Gonad metabolomics and blood biochemical analysis reveal differences associated with testicular oocytes in wild largemouth bass (Micropterus salmoides)

Abstract

Adverse reproductive effects associated with gonadal intersex among freshwater fish could hold considerable implications for population sustainability. Presence of testicular oocytes (TO) is the most common form of intersex and is widespread among centrarchids (sunfishes) of North America and other freshwater teleosts. Placing TO within the toxicological context of adverse outcome pathways (AOPs) to assess ecological risk is a priority for ecotoxicologists due to the association of TO with harmful chemical exposure and adverse reproductive effects in some cases. However, key event relationships between EDC exposure, incidence of TO, and apical outcomes have yet to be fully elucidated - in part due to a lack of knowledge of relationships between intersex gonad physiology and fish health. Understanding the physiological status of intersex fish is critical to assess ecological risk, understand mechanisms of induction, and to establish biomarkers of intersex in fish. In the present study, features of gonad metabolite profiles associated with TO in largemouth bass (LMB, Micropterus salmoides) from an impoundment in Georgia (USA) were determined using GC-MS-based metabolomics. Clinical blood biochemical screens were used to evaluate markers of fish health associated with TO. Results suggest that physiological changes in energy expenditure as well as relatively ‘feminized’ gonad lipid and protein metabolism may be related to the occurrence of TO in male LMB, and highlight the need to understand relationships between intersex and physical stressors such as elevated temperature and hypoxia. These results provide novel insight to AOPs associated with TO and identify candidate analytes for biomarker discovery.

Graphical Abstract

1. Introduction

Widespread reports of testicular oocytes (TO) among gonochoristic (fixed-sex) freshwater fishes have caused concern for the sustainability of ecologically and economically important fish populations around the world. Characterized by the presence of immature egg cells within the testicular tissue, TO is an intersex condition of male fish commonly considered a morphological abnormality and associated with exposure to estrogenic contaminants. The condition is highly prevalent among particular fish taxa in some locations, and evidence of adverse reproductive effects means that TO has the potential to elicit population-level effects – a major concern for ecotoxicologists and natural resource managers. For example, studies of common roach (Rutilus rutilus) in the United Kingdom have found associations between severe forms of TO and gonadal abnormalities leading to reduction in reproductive success (Jobling et al., 2002; Harris et al., 2011). Similarly, studies of wild rainbow darters (Etheostoma caeruleum) in North America have suggested that severe TO is associated with reduced fertilization success (Fuzzen et al., 2015). In addition, TO was detected in a population of fathead minnows (Pimephales promelas) that experienced reproductive failure following long-term exposure to a synthetic estrogen (Kidd et al., 2007), raising the possibility that TO contributed to the population crash. However, knowledge of intersex-induced reproductive failure is limited for most fish species and populations in which intersex is prevalent. For example, black bass species (Micropterus spp.) are among the most frequently reported taxa exhibiting TO across North America (Abdel-moneim et al., 2015). Severe forms of the condition (i.e., large portions of the testis consisting of vitellogenic oocytes) associated with adverse effects in some species are not known to occur among black bass, and little evidence of TO-induced physiological dysfunction exists in these taxa specifically, but the potential remains a major concern because of the widespread distribution of intersex black bass (Hinck et al., 2009). More information is needed regarding physiological differences related to the occurrence and severity of TO in black bass to assess ecological risk, understand the role of TO in adverse outcomes, and shed light on biochemical processes indicative of causative or parallel mechanisms of action.

Intersex, including TO, can result from exposure to endocrine disrupting chemicals (EDCs), but in many instances, chemical influences are unclear and contradictory evidence indicates the likely involvement of alternative factors. Laboratory studies have confirmed the potential for exposure to EDCs to induce intersex and other abnormalities in laboratory fish species such as Japanese medaka (Oryzias latipes) and zebrafish (Danio rerio), particularly during early life stages (reviewed by Leet et al., 2015). However, a great deal of variability exists in exposure scenarios such that variability in the incidence and magnitude of intersex responses in fish is difficult to predict even under controlled laboratory conditions. Further, TO has also been observed among laboratory control and reference groups (Grim et al., 2007), suggesting that intersex may sometimes develop independently of chemical factors. In wild fish, TO appears to be influenced by multiple factors, including species, timing and duration of exposure, chemical concentration and potency, and co-exposure to compounds that agonize or antagonize particular biochemical pathways. However, alternative non-chemical factors are suspected of involvement in many cases because associations with anthropogenic land use or specific contaminants are often inconclusive or inconsistent (Bahamonde et al., 2013). For example, TO appear to occur in black bass species even in the absence of obvious EDC exposure. Intersex black bass have been observed in relatively pristine areas with little evidence of chemical exposure (Iwanowicz et al., 2016; Kadlec et al., 2017), and a recent study found TO in smallmouth bass collected as far back as the 1880s (Grim et al., 2020), far before the extensive and widespread use of synthetic compounds. These findings suggest that intersex occurs naturally to some extent in certain species and that causes of intersex are more complex than currently understood.

An important aspect to understanding any potentially harmful biological condition or disease state relates to the biochemical mechanisms underlying its incidence. Identifying biochemical changes within the affected tissue may yield important information relevant to molecular initiating events, as well as associated adverse physiological states. Advanced analytical techniques, such as metabolomics, used to evaluate differences in small-molecule biochemical components of tissues, have provided important information about complex physiological conditions such as cancer and diabetes. Measuring levels of endogenous metabolites, collectively referred to as the metabolome, can lead to the discovery of biomarkers and yield important information regarding pathogenesis and molecular physiology. In a toxicological setting, metabolomics is commonly used to uncover mechanisms of toxicity to better understand how effects of chemical exposure are produced on a biochemical level. Investigation of the molecular physiology within the intersex gonad could provide insight into biochemical mechanisms of intersex induction and associated physiological effects. Metabolomics offers a unique opportunity to better understand the physiological state underlying intersex in fish whether TO is considered a pathological condition, a direct effect of chemical exposure, or a natural phenomenon. Confirmation of TO requires preservation and extensive histological evaluation of the majority of the gonadal tissue, limiting parallel investigations of the gonad to minimally destructive sampling methods. Metabolomics analysis can be conducted using very small (~50 mg) tissue samples to measure hundreds of biochemicals relevant to gonad physiology, while retaining adequate tissue for histological analysis.

In the present study, we evaluated the gonad metabolite profiles of wild LMB to reveal information relevant to adverse outcome pathways (AOPs) associated with TO. In addition, blood plasma biochemical profiling was employed to better understand organismal responses associated with TO and to evaluate its utility for classification of individuals with different levels of TO. Methods used in this study could be used to investigate intersex and other gonadal abnormalities in other species, and to compare incidences of intersex occurring under different circumstances to better explain variability observed in the field and in laboratory studies.

2. Methods

2.1. Fish collection

Adult LMB were collected by boat electrofishing from a single stream-fed impoundment (~20 ac.) in Wilkes Co., Georgia, USA during March 2015. This site was selected based on a previous report of a high prevalence (80%) of TO in male LMB, and no direct municipal or agricultural wastewater inputs are known or suspected to exist (Kellock et al., 2014). Fish targeted for the current experiment ranged from 200 to 375 mm total length and fish outside this range were released immediately. Although intersex male fish were the primary subject of this study, females were also collected for reference. Upon collection, a numbered zip-tie was placed around the mandible of each fish. Fish (no more than 5 at a time) were held alive in an aerated on-boat holding tank for ≤15 min prior to processing in the order in which they were collected. Each fish was euthanized with an overdose of neutral-buffered tricaine methanesulfonate solution (MS-222, 200 mg/L). Body weight (nearest 0.1 g) and total length (TL, nearest mm) were recorded for each fish, and relative weight (Wr) was calculated according to the Blackwell et al. (2000) method for LMB. Water quality variables (pH, DO, and temperature) measured during sampling (Hach HQ40D, USA) were 7.11, 9.97 mg/L, and 15.2 °C, respectively. All fish were handled in accordance with approved University of Georgia Institutional Animal Care and Use protocols (AUP # A2016 06–030-Y3-A3).

2.2. Tissue collection

After recording morphometrics, blood and gonadal tissue samples from each fish were immediately collected in the field. Blood was collected ventrally from the caudal vessels in heparinized syringes and expelled into heparinized 2 mL centrifuge tubes. Blood samples were kept on ice, transported to the University of Georgia Aquatic Toxicology Laboratory (Athens, Georgia, USA) for centrifugation (10 min. at 1500 xg, 4 °C), and two equal aliquots of plasma from each sample were transferred into separate tubes and frozen prior to biochemical profiling and vitellogenin analysis. Gonads from each male fish were weighed (nearest 0.001 g) to facilitate calculation of gonadosomatic index (GSI = [gonad weight (g) ÷ body weight (g)] × 100). Approximately 50 mg of gonadal tissue was excised from the left gonadal lobe of each fish for metabolomics. Gonad metabolomics samples were taken from a well-vascularized, medial region near the hilum, where TO might be expected to occur. Samples were placed in a chilled 2-mL cryogenic tube and immediately flash frozen in liquid nitrogen to prevent degradation of metabolites. Dissection tools and surfaces were wiped, rinsed thoroughly with methanol then de-ionized water, and dried between each fish to prevent cross-contamination. The remaining testicular tissue from all male fish was preserved in 10% neutral-buffered formalin solution for histopathological processing and evaluation. In the laboratory, metabolomics samples were stored at −80 °C until extraction.

2.3. Gonad histology

Testicular tissues preserved in formalin (≥96 h in 10% neutral-buffered formalin) were prepared for histology and evaluated for the presence of TO at the Aquatic Toxicology Laboratory at the University of Maryland (College Park, MD, USA) with the modified methods outlined by MacLeod et al. (2017). Briefly, preserved testes were dehydrated in alcohol, embedded in paraffin wax, sectioned at 6-mm, mounted on glass slides, and stained with hematoxylin and eosin (Presnell et al., 1997). For each male fish, three longitudinal step-sections were cut from the right lobe of the testis, and 5 transverse sections were cut from the left lobe, avoiding areas biopsied for use in another study. Slides were examined by light microscopy (Olympus BH2) for TO. Oocytes were enumerated for each slide, and the total number of TO detected per individual (TO count) was calculated as the sum of oocytes detected in both testis lobes. For statistical comparisons, male fish were placed into one of two groups based on TO count to determine differences between individuals with varying degrees of intersex: (1) individuals with TO count ≤20 were identified as ‘low TO,’ and (2) individuals with TO count >20 were identified as ‘high TO.’

2.4. Blood plasma biochemical profiling

Blood plasma samples were thawed and kept on ice prior to biochemical profiling. Samples were evaluated on a Hitachi® 912 Blood Chemistry Analyzer (Roche Diagnostics, Indianapolis, IN, USA) at the Clinical Pathology Laboratory at the University of Georgia College of Veterinary Medicine for the following 13 bio-chemical variables: alanine aminotransferase (ALT) activity, aspartate aminotransferase (AST) activity, alkaline phosphatase (ALP) activity, bicarbonate, calcium, phosphorous, magnesium, total protein, sodium, potassium, chloride, amylase, and glucose.

Plasma vitellogenin (VTG) was quantified by direct enzyme-linked immunosorbent assay (ELISA) using a striped bass monoclonal antibody (ND-3G2; Cayman Chemical, Ann Arbor, MI) with known sensitivity to largemouth bass VTG. The VTG standard used for the assays was purified from induced male largemouth bass at the Center for Human and Environmental Toxicology, University of Florida, Gainesville, FL. Briefly, samples (in duplicate) and standards (in triplicate) were diluted in phosphate buffered saline azide (PBSZ) and loaded into 96-well high-binding microplates (Costar, Corning Inc., Corning, NY), where they were held overnight at 4 °C in a humidified chamber. Plates were then washed 5 times with tris-buffered saline and Tween20 (TBST) before introduction of antibody diluted in blocking buffer, followed by incubation for 1 h at room temperature. Plates were washed another 5 times with TBST before a biotinylated secondary antibody (Biotin-SP-conjugated Affinipure Goat-anti-mouse IgG; Jackson ImmunoResearch, West Grove, PA) was added to each well. After incubation for 1 h at room temperature, plates were again washed 5 times in TBST and streptavidin-alkaline phosphatase diluted in blocking buffer (Invitrogen, Carlsbad, CA) was added to each well, and plates were again incubated for 1 h at room temperature. After a final wash (5× TBST), 4-nitrophenyl phosphate (Sigma-Aldrich) was added to each well and colour was allowed to develop for 30–45 min. Plates were read at 405 nm on a SpectraMax M2 microplate reader (Molecular Devices, Inc., Sunnyvale, CA), and concentrations of unknowns were estimated based on eight-point standard curves using SoftMax Pro ver. 5.4.1 software (Molecular Devices). Iterative runs during method validation indicated a limit of detection of ≤8.0 μg/mL.

2.5. Gonad extraction/derivatization

Metabolomics extractions and analyses were performed at the US EPA Center for Environmental Measurement and Modeling (CEMM, Athens, GA, USA). Extraction procedures followed those detailed by Viant (2007). Briefly, frozen gonad samples were randomized prior to extraction and homogenized with a tissuelyzer in a bi-phase solvent mixture of methanol/water (polar phase) and chloroform (non-polar phase). Samples were then centrifuged to separate phases, and final volumes of 500 μL (polar phase) and 210 μL (nonpolar phase) were decanted separately into amber glass GC vials. The polar phase used in the current study was evaporated under vacuum, and dried extracts were stored at (−80 °C) until derivatization. After thawing, extracts were derivatized with 100 μL methoxyamine hydrochloride solution (20 mg/mL in pyridine) and placed in a 60 °C oven for 150 min. Samples were vortexed for 1 min at 30 min intervals during incubation. After cooling, 100 μL BSTFA (N,O-bistrifluoroacetamide) with 10% TCMS (methyltrichlorosilane) was added to each sample. Samples were then placed in the oven again for 90 min and vortexed at 30 min intervals. Derivatized samples were then cooled to room temperature and transferred to micro-target inserts for GC/MS analysis.

2.6. GC–MS Metabolomics

Gonad extracts were analyzed on an Agilent® 7890 gas chromatograph (Agilent Technologies, CA, USA) linked to an Agilent 5975c mass spectrometer, and all data were collected and then processed using Chemstation®. Separation of metabolite derivatives was achieved with an Rxi-5Sil MS (30 m, 0.25 μm thickness, and 0.25 mm ID; Restek, PA, USA). All injections (2 μL) were made in the splitless mode. Injector temperature was set to 250 °C, transfer line temperature was held constant at 280 °C, source temperature was set to 200 °C, and trap and detector were set for 350 μA and 300 V, respectively. The carrier gas (helium) was maintained at a constant pressure of 65.20 kPa. The initial oven temperature (60 °C) was held for 2 min and then increased at 6 °C/min to 280 °C with a hold time of 3 min. Instrument blanks were run at the beginning and intermittently throughout the sequence to verify no instrument carry over, and 13 extraction method blanks were randomly included to verify no cross contamination. Mass spectra were acquired over a mass range of 25 to 800 m/z. Chromatograms were exported as netcdf files and imported into MetAlign (v. 041012) for data preprocessing and alignment. Distributor recommended parameters for fast scan analysis was used (Lommen, 2009). Following alignment, data were filtered and truncated in Excel as described in Niu et al. (2014).

2.7. Statistics (morphometrics)

Differences in mean morphometrics (total length, body weight, Wr, and GSI) were assessed with non-parametric procedures due to unequal sample sizes among sexual phenotype groups. Morphometric statistical analyses were conducted in R (http://www.r-project.org/, v. 3.4.2), and significant difference was assessed at α = 0.05 for all tests. Differences in total length, weight and Wr among all three groups (female, low TO, and high TO) were assessed using a Kruskal-Wallis test. Mann-Whitney U tests were performed to evaluate differences in mean total length, body weight, Wr, and GSI for all pair-wise comparisons (i.e., female vs. low TO, female vs. high TO, and low TO vs. high TO).

2.8. Statistics (blood plasma biochemistry)

Statistical analyses of blood plasma biochemistry data were performed in R. To evaluate mean differences among all three sexual phenotypes, a Kruskal-Wallis test was conducted for each blood plasma chemistry variable measured, and false-discovery rate (FDR) adjustments were applied to p-values (FDR-adj. p-value) to reduce the possibility of type I error. Additionally, receiver operating characteristic (ROC) curves were generated using the Biomarker Analysis module in Metaboanalyst 4.0 (Chong et al., 2018) to identify plasma chemistry variables associated with higher TO count and to evaluate diagnostic accuracy for classifying male fish into high TO and low TO groups. A generalized logarithm transformation was performed prior to analysis, and pairwise ratios of blood chemistry variables were computed for inclusion in ROC analysis. A t-test was used to compare mean values for each variable and variable ratio between high and low TO groups. Significance was assessed at α = 0.05, and area under the ROC curve (AUROC) was used to assess the relative usefulness of variables for discrimination of high and low TO groups, with an AUC value of 1.0 representing a test with perfect diagnostic accuracy (i.e. no false positives or false negatives). Plasma VTG concentrations below detection were calculated as ½ the detection limit for statistical purposes.

2.9. Statistics (metabolomics)

Gonad metabolomics data were analyzed statistically with Metaboanalyst 4.0. Prior to analysis, analyte relative abundance values were normalized by sum for each observation, mean-centered, and pareto scaled. Partial least squares discriminant analysis (PLS-DA) was used to assess overall differences among groups by evaluating spatial separation in a scores plot, and Q2 values were used to evaluate the utility of metabolite profiles for prediction of sexual phenotype. The top 25 variable importance in projection (VIP) scores were identified for relevant components and the relative directional response of gonad phenotype groups was noted for each analyte. A series of t-tests (α = 0.05) and fold-change analyses (fold-change threshold = 2.0) were conducted to evaluate differences between metabolite profiles of low TO and high TO groups. Metabolite annotations were performed for analytes of interest by comparing mass spectra to the NIST Spectral Library (v. 11) and a match quality threshold of ≥0.65.

3. Results

3.1. Fish collection and gonad histology

Thirty-eight adult LMB (16 males; 22 females) were collected during the sampling. Neither total length (p = 0.52), body weight (p = 0.47), nor Wr (p = 0.11) varied by sex. Histological evaluation of preserved testicular tissues revealed an intersex prevalence of 95.5% among males, with TO detected in all males except for one individual. TO count ranged from 0 to 276 per individual, with a total mean (±SD) of 37.7 ± 59.9 oocytes per individual. For individuals with TO count ≤20 (‘low TO’ group, n = 13, mean TO count ± SD = 10.0 ± 7.1) total length ranged from 261 to 325 mm, Wr ranged from 0.764 to 0.960, and GSI ranged from 0.177 to 0.606. For individuals with TO count >20 (‘high TO’ group, n = 9, mean TO count ± SD = 77.8 ± 79.3) total length ranged from 278 to 326 mm, Wr ranged from 0.760 to 0.957, and GSI ranged from 0.370 to 0.786. Mean total length (p = 0.76), body weight (p = 1.0), Wr (p = 0.09), and GSI (p = 0.13) were not significantly different between low TO and high TO groups.

3.2. Blood plasma biochemical profiles

Blood chemistry profiles were variable among individuals, but significant differences among sexual phenotypes were few (Table 1). Plasma VTG and calcium concentrations, significantly higher in females (FDR-adj p < 0.001), were the only blood chemistry analytes differing significantly among groups (Table 1). Plasma magnesium concentration was also generally higher in females, though marginally insignificant (FDR-adj p = 0.051; Table 1). Classical ANOVA revealed no significant differences in blood plasma analytes between low and high TO groups. However, evaluation of ROC curves identified variable ratios useful for classification of high TO and low TO males. The ratio of plasma ALP/AST was significantly different between low and high TO groups (p = 0.041) and also had the highest AUROC (0.778, Fig. 1), and the optimal cutoff for the ratio of ALP/phosphorus discriminated equally well (AUROC = 0.778) between TO groups, although differences were marginally insignificant (p = 0.056, not shown).

Table 1.

Mean (± SD) blood plasma biochemical profiles of female, low TO (≤20 testicular oocytes), and high TO (>20 testicular oocytes) male largemouth bass collected from an impoundment in Georgia (USA). ALP = alkaline phosphatase activity; ALT = alanine aminotransferase activity; AST = aspartate aminotransferase.

		Female	Male - High TO	Male - Low TO
Electrolytes	Sodium (mg/mL)	3.74 ± 0.10	3.72 ± 0.06	3.76 ± 0.08
	Chloride (mg/mL)	3.60 ± 0.16	3.71 ± 0.18	3.68 ± 0.24
	Potassium (μg/mL)	86.5 ± 38.4	83.0 ± 37.9	79.4 ± 32.6
	Bicarbonate (μg/mL)	530.1 ± 96.4	474.6 ± 117.3	450.6 ± 120.9
	Calcium (μg/mL)	165.7** ± 28.3	114.1 ± 11.6	118.8 ± 9.8
	Phosphorus (μg/mL)	117.1 ± 17.6	112.2 ± 13.7	105.6 ± 19.7
	Magnesium (μg/mL)	35.3* ± 3.5	30.9 ± 1.6	32.1 ± 4.1
Biomolecules	Glucose (mg/mL)	1.28 ± 0.55	1.31 ± 0.28	1.37 ± 0.39
	Cholesterol (mg/mL)	6.20 ± 2.04	6.11 ± 1.10	5.90 ± 1.19
	Total Protein (μg/mL)	34.06 ± 3.13	32.11 ± 3.06	33.08 ± 3.59
	Vitellogenin (μg/mL)	1342.33** ± 274.41	27.81 ± 22.22	24.51 ± 9.86
Enzyme Activities	ALP (mU/mL)	22.81 ± 9.25	19.67 ± 7.32	26.92 ± 10.22
	ALT (mU/mL)	8.25 ± 6.91	5.89 ± 5.60	5.54 ± 4.75
	AST (mU/mL)	77.25 ± 51.42	83.44 ± 47.17	72.62 ± 62.19

^*Marginally insigificant (FDR-adj p = 0.051).

^**Highly significant (FDR-adj p < 0.01).

Fig. 1.

Receiver operating characteristic (ROC) curve and 95% confidence band for the ratio of alkaline phosphatase (ALP) and aspartate phosphatase (AST) activity (generalized log transformed) in blood plasma of male largemouth bass with high TO (> 20 testicular oocytes) and low TO (≤ 20 testicular oocytes; A). e.g., Curves passing through the upper left corner have 100% diagnostic accuracy. Boxplot representing differences between high and low TO groups in plasma ALP/AST ratio (B). The red dot on the ROC curve, as well as the red dotted line in the corresponding boxplot represents the optimal cutoff value.

3.3. Metabolomics

Following processing and alignment, a total of 3597 spectral features were identified between retention times 6.5 and 35.0 min (Example chromatogram in Fig. 2). PLS-DA scores were used to aid in visualization of overall differences and in gonad metabolite profiles from females, low TO, and high TO male groups (Fig. 3). Results revealed clear separation of females from males, mainly along the first component. In comparison, differences between high and low TO groups were not as distinct. Separation between high and low TO groups in this model was incomplete, reflecting relatively similar gonad metabolite profiles between male groups. Interestingly, profiles of high TO males on average trended toward females along the first component. The top 25 variable importance in projection (VIP) scores for Component 1 indicated the relative importance of spectral features of interest for separation along the first component, as well as the directional response (elevated [↑], medial [•], or reduced [↓]) of each sexual phenotype for a given spectral feature (Table 2). Putative annotations included organic alcohols, amino acids, carbohydrates, carboxylic acids, and phosphate. Glycerol (VIP = 4.45) responses were elevated in females, medial in high TO males, and reduced in low TO males, as were amino acid analogs dimethyl glycine (VIP = 3.94) and aminobutyric acid (VIP = 4.33). Glutamine (VIP = 4.03) responses were elevated in high TO males, medial in low TO males, and reduced in females. However, most amino acids and analogs, including those with the highest VIP scores were reduced in females, medial in high TO males, and elevated in low TO males. These included pyroglutamic acid (VIP = 8.58), glycine (VIP = 5.12), alanine (VIP = 4.49), aminobutyric acid (VIP = 4.33), glutamine (VIP = 4.03), and dimethyl glycine (VIP = 3.94). The carbohydrates methyl glucopyranoside (VIP = 5.27) and an undetermined disaccharide (i.e., disaccharide 1; VIP = 4.79) were elevated in females, medial in low TO males, and reduced in high TO males. Responses of carboxylic acids were mixed, with succinic acid (VIP = 4.00) reduced in females and elevated in high TO males, and responses of two different spectral features of lactic acid (VIP = 3.82–5.22) mixed among groups. Of note, five spectral features putatively annotated as phosphate (VIP scores ranging from 4.02 to 18.22) held great importance for separation along Component 1, but directional responses varied among the five features. However, levels of phosphate, including the feature most important for separation along Component 1 (VIP = 18.22), were generally elevated in low TO males, medial in females, and reduced in high TO males. In addition, 7 spectral features for which annotations could not be assigned were also important for separation along PLS-DA Component 1.

Fig. 2.

Example gonad metabolomics total ion chromatogram from largemouth bass collected from impoundment in Northeast Georgia (USA) aligned and processed, along with select analyte annotations.

Fig. 3.

PLS-DA scores plot representing differences in individual gonad metabolite profiles of female (red) and male largemouth bass with varying levels of testicular oocytes (‘Low TO’ - males with low testicular oocyte count [≤ 20, blue]; ‘High TO’- males with high testicular oocyte count [> 20, green]) from a Georgia (USA) impoundment. Each point represents an individual fish, and distance between points represents overall differences in gonad metabolome.

Table 2.

Chemical class, putative annotation, and retention time for the top 25 PLS-DA variable importance in projection (VIP) scores for spectral features of Component 1, representing relative differences in gonad metabolite profiles of largemouth bass from a Georgia impoundment (Low TO, testicular oocyte count ≤20; High TO, testicular oocyte count >20). Columns on the right denote relatively elevated (↑), reduced (↓), and medial (•) responses among sexual phenotypes. Duplicate annotations are spectral bins representing the same putative metabolite.

^*Undetermined disaccharide.

^**Undetermined nucleoside phosphate.

T-tests and accompanying fold-change (FC) of testicular analytes indicated 45 spectral features with mean values varying significantly (p < 0.05 and FC > 2) between low TO and high TO males (Table 3). Putative annotations represented similar chemical classes to those important for PLS-DA separation of all groups, and four putative metabolites (glutamine, phosphate, lactic acid, and an undetermined nucleoside phosphate [i.e., “AXP”]) were found in common between both analyses (Table 3). Polyols (myoinositol, inositol phosphate, and myoinositol phosphate) were elevated in high TO males compared to low TO males (Table 3). Amino acids and analogs (glutamine, creatinine, amino isobutyric acid, valine, aspartic acid, glutamic acid, and pyroglutamic acid) were mostly elevated in high TO males, except for glutamine and creatinine which had the opposite trend (Table 3). Carbohydrates (undetermined monosaccharide phosphates 1, 2, 3, and 4, methyl-N-acetyl-glucosamide, and ribose 5-phosphate), carboxylic and carbamic acids (malonic acid, carbamic acid, and lactic acid), and hydroxy acids (malic acid) were generally elevated in high TO males, with the exception of monosaccharide phosphate 1, which was reduced in high TO males (Table 3). Nucleosides and derivatives were also generally elevated in high TO males, with exception of one undetermined nucleoside phosphate (AXP 4; Table 3). Phosphate and phosphate esters (O-phosphoethanolamine) exhibited the some of the highest fold-change values of any chemical class. Members of this class generally were elevated in high TO males, with considerable exception of three phosphate features that were greatly reduced (up to 2000-fold) in high TO males (Table 3). In addition, two features without reliable annotations were elevated in high TO males.

Table 3.

Chemical class, annotation, retention time, t-test p-value, and fold change (FC) values for gonad metabolite profile features differing significantly (p< 0.05, FC > 2.0) between low TO (≤20) and high TO (>20) male largemouth bass from a Georgia (USA) impoundment. Features are ordered by lowest p-value within each chemical class. On the right, log2 Fold change values are displayed for comparison (green, higher in High TO group; blue, higher in Low TO group). Duplicate annotations are different spectral bins representing the same putative metabolite.

^*Undetermined monosaccharide phosphate.

^**Undetermined nucleoside phosphate.

4. Discussion

In this study, we detected and putatively annotated features of interest in gonad metabolite profiles and evaluated differences in blood plasma biochemical profiles associated with increased number of TO in largemouth bass. To our knowledge, this is the first study to demonstrate the utility of metabolomics to identify biochemical differences in gonadal tissues of wild intersex fish. We putatively annotated several testicular analytes that are indicative of altered biochemical pathways in intersex testes and identified plasma biochemicals that indicate the physiological state associated with intersex fish. These results provide novel insight relevant to AOPs to elucidate cellular pathways associated with the intersex condition in male LMB and to identify candidate analytes for biomarker discovery.

Although differences in blood plasma biochemistry data between high TO and low TO groups of male fish were not significant (FDR-adj p > 0.05), results of ROC analyses suggest that ratios of plasma chemistry variables may be useful for discrimination between high and low TO males. The ratio of ALP/AST yielded a mean sensitivity (true positive rate) of 0.8 and 1-specificity (false positive rate) of 0.9, for an AUROC = 0.778 (Fig. 1A), indicating moderate accuracy in discriminating between high and low TO individuals (an AUROC of 1.0 with an ROC curve passing through the top left corner of the graph indicates 100% accuracy). This finding suggests that an elevated presence of TO in male LMB is associated with a decrease in plasma ALP and a corresponding increase in plasma AST activity. AST can be indicative of increased amino acid catabolism and energy expenditure in fish exposed to higher temperatures (Savoie et al., 2008) and is an established marker of hepatic ischemia-reperfusion injury, a form of liver damage related to hypoxia (Camargo et al., 1997). In rat cells exposed to hypoxic conditions, ALP showed a significant decrease in activity (Utting et al., 2006). In addition, Atlantic cod (Gadus morhua) ALP activity reportedly decreases with increasing temperature (Ásgeirsson et al., 1995). These results suggest that the ratio of plasma ALP/AST in intersex LMB may be indicative of physiological adversities and could serve as a candidate for biomarker discovery. The discovery of blood plasma biomarkers for TO would be a significant finding, considering current methods for determination of intersex are lethal, but the small amounts of blood needed for sensitive analytical instruments could be collected non-lethally. Repeating this work with larger sample sizes of LMB in other populations would be useful to validate these findings, and analysis of nonpolar metabolite profiles may yield additional information. In addition, interspecific comparisons may shed light on taxonomic differences in mechanisms of intersex induction, susceptibility, and related effects.

Metabolomics analysis revealed features of gonad metabolite profiles associated with sexual phenotype and the intersex condition. Separation of sexual phenotype groups in the PLS-DA model was achieved primarily between females from male groups (Fig. 3). However, on average, high TO males appeared to trend closer to females than did low TO males, suggesting that ‘feminized’ features of the gonad metabolome of high TO males may be indicative of increased presence of oocytes within the gonads. To identify these commonalities associated with oocytes, analyte directional response patterns (Table 2) increasing or decreasing in an order consistent with the level of oocytes in sexual phenotype groups (i.e. a medial [•] response of high TO individuals) serves to highlight putative metabolites of interest. Glycerol, which was elevated in females and reduced in low TO males is indicative of lipid metabolism and has been shown to be released by adipocytes following exposure to estrogen (Palin et al., 2003). Of note, the putative metabolites glycine, alanine, and pyroglutamic acid were lowest in females and highest in low TO males. These are important components of the fish egg proteins vitellogenin (Reading et al., 2009) and vitelline (Darie et al., 2004). A negative relationship between the relative abundance of these analytes with the number of oocytes in the gonads could reflect increased utilization of free amino acids for egg protein synthesis, although plasma VTG levels did not significantly differ between low and high TO groups. Previous studies have reported increased levels of alanine in fish and fish cells exposed to estrogenic EDCs (Davis et al., 2016; Ekman et al., 2008; Teng et al., 2013). Hence, the lack of differences in plasma VTG and negative relationship between the level of TO and levels of testicular alanine observed in the present study suggest that intersex may be related to pathways other than those activated by the estrogen receptor.

Pairwise analyses of low TO and high TO males suggest that features of the gonad metabolome associated with increased presence of oocytes in the testes (Table 3) may reflect differences in energy metabolism, protein synthesis, and phospholipid production between groups. Levels of amino acids were generally elevated in the gonads of high TO males (with the exception of glutamine and creatinine, which were reduced), which likely reflects differences in protein and peptide biosynthesis. Glutamine, glutamic acid, aspartic acid, and pyroglutamic acid were all significantly different between high TO and low TO groups. These amino acids are primary substituents of gonadins, a class of peptides with demonstrated ability to alter steroid hormone levels in male rats (Ruiz-Alcaraz and del Rio-Garcia, 2005). Altered levels of these amino acids in intersex LMB could be related to differences in gonadin metabolism. Generally, increased levels of phosphorylated carbohydrates in testes of high TO males may indicate differences in glycogenolysis or pentose phosphate pathways. Levels of ribose-5 phosphate, elevated in testes of high TO males, is a product of the pentose phosphate pathway (PPP). PPP plays an important role for oocyte development in other species, including protection of spermatogenic tissue from oxidative stress (Williams and Ford, 2004). In addition, PPP also plays an important role in lipogenesis for oocyte development. For example, female mice exposed to an androgen experienced reduced oocyte lipid content through inhibition of PPP (Jimenez et al., 2013). Thus, activation of PPP in intersex fish could be associated with oxidative stress and/or the production of lipids for oocyte development. Spectral features representing phosphate were not all in agreement, so false annotation of some peaks is possible due to errors in alignment and/or unresolved analytes. However, two features putatively annotated as phosphate were highly reduced in high TO males and showed the largest magnitude of change (131.7- and 2072.3-fold) of any feature measured. Phosphorylated sugars and alcohols, nucleoside phosphates, and O-phosphoethanolamine were elevated in high TO males, which may suggest increased incorporation of phosphates for energy storage and synthesis or modification of more complex biomolecules such as nucleic acids and lipids.

Overall, results of this study suggest that higher levels of TO in LMB seem to be associated with physiological stress. Chemical contaminants were not measured in this study, so exposure to unknown chemicals may have played a role in the occurrence of TO. However, this study site is not known to have a history of chemical contamination, and toxicological factors are not suspected as a cause of intersex in this case. In small, un-aerated impoundments of the Southeastern United States such as the one sampled in the current study, elevated temperatures and hypoxic conditions are fairly common during warmer periods, which can contribute to physiological stress (Esch and Hazen, 1980), particularly when combined with stresses associated with spawning. In a survey of LMB from impoundments of Georgia (USA), including the one sampled in the present study, Bringolf et al. (2015) found TO to be closely associated with decreased impoundment surface area, suggesting that higher nutrient levels, higher temperatures and/or lower levels of dissolved oxygen typical of smaller impoundments may play a role. Glycogenolysis and glycolysis can be induced by a variety of physiological stressors in fish, including thermal and hypoxic stress (Mukundan et al., 1986). For example, environmental hypoxia induces glycogenolysis (Wright et al., 1989), as well as glycolysis pathways, transgenerational testicular abnormalities, and epigenetic effects in fish (Wang et al., 2016). Therefore, similar mechanisms may be involved with the occurrence of intersex in fish. Signs of elevated glycogenolytic and glycolytic activity associated with intersex in the present study, such as elevated levels of testicular sugar-phosphates, lactic acid, and other carboxylic acids, in addition to reduced blood plasma ALP/AST ratio may indicate the involvement of hypoxic and/or thermal stress. Because all fish were collected from the same impoundment, differences observed within the sampled population could be attributed to individual-level variability in sensitivity to thermal and/or hypoxic stress or perhaps variation in habitat selection within different thermal strata of the impoundment.

The potential involvement of temperature (and/or the associated relationship with dissolved oxygen) in the induction of TO would not be surprising, considering evidence that temperature may alter sex steroid production in the gonads (Kime and Hyder, 1983; Kime and Manning, 1986; Manning and Kime, 1985) as well as sex determination and differentiation (Conover and Kynard, 1981; Nomura et al., 1998; Römer and Beisenherz, 1996; Strussmann et al., 1996a, Strussmann et al., 1996b) in various species. LMB are extremely adaptable to different environments and exhibit a relatively wide range of thermal tolerance (Mulhollem et al., 2015). However, variability is likely to exist among individuals and populations, which may in part be related to intra- and inter-population variability of intersex prevalence and severity among LMB. Furthermore, interspecies differences in thermal tolerance may help to explain relatively high levels of intersex reported in closely related species (e.g., smallmouth bass, M. dolomieu). Centrarchids, which predominate reports of intersex in North America, nest in shallow areas with slow waterflow during in the spring/summer (Bain et al., 1988; Clark et al., 2008), predisposing them to higher temperatures, particularly within small, confined bodies of water. If higher temperatures are related to the induction of intersex, this nesting behavior may partially explain why TO is so common in centrarchids.

In this study, we identified blood plasma biochemistry variables related to fish health, as well as features of the gonad metabolome associated with increased presence of TO in male LMB. Based on these results, the occurrence of TO in this population of LMB appears to be associated with differences in energy expenditure, gonad lipid and protein metabolism, and thermal/hypoxic stress. This information contributes to our understanding of the physiological state associated with TO in LMB and provides insight into biochemical pathways altered in intersex fish. Methods used in the present study could be applied to other species and populations of interest to investigate mechanisms of aberrant gonad morphologies and associated effects on reproductive health. For example, metabolomics analysis of gonadal tissues may have utility to investigate linkages between exposure to EDCs, incidences of intersex, and adverse reproductive effects in fish and other taxa known to exhibit intersex. Further, indications of thermal and hypoxic stress in individuals with elevated TO counts highlight the need to understand the relationships between intersex and non-chemical factors, such as elevated water temperature, hypoxia, and associated measures of fish health. Whether temperature and/or hypoxia are directly or indirectly connected with the incidence intersex in the present work is unknown, but understanding the relationship between non-chemical stressors and the development of intersex could aid in explaining extraneous variability in intersex occurrence observed among wild fish populations. Further, better understanding the role of TO in the toxicological context of AOPs will inform efforts to assess the ecological risk associated with intersex. If elevated water temperature is directly involved in the induction of intersex conditions in fish, it could exacerbate existing feminizing effects of EDC exposure, but could also support suspicions that intersex may occur independently of chemical exposure. Moreover, direct involvement of water temperature and/or hypoxia in the occurrence of intersex would hold implications related to rising global temperatures due to climate change. This work contributes to a growing body of evidence suggesting that global climate change may exacerbate effects of EDC exposure (Brown et al., 2015; Hooper et al., 2013; Keller et al., 2015), which could adversely affect the health of global fish populations.