Identification of avian wax synthases

Background Bird species show a high degree of variation in the composition of their preen gland waxes. For instance, galliform birds like chicken contain fatty acid esters of 2,3-alkanediols, while Anseriformes like goose or Strigiformes like barn owl contain wax monoesters in their preen gland secretions. The final biosynthetic step is catalyzed by wax synthases (WS) which have been identified in pro- and eukaryotic organisms. Results Sequence similarities enabled us to identify six cDNAs encoding putative wax synthesizing proteins in chicken and two from barn owl and goose. Expression studies in yeast under in vivo and in vitro conditions showed that three proteins from chicken performed WS activity while a sequence from chicken, goose and barn owl encoded a bifunctional enzyme catalyzing both wax ester and triacylglycerol synthesis. Mono- and bifunctional WS were found to differ in their substrate specificities especially with regard to branched-chain alcohols and acyl-CoA thioesters. According to the expression patterns of their transcripts and the properties of the enzymes, avian WS proteins might not be confined to preen glands. Conclusions We provide direct evidence that avian preen glands possess both monofunctional and bifunctional WS proteins which have different expression patterns and WS activities with different substrate specificities.

Production of wax esters has already been observed in preen gland membranes of chicken [20] and goose [21] in vitro, so it could be assumed, that genes essential for wax ester biosynthesis are expressed in preen gland tissue. The respective genes have not been identified in birds yet, but wax ester synthase sequences (acyl-CoA: alcohol acyltransferases, AWAT, WS) have already been described in other organisms including mammals [22,23], plants [24][25][26][27], bacteria [28][29][30] and protozoa [31]. Mammalian enzymes with wax synthase activity have been found within members of both DGAT1 and DGAT2 type acyltransferase families [22,32]. DGAT (acyl-CoA:diacylglycerol acyltransferases) catalyze the final step in storage lipid biosynthesis of TAG, but the human DGAT1 is capable of synthesizing wax monoesters, diesters and retinylesters as well [32,33]. Human wax synthases AWAT1 and AWAT2 belong to the DGAT2 type family. Like DGAT1, AWAT2 is a multifunctional acyltransferase which shows in vitro acyl-CoA:monoacylglycerol acyltransferase (MOGAT), DGAT, WS and retinylester synthase activities [33]. Bacterial wax synthases are at least bifunctional enzymes conferring WS activity next to DGAT and low MOGAT activity [28,29,34].
Wax esters are excellent lubricants because of their high stability under high temperature and pressure and high resistance to hydrolysis [35]. Unlike saturated longchain monoesters, mono-unsaturated monoester or diester waxes combine good lubricity with good thermal and oxidative stability, high viscosity indices [36] and stability against lipases [37]. To achieve the renewable production of wax esters, currently attempts are made to identify new enzymes catalyzing respective esterification reactions [22,25,31,38]. Production of wax esters in oil crops [35,39] or microorganisms [38,40] might in future be able to surrogate fossil materials in technical industry.
Our studies identified WS genes of chicken, goose and barn owl, members of different bird families with distinct preen wax compositions. As the chicken genome was fully sequenced and assembled in March 2004 by the National Human Genome Research Institute [41,42], it served as a starting point for the identification of avian WS genes. Several sequences were successfully cloned and functionally analyzed in yeast cells.

Identification of putative wax synthases from avian organisms
Sequence similarity based searches conducted with human AWAT1, AWAT2 and DGAT1 sequences as queries against the annotated Gallus gallus proteome resulted in five full-length sequences. Using mRNA isolated from preen glands as starting material, we succeeded in cloning the respective cDNAs of GgWS2

In silico analysis of putative avian wax synthases
Comparison of the avian WS protein sequences with respective sequences from different organisms gave the results illustrated in Figure 1. It reveals that GgWS1 and GgWS2 possess the highest similarity to human acyltransferases of the DGAT2-type (up to 60%) while GgDGAT1 shows the highest identity to DGAT1 family members, namely 62% to human acyl-CoA:cholesterol acyltransferase (ACAT) HsACAT1, 43% to HsACAT2 and 16% to HsDGAT1. Avian WS4 and WS5 proteins share less than 15% sequence identity to both DGAT1 and DGAT2 family members and build an own branch of enzymes ( Figure 1). WS4 and WS5 proteins comprise 55% identity to each other and WS5 proteins are most similar to transmembrane protein 68, a protein of unknown function found in various organisms ( Figure 1).
The relation to the different mammalian acyltransferase families was reflected in further characteristics of the avian proteins like the molecular mass, the transmembrane structure and acyltransferase motifs. Conserved domain search revealed that all DGAT1 family members possessed an MBOAT (membrane bound Oacyltransferase) superfamily motif (Pfam cl00738) and the FYxDWWN motif (Figure 2a), which is a predicted acyl-CoA binding site identified in mammalian DGAT1  family members [43,44]. Furthermore, GgDGAT1 and the human ACAT proteins contained the (H/Y)SF motif [43] which might play a role in sterol binding. Unlike DGAT1 family members, DGAT2 family members like GgWS1 and GgWS2 were found to contain an acyltransferase superfamily motif (Pfam cl00357), that was found in WS4 and WS5 as well. In addition, the HPHG motif which is typical for DGAT2 family members and likely comprises a part of the active site [45] was also found in GgWS1 and GgWS2, while in WS4 and WS5 homologs the motif was modified to YYHG and FYHG, respectively ( Figure 2b). GgWS1 and GgWS2 have masses of about 40 kDa with one predicted N-terminal transmembrane domain (TMD) and their C-terminal parts stretched into the cytosol like other DGAT2 related proteins. On the other hand, GgDGAT1 is a 60 kDa protein with several predicted TMDs typical of DGAT1 related proteins. With regard to the transmembrane structure, the avian WS4 homologs resemble DGAT2 family members unlike WS5 homologs, which differed insofar as the TMD prediction indicated that the C-terminal part of the protein was not located in the cytosol.

Expression profiles of wax synthase sequences in chicken
Expression profiles of four chicken sequences were analyzed with different chicken tissues to reveal whether the sequences were preferentially expressed in preen glands. As the use of oligo(dT) primers did not lead to the amplification of any wax synthase sequence, genespecific primers were used to synthesize the respective cDNAs from 1 μg of RNA. The comparison of products from the different tissues indicated that GgWS1 and GgDGAT1 were almost exclusively expressed in the preen gland, while GgWS2 and GgWS4 were expressed in the four analyzed tissues ( Figure 3).

Functional expression in yeast
To analyze the identities of the avian proteins, the respective cDNAs were expressed in a yeast mutant strain lacking TAG synthesis. Transgenic yeast cultures were supplemented with fatty alcohols from 10 to 18 carbon atoms to enable wax ester production. Lipid analyses of the transgenic yeast cells suggest that most of the avian proteins were functionally expressed and caused accumulation of storage lipids in significant but different levels and patterns ( Figure 4a). Expression of GgWS1 resulted in the highest levels of wax esters which comprised 2 μmol wax ester/g fresh weight, while TAG was formed in very low levels only (Additional file 4). GgDGAT1 expressing yeast cells produced almost 1 μmol wax esters/g but no TAG (Additional file 5), whereas all WS4 homologs catalyzed the synthesis of higher levels of TAG (250 to 500 nmol/g) than of wax esters (100 to 250 nmol/g). The produced wax esters contained mainly dodecanol (12:0-OH) and tetradecanol (14:0-OH) esterified with palmitoleic (16:1) and oleic (18:1) acid irrespective of the expressed sequence. In contrast, TAG produced by WS4 homologs consisted of almost equal amounts of saturated and unsaturated fatty acyl residues. GgWS2 expressing cells contained storage lipids in very low levels similar to those of the control cells (Additional file 4). Additional feeding of GgWS2 expressing yeast cells with myristic acid (14:0) gave a 5fold increase in wax production while the wax level of control strains was not affected. Such stimulation was also observed in GgWS4 expressing cells (Additional File 6).
To reconstitute the avian wax biosynthesis in yeast cells, WS were co-expressed with the recently identified fatty acyl-CoA reductase GgFAR1 [46]. This enzyme has a specificity for 16:0-acyl-chains, but is also able to produce 14:0-OH in yeast cultures supplemented with 14:0-FA [46]. GgWS4 was chosen for co-expression because it performed the highest WS activities of all analyzed WS enzymes with long chain acyl-acceptors. Supplementation with 14:0-FA of cultures expressing both GgWS4 and GgFAR1 resulted in the production of fatty alcohols, 350 nmol TAG/g and 550 nmol wax esters per g fresh weight, while expression of GgFAR1 alone led only to the production of fatty alcohols (Figure 4b).

Properties of avian proteins
To analyze the substrate specificities of the different avian proteins, in vitro assays were performed with  Figure 2a shows the alignment of the FYxDWWN motif which is a potential acyl-CoA binding motif in proteins of the DGAT1 family [44]. Figure 2b represents the partially modified HPHG motif representing a potential part of the active site in DGAT2 family proteins [45]. NCBI  membranes of yeast cells expressing the respective sequences as enzyme source. WS activity was detectable in membranes, but not in soluble fractions, which is in line with the predicted transmembrane domains. Under standard assay conditions the WS activities were constant for at least 2 μg protein in an incubation time of 20 minutes at 35°C (data not shown). This holds true with regard to yeast membranes harboring GgWS1, GgWS2 and the WS4 proteins. Unlike GgWS5, which did not show acyltransferase activities under any conditions and with any substrates tested, GgDGAT1 was catalytically active in yeast cells, but not in isolated yeast membranes (Additional File 7). Addition of detergents, divalent cations or bovine serum albumin did not improve but partially inhibited incorporation rates of the labeled 16:0-acyl-groups into lipophilic reaction products (data not shown).
GgWS1, GgWS2 and the avian WS4 proteins showing WS activity were further analyzed concerning their reaction products and their acyl-donor and -acceptor specificities. As given in Figure 5, GgWS1 and GgWS2 formed wax esters as main reaction products, while WS4 homologs possessed both WS and DGAT activities. These results which were in line with yeast expression experiments ( Figure 4) clearly demonstrate the mono-or bifunctionality of the respective avian WS (Additional File 7).
With regard to the acyl-CoA specificities, avian enzymes were most active with saturated acyl-CoA thioesters of 14 to 18 carbon atoms, while the WS and DGAT activities were rather low with shorter or unsaturated thioesters. GgWS1 and GgWS2 were almost exclusively active with 16:0-CoA while the avian WS4 homologs produced the highest wax amounts with 16:0or 18:0-CoA and the highest TAG levels with 14:0-CoA (Figure 5a and 5b). Hence bifunctional WS, especially those of goose and barn owl, displayed different acyl-CoA specificities regarding their WS and DGAT activities. In addition to straight-chain acyl-donors, we assayed 2-methyl-branched acyl-CoA thioesters of 14 to 18 carbon atoms as especially barn owls contain many methylbranched fatty acids in their preen gland secretion [19]. To detect WS activity with these substrates we had to increase the protein amount and incubation time. Under these conditions GgWS1 and GgWS2 showed appreciable activities with branched-chain acyl-CoA in comparison to 16:0-CoA, while activity of WS4 proteins were very low in comparison to 16:0-CoA (Figure 5c). GgWS1 displayed the highest activity with 2-methylbranched 16:0-CoA that comprised 70% of the activity determined with 16:0-CoA under the same conditions. GgWS2 performed lower WS activities than GgWS1 but the substrate specificities of both enzymes were very similar to each other (Figure 5c).
In summary, we have identified both monofunctional and bifunctional WS enzymes from birds, which differ in their substrate specificities, especially with regard to branched-chain substrates.

In vitro analysis of WS activities in preen gland membranes
To analyze the in vitro WS activities of preen glands, assays were performed with isolated membrane preparations. Strong WS activity could be detected in membranes of chicken preen glands, but comparably low activities were obtained in skin or liver preparations, suggesting that WS activity is restricted to preen gland tissue. Figure 7a displays the acyl-donor specificities of avian WS using preen gland membranes of chicken, goose and barn owl as enzyme source. The highest WS activities were obtained with 14:0-CoA and almost no activity with 18:1-CoA. In contrast to chicken and goose, barn owl performed high activities with 2-methylbranched acyl-CoA thioesters as well (Figure 7a).
The comparison of different acyl-acceptors proved that membranes of preen glands of all tested species were able to catalyze the esterification of straight-chain alcohols, especially 12:0-OH, or branched-chain alcohols like geranylgeraniol or 3,7-diMe-8:0-OH (Figure 7b). In addition to WS, membranes of chicken preen glands exhibited strong DGAT activity and also high diester synthase activity with 1,2-dodecanediol.

Discussion and Conclusion
In this study we present the identification and characterization of the first avian wax synthase sequences. Although WS activity has already been demonstrated in vitro for cell free preparations from preen glands of chicken and turkey [47], neither a specific enzyme catalyzing this reaction nor the respective gene has been identified yet. Six of nine proteins identified by sequence homologies catalyzed wax ester syntheses in vivo or in vitro (GgDGAT1, GgWS1, GgWS2, GgWS4, AdWS4 and TaWS4) and WS4 homologs additionally showed DGAT activity.
With most avian proteins, in vivo and in vitro experiments gave consistent results. WS activity of GgDGAT1, however, was clearly detectable in yeast cells (Figure 4), but not in enzymatic assays with yeast membranes, in spite of enzymatic assays being more sensitive than feeding experiments. Human ACAT1, to which GgDGAT1 displays the highest sequence identity, was functionally expressed in yeast cells and showed in vitro ACAT activity with cholesterol and oleate [43], but GgDGAT1 did not show ACAT activity. Perhaps we have not tested GgDGAT1 under conditions suitable for this enzyme so far, or the chicken enzyme has a very low protein stability under in vitro conditions and is rapidly degraded by yeast proteases. This still remains to be determined.
The protein termed WS5 is highly conserved among vertebrates and identical in different bird species. It contains a lysophospholipid acyltransferase motif (cl00357 and cd07987), but the enzymatic activities have not been studied in any organism to date. Our experiments could not confirm any acyltransferase activity in vivo or in vitro.
All avian WS are found to be most active with saturated medium-chain alcohols (10:0-OH to 12:0-OH) and saturated long-chain acyl-CoA thioesters (14:0 to 18:0), but they differ in their reaction products and their activities with certain substrates. GgWS4, like the respective homologs from goose and barn owl, are bifunctional enzymes which catalyze both WE and TAG biosynthesis. They are more active with unsaturated than with saturated long-chain alcohols and effectively utilize branched-chain alcohols like isoprenols as acyl-acceptors, especially AdWS4 and TaWS4, but they show hardly any activities with branched-chain acyl-CoA. In contrast to the WS4 homologs, GgWS1 and GgWS2 are monofunctional enzymes which are inactive with saturated and unsaturated alcohols of more than 14 carbon atoms ( Figure 6), but are active with methyl-branched alcohols and branched-chain acyl-CoA thioesters ( Figure  5c). The ability of the chicken enzyme to utilize branched-chain substrates was surprising, as these components are typical for wax esters of barn owl and goose preen glands, but not for those of chicken. Perhaps WS1 and WS2 homologs from barn owl and goose might be even more active with branched-chain substrates than the chicken enzymes. These homologs likely caused the relatively high WE formation rates determined in assays with preen gland membranes of barn owl (Figure 7), while the relatively low activities of preen gland membranes of goose might be due to the wax esters containing tetramethylated but not monomethylated acyl-groups like barn owl glands [18,19].
The secretion of chicken preen glands is rich in 2,3-diester waxes [14,20] and preen gland membranes of chicken, unlike those of goose and barn owl, effectively catalyze the esterification of 1,2-dodecanediol ( Figure 7). On the other hand, diester synthase activities with 1,2-dodecanediol were not detected with transgenic yeast membranes harboring one of the various avian enzymes even when WS assays with yeast membranes were run under conditions identical to those with preen gland membranes. These data suggest that diester waxes might be formed by a WS in chicken which is unrelated to known WS classes or by a related protein, which could not be detected in yeast membranes so far, like GgDGAT1. Anyhow, chicken preen gland membranes displayed high monoester synthase activities (Figure 7), which is in line with previous findings [47] and supports our results of the WS enzymes expressed in yeast (Figures 5 and 6).

Identification of putative wax synthases
BLASTp (protein-to-protein Basic Local Alignment Search Tool) [48] studies were undertaken to investigate the predicted Gallus gallus proteome for putative wax synthases using the NCBI server [49]. Human wax synthase sequences HsAWAT1 [NCBI: NP_001013597] and HsAWAT2 [NCBI: NP_001002254] were used as query sequences for DGAT2 family members and human HsDGAT1 [NCBI: NP_036211] was used as query for DGAT1 like proteins. The obtained results were drawn back on to the originating mRNA sequences, which were used for generating cloning PCR primers.

Phylogenetic analysis and structure prediction
Sequence analyses were carried out using ClustalX2 [51] and GeneDoc [52] software. Phylograms were computed with MEGA5 [53] and neighbor-joining method [54] with 1000 bootstrap replicates using p-distance method. All gaps were deleted for computation of evolutionary distances.
Molecular mass and isoelectric points were calculated using ProtParam [55] on the ExPASy Server [56]. Transmembrane helices of avian proteins and mammalian wax synthases were predicted using TMHMM software [57][58][59][60]. Predictions were compared to Kyte Doolittle plots [61] with window parameters of 19 which revealed similar results. Acyltransferase superfamily motifs and putative acyl-acceptor binding pockets were discovered by NCBI conserved domain search [62,63].
Extraction of yeast cells was performed according to Bligh and Dyer [64]. The lipid extracts were separated by TLC on preparative TLC plates (Silica Gel 60 plates 0.5 mm thickness, Merck) in heptane/diethyl ether/ acetic acid (90/30/1, v/v/v) and visualized under UV light after spraying with dichlorofluorescein (0.3% (w/v) dissolved in isopropanol) [65]. Myristoyl-dodecanoate (Sigma Aldrich) and TAG isolated from sunflower oil were used as standards.

GC analysis of WE and TAG
Bands co-chromatographing with the WE and TAG standards were transmethylated in 0.5 M sulfuric acid and 3% dimethoxypropane in methanol for 1 hour at 80°C together with 250 nmol docosanoic acid as internal standard. Fatty acid methyl esters (FAMEs) and fatty alcohols were extracted with heptane, concentrated and analyzed via gas chromatography (GC) with flame ionization detection (FID). For quantification of WE the total amount of fatty alcohols in the fractions was summarized, for quantification of TAG the sum of FAMEs was divided by 3.
GC-FID analysis was carried out using the HP6890 gas chromatograph equipped with an OPTIMA225 column (Macherey & Nagel) (25 m length, 0.25 mm diameter, 0.25 μm film thickness). 1 μl of the extract was analyzed in splitless injection with N 2 as carrier gas (constant flow, 0.9 bar pressure, total column flow 1 ml/min) and inlet and detector temperatures of 260°C. A temperature program was carried out starting at 120°C, 8°C/min to 144°C, 4°C/min to 240°C. Peaks were identified by comparison of the respective retention times with those of standard substances of different fatty alcohols and FAMEs (Sigma Aldrich).

Preparation of yeast membranes and in vitro wax synthase assay
Membrane preparation and WS assays were performed as described previously [67]. Briefly, transgenic yeast cells were harvested, washed in Tris-H 2 SO 4 (50 mM, pH 7.6), frozen and disrupted. Cell supernatants were combined and sonicated, cell debris was sedimented (2,500 × g, 15 min and 4°C) and the membranes were isolated from the supernatant by high speed centrifugation (1 h, 140,000 × g, 4°C). The sedimented membranes were resuspended in Tris-H 2 SO 4 buffer and stored in aliquots at -80°C. The protein concentration was determined [68].
WS activity was measured with unlabeled acyl-acceptors and labeled acyl-CoA-thioesters as outlined before [63]. The following acyl-CoA thioesters were used: [ Stymne and members of his laboratory, SLU Alnarp, Sweden. The reaction mixture of standard assays consisted of 10 mM BIS-Tris-propane buffer (pH 9), 13 μM [1-14 C]-labeled acyl-CoA and 300 μM acyl-acceptor (Sigma Aldrich) using 2 to 4 μg protein of total yeast membrane fractions as enzyme source. Acyl-acceptors were dissolved in heptane and evaporated to dryness in the reaction tubes before addition of further assay components. Incubation was carried out at 35°C for 20 minutes. Lipids were extracted, applied to TLC silica gel plates (Merck) and chromatographed in heptane/diethyl ether/acetic acid (90/20/1 v/v/v). The bands were visualized with the FLA-3000 bioimager system (Fujifilm) and quantified in a liquid scintillation counter LS 6500 (Beckman Coulter). DGAT activity was measured as byproduct of WS activity without addition of DAG, endogenous substrates were used for TAG synthesis.
Before calculation of relative activities, background activities of membranes from yeast control strains were subtracted.

WS assay with isolated preen gland membranes
Membranes from dissected preen glands were prepared as described before [21]. Briefly, the secretions were removed, the tissues were cut into small pieces, homogenized in NaK-buffer (0.1 M Na-K-phosphate buffer, pH 7.6, 0.5 mM DTT, 1 mM MgCl 2 ) using an Ultra-Turrax three times for 15 seconds. Residual fragments were removed by centrifugation at 2,500 × g for 15 min and membranes were sedimented by ultracentrifugation at 140,000 × g for one hour. The membranes were resuspended in an appropriate volume of NaK-buffer. The protein content of the prepared membrane fractions was determined [68] and about 50 μg protein was used in the WS assay as outlined above, but incubation time was extended to 2 hours. Additional file 7: TLC analysis of lipophilic reaction products from WS assays with yeast membranes. Assays were performed with 16:0-CoA and 10:0-OH under standard conditions using membranes of yeast cells expressing one of the respective sequences. Reaction products were extracted from the assays, separated by TLC and visualized using the FLA-3000 imaging system. The analysis is representative of several repetitions.