Tgl4p and Tgl5p, Two Triacylglycerol Lipases of the Yeast Saccharomyces cerevisiae Are Localized to Lipid Particles*

Triacylglycerol (TAG) lipases are required for mobilization of TAG stored in lipid particles. Recently, Tgl3p was identified as a major TAG lipase of the yeast Saccharomyces cerevisiae (Athenstaedt, K., and Daum, G. (2003) J. Biol. Chem. 278, 23317–23323). Here, we report the identification of Tgl4p and Tgl5p as additional TAG lipases of the yeast. Both polypeptides, encoded by open reading frames YKR089c/TGL4 and YOR081c/TGL5, share 30 and 26% homology, respectively, to Tgl3p. Cell fractionation experiments and microscopic inspection of strains bearing Tgl4p-GFP and Tgl5p-GFP hybrids demonstrated that both proteins are localized to lipid particles similar to Tgl3p. A 1.7-fold increased amount of TAG enriched in myristic and palmitic acids and the reduced mobilization rate of TAG from tgl4Δ in the presence of the fatty acid synthesis inhibitor cerulenin demonstrated the lipolytic function of Tgl4p in vivo. In contrast, neither the total amount of TAG nor the TAG mobilization rate after addition of cerulenin was affected in tgl5Δ cells. However, the enrichment of C26:0 esterified to TAG of tgl5Δ, an additional increase of TAG in the tgl4Δtgl5Δ double deletion mutant compared with tgl4Δ, and the impairment of TAG mobilization in the tgl4Δtgl5Δ strain in the presence of cerulenin suggested that also Tgl5p functions as a TAG lipase in vivo. Most importantly, the purified His6-tagged Tgl4p and Tgl5p hybrids exhibited TAG lipase activity demonstrating their function in vitro. In summary, our data obtained by biochemical, molecular, and cell biological analyses unambiguously identified Tgl4p and Tgl5p as novel TAG lipases of yeast lipid particles with certain enzymatic specificities.

In contrast, neither the total amount of TAG nor the TAG mobilization rate after addition of cerulenin was affected in tgl5⌬ cells. However, the enrichment of C26:0 esterified to TAG of tgl5⌬, an additional increase of TAG in the tgl4⌬tgl5⌬ double deletion mutant compared with tgl4⌬, and the impairment of TAG mobilization in the tgl4⌬tgl5⌬ strain in the presence of cerulenin suggested that also Tgl5p functions as a TAG lipase in vivo. Most importantly, the purified His 6 -tagged Tgl4p and Tgl5p hybrids exhibited TAG lipase activity demonstrating their function in vitro. In summary, our data obtained by biochemical, molecular, and cell biological analyses unambiguously identified Tgl4p and Tgl5p as novel TAG lipases of yeast lipid particles with certain enzymatic specificities.
The neutral lipids triacylglycerols (TAG) 2 and steryl esters (SE) serve as an energy source and/or a source of building blocks (fatty acids and sterols) needed for membrane biogenesis. In addition, they are required as precursors for the synthesis of specific lipophilic components, e.g. steroid hormones or prostaglandins in higher eukaryotes (1)(2)(3)(4). Because TAG and SE are unable to integrate into phospholipid bilayers they cluster and form the hydrophobic core of so-called lipid particles. The structure of this cell compartment is rather simple and reminiscent of lipoproteins in mammals, consisting of a hydrophobic core of neutral lipids that is surrounded by a phospholipid monolayer with a small amount of proteins embedded (reviewed in Ref. 5).
Mobilization of neutral lipids from lipid particles is catalyzed by TAG lipases and SE hydrolases. Some proteins associated with the phospho-lipid monolayer of lipid particles, such as mammalian perilipins or oleosins of plants, were assumed to be involved in this process by serving as docking and/or activating proteins for hydrolytic enzymes. As an example, Londos and co-workers (6,7) demonstrated that perilipin, the most abundant protein of mammalian lipid particles, accelerated lipolysis catalyzed by the cytosolic hormone-sensitive lipase.
Localization of lipolytic enzymes in eukaryotes depends on the type of cell studied. As an example, Lehner et al. (8) reported the presence of a TAG lipase on cytosolic lipid droplets of pig liver cells. Spalinger et al. (9) demonstrated lipase activities in the cytosolic fraction of Caco2 cells, but also in the apical brush border membrane and other organelles. Information about the intracellular localization of plant lipolytic enzymes is rather limited, although many plant TAG lipases were isolated and characterized. It was reported that lipases of maize kernel were associated with oil bodies during germination (10). Similar to perilipins in mammalian cells the major proteins of oil bodies in plants, the oleosins, appear to act as receptors for binding and/or activation of TAG lipases (11). They seem to maximize the surface area of oil bodies, thus accelerating the mobilization of the storage oil from the hydrophobic core by lipases. Analysis of the lipid particle proteome of the yeast Saccharomyces cerevisiae did not identify polypeptides homologous to docking proteins like mammalian perilipins and oleosins of plants (12). Nevertheless, some yeast lipid particle proteins were identified as enzymes catalyzing degradation of TAG and SE (13)(14)(15). So far, Tgl3p has been the only TAG lipase identified in the yeast S. cerevisiae. In contrast to the mammalian hormone-sensitive lipase and plant lipases, Tgl3p is permanently associated with lipid particles and exhibits lipolytic activity independent of activator proteins (13). Detection of a small but significant TAG lipase activity in a tgl3⌬ deletion mutant (13) suggested the presence of at least one additional TAG lipase. This observation led us to search for proteins homologous to Tgl3p. In the present study, we report identification of Tgl4p and Tgl5p, two homologs of Tgl3p, as novel yeast TAG lipases localized to lipid particles. This conclusion is based on results from experiments in vitro and in vivo performed with wild-type, the single deletion strains tgl4⌬ and tgl5⌬, and strains bearing multiple mutations of TGL3, TGL4, and TGL5 as well as strains bearing tagged hybrids of the two novel TAG lipases.

MATERIALS AND METHODS
Strains and Culture Conditions-Yeast strains used throughout this study are listed in TABLES ONE and TWO. S. cerevisiae cells were grown aerobically in Erlenmeyer flasks at 30°C in YPD medium containing 2% glucose (Merck), 1% yeast extract (Oxoid), and 2% peptone (Oxoid). One hundred milliliters of culture medium were inoculated to an optical density (A 600 ) of 0.1 from a preculture grown aerobically for 48 h. Growth was monitored by measuring A 600 . For localization studies cells were grown in synthetic minimal 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 His 6 -tagged Tgl4p and Tgl5p from S. cerevisiae in the yeast Pichia pastoris, cells were grown on buffered minimal methanol medium (BMM10) containing 1.34% yeast nitrogen base, 4 ϫ 10 Ϫ4 % biotin (Sigma), 5% methanol (Merck), and 200 mM potassium phosphate (pH 6.0). For sporulation of S. cerevisiae a medium containing 0.3% potassium acetate (Merck) and 0.02% raffinose (Merck) were inoculated with a colony of a diploid strain grown on solid medium and shaken at room temperature. Sporulation efficiency was monitored by microscopic inspection.
For the construction of Tgl4p-GFP and Tgl5p-GFP, the GFP(S65T) sequence with the selectable markers, kanMX6 or His3MX6, were inserted into the chromosomal DNA of the coding region of TGL4 and TGL5, respectively, at either the N or C terminus. The fragment for insertion was amplified by PCR with the plasmid pFA6a-kanMX6-PGAL1-GFP(S65T) (pFA6a-His3MX6-PGAL1-GFP(S65T)) for N-terminal, or pFA6a-GFP(S65T)-kanMX6 (pFA6a-GFP(S65T)-His3MX6) for C-terminal tagging (16). Primers used for amplification are listed in TABLE THREE. PCR was performed as described previously (13). PCR fragments were ethanol precipitated, and 400 -700 ng were used for transformation of FY1679 applying the high-efficiency lithium acetate transformation protocol (17). Transformed cells were grown in YPD at 30°C overnight and then plated on YPD containing 200 mg of G418 (Invitrogen) per liter. For transformation with the insertion cassette containing the His3MX6 marker, cells were spread on HIS Ϫ plates immediately after transformation. After incubation for 2-3 days, large colonies were transferred to fresh selective plates. Only those clones that yielded larger colonies were considered as positive transformants and further tested for correct integration of the respective fusion cassette. Correct replacement was verified by analytical PCR of whole yeast cell extracts (18).
His 6 -tagged hybrids of Tgl4p and Tgl5p, respectively, were heterologously expressed in the yeast P. pastoris under the AOX1 promoter. Six codons coding for the amino acid histidine were integrated after the start codon AUG into the forward primer for the respective gene (TABLE THREE). The fragments for insertion were amplified by PCR from genomic DNA of a wild-type S. cerevisiae strain and integrated into the plasmid pPIC3.5K (Invitrogen). After linearization of the constructed plasmid the DNA fragment was transformed into the P. pastoris wild-type strain GS115 (Invitrogen) by electroporation. Transformed cells were spread on HIS Ϫ plates, and after 2-3 days larger colonies were transferred to fresh HIS Ϫ and G418 plates, respectively. Only clones that yielded large colonies on both selective plates were considered as positive transformants and tested for correct integration of the respective fusion cassette.
Isolation and Characterization of Subcellular Fractions-Lipid particles were isolated at high purity from yeast cells grown to the early stationary phase by the method of Leber et al. (19). Microsomal fractions, mitochondria, and cytosol were prepared as described by Zinser et al. (20). Relative enrichment of markers and cross-contamination of fractions were similar to those reported previously (21).
Protein Analysis-Protein was quantified by the method of Lowry et al. (22) using bovine serum albumin as a standard. Proteins were precipitated with 10% trichloroacetic acid and then solubilized in 0.1% SDS, 0.1 M NaOH. Prior to protein analysis samples of the lipid particle fraction were delipidated. Non-polar lipids were extracted with 3 volumes of diethyl ether, the organic phase was withdrawn, residual solvent was removed under a stream of nitrogen, and proteins were precipitated from the aqueous phase as described above.
SDS-PAGE was carried out by the method of Laemmli (23). Samples were dissociated at 37°C to avoid hydrolysis of polypeptides that may occur at higher temperature. Western blot analysis was performed as described by Haid and Suissa (24). Immunoreactive proteins were detected by enzyme-linked immunosorbent assay using rabbit or mouse antisera as the first and goat anti-rabbit or goat anti-mouse IgG, respectively, linked to peroxidase or alkaline phosphatase as the second antibody.  Purification of His 6 -tagged Proteins and Measurement of Enzymatic Activity-The isolated mitochondrial fraction (130 g of protein) of P. pastoris containing the His 6 -tagged Tgl4p and the cytosol (230 g of protein) of P. pastoris containing the His 6 -tagged Tgl5p, respectively, were solubilized on ice with sucrose monolaurat at a final concentration of 5 mM for 30 min. The solution was put on a Hi-Trap Chelating column (Ni 2ϩ ; Amersham Biosciences) and the flow through was collected. The column was washed with 8 volumes of 10 mM phosphate buffer (pH 7.4), 0.5 M NaCl, 10 mM imidazol prior to a stepwise elution with 0.5 ml of elution buffer, each. The buffer for starting the elution contained 10 mM phosphate buffer (pH 7.4), 0.5 M NaCl, 25 mM imidazol. In each of the subsequent steps the imidazol concentration was increased by 25 mM up to 250 mM imidazol.
TAG lipase activity was measured as described previously (13). In brief, radioactively labeled TAG was synthesized by growing a wild-type yeast strain overnight in the presence of [ 14 C]palmitic acid. Lipids were extracted (25), separated by TLC (12), and bands containing TAG were scraped off the plate, extracted, and quantified. To measure enzymatic activity of TAG lipases, 8 g of TAG (approximately 40,000 cpm) were dried under a stream of nitrogen, 150 l of 0.1 M potassium phosphate buffer (pH 8.0) and 20 l of a bovine serum albumin solution (20 mg/ml) were added and sonicated for 4 min in a sonicator water bath. Then, 50 l of 0.2 M MgCl 2 was added, and the mixture was pre-warmed at 30°C. Samples containing TAG lipases were added and an aliquot was immediately taken as a blank. After incubation for 30 min at 30°C, the reaction was stopped by addition of 3 ml of chloroform/methanol (2:1; v/v), and lipids were extracted (25). Lipids were separated by TLC using the solvent system light petroleum/diethyl ether/acetic acid (70:30:2; per volume) (12), bands were visualized with iodine vapor and scraped off the plate, and radioactivity in TAG, fatty acids, and diacylglycerols was measured by liquid scintillation counting.
SE hydrolase activity was determined essentially as described by Taketani et al. (26) and Zinser et al. (27). In brief, 16.5 g of cholesteryl oleate dissolved in CHCl 3 /MeOH (2:1, v/v) and 5 Ci of cholesteryl [1-14 C]oleate were mixed and taken to complete dryness under a stream of nitrogen. Then, 250 l of 0.1 M Tris/Cl Ϫ buffer (pH 7.4) and 20 l of a bovine serum albumin solution (20 mg/ml) were added and sonicated for 4 min. The mixture was pre-warmed to 30°C and 450 l of the selected fractions obtained by column chromatography (see above) were added. After incubation for 30 min at 30°C, the reaction was stopped by the addition of 4 ml of chloroform/methanol (2:1; v/v).
Lipids were extracted by the method of Folch (25) and neutral lipids were separated by TLC as described above (12). Bands corresponding to free fatty acids and SE were scrapped off the thin-layer plates, and radioactivity was determined by liquid scintillation counting.
Lipid Analysis of Whole Cell Extracts-Lipids of whole yeast cells were extracted by the procedure of Folch et al. (25) after disruption of cells with glass beads. Analysis of neutral lipids was performed as described previously (12). In brief, quantification of ergosterol and ergosteryl esters separated by TLC was carried out by densitometric scanning at 275 nm with ergosterol as a standard. TAGs were visualized on TLC plates by post-chromatographic staining and quantified by densitometric scanning at 400 nm with triolein as a standard. Fatty acids of TAGs were analyzed by gas liquid chromatography. TAG fractions isolated by TLC were subjected to methanolysis using BF 3 /methanol (14%), thus converting fatty acids to their methyl esters (28). Gas liquid chromatography analysis of fatty acid methyl esters was performed as described before (12) using a Hewlett-Packard 6890 equipped with a flame ionization detector operated at 320°C and a capillary column (Hewlett-Packard 5; 30 m ϫ 0.32 mm ϫ 0.25-m film thickness). Fatty acids were identified by comparison to commercial fatty acid methyl ester standards (NuCheck, Inc., Elysian, MN).
In Vivo Mobilization of Triacylglycerols and Steryl Esters-To measure mobilization of neutral lipids in vivo, cells were pre-grown for 20 h in synthetic minimal medium containing 2% glucose as the carbon source. Then, fresh synthetic minimal was inoculated with the pre-grown culture to an A 600 of 3. Either cerulenin at a final concentration of 10 g/ml for mobilization of TAG, or terbinafine at a final concentration of 50 g/ml for mobilization of SE were added from ethanolic stock solutions. Control incubations contained the equivalent volume of ethanol only. At the time points indicated, 10-ml aliquots of the culture were withdrawn and cells were harvested by centrifugation on a tabletop centrifuge. After freezing the pellet at Ϫ70°C for at least 1 h, cells were disintegrated by glass beading, and lipids were extracted and analyzed as described above.
Fluorescence Microscopy-Cells grown on synthetic minimal containing galactose and glucose, respectively, were harvested by centrifugation, washed, and put onto microscope slides. Staining with Nile Red (Sigma) was performed on the slide by mixing the cell suspension with a solution of 0.1 mg/ml Nile Red in ethanol. Microscopic analysis was performed using a Zeiss Axiovert 35 microscope with a 100-fold oil immersion objective and a UV lamp. The detection range of 450 to

Primers used for the construction of Tgl4p-GFP and Tgl5p-GFP
The underlined sequences are homologous to the plasmid pFA6a; double underlined sequences are restriction sites; bold sequences encode the His 6 tag.

RESULTS
Recently, Tgl3p was identified as the first and major TAG lipase of the yeast S. cerevisiae in our laboratory (13). However, the existence of additional TAG lipase(s) in the yeast was hypothesized, because a small but significant TAG lipase activity was still present in a tgl3⌬ deletion mutant. Indeed, homology searches led to the identification of the two unassigned open reading frames YKR089c and YOR081c (in the following named TGL4 and TGL5), which share 30 and 26% homology, respectively, with TGL3. The multiple alignments of the primary sequences of Tgl3p and its homologs Tgl4p and Tgl5p are shown in Fig. 1A. Similar to Tgl3p, Tgl4p and Tgl5p harbor the consensus motif GXSXG, which is conserved in lipolytic enzymes, but lack additional motifs characteristic for lipases from other sources. Kyte-Doolittle blots of Tgl4p and Tgl5p (Fig. 1B) revealed some hydrophobic domains in the polypeptides but no transmembrane spanning regions, a common feature of proteins localizing to lipid particles (12).
Phenotypic Analysis of Strains Deleted of TGL4 and TGL5-Analysis of the growth phenotype revealed that tgl4⌬ and tgl5⌬ single mutants, the tgl4⌬tgl5⌬ double deletion strain, and the tgl3⌬tgl4⌬tgl5⌬ triple mutant (see TABLE ONE) grew like wild-type (data not shown). The increased expression not only of TGL3, but also of TGL4 and TGL5 during sporulation (www.proteom.com/ YPD), however, tempted us to investigate the involvement of these proteins in spore formation in more detail. No spores were formed from a homozygous diploid strain deleted of TGL3 after 21 days. In contrast, tgl4⌬/tgl4⌬ and tgl5⌬/tgl5⌬ formed spores after 3 to 4 days similar to wild-type. Analysis of the sporulation efficiency using the homozygous double deletion strain tgl4⌬tgl5⌬/tgl4⌬tgl5⌬ revealed only poor spore formation after 16 days. Thus, deletion of both TGL4 and TGL5 resulted in a negative effect on sporulation. Not unexpectedly, the homozygous tgl3⌬tgl4⌬tgl5⌬/tgl3⌬tgl4⌬tgl5⌬ triple deletion mutant was unable to sporulate. Subcellular Localization of Tgl4p and Tgl5p-Storage of TAG, the substrate of TAG lipases, occurs exclusively in the hydrophobic core of lipid particles. Thus, localization of Tgl4p and Tgl5p close to their substrate similar to the major TAG lipase Tgl3p (13) was anticipated. To test this hypothesis, GFP hybrids of Tgl4p and Tgl5p were constructed, and localization of these polypeptides within the cell was determined.
Fluorescence microscopic inspection demonstrated that both proteins were indeed associated with lipid particles (Fig. 2, A and B) confirming data provided previously in a genome wide study by Huh et al. (29). To rule out the possibility that localization of Tgl3p, Tgl4p, and Tgl5p in lipid particles depended on the correct targeting of all three polypeptides, GFP hybrid localization of each single Tgl protein was tested in strains bearing double deletions of the two other respective open reading frames (see "Materials and Methods"). All three GFP hybrids were localized to lipid particles independently of each other (data not shown) indicating that these polypeptides do not necessarily form a functional or structural complex.
In addition to microscopic inspection, subcellular localization of Tgl4p and Tgl5p was investigated by Western blot analysis of isolated organelle fractions. Only the Tgl5p-GFP hybrid could be detected by this approach. It was localized to the lipid particle fraction (Fig. 2C) confirming data obtained by fluorescence microscopy. No signal was found in any fraction of a strain bearing a Tgl4p-GFP hybrid, most likely because of the poor expression of this peptide under growth conditions required for lipid particle isolation.
Lipolytic Activity of Tgl4p and Tgl5p-Sequence homology with the major TAG lipase Tgl3p (13) suggested that Tgl4p and Tgl5p may also exhibit lipolytic activity. Using isolated lipid particle fractions of strains deleted of TGL4 or TGL5, respectively, in a tgl3⌬ background as enzyme sources for TAG lipase assays, however, did not yield unambiguous results. We concluded from this experiment that Tgl4p and Tgl5p exhibited only low TAG lipase activities. Therefore, His 6 hybrids of Tgl4p and Tgl5p were constructed, purified by affinity chromatography, and subjected to enzymatic analysis (see "Materials and Methods"). Fig. 3, A and B, shows the gel electrophoretic pattern of fractions that were obtained by this method. The final product of this procedure appeared to be non-homogenous. However, immunochemical analysis of these preparations using an antibody directed against His 6 demon-  strated that all polypeptides of low molecular mass detected by SDS-PAGE separation of these samples were degradation products of the hybrid proteins (data not shown). This observation indicated that Tgl4p and Tgl5p are rather labile polypeptides. Specific activities of 0.5 Ϯ 0.04 g/mg min Ϫ1 for purified Tgl4p and 5.7 Ϯ 0.4 g/mg min Ϫ1 for purified Tgl5p were determined. In comparison, purified Tgl3p exhibited a specific activity of 12.3 g/mg min Ϫ1 (13). Nevertheless, this analysis unambiguously identified Tgl4p and Tgl5p as novel TAG lipases of the yeast.
We also performed SE hydrolase assays with isolated Tgl4p and Tgl5p. No significant activity was measured with both polypeptides.
Analysis of Neutral Lipids in Total Cell Extracts of Cells Deleted of TGL4 and TGL5-Previous work from our laboratory (12) had demonstrated that deletion of TGL3 led to an increased cellular level of TAG. TABLE FOUR shows that with the exception of tgl5⌬ all strains tested also contained a higher amount of TAG than wild-type. All strains with a higher amount of TAG exhibited also an increased ratio of TAG to SE. Noteworthy, this is also true for the tgl4⌬tgl5⌬ double deletion mutant. In addition, the cellular level of SE was increased over wild-type in all strains with the exception of tgl4⌬.
To analyze fatty acids in TAG from tgl4⌬ and tgl5⌬ mutants, neutral lipids were separated by TLC and fatty acids from the TAG fraction were analyzed by gas liquid chromatography (see "Materials and Methods"). For providing a better assessment of the fatty analysis data from strains deleted in combination with TGL3, data of the tgl3⌬ single deletion mutant are included in TABLE FIVE. The tgl3⌬ deletion strain contained a higher amount of C14:0, C16:0, and C26:0 and a lower amount of C18:0 than wild-type. In the tgl4⌬ mutant the amount of C14:0 and C16:0 was also higher than in wild-type, but amounts of other fatty acid species were similar to the control. In contrast, tgl5⌬ exhibited a marked increase of the C26:0 species in TAG, whereas amounts of other fatty acids were unchanged. Similar to all mutants deleted in combination with TGL3, tgl4⌬tgl5⌬ exhibited an increased amount of C14:0, C16:0, and C26:0 and a decrease of C18:0 and C18:1 confirming the synergistic effect of Tgl4p and Tgl5p. Moreover, these observations suggested that the three yeast TAG lipases exhibit certain degrees of specificity.
In Vivo Mobilization of TAG in Single and Multiple Mutants Deleted of TGL4 and TGL5-Changes in the lipid pattern of strains deleted of TGL4 and TGL5 strongly suggested that these polypeptides also exhibit lipolytic activity in vivo. To demonstrate this function, mobilization of TAG was tested by growing tgl4⌬ and tgl5⌬ strains in the presence of the fatty acid synthase inhibitor cerulenin. Whereas mobilization of TAG from lipid particles of tgl4⌬ was slower than in wild-type (Fig. 4B), no alteration of the mobilization rate was detected with tgl5⌬ (Fig. 4D). This result is in line with the more or less unchanged neutral lipid pattern of tgl5⌬. A strain deleted of both TGL4 and TGL5, however, mobilized TAG only very slowly (Fig. 4F). This is another indication for the synergism of these two TAG lipases. The triple deletion mutant tgl3⌬tgl4⌬tgl5⌬ did not mobilize TAG at all in the presence of the fatty acid synthesis inhibitor cerulenin (Fig. 4H). In growth experiments, tgl4⌬ and tgl5⌬ did not exhibit hypersensitivity to cerulenin (Fig. 4, A and C). In contrast, the tgl4⌬tgl5⌬ double deletion mutant and the tgl3⌬tgl4⌬tgl5⌬ triple deletion strain were more sensitive to the inhibitor than wild-type (Fig. 4, E and G).

Analysis of neutral lipids in BY4741 (wild-type) and deletion mutants of lipases
Mean values of three independent measurements with a mean deviation of Ϯ 10% are shown.    NOVEMBER 11, 2005 • VOLUME 280 • NUMBER 45

Triacylglycerol Lipases of the Yeast
Because the total amount of SE was increased in strains bearing deletions of TGL3 and TGL4 compared with wild-type (see TABLE FOUR) we also tested mobilization of SE in vivo in the respective mutants. When terbinafine, an inhibitor of squalene epoxidase (Erg1p) (30), was added to strains deleted of TGL3, TGL4, and/or TGL5, SE of lipid particles were mobilized at the wild-type rate, and no hypersensitivity of the mutants against this drug was observed (data not shown). This observation parallels the finding that purified Tgl4p and Tgl5p did not exhibit SE hydrolase activity in vitro (see above). These results also confirm the view that Yeh1p, Yeh2p, and Tgl1p are the only SE hydrolases in the yeast (14).

DISCUSSION
Here, we provide evidence that two homologs of the major TAG lipase Tgl3p (13), the polypeptides Tgl4p and Tgl5p, are additional TAG lipases of the yeast S. cerevisiae. This was concluded from data obtained by a combined biochemical, cell biological, and molecular biological approach including enzymatic analysis with isolated polypeptides, localization experiments with GFP-tagged Tgl proteins, lipid profiling of single and multiple deletion strains, and in vivo experiments under conditions of TAG depletion.
Whereas all types of experiments showed that Tgl4p fulfilled all criteria of a TAG lipase, some uncertainties concerning Tgl5p were realized. Although isolated Tgl5p exhibited TAG lipase activity in vitro, no effect of a TGL5 deletion on the mobilization of TAG in vivo was observed (see Fig. 4). Only the enhanced inhibitory effect on TAG mobilization in the tgl4⌬tgl5⌬ double deletion mutant compared with the tgl4⌬ mutant suggested that also Tgl5p is involved in TAG lipolysis of the living cell. The substrate specificity of Tgl4p for TAG containing myristic and palmitic acids, and of Tgl5p for substrates containing C26:0 as deduced from fatty acid patterns of TAG from tgl4⌬ and tgl5⌬ (see TABLE FIVE) may explain the result of the in vivo experiment. Because a significant amount of myristic and palmitic acids but only traces of C26:0 are esterified to TAG, the mobilization defect may become evident in tgl4⌬ but not in tgl5⌬. Several TAG lipases of different sources exhibit substrate selectivity toward fatty acyl chains of defined length (31,32). As an example, Schmitt et al. (33) demonstrated the critical role of specific amino acids of the fatty acid binding site of a TAG lipase from Candida rugosa that determines chain length specificity.
Localization studies demonstrated that all TAG lipases of S. cerevisiae identified so far are lipid particle proteins (see Ref. 13 and Fig. 2). Most recently, two other lipid particle proteins were identified as SE hydrolases (14,15). The presence of lipolytic enzymes close to their substrate is advantageous upon requirement of fatty acids and/or sterols for membrane and cell proliferation. Immediate access of these substrates to the respective enzymes may guarantee a rapid response. On the other hand, this spatial arrangement of enzymes and substrates raises the question, how TAG lipases can be prevented from permanent lysis of TAG stored in lipid particles. In analogy to other experimental systems, e.g. neutral lipid metabolism in higher eukaryotic cells, we can only speculate at present that a lipolysis/re-esterification cycle of TAG stored in lipid droplets (34,35) may occur also in yeast. The fact that yeast lipid particles do not only contain TAG lipases but also the TAG synthase Dga1p (36) supports this model. It has to be taken into account that deletion of Dga1p reduces the amount of TAG only to 70% of the wild-type level even in the presence of TAG lipases (36). This can be explained by (i) presently unknown mechanisms regulating the activity of TAG lipases, and/or (ii) permanent compensation of TAG degradation catalyzed by the Tgl proteins through transport of newly synthe-sized TAG from the endoplasmic reticulum, the second site of TAG synthesis (37).
Presently, we can only speculate why different TAG lipases are localized to yeast lipid particles. One reason could be the different substrate specificity of Tgl3p, Tgl4p, and Tgl5p toward fatty acids of TAG. Activation of one of these lipases could liberate specific fatty acids that are required for specific purposes. As an example, activation of Tgl5p exhibiting a preference for C26:0 could provide this long chain fatty acid species for ceramide synthesis (38). In contrast, the major TAG lipase Tgl3p exhibiting a broader substrate specificity, might supply fatty acids to general acylation processes.
In the past decade many lipid particle proteins were identified as enzymes contributing to lipid metabolism (12)(13)(14)(15)36). Thus, the original idea that lipid particles are simple storage compartments for neutral lipids serving as energy sources and/or sources of components needed for membrane biogenesis has to be revised. It has now become clear that lipid particles significantly contribute to cellular metabolism. Recently, Yang and co-workers (39) showed that neutral lipids are also involved in the on-or offset of apoptosis in the yeast. Data provided in this study suggest a "new" function of TAG, namely the involvement in sporulation. The sporulation defect of homozygous tgl3⌬/tgl3⌬ and tgl4⌬tgl5⌬tgl4⌬tgl5⌬ deletion mutants and the fact that expression levels of TGL3, TGL4, and TGL5 are increased during sporulation (www.proteom.com/YPD), indicate that mobilization of TAG from lipid particles may be required for the formation of spores. To support this hypothesis, a homozygous diploid quadruple mutant deleted of DGA1, LRO1, ARE1, and ARE2, whose gene products contribute to TAG and SE synthesis (37) and thus to lipid particle formation, was constructed. Similar to tgl3⌬/tgl3⌬, this dga1⌬lro1⌬are1⌬are2⌬/ dga1⌬lro1⌬are1⌬are2⌬ strain was impaired in spore formation. 3 This result is in line with the recent observation by Jandrositz et al. (15) that a homozygous diploid strain deleted of the lipid particle located SE hydrolase, Tgl1p was also unable to sporulate. Thus, an additional essential role of neutral lipids in yeast spore formation may be anticipated.