Mercury Exposure Associated with Altered Plasma Thyroid Hormones

Feb 11, 2014 - Department of Biology, California State University, Fresno, 2555 East San Ramon ... circulating hormones.9,11 Mercury may also alter th...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/est

Mercury Exposure Associated with Altered Plasma Thyroid Hormones in the Declining Western Pond Turtle (Emys marmorata) from California Mountain Streams Erik Meyer,† Collin A. Eagles-Smith,*,‡ Donald Sparling,§ and Steve Blumenshine† †

Department of Biology, California State University, Fresno, 2555 East San Ramon Avenue, Fresno, California 93740, United States U.S. Geological Survey, Forest and Rangeland Ecosystem Science Center, 3200 Southwest Jefferson Way, Corvallis, Oregon 97331, United States § Department of Zoology, Southern Illinois University, Life Science II 257B, Carbondale, Illinois 62901, United States ‡

S Supporting Information *

ABSTRACT: Mercury (Hg) is a global threat to wildlife health that can impair many physiological processes. Mercury has welldocumented endocrine activity; however, little work on the effects of Hg on the thyroid hormones triiodothyronine (T3) and thyroxine (T4) in aquatic wildlife exists despite the fact that it is a sensitive endpoint of contaminant exposure. An emerging body of evidence points to the toxicological susceptibility of aquatic reptiles to Hg exposure. We examined the endocrine disrupting potential of Hg in the western pond turtle (Emys marmorata), a long-lived reptile that is in decline throughout California and the Pacific Northwest. We measured total Hg (THg) concentrations in red blood cells (RBCs) and plasma T3 and T4 of turtles from several locations in California that have been impacted by historic gold mining. Across all turtles from all sites, the geometric mean and standard error THg concentration was 0.805 ± 0.025 μg/g dry weight. Sampling region and mass were the strongest determinants of RBC THg. Relationships between RBC THg and T3 and T4 were consistent with Hg-induced disruption of T4 deiodination, a mechanism of toxicity that may cause excess T4 levels and depressed concentrations of biologically active T3.



INTRODUCTION The risk of mercury (Hg) to wildlife is a pervasive conservation threat on a global scale,1−3 and evidence of toxicological impairment at progressively lower Hg concentrations continues to mount.4−6 Moreover, the expanding breadth of wildlife species studied in relation to Hg-induced toxicity has improved our understanding of the prevalence of Hg risk across the taxonomic spectrum. Recent studies suggest that reptiles and amphibians are emerging as a particularly threatened group that may be negatively impacted at environmentally relevant Hg exposures.7,8 Mercury toxicity in reptiles and amphibians can manifest as a broad range of deleterious endpoints.9 Little is known about the underlying physiological mechanisms that drive these toxic responses, but many are linked to one or more of the major endocrine systems.10−12 The hypothalamic− pituitary−thyroid (HPT) axis may be particularly important because it regulates vertebrate growth and metabolism, and methylmercury (MeHg) binds with thyroid sulfhydryl groups, preferentially leading to the accumulation of Hg in the thyroid gland.11 Thus, understanding how Hg influences thyroid function is important for an improved understanding of the toxicological risk of Hg to wildlife. © 2014 American Chemical Society

The accumulation of mercury in the HPT axis disrupts the production of thyroid hormones thyroxine (T4) and triiodothyronine (T3) in a dose-dependent fashion, stimulating compensatory feedback loops that can lead to an imbalance in circulating hormones.9,11 Mercury may also alter thyroid hormones by targeting selenoenzyme synthesis in extrathyroidal tissues,13 where conversion of T4 to T3 primarily occurs in the pancreas, kidney, liver, and muscle of reptiles.14 Because thyroid hormones play a critical role in growth, development, and reproduction,15 disruption of the reptilian thyroid system interrupts these essential functions, as well as others, such as thermoregulation, metabolic homeostasis, and gene expression.12,16 Although limited data exist on the effects of Hg in reptiles, Hg-exposed northern watersnakes (Nerodia sipedon) exhibited several altered feeding behaviors17 but no evidence of altered reproductive endpoints.18 Similarly, common snapping Received: Revised: Accepted: Published: 2989

November 13, 2013 January 29, 2014 February 11, 2014 February 11, 2014 dx.doi.org/10.1021/es4050538 | Environ. Sci. Technol. 2014, 48, 2989−2996

Environmental Science & Technology

Article

turtles (Chelydra serpentina) experienced decreased hatchling success and increased infertility at higher Hg exposures.19 Although turtles are the most studied group in reptile toxicology,20 and Hg has been investigated in many turtle ecotoxicology studies,7 few studies have addressed the endocrine disrupting effects of Hg in reptiles. This is particularly important because many reptile species are undergoing rapid declines worldwide due to a combination of factors, including contaminant exposure.21 Turtles may exhibit unique risk to MeHg toxicity because they often live in aquatic habitats such as wetlands, bogs, and low-gradient rivers where MeHg production can be elevated, some species prey extensively on higher-order invertebrates and fish, and they can be very long-lived, accumulating Hg in their tissues over a span of decades. However, even with the emerging understanding of the effects of Hg on these groups, very little is known about exposure and toxic responses to Hg in many of the most susceptible reptile species because most of the work on reptilian Hg toxicology to date has focused on relatively common species that can be readily studied in the laboratory. In this study, we investigated Hg bioaccumulation and endocrine impairment in the western pond turtle (Emys marmorata), a reptile that has declined across its native range of California, Oregon, and Washington and is now listed as a Species of Special Concern in California, sensitive in Oregon, and endangered in Washington. Moreover, it is classified as vulnerable on the IUCN Red List.22 Importantly, many of its populations in California reside in areas that have been impacted by legacy Hg associated with historic gold mining activities. Despite the proximity of E. marmorata populations to Hg-impacted waterbodies, no studies have yet assessed Hg exposure in this species.

in the volume of whole blood that could safely be drawn from each animal and needed to balance the sample needs of analyses of Hg, T3, and T4. Second, across taxa it is well recognized that RBCs serve as the primary carrier of Hg within the blood, and most of the Hg in blood is bound to the cellular fraction;27 thus, the cellular fraction is representative of circulating blood THg concentrations. Finally, the proportion of blood volume that the cellular fraction comprises (i.e., hematocrit) can vary seasonally, among genders, and with the individual health of an animal;28 thus, the whole blood THg concentration can be biased by several factors that could influence assessments of relative exposure.29 Therefore, RBC THg concentrations on a dry weight basis provide a potentially more accurate index of exposure among individuals. Laboratory T3 and T4 Analyses. Free T3 and T4 were measured using a solid phase competitive enzyme-linked immunosorbent assay (BQ Kits, Inc.). Free T3 was measured because it is primarily responsible for biological activity in animals, found in plasma exclusively from T4 deiodination in extrathyroidal tissues,9 and less dependent upon natural factors that confound total T3 bound to carrier proteins.13 Validation of E. marmorata thyroid hormones consisted of pooling plasma from four different samples and making a serial dilution ranging from full strength to 1/32 (T3) or 1/8 (T4) of full strength. Validation details and curve responses are provided in the Supporting Information. The samples, assay buffer, and free T3 or T4 enzyme conjugates were added to wells coated with antiT3 or -T4 monoclonal antibodies. Free T3 or T4 in the turtle’s plasma competed with the T3 or T4 enzyme conjugate, respectively, for binding sites. Unbound T3 was washed off. Upon addition of a substrate, the intensity of the color reaction was inversely proportional to the concentration of either free T3 or T4 in the samples. A standard curve was prepared relating color intensity to the concentration of free T3 (nanograms per milliliter). For T4, light absorption at 450 nm was compared to serial dilutions ranging from 0 to 18 μg/ dL. Further details of the analytical procedure can be found in the Supporting Information. Laboratory THg Analysis. Red blood cell (RBC) samples were analyzed on a dry weight basis. Each sample was dried in a preweighed quartz sample vessel (to the nearest 0.0001 g, Ohaus Adventurer Balance, model AR0640, Ohaus Corporation, Pine Brook, NJ) at 50 °C for 48 h or until the sample was dried prior to analysis. We analyzed each sample for THg on a Milestone tricell DMA-80 Direct Mercury Analyzer (Milestone Inc., Monroe, CT) following the procedure of Herring et al.30 We calibrated the instrument prior to analysis using a certified Hg standard solution (Inorganic Ventures, Christiansburg, VA) and evaluated the accuracy and precision within each analytical batch with certified reference materials [either dogfish muscle tissue (DORM-3) or dogfish liver (DOLT-4) from the National Research Council of Canada, Ottawa, ON]. Batches consisted of 30 total samples, two of which were run in duplicate for a total of 32 samples per batch. The eight remaining places in each autosampler tray consisted of quality assurance and quality control samples. Recoveries averaged 100.17 ± 2.5% (n = 9) and 101.25 ± 0.9% (n = 6) for certified liquid standard calibration checks and certified reference materials, respectively. The absolute relative percent difference for all duplicates averaged 5.6 ± 2.6% (n = 6). Statistical Analyses. All statistical procedures were conducted in R 3.0.1 statistical software.31 We employed linear mixed effects models (LME) using the nlme package in R to



MATERIALS AND METHODS Study Sites. We sampled 79 individual E. marmorata from seven sites across a 700 km latitudinal gradient, representing two regions in California (Figure S1 of the Supporting Information). The southern region consisted of four lotic sites in the southern Sierra Nevada foothills where there was limited historical gold mining activity. In contrast, the northern region consisted of two lotic sites in the Klamath Mountains and one in the California Coast Range, where extensive gold and Hg mining occurred historically.23 Field Sampling of Turtles. We captured turtles from August to October 2011 by hand (snorkeling) in lotic habitats. Search methods and handling procedures were described by Meyer et al.24 We recorded straight-line maximum carapace length (millimeters), mass (grams), sex (male, female, or unknown), and life stage (adult or juvenile) following standardized measures.25 We recorded the time of capture to account for the influence of hormonal circadian rhythms.26 Handling time was restricted to 0.05). For the T3 models, residuals were

examine the factors influencing blood THg and plasma T3 and T4 concentrations.32 For modeling RBC THg concentrations, a global LME model included mass, sex, and region as fixed effects and site nested within region as a random effect. The global model for plasma T4 included RBC THg concentrations, time of capture, sex, mass, and THg*sex interaction as fixed effects, with site as a random effect. The global T3 model included the same structure, except for the addition of T4 as a fixed effect to account for the dependence of T3 on T4. Akaike’s Information Criterion (AIC) was calculated to examine the LME models that best explained the variation in RBC THg and plasma T3 and T4 concentrations. For modeling each response variable, we a priori constructed a global model and several candidate models using different combinations of fixed effects. For modeling RBC THg concentrations, there were seven candidate models and a eighth null (intercept only) model (Table 1). For plasma T3 concentrations, there were nine candidate models and a tenth null (intercept only) model, and for T4 concentrations, there were 11 candidate models and a twelfth null (intercept only) model (Table 1). Using the AICcmodavg package in R, model ranking tables were obtained 2991

dx.doi.org/10.1021/es4050538 | Environ. Sci. Technol. 2014, 48, 2989−2996

Environmental Science & Technology

Article

normally distributed without data transformation (p > 0.05). A likelihood ratio test (LR) found that all potential LME models benefitted from a random effect (p ≤ 0.05) when they were refit with an ordinary linear regression.35 Animal handling and care approval was obtained through the Institutional Animal Care and Use Committee (File 145), California State University, Fresno. Research was also completed with a California Department of Fish and Wildlife Scientific Collecting Permit (11633) and National Park Service permits (WHIS-2011-SCI-0007 and SEKI-2011-SCI-0022).



Figure 2. Partial residual plot of the relationships between western pond turtle (E. marmorata) plasma thyroxine (T4) and THg concentrations in red blood cells. Partial residual plots illustrate the relationship between two continuous variables after controlling for other factors that can influence the dependent variable.

RESULTS We captured and collected sufficient blood from 79 turtles across all locations. Concentrations of THg exceeded the detection limit (0.1 ng of Hg) in all RBC samples. The geometric mean [±standard error (SE]) RBC THg concentration across all sites and sexes was 0.805 ± 0.025 μg/g dry weight (dw). Across sites, the geometric mean THg concentration was highest at Clear Creek (1.037 ± 0.108), followed by South Fork Trinity River (0.962 ± 0.063), Tyler Creek (0.879 ± 0.047), Mad River (0.836 ± 0.054), North Fork Kaweah River (0.723 ± 0.026), Sycamore Creek (0.715 ± 0.031), and Jose Creek (0.607 ± 0.037) (Figure 1).

Besides the relationship between THg and T4, the most parsimonious model showed that both capture time and turtle mass had negative relationships with T4 (Figure 3a,b). Finally,

Figure 1. Geometric mean (±SE) western pond turtle (E. marmorata) red blood cell total mercury (THg) concentrations at several study locations in two regions of California. Figure 3. Partial residual plots of the relationships between western pond turtle (E. marmorata) plasma thyroxine (T4) and (A) body mass and (B) capture time. Partial residual plots illustrate the relationship between two continuous variables after controlling for other factors that can influence the dependent variable.

T4 Analysis. The most parsimonious model explaining plasma T4 concentrations was the global model, which included THg, capture time, sex, mass, and a THg*sex interaction [AICcWt = 0.31 (Table 1)]. This model was 1.24 times more likely than the next best, which included THg, capture time, sex, and mass (ΔAICc = 0.44). Two additional competing models include a model containing capture time, sex, and mass (ΔAICc = 1.22) and a model containing only turtle sex (ΔAICc = 1.84). The global model explained the most variation based on both fixed effects only and fixed and random effects [R2GLMM(m) = 0.26; R2GLMM(c) = 0.48] and predicted that plasma T4 concentrations increased positively with THg concentrations after controlling for other factors influencing T4 (Table S1 of the Supporting Information and Figure 2). Additionally, four of the five highest R2GLMM(c) values were models that included THg as a fixed effect, indicating the importance of Hg exposure on T4 and controlling for influential variables when trying to explain the relationship of THg and T4 hormones (Table S1 of the Supporting Information).

male (backtransformed least squared mean ± SE, 0.259 ± 0.083 μg/dL) and juvenile (0.791 ± 0.24) turtles had T4 concentrations higher than those of females (0.153 ± 0.133) after statistically controlling for all other fixed and random effects. T3 Analysis. There were four primary competing models explaining T3 concentrations in turtle plasma. The top model contained only THg concentrations [AICcWt = 0.32 (Table 1)]. This model was 1.78 times more likely than the next best model, which contained only capture time (ΔAICc = 1.15). Using ΔAIC values of ≤2.0 as a cutoff for determining likely plausible models, those candidate models featuring only turtle mass (ΔAICc = 1.92) followed by the null (intercept only) model (ΔAICc = 2.01) should be considered as probable alternatives to the most parsimonious model.36 The competing models containing only capture time and only mass also 2992

dx.doi.org/10.1021/es4050538 | Environ. Sci. Technol. 2014, 48, 2989−2996

Environmental Science & Technology

Article

provided a reasonably good fit to the data, and the plausibility of the null model suggests that additional relevant variables may not have been measured in this study. Importantly, in the top model, T3 hormones exhibited a slightly negative relationship with THg after controlling for the random site effects (Table S1 of the Supporting Information and Figure 4). Models containing THg as an explanatory

Figure 5. Partial residual plot of the relationship between red blood cell total mercury (THg) concentration and body mass. Partial residual plots illustrate the relationship between two continuous variables after controlling for other factors that can influence the dependent variable. Empty symbols depict data for turtles captured in the southern region and filled symbols data for turtles captured in the northern region.

the spatial exposure patterns, age, and feeding ecology of turtles. Moreover, we found evidence that Hg exposure in turtles was negatively correlated with plasma T3 and positively correlated with T4, suggesting that MeHg may influence the endocrine system in turtles, specifically altering the thyroid hormone response. Spatial heterogeneity is often a primary driver of Hg exposure in aquatic organisms, because of variation in both sources of inorganic Hg and MeHg production within a watershed.37−40 In this study, the most parsimonious THg model found that capture region and body mass were the best predictors of variation in RBC THg concentrations. There was strong evidence that differences in turtle THg concentrations were driven by variations in the availability of Hg within the distinct study regions, which may be indicative of differences in either inorganic Hg distribution or facilitation of MeHg production across the study area. The spatial differences we documented across the study area are consistent with other Hg studies in California and are likely driven largely by the disparity in historic Hg and gold mining.41 The northern region has a well-documented legacy of gold mining, where Hg was used for instream recovery of gold. In the northern region, two gold mines and one Hg mine were located upstream of the Mad River site, 79 gold mines and one Hg mine were located upstream of the South Fork Trinity River site, and 99 gold mines were located upstream of the Clear Creek site. Conversely, the only southern region site that contained historic gold mines was the North Fork Kaweah River, which had nine gold mines within its watershed. However, this area is void of many mineral resources, and the mines may have been more exploratory in nature.42 After controlling for body mass effects, we found RBC THg concentrations in turtles from the southern region were 28% lower than the concentrations in those from the northern region. This spatial pattern for exposure is consistent with regional fish advisories in California and provides further evidence of widespread Hg exposure in multiple aquatic and aquatic-related species in the state.40,41,43 Although spatial heterogeneity and site conditions are important drivers of the bioaccumulation of Hg in aquatic ecosystems, turtle feeding ecology can also dictate the degree of bioaccumulation within a site. We found that Hg concentrations in E. marmorata were higher in larger turtles across both regions, a relationship that is consistent with other studies of turtle exposure in which body mass was positively related to blood THg concentrations.37,44 However, bivariate linear models used in those studies generally resulted in little

Figure 4. Partial residual plot of the relationship between western pond turtle (E. marmorata) plasma free triiodothyronine (T3) and red blood cell total mercury (THg) concentrations after controlling for site effects. Partial residual plots illustrate the relationship between two continuous variables after controlling for other factors that can influence the dependent variable.

variable accounted for a cumulative 42% of the total AICcWt. Additionally, the five highest R2GLMM(c) values were all associated with models that included THg as a fixed effect, suggesting the importance of THg concentrations in explaining the variation in T3. However, the top model explained only 35% of the variation in T3 [R2GLMM(c) = 0.35] and all models had an R2GLMM(m) of