Yeast strains used throughout this study are listed in Table 1. Y. lipolytica cells were grown aerobically in Erlenmeyer flasks at 30 °C in medium containing 0.67% yeast nitrogen base (US Biological), 0.1% yeast extract (Oxoid), 0.5% NH₄Cl (Merck), 50 mM potassium phosphate buffer (pH 6.8), 0.01% leucine (Roth) and either 2% glucose (Merck) or 1% oleic acid (Merck) dissolved with 1% Brij58 as a carbon source. Fifty milliliters of culture medium were inoculated to an optical density (OD₆₀₀) of 0.01 from an over night culture grown aerobically. Growth was monitored by measuring OD₆₀₀.

For heterologous expression studies S. cerevisiae cells transformed with plasmids containing either LRO1 or DGA1 of Y. lipolytica were grown aerobically in Erlenmeyer flasks at 30 °C in synthetic minimal (SM) media containing 2% glucose or 2% galactose (Merck), respectively, and 0.67% yeast nitrogen base (US Biological) supplemented with the appropriate amino acids.

For heterologous expression of either Dga1p or Lro1p of Y. lipolytica in the quadruple deletion mutant lro1Δdga1Δare1Δare2Δ of the budding yeast S. cerevisiae the respective open reading frames YALI0E32769g (DGA1) and YALI0E16797g (LRO1) were inserted into the pYES2 vector (Invitrogen). Insertion cassettes were obtained by PCR using genomic DNA derived from a Y. lipolytica wild-type strain as template. Primers used for amplification are shown in Table 2. PCR products were purified with NucleoSpin purification kit (Marcherey-Nagel). The plasmid was cleaved with BamHI and EcoRI. PCR products and the cleaved plasmid were purified by agarose gel electrophoresis and fragments recovered from the gel with NucleoSpin gel extraction kit (Macherey-Nagel). The cleaved plasmid and the purified PCR fragment of either DGA1 or LRO1 were used for transformation of the quadruple deletion mutant lro1Δdga1Δare1Δare2Δ of S. cerevisiae. The respective plasmids were formed upon homologous recombination. The correct insertion of the PCR fragments into the plasmid was verified by colony PCR with control primers listed in Table 2.

The quadruple mutant lro1Δdga1Δare1Δare2Δ of the budding yeast S. cerevisiae was obtained by crossings and tetrad dissection of the respective single deletion strains (EUROSCARF strain collection). The disruption of all four genes in the respective mutant was verified by control PCR and confirmed by the absence of TAG and steryl esters resulting in a lack of lipid particles (see below).

Homogenates and microsomes of the strains listed in Table 1 were isolated as described previously ^[16]. Strains of Y. lipolytica grown on oleic acid containing media were washed three times with 0.5% BSA solution prior to application of the protocol for spheroplast and organelle preparation. Spheroplasts were prepared by treatment with zymolyase as described by Daum et al. ^[17]. Microsomal fractions were obtained by following the protocol of Zinser et al. ^[16].

For protein quantification polypeptides were precipitated from the aqueous phase with trichloroacetic acid at a final concentration of 10%. The protein pellet was solubilized in 0.1% SDS – 0.1 M NaOH at 37 °C. Protein quantification was performed by the method of Lowry et al. ^[18] using bovine serum albumin as a standard.

Total lipids of cells grown to the stationary phase (24 h) were extracted by the procedure of Folch et al. ^[19]. Strains of Y. lipolytica grown in the presence of oleic acid were washed three times with 0.5% BSA solution prior to lipid extraction. Analysis of neutral lipids was performed as described in ^[20]. In brief, lipids were separated by thin-layer chromatography (TLC) using the solvent systems petroleum ether-diethyl ether-acetic acid (25:25:1; per vol.; developed to 1/3 of the total distance) and petroleum ether-diethyl ether (49:1; vol./vol.; total distance). Neutral lipids were visualized on TLC plates by post-chromatographic staining. Plates were dipped for 6 s into a developing reagent consisting of 0.63 g of MnCl₂.4H₂O, 60 ml of water, 60 ml of methanol, and 4 ml of concentrated sulphuric acid, briefly dried, and heated to 105 °C for 30 min. Subsequently, TAG were quantified by densitometric scanning at 400 nm with triolein (NuCheck, Inc., Elysian, MN) as a standard.

Y. lipolytica cells were grown either on glucose or on oleic acid containing medium for 24 h prior to microscopic inspection. To analyze the functionality of DGA1 and LRO1, respectively, upon heterologous expression in the S. cerevisiae quadruple mutant lro1Δdga1Δare1Δare2Δ transformants were grown to the stationary phase in Ura^- minimal medium containing either glucose or galactose as the carbon source. Cells were harvested by centrifugation washed twice and put onto microscope slides. By mixing the cell suspension with a solution of 1 mg/ml Nile Red® (Sigma) in ethanol, cells were stained directly on the slide. Microscopic analysis was performed on a Zeiss Axiovert 35 microscope using a 100-fold oil immersion objective, a UV lamp, and a detection range between 450 and 490 nm. Images were taken with a CCD camera.

The diacylglycerol acyltransferase assay was performed as described previously ^[7]. In brief, enzymatic activity was measured in a final volume of 400 μl containing 10 nmol dipalmitoylglycerol, 100 nmol unlabeled (Sigma) and radioactively labeled [1-¹⁴C]oleoyl-CoA or [1-¹⁴C]palmitoyl-CoA (0.04 μCi; Perkin Elmer), 400 μg BSA, 30 μg dithiothreitol, 0.25% Triton X-100, 150 mM Tris/Cl^- (pH 7.0), 15 mM KCl, 15 mM MgCl₂ and either homogenate (0.7 to 0.9 mg) or microsomes (0.1 to 0.3 mg) as the respective enzyme source. For measuring the phospholipid:diacylglycerol acyltransferase activity the assay mixture was essentially the same as for the diacylglycerol acyltransferase assay but the amount of dipalmitoylglycerol was increased to 20 nmol, [¹⁴C]phospholipids (75,000 dpm; approx. 25 μg) were used as the labeled substrate and acyl-CoA was omitted. Both assays were carried out at 30 °C and the reaction was terminated after 30 min by the addition of 3 ml chloroform-methanol (2:1; vol./vol.). Lipids were extracted ^[19], separated by thin-layer chromatography with the solvent system petroleum ether-diethyl ether-acetic acid (70:30:2; per vol.), scraped off the plate and radioactivity was measured by liquid scintillation counting in 8 ml liquid-scintillation cocktail containing 5% water.

Radioactively labeled phospholipids were synthesized by incubation of a S. cerevisiae strain with 0.1 mCi of [1-¹⁴C]palmitic acid for 24 h at 30 °C in 100 ml medium containing 2% glucose (Merck), 1% yeast extract (Oxoid) and 2% peptone (Oxoid). Lipids were extracted by using the method of Folch et al. ^[19] and separated by thin-layer chromatography. The band containing phospholipids was scraped off the plate and phospholipids extracted from the silica gel with chloroform-methanol (1:4; vol./vol.). Label-distribution between major glycerophospholipids: phosphatidylcholine 45%, phosphatidylethanolamine 30%, phosphatidylinositol 14%, phosphatidylserine 11%.

Analysis of the lipid particle proteome of Y. lipolytica^[12] revealed the presence of a protein encoded by the open reading frame YALI0E32769g sharing 50% identity and 71% homology to the acyl-CoA:diacylglycerol acyltransferase Dga1p of the budding yeast S. cerevisiae and 51% identity and 72% homology to the respective enzyme in the fission yeast S. pombe (blast search: http://www.ch.embnet.org). In addition, the sequence of the potential TAG synthase of Y. lipolytica reveals homology to acyl-CoA:diacylglycerol acyltransferases of mammals, e.g., mouse, rat and human (56% homology). The predicted molecular mass of the Dga1p homolog of Y. lipolytica is 57.8 kDa (514 amino acids) with a pI of 8.7 (http://www.ch.embnet.org/software/TMPRED_form.html). Noteworthy, this protein contains after the first ~ 120 amino acids an insertion of 113 and 155 amino acids, respectively, compared to Dga1p of S. cerevisiae and S. pombe (Fig. 2A).

The second major TAG-synthase of the budding yeast S. cerevisiae is Lro1p (lecithin:cholesterol acyltransferase (LCAT) related ORF). Homology searches in the genome of Y. lipolytica with LRO1 of the budding yeast as a query highlighted the ORF YALI0E16797g as a sequence encoding a potential acyl-CoA independent TAG synthase. The respective protein encoded by YALI0E16797g shares 51% identity and 68% homology to Lro1p of S. cerevisiae, and 46% identity and 64% homology to Plh1p, the Lro1p counterpart of S. pombe (Fig. 2B). Furthermore, the potential TAG synthase of the oleaginous yeast is related to lecithin:cholesterol acyltransferases (LCAT) of mammals, e.g., rabbit, mouse, human. The Lro1p homolog of Y. lipolytica is formed of 648 amino acids with an estimated molecular mass of 72.6 kDa and a pI of 5.5 (http://www.ch.embnet.org/software/TMPRED_form.html).

For both potential TAG synthases of the oleaginous yeast a transmembrane domain is predicted at the N-terminus (Fig. 2C). Similarly, Dga1p and the phospholipid:diacylglycerol acyltransferase of the budding and fission yeast, respectively, contain an N-terminal transmembrane domain (Kyte Doolittle blot: http://fasta.bioch.virginia.edu/).

Due to the homologies of the respective gene products to already identified TAG synthases of other eukaryotes these unassigned genes of Y. lipolytica are named hereafter DGA1 (YALI0E32769g) and LRO1 (YALI0E16797g).

Since TAG and steryl esters lack charged groups they do not fit into the phospholipid bilayer of membranes and are sequestered from the cytosolic environment in the hydrophobic core of so-called lipid particles. In the oleaginous yeast Y. lipolytica the hydrophobic core of lipid particles consists mainly of TAG ^[12]. To analyze whether deletion of DGA1 and/or LRO1 affects the size and/or number of lipid particles, cells grown to the stationary phase were stained with the fluorescent dye NileRed® specific for these compartments (see Materials and methods). Whereas the number and size of lipid particles in cells of the single deletion mutants lro1Δ and dga1Δ grown on glucose containing medium are similar to control, a higher number of small lipid particles is formed in the double deletion mutant lro1Δdga1Δ (Fig. 3). In contrast, no significant differences in the formation of lipid particles are observed between the single deletion mutants lro1Δ and dga1Δ, the double deletion mutant lro1Δdga1Δ and control when oleic acid serves as a carbon source (data not shown).

Cells of the single deletion mutants lro1Δ and dga1Δ, and the double deletion mutant lro1Δdga1Δ were grown to the stationary phase either on glucose or oleic acid containing medium, lipids of total cells extracted and analyzed for the neutral lipid composition (see Materials and methods). The single deletion mutant lro1Δ accumulates TAG to approx. 75% of wild-type level independent whether glucose or oleic acid serves as a carbon source (Fig. 4). In contrast, the amount of TAG is only significantly reduced in lipid extracts of dga1Δ cells grown on glucose containing medium (90% of control) but not when oleic acid serves as a carbon source. By growing cells of the double deletion mutant lro1Δdga1Δ on glucose containing medium the TAG level decreases to 30% of control. In contrast, in cell extracts of the double deletion mutant grown on oleic acid medium the amount of TAG is reduced to 85% of wild-type level. Since TAG are still formed in the double deletion mutant lro1Δdga1Δ, Y. lipolytica harbors at least one additional enzyme catalyzing the acylation of diacylglycerol.

Even though all results presented so far are pinpointing that Dga1p and Lro1p of the oleaginous yeast are indeed TAG synthases, it cannot be excluded that one of these proteins or both are involved in the regulation of TAG synthesis. Thus, heterologous expression experiments were performed.

As already mentioned above, neutral lipids such as TAG and steryl esters do not fit into membrane structures and are sequestered in lipid particles. In the budding yeast S. cerevisiae four enzymes catalyze the acylation of diacylglycerol yielding TAG, i.e., Lro1p, Dga1p, Are1p and Are2p ^[6]. The preferential substrates of Are1p and Are2p, however, are sterols. Since Are1p and Are2p are the only steryl ester synthases in the budding yeast, the quadruple mutant lro1Δdga1Δare1Δare2Δ lacks not only TAG but also steryl esters and consequently does not form lipid particles. To investigate the function of Dga1p and Lro1p of Y. lipolytica the respective genes were cloned into the vector pYES2 and heterologously expressed in the quadruple mutant lro1Δdga1Δare1Δare2Δ of S. cerevisiae (see Materials and methods). The GAL1 promoter of the plasmid allows induction of gene expression by growing cells on galactose containing medium. Staining cells of both plasmid bearing lro1Δdga1Δare1Δare2Δ mutants with the lipophilic fluorescent dye NileRed® and microscopic inspection demonstrate the formation of lipid particles upon galactose induction (Fig. 5A, panels c and d). Whereas heterologous expression of LRO1 in the quadruple deletion mutant background leads to the formation of lipid particles similar in size to the ones present in the corresponding S. cerevisiae wild-type strain, the size of the particles is significantly increased in the quadruple mutant expressing DGA1. The number of lipid particles formed upon heterologous expression of either DGA1 or LRO1 in the lro1Δdga1Δare1Δare2Δ quadruple mutant is limited to 1 to 2. As a negative control, no lipid particles are formed in the quadruple mutant lro1Δdga1Δare1Δare2Δ bearing the empty vector (Fig. 5A, panel b) or when transformants are grown on glucose containing minimal media (data not shown). Lipid analyses of total cells of the quadruple mutant lro1Δdga1Δare1Δare2Δ expressing either DGA1 or LRO1 of Y. lipolytica demonstrate the presence of TAG (Fig. 5C). The prominent TAG band in the lipid extract of the transformant expressing DGA1 (Fig. 5C, lane 5) parallels the bigger size of lipid particles observed by fluorescence microscopy (Fig. 5A, panel c). The absence of steryl esters in lipid extracts of strains heterologously expressing DGA1 or LRO1 demonstrates that sterols do not serve as substrate of Dga1p and Lro1p (Fig. 5B).

To elucidate the mechanism employed by Lro1p and Dga1p, respectively, for TAG formation enzyme assays specific for acyl-CoA dependent and acyl-CoA independent acylation of diacylglycerol were performed with transformants heterologously expressing DGA1 or LRO1 (see Materials and methods). In presence of oleoyl-CoA TAG synthase activity of homogenate of the S. cerevisiae quadruple mutant lro1Δdga1Δare1Δare2Δ heterologously expressing DGA1 is 45-fold increased compared to the respective corresponding wild-type of the quadruple mutant bearing the empty plasmid (Fig. 6A). However, in the assay specific for acyl-CoA dependent acylation some label is also incorporated into TAG when homogenate of lro1Δdga1Δare1Δare2Δ heterologously expressing LRO1 serves as the enzyme source (60% of S. cerevisiae wild-type level). Since oleic acid is the most abundant fatty acid occurring in TAG of Y. lipolytica (52.5%) ^[12], enzymatic measurements for acyl-CoA dependent TAG synthesis were performed with oleoyl-CoA as a substrate. The second major fatty acid bound to the glycerol backbone of TAG is palmitic acid (22.1%). Comparison of the specific activities of Dga1p measured either in the presence of oleoyl-CoA or palmitoyl-CoA revealed a clear preference for the former fatty acid (479 ± 46 nmol/mg ∗ min vs. 108 ± 13 nmol/mg ∗ min).

By using phospholipids as acyl-donor microsomes of the quadruple mutant lro1Δdga1Δare1Δare2Δ containing LRO1 of the oleaginous yeast incorporate 1.4 times more radioactive label into TAG than microsomes of the corresponding S. cerevisiae wild-type strain bearing the empty vector (Fig. 6B). Whereas substantial amounts of labeled TAG were formed in the presence of the phospholipid mixture used for the acyl-CoA independent assay (see Materials and methods), only traces of labeled TAG were formed when individual phospholipids (phosphatidylcholine, phosphatidylethanolamine, phosphatidylinositol, phophatidylserine) were used as acyl-donor. Especially phosphatidylcholine turned out to be a very poor substrate (data not shown). In the phospholipid:diacylglycerol acyltransferase specific assay microsomes were used as the enzyme source, because only traces of label were incorporated into TAG when homogenates of the respective strains were used as an enzyme source.

Results of heterologous expression experiments revealed that Dga1p catalyzes TAG synthesis via the acyl-CoA dependent pathway, whereas Lro1p prefers phospholipids as acyl-donor. To study TAG synthesis in Y. lipolytica in more detail homogenates of the single deletion strains lro1Δ, dga1Δ, the double deletion mutant lro1Δdga1Δ and the corresponding wild-type strain were used as an enzyme source in assays specific for either acyl-CoA dependent or acyl-CoA independent acylation of diacylglycerol (see Materials and methods). By using homogenate of the single deletion mutant lro1Δ (i.e., Dga1p is present) grown on glucose containing medium as the enzyme source TAG are only formed in the presence of oleoyl-CoA (Fig. 7A), but not when phospholipids serve as acyl-donor. In contrast, homogenate of dga1Δ cells (i.e., Lro1p is present) grown in the presence of glucose uses phospholipids as the preferred substrate for TAG synthesis. However, this enzyme source incorporates also a small but significant amount of radioactive label into TAG in the acyl-CoA dependent assay. The finding that under the same conditions homogenate of lro1Δdga1Δ cells grown on glucose containing medium lacks TAG synthase activity indicates that Lro1p of dga1Δ catalyzes the incorporation of the labeled fatty acid into TAG. This is in line with the observation that upon heterologous expression of LRO1 in the S. cerevisiae mutant lro1Δdga1Δare1Δare2Δ homogenate of these cells forms TAG in the presence of acyl-CoA. Even though TAG are present in lro1Δdga1Δ cells of the oleaginous yeast grown on glucose containing medium (Fig. 4), in vitro homogenate of the respective mutant cells lacks also TAG synthase activity when phospholipids are provided as acyl-donor (Fig. 7A). In the budding yeast S. cerevisiae it has been shown that the contribution of Lro1p and Dga1p to TAG synthesis depends on the growth stage of the cells ^[5]. Whereas Lro1p is the major TAG synthase in cells of the exponential phase, TAG synthase activity of Dga1p is higher in cells entering the stationary phase. To test whether the same is true for Lro1p and Dga1p of Y. lipolytica, homogenate of wild-type cells of the exponential growth phase (OD₆₀₀ ~ 1.2) and the stationary phase (OD₆₀₀ ~ 5.4) were used as an enzyme source in the respective TAG synthase assays. Indeed, acyl-CoA independent TAG synthase (Lro1p) activity decreased by ~ 25% (A_spec = 1.4 ± 0.06 pg/mg ∗ min vs. A_spec = 1.1 ± 0.05 pg/mg ∗ min) and that of Dga1p increased by ~ 20% (A_spec = 8.3 ± 0.6 pmol/mg ∗ min vs. A_spec = 10 ± 1.8 pmol/mg ∗ min).

In analogy to the enzyme assays performed with homogenates of cells grown on glucose containing media, homogenates of cells grown on oleic acid containing media were tested for TAG synthase activities via both mechanisms (Fig. 7B). In the Lro1p-specific assay enzymatic activity of homogenate of dga1Δ cells (i.e., Lro1p is present) grown in the presence of oleic acid is 1.7-fold increased compared to wild-type. In contrast to homogenate of lro1Δ cells grown on glucose containing medium which lacks phospholipid:diacylglycerol acyltransferase activity (Fig. 7A), a small amount of label is incorporated into TAG via this pathway when the respective enzyme source of cells grown in the presence of oleic acid is used. However, under the same conditions homogenate of the double deletion mutant lro1Δdga1Δ is inactive. In the acyl-CoA dependent TAG synthase assay homogenates of both single deletion mutants and the double deletion mutant lro1Δdga1Δ form TAG, however, with lower efficiency than control. Noteworthy, in contrast to homogenate of lro1Δdga1Δ cells grown on glucose containing medium which do not incorporate the provided acyl-CoA group into TAG (Fig. 7A), homogenate of the respective mutant cells grown on oleic acid containing medium exhibit TAG synthase activity via the acyl-CoA dependent pathway. Since homogenates of both dga1Δ cells as well as lro1Δdga1Δ cells contain 35% of acyl-CoA dependent TAG-synthase activity compared to control (Fig. 7B), the observed acylation activity cannot be ascribed to Lro1p. Thus, in lro1Δdga1Δ cells grown on oleic acid containing medium at least one additional acyl-CoA dependent enzyme contributes to TAG synthesis.