Identification of a Saccharomyces Gene,LCB3, Necessary for Incorporation of Exogenous Long Chain Bases into Sphingolipids*

To identify genes necessary for sphingolipid synthesis in Saccharomyces cerevisiae we developed a procedure to enrich for mutants unable to incorporate exogenous long chain base into sphingolipids. We show here that a mutant strain, AG84-3, isolated by using the enrichment procedure, makes sphingolipids from endogenously synthesized but not from exogenously supplied long chain base. A gene termed LCB3 (YJL134W, GenBank designation X87371x21), which complements the long chain base utilization defect of strain AG84-3, was isolated from a genomic DNA library. The gene is predicted to encode a protein with multiple membrane-spanning domains and a COOH-terminal glycosylphosphatidylinositiol cleavage/attachment site. Deletion of thelcb3 gene in a wild type genetic background reduces the rate of exogenous long chain base incorporation into sphingolipids and makes the host strain more resistant to growth inhibition by long chain bases. Only one protein in current data bases, the S. cerevisiae open-reading frame YKR053C, whose function is unknown, shows homology to the Lcb3 protein. The two proteins are not, however, functional homologs because deletion of theYKR053C open reading frame does not impair long chain base utilization or enhance resistance of cells to growth inhibition by long chain bases. Based upon these data we hypothesize that the Lcb3 protein is a plasma membrane transporter capable of transporting sphingoid long chain bases into cells. It is the first candidate for such a transporter and the first member of what appears to be a new class of membrane-bound proteins.

To identify genes necessary for sphingolipid synthesis in Saccharomyces cerevisiae we developed a procedure to enrich for mutants unable to incorporate exogenous long chain base into sphingolipids. We show here that a mutant strain, AG84-3, isolated by using the enrichment procedure, makes sphingolipids from endogenously synthesized but not from exogenously supplied long chain base. A gene termed LCB3 (YJL134W, GenBank designation X87371x21), which complements the long chain base utilization defect of strain AG84-3, was isolated from a genomic DNA library. The gene is predicted to encode a protein with multiple membrane-spanning domains and a COOH-terminal glycosylphosphatidylinositiol cleavage/attachment site. Deletion of the lcb3 gene in a wild type genetic background reduces the rate of exogenous long chain base incorporation into sphingolipids and makes the host strain more resistant to growth inhibition by long chain bases. Only one protein in current data bases, the S. cerevisiae open-reading frame YKR053C, whose function is unknown, shows homology to the Lcb3 protein. The two proteins are not, however, functional homologs because deletion of the YKR053C open reading frame does not impair long chain base utilization or enhance resistance of cells to growth inhibition by long chain bases. Based upon these data we hypothesize that the Lcb3 protein is a plasma membrane transporter capable of transporting sphingoid long chain bases into cells. It is the first candidate for such a transporter and the first member of what appears to be a new class of membrane-bound proteins.
Sphingolipids comprise a diverse group of complex lipids found primarily in the plasma membrane of all known eukaryotes. Knowledge of the structures and functions of sphingolipids has expanded at an increasing rate, but much less is known about the enzymes and genes needed for synthesis of sphingolipids. In one approach to identify sphingolipid biosynthetic genes, our laboratory devised a unique procedure for isolating mutants of Saccharomyces cerevisiae defective in sphingolipid synthesis. As described here, this approach led to the isolation and characterization of a strain defective in the incorporation of exogenous sphingoid long chain bases into sphingolipids.
Synthesis of the long chain base component of sphingolipids ( Fig. 1) begins with the condensation of serine and palmitoyl-CoA to yield 3-ketosphinganine. This essentially irreversible reaction is catalyzed by serine palmitoyltransferase (3-ketosphinganine synthase (EC 2.3.1.50); for review, see Ref. 1), encoded by the LCB1 (2) and LCB2 (3) genes. The 3-ketosphinganine is converted to the long chain base sphinganine (dihydrosphingosine) which is N-fatty-acylated to yield dihydroceramide. Dehydrogenation of dihydroceramide in animals yields ceramide containing the long chain base sphingosine, which is rapidly converted to sphingolipids by the addition of polar components to the 1-hydroxyl group. The predominant ceramides in fungi and plants contain N-␣-hydroxyfatty-acylphytosphingosine (4) formed by undefined hydroxylation reactions. Phytosphingosine (PHS) 1 lacks the 4,5-double bond found in sphingosine and has instead an hydroxyl group at the 4 position (Fig. 1).
In S. cerevisiae and other fungi the 1-hydroxyl of phytoceramide is modified by addition of myo-inositol phosphate to form IPC which is then mannosylated to yield mannose-inositol-Pceramide (MIPC). The final step in S. cerevisiae sphingolipid synthesis is the addition of inositol-P to MIPC to yield the major sphingolipid mannose-(inositol-P) 2 -ceramide (M(IP) 2 C; Refs. 5-7).
We describe here a rationale and procedure for isolating mutants defective in synthesis of intermediates formed between 3-ketosphinganine and MIPC (Fig. 1). The procedure was designed to avoid the potential problems of growth inhibition and killing of cells by the buildup of a sphingolipid intermediate. It is known that exogenous N-acetyl-sphingosine (C 2ceramide) inhibits growth of S. cerevisiae cells (8,9), as does exogenous PHS. 2 Growth inhibition and killing were avoided by using a sphingolipid compensatory (SLC) strain that can grow without making sphingolipids, unlike normal yeast strains that require sphingolipids for growth and viability (10). SLC strain 7R6 carries two mutations, a deletion of the LCB1 gene and a point mutation that creates the suppressor gene SLC1-1 (11). In the absence of exogenous long chain base, the suppressor gene causes cells to make novel glycerophospholipids that compensate for one or more functions of sphingolipids essential for vegetative growth (12). The lcb1 mutation blocks the first committed step in sphingolipid synthesis, so no toxic sphingolipid intermediate should accumulate when cells are grown on medium lacking a long chain base. Thus if an SLC strain becomes mutated in, for example, the ceramide synthase gene ( Fig. 1) the mutant strain should continue to grow without making sphingolipids because of the SLC1-1 suppressor gene and should not accumulate a toxic long chain base intermediate because of the lcb1 mutation.
The mutant derivative of 7R6 described here has an impaired ability to incorporate exogenous long chain bases into sphingolipids because of a mutation in the LCB3 gene (long chain base). The Lcb3 protein is predicted to have several membrane-spanning domains, suggesting that it is a membrane-bound transporter or a facilitator of long chain base uptake. It is structurally but not functionally homologous to another yeast protein of unknown function. The Lcb3 protein is not homologous to any other predicted protein, thus it appears to be the first member of a new class of membrane-bound proteins with a potential transporter function.

Strains, Plasmids, Media, and Reagents
S. cerevisiae strains used in these studies are described in Table I. Culture conditions have been described previously (13). Reagents hygromycin B, canavanine, DL-erythrosphinganine, stearylamine, and DLthreosphinganine were from Sigma. Cerulenin was from Makor Chemicals Ltd. Sphingofungin B was from Merck. N-Acetylsphingosine was from Matreya, Inc. D-Erythrosphingosine was from Serdary Research Laboratories. For phytosphingosine, see Ref. 14.
The LCB3 gene was isolated from a S. cerevisiae genomic DNA (strain 4R3, 10) library made by ligating Sau3AI-cut DNA fragments into the BamHI site of pRS315 (CEN6, LEU2, 15). The library was transformed (16) into strain AG84-3, and Leu ϩ transformants were selected on minimal agar plates lacking leucine, buffered with 50 mM sodium succinate, pH 5.5, and incubated at 30°C for 10 days. About 15,000 transformants were tested by replica plating for colonies able to grow at 37°C on defined agar plates lacking leucine and sodium succinate but containing 25 M PHS. These conditions permitted growth of the parental strain 7R6 but restricted growth of the mutant derivative AG84-3. One out of 15 colonies that grew under these conditions contained a plasmid, pJAB15-1, with a 6.5-kb insert, which, when isolated by transformation into Escherichia coli and retransformed into strain AG84-3, gave transformants able to grow at 37°C in the presence of PHS. The DNA sequence of a portion of the yeast insert was determined and compared with the GenBank data base using the BLAST algorithm (17). Predicted restriction sites were used to subclone portions of the insert, carrying the complementing activity, into pRS315, yielding a 2.4-kb SpeI-MscI DNA fragment containing only one large open reading frame, YJL134W. Other subclones failed to complement.
The lcb3-⌬1 deletion allele has the 522 base pairs between the BamHI and the NsiI sites of the LCB3 coding region replaced with pRS305 (LEU2, 15) to give MPRS305. MPRS305 was constructed by cloning a 3.1-kb PstI-BamHI DNA fragment from pJAB15-1, containing the promoter and 5Ј end of the LCB3 gene, into pRS305 cut with the same restriction endonucleases. The resulting plasmid was cut with PstI and MscI and ligated with a 1-kb NsiI-MscI DNA fragment, containing the 3Ј-coding and flanking region of LCB3, obtained from pJAB15-1. Digestion of MPRS305 with MscI gives a linear DNA fragment with ends homologous to the 5Ј-and 3Ј-noncoding region of LCB3. Yeast strains carrying the lcb3-⌬1 allele were made by transformation with this linear molecule followed by selection for Leu ϩ cells. Replacement of LCB3 by the deletion allele was verified by Southern blotting.
A wild type allele of LCB3 was tagged at the HpaI site, located about 0.8 kb upstream of the LCB3 coding region, with the S. cerevisiae LEU2-selectable marker gene. The tagged allele was transformed into Leu Ϫ yeast cells as a 3.6-kb MscI DNA fragment with selection for Leu ϩ cells.
Construction of pLCB1-5 began by inserting the S. cerevisiae ADE2 gene as a PstI-SpeI DNA fragment into pRS315 (15) cut with the same restriction endonucleases. The resulting plasmid was cut with NaeI and SacI and ligated to the LCB1 gene obtained from pTZ18-LCB1 (2) as an NruI-SacI DNA fragment. A 2.3-kb DNA fragment containing the YKR053C open reading frame was made from genomic DNA by using the polymerase chain reaction and cloned into pRS315 as an EcoRI-PstI fragment to give pJAB2.

Isolation of Strain AG84-3
Strain 7R6 was grown overnight in PYED (1% yeast extract (Difco), 2% Bacto-Peptone (Difco), 2% glucose, 50 mM sodium succinate (pH 5.0), inositol (50 mg/liter), and potassium phosphate monobasic (0.5 g/liter)) plus 25 M PHS (referred to here as the medium) and then mutagenized with ethylmethanesulfonate to give 20% killing (18). Mutagenized cells were diluted to an absorbance at 600 nm (A 600 ) of 0.4 with medium, incubated with shaking at 30°C for 7 h, during which time the A 600 increased to 2.5, centrifuged, resuspended in 2 ml of medium, and sonicated using a microtip (Heat Systems-Ultrasonic) for 2 min to disrupt clumped cells. One ml of cells was layered on 4 ml of 30% sodium diatriazoate (18) and centrifuged at 10°C in a Sorvall RT6000B centrifuge for 4 min at 2,000 rpm. Most cells were at the interface, but a faint pellet of dense cells was present at the bottom of the tube. The liquid was aspirated, and the cell pellet was carefully resuspended in 0.5 ml of medium to avoid mixing with cells stuck to the side of the tube. Resuspended pellets from two tubes were mixed and recentrifuged on 30% sodium diatriazoate. The cell pellet was resuspended in medium, and about 500 cells were spread on PYED plates containing 25 M PHS. Two days later only about 15 colonies/plate were visible, suggesting that only 3% of the dense cells were viable.

Measurement of Sphingolipid Synthesis in Vivo
Protocol 1-Sphingolipid synthesis was measured by growing cells to saturation at 30°C in PYED, diluting to an A 600 of 0.1 with fresh PYED medium containing 10 g/ml of inositol and 10 Ci/ml myo-[2-3 H]inositol (20 Ci/mmol, DuPont NEN), and grown at 30°C overnight to an A 600 of 1. Cells were washed with 25 ml of 0.1 M sodium phosphate, pH 5.5, by centrifugation and resuspended in the same buffer at 50 A 600 units/ ml. A reaction containing 0.4 ml of 0.5% tergitol, 0.4 ml of 0.5 M sodium succinate, pH 5.5, 0.1 ml of 40% glucose, and 10 l of 2.5 mM PHS in 95% ethanol was initiated by addition of 0.2 ml of the radiolabeled cells. At time 0 a 0.2-ml sample and at the indicated times a 0.5-ml sample of radiolabeled cells were treated with trichloroacetic acid at a final con- centration of 5% for 20 min at 0°C. Cells were centrifuged, washed three times with 1 ml of 5% tricholoracetic acid and finally with water. Cell pellets were extracted with 1 ml of 95% ethanol, water, diethyl ether, pyridine, concentrated NH 4 OH (15:15:5:1:0.018, v/v) for 60 min at 60°C (19). After centrifuging while warm, a 0.5-ml sample of the supernatant fluid was evaporated under a stream of N 2 , dissolved in 1 ml of monomethylamine reagent, and heated for 60 min at 52°C to deacylate the acylester phospholipids (20). The sample was dried and dissolved in 0.5 ml of chloroform:methanol:water (16:16:5). Qualitative analysis of 25-l samples was carried out by silica gel thin layer chromatography on 20-cm Whatman LK5 plates developed with chloroform, methanol, 4.2 N NH 4 OH (9:7:2). Radioactivity was measured by using a BioScan apparatus. Alternatively, quantification of radioactive products was achieved by ascending chromatography of 5-l samples on Whatman SG-81 paper washed with EDTA (21) and developed with chloroform, methanol, 4.2 N NH 4 OH (9:7:2). The lanes were cut into 1-cm zones and counted in a liquid scintillation spectrometer.
Protocol 2-Cells were pregrown in nonradioactive minimal medium lacking leucine as described in protocol 1, resuspended at an A 600 of 0.5/ml in fresh medium containing 10 g/ml inositol and 20 Ci/ml myo-[2-3 H]inositol, and aerated at 30°C for the indicated times. Sphingolipids were processed and quantified after chromatography on Whatman SG-81 paper as described in protocol 1.
Protocol 3-Cells were grown to an A 600 of 0.8 in PYED medium at 30°C, 2.5 A 600 units of cells were centrifuged in a 50-ml plastic centrifuge tube, resuspended in 4 ml of medium containing 1% peptone (Difco), 1% yeast extract (Difco), 0.05% tergitol, 0.01 mg/ml inositol, 0.1% KH 2 PO 4 , 2% glucose, 50 mM sodium succinate, pH 5.5, 0.91 M , and shaken at 30°C. The resuspension medium was prepared by drying the sphinganine in the plastic tube under a stream of N 2 , adding the other ingredients, and sonicating in a water bath for 2 min to disperse the sphinganine uniformly. At 2 and 4 h, 0.25 ml of 100% trichloroacetic acid was added per tube to stop the reaction. The samples were transferred to small glass tubes containing 50 A 600 units of nonradioactive carrier cells, centrifuged at 3,000 ϫ g for 5 min, and washed once with 3 ml of 5% trichloroacetic acid and three times with 3 ml of water. Lipids were extracted with ethanol:water:diethyl ether:pyridine:concentrated NH 4 OH, processed, and dissolved in 0.5 ml of chloroform: methanol:water as described in protocol 1. Samples were chromatographed on a 0.5-ml bed of AG-X4 (acetate form, Bio-Rad) packed in a Pasteur pipette. The column was packed in water, equilibrated in methanol, and eluted with 3 ml of chloroform:methanol:water (16:16:5) followed by dilution of sphingolipids with 4 ml of the same solvent containing 0.4 N NH 4 OH. The latter eluate was dried and dissolved in 0.5 ml of chloroform:methanol:water (16:16:5), and 10-l portions were chromatographed on SG-81 paper and analyzed as described for protocol 1.

Determination of Minimum Inhibitory Concentration
Cells were grown overnight in minimal medium at 30°C and diluted with fresh medium so that following addition of an inhibitory compound the final concentration was 1-2 ϫ 10 Ϫ4 /ml, and the culture volume was 200 l/microtiter well. Inhibitory compounds were tested in 2-fold dilution series in covered microtiter plates incubated without shaking at 30°C for 40 h. The minimum inhibitory concentration is the lowest concentration of inhibitor yielding no visible cell pellet.

Uptake of Long Chain Bases
Cells were grown to an A 600 of 0.6 -0.8 at 30°C in minimal medium, centrifuged, resuspended in medium lacking tergitol at an A 600 of 5 units/ml, and diluted into 7.2 ml of minimal medium containing 0.056% tergitol and 11.11 M [ 3 H]sphinganine (4,000 cpm/nmol) in a 50-ml plastic tube that was incubated with shaking at 30°C. At various times duplicate 0.5-ml samples were diluted into 5 ml of ice-cold dilution buffer (0.02 M cyclodextrin (U. S. Biochemical Corp.), 0.05% tergitol, 10 M sphinganine), filtered (0.4-m pore size, Nucleopore Corp., Pleasanton, CA) to separate cells from the free radioisotope, and washed on the filter with 3 ϫ 5 ml of ice-cold dilution buffer. Radioactivity on the filter was measured by liquid scintillation spectrometry.

RESULTS
Isolation of Strain AG84-3-Strain AG84-3 was isolated by using the procedure described under "Experimental Procedures" to enrich for strains defective in sphingolipid synthesis. Briefly, when SLC strains like 7R6 are fed PHS they make sphingolipids and have a normal buoyant density. If, however, a mutation has blocked sphingolipid synthesis the density of the cell should increase because sphingolipids will not be made in the presence of PHS. Using this strategy, SLC cells having a higher density following growth in the presence of PHS were enriched, and putative mutants were screened to differentiate those specifically defective in sphingolipid synthesis from those defective in other lipid biosynthetic pathways which might also affect cell density. The screen was based upon the observation that strain 7R6 cannot grow at low pH when it lacks sphingolipids (no PHS present in the medium) but can grow when allowed to make sphingolipids (PHS present in the medium) (22). Thus, about 100 mutants unable to grow on PYED plates at pH 4.1 either in the presence or absence of 25 M PHS were identified.
Mutants specifically defective in sphingolipid synthesis rather than glycerophospholipid synthesis were identified by looking for decreased incorporation of [ 3 H]inositol into inositolcontaining sphingolipids relative to phosphatidylinositol (PI) (23). By this assay strain AG84 appeared to be specifically defective in sphingolipid synthesis.
It became necessary during the course of our experiments to transform AG84 with plasmid DNA. However, the strain transformed poorly, less than 10 transformants/g of plasmid DNA. To obtain a derivative of AG84 with enhanced transformability we selected derivatives that could grow on PYED plates at 37°C. The rationale for this approach was to isolate a more stress-tolerant strain that would resist killing during the transformation procedure. Several 37°C-resistant derivatives were tested for transformability, and one, AG84-3, was used for all studies presented here. Strain AG84-3 Cannot Make Sphingolipids When Fed PHS-Synthesis of inositol-containing sphingolipids in AG84-3 cells was measured using an in vivo radiolabeling procedure. Cells were cultured overnight, without PHS to prevent sphingolipid synthesis, but with [ 3 H]inositol to label PI, a substrate for synthesis of IPC and M(IP) 2 C. Cells were separated from unincorporated radioisotope and incubated with or without PHS in a buffered solution containing glucose. The parental strain 7R6, but not the presumptive mutant strain AG84-3, should exhibit a PHS-dependent transfer of the radiolabel from PI to ceramide to yield [ 3 H]IPC and to [ 3 H]MIPC to yield [ 3 H]M(IP) 2 C. Parental strain 7R6 behaved as expected because the amount of 3 H-labeled sphingolipids synthesized was linearly dependent upon the concentration of PHS in the reaction (Fig. 2). In contrast, AG84-3 cells made no detectable sphingolipids during a 1-h (Fig. 2) or a 3-h (data not shown) incubation in the presence of 2.5 M PHS. At 25 M PHS, AG84-3 cells made sphingolipids, but at a lower level than 7R6 cells at both the 1-h (Fig. 2) and the 3-h time point (data not shown). Both strains showed deacylation products near the origin derived from deacylation of PI, indicating that synthesis of PI was normal (data not shown). From these data we conclude that AG84-3 cells have a defect in sphingolipid synthesis, specifically, they are less effective in transferring inositol-P from PI to ceramide to form IPC and to MIPC to form M(IP) 2 C (see Fig. 1). This behavior could be due to a defect in uptake of PHS from the culture medium, in synthesis of ceramide, or in transfer of inositol-P from PI to ceramide and MIPC.
To differentiate between a defect in PHS uptake and a defect in the biosynthetic pathway, a wild type copy of LCB1, encoding a subunit of serine palmitoyltransferase (Fig. 1), was introduced into strain AG84-3 (deleted for lcb1) on a plasmid, and sphingolipid synthesis was measured. If AG84-3 is defective in the transport of PHS then the LCB1 gene should restore sphingolipid synthesis because the strain will be able to make endogenous PHS, whereas the gene will not restore sphingolipid synthesis if the strain is defective in a later step in the biosynthetic pathway. Cells were cultured with [ 3 H]inositol but not with PHS, and the amount of radiolabeled sphingolipids was measured as a function of time. AG84-3 cells (lcb1) carrying the LCB1 gene on a plasmid were able to synthesize sphingolipids as well as the parental strain 7R6 (LCB1) (Fig. 3). We conclude from these data that strain AG84-3 has a defect either in the uptake of PHS from the culture medium or in the intracellular transport of PHS to the site of ceramide synthesis in the endoplasmic reticulum.

Isolation of a Gene That Restores PHS Incorporation into Sphingolipids in AG84-3 cells-A gene enabling strain AG84-3
to grow under restrictive conditions, 37°C in the presence of 25 M PHS, was isolated from a genomic library and designated LCB3. The rate of PHS incorporation into sphingolipids was measured in AG84-3 cells transformed with a centromeric vector carrying the cloned LCB3 gene to determine if the gene restored the incorporation rate to normal. Two different isolates of AG84-3 transformed with the LCB3 gene behaved identically and showed a rate of sphingolipid synthesis similar to strain 7R6, the parent of AG84-3 (Fig. 2). We conclude from these data that the LCB3 gene complements the defect in sphingolipid synthesis in AG84-3 cells and restores the rate of PHS incorporation into sphingolipids to the wild type level seen in 7R6 cells.
LCB3 is predicted to encode a protein of 409 amino acids which lacks similarity to any known protein. It does show 53% amino acid identity (Fig. 4) to another S. cerevisiae open reading frame, YKR053C, whose predicted amino acid sequence also shows no similarity to any known protein or sequence motif.
By one method (24) there are eight predicted transmembrane helices in the Lcb3 protein, and four of these (indicated in Fig.  4) are also predicted to be membrane-associated by two other methods (25,26). Based on these predictions, the Lcb3 protein is most likely a membrane-bound protein having at least four and perhaps as many as eight transmembrane helices. Both the Lcb3 protein and the protein predicted to be encoded by open reading frame YKR053C have a putative glycosylphosphatidylinositiol cleavage/attachment site at their COOH terminus (Fig. 4).
At this point we wanted to verify that mutation of the lcb3 gene and not some other gene, whose defect could be compensated indirectly by the LCB3 gene, was responsible for the sphingolipid synthesis defect in strain AG84-3. Segregation of the putative lcb3 mutant gene by tetrad analysis was not possible because strain AG84-3 does not sporulate when mated to another strain mutated in lcb1 and carrying the SLC1-1 suppressor gene (a strain like parental strain 7R6 but of the opposite mating type). Instead of tetrad analysis, two other approaches were used to verify that the reduced rate of PHS incorporation into sphingolipids is due to inactivation of the lcb3 gene in AG84-3 cells.
First, if the cloned LCB3 gene is allelic to the mutated gene blocking incorporation of PHS into sphingolipids in strain AG84-3, then deletion of the lcb3 locus in strain AG84-3 should not relieve the block. Three independent derivatives of strain AG84-3 carrying the lcb3-⌬1 allele were constructed and found to behave the same. Only data for one strain (RCD104) will be presented. Like strain AG84-3, it made only very small amounts of sphingolipids when fed [ 3 H]sphinganine (Fig. 5, top  panel), indicating that the inability of strain AG84-3 to make sphingolipids at a normal rate was most likely due to mutation of lcb3 and not some unknown gene. Furthermore, since strains carrying the lcb3-⌬1 deletion allele behave like strain AG84-3, it would appear that the mutation in the AG84-3 lcb3 gene inactivates the gene rather than producing, for example, a dominant negative allele whose protein product binds to and inactivates another protein necessary for incorporation of sphinganine into sphingolipids.
In a second experimental approach, a wild type LCB3 gene, tagged with the LEU2 selectable marker gene, was used to replace the lcb3 locus in strain AG84-3. The resulting strains (AGW-1 through AGW-6) were tested for growth at 37°C on minimal agar plates lacking uracil but containing 25 M PHS. Unlike strain AG84-3, which cannot grow under these conditions, all of the AGW strains grew (data not shown), suggesting that they could make sphingolipids. Direct measurement showed that incorporation of exogenous [ 3 H]sphinganine into sphingolipids was restored in strain AGW-2 (Fig. 5, top panel). We conclude from these experiments that an inactivating mutation in the lcb3 gene prevents exogenous long chain bases from being incorporated into sphingolipids at a normal rate in AG84-3 cells.
Phenotypes of Wild Type Cells Carrying an lcb3 Deletion Allele-All of the experiments described so far were performed in cells deleted for lcb1 and carrying the SLC1-1 suppressor gene (strains 7R6, AG84-3, and their relatives). These strains grow slowly and die even under optimal conditions, and they could have accidentally acquired an unknown mutation that enhanced growth and played an unsuspected role in sphingolipid synthesis. To avoid such potential complications, the consequences of the lcb3 mutation were examined in a wild type background containing no known mutations in lipid metabolic genes. Diploid strain YPH274 was transformed with a DNA fragment carrying the lcb3-⌬1 deletion allele, Leu ϩ transformants were selected, and their DNA analyzed by Southern blotting to verify that one wild type LCB3 allele had been replaced by the lcb3 deletion allele (data not shown).
The LCB3 gene was found to be nonessential because heterozygous diploids when sporulated give tetrads with four viable spores, two of which were Leu ϩ , indicating the presence of the lcb3-⌬1 deletion allele, and two were Leu Ϫ , indicating the presence of the LCB3 allele. Haploid strains derived from the same tetrad and carrying the lcb3-⌬1 deletion allele (strain FIG. 4. Protein sequence alignment. The Lcb3 protein and the protein predicted to be encoded by open reading frame YKR053C were aligned (29) with identical amino acids indicated by a vertical bar and chemically similar amino acids by a dot. Predicted membrane-spanning domains (25,26) are underlined, and predicted glycosylphosphatidylinositiol cleavage/attachment sites (30) are indicated in bold.
FIG. 5. The LCB3 gene restores sphingolipid synthesis in strain AG84-3. Sphingolipid synthesis was measured by following the incorporation of exogenous [ 3 H]sphinganine (0.91 M) into sphingolipids for 4 h as described under "Experimental Procedures" using protocol 3. Sphingolipids were analyzed qualitatively by thin layer chromatography on Whatman LK5 plates with radioactivity localized using a BioScan apparatus as described in protocol 1. Brackets indicate the y axis scale in cpm. Peaks A, B, and C are M(IP) 2 C species 3, 2 and 1, respectively (4); peak D is an unknown; peak E is MIPC-3; peaks F, G, and H are IPC-3, 2 and 1, respectively; O is the origin. RCD101) or the wild type LCB3 allele (strain RCD100) grew in PYED medium with the same doubling time at 30°C, at pH 4.1, and in the presence of 0.75 M NaCl, indicating no effect of the lcb3 mutation by itself on the stress responses known to be defective in 7R6 cells when they lack sphingolipids (22).
If the Lcb3 protein transports long chain base into cells we would predict that the lcb3 deletion strain RCD101 should have a decreased rate of long chain base uptake. This prediction was examined by measuring incorporation of [ 3 H]sphinganine into sphingolipids containing inositol-P. At a low concentration of sphinganine (0.91 M) the lcb3-⌬1 mutant strain RCD101 made only low levels of sphingolipids compared with wild type RCD101 cells (Fig. 5, lower panel).
To quantify more carefully the effect of the lcb3-⌬1 deletion mutation, the time course of [ 3 H]sphinganine incorporation into sphingolipids was measured. At a low concentration of sphinganine (0.65 M), the lcb3 mutant strain showed almost no incorporation of radiolabel into sphingolipids over a 4-h time period, whereas the wild type strain showed a time-dependent increase in incorporation (Fig. 6). At a higher concentration of sphinganine (20.6 M), the lcb3-⌬1 mutant strain showed a low level of incorporation which increased with time, but the level was less than half the wild type level (Fig. 6). We conclude from these data that the Lcb3 protein is necessary for a normal rate of incorporation of exogenous long chain base into sphingolipids.
If the Lcb3 protein transports sphingoid long chain bases into the cell, we would expect strains lacking the protein to be more resistant to growth inhibition by long chain bases and related hydrophobic compounds. As predicted, the lcb3-⌬1 deletion strain, RCD101, was more resistant to growth inhibition by several types of long chain bases, the water-soluble ceramide analog N-acetylsphingosine, and stearylamine than was the LCB3 strain, RCD100 (Table II). In addition, the deletion strain was more resistant to the toxic arginine analog canavanine and to the antibiotic hygromycin B. Both strains were toxified to the same degree by the serine palmitoyltransferase inhibitor (Fig. 1) sphingofungin B (23) and by the antibiotic cerulenin (Table II).
Because the Lcb3 protein is predicted to be homologous to the putative protein encoded by the YKR053C open reading frame, we determined if the two proteins were functionally related. A strain, RCD102, deleted for the YKR053C DNA sequence in a wild type genetic background, had the same level of resistance to long chain bases and other inhibitors of growth as the nondeleted strain, RCD100 (Table II). Likewise, the deleted strain had the same ability to incorporate exogenous [ 3 H]sphinganine into sphingolipids as the nondeleted strain (Fig. 6).
Possible synergistic interaction between the Lcb3 and the YKR053C proteins was examined using a strain (RCD103) deleted for both coding regions. The double deletion strain had the same level of resistance to long chain bases and related compounds as the strain deleted only for lcb3 (compare strains RCD101 and RCD103, Table II) and the same rate of incorporation of exogenous [ 3 H]sphinganine into sphingolipids as the lcb3-deleted strain (Fig. 6). Based upon these data the protein encoded by YKR053C is not functionally related to the Lcb3 protein, at least not for the two phenotypes examined.
Based upon the reduced rate of [ 3 H]sphinganine incorporation into the sphingolipids of lcb3-defective cells over a 4-h period compared with wild type cells (Fig. 6), we anticipated that the initial rate of [ 3 H]sphinganine uptake would also be reduced in the lcb3-defective cells compared with wild type. Various conditions for measuring the initial rate of uptake were examined, but none proved satisfactory. For example, uptake during the first 3 min in wild type LCB3 (RCD100) and lcb3-deleted cells (RCD101) was masked by what appeared to be rapid binding or adsorption of [ 3 H]sphinganine to the cell surface, since wild type cells heated to 65°C for 15 min bound at least half as many counts as did unheated cells (data not shown). In addition, no reproducible difference in uptake between nonheated wild type and lcb3 mutant cells could be detected.

DISCUSSION
Based upon the available data we conclude that the LCB3 gene encodes a membrane-bound protein necessary for utilization of exogenous long chain bases. The Lcb3 protein is predicted to be localized in the plasma membrane where we hypothesize that it transports or facilitates uptake of long chain bases. Alternatively, it could be bound to an internal membrane where it would promote delivery of long chain bases to the site of cermide synthesis in the endoplasmic reticulum. LCB3 is the first gene in any organism to be implicated in long chain base transport.
Our conclusions and hypotheses are based upon an analysis of strain AG84-3, which is unable to make sphingolipids when fed PHS (Fig. 2). Failure to use exogenous PHS is not due to a block in one of the later steps in the sphingolipid biosynthetic pathway (Fig. 1) because the pathway is functional; complementation of the lcb1 mutation in AG84-3 cells with the LCB1 gene restores sphingolipid synthesis (Fig. 3). In addition, there was no difference in ceramide synthase (Fig. 1) activity between wild type RCD100 cells and cells deleted for either the lcb3 gene or the ykr053c open reading frame (data not shown).
We identified the mutant gene in AG84-3 cells, termed LCB3, by complementing the growth defect with a plasmid selected from a genomic DNA library. The Lcb3 protein is predicted to have four transmembrane domains by two algorithms (Fig. 4) and up to eight transmembrane domains by another algorithm (24) and is, therefore, most likely a membrane-bound protein. Another algorithm predicts that the protein is probably located in the plasma membrane (27). These predictions are consistent with the idea that AG84-3 cells are defective in a protein necessary for uptake of PHS from the culture medium.
The role of the LCB3 gene in PHS utilization was examined in a wild type, non-SLC genetic background to avoid the possibility that an unidentified mutation in strain AG84-3 was reducing PHS incorporation into sphingolipids. Deletion of the lcb3 gene in a wild type genetic background reduced the rate of [ 3 H]sphinganine incorporation into sphingolipids so that almost no incorporation was observed at a low substrate concentration (0.65 M, Fig. 6) whereas some incorporation occurred at a higher concentration (20.6 M, Fig. 6). These results support the hypothesis that the Lcb3 protein transports or facilitates uptake of long chain bases into cells. The finding that the lcb3-deleted strain RCD101 is more resistant than the nondeleted strain RCD100 to growth inhibition by long chain bases (Table II) further supports this hypothesis.
Attempts to measure directly the initial rate of long chain base uptake were not successful for at least two reasons. First, in the wild type genetic background even heat-killed cells contained a large amount of radioactive long chain base (sphinganine) after a 3-min incubation (data not shown), suggesting rapid binding to the cell surface, perhaps to the cell wall. The pmol of sphinganine bound/A 600 of heat-killed cells by 3 min approaches the total long chain base content of cellular sphingolipids (28) and thus masks initial PHS uptake. Second, in a SLC genetic background the rapid putative surface binding was absent both in cells containing or lacking sphingolipids. Instead, there was a gradual increase over at least a 30-min period in the pmol of sphinganine bound/cell, both in heatkilled and nonkilled cells (data not shown). Because of these problems it was not possible to measure the initial rate of long chain base uptake.
Since wild type S. cerevisiae cells can make sphingoid long chain bases there would appear to be no need for the Lcb3 protein. However, it may normally perform an unidentified physiological function that is not necessary under the growth conditions we examined. Alternatively, it may transport other compounds necessary for growth or survival under special, nonlaboratory circumstances. Cells lacking the lcb3 gene were more resistant to growth inhibition by the water-soluble, uncharged ceramide, N-acetylsphingosine, by the antibiotic hygromycin B, and by the arginine analog canavanine (Table II), indicating that the Lcb3 protein can recognize a wide range of compounds differing in charge, size, and hydrophobicity. Cells lacking the lcb3 gene were not more resistant to the serine palmitoyltransferase inhibitor sphingofungin B, an analog of sphingoid long chain bases (23), nor to the antibiotic cerulenin (Table II), suggesting that the Lcb3 protein does not transport these compounds into cells.
A search of the NCBI data bases revealed that the predicted Lcb3 protein was homologous to the protein predicted to be encoded by the S. cerevisiae open reading frame YKR053C but not to any other predicted proteins. The amino acid identity and similarity between these two proteins extends throughout their sequence, and both are predicted to have several membrane-spanning domains and a glycosylphosphatidylinositiol cleavage/attachment site at their COOH terminus (Fig. 4). Despite their similarity in primary structure, they appear not to be functionally similar because a ykr053c deletion mutant did not show reduced incorporation of [ 3 H]sphinganine into sphingolipids like the lcb3 deletion mutant (Fig. 6), nor did the ykr053c deletion increase resistance to growth inhibition as did the lcb3 deletion (Table II). Based upon its primary sequence similarity to the Lcb3 protein, the protein encoded by YKR053C is likely to be a plasma membrane-bound protein. Analysis of a mutant strain deleted for both the lcb3 gene and the YKR053C open reading frame indicates that there is no synergy between the two genes because the double mutant and the lcb3 single mutant show the same reduction in incorporation of [ 3 H]sphinganine into sphingolipids (Fig. 6) and the same level of resistance to growth inhibition by long chain bases and other compounds (Table II).
In summary, a gene, LCB3, has been shown to be necessary for a normal rate of incorporation of exogenous long chain bases into sphingolipids. The Lcb3 protein is predicted to be located in the plasma membrane where it most likely transports long chain bases and other compounds into the cell. However, we cannot exclude the possibility of the Lcb3 protein being bound to an internal membrane such as the endoplasmic reticulum where it delivers long chain bases to the site of ceramide synthesis. The Lcb3 protein appears to represent a new class of membrane-bound protein because it shows no homology to any protein with a known function.