Introduction

In eukaryotic cells, the front-end desaturases (Fads) catalyse the introduction of a double bond at fixed number of carbons from the carboxyl group. A typical Fads possesses characteristic features such as three histidine-rich boxes, transmembrane regions, and an N-terminal cytochrome b5 domain containing the heme-binding motif HPGG. Long-chain polyunsaturated fatty acids (LC-PUFA) such as arachidonic acid (ARA, 20:4n-6), eicosapentaenoic acid (EPA, 20:5n-3) and docosahexaenoic acid (DHA, 22:6n-3) are generated from shorter polyunsaturated fatty acids (PUFA) through the actions of Fads and elongases (Elovl). LC-PUFA are crucial for maintenance of cellular membrane integrity, as precursors for eicosanoids, regulation of gene expression and signal transduction pathways. In aquaculture, the interest to decipher the activities of the LC-PUFA biosynthesis enzymes and transcription factors involve in LC-PUFA biosynthesis is driven by the pressing need to improve performance of vegetable oils in aquafeeds1. These oils are lacking in LC-PUFA but rich in C18 PUFA. Understandably, a greater understanding of distinctive bioconversion capacities of C18 PUFA to LC-PUFA in different farmed species could theoretically improve the strategy of employing vegetable oil as feeds2.

Among the vertebrates, numerous studies have reported the molecular cloning and characterisation of Fads from different fish species1. Fads from mammals are principally mono-functional, with Δ5 and Δ6 desaturations being carried out by two separate genes, Fads1 and Fads2, respectively3. In fish, fads1 have so far been isolated from only one species, a basal gnathostome, leading to the hypothesis that a complete loss of fads1 in teleosts have occurred following gnathostome radiation4. Isolation and characterisation of teleost fads2 to date have reported a broad range of substrate specificities, with Δ4, Δ5, Δ6 and Δ8 desaturation capacities reported. In addition, both mono-functional and bi-functional fads2 have been isolated from a myriad of teleost species5,6,7. As compared to freshwater species, marine teleosts seemed to display a lesser capacity for bioconversion of C18 PUFA to LC-PUFA, due to the loss of crucial fads from its genome or low activities of fads2 or elovl8. This was postulated to be a result of richer LC-PUFA content in the marine environment, causing diminished capacity for LC-PUFA biosynthesis. Subsequent works however advocated the influence of feeding niche on LC-PUFA biosynthesis, when a catalogue of fads2 and elovl were isolated from marine fish species which possess a diet with limited DHA intake6,9.

Despite the multiple interests in molecular characterisation of fads in teleost, there is a paucity of understanding on its regulation at the transcriptional level. It is conceivable that differences in regulatory activities by dissimilar transcription factors responding to different cues including nutritional status is responsible for the different LC-PUFA biosynthesis capacity in various fish species. These transcription factors themselves are subjected to regulation by the nutritional status of the animal. Therefore, efficient use of vegetable oils in aquafeeds will benefit from the ability to decipher the link between the overall lipid metabolic pathways modification caused by the diet and the subsequent influence on lipid homeostasis genes at the transcriptional level. To date, promoter sequences from teleost fads2 have been isolated from Atlantic salmon10, Atlantic cod10, rainbow trout, European sea bass11, Japanese seabass12, large yellow croaker13 and rabbit fish14. Firstly, these works collectively showed the presence of binding sites for transcription factors known to play a role in mammalian cholesterol and lipogenic synthesis pathways such as sterol regulatory element-binding protein (Srebp), nuclear factor Y (NF-Y) and ubiquitous transcription factor Sp1 in the fads2 promoter10,11,12,13,14. Secondly, these studies suggested an interplay between different transcription factors to mediate the regulation of fads2 transcription.

The zebrafish has gained reputation as a useful model in understanding the role of lipids and fatty acids during development15,16. Largely driven by the existence of multiple homologous genes involved in conserved lipid and lipoprotein metabolism network, zebrafish have been used to gain insight on lipid adsorption, atherosclerosis, fatty liver disease and obesity15,17. Zebrafish tissues are reported to contain significantly higher concentrations of PUFA than equivalent mammalian tissues18,19. Molecular cloning, functional characterisation and expression profile of fads2 and several elovl genes during embryogenesis have been reported in zebrafish5,20,21,22,23. However, our knowledge on the function of fads2/elovl and LC-PUFA during development is still fragmentary. Given how important transcriptional regulation in influencing the activities of fads2 may be and how little is known about the regulators and cellular conditions for expression, it is opportune to use zebrafish as a model to fill these gaps.

In this present study, we described the cloning and characterisation of the zebrafish fads2 promoter. This includes identification and functional characterisation of pivotal regulatory elements responsible for driving transcriptional activities using luciferase reporter gene in a zebrafish cell line. In addition, we also carried out an in vivo characterisation of the zebrafish fads2 promoter using a transient approach to track its expression in developing zebrafish embryos.

Results

Cloning and characterisation of the zebrafish fads2 promoter

We isolated a putative promoter region of 1754 bp, comprising 19 bp of the 5′-UTR of zebrafish Δ5/ Δ6 fads25 and 1735 bp of the upstream region. For positioning purpose, +1 was given to the putative transcription start site (TSS) derived from comparison of the 5′-end of full-length cDNA (GenBank accession number: NM_131645.2) and DNA sequences of the fads2 gene. Within this fads2 promoter region, a classic transcription initiation element, the TATA box was identified at the location of −26 bp (Supplementary data 1).

Zebrafish Srebp upregulates fads2 promoter activity

Co-transfection of ZFL cells with the fads2-1735 promoter-luciferase reporter plasmid and expression plasmid containing either the nuclear form of zebrafish Srebp1 (nSrebp1) or Srebp2 (nSrebp2) proteins, respectively, resulted in significantly higher reporter activities than cells transfected with empty pcDNA3.1 expression plasmid (Fig. 1). Among the two isoforms, the zebrafish nSrebp2 protein resulted in higher luciferase reading. This signified the importance of Srebp proteins in driving the expression of zebrafish fads2, and subsequent luciferase reporter transactivation assays were performed with the overexpression of zebrafish nSrebp2 protein.

Figure 1
figure 1

Effects of nSrebp proteins on the zebrafish fads2 promoter activation. ZFL cells were co-transfected with nSrebp1 or nSrebp2 expression plasmid (empty pcDNA3.1 expression plasmid as control), and pGL3 promoter-luciferase reporter plasmid fused with the zebrafish fads2 desaturase promoter (−1735/+19 bp). Results represent normalised luciferase activity (to Renilla luciferase activity) of fads2 promoter in the ZFL cells relative to empty pGL3-Basic plasmid. Values are means ± SD (n = 3); P < 0.05 by one-way ANOVA.

Full size image

Identification of cis-regulatory elements within the −214/−67 bp region of zebrafish fads2 promoter

To identify the presence of any potential cis-regulatory element, the nucleotide sequence within the −214/−67 bp region of zebrafish fads2 promoter was aligned and compared to corresponding sequences from rainbow trout13 (GenBank accession number: KT781408.1), Atlantic salmon10 (GenBank accession number: AY736067.2), European seabass11 (GenBank accession number: FP671139.1), large yellow croaker13 (GenBank accession number: KT781409.1), Japanese seabass12 (GenBank accession number: KT781410.1), Atlantic cod10 (GenBank accession number: FJ859898.1), human24 (GenBank accession number: AP002380.3) and mouse (GenBank accession number: AC135670.3). The alignment disclosed the presence of several conserved putative transcription factor binding sites, namely two inverted CCAAT boxes, an inverted ubiquitous transcription factor SP1 site and a sterol regulatory element (SRE) within the cloned promoter region (Fig. 2). The result also showed that the CCAAT boxes were highly conserved across the mammalian and fish species while the SRE (CTCGAATGATC) found in the zebrafish fads2 promoter was identical to those in all the teleost species but differed from the mammalian SRE (CTCTGCTGATC) by three consecutive nucleotides. The SP1 binding site is less conserved amongst the mammalian and fish species, with two SP1 binding site in the mammalian Fads2 promoters, while teleost fads2 promoters typically contain either a single SP1 site or none.

Figure 2
figure 2

Conserved cis-regulatory elements within the core promoter regions of the fads2 genes from zebrafish and other vertebrate counterparts. Sequence alignment was performed with web-based program, MAFFT (version 7). Numbers indicating the sequence positions are relative to TSS (+1), based on the cDNA sequence information deposited in the GenBank (Danio rerio: CU694371.13; Oncorhynchus mykiss: KT781408.1; Salmo salar: AY736067.2; Dicentrarchus labrax: FP671139.1; Larimichthys crocea: KT781409.1; Lateolabrax japonycus: KT781410.1; Gadus morhua: FJ859898.1; Mus musculus: AC135670.3 and Homo sapiens: AP002380.3). Conserved regions for the SP1 binding site, CCAAT boxes and SRE consensus sequence are highlighted in grey, while identical nucleotides are indicated with asterisks.

Full size image

−161/−67 bp region is critical for zebrafish fads2 promoter activation

We sought to determine a minimal fads2 promoter fragment capable of activating transcription by fusing a series of 5′–deleted promoter fragments to the luciferase reporter gene. Upon normalisation to values obtained with a promoter-less pGL3.1 plasmid, the intact fads2 promoter (−1735/+19 bp) drove a luciferase activity of approximately 200-fold (Fig. 3). Deletion of promoter up to 161 bp upstream of the TSS did not greatly diminish transcription activities, notwithstanding some reduction in luciferase readings as compared to the longer fragments. However, further deletion of the sequence within −161/−67 bp almost completely abolished the promoter activity, which indicated a mandatory site within the region of −161 to −67 bp for activation of the fads2 activities. This region contains both CCAAT boxes and the SRE (Supplementary data 2). This also means the removal of the SP1 was not detrimental to the activation of zebrafish fads2 promoter.

Figure 3
figure 3

5′-deletion analysis of the zebrafish fads2 promoter. ZFL cells were transiently co-transfected with the fads2 5′-deletion promoter-luciferase reporter plasmids, nSrebp2 expression plasmid together with Renilla luciferase reference plasmid pRL-SV40. 5′-deletion plasmids were named according to respective position in relation to TSS (+1) and are represented by horizontal line on the left. Non-coding exon is indicated with open boxes. Luciferase activity of fads2 promoter in ZFL cells is expressed as normalised luciferase activity (to Renilla luciferase activity) relative to empty pGL3-Basic plasmid. Values are means ± S.D. (n = 3) (right panel). Groups indicated with different letters are significantly different (Tukey’s test; P < 0.05).

Full size image

CCAAT boxes and SRE mediates fads2 promoter activity

We introduced single or multiple site mutations for the SRE, proximal CCAAT box (CATp) and distal CCAAT box (CATd) to the fads2-244 promoter-luciferase reporter plasmid, followed by the dual luciferase reporter assay in ZFL (Fig. 4). Firstly, results showed that both CATp and CATd could drive expression of promoter, as seen in the fragment with mutated SRE, although the expression was lower than the fragment with all three binding sites intact. In comparison, disrupting both proximal and distal CCAAT bindings sites resulted in total abolishment of luciferase expression, an observation which was recapitulated when all three binding sites were mutated. This implies an obligation for both CCAAT and SRE to optimally drive zebrafish fads2, although the former was still able to activate a reduced measure of transcriptional activities. When SRE was present, CATp seemed to play a more prominent role as compared to CATd in driving promoter activation. A similar pattern was also observed when SRE was disrupted.

Figure 4
figure 4

Mutated CCAAT box and SRE reduce transcriptional activity of the zebrafish fads2 promoter. Single or multiple site mutations (X) of SRE (), CATp (●) and CATd (○) were introduced into the fads2−244 plasmid. ZFL cells were transiently co-transfected with various mutated fads2−244 plasmids, nSrebp2 expression plasmid, together with Renilla luciferase reference plasmid pRL-SV40. Luciferase coding region is indicated by shaded box and activity of the fads2 promoter in ZFL cells is expressed as normalised luciferase activity (to Renilla luciferase activity) relative to the empty pGL3-Basic plasmid. Values are means ± SD of three independent experiments (n = 3) (right panel). Groups indicated with different letters are significantly different (Tukey’s test; P < 0.05).

Full size image

In vitro binding of ZFL nuclear proteins at the SRE and CCAAT regions in the zebrafish fads2 promoter

Based on the observations from the mutagenesis experiments, we ran EMSA to assess the binding capacities of transcription factors in ZFL nuclear extract to the fads2 promoter cis-regulatory elements. Results showed the formation of DNA-protein complex when ZFL cells nuclear extract was incubated with wild type oligonucleotide probe containing CATp binding site. There was no complex formed when the probe was incubated without the nuclear extract or when a probe with mutated CATp sequence was used with the nuclear extract. Similarly, the use of non-labelled wild type oligonucleotides in excess amount completely outnumbered the formation of labelled DNA-protein complex. A comparable pattern was shown with CATd sequence (Fig. 5).

Figure 5
figure 5

Binding of ZFL nuclear proteins to the proximal CCAAT box (CATp) and distal CCAAT box (CATd) on the zebrafish fads2 promoter. EMSA was performed with biotin-labelled oligonucleotides containing CATp or CATd box and nuclear extract of ZFL cells. Specific binding of DNA-protein complex was validated with site-mutated oligonucleotides (Lane 3) and 100-fold molar excess of non-labelled competition oligonucleotides (Lane 4).

Full size image

Unlike the CATp and CATd probes, DNA-protein complex was not observed when nuclear proteins were added into the sample containing oligonucleotides with only SRE site (result not shown). Taking this and the results from mutagenesis experiment into consideration, we used a longer oligonucleotide consisting of both SRE and the adjacent CATp. Results showed that binding of transcription factors and DNA occurred with the simultaneous presence of both the SRE and CATp binding sites (Fig. 6). Mutation on either site prevented the formation of any DNA-protein complex. When the wild type labelled probe was incubated with excessive non-labelled probe containing mutated SRE and intact CATp, a decrease in DNA-complex staining was observed due to competition for binding. This result encored the observation in cell-based luciferase assay which showed the requirement of CATp for activation of reporter gene expression. In contrast, when wild type labelled probe was incubated with excessive non-labelled probe with intact SRE and mutated CATp binding site, the formation of DNA-complex occured. Therefore, in the absence of any conjugated protein binding to CATp, the binding of Srebp to the SRE was perturbed. Taken together, the results from mutagenesis and EMSA suggested a functional role for CCAAT box in the Srebp dependent transcriptional activation of zebrafish fads2 promoter.

Figure 6
figure 6

Binding of ZFL nuclear proteins to the SRE on the zebrafish fads2 promoter. Biotin-labelled probes containing SRE and proximal CCAAT box (CATp) were incubated with the nuclear extract of ZFL cells. Specific binding of DNA-protein complex was validated with site-mutated oligonucleotides (Lanes 3 and 4) and 100-fold molar excess of non-labelled mutated competition oligonucleotides (Lane 5 and 6).

Full size image

Transient expression of fads2 promoter (−1735/+19 bp) in developing zebrafish embryos

We observed GFP signal in the yolk syncytial layer (YSL) of embryos beginning at 26 hpf for both the pZs−244 bp and pZs−1735 bp constructs, respectively. As embryos developed, higher percentage of embryos with YSL-GFP expression were observed, along with increase in GFP signal intensity (Fig. 7). At 120 hpf, more than 95% of injected embryos showed GFP expression at the YSL (Table 1). Therefore, a fads2 promoter fragment containing response elements such as SRE, CATp and CATd proven to be essential for activation of expression is now shown to drive a spatio-temporal expression of fads2 in the zebrafish YSL.

Figure 7
figure 7

Transient GFP expression of linearised fads2 promoter (−1735/+19 bp) localised in yolk syncytial later (YSL) of zebrafish embryos. Lateral view of 26, 48, 72 and 96 hpf of transient transgene embryo injected with linearised fads2 promoter (−1735/+19 bp) showing increased expression in YSL (A,C,E,G) and their corresponding control (B,D,F,H). Cross-sectional view of a 120hpf control embryo (K–N) and a 120hpf transient transgenic embryo (I,J) injected with linearised fads2 promoter (−1735/+19 bp) showing expression at YSL.

Full size image
Table 1 Transient expression of the linearised fads2 promoter (−1735/+19 bp) in zebrafish embryos. Figures are cumulative embryos scored from three independent microinjections at 120 hpf.
Full size table

Discussion

The zebrafish fads2 was first cloned and shown to have the capacity for in vitro Δ5/Δ6 desaturation5. The transcripts were reported to express in several tissues known for biosynthesis of LC-PUFA during embryonic development21. The mRNA transcript of zebrafish fads2 is also seemingly regulated by dietary PUFA intake25. Despite these studies, investigation on the regulation of fads2 expression at transcription level is lacking. Here, we cloned a 1754 bp promoter region of the zebrafish fads2 gene to determine response elements critical for control of fads2 expression. We showed here that like all known teleost fads2 promoters, zebrafish fads2 promoter harbours SRE and CCAAT binding elements10,11,12,13. Zheng et al.10 also showed a high degree of sequence conservation between mammalian, amphibian and teleost species, which indicates conserved mechanism of fads2 role. As with salmon10 and trout13, putative SP1 binding site was also identified in this region of zebrafish fads2 promoter. In contrast, all sequenced promoters of fads2 from marine teleost species so far do not contain SP1 element, which was postulated as a reason for lower fads2 expression in marine fish species11,12,13.

Results from the promoter deletion experiment demarcated regions responsible for transcriptional activation of the zebrafish fads2. One such region is the cis-element SRE, the binding site for Srebp transcription factors. Using mutagenesis approach, we further corroborate the obligation of the SRE region in activating transcription of zebrafish fads2. We also showed that this sequence forms a DNA-protein complex with nuclear extract of ZFL cells. In human, an E-box like SRE region, as opposed to the classical SRE region, was crucial for both activation and suppression of the FADS2 gene24. Srebps are prominent regulators of lipid, fatty acid and cholesterol biosynthesis, and were first identified as nuclear proteins possessing binding abilities to SRE of low density lipoprotein receptor26. Like mammals, fish have two distinct homologues of mammalian Srebp1 and Srebp227. In zebrafish, activation of hepatic Srebp induce the expression of both lipid and cholesterol biosynthesis genes28. Our results also showed that in ZFL cells, Srebp2 activated a higher level of fads2 transcription as compared to Srebp1. This is reminiscent of the findings in Atlantic salmon, where Srebp2 promoted higher fads2 activities as compare to Srebp129. The importance of Srebp in mediating teleost fads2 expression has been shown through the use of cell-based luciferase assays involving Srebp expression plasmids or mutation of the promoter’s SRE response element13,14,29. In terms of tissue localisation, there was a parallel pattern between the distribution of srebp and fads2 mRNA expression27,30. The role of Srebp as a sensory mediator between dietary LC-PUFA intake and LC-PUFA biosynthesis was also observed in several fish species wherein limited dietary LC-PUFA intake led to increased Srebp level, subsequently resulting in higher fads2 mRNA expression13,29,30,31,32. In rat and human, dietary PUFA inhibits the rate of Fads2 transcription through Srebp, probably by lessening the production of mature Srebp or accelerating the decay of Srebp mRNA24,33. Besides FADS2, Srebp have also been shown to regulate ELOVL24,34.

Through a combination of deletion and mutagenesis, we demonstrate the importance of two CCAAT motifs in the core promoter region of zebrafish fads2 for transcriptional activation. The CCAAT box, named after the five nucleotides customarily found in the binding sites, is one of the most ubiquitous binding element in eukaryotic promoters, with several corresponding binding proteins already isolated and characterised35. The NF-Y is the major protein recognizing the CCAAT box, and is obligatory for the transcription of a considerable number of genes with ubiquitous or tissues-specific expression pattern35. Functional CCAAT motifs recognised by NF-Y are among the most frequent boxes found in promoters36. Mutation of putative NF-Y binding sites in fads2 promoter of two marine teleost species also resulted in weaker transcriptional activities10,14.

Our present finding suggests that the putative NF-Y binding sites are required for formation of complex between Srebp and SRE, which then activates the transcription of reporter gene. This result resonated with earlier findings wherein Srebp was incapable of driving transcription of target genes if adjacent NF-Y binding sites were mutated37,38,39. By themselves, Srebp proteins are inherently weak transcriptional activators, which underlines the need for Srebp proteins to interact with additional transcriptional factors to regulate the expression of lipogenic genes38,40,41. Studies on promoters of cholesterol and fatty acid metabolism genes consistently highlight the role of NF-Y as a pivotal partner in Srebp-mediated regulation of transcriptional activities36,38,41,42,43. Elsewhere, studies on murine stearoyl-CoA desaturase (Scd) and human FADS2 promoters both showed a requirement for putative NF-Y binding sites located adjacent to SRE for transcriptional activation24,39. Promoters of teleost and amphibian fads genes also contain a conserved region bearing NF-Y and Srebp binding sites at similar positions relative to the transcriptional start site (TSS)10. It is speculated that direct interaction of Srebp and NF-Y, together with the binding of both these transcription factors to neighbouring sites on the DNA lend stability to the DNA-protein complex39. In addition, our results showed that removal of the putative SP1 binding region did not supress the luciferase expression. The interaction between Srebp and SP1 in the regulation of Fads have not been demonstrated this far24,33, although a synergistic type of activation between these two transcriptional factors have been elucidated in other lipid metabolism genes44. A role for SP1 in Srebp-mediated activation of Fads2 however cannot be ruled out as different isoforms of Srebp or nutritional conditions may utilise different co-regulatory factors to activate expression of Fads, as shown in promoters of fatty acid synthase45.

Fatty acids from yolk storage are critical for energy and structural components for both fish embryo and larval stages46. Together with the head region, yolk sac has been reported to contain majority of the lipids during zebrafish embryonic stages19. While saturated and monounsaturated fatty acids (SFA, MUFA) are consumed for energy, LC-PUFA are essential for cellular membrane integrity, regulation of gene transcription, regulation of signalling pathways and production of eicosanoids. Previously, the yolk was regarded as a non-metabolic active site for nutrient reserve, where molecules such as lipids are stored within yolk cells and transported when needed46,47. However, emerging findings employing zebrafish as the model organism, are recognising the yolk as site for active lipid metabolism preceding transportation16. Using a transient transgenic approach in zebrafish embryos, we showed that the −1735/+19 bp region of the fads2 promoter directed a specific expression in the YSL beginning at 26 hpf embryos. In support of this, fads2 and elovl transcripts were reported to be expressed in YSL at the 24 hpf stage, alongside brain tissues18,21. During early embryonic development, some of the blastomeres situated at surface of the yolk collapse and fuse with the yolk cells, leading to the formation of YSL48,49. It then undergoes a series of changes in terms of thickness and nuclei distribution, before going into the process of degradation upon the exhaustion of yolk supplies50. YSL governs major roles during development, including yolk metabolism, epiboly, specification of germ layers, organogenesis of endoderm organs and innate immunity50,51,52. A major role of YSL in teleost development is on the transport of nutrients from yolk to embryonic cells and tissues. Transcripts of genes related to lipid metabolism and transportation53,54,55,56,57,58, are expressed in YSL during early development to facilitate the hydrolysis and transport of yolk lipids. When these genes are disrupted, embryos show reduced yolk consumption, delayed embryonic growth and failure in lipid absorption58,59.

The presence of GFP signal in YSL also suggested a role for LC-PUFA, the product of fads2 promoter activities, in transportation of yolk lipids. Detailed analyses of lipid classes in zebrafish yolk imply active de novo synthesis of lipids, in which fatty acids are resynthesised into major lipid classes between 0–48 hpf stage, followed by packaging into very low-density lipoproteins for supply to whole embryo beyond 48hpf15,19. In Caenorhabditis elegans, transportation of yolk lipoprotein to oocytes was disrupted in mutants with PUFA deficiency, a malaise restored by feeding worms with n-6 PUFA60. These findings, viewed alongside our results showing localisation of fads2 promoter activity at the YSL, provides the collective interpretation that LC-PUFA is required for transportation of yolk lipids during development.

In adult teleost, the liver is a major site for LC-PUFA biosynthesis, with both fads and elovl transcripts being reported in adult zebrafish liver and early larval stages18,21. Since the injected embryos only showed expression in the YSL, the fads2 promoter fragment lacks the regulatory region responsible for directing expression of fads2 in other organs such as liver, intestine or brain. It is plausible that the regulatory elements for the expression of these tissues are located in more distal regions or/and introns61,62. Multiple studies have shown that recapitulation of endogenous expression by a promoter in zebrafish embryos depends on the length of the promoter fragment62,63,64.

In conclusion, we successfully isolated a 1754 bp upstream region of the zebrafish fads2 promoter and identified several response elements critical to drive the activation of transcription. We also showed the capacity of this fragment to direct GFP expression at the YSL region of 26 hpf and older stages embryos. Since zebrafish relies on yolk as sole lipid source prior to onset of exogenous feeding capacity, it potentially provides a platform to further elucidate the role of fads2 or LC-PUFA in transportation of yolk lipids across the YSL.

Materials and Methods

Cell line maintenance and subculture

The zebrafish liver cell line, ZFL (ATCC® CRL-2643™) was purchased from the American Type Culture Collection (ATCC, USA) and maintained in the complete growth medium at 28 °C following the protocol of ATCC. Routine subculture was performed once the ZFL cells have achieved 80 to 90% confluence in a 25 cm2 flask.

Zebrafish maintenance and embryo collection

Wild type zebrafish (AB line) were purchased from the Institute of Molecular and Cell Biology (IMCB, Singapore), and maintained in the ZebTEC Stand Alone System (Tecniplast, USA). The fish were kept on a 13:11 h light/dark cycle at 28.5 °C, and fed till visual satiation twice daily with a combination of commercial micro pellet (Aquadene, Malaysia), frozen bloodworms and Artemia nauplii. Breeding and embryo collection were carried out according to Westerfield (2000) with adaptation65.

Animal ethics approval

Husbandry, handling and use of animals in this manuscript comply with the guidelines and requirements of the Animal Ethics Committee, Universiti Sains Malaysia and were approved by the same committee (PA/ACSC/002/2011).

Zebrafish fads2 promoter cloning and promoter-luciferase reporter plasmid construction

A promoter region of 1754 bp, corresponding to −1735/+19 bp relative to the transcription start site of the zebrafish fads2 gene was amplified with PCR utilising the i-Taq™ Plus DNA Polymerase (iNtRON, Korea). DNA extracted from the ZFL cells was used as the template for the amplification, together with the forward primer F1735 containing the restriction site for KpnI, and a reverse primer R19 containing the restriction site for XhoI (Supplementary data 3). The reaction mixtures were subjected to a thermal cycling programme consisted of an initial denaturation step for 2 min at 94 °C, followed by 40 cycles of denaturation for 20 s at 94 °C, annealing for 20 s at 50 °C, and extension for 1 min/kb of product length at 72 °C, and a final extension step for 5 min at 72 °C. The resulted PCR products were gel purified, cloned into the pGL3-Basic luciferase reporter vector (Promega, USA) and sequenced. Similarly, for the 5′-deletion analysis, another five promoter-luciferase reporter plasmids containing the fads2 promoter region with deletions of different lengths at the 5′-end were generated by using a series of forward primers (F1145, F586, F244, F161 or F67) and a common reverse primer (R19).

Zebrafish Srebp expression plasmid construction

Total RNA was extracted from the adult zebrafish liver using the TRIzol® Reagent (Invitrogen, Germany) according to the manufacturer’s descriptions. First-strand cDNA was generated via reverse transcription PCR using the M-MLV Reverse Transcriptase (Promega, USA). PCR fragments corresponding to the N-terminal of zebrafish srebp-1 (nSrebp-1, 1-457 aa; NM_001105129.1) and srebp-2 (nSrebp-2, 1-464 aa; NM_001089466.1) were identified through alignment with the human SREBP homologues (Nakakuki et al., 2007), and amplified from the first-strand cDNA using forward primers containing the restriction site for KpnI and reverse primers containing a stop codon (TAG) for translation termination and the restriction site for XhoI (Supplementary data 4). PCR amplification was performed as abovementioned and the resulted PCR products were cloned into the CMV promoter-driven expression plasmid pcDNA3.1 (Invitrogen, Germany).

Transient transfection and dual luciferase reporter assay

Prior to the day of transient transfection, 2.5 × 104 ZFL cells per well were seeded in a 96-well clear bottom white walled plates (Thermo Scientific, USA) and incubated overnight for cell attachment. Transient transfection was carried out using transfection reagent Lipofectamine™ 2000 (Life Technologies, Germany) at a volume of 0.15 µl per 100 ng DNA. For nSrebp overexpression assay, ZFL cells were co-transfected with fads2-1735 promoter-luciferase reporter plasmid (100 ng/well), one of the nSrebp expression plasmids (50 ng/well) and Renilla luciferase reference plasmid, pRL-SV40 (50 ng/well). ZFL cells prepared for the 5′-deletion assay were co-transfected with one of the pGL3 promoter-luciferase reporter plasmids (equimolar to 100 ng fads2-1735 plasmid/well), pcDNA3.1-nSrebp2 (50 ng/well) and pRL-SV40 (50 ng/well). Co-transfection for mutation analysis was performed with one of the mutated pGL3 promoter-luciferase reporter plasmids (100 ng/well), pcDNA3.1-nSrebp2 (50 ng/well) and pRL-SV40 (50 ng/well). Lastly, nuclear extracts for EMSA (refer to section 2.7) were prepared from the ZFL cells seeded in a 25 cm2 flasks and transfected with pcDNA3.1-nSrebp2 (2 µg/ml). Briefly, plasmid DNA-Lipofectamine™ 2000 complexes in 50 µl Opti-MEM were equally distributed to each well and the ZFL cells were incubated for 4.5 h at 28.5 °C. The plasmid DNA-Lipofectamine™ 2000 complexes contained-culture medium was then replaced with fresh culture medium and the transfected ZFL cells were further incubated for 24 h before being proceeded to subsequent assay. The ZFL cells were harvested in 20 µl/well Passive Lysis Buffer and luciferase assay was carried out using the Dual-Glo® Luciferase Assay System (Promega, USA) according to the manufacturer’s protocols. All the luminescence measurements were performed in the GloMax® Multi + Detection System (Promega, USA). Firefly luminescence reading for each sample was first normalised to Renilla luminescence reading as the internal control of transfection efficiency, before it was further normalised to the control cells transfected with pGL3-Basic plasmid for the comparison of promoter activities.

Site-directed mutagenesis

Multiple base substitution mutations at the SRE site, proximal and distal CCAAT box sites were created in the fads2-244 plasmid employing the Muta-DirectTM Site-Directed Mutagenesis Kit (iNtRON, Korea) following the manufacturer’s instructions. Complementary mutagenic primers (Supplementary data 5) used in the mutagenesis were designed using web-based programme PrimerX (http://www.bioinformatics.org/primerx/). Multiple site mutations were carried out one after another by using previously constructed mutant plasmids as templates.

Electrophoretic mobility shift assay

Prior to electrophoretic mobility shift assay (EMSA), 3′-end of each complementary strand of the oligonucleotides (Supplementary data 6) were labelled with biotin using the Biotin 3′ End DNA Labelling Kit (Thermo Scientific, USA) following descriptions of the manufacturer. Subsequently, biotin-labelled double-stranded oligonucleotides (10 μM) were generated by mixing equal amounts of each complementary strand of the resulted oligonucleotides at least 1 h prior to EMSA. For non-labelled competition double-stranded oligonucleotides, equimolar of each complementary strand of oligonucleotides (10 µM) were added into 1X annealing buffer containing 10 mM Tris (pH 7.5), 1 mM EDTA and 50 mM NaCl, and the mixture was heated in the Thermomix (Eppendorf, Germany) for 5 min at 95 °C before gradually cooling down to room temperature over a period of approximately 1 h. Nuclear extract was prepared from the transfected ZFL cells using the NE-PER® Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, USA) according to the manufacturer’s instructions. For EMSA, the LightShift® Chemiluminescent EMSA Kit (Thermo Scientific, USA) was used and each binding reaction containing 1X binding buffer, 1 µg Poly (dI·dC), 5% (v/v) glycerol, 0.05% (v/v) NP-40, 50 mM KCl, 5 mM MgCl2, 20 mM EDTA, 5 µl nuclear extract and 100 fmol biotin-labelled oligonucleotides in a total volume of 20 µl ddH2O was prepared. For the negative control binding reaction, nuclear extract was excluded, and 100-fold molar excess of non-labelled oligonucleotides was included to the competition binding reaction. The DNA-protein binding complexes were then incubated at room temperature for 30 min before size-fractioning on a 5% (v/v) non-denaturing polyacrylamide gel at 4 °C in the Mini-PROTEAN II Slab Electrophoresis Cell (Bio-Rad, USA). Finally, biotin-labelled DNA was detected with the aid of Chemiluminescent Nucleic Acid Detection Module (Thermo Scientific, USA) and luminescent signal was captured using the CCD camera (Bio-Rad, USA).

GFP promoter-reporter plasmid construction

DNA fragments from the fads2−244 and −1735 promoter-luciferase reporter plasmids were PCR amplified using both forward and reverse primers containing the restriction site for XhoI and BamHI (Supplementary data 3), respectively. The resulted PCR products were gel-purified and cloned into the pZsGreen1-1 GFP promoter-reporter plasmid.

Transient expression of GFP promoter-reporter plasmid in zebrafish embryo

In vivo promoter activity of both promoter fragments was determined by microinjection of GFP promoter-reporter plasmids into zebrafish embryos, respectively. Prior to microinjection, GFP promoter-reporter plasmids were linearised with the restriction enzymes XhoI and NotI. Each pZsGreen1-1 GFP promoter-reporter construct was adjusted to 10 ng/µl with dH2O containing 1X Danieau’s buffer and 0.2% phenol red. Approximately, 4.6 nl of the DNA solution was delivered into embryos at early one-cell stage by using the Nanoliter 2000 Microinjector (World Precision Instrument, USA). The injected embryos were incubated in E3 medium at 28.5 °C. The embryos were periodically monitored under the MVX10 Fluorescence Macro Zoom Microscope equipped with ColorView III Soft Imaging System (Olympus, Japan) until 120 hpf. For cryosectioning, zebrafish young larvae were fixed with 4% paraformaldehyde, treated and sectioned at a thickness of 20 µm using the Leica CM1850 UV Cryostat (Leica Biosystems, Germany).