Expression in yeast and tobacco of plant cDNAs encoding acyl CoA:diacylglycerol acyltransferase


P. Bouvier-Navé, Institut de Biologie Moléculaire des Plantes, Département de Biologie Cellulaire et Moléculaire, Institut de Botanique, 28 rue Goethe, F-67083 Strasbourg cedex, France. Fax: + 33 3 8835 8484, E-mail:


During the course of a search for cDNAs encoding plant sterol acyltransferases, an expressed sequence tag clone presenting substantial identity with yeast and animal acyl CoA:cholesterol acyltransferases was used to screen cDNA libraries from Arabidopsis and tobacco. This resulted in the isolation of two full-length cDNAs encoding proteins of 520 and 532 amino acids, respectively. Attempts to complement the yeast double-mutant are1 are2 defective in acyl CoA:cholesterol acyltransferase were unsuccessful, showing that neither gene encodes acyl CoA:cholesterol acyltransferase. Their deduced amino acid sequences were then shown to have 40 and 38% identity, respectively, with a murine acyl CoA:diacylglycerol acyltransferase and their expression in are1 are2 or wild-type yeast resulted in a strong increase in the incorporation of oleyl CoA into triacylglycerols. Incorporation was 2–3 times higher in microsomes from yeast transformed with these plant cDNAs than in yeast transformed with the void vector, clearly showing that these cDNAs encode acyl CoA:diacylglycerol acyltransferases. Moreover, during the preparation of microsomes from the Arabidopsis DGAT-transformed yeast, a floating layer was observed on top of the 100 000 g supernatant. This fraction was enriched in triacylglycerols and exhibited strong acyl CoA:diacylglycerol acyltransferase activity, whereas almost no activity was detected in the corresponding clear fraction from the control yeast. Thanks to the use of this active fraction and dihexanoylglycerol as a substrate, the de novo synthesis of 1,2-dihexanoyl 3-oleyl glycerol by AtDGAT could be demonstrated. Transformation of tobacco with AtDGAT was also performed. Analysis of 19 primary transformants allowed detection, in several individuals, of a marked increase (up to seven times) of triacylglycerol content which correlated with the AtDGAT mRNA expression. Furthermore, light-microscopy observations of leaf epidermis cells, stained with a lipid-specific dye, showed the presence of lipid droplets in the cells of triacylglycerol-overproducer plants, thus illustrating the potential application of acyl CoA:diacylglycerol acyltransferase-transformed plants.


acyl CoA:cholesterol acyltransferase


ACAT-related gene product


diacylglycerol acyltransferase


ethylene responsive enhancer element


expressed sequence tag


phosphoglycerate kinase


plaque-forming units

Acyl CoA:diacylglycerol acyltransferase (DGAT; EC is a membrane-bound enzyme that catalyzes the last and only committed step in triacylglycerol synthesis, by using a diacylglycerol and a fatty acyl CoA as its substrates [1,2]. In animals, triacylglycerol synthesis is involved in numerous processes such as regulation of plasma triacylglycerol concentration, fat storage in adipocytes and milk production [3]. In yeast, triacylglycerols and steryl esters coaccumulate under certain physiological conditions in lipid particles (or lipid bodies), which constitute a fatty acid and sterol storage pool [4,5]. In plants, triacylglycerol synthesis is mainly involved in the generation of seed oils [6] but triacylglycerols are also stored in fruits, petals and mature pollen grains [7] where they might be important as a source of fatty acid for the rapid growth of the pollen tube [8].

DGAT-catalyzed esterification of diacylglycerol was proposed to be an important step in the control of plant triacylglycerol synthesis [9,10] and cloning of the DGAT gene would allow us to ascertain this role. In recent months two cDNA clones have been reported, the deduced proteins of which showed identity with known acyl CoA:cholesterol acyltransferases (ACATs). One called ACAT-related gene product (ARGP) was isolated from Homo sapiens[11] and another from Mus musculus[12]. Actually both clones were shown not to code for an ACAT. The H. sapiens cDNA encoding ARGP did not complement a yeast are1 are2 double-mutant defective in the two ACATs (ARE1 and ARE2) present in wild-type yeast. Moreover, the murine cDNA, when expressed in H5 insect cells, did not produce any ACAT activity but was unambiguously shown to encode a DGAT [12]. More recently an Arabidopsis thaliana cDNA presenting significant identity with murine DGAT was cloned by Hobbs et al. [13]. This cDNA was expressed in insect cell cultures and the microsomal preparation of these cells catalyzed the synthesis of [14C] triacylglycerol from [14C]dioleylglycerol and oleyl-CoA thus demonstrating its function. In addition, these authors showed that the cDNA AtDGAT is expressed most strongly in developing embryos and petals [13]. During the course of a search for an ACAT-like plant cDNA, we happened to isolate, independently and using a different procedure, the same AtDGAT, as well as its homolog from Nicotiana tabacum. Both cDNAs were expressed in yeast where we could accumulate and thoroughly characterize the product of the enzymatic reaction. Transgenic tobaccos expressing AtDGAT under the control of a constitutive promoter were generated in order to prove the function in planta. A strong increase in the triacylglycerol content of young growing leaves was shown to be correlated with the appearance of lipid droplets in the cells.

Experimental procedures

Strains, media and culture conditions

Escherichia coli. XL1blue recA-[recA1, lac-, endA1, gyrA96, thi, hsdR17, SupE44, relA1 (F′proAB, lac1q, lacZ ΔM15, Tn10)].

Saccharomyces cerevisiae. Two strains of common genetic background (can 1–100, his 3–11, 15, leu 2-3, 112, trp 1-1, ura 3-1) were used: SCY059 (MATα, met14ΔHpaI-SalI, are1ΔNA::HIS3, are2Δ::LEU2) and the corresponding wild-type SCY062 (MATα, ade2–1).

Yeast strains transformed with plasmid pYeDP60, harboring either no insert or the plant cDNA under study, were always grown at the same time. For sterol or triacylglycerol analysis, culture conditions were 3 days at 30 °C on minimal medium [6.7 g·L−1 yeast nitrogen base (Difco), 20 g·L−1 galactose] containing suitable supplements (50 µg·mL−1 each), i.e. Ade, Trp and Met for SCY059 (are1 are2) and His, Trp and Leu for SCY062 (wild-type). The cultures were then centrifuged and freeze-dried.

For subcellular fractionation, these transformed strains were grown for 3 days at 30 °C in minimum medium containing 20 g·L−1 glucose and the suitable supplements. The cultures were then centrifuged under sterile conditions, and cells were resuspended in a complete medium [10 g·L−1 yeast extract (Difco), 10 g·L−1 bactopeptone (Difco), 20 g·L−1 galactose] and grown overnight at 30 °C. This two-step procedure resulted in a high cell mass, suitable for microsome preparation.

Plasmid for yeast transformation

The plasmid pYeDP60 [14] was used to transform yeast strains. This plasmid contains an E. coli replication origin, a yeast 2 µm plasmid replication origin, an E. coli ampicillin-resistance gene and the yeast genes URA3 and ADE2. It utilizes an expression cassette including a galactose-inducible hybrid promoter and a phosphoglycerate kinase (PGK) terminator. Gene expression is driven by the upstream activating sequence of the yeast GAL10 and CYC4 genes.

Cloning of a cDNA encoding DGAT from A. thaliana

An EST cDNA clone (ABRC DNA Stock Center, Columbia, OM 43210-1002) showing significant identity with ACATs was sequenced. This 849 bp EST fragment was used to screen a cDNA library CD4-15 [15] supplied by the ABRC DNA Stock Center. In total, 500 000 plaque-forming units (p.f.u.) were screened with the EST clone labeled with 32P after random priming leading to 12 cDNA clones of different lengths. Among them 10 were larger than 1.5 kb. When cut with EcoRI these clones showed a digestion pattern suggesting that they possess an identical ORF but differ by the respective length of the 5′-UTR and 3′-UTR. Two of them were sequenced, the first (3D11) contained 1845 bp with an ORF of 1560 bp flanked by 5′-UTR and 3′-UTR of, respectively, 90 and 195 bp. The second (3D71) contained 2062 bp with an identical ORF flanked by 5′-UTR and 3′-UTR of, respectively, 173 and 429 bp.

Cloning of a cDNA encoding DGAT from N. tabacum

A cDNA library from a 3-week-old N. tabacum cv. Xanthi line LAB 1-4 calli derived from leaf protoplasts [16] was screened with the Arabidopsis EST clone of 849 bp. First, 250 000 p.f.u. were screened with an ORF from the Arabidopsis cDNA 3D11 leading to two clones (1A111 and 2A113). One of them (2A113) was sequenced and shown to contain 1354 bp and its deduced amino acid sequence (317 amino acids) displayed a strong (> 70%) identity with the Arabidopsis DGAT deduced protein sequence. This clone, called NtDGAT2, was likely uncomplete. After digestion with EcoRI and HindIII, a 1-kb fragment was randomly labeled with [32P]dCTP and used to screen again the tobacco library (250 000 p.f.u.), seven clones of different lengths were isolated. NtDGAT2 was again found among them and was shown to be identical to the cDNA found previously. Another clone presented a length (2099 bp) comparable with that of the Arabidopsis DGAT. After sequencing it was shown to encode a polypeptide of 532 amino acids presenting 68% identity with Arabidopsis DGAT and 86% identity with NtDGAT2. This cDNA was called NtDGAT1.

Reformatting and cloning DGAT cDNAs into the yeast expression vector pYeDP60

Deletion of the 5′-UTR and 3′-UTR of the DGAT cDNAs was performed by PCR amplification using specific primers.

Arabidopsis thaliana DGAT. Specific primers were designed to introduce a BamHI restriction site immediatly upstream of the initiation codon and a KpnI site immediatly downstream of the stop codon. Direct primer: 5′-atatatggatccATGGCGATTTTGGATTCTGC-3′. Reverse primer: 5′-tatataggtaccTCATGACATCGATCCTTTTCG-3′. The DGAT was amplified using 25 thermal cycles (1 min 93 °C, 2 min 56 °C, 3 min 72 °C) with Pyrococcus furiosus (Pfu) DNA polymerase under the standard conditions. The PCR product was digested with BamHI and KpnI and subsequently cloned into Bluescript to give At DGAT-pSK. The BamHI, KpnI insert in pSK was extracted and subcloned into BamHI, KpnI of pYeDP60 leading to AtDGAT–pYeDP60.

A. thaliana FLAG DGAT. In order to evaluate the amount of Arabidopsis DGAT expressed in yeast, an N-terminal FLAG epitope (MDYKDDDDK, epitope underlined) was fused to the DGAT protein. For this purpose a DNA molecule containing at 5′-end the appropriated nucleotide sequence was synthesized by PCR. Direct primer: atatatggatccATGGACTACAAGGACGACGATGACAAGGCGATTTTGGATTCTGCTGG. The reverse primer is the one used above to amplify the At DGAT. The PCR product was digested with BamHI and KpnI and subsequently cloned into pYeDP60 to give AtFLAGDGAT–pYeDP60.

N. tabacum DGAT1. Direct primer: atatatggatccATGGTGATCATGGAATTGCC. Reverse primer tatataggtaccTTAACGTGCACTGCTTTTCT. The tobacco DGAT1 was amplified using 25 thermal cycles (1 min at 93 °C, 2 min at 56 °C, 3 min at 72 °C) with Pfu DNA polymerase under the standard conditions. The PCR product was digested with BamHI and KpnI and subsequently cloned into pYeDP60 to give NtDGAT1–pYeDP60.

Caenorhabditis elegans DGAT. cDNA EST yk 453 a2 was kindly donated by Y. Kohara (Mishima, Japan). After sequencing, this clone perfectly matches a processed gene (Z75526) arising from the systematic sequencing of the genome of this nematode. The ORF was amplified by PCR under standard conditions (see above) using the following primers. Direct primer: atatatggatccATGCAAATGCGTCAACAAACG. Reverse primer: tatataggtaccTCAAATACCAACGGTTTGG. The PCR product was digested with BamHI and KpnI and subcloned into BamHI, KpnI of pYEDP60 leading to CeDGAT–pYeDP60.

Nucleotide sequence determination

Sequencing of cDNAs AtDGAT, AtFLAGDGAT, NtDGAT1 and CeDGAT, as well as their ORF derivatives made by PCR, was performed using an automatic sequencer Perkin-Elmer model 373 with T3, T7 primers, specific oligonucleotide sequence belonging to the sequenced DNA and a modified Taq polymerase capable of incorporating fluorescent dNTP. Complete sequencing of both DNA strands was performed. cDNAs were in pSK or in pYeDP60.

Transformation of yeast

Transformation was performed according to Schiestl and Gietz [17] with some modifications. A fresh yeast culture (initial absorbance = 0.2) was grown in complete medium YPG [10 g·L−1 yeast extract (Difco), 10 g·L−1 bactopeptone (Difco), 20 g·L−1 glucose] for 5 h. The cells were collected, washed twice with water and then with 1.5 mL of a 0.1 m lithium acetate solution in Tris/EDTA buffer (1 mm EDTA, 10 mm Tris/HCl, pH 7.5) and finally resuspended in 200 µL of the same solution. Salmon sperm was added as a DNA carrier (100 µg from a 10 mg·mL−1 solution in Tris/EDTA) after sonication (10 s) and boiling (20 min) to the plasmid DNA (1 µg). Competent yeast cells (50–80 µL) and 50 µL of 40% poly(ethylene glycol), 0.1 m lithium acetate solution in Tris/EDTA (pH 7.5) were added. The mixture was incubated for 30 min at 30 °C followed by 15 min at 42 °C. After centrifugation, the cells were resuspended in YPG (1 mL), incubated 1 h at 30 °C, collected and then plated (with 100 µL water) on minimum medium containing suitable supplements (50 µg·mL−1 each).

Plant expression vectors

A cDNA 3D11 encoding the ORF of At DGAT was subcloned into the BamHI–KpnI sites of a pBD 121 binary vector derivative engineered as described by Husselstein et al. [18]. In particular, the vector contained between the left and right T-DNA borders a kanamycin resistance gene and a multiple cloning site flanked in the 5′-region by a duplicated 35S promoter described by Kay et al. [19] and by a nopaline synthase terminator at the 3′-end. The At DGAT construct and the corresponding void plasmid were transferred from E. coli XL1 Blue into Agrobacterium tumefaciens LB4404 [20] by triparental mating with the helper plasmid pRK 2013 [21].

Plant transformation

Tobacco (N. tabacum L. var Xanthi) was transformed via Agrobacterium tumefaciens according to a modification [22] of the method reported by Horsch et al. [23]. In brief, leaf pieces were cocultivated with an overnight Atumefaciens culture on Murashige and Skoog medium supplemented with glucose (30 g·L−1), ANA (0.1 mg·L−1) and 6-benzylaminopurine (1 mg·L−1) for 2 days and then transferred onto the same medium supplemented with kanamycin (100 mg·L−1) as the plant-selective agent and cefotaxime (500 mg·L−1) to prevent further bacterial growth. Regenerated shoots were rooted on Murashige and Skoog medium with a half-reduced concentration of NH4NO3 and supplemented with sucrose (30 g·L−1), kanamycin (200 mg·L−1) and cefotaxime (200 mg·L−1). In vitro cultures were grown under a 16-h light period at 24 °C and an 8-h dark period at 20 °C. Primary transformants were subcultured as T1 vitro-plantlets every 7–8 weeks in order to generate leaf material in sufficient quantities for analytical processes. Primary transformants were then transferred to the greenhouse and grown under standard conditions.

RNA gel blot analysis

Total RNAs from leaf material of 6-week-old in vitro-grown primary transformants were extracted according to the method described by Goodall et al. [24]. Northern analysis was carried out by separating 10 µg RNA samples on formaldehyde gel as described previously [25] and blotting them onto nylon filters (Hybond-N+, Amersham). Randomly primed (Stratagene) 32P-labeled probes polymerized from the 3D11 cDNA were hybridized to the filters as recommended by the manufacturer. Filters were then washed in 0.2× NaCl/Cit containing 0.1% SDS at 65 °C before autoradiography. Filters were further washed in 0.1% SDS at 70 °C for 3 h and then hybridized to randomly primed 32P-labeled probes polymerized from a 18S rRNA as an internal standard.

Lipid analysis

Lipids were extracted from freeze-dried yeast cells or tobacco leaves by immersion in methylene chloride/methanol (2 : 1) and heating under reflux for 1 h. The extract was submitted to TLC in methylene chloride into sterol esters (Rf = 0.8–0.7), triacylglycerols (Rf = 0.5–0.4) and free sterols (Rf = 0.3–0.1) using cholesteryl palmitate, trioleylglycerol, lanosterol and cholesterol as references.

Sterol esters were hydrolyzed by heating under reflux for 1 h with methanolic KOH (6%). The sterols were then extracted with hexane. Both free sterols and sterols corresponding to sterol esters were acetylated according to Schmitt and Benveniste [26]. Sterol acetates were purified by TLC in methylene chloride, then quantified and identified by GC on a DB-1 capillary column (according to their relative retention time to the internal standard cholesterol), as described previoulsy [22].

Triacylglycerols purified by TLC from the lipid extract of either yeast or transgenic tobacco leaves were quantified using the colorimetric assay Perichrom Triglycerides GPO-PAP from Roche-Diagnostics (Meylan, France), according to McGowan et al. [27].

GC-MS was performed on a computerized gas-chromatograph mass spectrometer (Fison MD500) equipped with an on-column injector and a capillary column (30 m × 0.25 mm internal diameter) coated with DB5 (J&W Scientific). The different fragments obtained are designated by the ratio m/z and their relative intensity.

Lipid-specific staining of cells

Leaf epidermis was stained in a satured Sudan IV (dye no. 26106 according to the color index) solution in 70% (v/v) ethanol. Lipid droplets appear as orange spherical granules on light-microscopy.

Subcellular fractionation

Yeasts were grown for 3 days in 100 mL glucose minimum medium followed by 16 h in 200 mL galactose complete medium (see above). The harvested cells were then disrupted as described previously [28] except that KCl was omitted from the washing buffer and BSA (1%) was added to the disruption medium. The homogenate was centrifuged for 10 min at 12 000 g and the supernatant for 60 min at 100 000 g. The lipid layer floating at the top of the 100 000 g supernatant from the plant DGAT-transformed yeasts was carefully withdrawn as was the corresponding volume (4–6 mL) from the void plasmid-transformed yeast. The microsomal pellet was resuspended (2–5 mg protein·mL−1) in 0.1 m Tris/HCl pH 7 containing 20% (v/v) glycerol. In some experiments (Fig. 3A, C) microsomes and the floating lipid layer were washed by dilution in Tris/HCl buffer and recentrifugation. Coarse or washed fractions were kept at −80 °C for months without significant loss of activity.

Figure 3.

DGAT enzymatic activities in subfractions of yeast expressing AtDGAT. Yeasts were transformed with ▪ the void pYeDP60 plasmid, bsl00023AtDGAT in pYeDP60, or AtFLAGDGAT in pYeDP60. (A) DGAT activities (nmol triacylglycerol per mg protein per h) in washed microsomes from void plasmid-transformed or AtDGAT-transformed are1 are2 mutant. (B) Expression of the FLAG-tagged DGAT. Microsomes (6 µg protein) from wild-type yeast transformed with (1) AtFLAGDGAT or (2) AtDGAT and 100 000 g supernatant top fractions (5 µL) from wild-type yeast transformed with (3) AtFLAGDGAT or (4) AtDGAT were resolved by SDS/PAGE. Proteins were submitted to immunoblotting with an anti-FLAG serum. (C) DGAT activities (nmol triacylglycerol·mL−1·h−1) in twice-washed 100 000 g supernatant top fractions from void plasmid-transformed or At DGAT-transformed wild-type yeast. DOG, dioleylglycerol; DHG, dihexanoylglycerol.

Proteins were quantified using the Bio-Rad protein assay according to Bradford [29]. Western blots of microsomes (6 µg protein) or the floating lipid layer (5 µL) from AtFLAGDGAT-transformed yeast were achieved after SDS/PAGE, electrotransfer to a nylon membrane (Hybond C, Amersham) and immunoblotting with the anti-FLAG M2 mAb from IBI/Kodak.

Enzymatic assays

ACAT assays were performed according to Billheimer et al. [30] using either [14C]cholesterol or [14C]oleyl CoA.

DGAT was assayed according to Cases et al. [12] with some modifications. Diacylglycerol (either dioleylglycerol or dihexanoylglycerol, 400 µm) was resuspended in 0.1 m Tris/HCl buffer pH 7 containing 20% glycerol, by vigorous shaking. The enzymatic preparation, either microsomes (50 µg protein) or 100 000 g supernatant top fraction (40 µL) was then added. In the control assays, proteins were omitted. After the addition of [14C]oleyl CoA (20 µm, 200 000 c.p.m.) the reaction mixture (100 µL) was incubated for 10 min at 30 °C. Under these conditions the substrates were shown to be saturating, the reaction was linear with protein and time and the reaction yield was kept below 10%. The reaction was stopped by adding a mixture of methylene chloride (400 µL) and methanol (100 µL) containing trioleylglycerol (100 µg) and cholesteryl palmitate (100 µg) as carriers. After further addition of water (300 µL) and methylene chloride (600 µL), the organic phase was withdrawn and the water phase was extracted twice more with methylene chloride. The lipid extract was separated by TLC with trioleylglycerol as standard and the radioactive triacylglycerol bands (Rf 0.5–0.4 for incubations with or without dioleylglycerol, Rf 0.3–0.2 for those with dihexanoylglycerol) were detected with an automatic TLC-linear analyzer (Berthold), and scrapped off the plate for counting.

A large-scale, high-yield DGAT assay was set up for GC and GC-MS. The incubation mixture (2 mL) had the same composition as described above except that unlabeled oleyl CoA (40 µm) was used and a longer incubation time (2 h) was applied. After purification, the triacylglycerol formed from dihexanoylglycerol was quantified by GC using the same column and temperature program as for steryl acetates. GC-MS analysis yielded the following spectrum: 552 (M+; 0.1), 534 (M+-H2O; 0.2), 438 (5), 437 (M+-hexanoyloxy; 5), 393 (oleyl + 128; 3), 363 (2), 340 (2), 339 (oleyl + 74; 2), 337 (2), 283 (oleic acid + 1; 5), 271 (M+-oleyloxy; 49), 265 (oleyl; 6), 264 (10), 257 (3), 227 (hexanoyl + 128; 10), 173 (hexanoyl + 74; 21), 99 (hexanoyl; 100). These characteristic triacylglycerol fragmentation ions, as described by Barber et al. [31] and Christie [32], correspond unambiguously to 1,2-dihexanoyl 3-oleyl glycerol.


Cloning of a cDNA encoding DGAT from A. thaliana

In order to isolate a full-length cDNA clone encoding an Arabidopsis enzyme able to acylate plant sterols, we started from an EST clone (GenBank accession number AA042298). After sequencing, this EST cDNA was shown to contain 849 bp and the deduced protein sequence had 30 and 25% identity with H. sapiens[33] and S. cerevisiae[34] ACATs, respectively. This EST clone was used as a probe to screen an A. thaliana cDNA library, as described above. After screening 500.000 p.f.u., 12 cDNA clones of different lengths were isolated. Among them, 10 had a length compatible with those (≈ 2 kb) reported for mammalian and baker's yeast ACAT cDNAs. Two (3D11 and 3D71) were sequenced and contained 1845 and 2162 bp, respectively. They differed only by the respective length of their 5′-UTR and 3′-UTR. The 5′-UTR and 3′-UTR of 3D71 were 173 and 429 bp. Both were shown to encode a polypeptide of 520 amino acids (58 947 Da) presenting 25% identity with both H. sapiens and baker's yeast ACATs and 40% identity with an ARGP from H. sapiens[11] and a DGAT from Mus musculus[12] (Table 1). It was identical to that isolated by Hobbs [13].

Table 1. Comparison of the deduced amino acid sequence of A. thaliana DGAT with those of other acyltransferases showing % identity. The names and accession numbers are those in Fig. 1, except ScACAT1 and 2, which are the Saccharomyces cerevisiae sequences (P25268 and U51790, respectively) and CeACAT, which is the Caenorhabditis elegans sequence (Z68131).
NtDGAT  68100        
HsAGRP  40  38100       
MmDGAT  39  38  86100      
CeDGAT  36  37  47  47100     
HsACAT1  25  25  26  27  25100    
HsACAT2  24  26  25  26  25  49100   
ScACAT1  26  26  21  23  24  23  25100  
ScACAT2  26  27  23  24  26  28  24  49100 
CeACAT  20  19  24  24  21  34  31  23  25100

Cloning of a cDNA encoding DGAT from N. tabacum and comparison of its deduced protein sequence with those of other acyltransferases

A cDNA library from a 3-week-old N. tabacum cv. Xanthi line LAB 1-4 calli derived from leaf protoplasts [16] was screened using the A. thaliana EST clone. As described above, seven cDNA clones of different lengths were isolated. Among them one had a length (2099 bp) comparable with that of the Arabidopsis DGAT. It was shown to encode a polypeptide of 532 amino acids presenting 68% identity with Arabidopsis DGAT (Table 1). Strikingly, the first 100 amino acids from the N-terminal sequence show a low identity (19%) with the corresponding Arabidopsis sequence. After amino acid 135, the identity increases abruptly and becomes > 70%. The Kyte Doolittle hydrophobicity profile is very similar to that of the Arabidopsis DGAT with a hydrophilic moiety of 120 amino acids and 10 hydrophobic domains [13]. As in the Arabidopsis DGAT, and at exactly the same place, four leucines (L230, L237, L244, L251), are regularly spaced out by six variable amino acids forming a typical leucine zipper. This cDNA clone was called Nt DGAT1. Another cDNA clone had a length of 1354 bp. Its deduced amino acid sequence (317 amino acids) was much shorter than that of NtDGAT1 but had a high identity (86%) with it. This second cDNA (NtDGAT2) was of course incomplete but its presence in the library showed that there are at least two genes encoding DGAT in tobacco.

Sequence comparison of the DGATs from Arabidopsis and tobacco with the two human ACATs [33], murine DGAT [12], the putative human DGAT called ARGP [11] and an unknown cDNA from C. elegans (putative DGAT) is shown in Fig. 1. Remarkably, all these sequences present the same global pattern and are composed of a strongly hydrophilic moiety of ≈ 100 amino acids (in blue) followed by 10 hydrophobic domains (in red and boxed) which may correspond, in the case of DGATs, to nine membrane helix spanning domains as confirmed by the transmembrane predict program ( In addition all these cDNAs have several invariant domains in the second moiety of the deduced protein. A remarkably conserved domain (AELLCFGDREFYKDWW) is situated between amino acids 382 and 400 of the AtDGAT. This sequence is present in a hydrophilic domain situated between two supposed transmembrane helices. It is common to both DGATs and ACATs. Also worthy of interest is the presence in all DGATs and ACATs of an invariant serine (Ser254 in AtDGAT) which has been shown to be essential for ACAT activity [37]. However, the DGAT sequences can easily be distinguished from the ACAT sequences. The five DGAT sequences possess a remarkable motif consisting of four consecutive positively charged amino acids (four Arg in four of five DGATs, three Arg and one Lys in AtDGAT). This cluster is absent from all ACATs. Several other motives are also restricted to the five DGATs, one example is a sequence IERVLKL starting at Ileu352 of AtDGAT.

Figure 1.

Sequence alignment of various putative or demonstrated diacylglycerol or acyl CoA cholesterol acyltransferases. DGAT, acyl CoA:diacylglycerol acyltransferase; ACAT, acyl CoA:cholesterol acyltransferase; ARGP, ACAT-related gene product. Alignments were performed using the pileup program of the GCG package run with default parameters and the seqvu program. AtDGAT is the A. thaliana sequence (GenBank accession number AF051849). NtDGAT1 is the N. tabacum sequence (GenBank accession number AF129003). MmDGAT is the M. musculus sequence (GenBank accession number AF078752). HsARGP is the H. sapiens sequence (GenBank accession number AF059202). CeDGAT is the C. elegans sequence (GenBank accession number Z75526). HsACAT1 is the H. sapiens sequence (GenBank accession number L21934). HsACAT2 is the H. sapiens sequence (GenBank accession number AF059203). Hydrophilic domains are in blue. Ten hydrophobic segments (red) are boxed. Signs are placed under important amino acids (e.g. the Ser254 of the Arabidopsis cDNA) or highly conserved motives such as the invariant FGDREFYRDWWN segment. Clusters of arginines present in DGATs are underlined. Leucines involved in a putative leucine zipper are circled.

Organization of the gene encoding A. thaliana DGAT

In recent months a genomic sequence originated from chromosome II sequencing appeared in GenBank (AC005917) and was designated as ‘putative acyl CoA:cholesterol acyltransferase’. According to databases (gene prediction program grail), this sequence was processed into a polypeptide (441 amino acids) matching partially the protein (520 amino acids) deduced from the AtDGAT cDNA. Using the GTX and XAG rule defining introns boundaries for splicing, we reconstituted from a 5051-bp genomic sequence corresponding to the polypeptide of 441 amino acids, a cDNA perfectly matching our AtDGAT cDNA. According to this study the gene encoding the Arabidopsis DGAT was formed by 16 exons separated by 15 introns (Fig. 2). The sequence (817 bp) situated upstream of the ATG was submitted to the place signal scan program ( to identify cis-acting regulatory DNA elements possibly intervening in the regulation of DGAT expression. Interestingly an ethylene-responsive element (ATTTCAAA) [36] was shown to be present 719 bp upstream of the ATG. This important finding is discussed below.

Figure 2.

Organization of the gene encoding the diacylglycerol acyltransferase from A. thaliana. The part of the gene corresponding to the coding region (3020 bp) contains 16 exons and 15 introns. Exons are shown by continuous heavy lines, introns are shown by continuous thin lines. Parts of the gene upstream and downstream of the coding region are shown by broken lines. The numbers under the exons and above the introns are in bp. The motif ATTTCAAA at −719 bp is a putative EREE. The arrow indicates the translation start point determined by Hobbs et al. [13]. S,E,A: SacI, EcoRI, AseI. Letters b and e mark the 5′-end and the 3′-end of the longest cDNA (3D71).

If one now considers the sequence downstream of the ATG, it appears that the first exon (130 amino acids) corresponds exactly to the hydrophilic moiety of the protein. The second exon fits precisely with the first hydrophobic trans-membrane segment.

Expression of AtDGAT in a yeast mutant deficient in ACAT activity

Because this cDNA was first expected to encode an ACAT-like protein, it was expressed in the yeast double-mutant are1 are2under control of the GAL10-CYC1 promoter from the vector pYeDP60. Because in this mutant the two ACAT genes have been knocked out, sterol ester biosynthesis is absent [33] and microsomes from this yeast lack ACAT activity [32]. Sterol ester and free sterol contents were determined for AtDGAT-transformed are1 are2 and the void plasmid-transformed yeast. Both were devoid of sterol esters and had similar sterol profile (data not shown). Microsomes were prepared from both yeasts and ACAT activity was assayed. It was undetectable under conditions in which microsomes from the corresponding wild-type yeast, transformed or not with AtDGAT, displaid significant activity (160 pmol sterol esters per mg protein per min). Taken together, these results clearly showed that cDNA AtDGAT does not encode an ACAT.

Meanwhile, Cases et al. [12] described a mouse cDNA encoding a DGAT, with which AtDGAT shared 40% identity. Therefore, the DGAT activities of AtDGAT-transformed are1 are2 and void plasmid-transformed are1 are2 were measured and compared. Microsomes from both yeasts were incubated with [14C]oleyl CoA in the absence or presence of dioleylglycerol or dihexanoyl-glycerol (Fig. 3A). The results clearly show that although microsomes from the void plasmid-transformed yeast display a significant DGAT activity due to the yeast enzyme, microsomes from the AtDGAT-transformed yeast show a higher (more than double) activity, corresponding to the expression of both the yeast and the plant genes. These results further show that yeast and plant DGAT activities are measurable without added diacylglycerols, presumably because diacylglycerols are present in the enzymatic preparations or produced during the course of the incubation. The addition of dioleylglycerol or dihexanoylglycerol significantly increases both yeast and plant activities. Expression of AtDGAT in the corresponding wild-type yeast gave identical results (data not shown). Thus expression of cDNA AtDGAT in yeast produces a 2–3 times increase of DGAT activity in the microsomal fraction.

Expression of AtDGAT, AtFLAGDGAT and NtDGAT1 in wild-type yeast

The wild-type yeast was transformed with AtDGAT, a FLAG epitope-tagged version of AtDGAT called AtFLAGDGAT, a N. tabacum homolog of AtDGAT, called NtDGAT1, or the void plasmid. The triacylglycerol content of these strains was determined. The three DGAT-transformed yeasts had significantly higher triacylglycerol contents than the void plasmid-transformed yeast (25–80 versus 9 ± 3 nmol triacylglycerol per mg dry weight).

DGAT activity was measured in microsomes from yeast transformed with either AtDGAT, AtFLAGDGAT or NtDGAT1 in the presence of [14C]oleyl CoA and dioleyl-glycerol as substrates and shown to be significantly higher than that of the void plasmid-transformed yeast (20–24 versus 6 nmol triacylglycerol formed per mg protein per h). Taken together, these results obtained with the wild-type yeast confirm those obtained for AtDGAT with the mutant are1 are2, indicate that the FLAG epitope-tagged version of AtDGAT produces a functional protein and clearly show that NtDGAT1 encodes a DGAT.

The microsomal fractions from AtDGAT-transformed and AtFLAGDGAT-transformed yeast cells, were submitted to SDS/PAGE followed by a Western blot analysis using commercial anti-FLAG serum (Fig. 3B). The blot revealed a tagged protein with a molecular mass (54 kDa) close to that (59 kDa) calculated for AtDGAT according to its sequence.

Expression of AtDGAT in the floating lipid layer from the transformed yeast: identification of the enzymatic product

In the course of the preparation of microsomes from plant DGAT-transformed yeasts, we consistently observed the formation of a cloudy floating layer on top of the 100 000 g supernatants from the AtDGAT-tranformed, AtFLAGDGAT-transformed or, to a lesser extent, NtDGAT1-transformed yeasts, whereas the supernatant from the void plasmid-transformed yeast remained clear. These floating layers and the corresponding region of the 100 000 g supernatant from the void plasmid-transformed yeast were taken from the centrifugation tubes for analysis.

Extraction of lipids and purification of triacylglycerols by TLC revealed that the floating layer from the AtDGAT-transformed yeast had a much higher triacylglycerol mass (assessed by I2 visualization) than the 100 000 g supernatant top fraction from the void plasmid-transformed yeast (data not shown). Furthermore, the floating lipid layers from yeast transformed with AtDGAT, AtFLAGDGAT or NtDGAT1 were shown to display much higher DGAT activities than the 100 000 g supernatant top from the void plasmid-transformed yeast (5, 4 and 1 versus 0.1 nmol·mL−1·h−1). The presence of plant DGAT in the floating lipid layer from AtFLAGDGAT-transformed yeast was confirmed by SDS/PAGE followed by Western blot analysis (Fig. 3B).

The DGAT activities of the 100 000 g supernatant top fractions from yeast transformed with AtDGAT or the void plasmid were studied further. The activities were measured with [14C]oleyl CoA in the absence or presence of dioleylglycerol or dihexanoylglycerol (Fig. 3C). The 100 000 g supernatant top from the control yeast is almost devoid of activity, whereas the floating lipid layer from the AtDGAT-transformed yeast contains significant activity, 200–600 times higher than that of the control. Hence the floating lipid layer obtained from AtDGAT-transformed yeast exhibits a DGAT activity almost exclusively due to the plant gene. Because of the presence of BSA in the 100 000 g supernatant, specific activities could not be determined for this fraction. Its total DGAT activity represented one half to one sixth, according to the experiment, that of the corresponding microsomal fraction.

As observed with the microsomal preparations (Fig. 3A) the DGAT activity of the floating lipid layer is measurable without added diacylglycerols, probably for the same reasons. However, the addition of dioleylglycerol or dihexanoylglycerol significantly increases the activity of the plant enzyme expressed in yeast (Fig. 3C). Dihexanoylglycerol was again a good substrate and, because of its low molecular mass, the resulting triacylglycerol (Rf = 0.2) could be easily separated from the usual long-chain fatty acid triacylglycerols (Rf = 0.4) using TLC. This latter TLC band corresponds to unlabeled pre-existing triacylglycerols together with labeled triacylglycerols formed during the incubation from endogenous diacylglycerols. In order to identify unambiguously the product of the reaction catalyzed by the A. thaliana DGAT, we scaled up (20 times) the usual incubation mixture with dihexanoylglycerol and the floating lipid layer from AtDGAT-transformed yeast, lengthened the incubation time (to 2 h) and could thus purify 20 µg of the short-chain fatty acid triacylglycerol. GC-MS analysis of this pure product clearly confirmed its structure as 1,2-dihexanoyl 3-oleyl glycerol (see MS in Experimental procedures). It must be noted that the same high-scale, long-time incubation with the 100 000 g supernatant top fraction from the control yeast yielded 0.5 µg of this low molecular mass triacylglycerol. Under these conditions, the floating lipid layer from the AtDGAT-transformed yeast produced 40 times more 1,2-dihexanoyl 3-oleyl glycerol, which definitely proves the function of the protein encoded by AtDGAT.

Expression of AtDGAT in N. tabacum

AtDGAT cDNA encoding the ORF of the gene was introduced into N. tabacum under the control of a strong constitutive promoter (tandem CaMV 35S). Nineteen primary transformants grown in vitro on kanamycin-supplemented Murashige and Skoog medium were analyzed and compared with wild-type tobacco. The upper leaves from 6-week-old plants were separated into two batches: one for triacylglycerol dosage and the other for RNA gel blots.

The results shown in Fig. 4 clearly show that, in all cases, the expression of AtDGAT in tobacco is associated with a significant increase in the triacylglycerol content. In the best case, this increase was up to seven times the mean value measured for wild-type leaf material. The specificity of the expression was attested to by the fact that control tobacco did not exhibit any AtDGAT expression, showing that under our experimental conditions the AtDGAT probe did not hybridize with the endogenous NtDGAT RNA and allowed quantification of transgene expression only.

Figure 4.

Expression of At DGAT in N. tabacum.Lanes 1–19, primary transformants; lane 20, wild-type tobacco. (A) Triacylglycerol content. Triacylglycerols were extracted from young leaves, purified by TLC and quantified using a colorimetric assay. The column in lane 20 corresponds to the mean of three different wild-type plants (21 ± 5 nmol triacylglycerol per mg dry weight). (B) Northern analysis: totalleaf RNA from primary transformants (around 10 µg) was hybridized with a probe derived from the Arabidopsis DGAT. The arrow denotes the DGAT 1.5 kb transcript. (C) 18S rRNA hybridization of the same blot is shown to illustrate the relative amount of RNA loaded and transferred.

The primary transformants were then transferred to the greenhouse and grown under standard conditions. Leaf material was collected at the flowering stage (3.5 months old) and submitted to triacylglycerol dosage. The results were in full agreement with those obtained with the corresponding in vitro plantlets, i.e. an accumulation of triacylglycerols up to five times the control value (data not shown). To address the question of the intracellular location of the overproduced triacylglycerols, we stained petiole and limb epidermis peels from the wild-type and transgenic lines in an ethanolic solution of Sudan IV, a lipid-specific dye. Light-microscopy observations showed the presence of lipid droplets that appeared as orange spheres in the cytoplasm of stained cells from transgenic triacylglycerol-overproducer plants (Fig. 5C,D); control lines, as well as low-triacylglycerol-containing transgenic plants, showed very few (or none) of these lipid droplets (Fig. 5A,B). Taken together these data demonstrate that, as expected, AtDGAT is involved in triacylglycerol synthesis in planta.

Figure 5.

Light micrographs (× 850) of petiole or limb epidermis cells after staining in a solution of Sudan IV. (A) Wild-type petiole epidermis; (B) wild-type limb epidermis; (C) petiole epidermis of triacylglycerol-overproducer line 5; (D) line 5 limb epidermis. triacylglycerol dosage gave the following values for the samples shown in this figure: wild-type leaf material, 3.1 nmol triacylglycerol per mg dry weight; line 5 leaf material, 11.5 nmol triacylglycerol per mg dry weight.


As shown here cDNA clones encoding DGAT from Arabidopsis (AtDGAT) and tobacco (NtDGAT) have been characterized. This was carried out using their functional expression in yeast and in tobacco. At the time when this was performed, a similar study on the functional expression of AtDGAT in insect cells was published [13]. However, our work is original in the following aspects.

Protein sequences deduced from the cDNA clones

As shown in Table 1 the protein sequences deduced from the cDNAs or genes considered here can be divided into two families. The first is composed of proteins having ≈ 35–75% identity with the At DGAT. In this group we find the identified DGATs from Arabidopsis, tobacco and mouse, the human clone called ARGP, which was shown not to be an ACAT and suggested to be a DGAT [11], and a cDNA clone from C. elegans, which can be considered as a putative DGAT because of its relatively high identity (36%) with that of the Arabidopsis DGAT. The second group is composed of proteins having ≈ 25% identity with AtDGAT. In that group we find the two human ACATs (HsACAT1 and HsACAT2) and the two yeast ACATS (ScACAT1 and ScACAT2). As shown in the sequence alignments (Fig. 1), divergences between DGATs and ACATs are concentrated in the first moiety starting from the N-terminal part of the protein, while the second moiety of the protein is remarkably conserved among the seven sequences. These strong identities show clearly that ACATs and DGATs are closely related evolutionarily. This is not surprising because both enzymes catalyze the acylation by a fatty acyl CoA ester of strongly hydrophobic alcohols associated with endoplasmic reticulum: diacylglycerol for DGAT and cholesterol for ACAT. The other acyl CoA acyltransferases known to date, glycerophosphate acyltransferase, lysophosphatidate acyltransferase [38] and vindoline acetyltransferase [39], did not show any significant sequence identity suggesting that they are unrelated.

Expression of the Arabidopsis and tobacco DGATs in yeast

Whereas expression of the Arabidopsis cDNA cloned here, in the double mutant (are1 are2) defective in sterol acyltransferases, did not lead to either complementation of the mutation or detection of any ACAT activity at the microsomal level, a strong increase in the triacylglycerol content of the yeast and in the DGAT activity of its microsomes were observed. Because no yeast mutant defective in triacylglycerol synthesis has been reported to date [5] we used either are1 are2 or a wild-type yeast. Fortunately under the conditions used, the in vitro incorporation of [14C]oleyl CoA into triacylglycerols was significantly higher in microsomes from yeast transformed with AtDGAT-pYeDP60 than with the void vector (Fig. 3A). Under these conditions the increment of DGAT activity due to the expression of the plant cDNA culminated at ≃ 40 nmol triacylglycerol·h−1·mg protein−1, a value 102 times higher than that (300 pmol triacylglycerol·h−1·mg protein−1) reported in the insect expression system [13]. As significant endogenous DGAT activity was observed in control yeast, we tried to devise conditions to decrease this yeast activity. Such a goal was achieved using Triton X-100 in the incubation medium. Indeed the detergent almost completely inhibited the endogenous DGAT activity, but also decreased the plant DGAT activity in the AtDGAT-transformed yeast by a factor of 2 or more according to the experiments (results not shown).

The best results were obtained after the discovery that the floating lipid layer appearing during the course of the microsome preparation from yeast transformed with AtDGAT exhibited strong DGAT activity, whereas the corresponding zone of the 100 000 g supernatant from the yeast transformed with the void vector was almost devoid of activity (Fig. 3C). Such selective expression of AtDGAT in the floating lipid layer from the transformed yeast allowed us to demonstrate the net synthesis by AtDGAT of a specific triacylglycerol (1,2-dihexanoyl 3-oleyl glycerol), which could be separated thoroughly from endogenous triacylglycerols by TLC and clearly identified by GC-MS. Finally, the use of a N-terminal FLAG epitope in the cDNA AtDGAT allowed us to show that the enzymatic activity measured in transformed yeast can be ascribed to the presence of a protein with a molecular mass (54 kDa) compatible with that (59 kDa) deduced from the proteic sequence. As expected, this protein was shown to be present both in the microsomal fraction and in the floating lipid layer (Fig. 3B). Therefore, the yeast system appears to be appropriate for the expression of AtDGAT. It also allowed identification of a new DGAT from tobacco. The characterization of a DGAT from C. elegans is in progress in the same expression system.

As mentioned above, DGAT is a membrane-bound enzyme. In plants it is thought to be located in the endoplasmic reticulum [40–42] but its activity has also been suggested in oil body fractions from Brassica napus[43]. This dual localization of DGAT was also suggested in the case of yeast [44] and described in an oleaginous fungus [45]. As in animals, the lipid bodies (or oil bodies) of plants and yeast are believed to arise from specific microdomains of the endoplasmic reticulum membrane that contain the full set of triacylglycerol biosynthesis enzymes [7,42,46]. Hence, some of these enzymes might remain attached to the membrane of the lipid (oil) bodies after their budding [7]. Indeed such a dual localization has been described in yeast for some enzymes involved in triacylglycerol synthesis, glycerol 3-phosphate acyltransferase and 1-acylglycerol 3-phosphate acyltransferase [5], as well as enzymes of sterol biosynthesis, sterol methyl transferase ERG6 [47] and squalene epoxidase ERG1 [48]. The localization of the plant DGAT in At DGAT-transformed yeast is therefore not surprising even thought yeast DGAT is localized mainly in microsomes in our conditions of culture.

Expression of AtDGAT in tobacco

Although the gene AtDGAT was efficiently expressed in yeast to give an enzymatically active protein, this result did not give any information about the activity of this protein inside the plant. For a complete characterization it was necessary to show that the gene could be expressed in plants. To this purpose transgenic tobaccos expressing AtDGAT cDNA and overproducing triacylglycerols have been characterized. Although DGAT enzymatic assays were not performed here, it seems reasonable to assume that triacylglycerol overproduction corresponds to increased DGAT activity. Relevant to these considerations, an EMS-mutant of Arabidopsis showing a correlation between lower DGAT activity and poor triacylglycerol formation has been reported previously [49].

The accumulation of triacylglycerols in leaf material from transgenic DGAT plants was shown to be associated with the presence of lipid droplets in the cytoplasm of their cells (Fig. 5). Although this does not prove directly that the droplets contain exclusively the overproduced triacylglycerols and correspond to the well-defined plant lipid bodies [7], it is most likely that they are budding from the endoplasmic reticulum where triacylglycerols are synthetized. Synthesis and storage of triacylglycerols, for example in seeds, have been widely documented as a process in which these compounds accumulate in the endoplasmic reticulum membrane and bud off as droplets surrounded by a phospholipid monolayer [7]. We previously observed lipid droplets similar to those shown in Fig. 5, in the cytoplasm of cells from sterol-overproducer plant lines. Sterols, which are biosynthetized in the endoplasmic reticulum as triacylglycerols are, were shown to accumulate, in these plant lines, as fatty acid steryl esters [16,50,51].

Triacylglycerol accumulation has already been described in the case of a plant (Arabidopsis) expressing a yeast gene (encoding lysophosphatidic acid acyltransferase) [52]. Nevertheless, the present work constitutes the first report to show an accumulation of triacylglycerols in a transgenic plant expressing a plant gene (DGAT). This result suggests that DGAT activity would be limiting, at least in tobacco leaves, which normally produce small amounts of triacylglycerols. That the same stimulation of triacylglycerol synthesis that occurs in tissues such as tobacco flowers and seeds, but also in fruits and tubers in other species, remains to be shown.

These considerations raise the interesting problem of the regulation of the AtDGAT gene. The DGAT gene corresponding to the cDNA described here was sequenced recently on BAC F27F23 (GenBank accession number AC003058) and BAC F3 P11 (GenBank accession number AC005917) and was shown to be positioned on chromosome II. The region upstream of the translation starting point possesses interesting features. As shown recently [13] the transcription starting point is situated at −225 bp, leading to a leader sequence much longer than the average in higher plant genes. A similar situation has been observed in the gene encoding Δ7-sterol-C5-desaturase in Arabidopsis[18]. As shown previously [53] sequences contained within the 5′-transcribed, UTR may play a key role in regulation of hydroxymethyl glutaryl CoA reductase expression in Arabidopsis, therefore similar studies should be conducted in the DGAT gene. A characteristic ethylene responsive enhancer element (EREE) ATTTCAAA has been found in the promoter of genes activated by ethylene in carnation flower petals [36] and in ripening tomato fruit [54]. The same EREE motif has been found in the DGAT gene at −719 bp upstream to ATG. As triacylglycerols have been shown to accumulate in flower petals and DGAT was shown to be strongly expressed in Brassica napus flower petals [13], it may be suggested that the EREE motif would play an important role in the regulation of triacylglycerol synthesis.

Finally, it appears from the literature that triacylglycerol and steryl ester biosynthesis in animals and plants are closely interwoven. Doubtless the DGAT cDNAs described here constitute potent tools for studying these interactions. However, plant genes encoding the enzyme catalyzing the acylation of plant sterols have still to be characterized. Whilst a revision of this manuscript after it's review was in progress, another paper describing A. thaliana DGAT in the mutant line AS11 was published [55].


We are grateful to Dr Y. Kohara (National Institute of Genetics, Mishima, Japan) who kindly gave the EST cDNA yk453a2 from C. elegans and to Dr D. Pompon (Center National de la Recherche Scientifique, Gif sur Yvette) for allowing us to use the plasmid pYeDP60. We warmly acknowledge the skillful assistance of M. Schmitz for the transformation and in vitro culture of tobacco, A. Hoeft for the GC-MS, B. Bastian and L. Thiriet for typing the manuscript. We also wish to thank T. Husselstein, I. Benveniste and R. Bronner for their advice on yeast transformation, Western blotting and microscopic observation, respectively.


  1. Enzymes: acyl CoA:cholesterol acyl transferase (EC; acyl CoA:diacylglycerol acyltransferase (EC a web page is available at