Route of exposure influences pesticide body burden and the hepatic metabolome in post-metamorphic leopard frogs

Abstract

Pesticides are being applied at a greater extent than in the past. Once pesticides enter the ecosystem, many environmental factors can influence their residence time. These interactions can result in processes such as translocation, environmental degradation, and metabolic activation facilitating exposure to target and non-target species. Most anurans start off their life cycle in aquatic environments and then transition into terrestrial habitats. Their time in the aquatic environment is generally short; however, many important developmental stages occur during this tenure. Post-metamorphosis, most species spend many years on land but migrate back to the aquatic environment for breeding. Due to the importance of both the aquatic and terrestrial environments to the life stages of amphibians, we investigated how the route of exposure (i.e. uptake from contaminated soils vs. uptake from contaminated surface water) influences pesticide body burden for four pesticides (bifenthrin (BIF), chlorpyrifos (CPF), glyphosate (GLY), and trifloxystrobin (TFS)) as well as the impact on the hepatic metabolome of adult leopard frogs. Body burden concentrations for amphibians exposed in water were significantly higher (ANOVA p < 0.0001) compared to amphibians exposed to contaminated soil across all pesticides studied. Out of 80 metabolites that were putatively identified, the majority expressed a higher abundance in amphibians that were exposed in pesticide contaminated water compared to soil. Ultimately, this research will help fill regulatory data gaps, aid in the creation of more accurate amphibian dermal uptake models and inform continued ecological risk assessment efforts.

Graphical Abstract

1. Introduction

There are numerous types of pesticides used globally to combat insects, fungi, weeds, and other agricultural nuisances. Predominantly, based on usage, they are classed as herbicides, insecticides, fungicides, and other smaller classes (Gilliom et al., 2006; Potter et al., 2014). Herbicides are the most frequently applied pesticides in the United States, due to their accepted application on staple crops such as corn, soybeans, cotton, and wheat (Fernandez-Cornejo et al., 2014). While pyrethroids and neonicotinoids require lower field application rates and some formulations can be applied as seed treatments limiting broadcast applications and non-target exposures (Fernandez-Cornejo et al., 2014). In several studies across the US, bifenthrin is often the most frequently detected insecticide and is frequently detected at levels higher than LOECs determined for aquatic arthropods (Mimbs IV et al., 2016; Smalling et al., 2015). Fungicides are used on a wide variety of crops including many fruits, vegetables, and grains, as a preventative rather than a curative measure (Battaglin et al., 2011). In the U.S., four different fungicides were registered in 2006 to treat soybean rust, but by 2009, this number increased to 14 (Battaglin et al., 2011). Overall, the type of herbicide or formulation applied depends on the crop, targeted nuisance plants, and intended land use (Gilliom et al., 2006). Currently, pesticides are being applied at a greater extent than in the past due in part to the availability of more formulations and/or marketed varieties (Mann et al., 2009). While pesticides are typically applied in agricultural settings, approximately 20% of these compounds are approved for use for non-agricultural purposes (Stokstad and Grullón, 2013).

Amphibians are the most rapidly declining and threatened taxa globally, compared to other vertebrates such as avians and mammals (Stuart et al., 2004). Since the 1980s, over 120 species of amphibians have become extinct, and over one-third of amphibian species are currently threatened world-wide (Whitfield et al., 2007). It has been thoroughly demonstrated that amphibian populations are in decline and it is hypothesized that pesticide exposure is one of the primary causative factors in addition to habitat loss, pollution, climate change, and diseases (Aldrich et al., 2016; Christin et al., 2003; Lötters et al., 2014; Sparling et al., 2001; Stuart et al., 2004). This rapid decline in amphibian populations is not the result of a single stressor but rather the endpoint of numerous and complex stressor interactions (Buck et al., 2012).

Amphibians are a unique class of vertebrates due to their varied life histories and thus can be considered valuable indicators of ecosystem health. The anuran lifecycle most commonly includes embryonic and tadpole stages that develop in aquatic environments. Despite the relatively short duration of the lifecycle that takes place in water, the complexity and importance of these early developmental stages cannot be understated. Post-metamorphosis, most species spend many years on land but must migrate back to the aquatic environment for breeding (Salice et al., 2011; Venturino et al., 2003). This bi-phasic life history leaves amphibians vulnerable to pesticide exposure in both aquatic and terrestrial habitats. Since amphibians have relatively permeable skin, dermal absorption is the major route of exposure for these contaminants (Smith et al., 2007).

Amphibian skin is well known for being relatively permeable to both respiratory gases and water. Thus, many anurans lack a hydrophobic barrier due to the absence of epidermal scales and protective layers (i.e., thick stratum corneum) observed in other terrestrial species (Hillman et al., 2009). This has resulted in the ability to readily exchange ions, gases, and water with their surrounding environment; however, their dermal composition comes with consequences. Due to the ease of water moving between amphibians and the environment, it is challenging for anurans to remain constantly hydrated. Therefore, multiple physiological and behavioral adaptations have evolved to compensate for this ease of water loss (Barbeau and Lillywhite, 2005; Hillman et al., 2009; Suzuki et al., 2007). Since amphibians need to stay hydrated and do not physically imbibe water, they have developed a highly vascularized seat patch. This vascularized seat patch is in the posteroventral region of anurans and is the primary route of absorption of water (Bentley and Main, 1972). Thus, understanding how pesticide exposure to amphibians influences toxicity is necessary to accurately assess the ecological risks these compounds pose.

Among amphibians, aquatic contact can occur as embryos, larvae, or as adults returning to breed at a pond or water surface (Salice et al., 2011). Many of these ponds are situated around agricultural fields where pesticides are often applied (Mann et al., 2009). Typically, breeding season and embryonic and larval development occur during spring and into early summer. Anuran breeding season frequently coincides with the planting of crops and application of pesticides to these agricultural areas (Berger et al., 2011; review in Fryday and Thompson, 2012). This concurrent timing results in herbicides, fungicides, and insecticides being applied over fields, so that amphibians, along with other non-target species, are likely exposed to these compounds.

Metabolomics can be utilized in the elucidation of the consequence of pesticide exposure in amphibians across their life stages (Glinski et al., 2019; Snyder et al., 2017; Van Meter et al., 2018). Following pesticide exposure, differences in metabolite fluxes in the hepatic metabolome can aid in the identification of specific pathways that are potentially affected before overt toxicity occurs. Understanding the interplay of these impacted biological pathways can be used to predict or estimate the overall consequence of exposure to these pesticides in amphibians and other non-target species. The objective of our research was to assess pesticide exposure and subsequent dermal uptake in amphibians by measuring the body burdens of pesticides (n = 4) along with utilizing metabolomics to investigate biochemical fluxes resulting from exposure. Due to the importance of both the aquatic and terrestrial environments to the life stages of amphibians, we investigated how the route of exposure (i.e., uptake from contaminated soils vs. surface water) influences pesticide body burden as well as the impact on the hepatic metabolome in post-metamorphic leopard frogs (Lithobates sphenocephala). Ultimately, this research will lend further insight to the importance of pesticide contamination in these matrices and the risks they pose to amphibians during critical life stages.

2. Materials & Methods

Extraction and analytical solvents used for analysis of pesticides were purchased from Fisher Scientific (Pittsburg, PA, USA) and of >HPLC grade purity. Pesticide active ingredients and their corresponding metabolites were obtained from the U.S. Environmental Protection Agency’s National Pesticides Standard Repository (Fort Meade, MD, USA). Parent pesticides and their metabolites analyzed in the study were ≥98% purity and included bifenthrin (BIF), 4-hydroxy bifenthrin (4-OH BIF), chlorpyrifos (CPF), chlorpyrifos oxon (CPO), glyphosate (GLY), trifloxystrobin (TFS) and trifloxystrobin acid (TFSa). These pesticides were selected due to their high usage patterns with estimated US annual usage rates range from 0.5 million pounds per year (trifloxystrobin) to over 250 million pounds per year (glyphosate) (USGS, 2016) coupled to limited data available for uptake rates and biological perturbations in amphibians. These pesticides are frequently detected in bed sediments and water in amphibian habitats (see Gilliom et al., 2006; Smalling et al., 2012; Battaglin et al., 2016). Relevant physiochemical parameters for the pesticides are detailed in Table 1.

Table 1.

Summary of the physiochemical properties (measured or estimated) for the pesticides used in the current study.

Chemical	CAS #	Abbreviation	Water Solubility (mg/L)a	Organic Carbon Partition Coefficient (Ka)	Log Kowa	Log BCFa	Log BAFa	BCF (Soil)	BCF (Water)
Bifenthrin	82657–04–3	BIF	0.1	2.27×106	8.15	2.3	4.671	0.33	1
Chlorpyrifos	2921–88–2	CPF	1.12	7283	5.11	2.94	3.327	2.84	11.5
Glyphosate	1071–83–6	GLY	1.05×104	1	−4.77	0.5	−0.049	ND	ND
Trifloxystrobin	141517–21–7	TFS	0.39	3.04×106	6.62	2.636	2.865	0.1	1.73

^aValues were obtained from EPI suite (U.S. EPA, 2012).

2.1. Dermal Exposure Experiments

Egg masses of southern leopard frogs (L. sphenocephala) were collected from vernal pools in Queen Anne’s County (MD, USA) and taken back to Washington College for rearing. Amphibian rearing was previously described in detail in Van Meter et al. (2019). Soil was classified as Orangeburg loamy-sand (OLS), collected in 2014 from the Joseph Jones Ecological Research Center in Newton, GA, sieved and stored at 4 °C until use. In-house spring water was allowed to degas overnight prior to its use in the water exposure experiments. Individual pesticides were mixed in 2 L amber glass bottles at a nominal concentration of 1 ppm (μg/mL), except glyphosate which was maintained in high density plastic due to its propensity to adsorb to glass. A concentration of 1 ppm was chosen as the target because this represents the upper range of on-field exposure concentrations. Although the range of environmentally relevant instream concentrations for these chemicals is 0.0001–0.01 ppm (e.g., Battaglin and Fairchild, 2002; Battaglin et al. 2005; Carpenter et al., 2016); reasonable agricultural on-field exposure concentration estimates for runoff pooled in trenches, ditches, puddles and irrigation ponds are often 2–3 orders of magnitude higher than instream concentrations (e.g., Bennett et al., 2005; EFSA, 2011). For the water exposures, 50 mL of 1 ppm active ingredient pesticide was added to a Pyrex® bowl and then arranged using a completely randomized design. The soil experiments were similar to Van Meter et al. (2016), briefly a thin layer of soil (~150 g) was added to the bottom of a Pyrex® bowl, followed by the application of a pesticide at 50 mL at 1 ppm to the top of the soil using an aerosolized spray bottle. Applying 50 mL at 1 ppm ensured that the water and soil exposures contained an equal amount of pesticide for dermal route exposure comparisons by pesticide. All exposures were conducted in glass bowls, again, except glyphosate exposures which were conducted in plastic Teflon® containers. Following application of pesticides to soil and water bowls, individual leopard frogs (~2 g; 60–90 days post-metamorphosis) were placed in each bowl and the bowls were covered with a mesh screen secured with a rubber band for the duration of the exposure 8 h (5 pesticide treatments × 2 media × 8 reps; N = 80). After exposure, amphibians were euthanized by submersion in liquid nitrogen and stored at −80 °C until processing and extraction.

2.2. Water Extraction for Pesticide Analysis

Following exposure water samples were extracted similar to (Glinski et al., 2018c), all water samples were spiked with 10 μL of a 1000 ppm tetraconazole solution and then passed through a solid phase extraction (SPE; Oasis HLB 6 cc, 200 mg) cartridge that was pre-conditioned with 6 mL each of methanol (MeOH) and water (18.2 MΩ) under pressure on a vacuum manifold. After the sample cleared the SPE column, 10 mL of water (18.2 MΩ) was used to wash the bowl and tubing and added to the SPE cartridge. The SPE cartridge was then vacuum dried for 45 min before being wrapped in aluminum foil and placed in the −20 °C freezer until pesticide elution. Each control, BIF, CPF, and TFS SPE cartridge was eluted with 5 mL of MeOH and 5 mL of dichloromethane, sequentially into glass cuvettes. Samples were blown down to 1 mL and transferred to a 2 mL glass vial along with three 200 μL solvent rinses. Water samples were blown to dryness and reconstituted in 1 mL of methyl tert-butyl ether (MTBE) for GC–MS analysis of parent pesticides. Nominal concentrations of glyphosate were not analyzed in the current study due to analytical constraints.

2.3. Soil Extraction for Pesticide Analysis

Previous pesticide extraction methods for amphibians followed those outlined in (Van Meter et al., 2014). At the end of the exposure, three composite soil samples were collected from the top layer of each bowl/container. Soil samples (~5 g) were spiked with 3 μL of 1000 ppm tetraconazole (internal standard; I.S.) for CPF and TFS. For the BIF exposure studies, samples were also spiked with 3 μL of a 1000 ppm stock of permethrin as I.S. for pyrethroids. Soil samples were extracted by the addition of 5 mL of MeOH. Samples were vortexed for a few seconds, sonicated in a chilled water bath (20 min), and finally centrifuged (3250 rpm) for 10 min. The supernatant was transferred to a 20 mL scintillation vial and evaporated to 1 mL under a gentle stream of nitrogen. The soil was extracted again following the same procedure. To the concentrated methanolic solution, 10 mL of water (18.2 MΩ), 3 mL of MTBE, and ~1 g of sodium sulfate was added. Samples were vortexed for 10 s and allowed to gravimetrically separate into 2 layers. The top organic layer was transferred to a 2 mL vial for analysis by GC–MS.

2.4. Hepatic Metabolomics and Amphibian Extraction for Pesticide Analysis

Amphibian pesticide analysis and its corresponding hepatic metabolomics have been previously described in Van Meter et al. (2018) and Glinski et al. (2019). Approximately 20 mg of liver tissue was excised from thawed amphibians after a final weight was obtained. Liver samples were placed into a pre-labeled 2 mL centrifuge tube which was immediately placed on ice. Extraction and analysis of endogenous hepatic metabolites using a bi-phasic protocol was described in Viant (2007). Briefly, amphibian liver samples were extracted with a bead homogenizer using step-wise additions of methanol:water (75:25 v:v) and chloroform resulting in both a polar and nonpolar metabolite-containing layer. Aqueous extracts were transferred to GC–MS vials and evaporated to dryness and stored at −20 °C until derivatization and metabolomic analysis.

Whole body amphibian homogenates were extracted similarly to soil samples described above. Briefly, internal standards were added, and samples were dehydrated on a freeze dryer for 24 h. MeOH was added, in two additions to the homogenates, and each layer was transferred to a 20 mL scintillation vial and blown down to 1 mL under a steady stream of nitrogen gas. Following each series of vortexing, sonication and centrifugation, the supernatants was combined into the same scintillation vial. Again, the samples underwent liquid-liquid extraction with 10 mL of water (18.2 MΩ), 3 mL of MTBE, and ~1 g of sodium sulfate. Each sample was vortexed and then allowed to settle to obtain two layers. The top organic layer was transferred to a 2 mL vial for GC-MS analysis.

2.5. Parent Pesticide GC-MS Analysis

Pesticide active ingredients were analyzed on an Agilent 6890 gas chromatograph coupled to an Agilent 5973 MSD mass spectrometer. The instrument was equipped with a DB-5MS column (30 m, 0.25 μm thickness, and 0.25 mm ID, Agilent Technologies, CA) and 2 μL injections were made in splitless mode. Helium was used as the carrier gas with a constant flow of 1.0 mL/min while the inlet temperature was 280 °C and the source was 230 °C. Initial conditions were held at 80 °C for 2 min and then ramped 10 °C/min to 300 °C with a final hold time of 6 min (total run time = 30 min). All compounds were analyzed in selected ion monitoring mode (SIM); BIF – 181 m/z, CPF – 197 m/z, TFS – 116 m/z, permethrin (IS) – 183 m/z, and tetraconazole (IS) – 336 m/z. Standards were analyzed at the beginning and end of the run, in addition to QA/QC and instrumental blanks analyzed intermittently every 5–10 samples and none of our target analytes were detected in any blanks. The lowest standards used in the calibration curves that were three times the signal to noise were 0.0019 ppm for TFS and BIF, and 0.0078 ppm for CPF. Body burden concentrations of glyphosate were not analyzed in the current study due to analytical constraints.

2.6. Pesticide Metabolite LC-MS Analysis

The metabolites chlorpyrifos oxon (CPO) and trifloxystrobin acid (TFSa) were quantified on an Agilent 1100 HPLC linked to an Agilent 6120 quadrupole mass spectrometer equipped with an Eclipse XDB-C18 column (3.5 μm particle size, 3.0 × 150 mm; Agilent Technologies, CA, USA). Samples (in MTBE) were blown to dryness under nitrogen gas and reconstituted in 30% MeOH. Initial chromatographic conditions were 70% water with 0.1% formic acid (A) and 30% acetonitrile with 0.1% formic acid (B). These conditions were held for 2 min then ramped to 90% B over 16 min and held for 4 min before returning to starting conditions (total run time = 30 min). Both metabolites were analyzed in positive selected ion monitoring mode; CPO – 336 m/z, TFSa – 395.1 m/z, while the internal standard, tetraconazole, was monitored at 372 m/z. Standard calibration curves were analyzed at the beginning and end of each run, while QA/QC and analytical blanks were analyzed throughout the entire sequence, every 5–10 samples and none of our target analytes were detected in any blanks. The lowest standards used in the calibration curves that were three times the signal to noise were 0.0010 ppm for TFS and CPF.

2.7. Sample Derivatization for GC/MS Metabolomic Profiling

Lyophilized aqueous extracts were allowed to reach room temperature prior to sequential derivatization with methoxyamine-HCl (MeOX; 20 mg/mL in pyridine) and bis(trimethylsilyl)trifluoroacetamide with 10% trimethylchlorosilane (BSTFA + 10% TMCS). For derivatization, 75 μL of MeOX was added to each sample, vortexed and heated at 60 °C for 2.5 h. Samples were removed from the oven every 30 min and vortexed. Following methoxyamination, 75 μL BSTFA + 10% TMCS were added and samples were again incubated at 60 °C for 1.5 h. Derivatized samples were transferred to micro-target inserts and analyzed by GC–MS within 36 h.

2.8. GC-MS-based Metabolomics

The derivatized, aqueous liver samples were analyzed on an Agilent 6890 gas chromatograph (GC) coupled to a 5973 mass selective detector equipped with an Agilent DB-5MS column (30 m, 0.25 μm thickness, and 0.25 mm ID). All injections, 2 μL, were made in the splitless mode. The injector temperature was 250 °C and the transfer line was held constant at 280 °C, while the MS source and MS quad were held at 230 °C and 150 °C, respectively. The carrier gas was UHP grade helium and maintained at a constant flow of 0.8 mL/min. The oven temperature program began at 60 °C and was held for 2 min, ramped at 8 °C/min to 300 °C and finally held for 5 min. Mass spectra were acquired from 50 to 650 m/z, with QA/QC and blanks analyzed every 5 samples. Acquired chromatograms were exported as netcdf files, aligned, and processed using developer-recommended parameters for an Agilent GC–MS in MetAlign 041012 (Lommen, 2009). Afterwards, Excel® was used to filter and remove duplicate retention times as described in Niu et al. (2014). The resulting spreadsheet contained the most abundant m/z determined by the MS during each scan for every sample (in columns). Approximate number of spectral features remaining for each sample used for analysis was 6000 retention time:m/z pairs.

2.9. Data Analysis

Body burden concentrations were tested to determine if the soil and water exposure route treatments were significantly different. A one-way blocked analysis of variance (ANOVA) was performed to compare the body burdens resulting from soil and water-based exposures. A Type II ANOVA was performed, as no interaction between water and soil exposures was expected or tested. Pesticide was used as the blocking variable, methanol (not detected) and glyphosate (not analyzed) were not analyzed as part of the ANOVA. The experimental design was complete and balanced, with 8 observations for each treatment and pesticide combination. Data were tested for normality with Shapiro-Wilkes with the null hypothesis being that the data came from a normal distribution. Pesticides with positive skewness were transformed via a reciprocal distribution prior to the ANOVA. The Levene test was used to confirm homogeneity of variances.

Student’s t-test filtered chromatograms, in Excel®, along with MetaboAnalyst 3.0 (www.metaboanalyst.ca/) were used for metabolite identification and subsequent pathway analysis (Xia and Wishart, 2016). The summed absolute abundance of retention time:m/z pairs passing a Student’s t-test were used to construct bar graphs representing the ‘total impact on the metabolome’ for each control:pesticide pairwise comparison. In MetaboAnalyst, truncated spectra were imported and subjected to ANOVA (p < 0.05) and both principle component analysis (PCA) and partial least squares discriminant analysis (PLS-DA). PCA was first used to visualize pesticide-clustering and potential outlier identification while PLS-DA was used to quantitate class separation (via cross validation methods) and generate variable influence in projection (VIP) plots to identify metabolites responsible for class separation. PLS-DA models with cumulative predictability (Q2) of >0.5 were considered valid models and a VIP score of >1.0 was considered a significant metabolite or retention time:m/z pair. Statistically significant metabolites were putatively identified by searching the NIST 2014 spectral database. Spectral similarities above 70 were considered adequate matches and metabolites assigned a putative identification.

2.10. Pathway Analysis

Perturbations in biological pathways (i.e., pathway analysis) were performed with MetaboAnalyst’s metabolite set enrichment analysis module. This analysis focuses on metabolite alterations that are considered topologically important (Xia and Wishart, 2016). Metabolites were searched as pathway-associated metabolite sets (SMPDB) without species specificity.

3. Results

3.1. Pesticide Body Burden

To determine if route of exposure influences pesticide body burden in post-metamorphic leopard frogs, amphibians were exposed to pesticides in either contaminated water (BIF_water, CPF_water, GLY_water, TFS_water) or via contact with contaminated soil (BIF_soil, CPF_soil, GLY_soil, TFS_soil). After 8 h of exposure the media was analyzed for pesticide concentrations (Fig. 1A). Nominal water for bifenthrin was 0.49 ± 0.18 (mean ± standard error) ppm while the soil concentration was 0.33 ± 0.02 ppm. While nominal concentrations for CPF in soil and water were 0.25 ± 0.02 and 0.21 ± 0.09 ppm, respectively. TFS nominal water concentration was 0.56 ± 0.03 ppm and soil concentration was 0.26 ± 0.03 ppm. There were minimal amounts of CPO measured in the CPF water and TFSa was found in both water and soil. Therefore, based on measured nominal concentrations target exposures were 49, 21, 56% for water and 100, 76, 79% for soil exposures (BIF, CPF, TFS, respectively). Body burden concentrations for amphibians exposed in water scenarios are reported below for each pesticide (mean ± standard error), concentrations were statistically significantly higher (p < 0.001) compared to amphibians exposed to contaminated soil overall. Individually, the parent pesticides in the current study and the metabolite TFSa were significant (Fig. 1B). The body burden concentration for the BIF_water exposure was observed to be 0.49 ± 0.03 μg/g and for the BIF_soil 0.11 ± 0.02 μg/g. CPF_water and CPF_soil derived body burden concentrations were 2.42 ± 0.22 μg/g and 0.71 ± 0.10 μg/g (respectively). CPF’s metabolite, CPO had body burden concentrations of 0.53 ± 0.12 μg/g for water and 0.29 ± 0.06 μg/g for soil exposures. TFS_water and TFS_soil body burden concentrations were 0.97 ± 0.08 μg/g and 0.12 ± 0.02 μg/g (respectively). While TFSa, the metabolite of TFS had a body burden concentration of 0.03 ± 0.01 μg/g in soil and 0.06 ± 0.01 μg/g in water. However, the metabolite 4-OH BIF was not detected in any samples and glyphosate was not analyzed. The pesticide determined at the highest concentration in frogs was CPF in water exposures while the lowest pesticide body burdens were calculated for BIF and TFS in the contaminated soil exposures. Interestingly, the water solubility for CPF is greatest and the KOC for BIF had the highest reported values of the studied pesticides (Table 1).

Figure 1.

Boxplot of active ingredients and their corresponding metabolite after 8 h of exposure in pesticide contaminated water or soil A) media concentrations (μg/g for soil and μg/mL for water); B) body burden concentrations (μg/g). Pesticides (bifenthrin, 4-hydroxy bifenthrin, chlorpyrifos, chlorpyrifos oxon, trifloxystrobin, and trifloxystrobin acid) that were analyzed.

3.2. Amphibian Metabolomics

Student’s t-test filtered chromatograms were created to visualize differences in metabolites across each exposure group when compared to controls (Fig. 2A & B). These chromatograms exhibit a higher number of (absolute) metabolites in water-exposed amphibians compared to soil-exposed amphibians. In addition to there being a markedly higher number of spectral features statistically different in the water-exposed amphibians these same pesticides (BIF, GLY, and TFS) also expressed a higher sum for absolute abundance in water-exposed amphibians compared to contaminated soil-exposed amphibians (Fig. 3A & B). Chlorpyrifos was the only pesticide that had a lower number of perturbed spectral features and total abundance in water than soil-exposed amphibians. This same phenomenon was also observed when t-test filtered chromatograms were aligned for soil-vs. water-exposed amphibians for each pesticide (Fig. 4).

Figure 2.

Student’s t-test filtered chromatograms illustrating the difference in metabolites across each exposure group when compared to controls A) soil exposed; B) water exposed). Peaks that are negative represent metabolites that are in higher abundance in non-exposed amphibians. The average response of controls was subtracted from the average response of exposed amphibians for each retention time:m/z pair.

Figure 3.

Plots of the number of significant spectral features (top) and the sums of their absolute abundance (bottom) for A) contaminated soil-exposed; B) and water-exposed amphibians.

Figure 4.

t-Test filtered chromatograms generated by subtracting the average response of soil-exposed frogs from their water-exposed equivalents at each retention time:m/z pair. Negative peaks represent metabolites that are higher in water-exposed amphibians.

To further aid in the identification of metabolites responsible for class separations by pesticide, a VIP score >1.0, derived from PLS-DA models, was considered to represent a significant metabolite. BIF had the most metabolites identified with 60, while for CPF, GLY and TFS there were 51, 54, and 50 metabolites, respectively (Table 2). Overall, out of the 80 significant metabolites that were identified, 29 were present in all four pesticides, with many of them being amino acids indicative of all the potential pathways that could be differently impacted in amphibians. Furthermore, out of these 29 metabolites only six of these metabolites (lysine, N-acetyl-D-glucosamine, phosphate, pyroglutamic acid, threonine, and tryptophan) were observed to be in the same direction for all pesticide exposures.

Table 2.

List of peaks with a VIP score ≥1 from the PLS-DA models constructed across each pesticide exposure (italic and red represent higher abundance in soil; bold and blue represent higher abundance in water).

Metabolite ID	BIF	CPF	GLY	TFS	Metabolite ID	BIF	CPF	GLY	TFS	Metabolite ID	BIF	CPF	GLY	TFS
11-Octadecenoic acid	1.0				Glucose	2.7	3.5	2.5	7.9	Norvaline			2.2
2-Deoxy-D-ribose	1.8				Glutamic acid	1.5	1.1	3.8	1.6	Octadecanoic acid	1.4
2’-Deoxyinosine	1.3	1.1			Glyceric acid	1.6	1.5			Ornithine	1.4	2.8	2.0	6.2
2-Ketoisocaproic acid		1.0			Glycerol	6.8			10.8	Oxalic acid	2.2	1.0		2.5
3-Ketobutyric acid		4.0			Glycerol-2-phosphate			1.0	1.9	Phenylalanine	1.9	1.6	3.4	1.9
4-Hydroxybutyric acid			1.1		Glycerophosphoric acid	1.2	2.1	1.4	2.4	Phosphate	7.9	4.8	9.9	7.1
9,12-Octadecadienoic acid	2.3		3.0	1.3	Glycine	2.9		5.8	1.0	Phosphoethanolamine	1.7		1.3
Adenine	2.8	8.1	6.3	2.3	Glycyl-glutamate	2.0				Phosphoric acid	7.9		3.8	7.4
Adenosine	1.1				Glycyl-glycine		1.2			Proline	3.3			3.6
Alanine	1.0	2.9	2.6		Guanine	1.0		1.3	1.6	Pyroglutamic acid	2.7	1.9	7.4	4.2
Allose	6.4	1.8	2.2	3.8	Guanosine	1.8	2.3	1.3	2.0	Ribose 5-phosphate		1.1
Arabinonic acid	1.1		1.1		Hexadecanoic acid	2.3		1.2		Sarcosine	1.2	3.0		1.5
Arabitol		2.5			Hypoxanthine	2.4	1.6	3.7	1.4	Sedoheptulose				4.5
Aspartic acid	4.6	2.2	6.5	3.8	Inosine	1.5	2.2	3.2	1.5	Serine	2.3	1.9	5.3	2.8
Benzoic acid	1.0				Isoleucine	3.3		2.5	3.8	Sucrose		1.7	2.1	1.1
Cadaverine	2.0	2.1		1.6	Lactic acid	1.1	2.9	7.2	6.0	Tetracosanoic acid		5.9
Chenodeoxycholic acid		2.2			Lactose	1.6	1.7	1.6		Threitol		1.7	1.7
Creatinine	1.3		1.7		Lactulose		1.2	1.4		Threonine	4.5	2.8	4.0	2.8
Cytosine	1.0		3.2		Leucine	2.4	2.0	2.8	1.7	Thymine	1.9	1.5	2.5	1.5
Fructose	2.2	2.8	1.4	1.5	Lysine	7.4	2.5	8.1	6.9	Tryptophan	2.0	1.1	1.9	2.3
GABA	1.9	1.2		2.4	Malic acid		1.2	1.7		Turanose		2.8		2.6
Galactitol			1.1		Maltose	4.6	4.6	7.6	4.5	Tyrosine	5.0	3.6	5.5	5.7
Galactonic acid	2.3			2.4	Mannobiose	1.9	1.5	3.0	4.2	Urea	2.5		1.9	1.5
Galactose	1.6	1.9	2.5	1.7	Mannose	3.1	1.1	4.3	1.2	Valine	2.2	3.6	4.0	4.8
Glucitol	1.6			1.1	Myo-inositol	3.5	3.0	7.1	1.9	Xanthine	1.6		1.4
Gluconic acid	1.6		2.2		N-a-Acetyl-L-lysine		1.4	4.0	1.7	Xylopyranose		1.0	1.2
Gluconic acid phosphate	1.1				N-Acetyl-D-glucosamine	1.4	1.3	1.0	1.2
Gluconolactone				4.8	Norleucine	2.3			2.3	Total perubated (%)	76	62	66	61

In the current study, BIF resulted in an upregulation of various metabolites (cytosine, fructose, galactose, inosine, lactic acid, oxalic acid, alanine, serine, and urea) when amphibians were exposed in water compared to contaminated soil (Table 2). Additionally, all four nucleic acids were observed to have higher abundance when amphibians were exposed to BIF_water than BIF_soil. Amphibians in the CPF_soil had higher abundance of hepatic metabolites such as adenine, glucose, maltose, lactic acid, with a higher abundance of valine, tyrosine, thymine, myo-inositol, aspartic acid, GABA, and leucine in CPF_water. GLY_water amphibians, expressed higher abundance of adenine, inosine, lactic acid, and tryptophan with lower abundances of aspartic acid, cytosine, galactose, lysine, glycine, threonine, and tyrosine compared to GLY_soil (Table 2). TFS_soil expressed higher abundances for aspartic acid, lysine, ornithine, pyroglutamic acid, serine, and threonine. While amphibians in TFS_water had higher abundances of hepatic metabolites for GABA, glucose, glycerol, lactic acid, maltose, proline, and valine.

3.3. Pathway Analysis

Pathway enrichment analysis was conducted across all metabolites identified in both soil and water exposures. Pathways identified as significant are included in Table 3 and Supplementary Fig. 1. Enrichment analysis was then used to compare pathways that were upregulated in each exposure media. For amphibians in BIF_water, 9 pathways were similar to the global analysis, while perturbed metabolites from BIF_soil also identified galactose metabolism and lactose degradation as significant. For CPF_water exposures, 7 pathways were identified in the global analysis, while in the CPF_soil no pathways were statistically significant. In GLY exposures there were more pathways impacted in the soil exposures (n = 6) than the water exposures (n = 4) compared to the global analysis. Only two of these pathways were similar including galactose metabolism and urea cycle. This same phenomenon was noticed for TFS, with more pathways impacted in the global analysis for soil (5) than the water exposures (4). The two similar pathways that were impacted were purine metabolism and ammonia recycling. All of these exposure-specific enrichment pathways are included in Supplemental Fig. 1.

Table 3.

Global pathway enrichment analysis across all metabolites identified in both soil and water exposures.

Metabolite Set	Total	Hits	Expect	P value	Holm P	BIFs	BIFw	CPFs	CPFw	GLYs	GLYw	TFSs	TFSw
Galactose Metabolism	38	11	3.04	9.02E-05	0.0088	x	x		x	x	x		x
Purine Metabolism	74	14	5.93	0.00131	0.127		x		x	x		x	x
Lactose Degradation	9	4	0.721	0.00353	0.339	x	x		x		x		x
Urea Cycle	29	7	2.32	0.00599	0.569		x		x	x	x	x
Glutathione Metabolism	21	5	1.68	0.0213	1		x			x
Ammonia Recycling	32	6	2.56	0.0367	1		x			x		x	x
Glycine and Serine Metabolism	59	9	4.72	0.0394	1		x		x			x
Malate-Aspartate Shuttle	10	3	0.801	0.0394	1				x	x		x
Alanine Metabolism	17	4	1.36	0.0407	1		x		x
Glycerolipid Metabolism	25	5	2	0.0432	1		x				x

4. Discussion

Anurans come in contact with both soil and water contaminants due to their life cycle encompassing both terrestrial and aquatic environments (Bishop et al., 1999). Many amphibian species spend a significant portion of their early life cycle in contact with the aquatic environment. Numerous amphibian species, post-metamorphosis, inhabit terrestrial areas far away from breeding grounds only returning to these areas for reproductive purposes. Interestingly, many of these breeding grounds can be found within agricultural landscapes (Lenhardt et al., 2015). Therefore, amphibians are often likely to be exposed to pesticides as a consequence of direct or indirect agricultural application (Houlahan and Findlay, 2003). Amphibians are known to breed and deposit eggs in agricultural ponds and streams where runoff and rainwater can contaminate the water sources (Glinski et al., 2018c; Houlahan and Findlay, 2003). Due to the permeability of the adult amphibian dermis, pesticides can be readily taken up into their body via contaminated soil (Glinski et al., 2018a; Glinski et al., 2018b; Glinski et al., 2019; Van Meter et al., 2019; Van Meter et al., 2015; Van Meter et al., 2016; Van Meter et al., 2014). During mating, it is not unusual for a frog to be in/near the water while calling for a mate repeatedly across several nights (Collins and Wilbur, 1979). When they find a mate, amplexus can take several hours, so an 8 h exposure, as conducted in the current study, can be considered a realistic time scenario (Wells, 1977; Woodruff, 1976).

4.1. Pesticide Body Burden

This study was conducted to assess if pesticide exposure via contaminated water vs contaminated soil would result in higher body burden concentrations terrestrial phase amphibians. The body burden concentrations for amphibians exposed in BIF_water were observed to be four times higher than in amphibians exposed to contaminated soil. The BIF_soil amphibians had a body burden of 0.11 ± 0.02 μg/g which was similar to the concentrations observed for the 1/10th maximum application of BIF (0.08 μg/g) in a study done by Glinski et al. (2019), in addition to a study by Battaglin et al. (2016) where the highest concentration in amphibian tissues found in the wild was 90.1 μg/kg. Overall, pesticide exposure (for both parents and metabolites) in contaminated water frequently resulted in higher body burdens compared to contaminated soil exposures. This could potentially be due to an increase in surface area being exposed in the water treatments even though the volume was small. Upon physical observation of the amphibians in the water exposures, water levels could have potentially reached their dorsal surface depending on their size. Correlatively, more metabolite perturbations were observed in the metabolome of amphibians exposed to BIF_water and CPF_water.

4.2. Amphibian Metabolomics

Pyrethroids, as a class, are known to inhibit ATPase enzyme production and this mechanism is the driving force behind efforts to understand the increased susceptibility of aquatic organisms (Clark and Matsumura, 1982). In the current study, BIF resulted in an upregulation of several amino acids (alanine, glutamic acid, glycine, isoleucine, proline, tryptophan, and valine) when amphibians were exposed in water compared to contaminated soil (Table 2). Additionally, all four nucleic acids were observed to have higher abundance when amphibians were exposed to BIF_water than BIF_soil. These same amino acids were identified to be perturbed in several other metabolomic studies spanning different taxa. BIF altered metabolites in the livers of mice primarily involved in amino acid metabolism and energy production, where the main metabolites being perturbed were isoleucine, threonine, valine, and glutamine (Miao et al., 2017). Similarly, in Xenopus laevis tadpoles exposed to BIF there was a decrease observed for GABA and glutamic acid in brain tissue when compared to controls which impacted the glutamatergic-GABAergic system (Li et al., 2020). In comparison, in goldfish, NMR-based metabolomics revealed that lambda-cyhalothrin, another synthetic pyrethroid, modified concentrations of metabolites such as alanine, leucine, isoleucine, valine, and myo-inositol (Li et al., 2014). Interestingly, many of these metabolites were identified in brain tissue, the perceived target organ, interfering with the glutamate-glutamine-GABA axis and eliciting oxidative stress ultimately leading to the misutilization of amino acids (Li et al., 2014).

Although the current study examined liver tissue due to it being the primary route of metabolism or the first pass effect for many pesticides that enter the body, many of the same metabolites were identified to be perturbed. Additionally, GABA was observed to be downregulated, while on the other hand, glutamic acid was upregulated in amphibians exposed to BIF_water which could result in the inhibition of glutamate decarboxylase activity (Xu et al., 2015a). Therefore, in amphibians, both amino acid metabolism and energy production pathways would be affected with more of an impact occurring in water-exposed amphibians. Furthermore, the upregulation of both glutamic acid and lactic acid in amphibians via BIF_water would suggest there was an interruption in the glutamate-glutamine-GABA axis but also that amphibian tissues were potentially hypoxic leading to oxidative stress (Glinski et al., 2019).

In the current study, amphibians in the CPF_soil had higher abundance of hepatic metabolites such as glucose and lactic acid, with a higher abundance of alanine in CPF_water (Table 2). In the common carp exposed to CPF, an upregulation of glucose, alanine, and lactate, was observed which supported the authors’ hypothesis that the upregulation of glucose was resulting from increased gluconeogenesis. An impairment in this pathway would ultimately decrease the energy provided to the organism while the increase in both alanine and lactate would ensue due to enhanced muscle contractions (Kokushi et al., 2015). Therefore, in amphibians that were exposed to CPF there would be an increase in gluconeogenesis and a decrease in energy.

Amphibians in the CPF_water had higher abundance for GABA, aspartic acid, glutamic acid, leucine, tyrosine, and valine when compared to their soil-exposed counterparts. In salmon hepatocytes, there was an upregulation of branched chain amino acids (BCAAs) (Olsvik et al., 2015) following exposure to CPF. Similar amino acids were also perturbated in the common carp (Cyprinus carpio) including but not limited to valine, leucine, and isoleucine (Kokushi et al., 2015). Additionally, in rat brains, leucine was noted to be an indicative biomarker of CPF exposure (Xu et al., 2015b). Taken together, the upregulation of BCAAs in amphibians could be indicative of oxidative damage followed by degradation of proteins where the organism was trying to restore energy levels (Gómez-Canela et al., 2017; Olsvik et al., 2015). Furthermore, these amino acids (valine, leucine, isoleucine, and tyrosine) are all part of the neurotransmitter system (Xu et al., 2015b), while Liu et al. (2015) mentioned that aspartate and glutamate are excitatory neurotransmitters. Therefore, in amphibians, the perturbation of these amino acids could imply that CPF impacted neurotransmission along with energy levels and caused oxidative damage resulting in the degradation of proteins. Since the amphibians in CPF_water had higher abundance of these amino acids than amphibians exposed in soil, this suggests the effect could be greater to those amphibians exposed in water.

In GLY_water amphibians, a higher abundance of alanine, glucose, and maltose was observed, with lower abundances of isoleucine, leucine, phenylalanine, and tyrosine compared to GLY_soil (Table 2). The same phenomenon was observed in two species of earthworms, using NMR metabolomics, biomarkers of GLY application included elevated levels of glucose, maltose, and alanine coupled with decreased levels of leucine, phenylalanine, and tyrosine (Rochfort et al., 2009). Li et al. (2016) also observed a decrease in BCAAs, phenylalanine, and tyrosine, with an elevation in glucose and maltose levels. Chronic GLY exposure in goldfish demonstrated that metabolomics combined with correlation network analysis revealed changes in amino acid metabolism, energy metabolism, and oxidative stress (Li et al., 2017). In rat serum, leucine and valine levels were decreased after oral exposure to GLY (de Souza et al., 2017). Mesnage et al. (2017) conducted a two-year, low GLY formulation dose, oral study in rats where they observed higher levels of glycine, aspartate, and several proline derivatives. These perturbations are indicative of an overall increased metabolism specifically in relation to heightened nucleic acid turnover, purine and pyrimidine metabolism, cell proliferation and organ regeneration in amphibians (Mesnage et al., 2017). In concordance, these biochemical pathways were magnified in amphibians that were exposed to GLY_water relative to those in GLY_soil.

In the current study, there was a plethora of sugar isomers (i.e., fructose, galactose, glucose, and maltose) that expressed a higher abundance in TFS_water amphibians which suggests that the energy metabolism in amphibians was impacted including but not limited to glucogenesis. Overall, it should be noted that there was very limited eco-biological information on trifloxystrobin exposures and related strobilurins in any species.

4.3. Pathway Analysis

Based on metabolite enrichment analysis, both galactose metabolism and lactose degradation encompassing perturbations in sugars were observed to be significantly impacted in all of the water exposures (Table 3). Fluxes observed in sugars are known to affect energy metabolism. Whereas the galactose metabolism pathway contains the conversion of galactose to glucose, which is necessary for cellular energy production and is a crucial part in subsequent macromolecule formation (Coelho et al., 2015; Petry and Reichardt, 1998). Also, galactose, glucose, and lactose levels could be elevated to meet the increasing demand for energy, resulting from pesticide-induced biological stress, however, it could also be due to the breakdown of amino acids via gluconeogenesis (Nagato et al., 2016). The galactose metabolism pathway was the most significantly impacted pathway in the enrichment pathway analysis.

Overall, an upregulation was observed in this pathway, suggesting that the energy metabolism was in high demand in amphibians that were exposed to pesticide contaminated water. The increase in lactose degradation pathway (lactose conversion into galactose and glucose via a hydrolytic cleavage with β-galactosidase) could be the result of the amphibians using anaerobic respiration for energy (Kaznadzey et al., 2018; Nagato et al., 2016). Since an increase in lactose was observed under high responses to pesticide exposures it suggests that the amphibians exposed to pesticide treated water are succumbing to these perturbations (Nagato et al., 2016). The galactose metabolism pathway was also observed to be impacted in amphibians exposed to a mixture of herbicides and a mixture of pesticides (Van Meter et al., 2018). This was similar to previous experiments involving various species (i.e., earthworm, zebrafish, mice, rats) where they demonstrated that energy metabolism are impacted by CPF (Baylay et al., 2012; Kokushi et al., 2015; Xu et al., 2015b; Zhao et al., 2016).

Furthermore, upregulation in intermediates involved in the urea cycle and ammonia recycling were observed across numerous pesticide exposures, regardless of route of exposure in the current study (Table 3). The urea cycle is the conversion of toxic ammonia into urea which primarily takes place in the liver (Dimski, 1994). Ammonia enters the cell along with carbon dioxide to form carbamoyl phosphate, along with ornithine progressing through the cycle with aspartate to form fumarate, urea, and ornithine (Dimski, 1994). The main byproduct of cellular metabolism is ammonia, where this compound becomes recycled for the biosynthesis of new amino acids (Dimski, 1994; Spinelli et al., 2017). An upregulation in urea cycle intermediates was also observed as an endpoint of transcriptomic analysis of tadpoles exposed to atrazine (Zaya et al., 2011). Due to high levels of stress in amphibians, a viable source of energy would be protein degradation which leads to an increase in amino acids and subsequently higher ammonium levels indicating an upregulation in both pathways.

The purine metabolism pathway provides the foundation of the synthesis of nucleotides and nucleosides, in addition to providing energy metabolism to promote cell proliferation and survival (Pedley and Benkovic, 2017). This pathway was enriched by annotated metabolites such as adenine, deoxyinosine, guanine, guanosine, hypoxanthine, inosine, phosphate, ribose-5-phosphate, and xanthine, in the current study. The enrichment of this pathway was observed in five different exposure treatments, whereas the upregulation of the purine pathway would suggest that excess energy was required to keep cells alive resulting from toxic insult. This pathway was also significantly impacted in tadpoles (A. americanus and H. versicolor) exposed to the herbicide atrazine (Snyder et al., 2017) and amphibians exposed to a mixture of herbicides (Van Meter et al., 2018).

4.4. Exposure Implications

The experimental tissue residue data demonstrate that the rate of amphibian pesticide uptake is approximately 2–4 times higher from shallow water sources (e.g., puddles, irrigation ditches) compared to soil. Characterizing the variability in exposure concentrations for amphibian rehydration events from different water sources is a known uncertainty (EFSA et al., 2018; Fryday and Thompson, 2012) and this study is a significant contribution to quantifying their relative importance. However, our experimental water:soil uptake ratios will not directly transfer to the same ratio for field doses received for terrestrial-phase amphibians due to water volume and behavioral considerations. These behavioral considerations include relative time period spent in terrestrial versus aquatic environments and the degree of aquatic exposures that occur in shallow, stationary water sources versus larger, flowing water bodies. From a water source perspective, amphibians may frequent in-field puddles, ditches, ponds or proximate streams for breeding or rehydration purposes. Exposures that occur immediately after a spray event in shallow water bodies may be expected to yield the highest exposure concentrations, however, runoff events from rainfall in the days and even weeks post-exposure may continue to mobilize significant exposures at lower concentrations. Some amphibian species also may rehydrate completely from soil moisture sources and therefore may not frequent a water body at all during some periods. The experimental exposures we conducted are most similar to irrigation ditches and/or in-field puddle exposures given the shallow amount of water to which the pesticides were directly applied and are expected to be episodic and in the upper range of field concentration experienced by amphibians.

4.5. Conclusions

The decline of amphibian populations is well documented, and a primary causative factor is the continued and increased use of pesticides across the globe. Most available studies focus on the acute effects of pesticide exposure on larval amphibians with limited data on juvenile and post-metamorphic parameters, and even fewer data in adult frogs.

In the current study, terrestrial phase amphibians were used to determine relative pesticide bioavailability from exposure in contaminated media (soil, water) and the resulting impact on the hepatic metabolome. The experimental design incorporates both dermal absorption scenarios and biomarker elucidation. We conclude that during periods of aquatic exposure during breeding season, pesticide uptake and associated changes in the amphibians’ metabolome are more pronounced for shallow water versus terrestrial exposures for a given exposure event. Understanding the impact of dermal routes of exposure in amphibians is relevant and remains a source of exposure assessment uncertainty in ecological risk assessments.

The data gained from this research will fill regulatory knowledge gaps and better allow amphibians to be more accurately assessed for the pesticide registration process. While amphibians are moving through terrestrial and aquatic habitats, they are continuously exposed to varying pesticide concentrations and bioavailability rates via dermal exposure. This presents complications and significant uncertainties for calculating cumulative exposures within different agroecosystem habitats and across amphibian species with different behavior and life history traits. Our research directly targets these uncertainties and can lead to the creation of more accurate dermal exposure and uptake models in amphibians and therefore inform risk assessment efforts for these and other non-target species.

Highlights.

• Juvenile amphibians were exposed to pesticide contaminated water or soil at 1 μg/mL

• Exposures in water resulted in higher pesticide body burden concentrations

• Water exposures led to a higher abundance in hepatic metabolites being perturbated

• Exposures in water resulted in more biological perturbations across all pesticides