Mercury bioaccumulation, genotoxic and biochemical biomarkers reveal the health status of yellow-spotted Amazon River turtles (Podocnemis unifilis) in an environmental protection area in the Amazon

ABSTRACT Chelonians are considered good bioindicators of environmental quality. The assessment of the health status of chelonian populations in the Amazon is important because they are traditionally consumed in large numbers in riverine communities and sustainable use reserves. The present study aimed to evaluate the health of Podocnemis unifilis (Testudines, Podocnemididae) in an environmental protection area in the Amazon region in Brazil. We analyzed the biomarkers lipoperoxidation, carbonylation of proteins, occurrence of micronuclei and erythrocytic nuclear abnormalities, quantified metallothioneins, and evaluated mercury bioaccumulation. We generated pioneering data regarding biomarkers in wild Amazonian freshwater turtles. All biomarker responses did not vary significantly between the sexes. The occurrence of oxidative and genotoxic damage, as well as concentrations of metallothioneins was low compared to other studies. In addition, the bioaccumulation of mercury in the muscle of the animals was below the limit recommended for human consumption by the World Health Organization. Our results provide baseline data on mercury bioaccumulation and biomarker responses that can be useful for future comparisons with other freshwater turtles. The data also provide evidence that the sustainable exploitation of these turtles in the study area (Piagaçu-Purus Sustainable Development Reserve) does not pose a risk to local riverine communities, as the detected mercury concentrations are safe for human consumption. In this sense, our results highlight the importance of protected areas for the conservation of healthy turtle populations, at the same time ensuring the health of the human populations that use them as a food resource.


INTRODUCTION
Biomarkers are widely used in biomonitoring studies, as they represent useful tools for assessing the health of aquatic organisms, including turtles (Labrada-Martagón et al. 2011;Camacho et al. 2013;Casini et al. 2018). Molecular and cellular biomarkers have been used in studies related to reptile protection and conservation initiatives, especially involving marine and freshwater turtles (Schneider et al. 2010;Camacho et al. 2012Camacho et al. , 2013Zapata et al. 2016).
Oxidative stress is an imbalance in cellular homeostasis between antioxidant defense systems and oxidizing agents, and is commonly known as the route of greatest production of reactive oxygen species (ROS) (Dalle-Donne et al. 2003;Yan and Forster 2011). The imbalance generated by excess ROS can be caused by a number of pollutants, such as mercury (Monteiro et al. 2010), organochlorine pesticides and other non-essential metals (Labrada-Martagón et al. 2011). The process culminates in the oxidation of molecular components essential to life, such as lipids (lipoperoxidation), proteins (carbonylation of proteins) and DNA, and can thus directly affect the health of organisms (Dalle-Donne et al. 2003;Yan and Forster 2011). Oxidative damage to cellular lipid components causes severe disturbances to the cell membrane system, which ultimately results in loss of membrane integrity and permeability (Stark 2005, Monteiro et al. 2010. In an interconnected way, the final products of lipoperoxidation, malondialdehyde (MDA) and 4-hydroxenonenal (HNE), can oxidize proteins, forming carbonyl groups in their amino acid side chains, and generating partial or total loss of biological activity (Dalle-Donne et al. 2003;Yan and Forster 2011).
ROS can establish genotoxic effects, such as DNA breaks, which compromise the expression and conservation of genetic information, which can be evidenced in morphological changes in erythrocyte nuclei (Evans et al. 2004;Finlayson et al. 2019). Metallothioneins are low molecular weight proteins that act regulating the levels of essential and non-essential metals in tissues. This control indirectly allows reduction of ROS generation, since high concentrations of metals can affect the redox balance in the cellular environment (Schlenk et al. 2008).
Environmental protection areas are intended to ensure representative samples of ecologically viable populations and play an important role in the conservation of endangered species (Pantoja-Lima et al. 2014). These areas are appropriate to obtain data on health of the biota that can be used as reference parameters for areas where pollution and other environmental impacts can affect the health status of animals (Camacho et al. 2013;Casini et al. 2018). Yet there are few such data from protection areas.
Podocnemis unifilis (Troschel, 1848, Testudines: Podocnemididae) is one of the most abundant freshwater turtle species in the Amazon region (Vogt 2004) and is listed as vulnerable in the IUCN Red List of Threatened Species (IUCN 2017). It is traditional food resource of human populations in the region and is still widely consumed in riverine communities, including those in sustainable use reserves (Waldez et al. 2013). The species was chosen as a bioindicator to generate baseline health data from a protected area and ensure the safe consumption of turtles by the local communities (Kemenes and Pezzuti 2007;Fagundes et al. 2016).
In this context, the objectives of this study were: (1) to evaluate the health status of the yellow-spotted Amazon River turtle (Podocnemis unifilis) in an environmental protection area through biomarkers that indicate damage in macromolecules, nuclear morphological changes in blood erythrocytes and mercury levels in muscle; and (2) to provide information about food security for human populations in the area regarding mercury concentrations in the turtles.
After biometric measurements (maximum straight carapace length -MSCL and total wheight -TW), 2 mL blood samples were obtained via caudal or cervical puncture with heparinized syringes (25 x 7 mm) and stored in cryogenic tubes. The samples were later used to obtain blood smears on microscopy slides. The slides were dried at room temperature and, after 24 hours, were fixed in absolute ethanol for 30 minutes. After blood sampling, the individuals were euthanized at the collection site with a lethal intramuscular injection of propofol (CFMV 2002;CFBio 2012). Once death was confirmed, dissection was performed via removal of the plastron, and samples of the muscle were taken from the pectoral region. The samples were stored in liquid nitrogen and transported to the Laboratório de Ecotoxicologia Aquática na Amazônia at Instituto Nacional de Pesquisas da Amazônia (INPA), where they were stored in an ultra-freezer at -80 °C. Considering that the turtles in PP-SDR can be subjected to unknown upstream pollutant sources, we also sampled six animals (four females and two males) from the Bicho do Rio turtle breeding farm (BRF), located by the AM 70 highway, highway marker 27, Iranduba, Amazonas state, VOL. 52 (3)

Biochemical biomarkers
Muscle fragments (± 300 mg) were homogenized in sodium phosphate buffer (0.1 M, pH 7.5), at a ratio of 1:10 (weight:volume). The homogenate was centrifuged at 15,000 g for 30 minutes at 4 °C. Subsequently, aliquots for the analysis of lipoperoxidation and carbonylation of proteins were separated and stored in a -80 °C ultra-freezer.

Lipoperoxidation (LPO) -
The content of malondialdehyde (MDA) in the homogenate was measured by the content of substances that are reactive to thiobarbituric acid (TBARS) (Ohkawa et al. 1979). The quantification of MDA was obtained by means of the molar extinction coefficient of 1.56 x 10 6 m -1 cm -1 and expressed in nmoles of MDA gram of wet weight -1 , with spectrophotometry reading at 535 nm (Spectramax Plus, Molecular Devices, USA).

Protein carbonylation (PCA) -
The quantification of carbonylated proteins was performed using the reaction of DNPH (2,4-dinitrophenyl-hydrazine) with carbonyl groups in amino acid residues, and generated compounds such as dinitrophenyl-hydrazones that are detected by spectrophotometry at 370 nm (Spectramax Plus, Molecular Devices, USA) (Levine et al. 1994). The results were expressed in µmol.mg protein -¹.

Metallothioneins (MT) -
Muscle samples (± 100 mg) were homogenized in Tris-HCl/sucrose buffer (20 mm/500 mm, pH 8.6), phenylmethylsulfonyl fluoride (0.1 m, 0.5 mM) and β-mercaptoethanol (0.01%) at a ratio of 1:5 (weight:volume), and centrifuged at 15,000 g for 30 min at 4 °C. The MTs were analyzed using the method proposed by Viarengo et al. (1997). A series of precipitations, centrifugations and resuspensions was carried out, resulting in the isolation of the MTs in the pellet, with the addition of Ellman's solution (DTNB 0.4 mM, sodium phosphate buffer NaCl 0.2/2 M, pH 8). The MTs have sulfhydryl groups (SH) in their cysteine residues and, from the comparison with the reduced glutathione curve (GSH) as a reference, the absorbance was determined at 412 nm (Spectramax Plus, Molecular Devices, USA). The MT concentration was expressed in µg mg of protein -1 .
Protein concentration -Bradford's method was used for determination of the protein concentration at 595 nm (Spectramax Plus, Molecular Devices, USA). Bovine serum albumin was used as the standard (Bradford 1976).

Genotoxicity biomarkers
The blood smears were stained with Giemsa 10% to determine the frequency of micronuclei and ENAs. Two thousand cells per individual were used to identify morphological changes and count frequencies (Carrasco et al. 1990).

Total mercury (THg)
THg concentrations in the muscle were determined in six animals (three males and three females) using the methodology of Bastos et al. (2015). Approximately 500 mg (wet weight) of muscle was digested in a solution of H 2 SO 4 and HNO 3 (1:1) and KMnO 4 (5%). After digestion, KMnO 4 solution (5%) was added to the samples for 1h 30 min in a digestion block (Tecnal-Mod.007a, Piracicaba, São Paulo, Brazil) at 60 °C. The samples were cooled at room temperature (± 25 °C) and hydroxylamine hydrochloride (12%) was added. The digested samples were diluted with deionized water (Milli-Q Plus, Millipore, Bedford, MA, USA). The THg reading was obtained using an atomic absorption spectrophotometer (Flow Injection Mercury System -FIMS -400 -Perkin Elmer, Ueberlingen, Germany). The reference material used was DORM-4 (fish protein certified reference material for trace metals, NRC National Research Council, Canada), with an average recovery of 100% for each battery of samples and a detection limit for THg equal to 0.0007 mg kg -1 .

Statistical analysis
The response variables were submitted to the Kolmogorov-Smirnov normality test and the Levene test for homocedasticity. All variables were non-conformant with normality assumptions. The Mann-Whitney U test was used to compare biochemical biomarkers, genotoxicity biomarkers and THg between the sexes, and also between PP-SDR and BRF, and the weight and length of the animals between PP-SDR and BRF. All results were expressed as mean ± standard deviation, and test results were considered significant at p < 0.05 (Zar 1996). Since the variables did not conform to normal distribution, we also represented the data as median and interquartile range, and in point dispersion graphs (Weissgerber et al. 2015). The basic and agricolae packages of the statistical software R Core Team, version 3.4.1 (R CoreTeam 2017) and Statistics IBM-SPSS 22 were used to perform the analyses.

ACTA
AMAZONICA associated with responses to biomarkers, as it is related to age and, possibly, to the time of exposure to toxic substances (Schneider et al. 2011). Information about the size/age relationship is also important because some chelonians, such as P. unifilis, undergo ontogenetic variation in diet, which can make them more vulnerable to pollutants in certain growth stages (Balensiefer and Vogt 2006).
The response of biomarkers can vary between sexes in turtles. For example, females of Trachemys scripta (Thunberg & Schoepff, 1972) may bioaccumulate less metals than males as they are able to excrete metals through the eggs (Burger and Gibbons 1998). Higher concentrations of mercury were observed in the muscle of females compared to that of males in Podocnemis sextuberculata (Cornalia, 1849) (Schneider et al. 2011). Variation between sexes was also observed in the land iguana, Conolophus subcristatus (Gray, 1831), with males showing lower oxidative damage than females, indicating that females have lower antioxidant capacity, especially in the reproductive period (Costantini et al. 2009). In the present study, the similar response to biomarkers of males and females may be related to the diet of P. unifilis, as all sampled individuals were adults, which are predominantly herbivorous in both sexes (Balensiefer and Vogt 2006;Lara et al. 2012;Souza-Araujo et al. 2015).

Genotoxicity
Erythrocytes of all sampled infividuals presented normal nuclei, as well as micronuclei and nuclei in the following formats: kidney-shaped, lobed, protuberant, blebbed, vacuolate and segmented. There was no significant difference between the sexes in frequency of micronuclei and any nuclei format in PP-SDR (23 males, 11 females) and BRF (2 males, 4 females). Despite the difference in size and weight of animals sampled in PP-SDR and BRF, the frequency of genotoxicity biomarkers did not differ significantly between both sites (Table 2).

DISCUSSION
The relationship between biomarkers and animal size or sex has been studied for some reptiles (Costantini et al. 2009;Amaral et al. 2012 (Furtado-Filho et al. 2007). Levels of 9.54 ± 0.9 nanomol g w.w. -1 were obtained in red muscle of the negative control group of the freshwater turtle Trachemys scripta elegans (Wied, 1839) subjected to hypoxia (Willmore and Storey 1997). Our values were also lower than those recorded in the muscle of a sea turtle Chelonia mydas agassizii (Bocourt, 1868) (63.7 ± 7.4 nanomol g w.t. -1 ) caught in a relatively undisturbed feeding area in the Eastern Pacific (Valdivia et al. 2007). Thus, our results indicate that there is no evidence of oxidative damage to lipid components that might compromise the health of P. unifilis in PP-SDR.
PCA is the most common way to evaluate oxidative damage to the protein components of cells. The blood of healthy adults of Chelonia mydas (Linnaeus, 1758) under the influence of non-essential metals from the Atlantic Ocean contained levels of 2.48 ± 0.25 nanomol mg protein -1 (da Silva et al. 2016). Brain and liver of the negative control group of juvenile Chrysemys picta (Schneider, 1783) subjected to hypoxia and cold contained 3.7 ± 0.4 and 2.2 ± 0.2 nanomol mg protein -1 , respectively (Baker et al. 2007). Average PCA in muscle of Chinese soft-shelled turtles, Pelodiscus sinensis (Wiegmann, 1835) was 19 nanomol mg protein -1 in a group submitted to infection with two furunculosis pathogens, and 7 nanomol mg protein -1 in the control group (Li et al. 2021). To our knowledge, there are no published studies on PCA related to metals in turtle muscle, but the comparison with the aforementioned studies indicates that our results for PCA (1.63 ± 0.40 nanomol mg protein -1 ) represents a low level of oxidative damage.
MT are widely studied in aquatic organisms, but, regarding chelonians, there are only a few studies on marine turtles (Andreani et al. 2008;Sinaei 2016). MT in the liver and kidney of Chelonia mydas from Costa Rica and Caretta caretta (Linnaeus, 1758) from the Mediterranean and Adriatic seas were positively correlated with cadmium and copper, emphasizing the role of these proteins in metal detoxification (Andreani et al. 2008). MT concentration in the blood of C. mydas was positively correlated with the concentrations of mercury, cadmium, lead, copper and zinc and the turtles were capable of responding quickly and efficiently to environmental contamination (Sinaei 2016). Although these studies have used different methodologies, units of measurement and tissues, they demonstrated a relatioship between the levels of metals and MT content in tissues, especially the induction of these metalloproteins by mercury in aquatic organisms, as already evidenced in C. mydas (Sinaei 2016). Our results suggest that the levels of MT are also basal and normal for P. unifilis in PP-SDR. Studies that associate MT concentration with the bioaccumulation of other elements in turtle tissues are needed in this region.
ENA are a relevant biomarker as they indicate early stages of damage to genetic material (Shimizu et al. 1998;Crott and Fenech 2001) and, as such, can be useful for the establishment of protective measures that simultaneously reverse the damage and prevent the formation of micronuclei, that represent irreversible DNA damage. Casini et al. (2018) have generated data regarding these biomarkers of genotoxicity in C. caretta from the Mediterranean Sea (see Table 1), which is an environment with high levels of pollution by metals, petroleum products and insecticides. The authors reported high frequencies of biomarkers of genotoxicity in the sampled animals and observed a positive correlation with the concentrations of measured contaminants (polycyclic aromatic hydrocarbons). Based on Casini et al. (2018), we can consider that the oxidative damage observed in the genetic material of the species in PP-SDR is basal.
The frequency of micronuclei in P. unifilis from BRF (1.47 ± 0.74% 0 ) and PP-SDR (1.44 ± 0.59% 0 ) was lower than that reported for juvenile P. unifilis exposed to cadmium in the diet for 60 days (12 ± 5% 0 , n = 12), and in the respective control group (8 ± 1% 0 , n = 12) (Frossard et al. 2013). It was similar to the frequency reported for the control group of Trachemys callirostris (Gray, 1855) exposed to chemical contaminants (0.78 ± 0.58% 0 ; n = 20) (Zapata et al. 2016). Since this biomarker represents the final stages of genotoxicity (Shimizu et al. 1998;Crott and Fenech 2001;Yasui et al. 2015) the low frequency of micronuclei indicates normal cell cycle condition and replication of the genetic material. Thus, in comparison with the aforementioned studies, we reported the occurrence of basal and normal frequencies of micronuclei in erythrocytes of P. unifilis in PP-SDR.
Although our P. unifilis from PP-SDR were significantly larger than those from BRF, the levels of genotoxicity were similar between the sites and are assumed to be representative of basal frequencies. The results confirm the good environmental conditions maintained at BRF, where even the smallest and youngest animals, which also include items of animal origin in their diet, feed on uncontaminated prey. The low and similar levels of genotoxicity biomarkers indicate the absence of DNA strand break inducers in the environment both at BRF and at PP-SDR.
Mercury is a current concern for human and animal health in the Amazon region (Fadini and Jardim 2001) due to its high toxicity to tissular, neurological and cellular components of organisms (Chan et al. 2003;Barcelos et al. 2011). Mercury can enter aquatic ecosystems by anthropogenic and natural processes. In the Amazon, mining activity is the main anthropogenic source of mercury, however, there are natural sources in soils that are ultimately incorporated into aquatic systems (Fadini and Jardim 2001;Roulet et al. 2001). In the Purus River channel, analyses of water, river bed sediments, suspended solids and fish tissues have revealed that the VOL. 52(3) 2022: 254 -263 ACTA AMAZONICA lithogenic mercury intake is more significant than that from external sources (Brabo et al. 2003;Castro et al. 2016). THg levels in the muscle of P. unifilis in this study were similar to those found in P. unifilis in the Purus River (Schneider et al. 2015) and in the lower Xingu River (Souza-Araujo et al. 2015), and were lower than the concentrations observed in the Negro River basin (Schneider et al. 2010(Schneider et al. , 2011 and another study from the lower Xingu River (Pignati et al. 2018) (see Table 3).
The maximum limit of mercury consumption established for humans is 500 ppb or 0.5 µg g -1 (ANVISA 1998;WHO 2008). The consumption of turtles that live in areas contaminated by mercury and other metals can represent a risk to riverine populations. Chelus fimbriatus (Schneider, 1783) from the Negro River showed high mercury concentrations of 432 ± 195.5 ppb (Schneider et al. 2010). Higher concentrations of mercury tend to be observed in tissues of carnivorous species, such as C. fimbriatus, which can be attributed to the biomagnification of mercury in trophic chains. Podocnemis unifilis is a predominantly herbivorous species, therefore, low concentrations of mercury were expected in this species (Schneider 2009(Schneider , 2010. Indeed Pignati et al. (2018) and Souza-Araujo et al. (2015) observed concentrations twice as high as those observed in our study in P. unifilis from the Xingu River basin. Schneider et al. (2011) and Souza-Araujo et al. (2015) also found no significant variation of THg between the sexes in P.unifilis, which is attributed to both sexes belonging to the same trophic level (predominently herbivorous) (Lara et al. 2012;Souza-Araujo et al. 2015). A negative and significant correlation between mercury and body size was reported for young individuals of P. sextuberculata, and therefore smaller turtles had higher levels of mercury than larger individuals was attributed to ontogenetic alterations in the diet (Schneider et al. 2010). The absence of an expressive variation in size within our sample suggests that the animals have similar age and, therefore, similar diets and exposure times to the environment. Additionally, these turtles tend to achieve a rapid balance between the rate of elimination of mercury and the rate of consumption of this metal (Schneider et al. 2010(Schneider et al. , 2011. The consumption of subadult and adult P. unifilis is a very common practice in PP-SDR (Waldez et al. 2013), and about 16% of eggs from P. unifilis nests are protected by riverine communities through participatory management in the PP-SDR and consumed locally (Erickson et al. 2020a). The concentrations of mercury observed in our study are 45 times lower than the threshold for human consumption. These values suggest that the risk of mercury contamination from consumption of P. unifilis for the inhabitants of PP-SDR is minimal. However, it is important that mercury levels in P. unifilis are monitored in larger samples and wider spatial and temporal scales, as the species has a generalist behavior and phenotypic plasticity to respond to variable environmental conditions (Erickson et al. 2020a(Erickson et al. , 2020b. It would also be relevent to evaluate biomarkers and bioaccumulation of metals on a wider scale in congeneric Podocnemis, which generally exhibit more specialized behavior than P. unifilis and are restricted to narrower environmental conditions, such as P. erythrocephala (Batistella and Vogt 2008), P. sextuberculata (Haller and Rodrigues 2006), P. lewyana (Páez et al. 2009), P. vogli (Rueda-Almonacid et al. 2007) and P. expansa (Vanzolini 2003).

CONCLUSIONS
Our study generated pioneering data regarding biomarkers in Amazonian freshwater turtles collected in the field. We demonstrated low levels of oxidative and genotoxic damage, as well as low concentrations of metallothioneins and mercury in muscle tissue of P. unifilis in an environmental protection area. Biomarker response did not differ between sexes. The biomarkers analyzed suggest that there is no evidence of damage to the health of P. unifilis in the Piagaçu-Purus Sustainable Development Reserve, and that the reserve seems to be fulfilling its function of preserving the population of this turtle by maintaining a good quality environment. In addition, low concentrations of mercury in the tissues sampled indicate that the sustainable exploitation of these turtles does not pose a risk to local riverine communities, as the detected concentrations are safe for human consumption. Our results provide a set of data on mercury bioaccumulation and biomarker response that can be useful for future comparisons with freshwater turtles. We also provide evidence for the effectiveness and importance of protected areas for the conservation of healthy turtle populations, and also to ensure the health of the human populations that use them as a food resource. The biomarkers evaluated in this study have shown to be adequate tools for biomonitoring the health of Amazonian freshwater turtles and can be applied in biomonitoring of other protected areas. As a future perspective, other categories of biomarkers can be included in the analyses, as well as a larger sample sizes.