A Bifunctional Δ12,Δ15-Desaturase from Acanthamoeba castellanii Directs the Synthesis of Highly Unusual n-1 Series Unsaturated Fatty Acids*

The free-living soil protozoon Acanthamoeba castellanii synthesizes a range of polyunsaturated fatty acids, the balance of which can be altered by environmental changes. We have isolated and functionally characterized in yeast a microsomal desaturase from A. castellanii, which catalyzes the sequential conversion of C16 and C18 Δ9-monounsaturated fatty acids to di- and tri-unsaturated forms. In the case of C16 substrates, this bifunctional A. castellanii Δ12,Δ15-desaturase generated a highly unusual fatty acid, hexadecatrienoic acid (16:3Δ9,12,15(n-1)). The identification of a desaturase, which can catalyze the insertion of a double bond between the terminal two carbons of a fatty acid represents a new addition to desaturase functionality and plasticity. We have also co-expressed in yeast the A. castellanii bifunctional Δ12,Δ15-desaturase with a microsomal Δ6-desaturase, resulting in the synthesis of the highly unsaturated C16 fatty acid hexadecatetraenoic acid (16:4Δ6,9,12,15(n-1)), previously only reported in marine microorganisms. Our work therefore demonstrates the feasibility of the heterologous synthesis of polyunsaturated fatty acids of the n-1 series. The presence of a bifunctional Δ12,Δ15-desaturase in A. castellanii is also considered with reference to the evolution of desaturases and the lineage of this protist.

The biosynthesis of polyunsaturated fatty acids (PUFAs) 2 is catalyzed by two key enzymatic reactions, acyl chain elongation and fatty acid desaturation (1). This latter reaction brings about the insertion of double bonds (usually in a methylene-interrupted manner) into substrate fatty acids, thus altering their physiochemical and biological properties (2,3). The study of fatty acid desaturases has revealed a number of key observations. First, apart from the acyl carrier protein desaturases found in the stroma of higher plant plastids, all fatty acyl desaturases are membrane-bound (2). Second, all desaturases require molecular oxygen and an electron donor (usually cytochrome b 5 or ferredoxin), and finally, the primary deduced amino acid sequences of many diverse desaturases contain highly conserved motifs, namely the three histidine boxes hypothesized to act as di-iron co-ordinating centers (4).
Some distinct aspects of specific desaturases are also used to categorize the different enzymes into subgroups. For example, the presence of an N-terminal cytochrome b 5 domain is considered to be diagnostic for the "front-end" class of desaturases involved in the biosynthesis of long chain polyunsaturated fatty acids (5,6). Equally, mechanistic properties can be used to separate desaturases into those requiring acyl-CoA substrates, and those which utilize glycerolipid-linked substrates (i.e. where the fatty acid substrate is esterified to a glycerol backbone) (7). This distinction appears particularly marked in the case of microsomal desaturases; in animals such enzymes are predominantly acyl-CoA-dependent, whereas in higher plants, microsomal desaturases use phosphoglyceride substrates (7,8). Relatively little is known about the processes by which desaturases "measure" regiospecificity (i.e. the position of the introduced double bond), though a number of pioneering studies have biochemically defined the regiospecificity of several plant desaturases on the basis of either carboxy-or methyl-counting (9 -11). More recently, Covello and colleagues (12, 13) have exemplified a classification based on several different mechanisms for desaturation. This includes the ⌬ x desaturases, which introduce a double bond (⌬), x carbons from the carboxyl end, and -x enzymes, which desaturate x carbons from the methyl terminus. In these two classes, the enzymes appear to use the end of the fatty acid (either carboxyl or methyl terminus) as a chemical position of reference. Two additional classes of desaturase appear to use pre-existing double bonds as a reference point; these are the v ϩ x enzymes, where the double bond is introduced x carbons from the pre-existing double bond (v) toward the methyl-end of the fatty acid. Conversely, v Ϫ x desaturases introduce a double bond x carbons from the pre-existing double bond v toward the carboxyl end; these are the so-called front-end desaturases (5,9,14).

* This work was supported in part by grants from the Biotechnology and
Biological Sciences Research Council (BBSRC) of the United Kingdom and the European Union FP6 programme, via the LIPGENE project C Contract FOOD-CT-2003-505944). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. This study focuses on membrane desaturation in the protist Acanthamoeba castellanii. Acanthamoeba spp. are among the most prevalent protozoa found in the environment and are considered to be amphizoic organisms based on the ability to exist both as a free-living amoeba and a parasitic pathogen (15). A. castellanii is a small amoeba ubiquitous in soils, where it is able to adapt to temperature fluctuations and is thus capable of vegetative growth over a wide range of temperatures (10 -32°C). Unusually for an animal, A. castellanii can synthesize linoleic acid de novo (16); this then acts as a precursor for the synthesis of other (mainly C 20 ) n-6 polyunsaturated fatty acids. A. castellanii is a useful model to study several aspects of fatty acid desaturation. Firstly, previous studies have shown that changes in two environmental conditions (lowered temperature, increased O 2 concentration) independently resulted in the rapid conversion of oleic acid (18:1(n-9)) to linoleic acid (18: 2(n-6)) through the action of an inducible ⌬12-desaturase. This microsomal ⌬12-desaturase is shown to utilize oleate at the sn-2 position of phosphatidylcholine preferentially as a substrate (17). Additionally, A. castellanii synthesizes long chain PUFAs such as arachidonic acid (20:4(n-6)) via the alternative ⌬8 pathway, using an unusual C 20 ⌬8-desaturase instead of the more prevalent ⌬6-desaturase-dependent pathway (18,19). In view of these properties, we have isolated desaturases from A. castellanii and functionally characterized their enzyme activities. Here we report the identification of a novel type of desaturase, which carries out two sequential desaturations on both C 18 and C 16 substrates, the latter resulting in the synthesis of a highly unusual 16:3⌬ 9,12,15 (n-1) triene.
Nucleic Acid Manipulation and PCR-based Cloning-Total RNA was extracted from cells using a RNAeasy plant mini kit (Qiagen). First strand cDNA was synthesized from total RNA using the SMART RACE cDNA amplification kit (Clontech) according to the manufacturer's instructions. Single-stranded cDNAs were amplified with primers specific to each desaturase gene as follows: the reactions were heated to 95°C for 2 min followed by 30 cycles at 94°C for 30 s; 30 s at temperatures ranging from 55 to 72°C according to the primer design; 72°C for 2 min; and then a single step at 72°C for 10 min. PCR amplification products were cloned into a TOPO vector (Invitrogen) and verified by sequencing. cDNAs were amplified as described previously (21) with degenerate primers (shown below) designed to the histidine boxes of previously characterized desaturases (14). The 5Ј-and 3Ј-ends of putative desaturases were amplified with the SMART RACE cDNA amplification kit (Clontech).
The derived 5Ј-and/or 3Ј-cDNA sequence data were used to synthesize full-length copies of the putative desaturases. Genespecific primers were designed to the 5Ј-and 3Ј-ends of the coding regions of the corresponding nucleotide sequences, with restriction sites to facilitate cloning into the yeast and plant vectors (see below). Forward primers for cloning into yeast expression vector pYES2 (Invitrogen) were designed to contain a G/A at position Ϫ3 and a G at position ϩ4 to improve translation initiation in eukaryotic cells.
Functional Expression in Yeast-Open reading frames (ORFs) encoding putative desaturase activities were introduced in Saccharomyces cerevisiae strain W303-1A by a lithium acetate method. Cultures were grown at 22°C in the presence of 2% (v/v) raffinose, and expression of the transgenes was induced by the addition of galactose to 2% (w/v) in the presence of 0.5 mM corresponding non-esterified fatty acid and 1% (w/v) tergitol Nonidet P-40 (Sigma) as described (22).
Lipid Separation and Fatty Acid Analysis-Intact lipids were extracted according to Refs. 7 and 23. Lipid classes were separated by 2-D TLC as before (24) and identified by co-chromatography with authentic standards and by specific color reactions (25). Specific lipids were scraped off the plate, eluted, and either methylated for gas-liquid chromatography (GLC) or subjected to phospholipase A 2 (Naja mossambica; Sigma) digestion followed by TLC separation of the products. Total fatty acids of yeast samples were extracted and methylated as described previously (22) and analyzed (GLC) of methyl ester derivatives. Acyl-CoA analysis was carried out as described previously (22).
Mass Spectrometry-The 2,4-dimethyloxazoline (DMOX) derivatives (26) were submitted to GC-MS with a Hewlett Packard 5890 Series II plus gas chromatograph attached to an HP model 5989 MS engine. The latter was used in the electron impact mode at 70 eV with a source temperature of 250°C. The GC was fitted with on-column injection. For all derivatives, a capillary column of fused silica coated with Supelcowax 10 TM (30 m ϫ 0.25 mm, 0.25-m film; Supelco UK, Poole, UK) was used. The column temperature was programmed from 80°C (hold 3 min) to 180°C at 20°C/min, then to 250°C at 2°C/min, then for 15 min at 280°C. Helium was the carrier gas at a constant flow rate of 1 ml/min and initial head pressure of 9.4 p.s.i. The injector temperature was 80°C, and the transfer-line temperature was 260°C.

RESULTS
Using publicly available data derived from an A. castellanii EST sequencing project, we identified via BLAST searches one EST sequence (designated ACE00001892), which showed some low homology to previously identified ⌬12 fatty acid desaturase genes. This partial sequence was used to design 5Ј-and 3Ј-RACE primers to amplify the ends of this putative fatty acid desaturase from cDNA generated from A. castellanii RNA. The resulting deduced full ORF encoded a polypeptide of 407 amino acids and also contained the three "histidine box" motifs conserved in all membrane-bound desaturases (4). The putative A. castellanii desaturase was predicted to contain several (at least three) transmembrane domains, with a topology likely to place the three highly conserved histidine box motifs on the cytosolic face of the endoplasmic reticulum membrane. In agreement with this, the A. castellanii sequence lacked any obvious cleavable targeting signals but did contain a motif recognized as a potential C-terminal endoplasmic reticulum dilysine-type retention signal (K-H-H-K-A-H) (27). The putative desaturase did not contain a cytochrome b 5 domain, present in all known examples of front-end desaturases (14), including the recently characterized alternative pathway C 20 ⌬8-desaturase of A. castellanii (22). A comparison of the A. castellanii desaturase sequence (designated AcD12) with the most related microsomal desaturases is shown in Fig. 1A, and a phylogenetic tree constructed with other examples of more distantly related desaturases is shown in Fig. 1B. As shown in Fig. 1B, AcD12 is more similar to endoplasmic reticulum-localized desaturases and forms a separate branch within the group of FAD2 microsomal desaturases from plants, implying a lineage more closely associated with higher plants than fungi or invertebrates (14,28).
To elucidate the function of the A. castellanii putative desaturase, the entire ORF was cloned into the yeast expression vector pYES2 to give a construct designated pYAcD12, and the S. cerevisiae strain W303-1A was transformed with either pYAcD12 or the empty vector pYES2. In addition, the borage FAD2 microsomal ⌬12-desaturase (29) ORF was cloned into pYES2 and transformed in yeast as a further control. Transformed yeast cells were identified on the basis of uracil auxotrophy, and expression of the A. castellanii or borage desaturase was induced by the addition of galactose to the yeast cultures. Analysis of total fatty acid methyl esters (FAMEs) by GLC revealed that in the presence of galactose, yeast cells transformed with pYAcD12 produced additional abundant fatty acids (Fig. 2B), absent in uninduced cells ( Fig. 2A). The two most abundant novel peaks present in the FAMEs of yeast expressing this desaturase (Fig. 2B) were provisionally identified as linoleic acid (18:2⌬ 9,12 (n-6)) and hexadecadienoic acid (16:2⌬ 9,12 (n-4)), based on co-migration with authentic standards (data not shown). Additional peaks of identical retention time were also observed following the expression of the borage FAD2 ⌬12-desaturase (Fig. 2C). The exact mass of these novel fatty acid methyl esters was determined by GC-time-of-flight MS and found to match the known mass of 18:2 and 16:2 FAMEs (294 and 266 m/z, respectively). In the case of the A. castellanii desaturase, in addition to these two fatty acids, two further products could be observed (Fig. 2B); these additional peaks were not observed with the expression of the borage FAD2 ⌬12-desaturase (Fig. 2C). Of these, one co-migrated with an authentic standard for ␣-linolenic acid (ALA; 18:3⌬ 9,12,15 (n-3)) and had a m/z of 292. The fourth peak, which migrated after 16:2 but before 18:0, had a similar but not identical retention time to hexadecatrienoic acid from Arabidopsis (16:3⌬ 7,10,13 (n-3)). This peak also yielded the same mass (m/z 264) as for hexadecatrienoic acid methyl ester, making it likely that it represented a desaturation product of 16:2(n-4).
Further confirmation of the regio-positions of the double bonds of the non-native fatty acids synthesized in the yeast expression system was provided by GC-MS analysis of DMOX derivatives, with the mass spectra and predicted fragmentation of the DMOX derivatives of the fatty acids described above shown in Fig. 3. Fig. 3A shows the deduced structure for ALA (molecular ion of m/z 331), with the double bonds in positions 9, 12, and 15 located by gaps of 12 atomic mass units between m/z 196 and 208, 236 and 248, and 276 and 288, respectively. This is in agreement with a previous report (30). In the case of the C 16 dienoic acid (molecular ion of m/z 305), the similar presence of ion pairs of m/z 196, 208 and 236, 248 indicate the presence of double bonds at the ⌬9 and ⌬12 positions and therefore confirmed this fatty acid as 16:2⌬ 9,12 (n-4) (Fig. 3B). Analysis of DMOX derivatives of the presumed 16:3 fatty acid (Fig. 3C) yielded a molecular ion of m/z 303 and indicated that this C 16 fatty acid contained one further double bond, most likely, in either position 14 or 15. Double bonds near either end of the chain no longer give fragmentations with a gap of 12 atomic mass units to define their position (31). However, prominent fragments formed by cleavage ␤ to the double bonds, i.e. at m/z 182, 222, and 262, are supportive of these being present in positions 9, 12, and 15 of the chain, respectively. The mass spectra of the methyl ester and pyrrolidide derivatives (data not presented) are also consistent with this interpretation as is the GC retention time relative to that of the (n-4) analogue. Therefore, on mass spectrometric and biosynthetic grounds, this fatty acid was identified as 16:3⌬ 9,12,15 (Fig. 3C). By overloading the column on the GC-MS, a further minor component was identified as 18:3(n-1) or 18:3⌬ 11,14,17 by the mass spectrum of its DMOX derivative; the diagnostic ions for the double bonds were similar to those in the spectrum of the 16:3(n-1) derivative, except that they were located 28 atomic mass units higher as expected. Presumably, this C 18 fatty acid is produced by chain elongation of 16:3 (n-1). From these data we concluded that the A. castellanii desaturase encodes a novel bifunctional ⌬12,⌬15-desaturase, which recognizes both C 16 and C 18 substrates to generate n-4 (11.5%) and n-1 (7.0%) (from 16:1(n-7)) or (n-6) (16.0%) and n-3 (7.3%) (from 18:1(n-9)) in yeast cells expressing this desaturase (respective conversion rates shown in brackets).
The temporal accumulation of the different fatty acids generated by the A. castellanii desaturase was investigated by GLC analysis of FAMEs after relatively short periods of induction to establish whether the accumulation of ⌬12-desaturation products either preceded or was simultaneous with the accumulation of ⌬15-desaturated fatty acids. Analysis of yeast cells expressing the AcD12 was carried out after 1, 2, 3, 4, 5, and 6 h and onwards and revealed that whereas ⌬12-desaturated fatty acids were already detectable after only 1 h of induction, ⌬15desaturated fatty acids only became detectable after 3 h (Fig. 4; data only plotted for first 8 h). This is particularly obvious for the synthesis of either 18:2(n-6) or 18:3(n-3), where the accumulation of the former fatty acid started to plateau after 3 h at ϳ10% of total unsaturated fatty acids. However, it is only at this same time point that the first accumulation of 18:3 (1.3% of total unsaturated fatty acids) was observed; prior to this, no ⌬15-desaturation products were detected (Fig. 4). Maximal accumulation of 18:2(n-6) occurred at the terminal time point of 52 h (22.4% of total unsaturated fatty acids), compared with 7.1% of total unsaturated fatty acids for 18:3(n-3) at the same time point (data not shown). Similar kinetics was observed for the accumulation of 16:2(n-6) versus 16:3(n-3). These time courses are consistent with a normal precursor-product metabolic relationship, indicating the sequential synthesis of first ⌬12and then ⌬15-desaturated fatty acids.
The substrate specificity of the AcD12 was also examined in terms of whether or not the enzyme used phosphoglycerolinked acyl chains or acyl-CoA substrates. In the case of higher plant FAD2 microsomal ⌬12-desaturases, it is well known that these enzymes utilize substrate fatty acids esterified to a phosphoglycerolipid backbone, though the FAD2 enzyme does not display a preference for either sn-1 or sn-2 positions (2, 11). In contrast, biochemical characterization of ⌬12-desaturases from insects has indicated that such enzymes utilize acyl-CoA substrates (3). Previous biochemical characterization of A. castellanii membrane fractions had identified a microsomal n-6/ ⌬12-desaturase activity, which utilized oleate linked to the sn-2 position of PC as a substrate (17,20). Having isolated a cDNA, which might represent such an activity, we analyzed lipid species from transgenic yeast expressing the A. castellanii ⌬12,⌬15-desaturase. Polar lipids extracted from such cells revealed the presence of ⌬12and ⌬15-desaturated fatty acids in all phospholipid classes examined (PC, phosphatidylethanolamine, phosphatidylinositol, and PS). However, some enrichment of these products was observed in PC and phosphatidylethanolamine. Further analysis of the positional distribution of the ⌬12-desaturation products in PC revealed a strong preference for the sn-2 position, with a 5.4-fold accumulation of linoleic acid at this position (10.3% sn-2, 1.9% sn-1). Equally, the accumulation of 16:2(n-4) occurred with a 2.4-fold preference at the sn-2 position of PC. The distribution of ⌬15-desaturated fatty acids (either C 16 or C 18 ) did not display any clear preferential accumulation at either position; this is likely to represent lipid remodeling catalyzed by activities such as lyso-phophatidylcholine; acyl-CoA acyltransferases, which facilitate return acyl exchange between the acyl-CoA pool and the sn-2 position of PC (19).
To complement the polar lipid class analysis, acyl-CoA profiles were determined for yeast expressing the A. castellanii desaturase; identical methods have recently been used to identify an acyl-CoA-dependent ⌬6-desaturase from Ostreococcus tauri (8). As can be seen in Fig. 5A, in the absence of galactose to induce the expression of the AcD12, the acyl-CoA pool of S. cerevisiae contains only 14:0, 16:0, 16:1, 18:0, and 18:1 fatty acids. However, after expression of the desaturase was induced for 20 h, in the absence of any exogenous substrates, the presence of several new acyl-CoAs could be detected (Fig. 5B). These were identified as 16:2(n-4), 16:3(n-1), 18:2(n-6), and 18:3(n-3) on the basis of co-migration with authentic standards. The stoichiometry of the levels of these four novel acyl-CoA was very similar to that observed for the accumulated total fatty acids (i.e. there were more ⌬12-desaturation products than ⌬15-desaturation products). To further examine the possibility of the AcD12 using an acyl-CoA substrate, exogenous substrate (oleic acid in the form of sodium soap) was added to the yeast cultures. After only 5 min, this substrate was detected in the acyl-CoA pool (Fig. 5C), as witnessed by the significant enhancement to the levels of this acyl-CoA (Fig. 5, compare  panels B and C). However, the elevated levels of this acyl-CoA did not result in any increase in the levels of 18:2, as might be predicted if the desaturase directly used an acyl-CoA substrate. Rather, the levels of 18:2 and 18:3 in the acyl-CoA pool remained almost unchanged over the course of the following 24 h (Fig.  5, C-E). Equally, at the 24 h time point, oleoyl-CoA had almost returned to basal levels (presumably because the pulse of exogenous 18:1 had been exhausted), again with no alteration to the levels of 18:2 and 18:3 in the CoA pool (Fig. 5E).
As a corollary to the use of acyl-CoA profiling to determine nonphopholipid-linked desaturation, we employed an in vivo approach to monitoring substrate requirements. This system uses the principles and methodologies established by Domergue et al. (7) in which the presence of an acyl-CoA desaturase is detected by its ability to provide substrate for acyl-CoAdependent fatty acid elongation, whereas phospholipid-dependent desaturation does not directly generate such a substrate. A. castellanii AcD12 was co-expressed in yeast with the C 18 ⌬9-elongase from Isochrysis galbana (32). Thus, if the AcD12 enzyme utilized oleic acid-CoA, the resulting desaturation product (linoleic acid-CoA) would serve as a direct substrate for elongation by the I. galbana elongase (which only accepts 18:2 or 18:3 as substrates). Although expression of the A. castellanii enzyme generated significant levels of desaturation products, only 29% of these were elongated by the Isochrysis enzyme. In contrast, when 18:2 was supplied directly to yeast culture media, 57% of this substrate was elongated by the Isochrysis elongase, consistent with the activation of exogenously supplied fatty acids to acyl-CoA forms (cf. Fig. 5C). Thus, these data are indicative of the A. castellanii enzyme not carrying out desaturation reactions on acyl-CoA substrates but instead using fatty acids linked to phospholipids. Neither 16:3(n-1) nor 18:3(n-1) appear to have been identified previously, although a 16:3(n-1) fatty acid has been hypoth- esized to be involved in the biosynthesis of sorgoleone, a lipophilic p-benzoquinone present in the root exudates of Sorghum bicolor and known to act as an allelochemical (33). Some marine species accumulate other examples of (n-1) unsaturated fatty acids. In particular, 16:4⌬ 6,9,12,15 has been reported in a variety of marine diatoms (34 -36), dinoflagellates (37), and Rhodophyta (38). Both 16:4(n-1) and 18:4(n-1) are known to be minor but common constituents of fish oils; 18:5(n-1) has recently been detected in a fish oil concentrate. 3 To facilitate the further study of this unusual C 16 PUFA, we attempted to reconstitute its synthesis in yeast by co-expressing the bifunctional ⌬12,⌬15 Acanthamoeba-desaturase with the borage microsomal ⌬6-desaturase (22). As shown in Table 1, this coexpression generated low, but significant, levels of 16:4 (n-1). The presence of 16:4(n-1) was confirmed by co-migration with authentic standards (Thalassiosira pseudonana fatty acids (39)) and an expected exact mass for this FAME (m/z 264). As a comparison, the same borage ⌬6-desaturase was co-expressed with the borage FAD2 ⌬12-desaturase. This resulted in the accumulation of 16:2 (predominantly n-7, low n-4) but no (n-1) fatty acids (Table 1).

DISCUSSION
We have identified a bifunctional ⌬12,⌬15-desaturase from the free living soil amoeba A. castellanii. This desaturase appears to be distantly related to other microsomal ⌬12-desaturases, showing only moderate sequence homology to microsomal desaturases of higher plants and fungi (ϳ44% identity and ϳ54% similarity) and is very distantly related to the corresponding plastidial desaturases. It also shows lower homology (38% identity and 54% similarity) to the recently described bifunctional ⌬12,⌬15-desaturase from the fungus Gibberella fujikuroi (GenBank accession number DQ272516) (40) as well as other microsomal desaturases such as Arabidopsis -3, FAD6, and Caenorhabditis elegans FAT-2 and FAT-1. This separation between AcD12 and other desaturases may indicate that the bifunctional enzyme from A. castellanii represent a new lineage in the evolution of desaturases. Interestingly, the bifunctional desaturase identified from G. fujikuroi (synonym, Fusarium moniliforme) appears to act predominantly as a -3 desaturase and is restricted to C 18ϩ substrates (40). In that respect, the fungal desaturase does not generate 16:2(n-4) or 16:3(n-1) (40), highlighting the difference between that enzyme and the A. castellanii bifunctional desaturase described in this 3 W. W. Christie, unpublished data.  study. The AcD12 enzyme does not contain a cytochrome b 5 domain (Fig. 1A), which agrees with experiments using A. castellanii microsomes, where the desaturase component appears separate to the cytochrome b 5 protein (41). Given the bifunctional nature of the A. castellanii enzyme, it may represent an ancestral form, from which individual discrete activities were derived (as has been suggested in the case of the FAT-2 ⌬12and FAT-1 ⌬15-desaturases observed in C. elegans) (42). A similar scenario has been proposed for the bifunctional ⌬5,⌬6-desaturase identified from zebrafish (43), in which cytochrome b 5 fusion front-end desaturase is suggested to possibly represent the ancestral progenitor of distinct ⌬5and ⌬6-desaturases (42,44).
Several aspects of this bifunctional ⌬12,⌬15-desaturase are worthy of consideration. Firstly, the enzyme utilizes both C 16 and C 18 substrates. As a result of the bifunctional nature of the A. castellanii desaturase, not only are 18:2(n-6) and 16:2(n-4) synthesized but also 18:3(n-3) and 16:3(n-1). Therefore, the enzyme appears to catalyze two cycles of ⌬ x -type desaturation, inserting double bonds three carbons from pre-existing double bonds (sequential ⌬9 and ⌬12). An alternative activity, the v ϩ x desaturase specificity, appears to be ruled out by the inability to use cis-vaccenate (which should generate 18:2⌬ 11,14 ). In contrast, the parasitic protozoon Trypansoma brucei ⌬12-desaturase, which represents the closest evolutionary lineage to the protist A. castellanii, is shown to recognize C 16 , C 18 , and C 19 monounsaturated substrates, though no subsequent ⌬15-desaturation is reported (44). It is also very likely from the data presented in this present study that the A. castellanii ⌬12,⌬15-desaturase utilizes substrate fatty acids esterified to phospholipids, with preference for the sn-2 position during the ⌬12 activity. We found no evidence for the use of acyl-CoA substrates by this enzyme, which indicates that phosphoglycerolipid-linked desaturation, rather than acyl-CoA-dependent activities, exists in protist species like A. castellanii (17). The substrate preference of the G. fujikuroi bifunctional desaturase has not been described (40).
Given that glycerolipid-linked desaturation is generally considered to be a hallmark of higher plants, this may reflect some aspect of the evolutionary lineage of A. castellanii. Despite the absence in this organism of any trace of an organelle that resembles a chloroplast, lipid metabolism in A. castellanii has several plant-like traits (45). Polyunsaturated fatty acids such as 18:2(n-6) and 18:3(n-3) have been so far found mainly in chloroplasts or the cytosol of plants, algae and cyanobacteria. This might suggest that enzymes involved in the synthesis of these fatty acids (i.e. ⌬15/-3 and ⌬12/-6 fatty acid desaturases) could have been acquired by horizontal gene transfer from an oxygenic, phototrophic organism during endosymbiosis. Acanthamoeba spp. are well known predators of bacteria (via phagocytosis), whereas strains of A. castellanii isolated from soil and water sites exhibit predatory activity on cyanobacteria in culture (45). As cyanobacteria are considered to be closely related to the precursor of all chloroplasts in plants and algae, one possible explanation for the presence of ⌬12/-3-desaturase activity in Acanthamoeba would be that such genes entered an ancestral phagocytic amoeba through horizontal gene transfer, supporting the Doolittle hypothesis of "you are what you eat" (46). Previous biochemical studies on temperature-dependent oleate desaturation in A. castellanii identified a ⌬12-desaturase reaction as occurring on the sn-2 position of PC (17), in agreement with the functional characterization of the heterologously expressed A. castellanii desaturase reported here. However, it is also worth noting that an oleate desaturase induced by oxygen, rather than temperature, in A. castellanii appeared to be a -6 desaturase rather than a strict ⌬12-desaturase (41) raising the possibility that the A. castellanii genome encodes several oleate desaturases.
One remarkable aspect of the bifunctional activity of the A. castellanii desaturase is the generation of (n-1) fatty acids, such that the terminal double bond is positioned between the two carbons closest to the methyl-end of the fatty acid. Although the presence in S. bicolor of hexadecatrienoic acid (n-1) has previously been suggested as part of the biosynthesis of sorgoleone, the authors of that study noted that this unusual fatty acid must be synthesized by "as-yet-unknown fatty acid desaturases" (33). Whether S. bicolor contains an ortholog of the A. castellanii bifunctional desaturase, which generates The resulting percent fatty acid composition of transgenic yeast is given. Values represent the mean of three independent experiments with standard deviations; for several independent experiments (n ϭ 3), standard deviation is shown. Yeast strains containing episomal heterologous genes as shown: AcD12, A. castellanii ⌬12,⌬15-desaturase; BoD6, B. officinalis ⌬6-desaturase; BoD12, B. officinalis, FAD2 ⌬12-desaturase; Elo⌬9, I. galbana C 18 ⌬9-elongase. All strains were induced for expression via galactose. In two treatments, exogenous substrate in the form of 18:2 was added at a concentration of 500 M.
16:3(n-1), is currently unknown, as is the possibility that the amoeba may synthesize compounds analogous to sorgoleone. Another aspect of this current study was the demonstration of the synthesis of hexadecatetraenoic acid (16:4⌬ 6,9,12,15 (n-1)). The availability of in vivo-synthesized 16:3(n-1) allowed us to examine the possibility of synthesizing 16:4(n-1) through the co-expression of a ⌬6-desaturase with the A. castellanii bifunctional enzyme. This demonstrated a low, but significant, synthesis of 16:4(n-1), which previously has only been detected in marine organisms. Given the considerable interest in polyunsaturated fatty acids for the control of gene expression (47) or in the production of bioactive lipids (48), it would be of interest to determine whether (n-1) fatty acids have similar roles. Our proof-of-concept synthesis of 16:4(n-1) may represent the first steps toward addressing such questions.
A. castellanii synthesizes a range of polyunsaturated fatty acids, including C 20 PUFAs, though the role of such fatty acids (both n-6 and n-3) in the metabolism and lifecycle of the amoeba is not clear. The observation of a bifunctional ⌬12,⌬15desaturase in this amoeba may represent additional capability for the self-sufficient synthesis of a wide range of fatty acids, including n-3 PUFAs. Analysis of the fatty acid composition of A. castellanii confirmed the presence of n-3 fatty acids such as ␣-linolenic acid and eicosapentaenoic acid (data not shown). Irrespective of whether there is more than one ⌬12-desaturase in A. castellanii (see above and Ref.41), these n-3 polyunsaturates would be generated by activity of the ⌬12,⌬15-desaturase in vivo.
In conclusion, we have identified a bifunctional ⌬12,⌬15desaturase from the free living amoeba A. castellanii. In particular, this enzyme activity is capable of synthesizing n-6, n-3, n-4, and n-1 unsaturated fatty acids, and therefore, represents a new extension to the catalytic plasticity already found in the FAD2-like class of microsomal desaturases.