Dietary selenium disrupts hepatic triglyceride stores and transcriptional networks associated with growth and Notch signaling in juvenile rainbow trout

Aquatic Toxicology 180 (2016) 103–114

Contents lists available at ScienceDirect

Aquatic Toxicology journal homepage: www.elsevier.com/locate/aquatox

Dietary selenium disrupts hepatic triglyceride stores and transcriptional networks associated with growth and Notch signaling in juvenile rainbow trout Rosalinda Knight a , Vicki L. Marlatt b , Josh A. Baker c , Bonnie P. Lo c , Adrian M.H. deBruyn d , James R. Elphick c , Christopher J. Martyniuk a,∗,1 a

Canadian Rivers Institute and Department of Biology, University of New Brunswick, Saint John, New Brunswick, Canada Department of Biological Sciences, Simon Fraser University, 8888 University Drive, Burnaby, British Columbia, Canada c Nautilus Environmental Company Inc., 8664 Commerce Court, Burnaby, British Columbia V5A 4N71, Canada d Golder Associates Ltd., 2920 Virtual Way, Vancouver, British Columbia, Canada b

a r t i c l e

i n f o

Article history: Received 3 July 2016 Received in revised form 19 September 2016 Accepted 21 September 2016 Available online 22 September 2016 Keywords: Dietary Growth Transcriptomics Metabolites Lipid peroxidation Oxidative stress

a b s t r a c t Dietary Se has been shown to adversely affect adult fish by altering growth rates and metabolism. To determine the underlying mechanisms associated with these observations, we measured biochemical and transcriptomic endpoints in rainbow trout following dietary Se exposures. Treatment groups of juvenile rainbow trout were fed either control Lumbriculus variegatus worms or worms cultured on selenized yeast. Selenized yeast was cultured at four nominal doses of 5, 10, 20 or 40 mg/kg Se dry weight (measured dose in the worms of 7.1, 10.7, 19.5, and 31.8 mg/kg Se dw respectively) and fish were fed for 60 days. At 60 d, hepatic triglycerides, glycogen, total glutathione, 8-isoprostane and the transcriptome response in the liver (n = 8/group) were measured. Fish fed the nominal dose of 20 and 40 mg/kg Se dry weight had lower body weight and a shorter length, as well as lower triglyceride in the liver compared to controls. Evidence was lacking for an oxidative stress response and there was no change in total glutathione, 8-isoprostane levels, nor relative mRNA levels for glutathione peroxidase isoforms among groups. Microarray analysis revealed that molecular networks for long-chain fatty acid transport, lipid transport, and low density lipid oxidation were increased in the liver of fish fed 40 mg/kg, and this is hypothesized to be associated with the lower triglyceride levels in these fish. In addition, up-regulated gene networks in the liver of 40 mg/kg Se treated fish included epidermal growth factor receptor signaling, growth hormone receptor, and insulin growth factor receptor 1 signaling pathways. These molecular changes are hypothesized to be compensatory and related to impaired growth. A gene network related to Notch signaling, which is involved in cell–cell communication and gene transcription regulation, was also increased in the liver following dietary treatments with both 20 and 40 mg/kg Se. Transcriptomic data support the hypothesis that dietary Se increases the expression of networks for growth-related signaling cascades in addition to those related to fatty acid synthesis and metabolism. We propose that the disruption of metabolites related to triglyceride processing and storage, as well as gene networks for epidermal growth factor and Notch signaling in the liver, represent key molecular initiating events for adverse outcomes related to growth and Se toxicity in fish. © 2016 Elsevier B.V. All rights reserved.

1. Introduction

∗ Corresponding author. E-mail address: cmartyn@ufl.edu (C.J. Martyniuk). 1 Current address: Department of Physiological Sciences and Center for Environmental and Human Toxicology, University of Florida Genetics Institute, College of Veterinary Medicine, University of Florida, Gainesville, FL, 32611, USA. http://dx.doi.org/10.1016/j.aquatox.2016.09.014 0166-445X/© 2016 Elsevier B.V. All rights reserved.

Essential elements are required for normal biochemical processes in organisms. However, these same elements can result in adverse health effects at doses outside a normal physiological range. Selenium (Se) is a trace element with a relatively narrow concentration range for dietary requirements and toxicity (Lemly, 1992). Different forms of organic Se are essential micronutrients such as the selenoamino acids, selenocysteine and selenomethion-

104

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

ine (SeMet), which are found incorporated into biological matrices (Fan et al., 2002). In addition to these organic forms, inorganic dietary sources such as selenite further contribute to the internal milieu of Se in the organism. Natural activities that release Se into the environment include crustal weathering and volcanic activity (Mosher and Duce, 1987; Nriagu and Pacyna, 1988; Wen and Carignan, 2007). Activities such as coal, phosphate, and uranium mining can also leach Se into the environment (Ross, 1985; Hamilton and Buhl, 2004; Maher et al., 2010). Other anthropogenic activities that release Se into the environment include production of Se for commercial use, runoff from dietary supplements for livestock that are Se deficient, and discharge of pharmaceuticals containing Se such as anti-fungal and anti-dandruff shampoos (Maher et al., 2010). As the demand for these products grows, and mining activities intensifies, there can be elevated health risks associated with increased Se in the environment. Selenium can exert sub-lethal effects in fish that are related to metabolism and growth, however studies are conflicted on whether Se enhances or impairs growth. Larval rainbow trout exposed to SeMet (4.6 and 12.0 mg/kg dietary dose) for 90 days showed decreased weight and body length compared to control fish (Vidal et al., 2005). Conversely, Thomas and Janz (2011) reported that adult zebrafish fed SeMet in their diet (3.7 and 26.6 mg/kg dietary dose) for 90–100 days showed an increase in total length and body weight. Thus, further studies are warranted to clarify the role of Se on growth in different species. One hypothesis for impaired growth following Se exposure in fish is that Se is disrupting the biochemical energy storage capacity in the liver (Bennett and Janz, 2007; Driedger et al., 2010; Thomas and Janz, 2011; McPhee and Janz, 2014). Adult zebrafish fed 9.6 mg/kg and 26.6 mg/kg SeMet for 90–100 days showed elevated levels of whole body triglycerides and glycogen (Thomas and Janz, 2011). Elevated triglycerides following Se exposure were also observed in juvenile fathead minnow fed 5.4 mg/kg SeMet for 60 days, however 9.9 and 26.5 mg/kg SeMet resulted in no change in whole body triglyceride levels (McPhee and Janz, 2014). In the same study, fathead minnows fed 5.4–26.5 mg/kg SeMet showed a decrease in whole body glycogen levels (McPhee and Janz, 2014). Thus, it appears that in some cases depending on the dose, the levels of these metabolites are altered by Se treatment. A limitation of these studies is that whole body levels of glycogen and triglycerides were measured, and this may not accurately reflect changes in the liver which is the primary storage site for these key energy-related metabolites. A second hypothesis for impaired growth by Se in fish is oxidative stress. Oxidative stress occurs when there is an overproduction of reactive oxygen species (ROS) and there is a deficiency in antioxidants (Valko et al., 2007). This imbalance can have secondary effects on growth and metabolism. This hypothesis is supported by the observation that superoxide radicals were generated when rainbow trout embryos were exposed to waterborne SeMet after the development of the liver (Palace et al., 2004). Moreover, acute exposure of juvenile rainbow trout to waterborne selenite affected hepatic glutathione peroxidase activity, suggesting that proteins related to oxidative stress are compromised (Miller et al., 2007). Thus, Se may exert adverse effects via oxidative stress, and the induction or suppression of genes or enzymes involved in mitigating oxidative stress would support this mode of action. The objectives of this study were to determine the effects of a 60 d dietary Se exposure in juvenile rainbow trout to clarify the impact of Se on growth. Dietary exposure was selected via incorporation of Se into oligochaetes, as this more accurately mirrors a realistic exposure route for rainbow trout in the natural environment. We measured endpoints related to growth, hepatic glycogen and triglyceride, oxidative stress (genes and enzyme activity) and

the transcriptome to span multiple biochemical and molecular perspectives under the context of an ecologically-relevant outcome (i.e. growth). Rainbow trout were used for this study because they are a species of high economic and ecological relevance, and are found naturally in lakes and rivers in mining regions in North America (Holm et al., 2005). Determining the dose of Se at which there are adverse effects in individuals has implications for setting safe limits in aquatic environments, and contributes to conservation efforts of fish populations by identifying high Se sites requiring monitoring and for prioritizing remediation. 2. Methods 2.1. Experimental design Juvenile rainbow trout were obtained from the Miracle Springs Trout Hatchery (Mission, BC) and were transported to Nautilus Environmental Laboratory (Burnaby, BC). Nautilus Environmental is certified by the Canadian Association for Laboratory Accreditation and all experiments were conducted in accordance with guidelines set forth to ensure animal welfare (Animal Care Manual, Nautilus, 2014). A sub-sample of fish (n = 10 fish) were measured for weight (mean ± SD: 0.28 ± 0.10 g) and length (32 ± 3 mm) at the initiation of the study. All fish were then randomly distributed among exposure and control tanks. Rainbow trout were held in a controlled-environment room at 11 ± 2 ◦ C in dechlorinated municipal tap water that was continuously aerated. The photoperiod in the environmental room was 16 h of light and 8 h of dark. A flowthrough experimental set up was used and the flow into each tank was 75 ± 25 ml/min at test initiation. This was increased to 125 ± 25 ml/min on day 15 and increased again to 225 ± 25 ml/min on day 45 to account for the growth of the fish. Dissolved oxygen (10.5 ± 0.4 mg/L), conductivity (30 ± 2 ␮S/cm) and pH (6.9 ± 0.2) were measured daily during the experimental period on a single test container. Fish were allowed to acclimatize to laboratory water conditions for 2 weeks prior to dietary dosing with Se. Mortalities were recorded daily throughout the experiment. The experimental diet consisted of live Lumbriculus variegatus worms that were cultured in selenized yeast (SelenoExcell, Fresno, CA), which is a yeast that contains organic Se. By incorporating Se into an oligochaete prior to exposure, a more suitable representation of an environmental Se exposure was achieved for rainbow trout. Nominal dietary doses of Se were 5, 10, 20 and 40 mg/kg dw Se (63, 127, 253 and 506 nmol/g). These doses were chosen to achieve body burden levels that are ecologically-relevant in the fish (Fan et al., 2002; Muscatello et al., 2008). These doses were obtained by culturing the Lumbriculus worms for different time periods between 8 and 120 h. Live worms containing Se were stored at 4 ± 2 ◦ C prior to feeding. Fish were fed daily at approximately 5% wet weight. Weight and length of fish were taken at 15, 30, and 45 days. A total of 15 7-L tanks was used, with 3 replicate tanks for the control and each of the 4 treatments. At the 60 d time point, wet weight and fork length were recorded (∼n = 30 per treatment ±1–2 fish/tank). Livers were dissected from animals following euthanasia for biochemical and molecular endpoints. Whole body and liver tissue samples were immediately placed on dry ice and stored at −80 ◦ C until further processing. 2.2. Measurements of selenium in feed and rainbow trout whole bodies Total Se analysis was done on Lumbriculus worms and rainbow trout whole body using inductively coupled plasma dynamic reaction cell mass spectrometry (ICP-DRC-MS). Samples were mea-

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

105

sured by Applied Speciation and Consulting (Bothell, WA). Samples were digested following USEPA Method 3050B. Prior to analysis, a five-point calibration curve was constructed and verified by second source standards. In order to account for operational biases, samples were instrument-blank corrected. Samples were verified using the certified reference material DORM-3. Digested samples were measured following a modified USEPA Method 200.8. Five fish from each tank were used resulting in 15 whole body fish replicates for the control group and 15 for each of the 4 treatment groups; thus there were 75 whole bodies that were assessed for total Se. Measurements are reported as mean ± standard deviation and total Se is reported as mg/kg dw. The relative contributions of different Se species was not determined in the worms nor the fish.

TRIzol (Invitrogen, Burlington, ON) following the manufacturer’s protocol. Total RNA pellet was re-suspended in 30 ␮l of RNA secure Reagent (Ambion, Austin, TX) and the concentration (ng/␮l) determined using the NanoDrop 2000 Spectrophotometer (Thermo Scientific, Wilmington, DE). RNA quality was evaluated using the 2100 Bioanalyzer (Agilent, Santa Clara, CA). Only samples with RNA Integrity Number (RIN) (Schroeder et al., 2006) >7.0 were used for further processing to ensure high quality molecular analysis. The RIN mean ± SD was 9.9 ± 0.5 and 35 samples showed a RIN > 9.0. A total of 40 samples were used for microarray and real-time PCR analysis.

2.3. Glycogen and triglyceride measurements

Additional details on the microarray analysis are provided in the supplemental methods. Briefly, a commercially available Oncorhynchus mykiss 4 × 44 K Agilent microarray (Platform ID: GPL16819) developed by Krasnov et al. (2011) was used to measure the response of the transcriptome in the liver of individuals in all the treatment groups; treatment biological replicates were from control fish (n = 8), 5 mg/kg (n = 8), 10 mg/kg (n = 7), 20 mg/kg (n = 8), and 40 mg/kg (n = 9). A minimum of n = 2 or 3 fish were selected from each tank replicates to comprise the biological replicates analyzed for each treatment. This microarray platform contains 22 K genes spotted in duplicate. The Agilent One-Color MicroarrayBased Gene Expression Analysis protocol (Agilent) was used following our published methods (Ornostay et al., 2016). Microarray data are available in Gene Expression Omnibus (GSE67599, GPL16819). Intensity data for each microarray were imported into JMP® Genomics 6. Expression data were normalized using LOESS normalization with a smoothing factor of 0.2 (Smyth and Speed, 2003; Berger et al., 2004). The limit of the detection was determined to be 2.75. Differently expressed genes (DEGs) were identified by using a one-way analysis of variance (ANOVA) followed by a false discovery rate (FDR) set at alpha of 0.05. Venny (Oliveros, 2007) was used to identify differentially expressed genes in common between the different Se treatments. Genes that showed p < 0.05 were clustered based on their normalized intensities using hierarchical clustering and was done using JMP Genomics 6® and was performed using the Fast Ward algorithm (Milligan, 1980). Pathway analysis was conducted using Pathway Studios 9.0 (Ariadne, Rockville, MD, now Elsevier) using the Mammalian ResNet 9.0. Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005) was used to identify enriched gene sets. Additional details on the bioinformatics analysis are available in the supplemental methods.

Additional details are provided in supplemental methods. Briefly, hepatic glycogen in fish was determined using the Glycogen Assay Kit (BioVision, Mountain View, CA) (n = 12 biological replicates per experimental group, n = 4 fish per each replicate tank). The glycogen assay proceeded as per instructions from manufacturer. The absorbance of the plate was measured at Excitation 530 nm/Emission 590 nm on the FLX 800 Microplate Fluorescence Reader (Bio-Tek Instruments Inc., Winooksi, VT). All comparisons were relative to the control group, which was adjusted to a relative glycogen level of 1.0. The r2 of the standard curve was ≥0.99. The limit of detection for this assay was 1256 RFU (normalized value of 0.010). All standards were tested in triplicate and individual rainbow trout liver samples were tested in duplicate. Hepatic triglycerides were measured with the Triglyceride Quantification Colorimetric/Fluorometric Kit (BioVision) using the fluorometric protocol (n = 12 biological replicates per experimental group, n = 4 fish per each replicate tank). The absorbance of each sample was measured at excitation 530 nm/emission 590 nm on the FLX 800 Microplate Fluorescence Reader (Bio-Tek Instruments Inc., Winooksi, VT). Comparisons are relative to the control group, which was adjusted to a relative triglyceride level of 1.0. The r2 of the standard curve was 0.99. The limit of detection for this assay was 1800 RFU (normalized value of 0.19) based on the last point of the linear standard curve. All standards were tested in triplicate and individual rainbow trout liver samples were tested in duplicate. 2.4. Oxidative stress biomarkers Livers were dissected from euthanized individuals, cut in half, immediately frozen on dry ice, and stored at −80 ◦ C for 2 months. Livers analyzed for oxidative stress biomarkers were n = 6-15 per treatment, with a minimum of n = 2 individuals analyzed per tank. One half of the liver was subjected to the 8-isoprostane analysis and the other half was subjected to the total glutathione analysis. All standards and rainbow trout liver samples were tested in duplicate in each assay. The Thermo Scientific Coomassie (Bradford) Protein Assay Kit (Thermo Scientific, Rockford, IL, USA) was first used to quantify total protein at 595 nm (PowerWave 340 Microplate Spectrophotometer (Winooski, VT, USA)) in every sample. The 8Isoprostane EIA Kit (Cayman Chemical Company, Ann Arbor, MI, USA) was used to quantify free 8-isoprostane at 405 nm in each sample according to the manufacturer’s protocol. The ApoGSH Glutathione Colorimetric Detection Kit (Biovision Inc., Milpitas, CA, USA) was used to quantify total glutathione at 405 nm in each sample according to the manufacturer’s protocol. 2.5. RNA extraction and quality control Total RNA was extracted from frozen rainbow trout liver samples weighing 50–100 mg. The RNA was extracted with 1 ml of

2.6. Microarray analysis and gene network analysis

2.7. Real-time PCR analysis Real-time PCR was used to measure the relative mRNA levels of genes involved in energy storage, ATP production, and oxidative stress. The following genes were selected due to their role in glycolysis and gluconeogenesis: glucokinase (gckr), glucose-6-phosphatase 1 (g6pase1), fructose-1,6-bisphosphatase (fbp), pyruvate kinase (pk), cytosolic phosphoenolpyruvate carboxykinase (pck1) and mitochondrial phosphoenolpyruvate carboxykinase (pck2). Genes involved in glycolysis and gluconeogenesis were of key interest because they are directly involved in glucose breakdown and production, respectively, and it was hypothesized that these genes would be altered with Se based on other studies showing disruptions in energy utilization. Citrate synthase (cs) was chosen for further analysis based upon its role in the Krebs cycle. Glucose-6-phosphate dehydrogenase (g6pd) was chosen based upon its involvement in the pentose phosphate pathway. Fatty acid synthase (fasn) and ATP citrate lyase (acly) were chosen based upon their role in fatty acid synthesis. Phosphorylase, glycogen, liver (pygl) was chosen due to its role in glycogen

(a) Marlatt et al., 2014; (b) Pacitti et al., 2013; (c) Ducasse-Cabanot et al., 2007; (d) Kolditz et al., 2008; (e) Polakof et al., 2010 (f) Libran-Perez et al., 2013 (g) Kamalam et al., 2013. Abbreviations are as follows: fbp, fructose-1,6-bisphosphatase; g6pd, glucose-6-phosphate-1-dehydrogenase; gk, glucokinase; pck2, phosphoenolpyruvate carboxykinase (mitochonrial); cs, citrate synthase; fasn, fatty acid synthase; acly, ATP citrate lyase; pck1, phosphoenolpyruvate carboxykinase (soluble); pk, pyruvate kinase; g6pase1, glucose-6-phosphatase 1; gpx1a, glutathione peroxidase 1a; gpx1b1, glutathione peroxidase 1b1; gpx1b2, glutathione peroxidase 1b2; ˇactin, beta actin; ef1˛, elongation factor 1 alpha; rpl60 (rl7), 60S ribosomal protein L7; mfap2, microfibrillar-associated protein 2; cytb, cytochrome b; id2, Inhibitor Of DNA Binding 2; pygl, Glycogen phosphorylase (liver form); sst1a, Somatostatin-1A.

59 59 59 59 55 59 60 65 59 55 60 61 61 58 58 59 58 58 58 58 58 0.99 1.00 0.90 0.98 0.99 0.99 0.98 0.99 0.97 0.99 0.97 0.97 1.00 1.00 1.00 0.99 1.00 1.00 0.99 1.00 0.98 100 109.4 89.5 86.1 89.1 94.5 94.6 97.4 92.1 100.4 101.5 100.1 103.7 99.3 97.6 95 103.9 108.6 115.7% 95.8% 111.0% CGACATAACGCCCACCATAGG AGAGAGCATCTGGAGCAAGT GCCTTGAACCCTTTGGTCCAG CCCGTCTTCTGATAAGTCCAA CTCATGGTCACTGTGGATGG TCTTGTTGATGGTGAGCTGT CAGATTGGAGGCCAAGATGT CCACACCGAAAAAGCCGTTC GCCCCTGGCCTTTCCTATGT TACACAGCAGCATCCAGAGC TCATCATTCTTACAATTCTCCTGATG TTCGTTATTGCAGTTCTCCTGATGTC CTTCGTTCTTGCAGTTCTCCTGATG TCTTCTCCCTGTTGGCTTTG CCAGAGTGTAGGCGAGGAGA CCCTCCACAAAGTGAGTCGT CTCCACAGCTCTTCCCACAG AAAGAAAGATGCTCCGTTGG CCTGCTGTCATCCGTTATCA CCTGCATCTTCCTCCATCTC CCACGAGACATGTCGTCCAA GCTGGACCCTTCCATCGG CTCATGGTCCTCAGGTTTG GCACGGCTGAGATGCTCTTTG GTTGGTGCTAAAGGGCACAC GGCCAAGTACTGGGAGTTCA GAGACCTAGTGGAGGCTGTC CTGAAGCCCAGACAAGGAAG CCCAGTGCCTGTGGGAAAAC CCATCGTCGCGGTAACAAGA CTCAGTGGCGACAGAAAGG ATGAAATGGCTGGGAAAATAAAGA CAACATGTCTGGAAGTGAGTTCTACAACA ACCAGGCAAATGGCTGTATGTAAGAT CTGTCTTCCCCTCCATCGT ACAAGCTGAAGGCTGAGAGG GGCAGGATGACCAAGCAG GAGACAGAGCCCACAGAACC CCTCTCCTAAAAATCGCTAATGAC GCCCAGTATCCCTCAGAACA AACCGACACCTCCACTTCACC CCCTAGCCCTTGTCATCAGC AF333188 EF551311.1 AF053331 AF246149 TC89195 (Tigr) tcaa0001c.m. 06 5.1.om.4 (Sigenae) CA349411.1 NM 001124275.1 AF246146 tcay0019b.d.18 3.1.s.om.8 (Sigenae) HE687021 HE687022 HE687023 AJ438158.1 NM 001124339.1 NM 001165156.1 NM 001141303.1 AF125208.1 NM 001124723.1 XM 014211673.1 XM 014199842.1 fbpc g6pd gkf pck2c csd fasnd aclye pck1e pkc g6pase1g gpx1ab gpx1b1b gpx1b2b ˇactina ef1˛a rpl60 (rl7)a mfap2 cytb id2 pygl sst1a

182 177 169 141 229 186 149 154 158 77 250 241 250 270 249 212 289 246 238 296 187

R1 (5 -3 ) F1 (5 -3 ) Accession No.

Product Size

Efficiency (%)

R2

Temp. (C◦ )

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

Gene Symbol

Table 1 Rainbow trout (Oncorhynchus mykiss) primer sequences used for qPCR. The table shows the goodness of fit (R2 ) and efficiency (%) of the standard curve, as well as the accession number, estimated product size, and annealing temperature.

106

regulation in the liver. Three glutathione peroxidase isoforms (glutathione peroxidase 1a (gp1a), glutathione peroxidase 1b1 (gp1b1), glutathione peroxidase 1b2 (gp1b2)) were chosen because of their role in oxidative stress and because they contain selenocysteine in the active site. Primer sets were either obtained from published literature or designed using the Primer3 program (Untergasser et al., 2012) (Table 1). The TURBOTM DNase kit (Ambion, Austin, TX) was used to remove genomic DNA from RNA samples as per manufacturer’s protocol. Following this step, the cDNA (15 ␮l total volume) was synthesized from 1 ␮g total RNA using the iScript cDNA Synthesis Kit (Bio-Rad, Mississauga, ON) as per manufacturer’s protocol. Four samples did not receive enzyme (i.e., no reverse transcriptase controls, NRT) and were used as negative controls. Each of these NRTs was derived from two independent samples of RNA selected at random. To generate the cDNA, samples were incubated with the following conditions: 5 min at 25 ◦ C, 30 min at 42 ◦ C and 5 min at 85 ◦ C. Prior to use, the cDNA was first diluted with RNAse-DNase free water at a ratio of 1:20 for the quantitative real time PCR analysis. Samples were run in duplicate and n = 8 biological replicates (total n = 40 samples) for each treatment group. Standard curves from gene expression data were assessed for linearity (r2 > 0.97) and primer efficiency (90% − 110%). Ribosomal protein 60L (rpl60), elongation factor (ef1˛) and beta-actin (ˇactin) were used to normalize gene expression (Vandesompele et al., 2002), as the mean Cq values did not vary among experimental groups. Two independent cDNA reactions were performed due to the number of genes examined, and the M values (metric of reference gene stability) for the control genes were 1.06 (rpl60, ef1˛, and ˇactin) and 1.02 (rpl60, ˇactin) respectively. Normalized expression was extracted using CFX ManagerTM software (BioRad) using the relative Cq method (Hellemans et al., 2007). The qPCR protocol followed closely the suggestions of the MIQE guidelines (Taylor et al., 2010). All amplicons were verified as correct target genes by Sanger Sequencing at McGill University and Génome Québec Innovation Centre (Montréal, QB, CAN) (Supplemental Table 1).

2.8. Data analysis Assumptions of normality and homoscedasticity were tested using Shapiro-Wilk’s test and Bartlett Equality of Variance test. When assumptions were met, one-way ANOVA was used to analyze the data, followed by a Dunnett’s post hoc test for multiple comparisons to the control group. When assumptions were not met, the Kruskal-Wallis test was used followed by Dunn’s test. Statistical analysis was performed in PRISM 6.0. Graphs were generated using PRISM 6.0.

3. Results 3.1. Selenium concentrations and survival Table 2 reports target dietary Se doses, the measured dietary doses in the Lumbriculus worms, and the measured whole body Se concentration in the fish. Here we point out that nominal doses are used throughout the manuscript for convenience. Measured doses were relatively close to nominal doses in most cases and ranged within 1–30% of the targeted dose. There was a significant difference in whole body Se concentration among groups (F4,10 = 524.5, p < 0.001). Survival ranged from 87% to 89% across treatment groups and the control group. There were no significant differences in mortality amongst treatments.

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

107

Table 2 Targeted Se dietary doses, the measured dietary Se doses in Lumbriculus worms, and the measured whole body Se in rainbow trout (n = 15 per group) after 60 days of a feeding regime. Data are mean ± standard deviation. Nominal Dose (mg/kg dw)

Measured Dose in Worms (mg/kg dw)

Measured Whole Body Se (mg/kg dw)

Control 5 10 20 40

1.30 ± 0.15a 7.10 ± 4.41b 10.70 ± 6.29c 19.50 ± 3.53d 31.80 ± 4.05e

0.96 ± 0.17a 4.39 ± 0.37b 6.03 ± 0.92c 10.43 ± 0.77d 15.02 ± 2.55e

dw = dry weight. Letters indicate significant difference.

Fig. 1. The (A) length (mm) and (B) weight (g) of juvenile rainbow trout (Oncorhynchus mykiss) exposed to 0, 5, 10, 20, and 40 mg/kg dw nominal dose of Se. In each treatment, the bar is the mean value (± SEM). Different letters indicate significant difference from the control group.

3.2. Body weight and length Rainbow trout body length was significantly different among groups (F4,10 = 14.3, p < 0.001) at 60 days, and fish in the 20 and 40 mg/kg groups showed lower body length compared to the control group following a post hoc test (Fig. 1). Wet weight was also different among groups, and the 5, 20, and 40 mg/kg groups showed lower wet weight compared to the control group following a post hoc test (F4,10 = 13.2, p < 0.001).

lower triglyceride levels in the liver when compared to control fish (Fig. 2). 3.4. Total glutathione and 8-isoprostane No significant differences (p > 0.05) were detected between treatments in total glutathione or 8-isoprostane levels in the liver of rainbow trout. Individual values for each endpoint analyzed per group are presented in Fig. 3. 3.5. Gene expression analysis

3.3. Glycogen and triglycerides There was significant variation in hepatic glycogen levels among groups (H4 = 18.7, p < 0.001) (Fig. 2) and only individuals from the 10 and the 40 mg/kg group were different from each other following a post hoc test. There was also a significant difference in liver triglyceride concentrations (F4,55 = 36.5, p < 0.0001). A post hoc test determined that both the 20 mg/kg and 40 mg/kg treated fish had

Supplemental Table 2 summarizes the number of differentially expressed genes (DEGs) in each Se treatment group and the number of transcripts that were increased or decreased in abundance in each Se treatment. After a correction for multiple hypothesis testing (FDR = 0.05), there were 449 genes that were differentially expressed in one or more of the Se treatments (Supplemental Table 2). There were 98 DEGs that were in common and differentially

Fig. 2. Relative liver glycogen and triglycerides (mean ± SEM) in rainbow trout exposed to dietary selenium (nominal doses of 0, 5, 10, 20 and 40 mg/kg). Each point represents an individual fish and the control group is adjusted to a mean value of one for a relative comparison across treatments. Different letters indicate significant differences from controls (n = 12 fish/group).

108

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

Fig. 3. Liver total glutathione and 8-isoprostane levels (mean ± SEM) normalized to total protein in rainbow trout exposed to dietary selenium (nominal doses of 0, 5, 10, 20 and 40 mg/kg). Each point represents an individual fish. Sample sizes for total glutathione are: 0 mg/kg (n = 6), 5 mg/kg (n = 5), 10 mg/kg (n = 15), 20 mg/kg (n = 13), and 40 mg/kg (n = 15). Sample sizes for 8-isoprostane are 0 mg/kg (n = 6), 5 mg/kg (n = 15), 10 mg/kg (n = 15), 20 mg/kg (n = 12), and 40 mg/kg (n = 15).

expressed from controls in the two highest Se treatments but there was little overlap in DEGs from the two lowest Se treatments after an FDR correction for multiple tests. Examples of transcripts that showed high relative fold changes from the control group included putative transmembrane protein C6orf191 (-26.6-fold in the 40 mg/kg), hemoglobin subunit zeta (-24.3-fold in the 40 mg/kg), microfibrillar-associated protein 2 precursor (+58.7-fold change in the 40 mg/kg) and ELKS/RAB6Interacting/CAST Family Member 1 (+36.7-fold change in the 40 mg/kg). There were 140 DEGs that were differentially expressed in individuals from all four Se treatments (Supplemental Fig. 1). Of these 140 genes, 78 were increased in abundance in all doses and 59 were decreased in all doses. Thus, there was strong agreement with the directional change of common transcripts identified in all four treatment groups. Three genes showed differential expression in opposite directions for duplicate probes (i.e., one probe for the gene showed up-regulation in two doses while the duplicate probe showed down-regulation in the other two doses). Of interest, based on previously reported effects of Se on growth and oxidative stress, growth hormone receptor isoform 1 increased approximately

2-fold in all four Se treatments while glutathione peroxidase 2 was down-regulated in all four Se treatments. Based upon the number of genes that were affected in individuals from each treatment, there were more DEGs detected in the higher doses of Se (3795 and 3176 non FDR corrected DEGs for the 20 and 40 mg/kg Se dose, respectively) compared to the lower doses (755 and 652 DEGs for the 5 and 10 mg/kg Se dose, respectively). Data for transcripts for each dose are provided in Supplemental Information 1. Two major clades were evident from clustering (Fig. 4), indicated by the blue line in the figure; with few exceptions, the bottom clade contained individuals from the control, 5 mg/kg, and the 10 mg/kg groups, whereas the top clade contained individuals from the 20 mg/kg and the 40 mg/kg treatment groups. Separation based upon expression patterns for individuals within the 5 and 10 mg/kg treatments and the 20 and 40 mg/kg treatment was not as clear. This may reflect the fact that measured concentrations of Se were very similar for the 5 and 10 mg/kg group (feed contained 7.1 mg Se/kg and 10.7 mg Se/kg, respectively) and these individuals are responding in a similar way to the Se.

Fig. 4. Hierarchical cluster analysis of differentially expressed transcripts (p < 0.05). The bottom clade contains fish fed control worms and the two lowest Se doses (5 and 10 mg/kg nominal Se doses). The top clade contained fish fed the two highest Se doses (20 and 40 mg/kg nominal Se dose). The two major clades are indicated to the left and grouped by treatment (0, 5, and 10 versus 20 and 40 mg/kg).

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

109

Fig. 5. A Venn diagram representing the overlap of gene sets and sub-networks in juvenile rainbow trout following a 60 day exposure to the following Se nominal doses 5, 10, 20, and 40 mg/kg dw (p < 0.05). There were 139 enriched gene sets that were altered in the liver of individuals in at least one of the four Se treatments. There were 285 sub-network enrichment analysis (SNEA) pathways (cell processes) that were altered in individuals from at least one of the four Se treatments.

3.6. Pathways identified by gene set enrichment analysis All pathways are presented in Supplemental Information 2. There were 139 enriched gene sets that were altered in the liver of individuals in at least one of the four Se treatments (Fig. 5, left panel). There were 17 pathways identified by GSEA that were altered in the liver of individuals from the two highest Se doses (20 and 40 mg/kg Se dose). In the group fed the 40 mg/kg Se dose, the “fatty acid oxidation” pathway showed a median fold change of 1.14 (p = 0.01). In the group fed the 20 mg/kg Se dose, the “heme oxidation” pathway showed a median fold change of −1.34 (p = 0.02) while in the group fed 40 mg/kg Se dose, the “heme oxidation” pathway showed a median fold change of −1.16 (p = 0.0004). Based on

the transcriptome, Notch signaling was up-regulated at the mRNA level in both the 20 and 40 mg/kg Se treatments by ∼7% (p = 0.007) (Fig. 6). Transcriptomic data suggested that insulin-like growth factor 1 receptor (IGF1R) and epidermal growth factor receptor (EGFR) pathways were altered in individuals fed Se. Five IGF1R and EGFR related pathways were increased in individuals fed the 5 and 10 mg/kg Se doses, one EGFR related pathway was decreased in individuals fed the 20 mg/kg Se dose, and nine IGF1R and EGFR related pathways were increased in individuals fed the 40 mg/kg Se dose. Taken together, there appeared to be an overall induction of IGF1R and EGRF related pathways with Se. Additionally, there was a growth hormone receptor pathway and a vascular endothe-

Fig. 6. The Notch signaling pathway in the liver of rainbow trout following dietary Se at 40 mg/kg (nominal dose). A number of downstream cell processes are affected by Notch and include glucose metabolism and triacylglycerol degradation (increased) and telomere maintenance, lipid transport (decreased). Red indicates that the gene is increased relative to the control, blue indicates that the gene is decreased relative to the control, and gray indicates the genes was not measured or was below detection limit of the microarrays. All abbreviations are provided in Supplemental Information 3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

110

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

other processes affected by Se and based on the transcriptome response, the process of chromosome condensation was dramatically increased 77% with 20 mg/kg dietary Se (Supplemental Fig. 2). Cell processes involving fatty acids and lipids were also upregulated in individuals from at least one Se treatment. These included processes such as “lipid export”, “long chain fatty acid transport”, “fatty acid import”, “lipid transport”, and “low density lipid (LDL) oxidation” (Table 3,Supplemental Fig. 3) 3.7. Quantitative real time PCR

Fig. 7. Gene set enrichment analysis suggested that growth hormone receptor signaling is altered in rainbow trout (Oncorhynchus mykiss) livers. The pathway GHR −> ELK-SRF/MYC signaling was increased at the transcriptional level after a 60-day exposure to a 40 mg/kg dw nominal Se dose. This pathway had a median fold change of 1.08 (p = 0.04). Red indicates that the gene is increased relative to the control, blue indicates that the gene is decreased relative to the control, and gray indicates the genes was not measured or was below detection limit of the microarrays. All abbreviations are provided in Supplemental Information 3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

lial growth factor receptor pathway that were increased at the gene level in individuals fed the 40 mg/kg Se dose (1.08-fold, p < 0.037) (Fig. 7). Supplemental Table 3 provides additional information on these growth pathways including their corresponding fold change and p-value. There were 285 pathways (cell processes) identified by subnetwork enrichment analysis that were altered in individuals from at least one of the four Se treatments (Fig. 5, right panel). There were six cell processes that were altered in individuals from all Se doses. These processes included “long-chain fatty acid transport”, “complement activation − alternative pathway”, “opsinization”, “tissue invasion”, and “complement activation”. As an example of

There was no significant difference among groups for acly, cs, fasn, fbp, g6pd, g6pase1, gk, pck1, pck2, or pygl (Supplemental Fig. 4–6). There were also no differences detected in the relative mRNA levels for gpx1a, gpx1b1, or gpx1b2 across treatment groups (Supplemental Fig. 7). Additional targets were measured by real-time PCR to verify microarray data based on differential expression and/or high fold change. These genes included cytochrome b (cytb), Inhibitor Of DNA Binding 2, Dominant Negative Helix-Loop-Helix Protein (id2), microfibrillar-associated protein 2 (mfap2), and somatostatin-1A precursor (sst1a). There were differences in transcript levels among groups for id2 (Kruskal-Wallis test H4 = 13.4, p = 0.009) and mfap2 (H4 = 16.7, p = 0.002), while there was no difference in mRNA abundance for cytb (H4 = 6.12, p = 0.19) nor sst1a (H4 = 5.72, p = 0.22). There was good congruence between the two techniques in terms of relative abundance of transcripts in each treatment (Fig. 8). For example, relative mRNA levels for id2 and mfap2 showed significantly higher expression in the 20 and 40 mg/kg Se treatments compared to the control and lower dose groups. Based on both techniques, transcript levels for sst1a were highest in both the 10 and 20 mg/kg group. Thus, patterns in relative changes in mRNA levels were comparable across methods with the exception of cytb mRNA. 4. Discussion This study aimed to determine the biochemical and genomic effects of dietary Se exposure in juvenile rainbow trout. Following this 60 day Se dietary exposure, there was a significant accumulation of Se in rainbow trout. The two lower Se treatments were below the USEPA (2016) criterion of 8.5 mg/kg for Se in fish tissues and the two higher doses were above the draft criterion. All four treatments were above the British Columbia guideline concentration of 4 mg/kg for Se in whole-body fish and effects on body weight were recorded at the two higher doses. In this study, hepatic glycogen was different between two treatment groups but did not change in comparison to the control fish. Previous studies have reported either an increase, decrease or no change in glycogen levels following Se treatments, and there are likely additional physiological parameters (i.e., energy

Table 3 Cell processes involving fatty acids that were significantly altered in rainbow trout (Oncorhynchus mykiss) livers following exposure to Se target doses of 5, 10, 20, and 40 mg/kg dw. Total # of neighbors are all the genes in the pathway, # of measured neighbors are those measured on the microarray, and median change is that of the entire pathway. Overall, there was an up-regulation of cell processes related to lipids and fatty acids storage/transport. Cell Process

Treatment

Total # of Neighbours

# of Measured Neighbours

Median change

p-value

long-chain fatty acid transport lipid export long-chain fatty acid transport fatty acids import lipid transport long-chain fatty acid transport long-chain fatty acid transport lipid transport LDL oxidation

5 mg/kg 5 mg/kg 10 mg/kg 10 mg/kg 10 mg/kg 20 mg/kg 40 mg/kg 40 mg/kg 40 mg/kg

11 111 11 101 368 11 11 368 89

8 81 8 67 241 8 8 241 59

1.20 1.08 1.37 1.06 1.04 1.79 1.64 1.13 1.15

0.03 0.04 p < 0.01 0.04 0.01 p < 0.01 p < 0.01 p < 0.01 p < 0.01

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

111

Fig. 8. Normalized expression of transcripts determined by real time PCR and normalized intensity determined by microarray in the liver of rainbow trout (Oncorhynchus mykiss). Transcripts are the following: (A) cytb (B) id2 (C) mfap2 and (D) sst1a. Significant differences are denoted by different letters. n = 6-8 fish per treatment for the real time PCR and n = 7-9 fish per treatment for the microarray. The bar indicates the median value of the group.

112

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

storage capacity, metabolic demands) that influence individual response to Se. Juvenile fathead minnows exposed to dietary SeMet (5.4–26.5 mg/kg) for 60 days showed a decrease in whole body glycogen, consistent with the decrease observed here in the liver of rainbow trout (McPhee and Janz, 2014). Conversely, adult female rainbow trout fed 8.5 mg/kg SeMet for 126 days showed no change in hepatic glycogen (Wiseman et al., 2011). Adult zebrafish fed SeMet showed varied responses with glycogen, with one study reporting an increase (Thomas and Janz, 2011) and another study reporting no change (Thomas et al., 2013) in whole body glycogen concentrations. It appears as though there may be dose-dependent and species specific responses for glycogen metabolism in adult fish following Se treatments and this component warrants additional studies. Hepatic triglycerides decreased in rainbow trout fed 20 and 40 mg/kg Se, but did not change in fish fed the lower doses of Se. Hepatic triglycerides were decreased by 27% and 26% compared to the control fish for the 20 mg/kg and 40 mg/kg Se, respectively. In contrast to the present study, juvenile fathead minnow fed 5.4 mg/kg SeMet showed increased whole body triglycerides while fathead minnows fed 9.9 and 26.5 mg/kg SeMet showed no change in whole body triglycerides (McPhee and Janz, 2014). Other studies also report increases in whole body triglycerides in adult zebrafish exposed to Se via the diet (Thomas and Janz, 2011; Thomas et al., 2013). In addition, the majority of studies described above measured whole body triglycerides, and tissue-specific responses can be masked when using the entire organism. The present study reports decreased hepatic triglycerides in rainbow trout, supporting the hypothesis that Se perturbs hepatic energy storage. There were no differences among treatments for endpoints related to oxidative stress (total glutathione, 8-isoprostane, and relative mRNA levels for glutathione peroxidase isoforms, glutathione peroxidase 1 (1a, 1b1 and 1b2)). As such, this study does not support the hypothesis that SeMet induces an oxidative stress response in trout. Selenium is also well known to act as an anti-oxidant and there is a fine balance between toxicity and anti-oxidant effects that can result in complex redox-related responses. Zee et al. (2016) also report that endpoints for oxidative stress were not affected in white sturgeon fed 5.6, 22.4 or 104.4 mg/kg SeMet. Oxidative stress was assessed in the livers of sturgeon by measuring gene expression and lipid hydroperoxide assay. In contrast, other studies support the hypothesis that a Se exposure can induce enzymes involved in the oxidative stress response (Palace et al., 2004; Miller et al., 2007; Lavado et al., 2012; Misra et al., 2012). In an immortal rainbow trout liver cell line, transcripts of three isoforms of glutathione peroxidase 1 and two isoforms of glutathione peroxidase 4 were increased in abundance when stimulated with organic Se (Pacitti et al., 2013), and juvenile yellowtail kingfish (Seriola lalandi) fed 4.9, 9.6, 15.4 and 20.9 mg/kg SeMet for ten weeks had higher glutathione peroxidase activity compared to controls (Le and Fotedar, 2014). One reason for discrepancies among studies is that there may be marked differences in the response of glutathione peroxidase isoforms; here we examined glutathione peroxidase 1 isoform, and not the other five known glutathione peroxidases found in some fish, such as zebrafish (Janz, 2011; Kryukov and Gladyshev, 2000). Discrepancies among studies may also arise for oxidative stress biomarkers due to additional mechanisms that do not include gene expression, for example changes in protein abundance or the activity of enzymes involved in specific aspects of oxidative stress that were not measured here (e.g., protein adducts or DNA damage by free radicals). This study provided an opportunity to compare gene expression responses to higher level whole organism and biochemical endpoints, such as growth and metabolite storage. One significant finding was that, based on the cluster analysis of transcripts, there was a distinct grouping of individuals into lower (0, 5, 10 mg/kg

Se) and higher (20 and 40 mg/kg Se) treatments. Thus, expression patterns appeared to be sensitive to the dose of Se and followed the division between doses in terms of the response observed with triglycerides (i.e. no change with lower doses and lower triglycerides with the higher doses of Se). Cell signaling pathways related to growth were in general upregulated with Se treatments. The growth factor pathways included epidermal growth factor and insulin-like growth factor 1 receptor pathways, and vascular endothelial growth factor receptor signaling. A growth hormone receptor pathway was also increased in the 40 mg/kg Se treatment. The overall induction of these growthrelated pathways was apparent at all Se doses except for one pathway in the 20 mg/kg Se treatment. In particular, EGFR signaling was a common theme that was increased at the transcript level. The roles of EGF receptor in the fish liver have yet to be elucidated and data are scarce on this key signaling pathway in teleost fishes. In zebrafish, EGF receptor has a low expression level in the liver relative to the gonad (Wang and Ge, 2004) but it is present in this tissue, suggesting that the liver is responsive to EGF-mediated signaling. In mammals, EGF pathways are involved in the stimulation of cell proliferation (Carpenter, 1987) and play a role in lipid regeneration (Scheving et al., 2014). In the present study, Se increased epidermal growth factor receptor pathways in the liver, and it is hypothesized that this response may be related to the changes in triglycerides. Furthermore, a growth hormone receptor signaling pathway was increased in the liver of rainbow trout fed 40 mg/kg Se. In both fish and mammals, the growth hormone receptor pathway is responsible for growth-promoting effects (Fauconneau et al., 1997). An increase in the growth hormone receptor pathway in response to Se exposure observed here may be a compensatory response to the decrease in body weight observed after 60 days. Our data suggest that Notch signaling is increased at the transcriptome level in the high dose Se exposures (20 and 40 mg/kg). Notch is a highly conserved intracellular signaling system that modulates numerous cell processes, including cell differentiation, cell fate, apoptosis, and cell growth among others (Baron, 2003; Andersson et al., 2011). Notch signaling pathway has also been implicated in the regulation of metabolites in transgenic mice models, in particular, regulation of gluconeogenesis and lipogenesis (Bi and Kuang, 2015). Interestingly, in the present study, rainbow trout fed high doses of Se showed increased Notch signaling and lower triglycerides in the liver. An increase in glucose and insulin levels have been observed in mice that have an overexpression in the Notch1 gene in the liver, and these transduced mice also have increased glucose-6-phosphatase expression that suggests gluconeogenesis regulation (Pajvani et al., 2011). Moreover, transduced mice with constitutively active Notch1 have increased liver size and higher triglycerides levels when compared to control mice; blocking Notch signaling in the liver decreased liver triglycerides (Pajvani et al., 2013). Song et al. (2016) demonstrated that diet induced obese mice with inhibited Notch1 signaling had improved fatty liver when compared to controls. In that study, transcripts involved in fatty acid oxidation were increased in mice with inhibition of Notch1 signaling, suggesting that the decrease in hepatic lipids may be due to increased fatty acid oxidation. Thus, there is evidence that fatty acid oxidation in the liver is regulated by Notch signaling. Based on the study by Pajvani et al. (2013), we hypothesize that Notch signaling is increased at the transcript level to compensate for lower hepatic triglyceride levels compared to controls. In general, transcriptional networks involving lipids were increased on average ∼10-20% in individuals exposed to all doses of Se. These cell processes included fatty acid transport, import, and LDL oxidation. These changes are noteworthy since triglyceride stores were decreased in fish treated with 20 mg/kg and 40 mg/kg Se. In addition, in the present study, transcripts of long-chain fatty

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

acid transport were increased in individual livers at all 4 doses. Thus, the increased expression of genes related to fatty acids may be a response to changing triglycerides in the liver. In mammals, fatty acids are transported into the mitochondria matrix to be oxidized so they can be used to generate energy (Reddy and Hashimoto, 2001). In rats, it has been demonstrated that increased enzyme activity involved in the shuttling of fatty acids into the mitochondrial matrix results in a concomitant decrease in triglycerides and increased fatty acid oxidation (Stefanovic-Racic et al., 2008), suggesting that the transport of fatty acids into the mitochondrial matrix is related to triglyceride content in the liver. The up-regulation in long chain fatty acid transport and fatty acid oxidation pathways supports the hypothesis that decreased triglycerides in the rainbow trout liver following dietary Se exposure may be a consequence of upregulating long chain fatty acid transport to shuttle fatty acids into the mitochondrial matrix for energy. Targeted genes related to energy production and storage showed no difference between control fish and those treated with Se. These genes included those in the citric acid cycle, glycolysis, gluconeogenesis, the pentose phosphate pathway and fatty acid synthesis. Thus, there was no evidence at the molecular level that these biological processes were affected by Se. This corroborates a study conducted in adult zebrafish fed SeMet, where citrate synthase and fatty acid synthase mRNA levels were not affected in the liver (Thomas et al., 2013). In contrast, rats fed 75 and 150 mg/kg Se had an increase in steady state mRNA levels of fatty acid synthase in the liver (Mueller et al., 2008). This may indicate that changes in energy storage genes after Se exposure occur differently in mammals and fish, and fish may require a higher Se dose to induce transcript changes in specific tissues. In summary, fish fed nominal doses of 20 and 40 mg/kg Se for 60 days had decreased hepatic triglycerides compared to controls, which corresponded to shorter length and lower body weight. Fish in all treatment groups showed altered lipid and growth pathways at the transcript level in the liver. This study tested Se effects at whole body Se concentrations both below and above the proposed USEPA (2016) Se criterion of 8.5 mg/kg dw Se and the British Columbia Se guideline of 4 mg/kg dw. Adverse effects on triglyceride stores were observed only in exposures above the USEPA criterion. There were a higher number of differentially expressed transcripts at the two higher Se doses following microarray analysis, suggesting a greater transcriptome response compared to that observed at the two lower doses. Biological responses in the two higher doses were therefore more similar at the molecular and biochemical level to each other compared to the lower doses of Se (e.g., at both 20 and 40 mg/kg, Notch signaling increased, triglyceride stores decreased, and length and weight were reduced). However, there were also some conserved effects across all the doses of Se. For example, growth-related signaling cascades and longchain fatty acid transport were processes that were differentially affected both below and above the USEPA criterion. We propose that decreased body weight of juvenile rainbow trout following exposure to 20 and 40 mg/kg Se is related to altered lipid profiles in the liver, and this may result in an up-regulation of pathways related to cell growth and proliferation to compensate for impaired growth.

Acknowledgements We thank Dr. Aleksei Krasnov for generously providing the rainbow trout microarray. Funding was provided by an NSERC Collaborative Research and Development grant (CRD grant 447720-2013 CRDPJ), New Brunswick Innovation Foundation RAI2014-049, Golder Associates, and a NSERC Graduate Scholarship to RK.

113

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.aquatox.2016.09. 014. References Andersson, E.R., Sandberg, R., Lendahl, U., 2011. Notch signaling: simplicity in design, versatility in function. Development 138, 3593–3612. Baron, M., 2003. An overview of the Notch signalling pathway. Semin. Cell Dev. Biol. 14, 113–119. Bennett, P.M., Janz, D.M., 2007. Bioenergetics and growth of young-of the-year northern pike (Esox lucius) and burbot (Lota lota) exposed to metal mining effluent. Ecotoxicol. Environ. Saf. 68, 1–12. Berger, J., Hautaniemi, S., Jarvinen, A., Edgren, H., Mitra, S., Astola, J., 2004. Optimized LOWESS normalization parameter selection for DNA microarray data. BMC Bioinf. 5, 194. Bi, P., Kuang, S., 2015. Notch signaling as a novel regulator of metabolism. Trends Endocrinol. Metab. 26, 248–255. Carpenter, G., 1987. Receptors for epidermal growth-factor and other polypeptide mitogens. Annu. Rev. Biochem. 56, 881–914. Driedger, K., Weber, L.P., Rickwood, C.J., Dube, M.G., Janz, D.M., 2010. Growth and energy storage in juvenile fathead minnows exposed to metal mine waste water in simulated winter and summer conditions. Ecotoxicol. Environ. Saf. 73, 727–734. Ducasse-Cabanot, S., Zambonino-Infante, J., Richard, N., Medale, F., Corraze, G., Mambrini, M., Robin, J., Cahu, C., Kaushik, S., Panserat, S., 2007. Reduced lipid intake leads to changes in digestive enzymes in the intestine but has minor effect on key enzymes of hepatic intermediary metabolism in rainbow trout (Oncorhynchus mykiss). Animal 1, 1272–1282. Fan, T., Teh, S., Hinton, D., Higashi, R., 2002. Selenium biotransformations into proteinaceous forms by foodweb organisms of selenium-laden drainage waters in California. Aquat. Toxicol. 57, 65–84. Fauconneau, B., Andre, S., Chmaitilly, J., LeBail, P., Krieg, F., Kaushik, S., 1997. Control of skeletal muscle fibres and adipose cells size in the flesh of rainbow trout. J. Fish Biol. 50, 296–314. Hamilton, S., Buhl, K., 2004. Selenium in water, sediment, plants, invertebrates, and fish in the blackfoot river drainage. Water Air Soil Pollut. 159, 2–34. Hellemans, J., Mortlier, G., Paepe, A.D., Speleman, F., Vandesompele, J., 2007. qBase relative quantification framework and software for management and automated analysis of real-time quantitative PCR data. Genome Biol. 8, R19. Holm, J., Palace, V., Siwik, P., Sterling, G., Evans, R., Baron, C., Werner, J., Wautier, K., 2005. Developmental effects of bioaccumulated selenium in eggs and larvae of two salmonid species. Environ. Toxicol. Chem. 24, 2373–2381. Janz, D.M., 2011. Selenium. In: Wood, C.M., Farrell, A.P., Brauner, C.J. (Eds.), Fish Physiology. Volume 31B. Homeostasis and Toxicology of Non-Essential Metals. Academic Press, London, pp. 327–374. Kamalam, B.S., Medale, F., Larroquet, L., Corraze, G., Panserat, S., 2013. Metabolism and fatty acid profile in fat and lean rainbow trout lines fed with vegetable oil: effect of carbohydrates. PLoS One 8, e76570. Kolditz, C., Borthaire, M., Richard, N., Corraze, G., Panserat, S., Vachot, C., Lefevre, F., Medale, F., 2008. Liver and muscle metabolic changes induced by dietary energy content and genetic selection in rainbow trout (Oncorhynchus mykiss). Am. J. Physiol.-Regul. Integr. Comp. Physiol. 294, R1154–R1164. Krasnov, A., Timmerhaus, G., Afanasyev, S., Jorgensen, S.M., 2011. Development and assessment of oligonucleotide microarrays for Atlantic salmon (Salmo salar L). Comp. Biochem. Physiol. Part D Genomics Proteomics 6, 31–38. Kryukov, G., Gladyshev, V., 2000. Selenium metabolism in zebrafish: multiplicity of selenoprotein genes and expression of a protein containing 17 selenocysteine residues. Genes Cells 5, 1049–1060. Lavado, R., Shi, D., Schlenk, D., 2012. Effects of salinity on the toxicity and biotransformation of L-selenomethionine in Japanese medaka (Oryzias latipes) embryos: mechanisms of oxidative stress. Aquat. Toxicol. 108, 18–22. Le, K.T., Fotedar, R., 2014. Toxic effects of excessive levels of dietary selenium in juvenile yellowtail kingfish (Seriola lalandi). Aquaculture 433, 229–234. Lemly, A., 1992. Guidelines for evaluating selenium data from aquatic monitoring and assessment studies. Environ. Monit. Assess. 28, 83–100. Libran-Perez, M., Lopez-Patino, M.A., Miguez, J.M., Soengas, J.L., 2013. In vitro response of putative fatty acid-sensing systems in rainbow trout liver to increased levels of oleate or octanoate. Comp. Biochem. Physiol. A-Mol. Integr. Physiol. 165, 288–294. Maher, W., Roach, A., Doblin, W., Fan, T., Foster, S., Garrett, R., Moller, G., Oram, L., Wallschlager, D., 2010. Environmental Sources, Speciation, and Partitioning of Selenium Ecological Assessment of Selenium in the Aquatic Environment. CRC Press. Marlatt, V.L., Sun, J., Curran, C.A., Bailey, H.C., Kennedy, C.K., Elphick, J.R., Martyniuk, C.J., 2014. Molecular responses to 17 beta-estradiol in early life stage salmonids. Gen. Comp. Endocrinol. 203, 203–214. McPhee, D.L., Janz, D.M., 2014. Dietary selenomethionine exposure alters swimming performance, metabolic capacity and energy homeostasis in juvenile fathead minnow. Aquat. Toxicol. 155, 91–100. Miller, L.L., Wang, F., Palace, V.P., Hontela, A., 2007. Effects of acute and subchronic exposures to waterborne selenite on the physiological stress response and

114

R. Knight et al. / Aquatic Toxicology 180 (2016) 103–114

oxidative stress indicators in juvenile rainbow trout. Aquat. Toxicol. 83, 263–271. Milligan, G., 1980. An Examination of the Effect of Six Types of Error Perturbation on Fifteen Clustering Algorithms., pp. 45. Misra, S., Hamilton, C., Niyogi, S., 2012. Induction of oxidative stress by selenomethionine in isolated hepatocytes of rainbow trout (Oncorhynchus mykiss). Toxicol. In Vitro 26, 621–629. Mosher, B., Duce, R., 1987. 1987: a global atmospheric selenium budget. J. Geophys. Res. −Atmos. 92, 13289–13298. Mueller, A.S., Klomann, S.D., Wolf, N.M., Schneider, S., Schmidt, R., Spielmann, J., Stangl, G., Eder, K., Pallauf, J., 2008. Redox regulation of protein tyrosine phosphatase 1B by manipulation of dietary selenium affects the triglyceride concentration in rat liver. J. Nutr. 138, 2328–2336. Muscatello, J.R., Belknap, A.M., Janz, D.M., 2008. Accumulation of selenium in aquatic systems downstream of a uranium mining operation in northern Saskatchewan, Canada. Environ. Pollut. 156, 387–393. Nriagu, J., Pacyna, J., 1988. Quantitative assessment of worldwide contamination of air, water and soils by trace-metals. Nature 333, 134–139. Oliveros, J.C., 2007. 2015 Venny. An Interactive Tool for Comparing Lists with Venn’s Diagrams. http://bioinfogp.cnb.csic.es/tools/venny/index.html. Ornostay, A., Marr, J., Loughery, J.R., Martyniuk, C.J., 2016. Transcriptional networks associated with 5-alpha-dihydrotestosterine in fathead minnow (Pimephales promelas) ovary. Gen. Comp. Endocrinol. 225, 23–32. Pacitti, D., Wang, T., Page, M.M., Martin, S.A.M., Sweetman, J., Feldmann, J., Secombes, C.J., 2013. Characterization of cytosolic glutathione peroxidase and phospholipid-hydroperoxide glutathione peroxidase genes in rainbow trout (Oncorhynchus mykiss) and their modulation by in vitro selenium exposure. Aquat. Toxicol. 130, 97–111. Pajvani, U.B., Shawber, C.J., Samuel, V.T., Birkenfeld, A.L., Shulman, G.I., Kitajewski, J., Accili, D., 2011. Inhibition of Notch signaling ameliorates insulin resistance in a FoxO1-dependent manner. Nat. Med. 17, 961–967. Pajvani, U.B., Qiang, L., Kangsamaksin, T., Kitajewski, J., Ginsberg, H.N., Accili, D., 2013. Inhibition of Notch uncouples Akt activation from hepatic lipid accumulation by decreasing mTorc1 stability. Nat. Med. 19, 1054–1060. Palace, V., Spallholz, J., Holm, J., Wautier, K., Evans, R., Baron, C., 2004. Metabolism of selenomethionine by rainbow trout (Oncorhynchus mykiss) embryos can generate oxidative stress. Ecotoxicol. Environ. Saf. 58, 17–21. Polakof, S., Medale, F., Skiba-Cassy, S., Corraze, G., Panserat, S., 2010. Molecular regulation of lipid metabolism in liver and muscle of rainbow trout subjected to acute and chronic insulin treatments. Domest. Anim. Endocrinol. 39, 26–33. Reddy, J., Hashimoto, T., 2001. Peroxisomal beta-oxidation and peroxisome proliferator-activated receptor alpha: an adaptive metabolic system. Annu. Rev. Nutr. 21, 193–230. Ross, H., 1985. An atmospheric selenium budget for the region 30-Degrees-N–90-Degrees-N. Tellus Ser. B-Chem. Phys. Meteorol. 37, 78–90. Scheving, L.A., Zhang, X., Garcia, O.A., Wang, R.F., Stevenson, M.C., Threadgill, D.W., Russell, W.E., 2014. Epidermal growth factor receptor plays a role in the regulation of liver and plasma lipid levels in adult male mice. Am. J. Physiol.-Gastrointest. Liver Physiol. 306, G370–G381. Schroeder, A., Mueller, O., Stocker, S., Salowsky, R., Leiber, M., Gassmann, M., Lightfoot, S., Menzel, W., Granzow, M., Ragg, T., 2006. The RIN: an RNA integrity number for assigning integrity values to RNA measurements. BMC Mol. Biol. 7, 3. Smyth, G., Speed, T., 2003. Normalization of cDNA microarray data. Methods 31, 265–273.

Song, N., Yun, U.J., Yang, S., Wu, C., Seo, C., Gwon, A., Baik, S., Choi, Y., Choi, B.Y., Bahn, G., Kim, S., Kwon, S., Park, J.S., Baek, S.H., Park, T.J., Yoon, K., Kim, B., Mattson, M.P., Lee, S., Jo, D., Park, K.W., 2016. Notch1 deficiency decreases hepatic lipid accumulation by induction of fatty acid oxidation. Sci. Rep. 6, 19377. Stefanovic-Racic, M., Perdomo, G., Mantell, B.S., Sipula, I.J., Brown, N.F., O’Doherty, R.M., 2008. A moderate increase in carnitine palmitoyltransferase 1a activity is sufficient to substantially reduce hepatic triglyceride levels. Am. J. Physiol.-Endocrinol. Metab. 294, E969–E977. Subramanian, A., Tamayo, P., Mootha, V., Mukherjee, S., Ebert, B., Gillette, M., Paulovich, A., Pomeroy, S., Golub, T., Lander, E., Mesirov, J., 2005. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. U. S. A. 102, 15545–15550. Taylor, S., Wakem, M., Dijkman, G., Alsarraj, M., Nguyen, M., 2010. A practical approach to RT-qPCR-Publishing data that conform to the MIQE guidelines. Methods 50, S1–5. Thomas, J.K., Janz, D.M., 2011. Dietary selenomethionine exposure in adult zebrafish alters swimming performance, energetics and the physiological stress response. Aquat. Toxicol. 102, 79–86. Thomas, J.K., Wiseman, S., Giesy, J.P., Janz, D.M., 2013. Effects of chronic dietary selenomethionine exposure on repeat swimming performance, aerobic metabolism and methionine catabolism in adult zebrafish (Danio rerio). Aquat. Toxicol. 130, 112–122. USEPA, 2016. Aquatic Life Ambient Water Quality Criterion for Selenium −Freshwater 2016. United States Environmental Protection Agency. https:// www.epa.gov/sites/production/files/2016-07/documents/aquatic life awqc for selenium - freshwater 2016.pdf. Untergasser, A., Cutcutache, I., Koressaar, T., Ye, J., Faircloth, B.C., Remm, M., Rozen, S.G., 2012. Primer3-new capabilities and interfaces. Nucleic Acids Res. 40, e115. Valko, M., Leibfritz, D., Moncol, J., Cronin, M.T.D., Mazur, M., Telser, J., 2007. Free radicals and antioxidants in normal physiological functions and human disease. Int. J. Biochem. Cell Biol. 39, 44–84. Vandesompele, J., De Preter, K., Pattyn, F., Poppe, B., Van Roy, N., De Paepe, A., Speleman, F., 2002. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol. 3, 0034.1. Vidal, D., Bay, S., Schlenk, D., 2005. Effects of dietary selenomethionine on larval rainbow trout (Oncorhynchus mykiss). Arch. Environ. Contam. Toxicol. 49, 71–75. Wang, Y., Ge, W., 2004. Cloning of epidermal growth factor (EGF) and EGF receptor from the zebrafish ovary: evidence for EGF as a potential paracrine factor from the oocyte to regulate activin/follistatin system in the follicle cells. Biol. Reprod. 71, 749–760. Wen, H., Carignan, J., 2007. Reviews on atmospheric selenium Emissions, speciation and fate. Atmos. Environ. 41, 7151–7165. Wiseman, S., Thomas, J.K., McPhee, L., Hursky, O., Raine, J.C., Pietrock, M., Giesy, J.P., Hecker, M., Janz, D.M., 2011. Attenuation of the cortisol response to stress in female rainbow trout chronically exposed to dietary selenomethionine. Aquat. Toxicol. 105, 643–651. Zee, J., Patterson, S., Wiseman, S., Hecker, M., 2016. Is hepatic oxidative stress a main driver of dietary selenium toxicity in white sturgeon (Acipenser transmontanus)? Ecotoxicol. Environ. Saf. 133, 334–340.