DGAT2 Is a New Diacylglycerol Acyltransferase Gene Family

Acyl CoA:diacylgycerol acyltransferase (EC2.3.1.20; DGAT) catalyzes the final step in the production of triacylglycerol. Two polypeptides, which co-purified with DGAT activity, were isolated from the lipid bodies of the oleaginous fungusMortierella ramanniana with a procedure consisting of dye affinity, hydroxyapatite affinity, and heparin chromatography. The two enzymes had molecular masses of 36 and 36.5 kDa, as estimated by gel electrophoresis, and showed a broad activity maximum between pH 6 and 8. Based on partial peptide sequence information, polymerase chain reaction techniques were used to obtain full-length cDNA sequences encoding the purified proteins. Expression of the cDNAs in insect cells conferred high levels of DGAT activity on the membranes isolated from these cells. The two proteins share 54% homology with each other but are unrelated to the previously identified DGAT gene family (designated DGAT1), which is related to the acyl CoA:cholesterol acyltransferase gene family, or to any other gene family with ascribed function. This report identifies a new gene family, including members in fungi, plants and animals, which encode enzymes with DGAT function. To distinguish the two unrelated families we designate this new class DGAT2 and refer to the M. ramanniana genes asMrDGAT2A and MrDGAT2B.

Diacylglycerol acyltransferase (DGAT) 1 is an integral membrane protein that catalyzes the final enzymatic step in the production of triacylglycerols in plants, fungi, and mammals (for reviews, see Refs. [1][2][3]. The enzyme is responsible for transferring an acyl group from acyl-coenzyme-A to the sn-3 position of 1,2-diacylglycerol (DAG) to form triacylglycerol (TAG). As the final step in TAG biosynthesis via the Kennedy pathway, it is the only step not involved in membrane biosynthesis. In plants and fungi, DGAT is associated with the mem-brane and lipid body fractions, particularly in oilseeds, where it contributes to the storage of carbon used as energy reserves. In animals, the role of DGAT is more complex. Triacylglycerols are synthesized and stored in several cell types including adipocytes and hepatocytes (4), but, in addition, DGAT may play a role in lipoprotein assembly and the regulation of plasma triacylglycerol concentration (4), as well as participate in the regulation of DAG levels (5,6).
Cases et al. (7) reported the first cloning of a DGAT gene from mouse. Using coding sequences from acyl CoA:cholesterol acyltransferase (ACAT), expressed sequence tag data bases were searched and a gene identified that shared 20% identity with the mouse ACAT. After cloning and expression of the gene in insect cells, no ACAT activity was detected in isolated membranes; however, using [1-14 C]oleoyl-CoA as substrate, a range of acceptors was examined and Cases et al. discovered DAG was the acceptor molecule, thus demonstrating DGAT activity. Hobbs et al. (8) reported the cloning of an Arabidopsis homologue of the mouse DGAT gene and confirmed the presence of DGAT activity in insect cells expressing the cDNA. Southern analysis indicated a single gene copy was present in Arabidopsis. Katavic et al. (9) and Zou et al. (10) also implicated this gene in seed oil production when an insertional mutation (AS11) in the gene was found to lower seed oil levels and decrease DGAT activity. The locus, at ϳ35 cM on chromosome II, was designated TAG1. Routaboul et al. (11) reported similar results identifying an Arabidopsis mutant (ABX45) harboring a frameshift mutation near the 5Ј end of the TAG1 reading frame. This mutation resulted in a complete change in coding sequence after the first 60 amino acids. With the identification of a single DGAT gene copy in Arabidopsis and the detection of DGAT activity even after a frameshift mutation disabled gene translation, Routaboul et al. concluded that another protein must be responsible for the remaining DGAT activity.
We chose the oleaginous fungus Mortierella ramanniana as our source material since the organism produces up to 80% of its dry weight as TAG when grown under nitrogen-limiting conditions. M. ramanniana had previously been identified as exhibiting high levels of DGAT activity, and is easily cultured in the laboratory (12,13). Our approach to the identification of DGAT involved protein purification, peptide sequencing, cloning of the corresponding cDNAs, and testing the gene products for DGAT function.
In this report a new class of proteins involved in TAG production was identified. Two polypeptides from M. ramanniana microsomes that co-purified with DGAT activity were sequenced and their corresponding cDNAs cloned. Expression of the cDNA sequences in insect cells conferred high levels of DGAT activity on membranes isolated from those cells. Although the genes encode proteins with DGAT activity, they are * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  unrelated to the previously identified DGAT gene family (now designated DGAT1), which is related to the ACAT gene family. We designate this new DGAT family DGAT2, and refer to the two genes in M. ramanniana as DGAT2A and DGAT2B. Based on nucleic acid comparison, we identified homologues of DGAT2 genes in fungi, plants, and mammals. We describe the cloning of several of these genes, their expression in insect cells, and confirmation of DGAT activity by enzyme assay.
Fungal Cultures-M. ramanniana was cultured as described by Kamisaka (12). Cells were harvested by passing 10 -13-day-old cultures through Miracloth and removing excess liquid by hand wringing. Wet packed cells were stored at Ϫ70°C.
Extraction-Lipid bodies were isolated from 70 -75 g of wet packed cells. Immediately prior to use, cells were thawed on ice and resuspended in 200 ml of Buffer A (10 mM potassium phosphate, pH 7.0, 1 M KCl, 0.5 M sucrose, 1 mM EDTA). Samples were lysed with an equal volume of 0.5-mm glass beads in a cell disrupter (Bead-Beater, Biospec Products, Bartlesville, OK) set on "Homogenize" for 45-90 s. The cell slurry containing glass beads was centrifuged at 500 ϫ g, the supernatant removed, and the pellets washed with another 5 ml of Buffer A. Following centrifugation, the supernatants from both centrifugations were combined. This was divided into six ultracentrifuge tubes (25 ϫ 89 mm), and each was overlaid with 5 ml of Buffer B (10 mM potassium phosphate, pH 7.0, 1 M KCl, and 0.3 M sucrose). Samples were centrifuged at 100,000 ϫ g at 4°C for 3 h. The lipid body fractions, floating on top of the overlays, were combined and solubilized in 50 ml of Buffer C (10 mM potassium phosphate, pH 7.0, 75 mM KCl, 0.5 M sucrose, and 1.5% Triton X-100). Non-solubilized material was removed by ultracentrifugation (90,000 ϫ g for 1.8 h). The floating lipid layer was discarded, and the supernatant containing the solubilized fraction (Triton X-100 extract) was retained for column purification.
DGAT Assay-DGAT activity was measured as the production of [ 14 C]triacylglycerol from [1-14 C]oleoyl-CoA and unlabeled dioleoyl-DAG. For non-solubilized samples, the reaction mixture (0.1 ml) consisted of enzyme extract, 3.67 M [1-14 C]oleoyl-CoA, and 1.5 mM 1,2-18:1 diacylglycerol in a buffer containing 10 mM potassium phosphate, pH 7.0, 100 -150 mM KCl, and 0.1% Triton X-100 (w/v). Assay mixtures were incubated at 25°C for 5 min and reactions terminated by adding 1.5 ml of heptane:isopropanol:0.5 M H 2 SO 4 (10:40:1, v/v/v). For solubilized samples 1,2-18:1 DAG was reduced to 0.5 mM, Triton X-100 was increased to 0.2%, and 300 M L-␣-phosphatidic acid was included. The L-␣-phosphatidic acid was required to recover activity following solubilization with detergent as described by Kamisaka et al. (13), except we found that 300 M rather than 500 M phosphatidic acid gave a greater stimulation of activity. Following solubilization, product formation was dependent on the addition of exogenous DAG. Under these conditions the reaction rate was linear with respect to time for up to 10 min.
After the assay was stopped, radiolabeled glycerolipids were isolated by adding 0.1 ml 1 M NaHCO 3 and 1 ml of heptane containing 15 nmol/ml triolein as a carrier. The mixture was vortexed and the upper organic phase was removed to a new glass vial. The organic extract was back-extracted with 1 ml 1 M NaCl. Forty percent of the final organic phase was removed for liquid scintillation counting and the remaining organic phase evaporated to dryness under nitrogen gas. The residue was resuspended in hexane and subjected to TLC on Silica gel-G with a preadsorbent loading zone (model 31011; Analtech, Newark, DE). The TLC plate was developed in hexane:diethyl ether:acetic acid (50:50:1, v/v/v), dried, and scanned by a radioimage analyzer (model 3000; AM-BIS, San Diego, CA) to determine the portion of radioactivity incorporated into TAG. Confirmation of TAG identity on the TLC plate was determined by comigration of the unlabeled triolein carrier and the [ 14 C]TAG following exposure to iodine vapor.
DGAT Purification-DGAT activity in the Triton X-100 extract was further purified by dye-binding chromatography on a Yellow 86-agarose column (2.5 cm ϫ 6.4 cm) equilibrated with 75 mM KCl in Buffer D (10 mM potassium phosphate, pH 7.0, 0.1% (w/v) Triton X-100, 10% (w/v) glycerol). The column was washed with five volumes of equilibration buffer at 2 ml/min, and then activity was eluted with 500 mM KCl in Buffer D. DGAT activity was stable to freeze/thaw at this stage of purification so eluted fractions were assayed immediately and active fractions stored at Ϫ70°C. Four preparations of Yellow 86-agarosepurified activity were combined and concentrated 12-fold by ultrafiltration (YM-30 membrane, Amicon, Beverly, MA). The activity was further purified by hydroxyapatite chromatography on a 1.0 cm ϫ 25.5 cm column equilibrated with 500 mM KCl in Buffer D. The column was washed with 40 ml of equilibration buffer before bound proteins were  eluted with a step gradient to 100 mM dipotassium phosphate in the equilibration buffer. Fractions containing DGAT activity were pooled and diluted 1:3.3 in Buffer D to reduce the KCl concentration from 500 to 150 mM. The diluted sample was applied to a heparin column (0.55 ϫ 4.7 cm) equilibrated with 150 mM KCl in Buffer D. The column was washed with five volumes of equilibration buffer at 0.5 ml/min, and bound proteins were eluted in a 10-ml linear gradient of 150 -500 mM KCl followed by 10 ml of 500 mM KCl in Buffer D at 0.25 ml/min. Fractions of 1.1 ml were collected. Protein Determination-The protein concentration of extracts was determined according to Bradford (14) using bovine serum albumin as standard.
SDS-PAGE-Polyacrylamide gradient gel electrophoresis (10 -13%) was carried out according to the method of Laemmli (15) with some of the modifications of Delepelaire (16). The resolving gel contained a 10 -13% linear gradient of acrylamide stock stabilized by a 0 -10% linear gradient of sucrose. Proteins were visualized by staining with silver according to the method of Blum et al. (17) Partial Amino Acid Sequence Determination-Proteins in active fractions eluting from the heparin step were precipitated with 10% trichloroacetic acid, washed with ice-cold acetone, and resuspended in SDS sample buffer. Samples were subjected to SDS-PAGE, and the gel was stained with Coomassie Blue. Protein bands at apparent molecular masses of 36 and 36.5 kDa were excised from the gel and sent to a commercial laboratory (Argo Bioanalytica, Morris Plains, NJ) for analysis. Gel slices were digested in situ with trypsin, and the resulting peptides were separated by reversed-phase HPLC. Amino acid sequencing was performed on a model 473 Protein Sequencer (Applied Biosystems, Foster City, CA).
Isolation of Total RNA and cDNA Amplification-Total RNA was prepared from wet packed cells essentially as described by Jones et al. (18). The RNA was then used to synthesize double-stranded amplified cDNA using the Marathon cDNA amplification kit (CLONTECH Laboratories, Inc., Palo Alto, CA).
Polymerase Chain Reaction-Degenerate oligonucleotides were synthesized on an oligonucleotide synthesizer (Applied Biosystems model 394) and used as primers in polymerase chain reaction. The peptide sequences used for synthesizing the corresponding coding and complementary oligonucleotides were designed according to the partial amino acid sequence obtained. The Marathon cDNA was used as a template. The amplification mixture consisted of template, polymerase chain reaction buffer, 200 -300 ng of each primer, 2.5 mM dNTP, and 1 unit of AmpliTaq Gold polymerase (PerkinElmer Life Sciences) in 50 l. The amplification program consisted of one 10-min hold at 95°C, 30 cycles of denaturation (94°C, 30 s), annealing (62°C, 10 s, 10% ramp to 50°C, 15 s), and primer extension (72°C, 2 min). Products of the reaction were separated on a 0.7% agarose gels, excised, and then purified according to the QIAPREP DNA extraction handbook (Qiagen, Santa Clarita, CA). The purified products were cloned into the pCR2.1-TOPO vector (Invitrogen, Carlsbad, CA).
Rapid Amplification of cDNA Ends (RACE)-RACE reactions were completed according to the instruction manual for Marathon cDNA amplification kit using oligonucleotides designed from the products of the degenerate PCR. Gel-purified RACE products were cloned into the pCR2.1-TOPO vector.
Cloning of DGAT2 Homologs-Data base searches of the predicted proteins from the public genomic data bases of Caenorhabditis elegans yielded three similar sequences. Searches of the public Saccharomyces cerevisiae predicted protein data base yielded one sequence. Searches of proprietary Arabidopsis expressed sequence tag data bases yielded partial sequences that were sufficient for PCR primer design. Total RNA was collected from these three organisms, and first strand cDNA libraries were created using the Marathon cDNA library kit (CLON- Solubilized lipid body proteins were applied to a Yellow 86-agarose column in buffer A containing 75 mM KCl. DGAT2 activity was eluted in buffer A containing 500 mM KCl. Protein content was determined according to the method of Bradford (14) and is reported as milligrams of protein per fraction. DGAT2 activity is reported as nanograms of TAG formed per minute per fraction. Active fractions from four Yellow 86-agarose columns were pooled and concentrated 12-fold by ultrafiltration. B, hydroxyapatite chromatography. The 500 mM KCl concentrate was applied to a hydroxyapatite column in Buffer D containing 500 mM KCl. The column was washed with equilibration buffer, and bound proteins were eluted with 0.1 M potassium phosphate in equilibration buffer. Active fractions present in the flow-through were pooled and diluted 1:3.3 to reduce the KCl concentration to 150 mM. C, heparin chromatography. The diluted hydroxyapatite flow-through was applied to a heparin column in Buffer D containing150 mM KCl. The column was washed with equilibration buffer, and DGAT2 activity was eluted in a linear gradient of 150 -500 KCl in Buffer D followed by a wash of 500 mM KCl in Buffer D. DGAT2 activity was resolved into two peaks. TECH). The primers in Table I were used to PCR-amplify each of the sequences. The PCR products were cloned into the pCR2.1-TOPO vector.
DNA Sequence Determinations-DNA sequence determinations were carried out using a modified protocol from Applied Biosystems. Sequence analyses were carried out using software of the Gen Codes Corp. (Ann Arbor, MI).
Expression of DGAT2 Genes in Insect Cells-The commercial BACto-BAC baculovirus expression system (Life Technologies, Inc.) was used to express full-length proteins in cultured insect (Sf9) cells. Fulllength DGAT2 open reading frames were amplified by PCR employing primers containing restriction sites at the 5Ј ends (NotI and SpeI to the sense primers and PstI to the antisense primers). The PCR products were cloned into the pCR2.1TOPO vector and sequenced to confirm the fidelity of the constructs. Full-length cDNAs in pCR2.1-TOPO vectors were digested with NotI and PstI and cloned into the NotI and PstI restriction sites of the pFASTBAC1 vector (Life Technologies, Inc.).
Insect cells (1 ϫ 10 6 cells/ml) were infected at a multiplicity of infection of 0.05-0.1 and harvested after 5 days at 27°C by centrifugation. Pelleted cells were resuspended in Buffer E (100 mM Tricine-NaOH, pH 7.8, 10% glycerol, 100 mM NaCl) and lysed by sonication (2 ϫ 10 s). Cell walls and other debris were pelleted by centrifugation and discarded. Membranes were harvested by centrifugation of the supernatant fraction (100,000 ϫ g for 1 h), and pellets were resuspended in Buffer E for enzyme assay. Assays were linear with respect to protein and time.
TAG Production in Insect Cells-Transformed insect cells were assayed for triacylglycerol and phosphatidylcholine by the following methods. An insect cell culture suspension was diluted to a standard optical density (usually 0.5) at an absorbance of 600 nm with culture medium. To a sample of 4.5 ml of insect cells in culture medium, 200 l glacial acetic acid, internal standards consisting of 12.5 g of C17:0 TAG and 25 g of C15:0 phosphatidylcholine, and 10 ml of chloroform: methanol (1:1, v/v) were added. After vortexing, the phases were separated by centrifugation (about 500 ϫ g, 5 min). The lower, organic phase was saved, and the upper, aqueous phase was re-extracted. The two organic extracts were combined and evaporated under nitrogen gas to a final volume of 0.4 ml. Twenty-five percent of the final volume was spotted onto a hard layer silica gel-GHL TLC plate with inorganic binder (Alltech Associates, Inc., Newark, Delaware). The TLC plate was developed for 30 min in hexane:diethyl ether:acetic acid (80:20:2, v/v/v) containing 20 mg/100 ml propyl gallate as an antioxidant. The plate was dried, sprayed with 0.001% primuline in 80% acetone, and the lipid bands identified under UV light. The TAG and phospholipid bands were scraped from the TLC plate into glass vials. The samples were methanolyzed in 2 ml of 5% H 2 SO 4 in methanol at 90°C for 2 h. After cooling, 2 ml of 0.9% NaCl and 0.50 ml of hexane were added and the top hexane layer analyzed for fatty acid methyl esters by gas chromatography (18). Table II. Initial steps included homogenization of the fungal paste, isolation of the lipid bodies by centrifugation, and solubilization of the membrane-bound proteins using the detergent Triton X-100. In the early stages of purification, high salt and detergent concentrations were necessary to maintain the solubility of the hydrophobic proteins. Enzyme activity was stable through the first column, Yellow 86-agarose (Fig. 1A), but was rapidly lost during subsequent purification. For that reason, scale-up occurred by pooling and concentrating the eluted fractions from four Yellow 86-agarose preparations.

Enzyme Purification-A summary of the purification of two proteins from M. ramanniana is presented in
In order to maintain maximal activity, subsequent chromatography was performed and fractions assayed on the same day. Significant purification was achieved using hydroxyapatite chromatography (Fig. 1B). Although DGAT activity did not bind the column, 64% of the protein present bound the column and was removed. Active fractions from the flow-through of the hydroxyapatite column were purified on heparin CL 6B-agarose (Fig. 1C). Two activity peaks eluted from the heparin column, one during the 100 -500 mM KCl gradient and one during the 500 mM KCl wash. Several protein bands (36.5, 36, 35, and 34 kDa) were associated with the first peak of activity (Fig. 2, fraction 22). The 34-kDa band did not correlate with DGAT activity in all chromatographic steps so it was eliminated (data not shown). The second peak had a higher specific activity (Table II) and contained a major protein band at 36 kDa by SDS-PAGE (Fig. 2, fraction 28). Three proteins (36.5, 36, and 35 kDa) were identified from the purification as potential DGAT candidates.
Partial Amino Acid Sequence Determination and Cloning of Purified DGAT2 Polypeptides-The three proteins associated with DGAT activity were gel-purified by SDS-PAGE, stained with Coomassie Blue, and then excised for protein sequencing. In-gel digestion of the proteins was performed using trypsin, and peptides were purified using reversed-phase HPLC. Examination of the peptide maps revealed that the 36.5-kDa map and the 35-kDa map were identical. Only peptides from the 36.5-kDa band were sequenced. The peptide map of the 36-kDa protein was significantly different than that of the 36.5/35-kDa proteins, and several of these peptides were sequenced.
Degenerate primers (Fig. 3), designed from the amino acid sequences generated from the 36-kDa peptide, were constructed in both sense and antisense orientations. These primers were employed in different combinations to amplify cDNA produced from M. ramanniana total RNA. PCR products were cloned into pCR2.1-TOPO and analyzed by DNA sequencing. Comparisons between peptide sequences obtained by Edman degradation not used to design the primers and the deduced amino acid sequences of PCR products were used to confirm the identity of the fragments. RACE using primers specific to these fragments was performed to yield a 1280-bp cDNA. This cDNA, which was designated DGAT2A (accession no. AF391089), contains a large open reading frame starting at bp 1. The most 5Ј ATG codon of this reading frame is located at bp 26, allowing for the translation of a 355-amino acid polypeptide (DGAT2A, Fig. 3). 2 A similar strategy was employed to clone the cDNA encoding the 36.5-kDa protein. We observed similarities between peptide sequences obtained from the 36-and 36.5-kDa polypeptides. Therefore, degenerate oligonucleotide primers were designed to the sequences of the 36.5-kDa peptide that had the least homology to the 36-kDa protein (Fig. 3). Evolutionary PCR, combined with RACE using primers specific to these fragments, was performed to yield a 1133-bp cDNA. This cDNA, which was designated DGAT2B (accession no. AF391090), contains a single large open reading frame from the 5Ј end to bp 1086. The most 5Ј ATG codon of this reading frame is located at position 40, which allows for the translation of a 349-amino acid polypeptide (DGAT2B, Fig. 3). 2 Both designated MrDGAT2 ATG codons are followed by a G residue, the consensus nucleotide for initiation of translation in eukaryotes at this position.
DGAT2A encodes a polypeptide of a calculated molecular 2 Sequence is also present in Patent Application WO 00/01713. mass of 40,602.5 Da, and a theoretical pI value of 9.18. DGAT2B encodes a polypeptide with a calculated molecular mass of 39595.49 Da, and a theoretical pI value of 9.40. These predicted molecular sizes fit very well with the apparent mass of the purified proteins, which indicates that DGAT2 polypeptides do not undergo major posttranslational proteolytic processing in vivo. The two polypeptides share 54% identity at the protein level (Fig. 3, top two sequences).
GenBank™ searches showed that these polypeptides are not sequence-related to the known DGAT1 or any other acyltransferases, but were members of a previously unannotated gene family present in major phyla of eukaryotes, in particular fungi, plants, animals, and basal eukaryotes (Fig. 3). An alignment of members from different major eukaryotic phyla shows that these sequences are approximately conserved in length and they co-align over large stretches, with about 10% of totally conserved residues dispersed throughout. This represents strong evidence for common evolutionary origin. A preliminary phylogenetic tree built from currently available sequences (Fig.  4) shows clustering of sequences by systematic relationship of species, indicating that DGAT2 gene variations, as found in Morteriella, C. elegans, and mammals, appear to be the result of relatively late gene duplications, having occurred after the divergence of the respective main branches of eukaryotes.
Insect Cell Expression and Characterization-The two putative DGAT genes identified in M. ramanniana were expressed in an insect cell system to determine if they indeed encoded polypeptides with DGAT activity. Membranes from baculovirus-infected insect cells expressing DGAT2 cDNAs were harvested and assayed for activity. A significant elevation in DGAT activity was detected relative to untransformed Sf9 cells for both DGAT2A and DGAT2B proteins of 94-and 37-fold, respectively. (Fig. 5A).
Full-length clones were obtained for several of the genes whose sequences showed homology to the M. ramanniana DGAT2 genes. The genes (S. cerevisiae DGAT2; C. elegans DGAT2A, 2B, and 2C; and Arabidopsis thaliana DGAT2) were selected from different phyla to test the relatedness of protein function. These cDNAs were expressed in insect cells, and the isolated membranes were assayed for DGAT activity. A 2-180fold increase in DGAT activity was observed, relative to untransformed Sf9 cells, confirming that the genes encode proteins which are related by function (Fig. 5, A and B). Since we did not tag the gene product, we were unable to determine the amount of each protein produced and normalize the data.
In addition to the observed increase in DGAT activity, we also detected a 2.7-fold increase in the amount of TAG present in insect cells expressing the M. ramanniana DGAT2A gene relative to untransformed Sf9 cells. When the samples were normalized with respect to phospholipid content, the -fold increase in TAG was 3.1. Results of the triacylglycerol analysis demonstrate that overexpression of the M. ramanniana DGAT2A gene leads to an increase in the production of triacylglycerols in insect cells.
We also investigated some of the enzymological properties of the expressed M. ramanniana DGAT2A and DGAT2B genes. The effect of pH on DGAT activity was evaluated from 4.0 to 11.0. The pH optimum for both enzymes was observed at 6.8. No differences were detected between the two polypeptides with respect to pH (data not shown). A difference was observed in their response to temperature. The temperature optimum for DGAT2A was 37°C, whereas DGAT2B did not demonstrate an optimum temperature (Fig. 6). The polypeptides were also characterized with respect to their ability to utilize two different acyl-coenzyme A donors, 18:1 and 12:0, and a range of diacylglycerol acceptors (6:0 through 18:0, even numbers, and 18:1) (Figs. 7, A and B). We detected an enhanced capacity for the utilization of medium-chain substrates (6:0 to 10:0) for both DGAT2A and DGAT2B proteins. Since we did not determine the specificity constants (V max /K m ) for the various substrates supplied, these data are preliminary and should be substantiated by further investigation. DISCUSSION We have isolated novel DGAT proteins from cells of the oleaginous fungus M. ramanniana. Following cell lysis, DGAT activity was associated with the lipid body fraction and detergent solubilization was required to release the membranebound proteins to permit their purification using traditional chromatographic techniques. A stimulation of DGAT activity in the homogenate was observed following the addition of the detergent Triton X-100. Using a five-step protocol, two proteins, 36 and 36.5 kDa by SDS-PAGE, were identified as being associated with DGAT activity. Final specific activity recoveries of 1.6 and 4.2%, respectively, were reported for the purest, most active fractions containing each protein. Expression of the cloned cDNAs in insect cells allowed the unambiguous confirmation of DGAT activity as being associated with the two polypeptides. Alignment of the two protein sequences indicates they share only 54% sequence similarity (Fig. 3, top two lines).
Our purification of M. ramanniana DGAT differs from that reported by Kamisaka and co-workers (19), who identified a 53-kDa protein (by SDS-PAGE) as DGAT. The two polypeptides we identified corresponded to 36-and 36.5-kDa (by SDS- PAGE). In addition, all other identified DGAT2 polypeptides from other species (Fig. 3) are approximately in the 33-42-kDa range. Since apparent and predicted molecular mass values match approximately, it is likely that the proteins we isolated represent unprocessed DGAT2 polypeptides. It is noteworthy that using our assay, the 36-and 36.5-kDa polypeptides were the only protein bands we observed that correlated with DGAT activity throughout purification.
An unexpected observation of the characterization of M. ramanniana DGAT2 proteins isolated from insect cells was the enhanced activity with medium-chain substrates. M. ramanniana produces TAG comprising primarily C18 acyl groups, yet more activity was detected when C6-C10 DAGs were provided as the acyl acceptor, especially when a medium chain donor (12:0-CoA) was used. Although absolute activity values cannot be compared between samples because of differences in the level of protein expression in different insect cell lines, DGAT2A appears to have greater specificity for medium-chain substrates relative to long-chain substrates than does DGAT2B. Whether the observed activities with medium-chain substrates are a unique feature of the M. ramanniana enzymes or an artifact due to differing solubilities of the hydrophobic substrates remains to be determined. If true, these findings offer intriguing possibilities for the use of M. ramanniana genes in the engineering of unusual fatty acids in plant seed oils.
A search of the sequence data bases using the deduced amino acid sequences of the two M. ramanniana clones revealed no homology with the previously identified DGAT1 gene family that is sequence-related to the ACAT gene family. Unidentified DGAT2 homologues were found in many eukaryotic species, namely animals, plants, fungi, and Leishmania, but were absent from the prokaryotes (Fig. 4). However, it is noteworthy to mention that several conserved signature amino acid residues of motifs D and E of the previously proposed acyltransferase superfamily (20) and motif IV of sn-glycerol-3-phosphate acyltransferase consensus (21) are also conserved in DGAT2 (see the sequence above the alignments in Fig. 3). Since acyl-CoA is the shared substrate used by all these diverse enzymes, we can only speculate that this motif might be related to acyl-CoA binding and might indicate a common origin.
Full-length clones were obtained for several homologues, and the expressed proteins were evaluated in insect cells. All of the homologues tested exhibited some level of DGAT activity, demonstrating that the genes in this family are related by function. These data confirm our discovery of a second DGAT gene family. The identification of a new DGAT gene family is consistent with previous biochemical observations (9,11). First, gene disruptions of DGAT1 (TAG1 locus) in Arabidopsis did not abolish DGAT activity completely or eliminate TAG production in seeds. Second, Smith et al. (22), working with DGAT1 knockout mice, concluded there may be an additional DGAT gene present in mammals when experimental data showed that TAG production still occurred in these animals. These data collectively supported the presence of an additional source for DGAT in plants and mammals. Cases et al. (35) report the cloning of a mouse DGAT2 cDNA, verify DGAT enzyme function in insect cells, and describe the DGAT2 mRNA distribution in mouse.
In addition to our discovery of a second DGAT gene family, a novel, alternative mechanism for the production of TAG has recently been reported in yeast (23,24). This pathway utilizes phospholipid, rather than acyl-coenzyme A, as a substrate for acyl transfer to DAG to produce TAG. The acyl-CoA-independent production of TAG during exponential growth in yeast was associated with the LOR1 gene (25,26). A knock-out of LOR1 resulted in the complete removal of the acyl-CoA-independent activity and a significant reduction in TAG accumulation. Dahlqvist designated this enzyme phospholipid:diacylglycerol acyltransferase (PDAT) since the enzyme apparently does not discriminate between phospholipid species supplying the acyl group. PDAT is structurally related to the lecithin:cholesterol acyltransferase family, and homologues of LOR1 appear to be common in eukaryotes. With this discovery, the contribution of PDAT as well as the newly discovered DGAT2 family to the overall production of TAG must be determined.
To date, three independent gene families (DGAT1, DGAT2, and PDAT) have been described that encode unique proteins with the capacity to form TAG, and all three are present in genomes of eukaryotes. It is possible the three gene families may play different roles in different species, in different tissues, or at different times during development. In yeast, for example, all three genes are present but their expression levels vary during different phases of the life cycle (26). In mice in which the DGAT1 gene was disrupted, certain tissues appeared to be more affected than others (22). For example, although the Dgat1 Ϫ/Ϫ mice showed only a 20% reduction in total carcass triglyceride, the female mice lost the ability to lactate. Examination of the breast tissue showed a severe reduction in lipid droplets, indicating DGAT1 plays a key role in this specific tissue. Dhalqvist et al. (26) proposed, in plant seeds, PDAT may be responsible for the selective shuttling of unusual fatty acids out of membrane lipids and into TAG. Microsomes isolated from developing seeds of species that produce large amounts of unusual fatty acids in their oil, such as ricinoleic acid in castor and vernolic acid in Crepis palaestina, preferentially incorporate these fatty acids into TAG. Further research is needed to elucidate the roles these three gene families play in different organisms.
TAG is an abundant molecule found in many forms of life most likely because of its high energy density. The ability to alter oil levels either up or down, depending on the species, is of commercial interest. For example, in humans fat storage has many implications in health maintenance and well-being and drug therapies are being developed to reduce its accumulation. In oilseeds, which are economically and nutritionally significant crops, increasing seed value by increasing stored TAGs is an important goal in agriculture. In this regard, the study of FIG. 7. DGAT2 substrate specificity profiles. Substrate profiles were obtained for DGAT2A and DGAT2B in insect cell membranes. Substrate specificity was determined with 18:1-CoA as acyl donor (A) and 12:0-CoA as acyl donor (B) using a range (6:0 -18:1) of DAG acceptors.
TAG synthesis is of consequence as we consider ways to manipulate the production of TAG. Researchers have successfully altered fatty acid composition of seed oils through biotechnology (28 -31); however, increasing fatty acid content has proved more elusive, although several reports have appeared in the literature containing preliminary evidence of success (32)(33)(34).
The manipulation of oil levels in model organisms can be achieved by expression of genes that increase DGAT activity. Expression of DGAT1 genes in insect cells (8), yeast (27), and tobacco leaf (27) and the overexpression of the PDAT gene in yeast (26) all resulted in an increase in TAG accumulation. We observed that expression of M. ramanniana DGAT2A in insect cells increased the total amount of TAG 2-3-fold in those cells. All of these genes show great potential as tools to increase TAG levels in oilseeds.