Hnf4α Is Involved in LC-PUFA Biosynthesis by Up-Regulating Gene Transcription of Elongase in Marine Teleost Siganus canaliculatus

The rabbitfish Siganus canaliculatus is the first marine teleost shown to be able to biosynthesize long-chain polyunsaturated fatty acids (LC-PUFA) from C18 PUFA precursors catalyzed by two fatty acyl desaturases (fad) including Δ4 Fad and Δ6/Δ5 Fad as well as two elongases (Elovl4 and Elovl5). Previously, hepatocyte nuclear factor 4α (Hnf4α) was demonstrated to be predominant in the transcriptional regulation of two fads. To clarify the regulatory mechanisms involved in rabbitfish lipogenesis, the present study focused on the regulatory role of Hnf4α to elovl5 expression and LC-PUFA biosynthesis. Bioinformatics analysis predicted two potential Hnf4α elements in elovl5 promoter, one binding site was confirmed to interact with Hnf4α by gel shift assays. Moreover, overexpression of hnf4α caused a remarkable increase both in elovl5 promoter activity and mRNA contents, while knock-down of hnf4α in S. canaliculatus hepatocyte line (SCHL) resulted in a significant decrease of elovl5 gene expression. Meanwhile, hnf4α overexpression enhanced LC-PUFA biosynthesis in SCHL cell, and intraperitoneal injection to rabbitfish juveniles with Hnf4α agonists (Alverine and Benfluorex) increased the expression of hnf4α, elvol5 and Δ4 fad, coupled with an increased proportion of total LC-PUFA in liver. The results demonstrated that Hnf4α is involved in LC-PUFA biosynthesis by up-regulating the transcription of the elovl5 gene in rabbitfish, which is the first report of Hnf4α as a transcription factor of the elovl5 gene in vertebrates.


Introduction
Long-chain polyunsaturated fatty acids (LC-PUFA) such as eicosapentaenoic acid (EPA; 20:5n-3), arachidonic acid (ARA; 20:4n-6) and docosahexaenoic acid (DHA; 22:6n-3) are cell membranes components, precursors of lipogenesis [1]. As signal molecules involved in metabolic pathways, LC-PUFAs are also very important to human health, which could respond to immune and inflammatory stimulation [2][3][4]. Fish are the major sources of n-3 LC-PUFAs from the human diet [5], while marine teleost mainly relies on feed rich in fish oil (FO) or fish meal (FM) to meet their requirement for LC-PUFA such as ARA, EPA and DHA. With an increase of global fish consumption, the finite ocean

Two Hnf4α Binding Sites Were Predicted in Rabbitfish Elovl5 Promoter
Using the bioinformatics software TRANSFAC ® and TF binding ® , two Hnf4α binding sites (+70~+81 and −84~−74) were predicted in the promoter region of D3 (−837 bp to +89 bp) of rabbitfish elovl5 (Table 1 and Figure 2B). Based on these results, we speculated that Hnf4α might be a potential factor that affects the activity of the rabbitfish elovl5 promoter. The position of each element is numbered relative to the presumed TSS. The bases underlined are the mutation sites for site-directed mutant, "×" denotes deletion.

Overexpression of Hnf4α Increased Elvol5 Promoter Activity
To explore the regulatory role of rabbitfish Hnf4α in elvol5 gene transcription, the effect of rabbitfish hnf4α overexpression on elvol5 promoter activity was determined. The recombinant plasmid pcDNA3.1-Hnf4α and progressive deletion mutants of elvol5 promoter or site-directed mutants were co-transfected into HEK 293T cells. The promoter activity of each progressive deletion mutant significantly increased with hnf4α over-expression, while the negative control pGL4.10 showed no response to Hnf4α treatment (p < 0.05) ( Figure 3). As to the site-directed mutants, D3-M1 showed no response to hnf4α overexpression compared with the wild type D3, while the promoter activity of D3-M2 was decreased after hnf4α overexpression ( Figure 3). region, which is relative to the transcription start site (TSS, +1). The bases with black background refer to Hnf4α-1, TSS, and Hnf4α-2, respectively. The bases underlined are downstream sequence of TSS.

Two Hnf4α Binding Sites Were Predicted in Rabbitfish Elovl5 Promoter
Using the bioinformatics software TRANSFAC ® and TF binding ® , two Hnf4α binding sites (+70 ~ +81 and −84 ~ −74) were predicted in the promoter region of D3 (−837 bp to +89 bp) of rabbitfish elovl5 (Table 1 and Figure 2B). Based on these results, we speculated that Hnf4α might be a potential factor that affects the activity of the rabbitfish elovl5 promoter. The position of each element is numbered relative to the presumed TSS. The bases underlined are the mutation sites for site-directed mutant, "×" denotes deletion.

Overexpression of Hnf4α Increased Elvol5 Promoter Activity
To explore the regulatory role of rabbitfish Hnf4α in elvol5 gene transcription, the effect of rabbitfish hnf4α overexpression on elvol5 promoter activity was determined. The recombinant plasmid pcDNA3.1-Hnf4α and progressive deletion mutants of elvol5 promoter or site-directed mutants were co-transfected into HEK 293T cells. The promoter activity of each progressive deletion mutant significantly increased with hnf4α over-expression, while the negative control pGL4.10 showed no response to Hnf4α treatment (p < 0.05) ( Figure 3). As to the site-directed mutants, D3-M1 showed no response to hnf4α overexpression compared with the wild type D3, while the promoter activity of D3-M2 was decreased after hnf4α overexpression ( Figure 3). . Effects of S. canaliculatus hnf4α over-expression on activity of elovl5 promoter deletion mutants and site-directed mutation in HEK 293T cells. The elovl5 promoter deletion mutants, sitedirected mutants and negative control were co-transfected with the overexpression plasmid pcDNA3.1-Hnf4α, while the control group was transfected with the empty vector pcDNA3.1. The negative control pGL4.10 is an empty vector with no promoter sequence upstream the reporter gene. Each plasmid complex was transfected in triplicate in three independent experiments. Significant differences compared with the corresponding control group were analyzed using Student's t-test; with * denoting p < 0.05. Figure 3. Effects of S. canaliculatus hnf4α over-expression on activity of elovl5 promoter deletion mutants and site-directed mutation in HEK 293T cells. The elovl5 promoter deletion mutants, site-directed mutants and negative control were co-transfected with the overexpression plasmid pcDNA3.1-Hnf4α, while the control group was transfected with the empty vector pcDNA3.1. The negative control pGL4.10 is an empty vector with no promoter sequence upstream the reporter gene. Each plasmid complex was transfected in triplicate in three independent experiments. Significant differences compared with the corresponding control group were analyzed using Student's t-test; with * denoting p < 0.05.

Electrophoretic Mobility Shift Assay
To further confirm whether Hnf4α in rabbitfish liver could bind to the promoter of elovl5, EMSA (electrophoresis mobility shift assay) was performed with rabbitfish hepatocytes cytoplasmic and nuclear proteins. The results indicated that the hepatocytes nuclear proteins bound to biotin-labeled probe and retarded their mobility (Figure 4 lane 2). When the assays were further performed using unlabeled probe (Figure 4 lane 3) as a specific competitor, the specific shift was abolished by excess unlabeled probe, which indicated specific binding of hepatocytes nuclear proteins to the probe ( Figure 4). The specific binding of Hnf4α to elovl5 was also confirmed by its super shift after the addition of Hnf4α antibody (Figure 4 lane 4). These results further suggested that Hnf4α specifically bound to the predicted binding sites (D3-M1) in the upstream of elovl5 and thereby might regulate elovl5 transcription.

Electrophoretic Mobility Shift Assay
To further confirm whether Hnf4α in rabbitfish liver could bind to the promoter of elovl5, EMSA (electrophoresis mobility shift assay) was performed with rabbitfish hepatocytes cytoplasmic and nuclear proteins. The results indicated that the hepatocytes nuclear proteins bound to biotin-labeled probe and retarded their mobility (Figure 4 lane 2). When the assays were further performed using unlabeled probe (Figure 4 lane 3) as a specific competitor, the specific shift was abolished by excess unlabeled probe, which indicated specific binding of hepatocytes nuclear proteins to the probe ( Figure 4). The specific binding of Hnf4α to elovl5 was also confirmed by its super shift after the addition of Hnf4α antibody (Figure 4 lane 4). These results further suggested that Hnf4α specifically bound to the predicted binding sites (D3-M1) in the upstream of elovl5 and thereby might regulate elovl5 transcription. . Band A is gel shift of DNA-protein complexes. Band B is the free probe. Band C is supershift of DNA-protein-antibody complexes. "+" means that the corresponding material in the row has been added, and "−" means that the material is not added.

Overexpression of Hnf4α Enhanced Elvol5 Gene Expression and LC-PUFA Biosynthesis in SCHL Cells
To further confirm the regulatory role of rabbitfish Hnf4α in elvol5 gene transcription, rabbitfish hnf4α mRNA synthesized in vitro was transfected into SCHL cells. The mRNA expression level of hnf4α and elvol5 was determined by qPCR, with the results showing that the mRNA levels of hnf4α and elovl5 significantly increased after hnf4α mRNA transfection ( Figure 5). We therefore analyzed the effect of hnf4α overexpression on fatty acids composition in the SCHL cells. The results from this analysis showed that the levels of ARA, EPA and DHA were significantly up-regulated (Table 2), and the conversion rates of 18:2n-6 to 20:2n-6 and 18:3n-3 to 20:3n-3, the two pathways catalyzed by Elovl5, were enhanced after hnf4α overexpressing ( Figure 6). . Band A is gel shift of DNA-protein complexes. Band B is the free probe. Band C is supershift of DNA-protein-antibody complexes. "+" means that the corresponding material in the row has been added, and "−" means that the material is not added.

Overexpression of Hnf4α Enhanced Elvol5 Gene Expression and LC-PUFA Biosynthesis in SCHL Cells
To further confirm the regulatory role of rabbitfish Hnf4α in elvol5 gene transcription, rabbitfish hnf4α mRNA synthesized in vitro was transfected into SCHL cells. The mRNA expression level of hnf4α and elvol5 was determined by qPCR, with the results showing that the mRNA levels of hnf4α and elovl5 significantly increased after hnf4α mRNA transfection ( Figure 5). We therefore analyzed the effect of hnf4α overexpression on fatty acids composition in the SCHL cells. The results from this analysis showed that the levels of ARA, EPA and DHA were significantly up-regulated (Table 2), and the conversion rates of 18:2n-6 to 20:2n-6 and 18:3n-3 to 20:3n-3, the two pathways catalyzed by Elovl5, were enhanced after hnf4α overexpressing ( Figure 6). Significant difference compared with the control group was analyzed using Student's t-test; with * denoting p < 0.05. Figure 6. Fatty acid conversion rates in SCHL cells transfected with hnf4α mRNA compared with control. White columns represent the control groups while the black columns are the experiment groups transfected with hnf4α mRNA. Results are means ± SEM (n = 3). Significant differences compared with the control group were analyzed using Student's t-test; with * denoting p < 0.05. mRNA or control. The Relative expression of hnf4α and elovl5 were analyzed by qPCR and normalized to 18S rRNA expression using the by 2 −∆∆Ct method. Results are means ± SEM (n = 3). Significant difference compared with the control group was analyzed using Student's t-test; with * denoting p < 0.05. Results are showed as means ± SEM (n = 3). Values in each row with different superscripts indicate significant difference (analyzed by ANOVA followed by paired t-test; p < 0.05). Figure 5. Q-PCR analysis of hnf4α and elovl5 gene expression level in SCHL cells transfected with hnf4α mRNA or control. The Relative expression of hnf4α and elovl5 were analyzed by qPCR and normalized to 18S rRNA expression using the by 2 −ΔΔCt method. Results are means ± SEM (n = 3). Significant difference compared with the control group was analyzed using Student's t-test; with * denoting p < 0.05. Figure 6. Fatty acid conversion rates in SCHL cells transfected with hnf4α mRNA compared with control. White columns represent the control groups while the black columns are the experiment groups transfected with hnf4α mRNA. Results are means ± SEM (n = 3). Significant differences compared with the control group were analyzed using Student's t-test; with * denoting p < 0.05. Figure 6. Fatty acid conversion rates in SCHL cells transfected with hnf4α mRNA compared with control. White columns represent the control groups while the black columns are the experiment groups transfected with hnf4α mRNA. Results are means ± SEM (n = 3). Significant differences compared with the control group were analyzed using Student's t-test; with * denoting p < 0.05.

Knockdown of Hnf4α Expression Reduced Elvol5 Expression in SCHL Cells
RNA interference assay was carried out so as to further investigate the regulatory role of Hnf4α on elvol5 gene expression in SHCL cells. First, the efficiency of the siRNA to silence hnf4α was evaluated by analyzing the mRNA levels of hnf4α using qPCR. The results indicated that the mRNA level of hnf4α was significantly down-regulated by about 51.1% at 24 h after hnf4α siRNA transfection (Figure 7). Meanwhile, the mRNA expression level of elovl5 decreased by about 44.5% compared with negative control group (NC) (Figure 7).

Knockdown of Hnf4α Expression Reduced Elvol5 Expression in SCHL Cells
RNA interference assay was carried out so as to further investigate the regulatory role of Hnf4α on elvol5 gene expression in SHCL cells. First, the efficiency of the siRNA to silence hnf4α was evaluated by analyzing the mRNA levels of hnf4α using qPCR. The results indicated that the mRNA level of hnf4α was significantly down-regulated by about 51.1% at 24 h after hnf4α siRNA transfection ( Figure 7). Meanwhile, the mRNA expression level of elovl5 decreased by about 44.5% compared with negative control group (NC) (Figure 7).  . The relative expression of hnf4α and elovl5 were analyzed by qPCR and normalized to 18S rRNA expression using the by 2 −∆∆Ct method. Results are means ± SEM (n = 3). Significant differences compared with the control group were analyzed using Student's t-test; with * denoting p < 0.05.

Intraperitoneal Injection of Hnf4α Agonists Increased Elvol5 and ∆4 Fad Expression and Fatty Acid Composition in Rabbitfish Liver
To further identify the regulatory role of hnf4α on rabbitfish LC-PUFA biosynthesis in vivo, Hnf4α agonists (Alverine and Benfluorex) were injected into the enterocoelia of juvenile rabbitfish. Real time qPCR results in liver samples showed that the gene expression levels of hnf4α, elovl5 and ∆4 fad significantly increased in Alverine and Benfluorex treatment groups compared to the negative control ( Figure 8). Moreover, the results of fatty acid composition in liver showed that there was a higher content of DHA and total HUFA in Alverine injection group compared with the negative control (Table 3), and the content of EPA in Benfluorex treatment group was also higher than that in negative control.

Intraperitoneal Injection of Hnf4α Agonists Increased Elvol5 and Δ4 Fad Expression and Fatty Acid Composition in Rabbitfish Liver
To further identify the regulatory role of hnf4α on rabbitfish LC-PUFA biosynthesis in vivo, Hnf4α agonists (Alverine and Benfluorex) were injected into the enterocoelia of juvenile rabbitfish. Real time qPCR results in liver samples showed that the gene expression levels of hnf4α, elovl5 and Δ4 fad significantly increased in Alverine and Benfluorex treatment groups compared to the negative control ( Figure 8). Moreover, the results of fatty acid composition in liver showed that there was a higher content of DHA and total HUFA in Alverine injection group compared with the negative control (Table 3), and the content of EPA in Benfluorex treatment group was also higher than that in negative control. Figure 8. Q-PCR analysis of hnf4α, elovl5 and ∆4 fad gene expression level in liver of juvenile rabbitfish injected with Hnf4α agonists (Alverine and Benfluorex) or control. The relative expression of hnf4α, elovl5 and ∆4 fad was analyzed by qPCR and normalized to 18S rRNA expression using the by 2 −ΔΔCt method. Control 1 was injected with 0.9% NaCl while control 2 was injected with 2.5% DMSO. Results are means ± SEM (n = 6). Significant differences were analyzed by ANOVA followed by Tukey's multiple comparison test; with * denoting p < 0.05. Figure 8. Q-PCR analysis of hnf4α, elovl5 and ∆4 fad gene expression level in liver of juvenile rabbitfish injected with Hnf4α agonists (Alverine and Benfluorex) or control. The relative expression of hnf4α, elovl5 and ∆4 fad was analyzed by qPCR and normalized to 18S rRNA expression using the by 2 −∆∆Ct method. Control 1 was injected with 0.9% NaCl while control 2 was injected with 2.5% DMSO. Results are means ± SEM (n = 6). Significant differences were analyzed by ANOVA followed by Tukey's multiple comparison test; with * denoting p < 0.05.

Discussions
To gain insight into the regulatory mechanisms of hepatocyte nuclear factor 4α (Hnf4α) in LC-PUFA biosynthesis of marine teleosts, previous studies we conducted in rabbitfish showed that Hnf4α targeted at ∆4 fad and ∆6/∆5 fad promoter directly and upregulated their gene expression [28][29][30]. Above all, such TF has been considered as one vital regulator involved in LC-PUFA biosynthesis. However, the influence of Hnf4α to of elovls gene transcription has not been studied and whether Hnf4α could directly regulate elovl5 expression was still unknown. Therefore, the present study focused on the regulatory role of Hnf4α in elovl5 gene transcription and LC-PUFA biosynthesis of rabbitfish S. canaliculatus.
Hnf4α is an important regulator of the key enzymatic genes involved in vertebrates LC-PUFA biosynthesis. As a vital TF involved in the regulation of lipid and cholesterol metabolism, HNF4α has been reported to activate the following targets: Apolipoprotein C-III (ApoCIII), Cholesterol 7α (Cyp7α) Hydroxylase [31,32], fatty acid synthase (FAS) [33], stearoyl-CoA desaturase (SCD) and ∆4 Fad [28]. Recently, one study identified a fragment of HNF4α binding to the core promoter of rabbitfish ∆6/∆5 fad, suggesting its potential modulation to this new target [29], while another study demonstrated that Hnf4α is involved in the transcriptional regulation of LC-PUFA biosynthesis by targeting ∆4 fad and ∆6/∆5 fads in rabbitfish [30]. elovl5 is another key enzymatic gene in LC-PUFA biosynthesis, which has attracted many researchers with lots of studies focused on its transcriptional mechanism in humans, mice and salmon. At present, SREBPs have been demonstrated as the major regulator in elovl5 transcription, and LXR might be another potential TF involved in such a process directly in salmon, while in mammals it was an indirect TF that influenced elovl5 expression [22,34]. The present study discovered a conservative element unit of NF-Y and SRE in elovl5 promoter, which was similar to the previous reports mentioned above, suggesting that SREBPs might be the main regulator in rabbitfish elovl5 transcription. Additionally, we have also identified the positive effect of Hnf4α on rabbitfish elovl5 expression through site-directed mutation, electrophoretic mobility shift assay, hnf4α overexpression, hnf4α knock-down by RNAi, as well as by drug treatment. The results were remarkable as this novel discovery expanded the regulatory range of Hnf4α target genes in lipogenesis. Thus, with the addition of elovl5, Hnf4α has now been demonstrated to positively regulate the complete enzymatic pathway in LC-PUFA biosynthesis (FAS, Fad, Elovl), suggesting its prominent role in lipid metabolism.
HNF4α could improve LC-PUFA biosynthesis in SCHL cells by increasing elovl5 gene expression. In general, HNF4α acted as a positive regulator in lipogenesis, as its feed-back regulation in energy metabolism maintains physiological homeostasis in organisms [35]. As a ligand-dependent TF, long chain fatty acids such as ALA, EPA and DHA are endogenous ligands for HNF4α, so binding to this nuclear receptor suppressed its activation to the target genes [36]. Some chemical ligands such as those used in the present study, i.e., Alverine and Benfluorex, could increase hnf4α gene expression and activate this TF as a ligand [37]. In relation to recent research on the regulation of elovl5 in the teleost Larimichthys crocea and Epinephelus coioides, feed-back regulation in LC-PUFA biosynthesis from the process of dietary lipid to fish metabolism and LC-PUFA (EPA and DHA) to hepatocytes was carried out through another nuclear receptor LXRα and its downstream target SREBP-1 [21,23]. Previous research in Salmon and SHK-1 cell lines also supported this regulation pattern at both the nutritional and cellular level [22,34]. The present study tested the feedback regulation model of Hnf4α in rabbitfish from the physiological level with chemical ligands and at the cellular level with fatty acid substrate conversion. The results indicated that Alverine and Benfluorex could activate hnf4α and elovl5 gene expression and then improve LC-PUFA biosynthesis in rabbitfish, while hnf4α mRNA overexpression revealed that hnf4α overexpression could improve LC-PUFA biosynthesis in SCHL cells. This observation further demonstrates the important role of hnf4α in rabbitfish LC-PUFA biosynthesis, which is a completely novel mechanism in vertebrate lipogenesis.
In conclusion, the elovl5 promoter of S. canaliculatus was cloned and characterized, moreover Hnf4α was demonstrated to be a TF of elovl5 in vertebrate for the first time, both discoveries have suggested a new regulatory mechanism of LC-PUFA biosynthesis in teleost.

Compliance with Ethical Standards
In the present study, we followed the requirement of the National Institutes of Health guide (NIH Publications No. 8023, revised 1978)

Cloning of 5 Flanking Sequence of Rabbitfish Elovl5
Genomic DNA was extracted from rabbitfish muscle with the proteinase K and phenol method as previously noted [38]. The Genome Walker TM Universal Kit (TaKaRa Bio, Tokyo, Japan) was used for elovl5 promotor cloning according to the manufacturer's instructions. Nested PCR was performed with the outer adaptor primer AP1 in the kit and a specific antisense primer E5UA0, while the secondary nested PCR reaction was carried out with the nested adaptor primer AP2 and specific antisense primer E5UA1 ( Table 4). The primers E5UA0 and E5UA1 were designed based on the mRNA sequence of elovl5 (GenBank: GU597350.1) [27]. After two rounds of PCR, the PCR product of the upstream sequence was recovered and isolated by gel extraction, then inserted into the pMD18-T Vector (TaKaRa Bio, Tokyo, Japan), and sequenced (Sangon Biotech Co., Ltd., Shanghai, China). The sequencing results revealed the presence of first non-coding exons in the 5 untranslated region (UTR) of elovl5, indicating that the PCR product was indeed the 5 flanking sequence of elovl5.

Bioinformatics Analysis
The conserved elements of NF-Y and SRE in rabbitfish S. canaliculatus elovl5 promoter were identified by alignment with the corresponding elovl5 promoter sequence from Salmo_salar (GU238431.1 and GU324549.1), Danio_rerio (NC_007124.7), Mus_musculus (NC_000075.6), and Homo_sapiens (NG_034263.1). Online software including JASPAR ® , TRANSFAC ® and TF Binding ® were used to analyze the promoter region of elovl5 for potential TF binding sites. The potential TF elements were obtained from the predicted results analyzed by the software.

Identification of Elovl5 Core Promoter through Progressive Deletion Mutation
To identify the core promoter region within the cloned 5 flanking sequence of rabbitfish elovl5, the candidate promoter was progressively deleted. PCR reaction was carried out using 2× pfu  (Figure 1). After construction, the vector consisted of insert fragments (D0, D1, D2, D3 and D4) and pGL4.10, and high Pure Plasmid Isolation Kit (Roche, Mannheim, Germany) was used to isolate the construct. Later, the transfection assay in human embryonic kidney (HEK293T) cells (Chinese Type Culture Collection, Shanghai, China) was carried out.

Functional Identification of the Two-Candidate Hnf4α Elements
To determine the potential effect of the predicted Hnf4α binding sites on promoter activity, recombinant plasmids with site-directed mutation of Hnf4α elements in the promoter was constructed. For the rabbitfish elovl5 promoter, deletion mutant D3 containing core promoter region was treated as wild-type and site-directed mutants were produced from this using the Muta-direct TM site-directed mutagenesis kit (SBS Genetech, Shanghai, China) according to the manufacturer's instruction. The strategy of site directed mutation is shown in Table 1 and the primers are shown in Table 4. The site-directed mutation plasmids from D3 are designated D3-M1 and D3-M2. The over-expression plasmid pcDNA3.1-Hnf4α contains the whole Open reading frame (ORF) of rabbitfish Hnf4α. All the recombinant plasmids were isolated with High Pure Plasmid Isolation Kit (Roche, Swiss) for use in transfection. HEK 293T cells were seeded onto 96-well plates at a density of 4 × 10 4 per well in a volume of 100 µL per well with High Glucose Dulbecco's Modified Eagle Medium (DMEM) (Gluta MAX) (Gibco, Thermo Fisher, Carlsbad, CA, USA) and 10% fetal bovine serum (Gibco, Life Technologies, Carlsbad, CA, USA), then cultured at 37 • C. Transfection was carried out with mutants of elvol5 promoter including D0, D1, D2, D3, D4, D3-M1, D3-M2 (100 ng/well), pGL4.75 (0.02 ng/well) and pcDNA3.1-Hnf4α (50 ng/well), with pGL4.10 used as vector control, following the method described previously [28]. Transfections were done in triplicates and three independent experiments. Cell culture medium was replaced with 75 µL DMEM + 10% FBS at 24 h after transfection. Luciferase assays were performed at 48 h after transfection with the Dual-Glo TM luciferase assay system (Promega, Madison, WI, USA), and luminescence was detected by a microplate reader (InfiniteM200 Pro, Tecan, Switzerland). The method for promoter activity calculation was the same as previously noted [28].

Electrophoretic Mobility Shift Assay (EMSA)
To confirm the binding of Hnf4α to the promoter of rabbitfish elovl5, nuclear and cytoplasmic proteins were extracted from rabbitfish hepatocytes with the Beyotime Nuclear Extract Kit (Beyotime Institute of Biotechnology, Haimen, China) and quantified by Modified BCA Protein Assay Kit (Sangon, Shanghai, China). The 29 bp 5 end biotin-labeled probe covering the predicted Hnf4α elements was designed and incubated with the proteins to determine whether Hnf4α interacted with the promoter of elovl5. Both the labeled and unlabeled probes in the experiment were obtained from Shanghai Sangon Biotech Co., Ltd., while the EMSA reaction system was performed with the Beyotime Chemiluminescent EMSA Kit (Beyotime Institute of Biotechnology, Haimen, China) according to the manufacturer's instructions. For the super shift assay, 1 µL antibody (Abcam, Cambridge, MA, USA) of Hnf4α was pre-incubated with nuclear or cytoplasmic proteins for 30 min at 0-4 • C. Samples obtained after the binding reaction were subjected to a 4% non-denaturing polyacrylamide gel electrophoresis and transferred onto a nylon membrane. The 5 end biotin-labeled probe was detected using a streptavidin-horseradish peroxidase conjugate and a chemiluminescent substrate. The signal was then detected by autoradiography with X-OMAT BT X-ray film (Kodak, Rochester, MN, USA).

In Vitro mRNA Transcription of Rabbitfish Hnf4α
The in vitro transcription of hnf4α mRNA was performed on a linearized DNA template containing T7 promoter and rabbitfish hnf4α cDNA sequence using the mMESSAGE mMACHINE ® T7 Ultra Kit (Ambion, Thermo Fisher, Carlsbad, CA, USA). The pcDNA3.1-Hnf4α plasmid previously constructed in our laboratory was used to produce the linearized DNA template [28]. Finally, the product containing the hnf4α mRNA was purified with MEGAclear TM Kit (Ambion, Austin, TX, USA) and used immediately or stored at −80 • C for later use.

Transfection of Rabbitfish Hnf4α mRNA and siRNA into SCHL Cells
The rabbitfish S. canaliculatus hepatocytes cell line (SCHL) established by our group [39] were seeded onto 6-well plates (Eppendorf, Hamburg, Germany) at a density of 1.2 × 10 6 per well in a volume of 2 mL Dulbecco's modified Eagle's medium (DMEM)-F12 medium (Gibco, Life Technologies, Carlsbad, CA, USA) supplemented with 10% foetal bovine serum (FBS) (Gibco, Life Technologies, Carlsbad, CA, USA) and 0.5% rainbow trout Oncorhychus mykiss serum (Caisson Labs; www.caissonlabs. com), and maintained at 28 • C. At 80% confluence, cells were transfected with 5 µg/well hnf4α mRNA using Lipofectamine TM Messenger-MAX TM Reagent (Thermo Fisher, Carlsbad, CA, USA). Medium was removed at 48 h post transfection, cells washed carefully with 1 mL PBS, and then total RNA extracted using 1 mL Trizol (Invitrogen, Carlsbad, CA, USA). For lipid extraction and fatty acids content detection, the SCHL cells were seeded onto six 100 mm dishes (Eppendorf, Hamburg, Germany) at a density of 7 × 10 6 cells per well in a volume of 8 mL (DMEM/F12 + 10% FBS + 0.5% rainbow trout Oncorhychus mykiss serum) and maintained at 28 • C. At about 24 h or 70% confluence, cells were then transfected with 6 µg/per dish hnf4α mRNA using Lipofectamine TM Messenger-MAX TM Reagent (Invitrogen, Carlsbad, CA, USA). Transfections were done in triplicates and three independent experiments. At 24 h post transfection, medium was replaced with 8 mL DMEM/F12 + 10% FBS + 0.5% rainbow trout Oncorhychus mykiss serum, and at 72 h post transfection, cells were treated with 1 mL Trypsin-EDTA (Invitrogen, Carlsbad, CA, USA), centrifuged at 1500× g for 2 min, and then fatty acids were extracted from the precipitate as described in Section 4.11.
In order to knockdown the expression of hnf4α in SCHL cells, 21-nucleotide small interfering RNA duplexes (siRNA) targeting hnf4α and negative control siRNA (Table 4) were chemically synthesized by Gene-Pharma Biotechnology Company (Suzhou, China). The siRNAs were diluted with DEPC-water to a final concentration of 125 mg/mL. SCHL cells were seeded onto 12-well plates (Eppendorf, Hamburg, Germany) at a density of 5 × 10 5 cells per well in a volume of 1 mL medium (DMEM/F12 + 10% FBS + 0.5% rainbow trout Oncorhychus mykiss serum) maintained at 28 • C, and after 24 h or 60% confluence, cells were then transfected with 40 pmol siRNA per well using Lipofectamine TM 2000 Reagent (Invitrogen, Carlsbad, CA, USA). At 24 h post transfection, media was removed, cells washed carefully with 1 mL PBS, and then total RNA extracted with 1 mL Trizol (Invitrogen, Carlsbad, CA, USA).