Janneche U. Skaare,‡,∥ Bjørn M. Jenssen,† and Eugen G. Sørmo*,† †
Department of Biology and §Department of Chemistry, Norwegian University of Science and Technology, Trondheim, Norway Norwegian School of Veterinary Science, Oslo, Norway ∥ National Veterinary Institute, Oslo, Norway ‡
S Supporting Information *
ABSTRACT: Thyroid hormones (THs) are essential for cellular metabolism, somatic growth and development, and reproduction. Mercury (Hg) entering aquatic systems and accumulated as highly toxic methylmercury (MeHg) represents a threat to wildlife and human health. Selenium (Se) is an essential element critical for TH activation and regulation. In organisms, binding of Hg in a Se−Hg complex results in a detoxification of Hg. However, formation of Se−Hg complexes also affects Se bioavailability, disrupting functions of Sedependent enzymes, such as TH deiodinases, which convert thyroxine (T4) to the physiologically active TH, triiodothyronine (T3). The main aim of the present study was to investigate how tissue Se:Hg molar ratios, tissue levels of Se and Hg, and other potential TH disruptive contaminants (metals and organic chemical compounds) affect plasma TH levels in free-ranging brown trout, Salmo trutta, from Lake Mjøsa (a Se-deprived lake) and Lake Losna (a reference lake), Norway. Among the wide range of potential TH disruptive pollutants investigated, tissue Se:Hg molar ratios in muscle and liver were the most significant predictors of plasma TH levels in the trout. Moreover, lower plasma levels of the biological active hormone, T3, in the Lake Mjøsa trout co-occurred with their low Se:Hg molar ratios. This suggests that Se availability is impaired by Hg and results in altered selenoenzyme activities and loss of optimal control of TH balance in freeranging freshwater fish.
■
INTRODUCTION Mercury (Hg) is considered a global pollutant, and elemental mercury (Hg0) is the predominant form of atmospheric Hg. Because Hg has a long residence time in the atmosphere, it is also transported to and deposits in remote regions far away from the sources.1,2 In aquatic sediments inorganic Hg can be converted to methylmercury (MeHg) that can be biomagnified in the aquatic food chain causing alarming levels in predatory fish, posing a threat to wildlife and human health.3,4 As a soft electrophile, MeHg has a strong affinity to soft nucleophiles, thiol and selenol groups. The toxicity of Hg compounds is therefore assumed to be associated with their high affinity for sulfur and selenium, causing efficient bindings to cysteine (thiol) and selenocysteine (selenol) moieties of proteins and enzymes, perturbing their functions.5,6 Prior to understanding the biological role of selenium (Se), it was well-established that Se counteracts Hg toxicity.6,7 This phenomenon was suggested to be a result of the sequestration of Hg by Se and the resulting Se−Hg complexes, rendering Hg inefficient to elicit toxicity. However, because formation of Se− Hg complexes also reduces the bioavailability of Se (i.e., sequestration of Se by Hg), this also affects Se physiology and Se-dependent body functions.7 Hence, following an increased © 2012 American Chemical Society
understanding of the biological role of Se, evidence suggests that the concept of Se protection against Hg toxicity occurs by ensuring sufficient amount of bioavailable Se so that normal selenoprotein and selenoenzyme synthesis is maintained. This implies that toxic effects accompanying exposure to Hg depend upon the tissue Se:Hg molar ratio of the organism, where Hg is more hazardous when in molar excess of Se. Because of the 1:1 stoichiometry of the Se−Hg interaction, a tissue Se:Hg molar ratio greater than 1 is suggested as a threshold for the protecting action of Se against Hg toxicity.4,7−9 Selenium, as a constituent of the amino acid selenocysteine (SeCys), is essential for a wide range of body functions, such as ensuring proper thyroid hormone function and antioxidant functions.10 SeCys is required as the active component of various selenoproteins and selenoenzymes. In contrast to other amino acids, SeCys is degraded and resynthesized during each cycle of protein synthesis. Selenide ions (Se2‑) formed when SeCys is degraded are vulnerable to the binding of heavy metals Received: Revised: Accepted: Published: 9027
March 28, 2012 June 28, 2012 July 13, 2012 July 13, 2012 dx.doi.org/10.1021/es301216b | Environ. Sci. Technol. 2012, 46, 9027−9037
Environmental Science & Technology
Article
such as Hg due to its very high affinity for these elements.7 When bound to Hg or other metals, Se is no longer available for selenoprotein synthesis.7,11 Hence, MeHg is believed to be a highly specific and irreversible selenoenzyme inhibitor, by either binding bioavailable Se from SeCys (selenide−Hg complexes) or due to binding to SeCys (selenol) moieties of selenoproteins and selenoenzymes.12 Thyroid hormones (THs) play an important role in the maintenance of a normal physiological state in vertebrates, influencing activities in a wider variety of tissues and biological functions than does any other hormone.13 In fish, THs have been linked to a multitude of functions such as development and somatic growth, cell metabolism, osmoregulation, and reproduction.13 Thyroxine (T4) is considered a prehormone and is required for the production of the more physiologically active TH, triiodothyronine (T3).13 T4 is produced in the thyroid gland, whereas the conversion of T4 to T3 by deiodinases predominately occurs in peripheral tissues such as the liver, kidney, muscles, and gills.13,14 Se is essential for TH function because SeCys constitutes the active site of deiodinases.14 Because depletions in circulatory TH levels have been associated with Se-deficiency,15 we hypothesize that sequestration of Se by an excess of Hg may cause depletions of circulatory TH levels and disruption of TH function.15,16 This may account for observed similarities in clinical manifestations of hypothyroidism, Se-deficiency, and MeHg poisoning.17 In freshwater food chains, the highest bioaccumulation of MeHg is generally observed in piscivorous (fish-eating) species, and particularly in large-sized and long-lived species such as the brown trout, Salmo trutta.4,18 MeHg concentrations in predatory brown trout from the largest lake in Norway, Lake Mjøsa, and other waterways in southeastern Norway are higher than current tissue-based criterions for the protection of humans given by the U.S. Environmental Protection Agency (0.3 mg/kg wet weight) and the European Food Safety Authority (0.5 mg/kg wet weight).19,20 Hg concentrations in these trout also exceed the wildlife threshold of 0.1 mg/kg.4,21 Soils and waterways of north-central Europe and Scandinavia are also relatively deprived of Se.22 Thus, Hg pollution in freshwater systems from this part of the world may be more hazardous to wildlife and humans than currently recognized. The main objective of the present study was to investigate how levels of Hg and Se and the Se:Hg molar ratio affect plasma TH levels in free-ranging brown trout from Lake Mjøsa (a Se-deprived lake) and Lake Losna (a reference lake).23 Because other pollutants than Hg may also affect TH function, we additionally included measurements of a wide-range of other potential TH disruptive pollutants, including various trace elements and organic compounds,13,24 into the statistical models. Particular emphasis was on the possible TH disruptive effects of the high contamination for polybrominated diphenyl ethers (PBDEs), hexabromocyclododecane (HBCD), and polychlorinated biphenyls (PCBs) in Lake Mjøsa.25,26
water. Water was replaced every 5−10 min to keep it oxygenrich. Within 30−45 min from capture, biometric measurements, blood sampling, and tissue collection were performed. Blood was sampled from the caudal vein using a heparinized syringe. Plasma was prepared and than stored below −20 °C in cryovials until analysis of TH levels. Muscle and liver samples were stored in plastic bags below −20 °C until analyzed for levels of trace elements. Muscle samples wrapped in aluminum foil were stored below −20 °C until analyzed for concentrations of organic pollutants. The condition factor (CF) of the fish was calculated according to Fulton’s condition factor K.27 The approximate age (years of age) of the trout was determined using both scales and otolith samples.28 Fishing permits were given by governmental authorities. Sampling of the trout was performed according to national legislations and sampling conducted by certified researchers. Determination of Thyroid Hormone (TH) Levels in Plasma. An assumption of the present study is that measured plasma TH concentrations were not significantly distorted by capture and handling stress. It appears that such stress has minimal effect on the circulating levels of both T4 and T3 in fish.29,30 Plasma samples were analyzed for total T4 (TT4), total T3 (TT3), free T4 (FT4), and free T3 (FT3) concentrations using commercial radioimmunoassay (RIA) kits (Coat-A-Count TT4, TT3, FT4, and FT3, Siemens Medical Solution Diagnostics, Los Angles, CA) at the Department of Biology, Norwegian University of Science and Technology (NTNU). A γ- scintillation counter (Cobra AutoGamma, model 5003, Packard Instrument Co., Dowers Grove, IL) was used to detect the bound radioactive antigen in the trout samples. Calibration curves and concentrations of TH in the samples were calculated by the software of the γ-counter (Spectra Works Spectrum Analysis Software). Plasma samples were analyzed in duplicates (TT3, FT3) or triplicates (TT4, FT4). The kits had analytical sensitivities of 0.31 pmol/L for FT3, 0.13 pmol/L for FT4, 0.11 nmol/L for TT3, and 3.2 nmol/L for TT4. Inter- and intraassay variance was calculated using the laboratory’s own reference material (bovine plasma) and the standard reference materials included in the RIA kits. Intra- and interassay coefficient of variation (CV) values of the reference materials were below 15% for all hormones. CV ranged from 0.1 to 7.5% in sample replicates for TT3. For FT3, CV was 0.02−14.3%. Very low levels of TT4 and FT4 in the trout plasma resulted in a larger variation of the concentrations in some of the samples. CV exceeded 35% for six of the samples of TT4 (35−44%) and five of the samples of FT4 (35−51% and 121%). However, the levels of TT4 and FT4 in these individuals were very low compared to those in the other individuals. Thus, although the reproducibility was low between the replicates of these specimens, analysis identified low T4 concentrations in these individuals. Indeed, statistical analysis revealed a high degree of correlation between TT4 and FT4 concentrations in the samples (r = 0.87, p < 0.0001). Determination of Trace Element Levels in Muscle and Liver. Analysis of trace element concentrations in the tissue samples (both muscle and liver) was performed at the Department of Chemistry, NTNU, according to methods described by Sørmo et al.4 Briefly, approximately 1 g of sample was weighed, and ultrapure water and concentrated nitric acid (HNO3) were added prior to digestion with a high-pressure microwave system (Milstone UltraClave, EMLS, Lautkirch, Germany). After cooling to room temperature, the digested
■
MATERIAL AND METHODS Sampling. Brown trout were captured from boats using fishing rods. Samples were obtained from the northern part of Lake Mjøsa (60°53′58′N, 10°41′31E), close to Lillehammer City, in May 2008 (n = 34), and in the Lake Losna (61°22′41′N, 10°15′28′E), in May 2009 (n = 14). The Gudbrandsdalslågen River connects the two lakes, and Lake Losna is situated approximately 50 km upstream of Lake Mjøsa. After capture, fish were kept alive in large tanks filled with 9028
dx.doi.org/10.1021/es301216b | Environ. Sci. Technol. 2012, 46, 9027−9037
Environmental Science & Technology
Article
isometric interconversion of the diastereomers.31 Thus, our result represents total HBCD. Standard procedures were used to ensure adequate quality assurance and control. The precision, linearity, and sensitivity of the analyses were within the laboratory’s accredited requirements. Matrices of uncontaminated trout filets spiked with all analytes had relative recoveries between 69 and 134% for PCBs and 96−139% for BFRs. Reproducibility and repeatability were validated with the laboratory’s own reference material, seal blubber, which was analyzed in each series of 12− 24 samples. Drift was controlled by analyzing standards every 10th sample. Blanks were below the detection limit for most analytes. However, because of a varying degree of contamination for BDE-153 and BDE-154 in blanks, the average of the blanks (n = 9) + two standard deviations was set as the limit of detection for these compounds. Limits of detection for the other analytes were set to 3 times the noise level. Concentrations of POPs are presented on a ww basis. Limit of detection ranged from 0.01 to 0.31 ng/g ww for BFRs and from 0.02 to 0.04 ng/g ww for PCBs. Statistical Analyses. Principal component analysis (PCA) using SIMCA P+ (version 12.0.0.0) (Umetrics, Umeå, Sweden) was used to investigate trends and patterns (clustering and orthogonal relationships) among the variables and the observations. Orthogonal partial least-squares (OPLS) regression using SIMCA P+ was used to model the effect of pollutants and elements and biometric measurements (Xvariables) on plasma TH levels in the trout (Y-variables). OPLS is a statistical tool that has been designed to deal with multiple regression problems where the number of observations are limited and the relationships of predictor X-variables are highly correlated (e.g., high degree of multicolinearity). The OPLS method is a modification of the partial least-squares (PLS) method.33 The model separates the variation of X into two parts, one that is linearly related (and therefore predictive of Y) and one that is orthogonal to Y. This partitioning of the X-data provides improved transparency and interpretability compared to conversional PLS. Variable importance in projection (VIP) values reflect the relative importance of each X-variable in the prediction model. VIP allows classifying the X-variables according to their explanatory power of Y. Predictors (Xs) with VIP-values larger than 1 are the most relevant for explaining Y. For significance testing the OPLS prediction, ANOVA of the cross-validation of residuals (CV-ANOVA) was applied.34 For follow-up statistical analyses, Pearson correlation, multiple regression, and univariate general linear models (GLMs) were performed using PASW Statistics (version 18.0.0) (IBM-SPSS Statistics). Multiple linear regressions were performed in the backward method, and data were diagnosed for multicolinearity according to the software description. Student’s t-test was selected to measure differences between the two populations. Data were log10-transformed to obtain normal distribution. The significance level was set to p < 0.05 for all tests.
samples were diluted with ultrapure water to 60 mL in polypropylene vials to achieve a final HNO3 concentration of 0.6 M. To determine concentrations of trace elements, highresolution inductively coupled plasma mass spectrometry (HRICP-MS) analyses were performed using a Thermo Finnigan model Element 2 instrument (Bremen, Germany). The instrument was calibrated using 0.6 M HNO3 solutions of matrix-matched multielement standards. Concentrations of metals and elements were calculated and determined using five-point linear curves created for each component of these standards. A multielement standard was randomly analyzed every 10th sample to control for the instrument drift. The accuracy of the method was verified by analyzing the certified reference material Oyster Tissue NIST 1566b (National Institute of Standards and Technology, Gaithersburg, MD). The concentrations found were within 90−115% of the certified values for all the elements. To assess possible contamination during sample preperation, blanks of HNO3 and ultrapure water were prepared using the same procedures as for the samples. Concentrations of trace elements are presented on a wet weight (ww) basis. Method detection limits (MDL) were calculated as described in Sørmo et al.4 and ranged from 0.01 μg/kg to 0.02 mg/kg ww for Au and Zn, respectively. Determination of Persistent Organic Pollutants (POPs) Levels in Muscle. Analysis of POPs in muscle samples (2−3 g) were performed at the Laboratory of Environmental Toxicology at the Norwegian School of Veterinary Science (NVH) in Oslo. The laboratory is accredited by Norwegian Accreditation (Kjeller, Norway) for the determination of brominated flame retardants (BFRs) and organochlorines (OCs) in biological material of animal origin according to the requirements of the NS-EN ISO/IEC 17025 (test 137). Samples were homogenized, extracted, and analyzed for concentrations of POPs using gas chromatography/mass spectrometry-negative ion chemical ionization detection (GC/ MS-NCI) according to methods described by Sørmo et al.31 and Murvoll et al.32 Briefly, the samples were weighed in 80 mL centrifugation tubes, and internal standards were added and extracted twice with cyclohexane and acetone (3:2) using ultrasonic homogenization. The supernatants of both extracts were pooled, and an aliquot was used for gravimetric lipid determination of the muscle lipid content. For each sample, the remaining supernatant was treated with concentrated sulfuric acid to remove lipids. The final cleaned-up concentrates were transferred to amber vials and analyzed on the GC/MS. For separation and detection of BFRs (PBDEs and HBCD), a volume of 1 μL was injected on a DB-5MS column as described by Sørmo et al.31 For the separation and detection of PCBs, a volume of 2 μL was injected on a nonpolar fused silica precolumn that was split into two separate columns (SPB-5 and SPB-1701; Superlo Inc.) as described by Murvoll et al.32 Concentrations of POPs were calculated using the Software MSD ChemStation G1701 version D.01.00 (Agilent Technologies) and determined using six to eight point linear calibration curves created for each component in the standards. The following compounds were determined: PBDEs (BDE-28, -47, -99, -100, -153, -154), HBCD, and PCBs (PCB-28, -47, -52, -66, -74, -87, -99, -101, -105, -110, -114, -128, -136, -137, -138, -141, -151, -153, -156, -157, -170, -180, -183, -189, -194, -199, -206). HBCD consists of three diastereomers (α-, β-, and γ-HBCD). At temperatures above 160 °C in the injection port, as used in this GC analysis, thermal rearrangement leads to
■
RESULTS Principal component analysis (PCA). The analysis resulted in a PCA model with three principal components (PC1−3; R2X = 0.56, Q2 = 0.36) (see Figures S1−S3, Supporting Information). The scores of the PCA indicated a clear separation of the observations by population, except for one fish from Lake Losna (L04) that was situated within the cluster of the Lake Mjøsa fish (see Figure S1, Supporting 9029
dx.doi.org/10.1021/es301216b | Environ. Sci. Technol. 2012, 46, 9027−9037
Environmental Science & Technology
Article
Table 1. Mean ± Standard Deviation (Minimum−Maximum) of Tissue Se:Hg Molar Ratios and Wet Weight Tissue Concentrations of Selected Trace Element and Persistent Organic Pollutant (POP) Variables in Brown Trout, S. trutta, from Lake Mjøsa and Lake Losna, Norwaya b
Se:Hg ratio muscle Se:Hg ratio liver Hg muscle (mg/kg)b Hg liver (mg/kg) Se muscle (mg/kg)b Se liver (mg/kg) Cd liver (mg/kg) PBDE muscle (ng/g) HBCD muscle (ng/g) PCB muscle (ng/g)
Lake Mjøsa
Lake Losna
0.98 ± 0.35 (0.39−1.88) 1.51 ± 1.14 (0.26−3.97) 0.69 ± 0.29 (0.33−1.81)* 1.74 ± 0.92 (0.76−4.95)** 0.23 ± 0.04 (0.17−0.35) 5.92 ± 5.10 (1.15−24.34) 0.12 ± 0.07 (0.05−0.39)* 178.6 ± 12.68 (30.99−452.4)*** 18.55 ± 12.68 (17.10−64.27)*** 62.51 ± 33.03 (23.03−161.0)***
2.83 ± 1.43 (1.32−5.34)*** 9.31 ± 7.50 (0.83−26.53)*** 0.52 ± 0.33 (0.16−1.08) 0.93 ± 0.72 (0.21−2.18) 0.43 ± 0.12 (0.26−0.68)*** 13.92 ± 9.72 (3.94−37.72)** 0.09 ± 0.03 (0.03−0.15) 5.56 ± 5.04 (0.96−16.72) 0.88 ± 0.83 (nd−2.54) 19.39 ± 17.63 (3.99−66.46)
a
See Tables S1 and S2 (Supporting Information) for concentrations of individual POPs and other trace elements measured. Asterisks devote statistical differences between the populations (Student’s t-test), *p < 0.05, **p < 0.001, ***p < 0.0001. bLake Mjøsa data from Sørmo et al.4 Abbreviations: PBDE = polybrominated diphenyl ethers, HBCD = hexabromocyclododecane, PCB = polychlorinated biphenyls, nd = not detected.
Table 2. Mean ± Standard Deviation (Minimum−Maximum) of Plasma Thyroid Hormone Levels, Age, and Biometric Data in Brown Trout, S. trutta, from Lake Mjøsa and Lake Losna, Norwaya TT4 (nmol/L) FT4 (pmol/L) TT3 (nmol/L) FT3 (pmol/L) age (year) body length (cm) body mass (g) condition factor (CF) lipid content (%) (muscle tissue)
Lake Mjøsa
Lake Losna
6.55 ± 4.91 (1.37−22.47) 1.43 ± 0.96 (0.14−4.54) 18.98 ± 8.15 (6.49−34.34) 12.84 ± 5.23 (5.10−25.62) 8.39 ± 0.99 (7.00−11.00) 68.06 ± 8.56 (52.00−89.00)* 3030 ± 961 (1200−5220) 0.94 ± 0.19 (0.55−1.39) 3.71 ± 3.03 (0.47−11.28)
6.49 ± 3.89 (1.70−17.62) 1.26 ± 0.88 (0.18−3.46) 22.42 ± 13.11 (9.85−63.00) 17.65 ± 9.34 (7.81−43.00)* 9.06 ± 1.33 (7.00−12.00) 59.86 ± 13.42 (45.00−82.00) 2295 ± 1609 (740−5002) 0.90 ± 0.11 (0.69−1.17) 2.51 ± 1.76 (0.58−5.68)
An asterisk devotes significance differences between the populations (Student’s t-test). *p < 0.05. Abbreviations: TT4 = total thyroxine (free + protein bound), FT4 = free thyroxine, TT3 = total triiodothyronine, FT3 = free triiodothyronine. a
PCB correlated with fish body size (r > 0.57, p < 0.0001) (statistical analyses for these pollutants were performed in each population because of their much higher levels in the Lake Mjøsa trout). Plasma TH levels, except for FT3 levels, tended to orient positively along PC2 in the PCA model, covarying with the nutritional status of the trout, i.e., muscle lipid content and condition factor (see Figure S1, Supporting Information). Hence, TT3 but not FT3 was positively correlated with muscle lipid content (r = 0.42, p = 0.003) and condition factor (r = 0.40, p = 0.005) (see Figure S2, Supporting Information). FT3 levels were about 30−35% lower in Mjøsa trout than in Losna trout (Table 2). TT3, TT4, and FT4 did, however, not differ in levels between the two populations. TT3 and FT3 correlated positively with tissue Se:Hg molar ratios and with tissue levels of the essential elements Co and Se (Table 3; see Figure S3, Supporting Information). Likewise, TT4 was positively correlated with the Se:Hg molar ratio. The TH levels correlated inversely with tissue levels of the toxic elements Hg and Cd (Table 3). Orthogonal Partial Least Squares (OPLS) Regression Analysis. Plasma TT4 and FT4 Levels. The OPLS model for plasma TT4 levels (Y-variable) was statistically significant (R2X = 0.33, R2Y = 0.25 Q2 = 0.13; CV-ANOVA, p = 0.048) and indicated that some the measured pollutants affect plasma TT4 levels in the trout. Hgliver had the highest VIP-value (VIP > 1.5), followed by Cdliver, Hgmuscle and Se:Hgmuscle, Crliver, Femuscle and PCBmuscle, and body length (VIP > 1) (Figure 1a). The
Information). This particular specimen was ID-tagged in Lake Mjøsa. Thus it is possible that this fish was a Lake Mjøsa trout that recently had migrated to Lake Losna. This trout was therefore grouped in the Lake Mjøsa population in the further statistical analyses. Tissue Se:Hg molar ratios and levels of Se, Hg, PBDE, HBCD, and PCBs were oriented along PC1 of the PCA model (see Figure S1, Supporting Information). The Lake Mjøsa trout differed from their Lake Losna counterparts by higher levels of Hg, PBDEs, HBCD, and PCBs, and conversely by lower Se levels and Se:Hg molar ratios (Table 1). The higher accumulation of Hg in the Lake Mjøsa trout was particularly evident for Hgliver, with levels about 2-fold higher than in the Losna trout (Table 1). Conversely, Semuscle and Seliver levels, respectively, were about 2- and 3-fold higher in Lake Losna trout than in Mjøsa trout. Hence, Se:Hgmuscle and Se:Hgliver, respectively, were therefore approximately 2.5- and 6-fold higher in Losna trout than in Mjøsa trout. Approximately half of the Mjøsa trout had Se:Hgmuscle and Se:Hgliver molar ratios 1.5), followed by Se:Hgliver and Hgliver, PBDEmuscle, HBCDmuscle, PCBmuscle and Hgmuscle, Comuscle and Coliver, and population (VIP > 1) (Figure 3a). Tissue Se in molar excess of Hg and higher tissue levels of Se and Co were associated with higher FT3 levels (Figure 3b; see Figure S4, Supporting Information). Conversely, higher tissue levels of Hg and POPs were associated with lower FT3 levels (Figure 3b). The OPLS model confirmed the statistically significant lower plasma FT3 levels in Mjøsa trout than in their Losna counterparts (Table 1). The OPLS model also indicated the possibility that high concentrations of POPs (e.g., PBDEs, HBCD, and PCBs) were inversely associated with FT3 levels (Figure 3b). However, additional testing using GLMs confirmed that Se:Hg molar ratio did affect FT3, whereas levels of POPs (tested individually) and the population affiliation of the trout did not affect plasma FT3 levels; exemplified by the GLM showing that Se:Hgmuscle (F1.44 = 4.98, p = 0.032) but not PBDE (F1.44 = 0.027, p = 0.88) and population (F1.44 = 0.011, p = 0.92) affected plasma FT3 levels in the trout (R2 = 0.25, F1.44 = 4.93, p = 0.005). This implies the lower FT3 levels in Mjøsa trout as a consequence of their lower tissue Se:Hg molar ratios and not because of their high POP concentrations.
Table 3. Bivariate Relationships of Tissue Se:Hg Molar Ratios and Tissue Se, Hg, Cd, and Co Levels with Plasma Thyroid Hormone Levels in Brown Trout, S. trutta, from Lake Mjøsa and Lake Losna, Norwaya muscle Se:Hg liver Se:Hg muscle Se liver Se muscle Hg liver Hg muscle Cd liver Cd muscle Co liver Co
TT3
FT3
TT4
r = 0.42 p = 0.003 r = 0.33 p = 0.031 r = 0.28 p = 0.048 ns
r = 0.50 p < 0.0001 r = 0.43 p = 0.004 r = 0.32 p = 0.029 r = 0.28 p = 0.070 r = −0.36 p = 0.011 r = −0.40 p = 0.007 r = −0.27 p = 0.060 r = −0.31 p = 0.041 r = 0.35 p = 0.015 r = 0.36 p = 0.018
r = 0.34 p = 0.019 ns
ns
ns
ns
ns
ns
r = −0.34 p = 0.017 r = −0.35 p = 0.021 ns
r = −0.28 p = 0.049 r = −0.29 p = 0.064 ns
r = −0.30 p = 0.053 ns
ns
ns
ns
r = −0.29 p = 0.047 r = −0.30 p = 0.050 r = −0.37 p = 0.010 r = −0.40 p = 0.007 r = 0.40 p = 0.004 r = 0.40 p = 0.008
FT4
ns
ns
a
Borderline significant correlations (0.1 > p < 0.05) are in italic; ns = non significant (p > 0.1). Abbreviations: TT4 = total thyroxine (free + protein bound), FT4 = free thyroxine, TT3 = total triiodothyronine, FT3 = free triiodothyronine, Se:Hg = tissue selenium−mercury molar ratio.
coefficient plot showed that tissue Se in excess of Hg was associated with higher TT4 levels, while elevated tissue levels of Hg and Cd (toxic metals), Cr, PCBs, and Fe as well as larger fish body sizes were associated with the lower TT4 levels in the trout (Figure 1b). The OPLS model for plasma FT4 was not statistically significant (not shown) but indicated a largely similar pattern in X-predictor VIP and coefficient values as in the corresponding OPLS model for TT4. Plasma TT3 and FT3 Levels. The OPLS model for plasma TT3 levels (Y) was statistically significant (R2X = 0.25, R2Y = 0.34, Q2 = 0.19; CV-ANOVA, p = 0.009), suggesting that the pollutants affect TT3 levels in the trout. Cdliver showed the highest VIP followed by Se:Hgmuscle (VIP > 1.5), Crliver, lipid contentmuscle, Coliver, condition factor (CF), Cdmuscle, Se:Hgliver, Hgliver, Hgmuscle, Semuscle, and Alliver (VIP > 1) (Figure 2a). Tissue Se in molar excess of Hg, higher tissue levels of Se and Co (essential elements), and higher lipid content and condition factors were associated with the higher TT3 levels (Figure 2b). Conversely, higher tissue levels of Hg and Cd (toxic metals), Cr, and POPs were associated with the lower TT3 levels (Figure 2b). The OPLS model confirmed the positive bivariate correlations between muscle lipid content and condition factor and TT3 in the trout. Although mean muscle lipid contents did not differ between the two populations (Table 2), the individuals of higher lipid contents were confined to the Lake Mjøsa population (see Figure S1, Supporting Information). This suggests the more pronounced effect of lipid contents accounting for higher TT3 levels in the Mjøsa trout, whereas higher tissue Co and Se levels and higher Se:Hg molar ratio in the Losna trout suggest the more pronounced effect of these variables for accounting for higher TT3 levels in this
■
DISCUSSION The findings in the present study support the assumptions that Hg induces TH disruptive effects affecting the biologically active TH, T3.17,35,36 Furthermore, the stronger disruptive effect of Hg in molar excess of Se than of Hg alone on plasma T3 levels suggests that Se-dependent functions may be particularly susceptible to Hg toxicity. Indeed, the positive relationships between tissue Se levels with FT3 and TT3, but not with FT4 and TT4 (Table 2), in the present study predict that the conversion of T4 to T3 is Se-dependent in brown trout. This is consistent with the evidence that SeCys, as in mammals, constitutes the active site in fish deiodinases.14 This implies that the higher the Se:Hg molar ratio is (increased amounts of bioavailable Se), the more likely it is that Sedependent functions such as TH function will be undisturbed by Hg. Moreover, the present study indicates that nonessential trace elements such as Hg in an area of low Se affect the TH function of brown trout to a larger extent than potential TH disruptive POPs, even when the latter are present at very high concentrations. Our present data also suggests that the effect of the Se−MeHg interaction predominately affects the extrathyroidal (or peripheral) TH balance in fish. In normal thyroid function, as the thyroid carrier proteins change, plasma TT3 levels also change, whereas FT3 levels remain constant.37,38 Measurements of FT3, therefore, correlate more reliably with clinical thyroid status than TT3.37 Indeed, in the present study, plasma TT3 deviated from FT3 by 9031
dx.doi.org/10.1021/es301216b | Environ. Sci. Technol. 2012, 46, 9027−9037
Environmental Science & Technology
Article
Figure 1. (A) Orthogonal partial least-squares (OPLS) regression variance of importance plot (VIP) reflecting the relative importance of metal levels, organic pollutants, Se:Hg molar ratios, population, and biometric data (X-variables) on plasma total thyroxine (TT4) levels (Y-variable) in brown trout, S. trutta, from Lake Mjøsa and Lake Losna, Norway. (B) OPLS regression coefficient plot summarizing the relationships between Xvariables and plasma TT4 levels. Negative coefficients reflect inverse relationships, whereas positive coefficients reflect positive relationships of the different X-variables with TT4 levels. Error bars are 95% confidence intervals. Abbreviations: Se:Hg = tissue selenium−mercury molar ratio, lipid = lipid muscle content, CF = Fulton’s condition factor, length = body length, BM = body mass, Pop = population, PBDE = total concentration of polybrominated diphenyl ethers, HBCD = total concentration of hexabromocyclododecane, PCB = total concentration of polychlorinated biphenyls.
correlating with the nutritional status of the trout. This increase in TT3 is consistent with reports that increased plasma T3 coincides with replenishment of lipid stores in salmonids following periods of natural starvation (e.g., during winter).39,40 Because the trout of the present study were captured in the spring (May) it is likely that they were in a period of feeding and lipid restoration. This may explain their large ranges in lipid contents (Table 1) and positive correlation between TT3 and
muscle lipid content. This positive correlation between TT3 and nutritional status was, however, not associated with a corresponding positive correlation between nutritional status and FT3. One explanation for this could be that levels of plasma TH carrying proteins, such as transthyretin/prealbumin (TTR), also in fish, are down-regulated during periods of food deprivation or starvation, thereby affecting circulatory TH levels.41 Indeed, in human medicine, measurement of 9032
dx.doi.org/10.1021/es301216b | Environ. Sci. Technol. 2012, 46, 9027−9037
Environmental Science & Technology
Article
Figure 2. (A) Orthogonal partial least-squares (OPLS) regression variance of importance plot (VIP) reflecting the relative importance of metal levels, organic pollutants, Se:Hg molar ratios, population, and biometric data (X-variables) on plasma total triiodothyronine (TT3) levels (Yvariable) in brown trout, S. trutta, from Lake Mjøsa and Lake Losna, Norway. (B) OPLS regression coefficient plot summarizing the relationships between X-variables and plasma TT3 levels. Negative coefficients reflect inverse relationships, whereas positive coefficients reflect positive relationships of the different X-variables with TT3 levels. Error bars are 95% confidence intervals. See Figure 1 for definition of abbreviations.
prealbumin (TTR) is extensively used as a sensitive marker for assessing changes in nutritional status of patients.42 This may affect TT3 (protein bound T3) to a greater extent than FT3 in plasma. The effects of nutritional condition (lipid content and condition factor) on plasma TT3 levels indicate the importance of accounting for confounding factors, such as nutritional status, when assessing effects of pollutants on circulatory TH levels. Moreover, the present study suggests the replenishment of lipid stores in trout during the spring does not necessarily
result in altered peripheral TH activity (i.e., altered plasma FT3 levels). In fish, more than in other vertebrates, most of the biotransformation of THs is controlled outside the thyroid gland, causing the deiodination of T4 to T3 to occur mainly in peripheral tissues rather than the thyroid gland itself.13,14 It is assumed that T3 generated by the liver is predominately transferred to the plasma, while T3 in other tissues will bind to TH receptors preventing the hormone from entering the plasma.13,14 However, in the present trout study Se:Hgmuscle and 9033
dx.doi.org/10.1021/es301216b | Environ. Sci. Technol. 2012, 46, 9027−9037
Environmental Science & Technology
Article
Figure 3. (A) Orthogonal partial least-squares (OPLS) regression variance of importance plot (VIP) reflecting the relative importance of metal levels, organic pollutants, Se:Hg molar ratios, population, and biometric data (X-variables) on plasma free triiodothyronine (FT3) levels (Y-variable) in brown trout, S. trutta, from Lake Mjøsa and Lake Losna, Norway. (B) OPLS regression coefficient plot summarizing the relationships between Xvariables and plasma FT3 levels. Negative coefficients reflect inverse relationships, whereas positive coefficients reflect positive relationships of the different X-variables with FT3 levels. Error bars are 95% confidence intervals. See Figure 1 for definition of abbreviations.
free-ranging trout. This supports a Se:Hg molar ratio ≈1 as a threshold for the protecting action of Se against Hg toxicity.4,7 This is probably because Se:Hg molar ratios below or approaching the 1:1 stoichiometry reflect physiological states where the bioavailability of Se is sufficiently reduced by Hg to perturb Se-dependent functions. Such Hg-induced functional Se-deficiency raises the concern that Hg toxicity could occur at more or less any level of Hg exposure, provided a concurrent low molar Se level in the animal.4 This implies that detrimental effects following Hg exposure in wildlife may occur at lower Hg exposure levels providing a low environmental background concentration of Se. Thus, toxic effects of Hg may occur even at
Semuscle were relatively stronger predictors of FT3 (and TT3) than were the cases for the Se:Hgliver and Seliver (Figures 2 and 3). This finding may imply that measurements of these trace elements in muscle tissues better predict their “whole body” TH disruptive effects than corresponding measurements in liver tissue. This also suggests that T3 and TH activity in muscle tissue, which contains >80% of the total content of T3 in fish,43 in all probability interacts with the plasma T3 pool in brown trout. Approximately half of the trout from Lake Mjøsa had Se:Hg molar ratios
