Cloning and functional characterization of Phaeodactylum tricornutum front-end desaturases involved in eicosapentaenoic acid biosynthesis Fre ´ de ´ ric Domergue 1, *, Jens Lerchl 2 , Ulrich Za¨ hringer 3 and Ernst Heinz 1 1 Institut fu ¨ r Allgemeine Botanik, Universita ¨ t Hamburg, Hamburg, Germany; 2 BASF Plant Science GmbH, BPS-A30, Ludwigshafen, Germany; 3 Forschungszentrum Borstel, Borstel, Germany Phaeodactylum tricornutum is an unicellular silica-less diatom in which eicosapentaenoic acid accumulates up to 30% of the total fatty acids. This marine diatom was used for cloning genes encoding fatty acid desaturases involved in eicosapentaenoic acid biosynthesis. Using a combination of PCR, mass sequencing and library screening, the coding sequences of two desaturases were identified. Both protein sequences contained a cytochrome b 5 domain fused to the N-terminus and the three histidine clusters common to all front-end fatty acid desaturases. The full length clones were expressed in Saccharomyces cerevisiae and characterized as D5- and D6-fatty acid desaturases. The substrate specificity of each enzyme was determined and confirmed their involvement in eicosapentaenoic acid biosynthesis. Using both desaturases in combination with the D6-specific elongase from Physcomitrella patens, the biosynthetic path- ways of arachidonic and eicosapentaenoic acid were recon- stituted in yeast. These reconstitutions indicated that these two desaturases functioned in the x3- and x6-pathways, in good agreement with both routes coexisting in Phaeodacty- lum tricornutum. Interestingly, when the substrate selectivity of each enzyme was determined, both desaturases converted the x3- and x6-fatty acids with similar efficiencies, indicating that none of them was specific for either the x3- or the x6-pathway. To our knowledge, this is the first report describing the isolation and biochemical characterization of fatty acid desaturases from diatoms. Keywords: front-end desaturase; diatom; polyunsaturated fatty acid; eicosapentaenoic acid. Diatoms are unicellular photosynthetic microalgae of the phytoplankton that are particularly important in ocean ecosystems; they are thought to be responsible for as much as 25% of global primary productivity and for an accord- ingly significant O 2 production [1]. Because most of diatoms are surrounded by a highly structured silica cell wall, they also play a key role in the biogeochemical cycling of silica [2]. Diatoms are used commercially for various purposes such as feeds in aquaculture, sources of polyunsaturated fatty acids (PUFAs) or pharmaceutical drugs [3]. Phaeo- dactylum tricornutum, a silica-less diatom, is one of the most widely utilized model systems for studying the ecology, physiology, biochemistry and molecular biology of diatoms [4]. This organism became even more attractive with the recent establishment of a procedure for its stable transfor- mation [4–6], which enabled the demonstration of its sensing system [7] and its conversion into a heterotrophic organism, opening the possibility to grow it by large-scale fermentation for commercial exploitation [8]. As its fatty acid composition contains up to 30% eicosapentaenoic acid (EPA, 20:5 D5,8,11,14,17 ) and only traces of docosahexaenoic acid (DHA, 22:6 D4,7,10,13,16,19 ), P. tricornutum represents an interesting alternative source for the industrial production of EPA, and gram-scale purification of this particular PUFA has already been achieved [9]. Very long chain PUFAs like EPA, DHA and arachidonic acid (ARA, 20:4 D5,8,11,14 ) have received great interest as such fatty acids are required in the human diet for normal health and development, particularly in the case of new-borns and infants [10,11]. Polyunsaturated fatty acids are important constituents of membranes and precursors of several biologically active eicosanoids, like prostaglandins and leukotrienes, which regulate many physiological functions in mammals [12]. For mammals lacking D12- and D15-fatty acid desaturases, 18:2 D9,12 and 18:3 D9,12,15 are considered to be essential fatty acids that must be supplied in the diet. Although humans can synthesize very-long chain PUFAs from these precursors, dietary changes over the last decades resulted in very high x6- to x3-fatty acid ratios that have negative impacts on both health and development [13,14]. In order to circum- vent this deficit in x3-fattyacidsandtopreservemarine reserves, alternative sources have been searched, among others the production of PUFAs in transgenic oilseed crops [15]. Such a goal has led to the identification of most of the genes coding for the enzymes involved in PUFA biosynthesis (fatty acid desaturases and elongases) in several organisms, such as the nematode Caenorhabditis elegans [16], the fungus Mortierella alpina [17,18] and the moss Physcomitrella patens [19,20], but so far not in any diatom, even though they represent an important group of primary x3-fatty acid producers. Correspondence to F. Domergue, Institut fu ¨ r Allgemeine Botanik, Universita ¨ t Hamburg, Ohnhorststrasse 18, 22609 Hamburg, Germany. Fax: +49 40 42816 254, Tel.: + 49 40 42816 373, E-mail: fredDo@botanik.uni-hamburg.de Abbreviations: PUFAs, polyunsaturated fatty acids; GLC-MS, gas liquid chromatography-mass spectrometry; ARA, arachidonic acid; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid; FAMEs, fatty acid methyl esters; FID, flame-ionization detector. *Present address: Plant Science Sweden AB, SE-26831 Svalo ¨ v, Sweden. (Received 18 April 2002, revised 24 June 2002, accepted 10 July 2002) Eur. J. Biochem. 269, 4105–4113 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03104.x Using various 14 C-labelled precursors, Moreno et al.[21] showed that P. tricornutum synthesized EPA de novo by elongation and aerobic desaturation of fatty acids. As shown in Fig. 1, EPA can be synthesized by the classical x6-pathway, the classical x3-pathway, a pathway relying on intermediates of both of these pathways and by an alter- native x3-pathway involving D9-elongation and D8-desat- uration [22]. Pulse-chase experiments with [ 14 C]linoleic acid suggested that the third route was the most active one, invol- ving successively D6-desaturation (18:2 D9,12 to 18:3 D6,9,12 ), x3-desaturation (18:3 D6,9,12 to 18:4 D6,9,12,15 ), D6-elongation (18:4 D6,9,12,15 to 20:4 D8,11,14,17 ) and a final D5-desaturation (20:4 D8,11,14,17 to 20:5 D5,8,11,14,17 [22]). P. tricornutum is a particularly interesting model for studies on PUFA biosyn- thesis because EPA represents up to 30% of the total fatty acids whereas all the intermediates of its pathway are only present in traces. This accumulation of only EPA may indicate that the elongase and desaturases responsible for its biosynthesis are very effective in channelling the different intermediates towards the final product. We therefore decided to use this organism for cloning the genes encoding the different fatty acid desaturases involved in EPA biosynthesis and identified two sequences coding for fatty acid desaturases with D5- and D6-regioselectivities. Using these two genes together with that of the D6-specific elongase from P. patens [20], the biosynthetic pathways of ARA and EPA were successfully reconstituted in yeast, showing that these desaturases were responsible for the D5- and D6-activities of both the x3- and x6-pathways. MATERIALS AND METHODS Materials Restriction enzymes, polymerases and DNA modifying enzymes were obtained from New England Bioloabs (Frankfurt, Germany) unless indicated otherwise. All other chemicals were from Sigma (St Louis, MO, USA). Phaeodactylum tricornutum culture P. tricornutum UTEX 646 was grown in f/2 culture medium supplemented with 10% organic medium [23] at 22 °C with aeration and photoperiods of 16 h of light (35 lEÆm )2 Æs )1 ). cDNA library construction Frozen cells of P. tricornutum were ground in the presence of liquid nitrogen and the resulting fine powder was resuspended in 2 mL of homogenization buffer (0.33 M sorbitol, 0.3 M NaCl, 10 m M EDTA, 10 m M EGTA, 2% SDS, 2% 2-mercaptoethanol in 0.2 M Tris/HCl pH 8.5). Phenol (4 mL) and chloroform (2 mL) were added succes- sively and the samples were shaken vigorously for 15 min at 40–50 °C. After centrifugation (10 min · 10 000 g), the aqueous phase was successively extracted with 2 mL of phenol/chloroform (1 : 1, v/v) and 2 mL of chloroform. 1/20 vol. of 4 M NaHCO 3 and 1 vol. of ice-cold isopropanol were added and the sample stored overnight at )20 °C. Nucleic acids were precipitated (30 min · 10 000 g), washed with 70% ethanol and resuspended in 80 m M Tris/borate pH 7.0 containing 1 m M EDTA. 1/3 vol. of 8 M LiCl was added and the sample incubated 30 min at 4 °C. After centrifugation (30 min · 10 000 g), the pellet was washed with ethanol 70% (v/v) and resuspended in RNase-free water. Poly(A) + RNA was isolated with Dynabeads (Dynal, Oslo, Norway) and cDNA first strand synthesis was achieved using the murine leukemia virus reverse transcriptase (Roche, Mannheim, Germany). The second strand synthesis was carried out by incubation with DNA polymerase I and Klenow enzyme, followed by RNaseH digestion. cDNA was blunted with the T4 DNA Fig. 1. Possible biosynthetic routes leading to EPA biosynthesis in Phaeodactylum tricornu- tum. The classical x6- and x3-pathways are framed and the alternative x3-pathway (involving D9-elongation and D8-desatura- tion) is shown with broken arrows. The most active route according to Arao and Yamada [22] is shown with wide arrows and with text in bold. 4106 F. Domergue et al. (Eur. J. Biochem. 269) Ó FEBS 2002 polymerase (Roche), and EcoRI/XhoIadapters(Pharmacia, Freiburg, Germany) were ligated with the T4 DNA ligase. After XhoI digestion, phosphorylation with the polynucleo- tide kinase (Roche) and gel separation, DNA molecules larger than 300 bp were ligated into vector arms and packaged into lambda ZAP Express phages using the Gigapack Gold Kit (Stratagene, Amsterdam, Netherlands). Random sequencing of the cDNA library After in vivo mass excision of the cDNA library, plasmid recovery and transformation of Escherichia coli DH10B (Stratagene), plasmid DNA was prepared on a Qiagen DNA preparation robot (Qiagen, Hilden, Germany) according to the manufacturer’s instructions and submitted to random sequencing by the chain termination method using the ABI PRISM Big Dye Termination Cycle Sequencing Ready Reaction Kit (Perkin-Elmer, Weiters- tadt, Germany). About 8400 clones were processed and annotated using the standard software package EST-MAX (Bio-Max, Munich, Germany), resulting in 3860 nonredun- dant sequences. PCR amplification In parallel, a PCR-based cloning strategy was followed using primer mixtures corresponding to the highly con- served histidine clusters present in membrane-bound desat- urases [24]. After in vivo mass excision of the P. tricornutum cDNA library, the resulting plasmid bank was used as template for PCR with different combinations of degenerate primers [19]. PCR amplifications were carried out using the Pfu DNA polymerase (Stratagene) and the following program: denaturation for 3 min at 96 °C; 5 cycles of 30 s at 96 °C, 30 s at 55 °C (decreasing by 3 °C in each successive cycle), 1 min at 72 °C; 30 cycles of 30 s at 96 °C, 30 s at 40 °C,1minat72°C; and a final extension step of 10 min at 72 °C. cDNA library screening Both methods yielded several sequences which were anno- tated to putative desaturases. For full length cloning via library screening, the sequence of interest was cloned into the pGEM-T vector (Promega, Madison, WI, USA) and a digoxigenin-labelled DNA probe of this fragment was synthesized by PCR. The cDNA library described above (about 5 · 10 5 plaques) was screened according to the manufacturer’s instructions (Boehringer, Mannheim, Ger- many, Stratagene). After two rounds of screening, several clones were isolated and the longest ones sequenced on both strands. Functional characterization in S. cerevisiae For functional characterization, the P. tricornutum cDNA clones were cloned in different yeast expression vectors. For this purpose, the different open reading frames (ORF) were modified by PCR to create appropriate restriction sites adjacent to the start and stop codons and to insert the yeast consensus sequence for enhanced translation [25] in front of the start codon. The amplified DNAs were first cloned into pUC18 using the SureClone Ligation Kit (Pharmacia). The ORFs were then released using the restriction sites created by PCR and cloned into the same sites of different yeast expression vectors. Using SpeI and SacI sites, the ORF of PtD5 was cloned behind the galactose-inducible promoter P GAL10 of the yeast expression vector pESC-LEU (Strata- gene), yielding pESC-LEU-PtD5. The ORF of PtD6 was cloned behind the constitutive promoter P ADH of the yeast expression vector pVT102-U [26] using BamHI and XhoI sites, yielding pVT102-U-PtD6. To obtain pESC-LEU- PSE1-PtD5, the ORF of pse1 was released from pY2PSE1 [20] using BamHI, cloned behind the galactose-inducible promoter P GAL1 of pESC-LEU before inserting the ORF of PtD5 as indicated above. C13ABYS86 Saccharomyces cerevisiae strain (leu2, ura3, his, pra1, prb1, prc1, cps [27]) was transformed with plasmid DNA by a modified PEG/lithium acetate protocol [28]. After selection on minimal medium agar plates without uracil or leucine, cells harbouring the yeast plasmid were cultivated in minimal medium lacking uracil or leucine but containing 2% (w/v) raffinose and 1% (v/v) Tergitol NP-40. The expression was induced by supplementing galactose to 2% (w/v) when the cultures had reached a D 600 of 0.2–0.3. At that time, the appropriate fatty acids were added to a final concentration of 500 l M , unless indicated otherwise. All cultures were then grown for a further 48 h at 20 °C, unless indicated otherwise, and used for fatty acid analysis. Fatty acid analysis Fatty acid methyl esters (FAMEs) were obtained by transmethylation of yeast cell sediments with 0.5 M sulphu- ric acid in methanol containing 2% (v/v) dimethoxypropane at 80 °C for 1 h. FAMEs were extracted in petroleum ether and analysed by gas-liquid chromatography (GLC) using a Hewlett-Packard 6850 gas chromatograph (Hewlett-Pac- kard, Palo Alto, CA, USA) equipped with a polar capillary column (ZB-Wax, 30 m · 0.32 mm internal diameter, 0.25 lm film, Torrance, CA, USA) and a flame-ionization detector (FID). Data were processed using the HP Chem Station Rev. A06.03. FAMEs were identified by compar- ison with appropriate reference subtances or by GLC-MS of FAMEs and 4,4-dimethyloxazoline derivatives as described previously [29]. RESULTS Isolation of two fatty acid desaturase-like cDNA clones Two full length clones encoding putative fatty acid desat- urases were isolated from the P. tricornutum cDNA library using a combination of PCR amplification, mass sequencing and library screening as described in the Materials and methods. The first clone was 1689 bp and contained an ORF of 1434 bp coding for a polypeptide of 477 amino acids. This polypeptide showed high sequence homologies to various D6-desaturases. It was, for example, 43% identical (59% similar) to the D6-desaturase from Pythium irregulare (GenBank accession number AAL13310) and 37% identical (53% similar) to the D6-desaturase from Mortierella alpina [30]. Accordingly, this clone was anno- tated PtD6 (for P. tricornutum delta 6-desaturase). The second clone was 1672 bp and contained an ORF of 1410 bp coding for a polypeptide of 469 amino acids; it had Ó FEBS 2002 D5- and D6-fatty acid desaturases from diatom (Eur. J. Biochem. 269) 4107 highest sequence similarities to the two D5-fatty acid desaturases recently cloned from Dictyostelium discoideum. For example, it was 21% identical (39% similar) to the second D5-desaturase from D. discoideum [31]. This clone was consequently annotated PtD5. Protein domains Figure 2 shows the amino acid sequences of the proteins encoded by PtD5 and PtD6 (PtD5p and PtD6p, respect- ively). Both proteins contain the three conserved histidine clusters characteristic for all membrane-bound desaturases [24]. They also contain a cytochrome b 5 domain fused at the N-terminus and an H to Q substitution in the third histidine-box, both of these features being typical of front- end desaturases [32]. Hydrophobic plots have indicated the presence of several long hydrophobic stretches and the four potential transmembrane helices fitting to the topological model developed for membrane-bound desaturases by Shanklin et al. [33] are shown. In good agreement with this model, the program developed by Nakai & Horton for prediction of cellular localization [34] indicated that both proteins were localized in the endoplasmic reticulum, although no di-lysine retention signal was present in the sequences. Figure 2 also shows the amino acid sequence of the free cytochrome b 5 (PtCytb5), which was obtained in full length from the mass sequencing. Whereas the cytochrome b 5 domain of PtD6p contains the eight invariant amino acids of the cytochrome b 5 superfamily (Fig. 2, dots; [35]), PtD5p has two substitutions (W to E and Y to F), with only the last one being conserved. Such substitutions have in fact already been found for four of the eight conserved amino acids as only the HPGG cluster is systematically present in all the cytochrome b 5 domains of front-end desaturases cloned so far. Another characteristic feature of the cytochrome b 5 domain of fused desaturase was observed in P. tricornutum. The sequence of the free microsomal cytochrome b 5 between the two highly conserved W and G (marked with arrows) contains 12 acidic amino acids (aspartate and glutamate), whereas the corresponding domain sequences in PtD5p and PtD6p have only eight and five acidic residues, respectively. Such a decrease in the number of acidic amino acids in fused cytochrome b 5 domains has been observed in Fig. 2. Amino acid sequences of PtD5p and PtD6p. For alignment the CLUSTAL X program was used (gap opening 12, gap extension 0.05) and the five highest BLAST scores of each desaturase (not shown) in order to obtain correct alignment of the histidine clusters. The conserved amino acids are on black back- ground. The eight amino acids which are usually conserved in the cytochrome b 5 superfamily are marked with dots, the three histidine boxes framed and the potential transmembrane domains underlined. The sequence data published here have been submitted to the DDBJ/EMBL/GenBank sequence data bank with the accession number AY082392, AY082393 and AF503284 for PtD5p, PtD6p and PtCytb5, respectively. 4108 F. Domergue et al. (Eur. J. Biochem. 269) Ó FEBS 2002 most front-end desaturases cloned so far and may be correlated with the necessary close interaction between this domain and the desaturase part [32]. Functional characterization in Saccharomyces cerevisiae The activities of the proteins encoded by these two cDNAs were confirmed by heterologous expression in S. cerevisiae. The open reading frames of PtD6 and PtD5 were cloned in the yeast expression vectors pVT102-U [26] and pESC-LEU (Strategene), respectively, and these constructs and the empty vectors were transformed into the S. cerevisiae strain C13ABYS86 [27]. Expression were carried out for 48 h at 20 °C in the presence of the potential substrates for D6- or D5-fatty acid desaturases, 18:2 D9,12 and 20:3 D8,11,14 ,respect- ively. In the presence of the empty vector pVT102-U and 500 l M linoleic acid, only the endogenous yeast fatty acids and the additional substrate could be detected indicating that yeast did not metabolize the exogenously added fatty acid (Fig. 3A). In contrast, in the yeast containing pVT102- U-PtD6, a new fatty acid corresponding to c-linolenic acid (18:3 D6,9,12 ) was detected. The structure of this fatty acid was confirmed by GLC-MS analyses demonstrating that PtD6p was a D6-desaturase. In pVT102-U, the expression of PtD6 was under the control of a constitutive promoter (P ADH ) which resulted in a high desaturation of linoleic acid (25– 30%). When PtD5p was expressed in the presence of 20:3 D8,11,14 , a peak corresponding to arachidonic acid (ARA, 20:4 D5,8,11,14 ) was detected (Fig. 3B). The structure was confirmed by GLC-MS analyses, demonstrating that PtD5p was a D5-desaturase. Specificity of Phaeodactylum tricornutum front-end desaturases The substrate specificity of each desaturase was subse- quently determined in the same transgenic yeast using different potential substrates (Table 1). The favourite sub- strate of PtD6p was linoleic acid, but a-linolenic acid (18:3 D9,12,15 ) was converted to stearidonic acid (18:4 D6,9,12,15 ) with similar efficiency. In the absence of exogenously fed fatty acids, 5–6% of the endogenous monounsaturated fatty acids of yeast were desaturated (16:1 D9 to 16:2 D6,9 and 18:1 D9 to 18:2 D6,9 ), similar to the D6-desaturases from P. patens [19] and M. alpina [17]. In order to check if PtD6p could also be responsible for the D8-desaturation activity involved in the alternative x3-pathway (see Fig. 1), similar to the bifunctional (D5/D6) fatty acid desaturase recently cloned from zebrafish [36], 20:3 D11,14,17 was assayed as a potential substrate. No conversion of this substrate could be detected indicating that PtD6p did not display any D8-desaturase activity (Table 1). The highest activity of PtD5p was obtained with 20:3 D8,11,14 as substrate (24,7% conversion), but PtD5p was also active with both 20:2 D11,14 and 20:3 D11,14,17 , as it has already been reported for the D5-desaturase from C. elegans [37]. PtD5p also showed low activities with 20:1 D11 and vaccenic acid (18:1 D11 ), but no activity was detected with 20:1 D8 (Table 1) or with oleic, linoleic and a-linolenic acid (data not shown). Reconstitution of PUFA biosynthetic pathways in Saccharomyces cerevisiae Using the two front-end desaturases from P. tricornutum and Pse1p, the D6-specific elongase from P. patens [20], the possibility of reconstituting the biosynthetic pathways of arachidonic and eicosapentaenoic acid in yeast was inves- tigated. Pse1p was used as it allows the synthesis of 20:3 D8,11,14 and 20:4 D8,11,14,17 from 18:3 D6,9,12 and 18:4 D6,9,12,15 , respectively [20]. When PtD5p was coexpressed with Pse1p in the presence of c-linolenic acid, 20:3 D8,11,14 and ARA were synthesized (Fig. 4A). About 50% of 18:3 D6,9,12 was elongated by Pse1p and 15% of the resulting 20:3 D8,11,14 was converted to 20:4 D5,8,11,14 by PtD5p. Simi- larly, when 18:4 D6,9,12,15 was exogenously fed, 20:4 D8,11,14,17 Fig. 3. Fatty acid profiles of yeast transformed with pVT102-U-PtD6 (A) or pESC-LEU-PtD5 (B). C13ABYS86 yeast strain transformed with either the empty vector (dashed line) or the different constructs (full line) was supple- mented with different fatty acids (A, 500 l M 18:2 D9,12 ;B,500 l M 20:3 D8,11,14 ) and grown for 48 h at 20 °C. FAMEs from the whole cells were prepared and analysed by GLC as indi- cated under Material and methods. Table 1. Substrate specificity of PtD5p and PtD6p expressed in Sac- charomyces cerevisiae. Yeast strain C13ABYS86 was transformed with the different constructs (pESC-LEU-PtD5 or pVT102-U-PtD6) and grown for 48 h at 20 °C in the presence of different fatty acid substrates (500 l M or as indicated). FAMEs from the whole cells were prepared and analysed by GLC as indicated under Material and methods. Desaturation (%) was calculated as (product · 100)/(educt + prod- uct) using values corresponding to percent of total fatty acids. Each value is the mean ± SD from three to five independent experiments. PtD5 PtD6 Substrate Desaturation Substrate Desaturation 18:1 D11 2.0 ± 0.2 16:1 D9a 5.7 ± 0.6 20:1 D8b 0 18:1 D9a 5.3 ± 0.5 20:1 D11 b 3.7 ± 0.7 18:2 D9,12 27.8 ± 3.6 20:2 D11,14 b 11.8 ± 1.0 18:3 D9,12,15 26.7 ± 3.3 20:3 D11,14,17 10.8 ± 0.3 20:3 D8,11,14 24.7 ± 3.3 20:3 D8,11,14 0 a In the absence of exogenously fed fatty acid. b In the presence of 1m M exogenously fed fatty acid. Ó FEBS 2002 D5- and D6-fatty acid desaturases from diatom (Eur. J. Biochem. 269) 4109 and EPA were detected (Fig. 4B). In the presence of a-linolenic acid (18:3 D9,12,15 ), the yeast containing the three heterologous genes synthesized 18:4 D6,9,12,15 ,20:3 D11,14,17 , 20:4 D8,11,14,17 and 20:5 D5,811,14,17 (EPA,Table2).Thepres- ence of 20:3 D11,14,17 resulted from the elongation of 18:3 D9,12,15 by Pse1p. Although the activity of Pse1p upon 18:3 D9,12,15 is usually low, the high proportion of the exogenously fed a-linolenic acid in the medium lead to a significant production of 20:3 D11,14,17 (Table 2). In addition to the unexpected synthesis of this fatty acid, traces of 16:2 D6,9 , 18:2 D6,9 and 20:4 D5,11,14,17 were detected (data not shown), in agreement with the specificities of both desatu- rases (Table 1). EPA was present and accumulated up to 0.23% of the total fatty acids, confirming that its complete biosynthetic pathway from a-linolenic acid, involving D6-desaturation, D6-elongation and D5-desaturation (x3-pathway), had been successfully reconstituted. About 16% of 18:3 D9,12,15 was converted by PtD6p, 24% of the resulting 18:4 D6,9,12,15 was then elongated by Pse1p and about 14% of the elongated product was finally desaturated to EPA by PtD5p. When 18:2 was the exogenously fed substrate, the synthesis of arachidonic acid (20:4 D5,8,11,14 ) was achieved, confirming that the three proteins were also active in the x6-pathway (Table 2). PtD6p converted about 23% of 18:2 D9,12 to c-linolenic acid, Pse1p elongated 14% of 18:3 D6,9,12 to 20:3 D8,11,14 and finally, PtD5p desaturated about 10% of the elongated product to ARA. Arachidonic acid represented 0.16% of the total fatty acids (Table 2) and, similar to the results obtained with the reconstitution of the EPA biosynthetic pathway, several side-products (20:2 D11,14 and traces of 16:2 D6,9 ,18:2 D6,9 and 20:3 D5,11,14 ) were present in the fatty acid profiles (data not shown). Selectivity of Phaeodactylum tricornutum front-end desaturases In order to evaluate if EPA was preferentially synthesized in P. tricornutum through the x6- or x3-pathway, the select- ivity of both front-end desaturases was investigated. In the x6-pathway, the D6-fatty acid desaturase acts on 18:2 D9,12 before the action of the D6-elongase followed by D5- and x3-desaturases, whereas in the x3-pathway, the D6-desat- urase converts the product of the x3-desaturase, 18:3 D9,12,15 , to 18:4 D6,9,12,15 (see Fig. 1). The selectivity of PtD6p was assayed in the presence of 125 l M of both 18:2 D9,12 and 18:3 D9,12,15 so that the four potential substrates of PtD6p Fig. 4. Fatty acid profiles of yeast transformed with pESC-LEU-PSE1-PtD5. C13ABYS86 yeast strain transformed with either the empty vector (dashed line) or pESC-LEU-PSE1- PtD5 (full line) was supplemented with 500 l M 18:3 D6,9,12 (A) or 18:4 D6,9,12,15 (B) and grown for 48 h at 20 °C.FAMEsfrom the whole cells were prepared and analysed by GLC as indicated under Material and methods. Table 2. Reconstitution of pathways for EPA and ARA biosynthesis in yeast. C13ABYS86 yeast strain was transformed with pVT102-U-PtD6 and pESC-LEU-PSE1-PtD5 or the corresponding empty vectors. The transformants were grown for 96 h at 20 °C in the presence of 250 l M 18:3 D9,12,15 or 18:2 D9,12 . FAMEs from the whole cells were prepared and analysed by GLC as indicated under Material and methods. Each value corresponds to the percent of total fatty acids and is the mean ± SD from three independent experiments. 18:3 D9,12,15 exogenously supplied 18:2 D9,12 exogenously supplied Fatty acid Empty vectors PtD6 + PSE1 + PtD5 Empty vectors PtD6 + PSE1 + PtD5 16:0 12.2 ± 0.3 12.0 ± 0.5 15.1 ± 0.2 15.6 ± 0.3 16:1 D9 21.2 ± 1.2 15.4 ± 1.4 14.4 ± 0.8 11.5 ± 0.3 18:0 5.9 ± 0.4 6.7 ± 0.6 4.3 ± 0.4 4.3 ± 0.3 18:1 D9 18.0 ± 0.6 17.4 ± 1.4 7.8 ± 0.1 9.0 ± 0.3 18:2 D9,12 – – 58.3 ± 0.1 43.5 ± 0.6 18:3 D6,9,12 – – – 11.1 ± 0.8 18:3 D9,12,15 42.6 ± 1.2 36.7 ± 1.9 – – 18:4 D6,9,12,15 – 5.2 ± 0.1 – – 20:2 D11,14 – – – 2.9 ± 0.1 20:3 D8,11,14 – – – 1.7 ± 0.1 ARA – – – 0.17 ± 0.03 20:3 D11,14,17 – 4.7 ± 0.6 – – 20:4 D8,11,14,17 – 1.5 ± 0.2 – – EPA – 0.23 ± 0.03 – – 4110 F. Domergue et al. (Eur. J. Biochem. 269) Ó FEBS 2002 (16:1 D9 , 18:1 D9 , 18:2 D9,12 and 18:3 D9,12,15 , Table 1) were present in similar proportions (Fig. 5A). PtD6p desaturated the intermediates of the x3- and x6-pathway, 18:3 D9,12,15 and 18:2 D9,12 , respectively, with similar efficiency. At the same time, it was also active on the monounsaturated fatty acids (16:1 D9 and 18:1 D9 ). Interestingly, the conversions of the four potential substrates by PtD6p observed in this experiment (6.3, 5.8, 33.5 and 25.6% for 16:1 D9 , 18:1 D9 , 18:2 D9,12 and 18:3 D9,12,15 , respectively) were very close to those obtained when feeding a single substrate (see Table 1). Therefore, PtD6p showed much higher activities towards the dienoic and trienoic intermediates of PUFA biosynthesis than towards the monounsaturated fatty acids, but it was not confined to either the classical x6- or x3-pathway. The D5-fatty acid desaturase is the last enzyme involved in the x3-pathway, converting 20:4 D8,11,14,17 to EPA, whereas in the x6-pathway, it converts 20:3 D8,11,14 to 20:4 D5,8,11,14 before an x3-desaturase synthesizes 20:5 D5,8,11,14,17 (see Fig. 1). To check the activity of PtD5p on both substrates, PtD5p was coexpressed with the elongase Pse1p in the presence of both c-linolenic acid and stearidonic acid (Fig. 5B). Pse1p activity led to the synthesis of 20:3 D8,11,14 and 20:4 D8,11,14,17 , both of them being further desaturated by PtD5p. In good agreement with Zank et al. [20], the elongase displayed no selectivity when facing two of its favourite substrates. Fifty percent of 18:3 D6,9,12 and 45% of 18:4 D6,9,12,15 were elongated by Pse1p, which is nearly identical to the data obtained upon feeding the substrates separately [20]. Similarly, PtD5p did not show any preference for 20:3 D8,11,14 or 20:4 D8,11,14,17 , as about 15% of each elongation product was desaturated. Conse- quently, like PtD6p, the D5-fatty acid desaturase from P. tricornutum has no preference for the x6- or the x3-pathway. These experiments confirmed that both front-end desaturases were acting simultaneously in the x3- and x6-pathway because they accept every potential substrate encountered. When experiments similar to those reported in Fig. 5 were conducted with the D5-desaturases from M. alpina or other D6-desaturases (from M. alpina, Homo sapiens, Borago officinalis, Ceratodon purpureus and P. patens), none of these front-end desaturases was specific for the x3- or the x6-fatty acids (data not shown). Although these desaturases differed in their level of activity and in substrate preference, all the D5- and D6-desaturases tested showed similar activities towards the two substrates provided simulta- neously (20:3 D8,11,14 and 20:4 D8,11,14,17 or 18:2 D9,12 and 18:3 D9,12,15 , respectively). These results concerning the selectivity of front-end desaturases suggest that such enzymes are in general not specifically restricted to one of the two classical x3- or x6-pathways as depicted in Fig. 1 but that the same enzymes are involved in both routes. DISCUSSION In the present study, we report the cloning and functional characterization of a D5- and a D6-fatty acid desaturase from P. tricornutum. Both contain the typical features of membrane-bound desaturases and a cytochrome b 5 domain fused to their N-terminal extremity (Fig. 2), similar to other front-end desaturases with the exception of those from cyanobacteria [32,33]. Such fused domains are also found at the C-terminal end of some D9-desaturases and in several hydroxylases, sulfite oxidases and nitrate reductases, all these proteins being members of the cytochrome b 5 super- family [38]. It should be noted that in a phylogenetic tree with the D5- and D6-desaturases from various organisms, PtD5p and PtD6p fell into different branches like the front- end desaturases from M. alpina,whereasinthecaseof C. elegans or man, each pair of front-end desaturases formed a separate branch (data not shown). This may indi- cate that the D5- and D6-desaturases from P. tricornutum and M. alpina have evolved separately a long time ago while a common ancestor has been functioning for a longer time in the case of man and C. elegans. Heterologous expression in S. cerevisiae was used to confirm the D5- and D6-regioselectivity of PtD5p and PtD6p, respectively (Fig. 3), and to determine each sub- strate specificity (Table 1). PtD6p desaturated 25–30% of both 18:2 D9,12 and 18:3 D9,12,15 (Table 1), while PtD5p was as active on 20:4 D8,11,14,17 as on 20:3 D8,11,14 when coexpressed with the D6-elongase from P. patens, Pse1p (Fig. 4). These specificities fit perfectly with their involvement in EPA biosynthesis. Because their coexpression with Pse1p in the presence of 18:2 D9,12 or 18:3 D9,12,15 led to the synthesis of ARA or EPA (Table 2), respectively, PtD5p and PtD6p can function in both the x3- and x6-pathway. Concerning selectivity, it was shown that neither PtD5p nor PtD6p showed any preference for x3- or x6-fatty acids (Fig. 5). Moreover, each desaturase displayed similar activities towards the different substrates, when they were provided in a mixture or fed as a single substrate. As corresponding results were obtained with several D5- and D6-desaturases from different organisms, indiscriminate use of x3- and x6-fatty acids may be a general feature of front-end desaturases. Fig. 5. Fatty acid profiles of yeast transformed with pVT102-U-PtD6 (A) or pESC-LEU- PSE1-PtD5 (B). C13ABYS86 yeast strain transformed with either the empty vector (dashed line) or the different constructs (full line) was supplemented with different fatty acids (A, 100 l M 18:2 D9,12 and 100 l M 18:3 D9,12,15 ;B,250l M 18:3 D6,9,12 and 250 l M 18:4 D6,9,12,15 ) and grown for 48 h at 20 °C. FAMEs from the whole cells were prepared and analysed by GLC as indicated under Material and methods. Ó FEBS 2002 D5- and D6-fatty acid desaturases from diatom (Eur. J. Biochem. 269) 4111 The results presented in Fig. 5B also show that the D6- elongase converted equally well 18:3 D6,9,12 and 18:4 D6,9,12,15 . The fact that the presence or absence of an x3-double bond has no effect on front-end desaturase and elongase activities may be correlated with the regiochemistry of the reactions catalysed. Such enzymes are acting at the carboxyl end of the molecule, whereas the structural difference between x3- or x6-fatty acids resides in the methyl end. Accordingly, it is probable that neither the positioning of the substrate in the reaction center nor the catalysis is affected by the presence of an x3-double bond. In this regard, the recent demon- stration that the human D6-desaturase is not only active on 18:2 D9,12 and 18:3 D9,12,15 but as well on 24:4 D9,12,15,18 and 24:5 D9,12,15,18,21 [39] suggests that front-end desaturases may tolerate some differences at the methyl end of their substrates and only require an appropriate carboxylic end, where the new double bond is to be inserted. It should be noted that the inverse situation was demonstrated in the case of x3-desaturases. Expression of the x3-desaturases from Brassica napus and C. elegans (FAT-1) in yeast clearly showed that both were insensitive to the fatty acid chain length and to the presence of double bonds proximal to the carboxyl end [40,41]. Similarly, a very low selectivity was demonstrated for the x3-desaturase FAT-1 as this desatur- ase expressed in mammalian cells converted almost all x6-fatty acids to the corresponding x3-fatty acids (18:2 D9,12 to 18:3 D9,12,15 , 20:2 D11,14 to 20:3 D11,14,17 , 20:3 D8,11,14 to 20:4 D8,11,14,17 , ARA to EPA and even 22:4 D7,10,13,16 to 22:5 D7,10,13,16,19 [42]). In P. tricornutum, the different intermediates of the EPA biosynthetic pathway are only present in trace amounts, suggesting that P. tricornutum has developed highly efficient mechanisms to achieve this specific fatty acid composition. Using in vivo labelling techniques, Arao & Yamada [22] showed that four different routes led to the synthesis of EPA in P. tricornutum and that the preferred route relied on intermediates of both the x6- and the x3-pathway (see Fig. 1). The results presented in Table 2 show that PtD5p and PtD6p are involved in both pathways and they support the preferred route, as the activity of PtD6p is higher with 18:2 D9,12 than with 18:3 D9,12,15 and that of PtD5p higher with 20:4 D8,11,14,17 than with 20:3 D8,11,14 . Nevertheless, the substrate specificities and selectivities of PtD5p and PtD6p cannot account for the nearly exclusive accumulation of EPA observed in P. tricornutum. Moreover, and in good agreement with Beaudoin et al. [43], the synthesis of side products suggests that all the enzymes involved in PUFAs biosynthesis are rather unspecific and modify all the different fatty acids they encounter (Table 2 and Fig. 5). Because of the relative insensitivity of front-end desaturases and elongases to the presence of an x3-double bond and of x3-desaturases to the structure of the carboxyl end of the fatty acids, the existence of separate x3- and x6-pathways as depicted in Fig. 1 becomes questionable. The results presented here on front-end desaturases together with those concerning x3-desaturase [42] and D6-elongase [20] are all in favour of a scenario in which the same enzymes are contributing to both pathways, which in fact do not exist separately under these conditions. As the activities of the enzymes modifying the carboxyl end (D5- and D6-desatu- rases and D6-elongase) and the methyl end of PUFAs (x3- desaturase) do not depend on the structure of the opposite end of the substrate, it is more likely that these enzymes act simultaneously at several steps of interconnected pathways. The resulting complex reaction network makes it impossible to separate an x6- from an x3-pathway as the x3-desatur- ase is pulling all the intermediates into the x3-pathway when simultaneously, the front-end desaturases and elongase are pushing towards the end-product. Accordingly, if these enzymes are expressed simultaneously in the same cellular compartment, it may not be appropriate to divide such a complex reaction network into separate x6- and x3-path- ways as depicted in Fig. 1. To conclude, we have cloned and functionally charac- terized a D5- and a D6-fatty acid desaturase involved in EPA biosynthesis. In addition, several lines of evidence presented in this study strongly suggest that front-end desaturases are not specific for the desaturation of a single fatty acid in a straight biosynthetic pathway but that they rather act simultaneously in both the x3- and x6-pathway and on other potential substrates. This low specificity leads in yeast to the synthesis of undesirable side-products, which implies that the organisms accumulating a single PUFA like P. tricornutum have developed highly effective strategies to channel all biosynthetic intermediates towards the accumulation of a single end-product. In view of these results, elucidating and understanding the regulatory mechanisms leading to these highly selective accumula- tions becomes a clear prerequisite in order to implement PUFA biosynthesis in oilseed crops such as linseed or rapeseed. ACKNOWLEDGEMENTS This research has been supported by a Marie Curie Fellowship of the European Community programme Human Potential under the contract number HPMF-CT-1999-00148. We thank Dr T. K. Zank for providing the pY2PSE1 clone and BASF Plant Science GmbH (Ludwigshafen, Germany) for providing the k-ZAP Express library and for performing the random sequencing. REFERENCES 1. Scala, S. & Bowler, C. (2001) Molecular insights into the novel aspects of diatom biology. Cell. Mol. Life Sci. 58, 1666– 1673. 2. Tre ´ ger,P.,Nelson,D.M.,VanBennekom,A.J.,DeMaster,D., Leynaert,A.&Que ´ guiner, B. (1995) The silica balance in the world ocean: a reestimate. Science 268, 375–379. 3. Apt, K.E. & Behrens, P. (1999) Commercial developments of microalgal biotechnology. J. Phycol. 35, 215–226. 4. Apt, K.E., Kroth-Pancic, P.G. & Grossman, A.R. (1996) Stable nuclear transformation of the diatom Phaeodactylum tricornutum. Mol. Gen. Genet. 252, 572–579. 5. Falciatore, A., Casotti, R., Leblanc, C., Abrescia, C. & Bowler, C. (1999) Transformation of Nonselectable Reporter Genes in Marine Diatoms. Mar. Biotechn 1, 239–251. 6. Zaslavskaia, L.A. & Lippmeier, J.C. (2000) Transformation of the diatom Phaeodactylum tricornutum (Bacillariophyceae) with a variety of selectable marker and reporter genes. J. Phycol. 36,379– 986. 7. Falciatore, A., d’Alcala, M.R., Croot, P. & Bowler, C. (2000) Perception of environmental signals by a marine diatom. Science 288, 2363–2366. 8. Zaslavskaia, L.A., Lippmeier, J.C., Shih, C., Ehrhardt, D., Grossman, A.R. & Apt, K.E. (2001) Trophic conversion of an obligate photoautotrophic organism through metabolic engineer- ing. Science 292, 2073–2075. 4112 F. Domergue et al. (Eur. J. Biochem. 269) Ó FEBS 2002 9. Molina Grima, E., Robles Medina, A., Gime ´ nez Gime ´ nez, A. & Iba ´ n ˜ ez Gonza ´ lez, M.J. (1996) Gram-scale purification of eicosa- pentaenoic acid (EPA, 20: 5n)3) from wet Phaeodactylum tricor- nutum UTEX 640 biomass. J. Appl. Phycol. 8, 359–367. 10. Carlson, S.E., Werkman, S.H., Peeples, J.M., Cooke, R.J. & Tolley, E.A. (1993) Arachidonic acid status correlates with first year growth in preterm infants. Proc. Natl Acad. Sci. U S A 90, 1073–1077. 11. Crawford, M. (2000) Placental delivery of arachidonic and docosahexaenoic acids: implications for the lipid nutrition of preterm infants. Am.J.Clin.Nutr.71, 275S–284S. 12. Kinsella, J.E., Lokesh, B., Broughton, S. & Whelan, J. (1990) Dietary polyunsaturated fatty acids and eicosanoids: potential effects on the modulation of inflammatory and immune cells: an overview. Nutrition 6, 24–44. 13. Simopoulos, A.P. (1991) Omega-3 fatty acids in health and disease andingrowthanddevelopment.Am.J.Clin.Nutr.54, 438–463. 14. Simopoulos, A.P. (1999) Evolutionary aspects of omega-3 fatty acids in the food supply. Prostaglandins Leukot. Essent. Fatty Acids 60, 421–429. 15. Abbadi, A., Domergue, F., Meyer, A., Riedel, K., Sperling, P., Zank, T.K. & Heinz, E. (2001) Transgenic oilseeds as sustainable source of nutritionally relevant C20 and C22 polyunsaturated fatty acids? Eur. J. Lipid Sci. Technol. 103, 106–113. 16. Napier, J.A. & Michaelson, L.V. (2001) Genomic and funtional characterization of polyunsaturated fatty acid biosynthesis in Caenorhabditis elegans. Lipids 36, 761–766. 17. Huang, Y.S., Chaudhary, S., Thurmond, J.M., Bobik, E.G. Jr, Yuan, L., Chan, G.M., Kirchner, S.J., Mukerji, P. & Knutzon, D.S. (1999) Cloning of delta 12- and delta 6-desaturases from Mortierella alpina and recombinant production of gamma-lino- lenic acid in Saccharomyces cerevisiae. Lipids 34, 649–659. 18. Parker-Barnes, J.M., Das, T., Bobik, E., Leonard, A.E., Thur- mond, J.M., Chaung, L.T., Huang, Y.S. & Mukerji, P. (2000) Identification and characterization of an enzyme involved in the elongation of n-6 and n-3 polyunsaturated fatty acids. Proc. Natl Acad.Sci.USA97, 8284–8289. 19. Girke, T., Schmidt, H., Zahringer, U., Reski, R. & Heinz, E. (1998) Identification of a novel delta 6-acyl-group desaturase by targeted gene disruption in Physcomitrella patens. Plant J. 15, 39– 48. 20. Zank, T.K., Zahringer, U., Beckmann, C., Pohnert, G., Boland, W., Holtorf, H., Reski, R., Lerchl, J. & Heinz, E. (2002) Cloning and functional characterisation of an enzyme involved in the elongation of delta 6-polyunsaturated fatty acids from the moss Physcomitrella patens. Plant J. 31, 1–15. 21. Moreno, V.J., De Moreno, J.E.A. & Brenner, R.R. (1979) Bio- synthesis of unsaturated fatty acids in the diatom Phaeodactylum tricornutum. Lipids 14, 15–19. 22. Arao, T. & Yamada, M. (1994) Biosynthesis of polyunsaturated fatty acids in the marine diatom, Phaeodactylum tricornutum. Phytochemistry 35, 1177–1181. 23. Guillard, R.R.L. (1975) Culture of phytoplankton for feeding marine invertebrates. In Culture of Marine Invertebrate Animals (Chanley, M.H., ed.), pp. 29–60. NY. Plenum Press. 24. Shanklin, J. & Cahoon, E.B. (1998) Desaturation and related modifications of fatty acids. Annu. Rev. Plant Physiol. Plant Mol. Biol. 49, 611–641. 25. Donahue, T.F. & Cigan, A.M. (1990) Sequence and structural requirements for efficient translation in yeast. Meth Enzymol. 185, 366–372. 26. Vernet,T.,Dignard,D.&Thomas,D.Y.(1987)Afamilyofyeast expression vectors containing the phage f1 intergenic region. Gene 52, 225–233. 27. Bro ¨ ker, M., Bauml, O., Gottig, A., Ochs, J., Bodenbenner, M. & Amann, E. (1991) Expression of the human blood coagulation protein factor XIIIa in Saccharomyces cerevisiae: dependence of the expression levels from host-vector systems and medium con- ditions. Appl. Microbiol. Biotechnol. 34, 756–764. 28. Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, J.G., Smith, J.A., Struhl, K., Albright, L.M., Coen, D.M.&Varki,A.(1995)Current Protocols in Molecular Biology. John Wiley and Sons, New York. 29. Sperling, P., Lee, M., Girke, T., Zahringer, U., Stymne, S. & Heinz, E. (2000) A bifunctional delta 6-fatty acyl acetylenase/ desaturase from the moss Ceratodon purpureus. A new member of the cytochrome b5 superfamily. Eur. J. Biochem. 267, 3801–3811. 30. Sakuradani, E., Kobayashi, M. & Shimizu, S. (1999) Delta 6-fatty acid desaturase from an arachidonic acid-producing Mortierella fungus. Gene cloning and its heterologous expression in a fungus, Aspergillus. Gene 238, 445–453. 31. Saito, T., Morio, T. & Ochiai, H. (2000) A second functional delta 5 fatty acid desaturase in the cellular slime mould Dictyostelium discoideum. Eur. J. Biochem. 267, 1813–1818. 32. Sperling, P. & Heinz, E. (2001) Desaturases fused to their electron donor. Eur. J. Lipid Sci. Technol. 103, 158–180. 33. Shanklin, J., Whittle, E. & Fox, B.G. (1994) Eight histidine residues are catalytically essential in a membrane-associated iron enzyme, stearoyl-CoA desaturase, and are conserved in alkane hydroxylase and xylene monooxygenase. Biochemistry 33, 12787– 12794. 34. Nakai,K.&Horton,P.(1999)PSORT:aprogramfordetecting sorting signals in proteins and predicting their subcellular locali- zation. Trends Biochem. Sci. 24, 34–36. 35. Lederer, F. (1994) The cytochrome b 5 -fold: an adaptable module. Biochimie 76, 674–692. 36. Hastings, N., Agaba, M., Tocher, D.R., Leaver, M.J., Dick, J.R., Sargent, J.R. & Teale, A.J. (2001) A vertebrate fatty acid desa- turase with delta 5 and delta 6 activities. Proc. Natl Acad. Sci. USA 98, 14304–14309. 37. Watts, J.L. & Browse, J. (1999) Isolation and characterization of a delta 5-fatty acid desaturase from Caenorhabditis elegans. Arch. Biochem. Biophys. 362, 175–182. 38. Napier, J.A., Sayanova, O., Sperling, P. & Heinz, E. (1999) A growing family of cytochrome b 5 -domainfusionproteins.Trends Plant Sci. 4, 2–4. 39. de Antueno, R.J., Knickle, L.C., Smith, H., Elliot, M.L., Allen, S.J., Nwaka, S. & Winther, M.D. (2001) Activity of human delta 5 and delta 6 desaturases on multiple n-3 and n-6 polyunsaturated fatty acids. FEBS Lett. 509, 77–80. 40. Reed, D.W., Schafer, U.A. & Covello, P.S. (2000) Characteriza- tion of the Brassica napus extraplastidial linoleate desaturase by expression in Saccharomyces cerevisiae. Plant Physiol. 122, 715– 720. 41. Meesapyodsuk, D., Reed, D.W., Savile, C.K., Buist, P.H., Ambrose, S.J. & Covello, P.S. (2000) Characterization of the regiochemistry and cryptoregiochemistry of a Caenorhabditis elegans fatty acid desaturase (FAT-1) expressed in Saccharomyces cerevisiae. Biochemistry 39, 11948–11954. 42. Kang,Z.B.,Ge,Y.,Chen,Z.,Cluette-Brown,J.,Laposata,M., Leaf, A. & Kang, J.X. (2001) Adenoviral gene transfer of Caenorhabditis elegans n-3 fatty acid desaturase optimizes fatty acid composition in mammalian cells. Proc. Natl Acad. Sci. USA 98, 4050–4054. 43. Beaudoin, F., Michaelson, L.V., Hey, S.J., Lewis, M.J., Shewry, P.R., Sayanova, O. & Napier, J.A. (2000) Heterologous recon- stitution in yeast of the polyunsaturated fatty acid biosynthetic pathway. Proc. Natl Acad. Sci. USA 97, 6421–6426. Ó FEBS 2002 D5- and D6-fatty acid desaturases from diatom (Eur. J. Biochem. 269) 4113 . genes encoding fatty acid desaturases involved in eicosapentaenoic acid biosynthesis. Using a combination of PCR, mass sequencing and library screening, the coding sequences of two desaturases. Cloning and functional characterization of Phaeodactylum tricornutum front-end desaturases involved in eicosapentaenoic acid biosynthesis Fre ´ de ´ ric Domergue 1, *,. organism for cloning the genes encoding the different fatty acid desaturases involved in EPA biosynthesis and identified two sequences coding for fatty acid desaturases with D5- and D6-regioselectivities.