Role and mechanism of the AMPK pathway in waterborne Zn exposure influencing the hepatic energy metabolism of Synechogobius hasta

Previous studies have investigated the physiological responses in the liver of Synechogobius hasta exposed to waterborne zinc (Zn). However, at present, very little is known about the underlying molecular mechanisms of these responses. In this study, RNA sequencing (RNA-seq) was performed to analyse the differences in the hepatic transcriptomes between control and Zn-exposed S. hasta. A total of 36,339 unigenes and 1,615 bp of unigene N50 were detected. These genes were further annotated to the Nonredundant protein (NR), Nonredundant nucleotide (Nt), Swiss-Prot, Kyoto Encyclopedia of Genes and Genomes (KEGG), Clusters of Orthologous Groups (COG) and Gene Ontology (GO) databases. After 60 days of Zn exposure, 708 and 237 genes were significantly up- and down-regulated, respectively. Many differentially expressed genes (DEGs) involved in energy metabolic pathways were identified, and their expression profiles suggested increased catabolic processes and reduced biosynthetic processes. These changes indicated that waterborne Zn exposure increased the energy production and requirement, which was related to the activation of the AMPK signalling pathway. Furthermore, using the primary hepatocytes of S. hasta, we identified the role of the AMPK signalling pathway in Zn-influenced energy metabolism.

Zinc (Zn) is a ubiquitous micronutrient required for the normal growth, reproduction and development of animals, including fish. As an essential ion for more than 300 enzymes, Zn plays key roles in many aspects of cellular metabolic processes, including carbohydrate, lipid and protein metabolism. During the past several decades, extensive studies have focused on the essential roles of Zn in various biological processes and its toxic effects on many organisms 1,2 . Despite a considerable tolerance to high doses of Zn in some organisms 2 , excessive Zn in the aquatic environment can be toxic 1 and has been reported to adversely impact growth, survival, reproduction, histological changes, metal bioaccumulation and the production of reactive oxygen species in fish species [3][4][5] . Accordingly, excessive Zn can pose a serious threat to the sustainable development of aquaculture. There are also studies about the effects of Zn on carbohydrate, lipid and protein metabolism [6][7][8][9] . However, the underlying molecular mechanism of waterborne Zn exposure in the perturbation of energy metabolism remains unclear.
In fish, the liver is one of the main sites of Zn bioaccumulation and plays a central role in energy metabolism. Protein, lipid and glucose are the major energy sources, and their balance may largely determine the energy homeostasis of the organism. Generally, disorder of energy metabolism is caused by an imbalance among energy intake from the diet, protein anabolism and catabolism, de novo fatty acid synthesis (lipogenesis)/glucose (gluconeogenesis) and fat catabolism via β -oxidation (lipolysis)/glucose breakdown (glycolysis). At present, accumulating evidence has demonstrated that cellular metabolic pathways and some kinase components play crucial roles in the regulation of energy homeostasis. Among them, AMP-activated protein kinase (AMPK)

Results
Illumina sequencing and sequence assembly. A total of 18 hepatic RNA samples, collected from 3 biological replicates of each treatment (control: C1, C2 and C3; Zn treatment: T1, T2 and T3), were subjected to RNA sequencing (RNA-seq). Approximately 340 million reads were generated, and every sample yielded 51.5 to 53.9 million clean reads (Supplementary Table 1). Of these, 97.44% of clean reads had quality scores greater than or equal to Q20 (the base quality score of 20 means an error probability of 1%, based on Phil Green's PHRED base-calling software). Furthermore, 36,339 unigenes were detected after assembly, including 5,669 clusters and 30,670 singletons. The total length for unigenes was 33,757,047 nucleotides (nt), and the average length was 929 nt. The N50 (median length of all non-redundant sequences, with higher N50 values indicating better quality of assembly) was 1615 nt (Supplementary Table 2). The length distribution of All-Unigene is shown in Supplementary Fig. 1A. All reads have been submitted to the Sequence Read Archive at NCBI (Accession Number: SRP073412).

Functional annotation and classification of unigenes.
To verify that we annotated the unigenes, all unigene sequences were searched in the Nonredundant protein (NR), Nonredundant nucleotide (Nt), Swiss-Prot, Kyoto Encyclopedia of Genes and Genomes (KEGG), Clusters of Orthologous Groups (COG) and Gene Ontology (GO) databases. The results are shown in Supplementary Table 3, based on the cut-off e-value < 0.00001. For the analysis of the protein coding region, the number of coding DNA sequences (CDSs) that mapped to the protein database was 22,694, and the number of predicted CDSs (ESTscan software) was 1,109 (Supplementary Table 3). The length distribution of CDS nucleotide sequences is shown in Supplementary Fig. 1B and C. Only unigenes mapped against CDSs were used in the subsequent analyses.
For the GO analysis, 255, 291 and 301 DEGs were grouped in the cellular component, molecular function and biological process categories, respectively. Within the biological processes, most of the DEGs were classified into cellular processes, followed by metabolic processes. The terms "cell" and "cell part" and "binding" were dominant in the categories of cellular components and molecular functions, respectively (Fig. 2). To characterize the functional analysis of DEGs, we performed a pathway analysis based on the KEGG database. Of the 945 DEGs, 412 had a specific KEGG pathway annotation, indicating that Zn exposure influenced different biological pathways. Furthermore, as a continuation of our previous study 16 , the present study focused on the analysis of pathways involved in energy metabolism, including lipid metabolism (Fig. 3), carbohydrate metabolism ( Fig. 4) and the AMPK pathway (Fig. 5). Some key DEGs involved in oxidative phosphorylation, apoptosis, stress and repair are summarized in Supplementary Table 5. In addition, a relative RPKM (reads per kilobase per million mapped reads) value was calculated for each DEG (RPKM gene /RPKM lowest ). A heatmap was constructed using log 2 (relative RPKM value) to visualize the differential expression patterns of genes involved in lipid metabolism ( Supplementary Fig. 3), carbohydrate metabolism ( Supplementary Fig. 4) and the AMPK pathway ( Supplementary Fig. 5). The DEGs with similar expression patterns were clearly clustered, and the abbreviations of genes are shown in the Abbreviation List of the Supplementary file.

Validation of DEGs.
To validate our RNA-seq results, 20 DEGs (15 up-and 5 down-regulated genes) were selected for validation by Q-PCR (real-time quantitative PCR). In all, except the CIDEC (cell death-inducing DFFA-like effector protein C) and HMGR (hydroxymethylglutaryl-CoA reductase) genes, the expression profiles of the other 18 genes exhibited similar trends both in RNA-seq and Q-PCR analysis ( Supplementary Fig. 6). The correlation coefficient between RNA-seq and Q-PCR results was 0.892 (p < 0.001).
Cell viability, intracellular TG (triglyceride) and ATP Content. Compared to the control, Zn and/or the inhibitor CC (Compound C) of the AMPK pathway had no adverse effects on cell viability except in the 33 μ M Zn + 200 nM CC group ( Supplementary Fig. 7). Compared to the control, 3.3 μ M Zn had no significant effect on intracellular TG content, but 33 μ M Zn significantly reduced the TG content. Compared to the single 33 μ M Zn-exposed group, CC pre-treatment and then Zn incubation resulted in a slight increase of TG, although the differences were not statistically significant (Fig. 6A). The intracellular ATP level declined with increasing Zn concentration (Fig. 6B). Compared to the single Zn exposure group, CC pre-treatment and then Zn incubation resulted in a slight reduction of intracellular ATP content, although the differences were not statistically significant.
Enzymatic activities. Compared to the control, 3.3 μ M Zn had a significant effect on the activities of ME (malic enzyme), CPT I (carnitine palmitoyl transferase I), G6PD (glucose-6-phosphate dehydrogenase), HK (hexokinase) and PEPCK (phosphoenolpyruvate carboxykinase), and 33 μ M Zn significantly increased the activities of ICDH (isocitrate dehydrogenase), CPT I, G6PD and HK, and reduced the activities of ME and FAS (fatty acid synthase) (Fig. 7). For all of the tested enzymes, a single CC incubation showed no significant effects on their activities. Compared to single Zn treatment, CC pre-treatment reduced the activities of CPT I, HK and PEPCK, increased the activities of FAS in the CC + 33 μ M group, and increased the activities of FAS and PEPCK in the CC + 3.3 μ M group.
mRNA expression levels of genes. In hepatocytes of S. hasta, compared to the control, a single CC incubation showed no significant effect on the mRNA levels of tested genes, except for TAK1 expression (Fig. 8), and 3.3 μ M Zn significantly reduced the expression of ACC (acetyl-CoA carboxylase) and 6PGD (6-phosphogluconate dehydrogenase) and increased the mRNA levels of AMPK, CPT I and PFK2 (phosphofructokinase 2). The mRNA levels of AMPK, LKB1/STK11 (serine/threonine-protein kinase 11), CaMKKβ , HK, PFK2, CPT I, G6PD and ICDH were significantly higher in the 33 μ M Zn-treated group compared to those in the control group. Compared to the control, 33 μ M Zn exposure down-regulated the mRNA levels of G6PC (glucose-6-phosphatase), GS (glycogen synthase), ACC, HSL (hormone sensitive lipase), ME and SREBP 1 (sterol-regulator element-binding protein 1). Compared to single Zn incubation, CC pre-treatment reduced the mRNA levels of AMPK and PFK2, up-regulated the mRNA levels of GS, HMGR and HSL in the CC + 33 μ M group, and reduced the mRNA levels of LKB1, G6PC, PEPCK, PFK2, G6PD and SREBP-1 in the CC + 3.3 μ M group but had no significant effect on the mRNA levels of other tested genes.

Discussion
At present, studies involving the effects of metal elements on fish often focus on oxidative damage, disease and apoptosis, and little is known about their effects and mechanisms on energy metabolism 1,2,16 . The present study provides crucial molecular insights into the mechanism of how Zn influences hepatic energy metabolism at the whole transcriptomic levels. Furthermore, using the primary hepatocytes of S. hasta, we identified a role of the AMPK signalling pathway in Zn-influenced energy metabolism in S. hasta.
In most fish species, the liver is the principal site involved in energy balance and lipid homeostasis 17 . Our recent study showed that waterborne Zn exposure reduced hepatic lipid deposition and influenced lipid The colours of ellipses were shaded according to significance level. Red: the mRNA levels of Zn-exposed fish were significantly higher than those in the control (Probability ≥ 0.8, and the absolute value of log 2 (Ratio) ≥ 1). Green: the mRNA levels of Znexposed fish were significantly lower than those in the control (Probability ≥ 0.8, and the absolute value of log 2 (Ratio) ≥ 1). Grey: not DEGs. The colours of ellipses were shaded according to significance level. Red: the mRNA levels of Zn-exposed fish were significantly higher than those in the control (Probability ≥ 0.8, and the absolute value of log 2 (Ratio) ≥ 1). Green: the mRNA levels of Zn-exposed fish were significantly lower than those in the control (Probability ≥ 0.8, and the absolute value of log 2 (Ratio) ≥ 1). Grey: not DEGs. metabolism of S. hasta 16 . However, due to the lack of transcriptomic information, the molecular mechanism remained unknown. In this study, using RNA-seq, we obtained many DEGs involved in lipid metabolism (Fig. 3). For example, the genes linked to fatty acid β -oxidation, such as CPT I, ACO (acyl-CoA oxidase) and ACSL (long-chain acyl-CoA synthetase), were up-regulated by Zn exposure, indicating increased lipolysis. Given that mitochondrial fatty acid β -oxidation is the major source of energy for the organism 18 , the present results likely reflected the enhancement of energy expenditure and demand after Zn exposure. Interestingly, Zheng et al. 9 observed that chronic waterborne Zn exposure reduced the CPT I mRNA level in yellow catfish Pelteobagrus fulvidraco. This difference may be due, at least partly, to the difference of species. The present study also indicated that the mRNA levels of genes involved in glycerolipid metabolic process, such as LPL (lipoprotein lipase), ATGL (adipose triglyceride lipase) and MGLL (acylglycerol lipase), were up-regulated. LPL hydrolyses TG present in plasma lipoproteins and supplies free fatty acids (FAs) for storage or for oxidation 19 . MGLL converts monoacylglycerides to free FAs and glycerol, and ATGL is one of the main enzymes mediating TG catabolism 20 . The enhanced expression of these genes indicated the increased requirement for fatty acids, which might further promote FA oxidation. For FA elongation, ELOVL4 (elongation of very long-chain fatty acids protein 4) participates The colours of ellipses were shaded according to significance level. Red: the mRNA levels of Zn-exposed fish were significantly higher than those in the control (Probability ≥ 0.8, and the absolute value of log 2 (Ratio) ≥ 1). Green: the mRNA levels of Zn-exposed fish were significantly lower than those in the control (Probability ≥ 0.8, and the absolute value of log 2 (Ratio) ≥ 1). Grey: not DEGs. in the biosynthesis of long-chain FAs 21 . Its expression was significantly down-regulated, possibly indicating that waterborne Zn exposure inhibited long-chain FA synthesis. However, the mRNA levels of genes related to lipogenesis, such as G6PD, 6PGD, ME 22 , ACC 23 and FAS 24 , were not significantly influenced by waterborne Zn exposure. In general, lipid homeostasis is characterized by the balance between lipolysis and lipogenesis. The present study indicated that Zn exposure mainly improved lipolysis rather than lipogenesis at the transcriptional level, indicating an increased energy expenditure.
Despite the low efficiency of carbohydrate utilization in fish, previous studies demonstrated the existence of a glucosensing system in fish and that carbohydrates are also essential for function 25,26 . In the present study, the changes in the mRNA levels of genes involved in gluconeogenesis did not follow a constant trend. As shown in Fig. 4, G6PC and PEPCK, two key regulatory enzymes in gluconeogenesis 27 , showed opposite changes at the transcriptional level. The present study indicated a reduction of GS expression and an increase in the PYG (glycogen phosphorylase) mRNA level after Zn exposure, indicating that Zn exposure promoted glycogen breakdown and inhibited glycogen synthesis. Of the three rate-limiting enzymes of glycolysis, the mRNA expression of HK was significantly up-regulated. Although PFK expression remained relatively stable, the mRNA level of PFK2, which activates PFK through fructose-2,6-bisphosphate production 28 , was significantly up-regulated. Thus, changes in the expression levels of these genes might indicate increased glycolysis. The tricarboxylic acid (TCA) cycle is a pivotal metabolic pathway that unifies carbohydrate, lipid and protein metabolism. KGDH (ketoglutarate dehydrogenase), ICDH and MDH (malate dehydrogenase) are critical enzymes in the TCA cycle involved in the production of NADH, which is in turn used by the oxidative phosphorylation pathway to generate adenosine triphosphate (ATP) 29 . The present study indicated that the mRNA level of KGDH was up-regulated. In addition, Zn exposure also up-regulated the expression of several key regulators involved in ATP biosynthesis, such as NADH dehydrogenase (NADH-Q, complex I), succinate dehydrogenase (SDH, complex III), cytochrome c oxidase (Cyt C Ox, complex IV) and three components (F-type subunit alpha, V-type proteolipid subunit and V-type S1 subunit) of ATP synthase (Supplementary Table 5). The increased ATP production is likely attributable to the increased metabolic expenditure for detoxification and the maintenance of homeostasis, as suggested by several researchers 6,8,9,22 .
The AMPK pathway is an intracellular master sensor and regulator of energy homeostasis and is activated by ATP depletion or a rise in the AMP/ATP ratio 11,12 . The present study indicated that Zn exposure significantly up-regulated AMPK expression (Fig. 5), suggesting the possibility of cellular ATP depletion. Similarly, Lemire et al. 29 noted that Zn reduced the ATP content in hepatocytes. We also found that Zn exposure also significantly up-regulated poly (ADP-ribose) polymerase (PARP) expression (Supplementary Table 5). PARP activation is an immediate cellular response to DNA damage, and activated PARP can deplete cellular ATP and NAD + contents to repair the damaged DNA 30 . In the present study, the mRNA levels of genes involved in the DNA fragmentation of apoptosis 31,32 , such as AIF (apoptosis-inducing factor), DFFA (DNA fragmentation factor, 45 kD, alpha subunit) and DFFB (DNA fragmentation factor, 40 kD, beta subunit), were significantly higher in the Zn-treated group than those in the control. In addition, Zn exposure changed the expression levels of calpain and caspase-3 (CASP3), key genes mediating cellular apoptosis 33 . Together, these changes suggested that excessive Zn exposure also induced DNA damage and apoptosis in the liver of S. hasta, in agreement with other reports 34 . Additionally, many pivotal genes involved in immune pathways, including NF-κ B (nuclear factor NF-kappa-B), chemokine and NFAT (nuclear factor of activated T cells), were up-regulated by Zn exposure (Supplementary Table 5). NF-κ B is an important redox-sensitive transcription factor that regulates gene expression involved in immune and inflammatory responses 35 . In the liver, chemokines recruit immune and non-immune cells into inflamed sites and promote wound healing 36 . The elevation of their expression levels indicated that Zn exposure influenced the immune system of S. hasta. Thus, all of the observations above indicated the Zn-induced increase of ATP depletion and the Zn-induced DNA damage and apoptosis. Taken together, it was plausible to imply that Zn exposure caused ATP depletion, which in turn activated AMPK to adjust the ATP-consumption and ATP-generation rates.
Generally, divalent metals can affect cellular calcium homeostasis and calcium signalling 37,38 . In the present study, waterborne Zn exposure up-regulated the expression of Ca 2+ -dependent proteins, such as calpain and calreticulin (CRT) 39,40 (Supplementary Table 5). The results might reflect an elevation of cellular free Ca 2+ concentration after waterborne Zn exposure. As we mentioned above, AMPK also responds to some protein kinases (AMPK kinase, AMPKK). Ca 2+ /calmodulin-dependent protein kinase kinase β (CaMKK β ), one of AMPK's upstream kinases 41,42 , senses changes in the intracellular Ca 2+ ion concentration 43 and has been described as the main kinase that activates AMPK in response to the elevation of intracellular Ca 2+ levels 41,42 . However, the present study indicated that waterborne Zn exposure did not significantly influence the mRNA levels of CaMKKβ (Fig. 5), indicating that activation between CaMKKβ and AMPK does not always remain constant.
The present study also indicated that Zn activated the insulin signalling pathway because of the up-regulation of the mRNA expression of InsR (insulin receptor), IRS (insulin receptor substrate) and PI3K (phosphatidylinositol-4,5-bisphosphate 3-kinase). However, insulin expression was decreased. In fact, AMPK is able to activate IRS and affect insulin sensitivity by modulating the mTOR (serine/threonine-protein kinase The data (mean ± SEM, n = 4) were normalized to the housekeeping genes TBP and TUBA and are expressed relative to the control treatment. Bars that share different lowercase letters indicate significant differences among groups (p < 0.05).
Because transcriptome analysis indicated that the AMPK signalling pathway played an important role in the Zn-induced change of energy metabolism, an in vitro experiment using primary hepatocytes was conducted to explore its mechanism. Among several key metabolic enzymes, the activities of ME and FAS were significantly decreased while the activities of CPT I and HK increased after Zn treatment, indicating the enhancement of lipolysis, glycolysis and the reduction of lipogenesis. In the present study, 33 μ M Zn incubation significantly influenced the mRNA levels of AMPK, G6PC, HK, GS, PFK2 and CPT I 6PGD, in agreement with the results of RNA-seq. As a consequence, the TG content of hepatocytes decreased in the 33 μ M group (Fig. 6). The intracellular ATP level declined with increasing Zn concentration, indicating the probable activation of the AMPK pathway, which further confirmed the results of the transcriptome analysis. Meanwhile, CaMKKβ expression was up-regulated, also indicating the activation of the AMPK pathway after Zn exposure, since CaMKKβ is an important kinase upstream of AMPK 41,42 (Fig. 8). Hormone-sensitive lipase (HSL), a key lipolytic enzyme, is a target gene for AMPK regulation 46 . The present study indicated that 33 μ M Zn exposure significantly down-regulated HSL expression. Except for oxidative processes, the free fatty acids released by lipolysis could be re-utilized for lipogenesis. Thus, down-regulated HSL expression contributed to ensure that the rate of lipolysis did not exceed the rate of lipogenesis and thus prevented unnecessary ATP consumption induced by excessive fatty acids 14 .
Similarly, previous studies demonstrated that HSL could be inhibited by AMPK activation 10,47 . Moreover, as shown in Figs 7 and 8, Zn incubation increased the activities and mRNA levels of G6PD and ICDH, the key regulatory enzymes involved in the production of NADPH, an essential material for lipid synthesis 21 . NADPH is also important in providing the reductive power necessary to regenerate antioxidants such as SOD and glutathione 48 . Therefore, the changes of these two NADPH-dependent enzymes likely contributed to produce large amounts of NADPH to cope with oxidative stress, as suggested by previous studies 48,49 . The present study indicated that, compared to single Zn treatment, CC pre-treatment reduced the activities of CPT I, HK and PEPCK and the mRNA levels of AMPK and PFK2. CC pre-treatment increased FAS activity and the mRNA levels of GS, HMGR and HSL in the CC + 33 μ M group and increased the activities of FAS and PEPCK and down-regulated the mRNA levels of LKB1, G6PC, PEPCK, PFK2 and G6PD in the CC + 3.3 μ M group. These findings indicated that these genes were potential targets of AMPK, in agreement with the report by Kahn et al. 13 . Compared to single Zn groups, CC pre-treatment significantly reduced the mRNA levels of AMPK in the CC + 33 μ M Zn group. Thus, our study suggested that CC addition caused the reversion of the Zn-induced change on AMPK expression in the higher Zn group and accordingly affected the expression of downstream genes. In the present study, compared to the single Zn exposure group, CC pre-treatment and then Zn incubation resulted in a reduction of ATP content, although the differences were not statistically significant (Fig. 6). Given the close correlation between AMPK and the ATP level, the result might indicate that AMPK activation contributed to maintain a relatively adequate ATP level. The present study indicated that the 33 μ M Zn + 200 nM CC group exhibited reduced cell viability. Studies have noted that intracellular ATP depletion induces cell death 50 . Taken together, the present study indicated that AMPK signals constitute a pivotal link between the ATP level and energy metabolism at the transcriptional level and that AMPK activation may be required for an ATP-conserving mechanism in response to Zn exposure.
In conclusion, this study evaluated the effects of waterborne Zn exposure on hepatic metabolism in S. hasta at the transcriptomic level and found that Zn exposure promoted catabolic processes and inhibited biosynthetic processes. AMPK played an important role in the Zn-induced changes in these genes' expression levels and pathways. Furthermore, in vitro evidence suggested that the changes of energy metabolism compensated for the enhancement of energy demands, which might be attributable to the ATP-conserving mechanism of the AMPK signalling pathway in response to Zn exposure.

Materials and Methods
The study consisted of two experiments. In Exp. 1, RNA-seq technology was used to explore the effects and mechanisms of waterborne Zn exposure on signalling pathways at the genome-wide level. Based on the results of Exp. 1, Exp. 2 was conducted to investigate the potential mechanism of AMPK pathways in Zn-influenced hepatic physiological changes of S. hasta. We assured that the experiments performed on animals and cells followed the ethical guidelines of Huazhong Agricultural University and confirmed that all experimental protocols were approved by Huazhong Agricultural University. Experiment 1: Transcriptome analysis. Fish and Zn exposure. S. hasta were obtained from a local marine water pond (Panjin, China), and the culture experiment was performed following the procedures described in our recent study 16 . Briefly, the fish were transferred to indoor cylindrical fibreglass tanks (300 L water volume) for two weeks of acclimation. After acclimation, 144 uniform-sized fish (initial mean weight: 11.3 ± 0.3 g, mean ± SEM) were randomly assigned to 6 fibreglass tanks with 24 fish per tank (200 L in water volume). They were exposed to two nominal Zn concentrations of zero (control, without extra Zn addition) and 8.3 μ M (0.75% of the 96 h 50% lethal concentration [LC 50 ] of Zn for S. hasta, Zheng et al. 8 ), respectively, with triplicates for each concentration. Zn was added as ZnSO 4 ·7H 2 O (AR, Shanghai Sinopharm Group Corporation, Shanghai, China) and was dissolved in distilled water for stock concentrations. Individual test solutions during the experiment were obtained by adding the appropriate volume of the primary stock to the dilution. The Zn concentrations in the test tanks were monitored twice every week by inductively coupled plasma atomic emission spectrometry (ICP-AES). The measured Zn concentrations for the two Zn treatments were 0.08 ± 0.03 and 8.4 ± 0.1 μ M, respectively.
The experiment was conducted in a semi-static aquarium system at ambient temperature with a natural photoperiod provided with continuous aeration to maintain the dissolved oxygen level near saturation. All fish were fed 6% of their biomass daily (two meals per day) with minced trash fish. After 15 min, the uneaten food was removed from the tanks. The amount of food consumed by the fish in each tank was recorded daily, and the Zn Scientific RepoRts | 6:38716 | DOI: 10.1038/srep38716 concentrations used here did not affect the feeding rate. Meanwhile, to ensure good water quality and to maintain waterborne Zn levels, water was renewed twice daily. Water quality parameters were monitored twice a week in the morning. The parameters were as follows: water temperature 23.7 ± 3.2 °C; pH 8.3 ± 0.2; dissolved oxygen 0.22 ± 0.01 mM; salinity 18.9 ± 0.4%; total hardness 78.6 ± 1.5 mM and total alkalinity 6.6 ± 0.3 mM. The experiment continued for 60 days.
Sampling and RNA isolation. At the end of the 60-day period, the fish were starved for 24 h before sampling. After euthanizing with MS-222 (tricaine methanesulfonate, 0.38 mM), 3 fish were randomly selected from each tank (Note: the remaining fish per tank were used for histochemical observation, the determination of enzymatic activities and mRNA analysis, as described in our previous study ref. 16). Liver samples were excised and immediately frozen in liquid nitrogen. Total RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. The RNA quality and quantity were measured using a NanoDrop 2000 (Thermo Scientific, Wilmington, DE, USA) and an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). All of the samples were standardized to 500 ng/μ L, and 3 RNA samples (equal volumes) from the same tank (the same biological replicate) were combined into one pool for transcriptome analysis. There were three replicate tanks (n = 3 biological replicates) for transcriptome analysis for each treatment.
Library preparation and Illumina sequencing. The transcriptome library construction and sequencing were performed at the Beijing Genome Institute (BGI, Shenzhen, China). RNA samples were digested by DNase I, and magnetic beads with Oligo (dT) were used to isolate mRNA (Dynabeads mRNA Purification Kit, Invitrogen, CA, US).
The mRNA was sheared to yield short fragments with Fragment Buffer (Ambion ® , ThermoFisher Scientific, MA, US), and the fragments were used as templates for cDNA synthesis. N6 primer, First Strand Master Mix and Super Script II reverse transcription (Invitrogen, CA, US) were mixed together and incubated at 25 °C for 10 min, 42 °C for 30 min and 70 °C for 15 min to synthesize the first strand cDNA. Then, the Second Strand Master Mix (Invitrogen) was added to the solution, which was then incubated at 16 °C for 2 h to synthesize the second strand cDNA. The cDNA fragments were purified with a QIAquick PCR Purification Kit (QIAGEN, Germany). The purified cDNA fragments were resolved with EB buffer and used for end reparation and single nucleotide A (adenine) addition. They were then connected to adapters. Afterwards, the suitable fragments (300-350 bp in size) were selected for the PCR amplification. The amplified products were purified to create cDNA libraries. The Agilent 2100 Bioanalyzer and the ABI StepOnePlus Real-Time PCR System were used to quantify and qualify the sample libraries. Finally, the qualified libraries were amplified on a cBot to generate the cluster on the flowcell (TruSeq PE Cluster Kit V3-cBot-HS, Illumina, San Diego, CA, USA). The amplified flowcell was sequenced on a HiSeq 2000 System (TruSeq SBS KIT-HS V3, Illumina). In total, six libraries were generated in the present study and were run in two lanes on the Illumina platform.
Assembly and functional annotation. Following sequencing, the raw image data were obtained and transformed by base calling for sequence data (90-bp raw paired-end reads). After filtering adaptor sequences and low-quality bases from the reads, we used the clean reads (above 4 G per sample) for bioinformatics analysis. De novo transcriptome assembly was achieved using Trinity 51 . The resulting sequences were considered to be unigenes. To annotate the transcriptome, we performed the BLASTx alignment (e-value < 10 −5 ) between unigenes and protein databases, including NR, Nt, Swiss-Prot, KEGG and COG. Genes were identified according to the best hits against known sequences. GO functional annotation was accomplished with Blast2GO software 52 , and further gene classifications were performed using the WEGO program 53 .
DEGs and Q-PCR validation. The gene expression levels were calculated using the RPKM method 54,55 . DEGs were screened using Noiseq Method55 with a threshold of 0.8 (diverge probability ≥ 0.8). For pathway and GO enrichment analysis, all DEGs were mapped to terms in GO and the KEGG database.
Twenty candidate genes involved in lipid and carbohydrate metabolism, the AMPK pathway and signal transduction were selected for real-time quantitative PCR (Q-PCR) validation. Total RNA from control and Zn-exposed groups (n = 3 replicate tanks, two fish were sampled for each tank, and samples were not pooled) was extracted as described above. Total RNA was quantified spectrophotometrically, and the integrity was assessed by agarose gel electrophoresis. First-strand cDNA was synthesized using a PrimeScript TM RT reagent Kit with gDNA  Supplementary Table 4. The Q-PCR parameters consisted of initial denaturation at 95 °C for 30 s, followed by 40 cycles at 95 °C for 5 s, 57 °C for 30 s and 72 °C for 30 s. All reactions were performed in duplicate, and each reaction was confirmed to contain a single product of the correct size by agarose gel electrophoresis. A non-template control and dissociation curve were performed to ensure that only one PCR product was amplified and that stock solutions were not contaminated. Standard curves were constructed for each gene using serial dilutions of stock cDNA. The amplification efficiencies of all genes were approximately equal and ranged from 97% to 102%. A set of seven housekeeping genes, including β -actin, GAPDH (glyceraldehyde-3-phosphate dehydrogenase), RPL7 (ribosomal protein L7), 18 S rRNA, HPRT (hypoxanthine-guanine phosphoribosyl transferase), UBCE (ubiquitin-conjugating enzyme) and TUBA (tubulin alpha chain) were selected from the literature 56 to test their stability of mRNA expression. The mRNA expression levels of β -actin and RPL7 were the most stable under the experimental conditions, as suggested by geNorm software 56 . Thus, the expression levels of each tested gene were normalized to the geometric mean of the best combination of β -actin and RPL7. The fold changes in relative expression to the control were calculated using the 2 −ΔΔCt method 57 .
Scientific RepoRts | 6:38716 | DOI: 10.1038/srep38716 Experiment 2: Treatments in vitro. Hepatocyte culture and treatments. Hepatocytes were isolated from juvenile S. hasta according to our recent study 58 , and the cells were counted in a haemocytometer. Trypan blue exclusion was utilized to evaluate cell viability, and only those cultures with more than 95% cell viability were accepted for the subsequent experiment. The freshly isolated hepatocytes were seeded at a density of 1 × 10 6 cells/mL onto 25 cm 2 flasks and kept at 28 °C in a CO 2 incubator (0.5% CO 2 ). For each culture, a pool of cells from four fish was used.
For the Zn-exposed experiment, hepatocytes of S. hasta were incubated with Zn and/or CC (Dorsomorphin, Selleck S7306, Selleck Chemicals, Houston, TX, USA), a widely used AMPK inhibitor. Here, six groups were designed as follows: control, 3.3 μ M Zn (0.75% of the 96-h IC 50 of Zn for S. hasta hepatocytes. Here, IC 50 represents concentration resulting in 50% inhibition of cell growth), 33 μ M Zn (7.5% of the 96 h IC 50 ), 200 nM CC, 3.3 μ M Zn + 200 nM CC and 33 μ M Zn + 200 nM CC. The groups containing CC were pre-treated with CC for 1 h prior to the addition of Zn. The 96 h IC 50 (436.78 μ M) of Zn for S. hasta hepatocytes was obtained from our preliminary experiment. The concentration of inhibitor was selected according to our preliminary experiment and according to previous in vitro studies 59,60 . The hepatocytes were maintained in M199 medium (M199, Gibco/ Invitrogen, UK) containing 1 mmol/L L-glutamine, 5% (v/v) foetal bovine serum (FBS, Gibco/Invitrogen, UK), penicillin (100 IU/mL) and streptomycin (100 μ g/mL). Each treatment was performed in quadruplicate.

Analysis of enzymatic activity and Q-PCR.
For assays of the activities of FAS, G6PD, 6PGD, ICDH and ME, the cells were homogenized by sonication in extraction buffer (0.02 M Tris-HCl, 0.25 M sucrose, 2 mM EDTA, 0.1 M sodium fluoride, 0.5 mM phenyl methyl sulphonyl fluoride, and 0.01 M β -mercaptoethanol, pH 7.4). FAS activity was determined by the method of Chakrabarty and Leveille 61 , G6PD activity following the method of Barroso et al. 62 , and 6PGD activity according to the method of Hisar et al. 63 . The ICDH and ME activities were measured according to Pierron et al. 64 . CPT I activity was measured according to a modified protocol from Morash et al. 65 . Briefly, the cells were homogenized by sonication in extraction buffer (250 mM sucrose, 1 mM EDTA, 20 mM HEPES, and 0.5% bovine serum albumin [BSA], pH 7.4). The reaction mixture contained 0.1 mM 5,5′ -dithiobis (2-nitrobenzoic acid) (DTNB), 5 mM L-carnitine and 0.1 mM palmitoyl-CoA. CPT I activity was measured in the forward direction (formation of palmitoylcarnitine) by monitoring the initial rate of CoA-SH release with DTNB at 412 nm. The HK and PEPCK activities were measured according to Polakof et al. 26 . The protein content was measured following the method of Bradford 66 , with BSA used as the standard. One unit of enzyme activity was defined as 1 μ M of substrate converted to product per minute at 28 °C and was expressed as mU mg −1 soluble protein.
Q-PCR for the in vitro experiment was performed using the protocol described above. Among a set of eight housekeeping genes (β -actin, GAPDH, RPL7, 18 S rRNA, HPRT, TBP, TUBA and UBCE), the mRNA expression of TBP and TUBA proved to be the most stable across the experimental conditions in vitro according to geNorm software. Primers are given in Supplementary Table 4.
Statistical analysis. Statistical analyses were performed with SPSS 19.0 software (SPSS, Michigan Avenue, Chicago, IL, USA). The results are presented as the means ± SEM (standard errors of means). Prior to statistical analysis, all data were evaluated for normality using the Kolmogorov-Smirnov test. Bartlett's test was performed for testing the homogeneity of variances. Then, the data were subjected to one-way ANOVA followed by Turkey's multiple range tests. Significant differences were established at p < 0.05.