Vectorial acylation in Saccharomyces cerevisiae. Fat1p and fatty acyl-CoA synthetase are interacting components of a fatty acid import complex.

In Saccharomyces cerevisiae Fat1p and fatty acyl-CoA synthetase (FACS) are hypothesized to couple import and activation of exogenous fatty acids by a process called vectorial acylation. Molecular genetic and biochemical studies were used to define further the functional and physical interactions between these proteins. Multicopy extragenic suppressors were selected in strains carrying deletions in FAA1 and FAA4 or FAA1 and FAT1. Each strain is unable to grow under synthetic lethal conditions when exogenous long-chain fatty acids are required, and neither strain accumulates the fluorescent long-chain fatty acid C(1)-BODIPY-C(12) indicating a fatty acid transport defect. By using these phenotypes as selective screens, plasmids were identified encoding FAA1, FAT1, and FAA4 in the faa1Delta faa4Delta strain and encoding FAA1 and FAT1 in the faa1Delta fat1Delta strain. Multicopy FAA4 could not suppress the growth defect in the faa1Delta fat1Delta strain indicating some essential functions of Fat1p cannot be performed by Faa4p. Chromosomally encoded FAA1 and FAT1 are not able to suppress the growth deficiencies of the fat1Delta faa1Delta and faa1Delta faa4Delta strains, respectively, indicating Faa1p and Fat1p play distinct roles in the fatty acid import process. When expressed from a 2-mu plasmid, Fat1p contributes significant oleoyl-CoA synthetase activity, which indicates vectorial esterification and metabolic trapping are the driving forces behind import. Evidence of a physical interaction between Fat1p and FACS was provided using three independent biochemical approaches. First, a C-terminal peptide of Fat1p deficient in fatty acid transport exerted a dominant negative effect against long-chain acyl-CoA synthetase activity. Second, protein fusions employing Faa1p as bait and portions of Fat1p as trap were active when tested using the yeast two-hybrid system. Third, co-expressed, differentially tagged Fat1p and Faa1p or Faa4p were co-immunoprecipitated. Collectively, these data support the hypothesis that fatty acid import by vectorial acylation in yeast requires a multiprotein complex, which consists of Fat1p and Faa1p or Faa4p.

Biological membranes are complex in both their protein and lipid compositions. This complexity is essential and contributes to the barrier function of the membrane and to selectively regulated transport of molecules into and out of the cell. Unlike hydrophilic molecules such as sugars and amino acids, hydrophobic fatty acids are able to dissolve in the membrane, and as a consequence, the processes governing their regulated movement across membranes are likely to be quite distinct. Recent investigations into the problem of fatty acid transport have intensified due to findings that exogenous fatty acids influence a number of important cellular functions, including signal transduction and transcriptional control. To date, several distinct membrane-bound and membrane-associated proteins have been identified as components of fatty acid import systems in eukaryotic cells. Most notable among these are fatty acid translocase (FAT, 1 the murine homologue to CD36) (1, 2), fatty acid transport protein (FATP) (3), and fatty acyl-CoA synthetase (3)(4)(5)(6). FAT was identified following protein modification using sulfo-N-succinimidyl oleate (7), whereas FATP and fatty acyl-CoA synthetase were both identified using expression cloning (3). Both FAT and FATP have been claimed to be fatty acid transport proteins (1,8,9). Despite these claims, there is controversy surrounding the classification of FAT/ CD36 and FATP as bona fide integral membrane-bound fatty acid transporters (10). Indeed, there are gnawing questions as to whether these proteins actually function as components of a fatty acid delivery system (i.e. FAT/DC36) or as components of a utilization driven fatty acid import system (i.e. FATP), which also includes fatty acyl-CoA synthetase (2,4,8,10,11). In this regard, proteins identified as required for fatty acid transport may function not as transport proteins per se but in an alternative manner, perhaps by promoting selectivity and specificity of fatty acid delivery to downstream metabolic events.
The best characterized fatty acid transport system is that found in Escherichia coli (4). In this case, the specific integral outer membrane protein, FadL, is required for long-chain fatty acid binding and transport across that membrane. The fatty acid ligands must then traverse the bacterial periplasmic space and the inner membrane. No inner membrane proteins have been identified that are required for this process. On the basis of studies defining the energetics of fatty acid transport, we suggested protonated fatty acids flip across the inner membrane and are subsequently abstracted from the inner membrane concomitant with activation by fatty acyl-CoA synthetase (12). In this manner, exogenous fatty acids are metabolically trapped as CoA thioesters upon transport, which in turn generates a concentration gradient further driving the system. Overath and colleagues (5) coined the term "vectorial acylation" to describe this process at the time they identified the structural gene for the E. coli fatty acyl-CoA synthetase (fadD). This postulate was initially expanded by Frerman and Bennett (6) and subsequently by our laboratory (4,12) as the underlying mechanism driving long-chain fatty acid transport in bacteria. Although at the time the model of vectorial acylation was proposed the bacterial fatty acid transporter FadL had not been identified, our subsequent studies have clearly shown that both FadL and fatty acyl-CoA synthetase are required for fatty acid transport in E. coli.
By using the yeast Saccharomyces cerevisiae as a model eukaryotic system, we have recently shown the fatty acyl-CoA synthetases Faa1p or Faa4p function in the fatty acid transport system presumably by activating exogenous fatty acids concomitant with transport (11). This finding presents somewhat of a conundrum as we have also shown that long-chain fatty acid import in yeast requires Fat1p, the yeast orthologue of the murine FATP1 (13). One of the central questions we are now faced with is to determine the mechanisms by which Fat1p and fatty acyl-CoA synthetase (Faa1p and/or Faa4p) work in concert to promote fatty acid import. A similar situation appears to be operational in murine adipocytes, where there are data supporting a functional association of mmFATP1 with fatty acyl-CoA synthetase (3,15). We suggest vectorial acylation is one general mechanism of fatty acid import, which functions to promote the regulated import and metabolic trapping of exogenous long-chain fatty acids.
In our prior investigations into fatty acid import in yeast, we used reverse genetic approaches to demonstrate this process requires the yeast orthologue of mmFATP (Fat1p) and fatty acyl-CoA synthetase (Faa1p or Faa4p) (11,13,14). Despite the information gleaned from these studies, there are no data demonstrating these proteins function cooperatively in a physical complex, and there is no information as to whether there are additional proteins involved in mediating the regulated import of exogenous long-chain fatty acids. In the present work, we sought to identify additional components required for fatty acid transport and to confirm the importance of Fat1p and fatty acyl-CoA synthetase (Faa1p and Faa4p) by using a genetic approach. A valuable molecular-genetic method for the identification of participants in multicomponent cellular processes is the selection of plasmid-encoded multicopy extragenic suppressors (16). The rationale behind this approach is that the altered phenotype resulting from a deficiency in one participant can be suppressed by overexpression of another participant required for the same process (16). In this manner, we sought to identify plasmid-encoded multicopy extragenic suppressors of the deficiency in fatty acid import caused by deletion of FAT1 and/or FAA1 and FAA4. We report that plasmids encoding Fat1p, Faa1p, and Faa4p were identified in a screen for multicopy extragenic suppressors of the transport and activation deficiency of a faa1⌬ faa4⌬ strain, and plasmids encoding only Fat1p and Faa1p were identified as multicopy extragenic sup-pressors of the transport deficiency of a faa1⌬ fat1⌬ strain. Additional biochemical evidence is provided demonstrating Fat1p and acyl-CoA synthetase interact in a physical complex. This work establishes for the first time a genetic, physical, and functional linkage between Fat1p and fatty acyl-CoA synthetase and substantiates the hypothesis that these proteins, perhaps exclusively, are required for long-chain fatty acid transport in yeast.

EXPERIMENTAL PROCEDURES
Strains, Media, and Materials-The S. cerevisiae strains used in this study are listed in Table I. The fat1⌬::G418 mutation was introduced by transformation of the strain of interest with linear DNA generated by amplification of the kanamycin resistance cassette (resulting in G418 resistance) using oligonucleotides complementary to both FAT1 and the cassette as described (17). The oligonucleotide for the coding strand was 5Ј-CACTGTCAAGAAGGGCAAGAAGGCAGCAGTATGGCTTGGGCA-TAGGCCACTAGTGGATCTG-3Ј, and the oligonucleotide for the template strand was 5Ј-CCACTGGATCATTCGTAAGTGATCCTGAAA-CAAACCATTCAGCAGCTGAAGCTTCGTACGC-3Ј. Chromosomal replacement of the native gene was confirmed by Southern analysis of chromosomal DNA from the transformants by comparison to DNA obtained from the parental strain. Yeast strains were transformed by the lithium acetate method (18).
Yeast extract, yeast peptone, and yeast nitrogen base were obtained from Difco. Oleic acid was obtained from Sigma. 3 H-or 14 C-labeled fatty acids were from PerkinElmer Life Sciences and American Radiochemicals. C 1 -BODIPY-C 12 was purchased from Molecular Probes. Enzymes required for all DNA manipulations were from Promega, Invitrogen, New England Biolabs, U. S. Biochemical Corp., or Roche Molecular Biochemicals. Anti-V5 antibody and anti-T7 antibodies were purchased from Invitrogen and Novagen, respectively. Anti-Pma1p was the gift of Dr. Gü nther Daum (Technische Universitä t Graz, Graz, Austria).
Complementation of faa1⌬ faa4⌬ and faa1⌬ fat1⌬ Using Multicopy Extragenic Suppression-Cells of the faa1⌬ faa4⌬ strain or faa1⌬ fat1⌬ strain were rendered competent using lithium acetate as noted above, transformed with a yeast multicopy library in YEp24, and transformants selected on YNBD containing the appropriate supplements but lacking uracil (19). Thirty thousand individual Ura ϩ transformants were selected from the library and were screened for growth following replica plating on YPD plates containing 45 M cerulenin and 100 M oleic acid (YPD-CER-OLE). Transformants that were able to grow on YPD-CER-OLE were colony-purified on the same media and phenotypes validated on YPD-CER-OLE. Plasmids were isolated from those that retained positive growth on all three media and retransformed into the faa1⌬ faa4⌬ and faa1⌬ fat1⌬ strains. Additionally, the same plasmids were propagated in the E. coli strain DH5␣, purified using Qia-Prep columns (Qiagen), and sequenced using two plasmid-specific primers flanking the insert (upstream, 5Ј-GGAGCCACTATCGACTACGC-3Ј; downstream, 5Ј-CCTGTGGCGCCGGTGATG-3Ј) using an Applied Biosystems automated fluorescence DNA sequencer. The sequences obtained were compared with the Saccharomyces genome data base for identification.
Assessment of Fatty Acid Import Capacity-Fatty acid import was assessed using confocal laser scanning microscopy to detect accumulation of the fluorescent long-chain fatty acid analogue 4,4-difluoro-5methyl-4-bora-3a,4a-diaza-s-indacene-3-dodecanoic acid (C 1 -BODIPY-C 12 ) as described previously (13). Following growth under selective conditions, cells were harvested, washed with phosphate-buffered saline (PBS) and resuspended in 0.1 volume of PBS. All steps were performed at room temperature. Washed cells were incubated with 10 M C 1 -BODIPY-C 12 for 60 s, washed with PBS containing 50 M fatty acid-free bovine serum albumin (two times), PBS, resuspended in PBS, and visualized on an NORAN-OZ confocal laser scanning microscopy, interfaced with a Nikon Diaphot 200 inverted microscope equipped with a PlanApo ϫ60, 1.4 NA oil-immersion objective lens. The instrument settings for brightness, contrast, laser power, and slit size were optimized for the brightest sample to ensure that the confocal laser scanning microscopy was set for its full dynamic range. The same settings were used for all subsequent image collections. Quantification of Fatty Acyl-CoA Synthetase Activities-Cells were grown from overnight cultures in YNBD (with appropriate supplements) and grown to A 600 of 1.0. Following growth, cells were harvested by centrifugation, washed twice with PBS, and resuspended to a density of 1.2 ϫ 10 9 cells/ml in 200 mM Tris-HCl, pH 8.0, 4 mM EDTA, 5 mM 2-mercaptoethanol, 10% glycerol, 0.01% Triton X-100, 0.5 mM phenylmethylsulfonyl fluoride, 4 M pepstatin A, and 8 M leupeptin. The cells were lysed by vigorously vortexing the cell suspension containing glass beads for 1 min, 5 times at 0°C. Samples were clarified by centrifugation (1,500 ϫ g, 5 min, 4°C), and supernatants were used to assess fatty acyl-CoA synthetase activities as described (20). The reaction mixtures contained 200 mM Tris-HCl, pH 7.5, 2.5 mM ATP, 8 mM MgCl 2 , 2 mM EDTA, 20 mM NaF, 0.01% Triton X-100, fatty acid dissolved in 10 mg/ml ␣-cyclodextrin (final concentrations of fatty acids were 50 M), 0.5 mM coenzyme A, and cell extract in a total volume of 0.5 ml. The reactions were initiated by the addition of coenzyme A, incubated at 30°C for 20 min, and terminated by the addition of 2.5 ml of isopropyl alcohol, n-heptane, 1 M H 2 SO 4 (40:10:1). The radioactive fatty acid was removed by organic extraction using n-heptane. Acyl-CoA formed during the reaction remained in the aqueous fraction and was quantified by scintillation counting. Protein concentrations in the cell extracts were determined using the Bradford assay and bovine serum albumin as a standard (21). The values presented represent the average from at least three independent experiments performed in duplicate. All experiments were subjected to analysis of variance (StatView, SAS Institute, Inc.).
Negative Dominance of Mutant Fat1p Over Fatty Acyl-CoA Synthetase-The sequence encoding the C-terminal 125 amino acids (residues 545-669) of Fat1p was cloned in-frame to the T7 epitope tag of the yeast expression vector YEpGALSET983 to generate YEpDB213. The resulting T7 Fat1p 125C fusion was expressed under the control of the GAL10 promoter. To test for negative dominance, YEpDB213 was transformed into YB332. Cells transformed with the vector (YEpGALSET983) served as a control. The cells were pre-grown in YNBD (without leucine) overnight. The culture was harvested by centrifugation and resuspended to a cell density of 0.1 A 600 in 50 ml of YNB containing 2% galactose and 2% raffinose to induce expression of T7 Fat1p 125C . When the density reached 1.0 A 600 , cells were harvested by centrifugation, washed once in PBS, and resuspended in 1 ml of breaking buffer (200 mM Tris, pH 8.0, 4 mM EDTA, 10% glycerol, 5 mM ␤-mercaptoethanol, 0.01% Triton X-100, 0.5 mM phenylmethylsulfonyl fluoride, 4 M pepstatin A, and 8 M leupeptin). The cells were lysed by vortexing with glass beads and assayed for long-chain acyl-CoA synthetase activity as detailed above.
Two-hybrid Analysis of Fat1p and Faa1p-The yeast two-hybrid system was used to test Faa1p-Fat1p interaction (22). The bait plasmid vector used was pEG202; the trap plasmid vector was pJG4-5, and the reporter plasmid was pSH18-34T. To generate the full-length Faa1pbait fusion protein, the coding sequence of FAA1 was amplified using the upstream primer 5Ј-AGACCCATGGATGGTTGCTCAATATACC-G-3Ј and the downstream primer 5Ј-AAATGTTGGCGGCCGCAGACG-AACTATAAACGGC-3Ј. The amplified DNA fragment was cleaved with NcoI and NotI and ligated into pEG202 cleaved with the same enzymes. For the trap plasmids, a single primer was used to amplify DNA at the 3Ј end of the gene including the termination codon encoding amino acid 669, 5Ј-GAACATCCTCGAGTAATTTAATTGTTTGTGC-3Ј, whereas unique primers were used to amplify DNA at the 5Ј ends. These included 5Ј-TTTTTAGCGCGCAATACTAAAGGCACTCCG-3Ј to generate a peptide from amino acids 169 to 669 of Fat1p (Fat1p 500C ) and 5Ј-GAA-GATGAATTCACGGCCAGTAACAAAGAAC-3Ј to generate a peptide from amino acids 544 to 669 of Fat1p (Fat1p 125C ). The amplified DNA fragments were digested with the appropriate restriction enzymes and ligated into pJG4-5.
To test interaction, the Faa1p bait plasmid and the target trap plasmid of interest were transformed into yeast strain W303B carrying the reporter plasmid pSH18-34T. The reporter plasmid pSH18-34T contains the lacZ gene encoding ␤galactosidase driven by a promoter controlled by eight LexA operators. To maintain each plasmid, transformants were selected and maintained on YNBD media lacking uracil (for pSH18-34T), histidine (for the pEG202-derived bait), and tryptophan (for the pJG4-5-derived traps). Expression of ␤-galactosidase activity was measured using the liquid assay employing o-nitrophenyl ␤-D-galactopyranoside as substrate as described previously (23). For these experiments, cells were grown overnight in YNBD (without uracil, histidine, or tryptophan) and subcultured to A 600 0.02-0.1 in 10 ml of YNB containing 2% galactose and 2% raffinose. Growth was continued until the A 600 reached 0.5-1.0, at which time was stopped by placing the cultures on ice. Aliquots of cells (1 ml) were harvested by centrifugation (14,000 rpm for 3 min). The cell pellets were resuspended in 200 l of 0.1 M Tris, pH 7.5, containing 0.05% Triton X-100. The sample was frozen on dry ice and stored at Ϫ80°C prior to assay. All experiments defining ␤-galactosidase activities were performed in duplicate at least five times as described previously (23); the data were analyzed using paired t-tests against cells containing the bait (Faa1p), the trap vector (pJG4-5), and the reporter (StatView, SAS Institute, Inc.).
Co-immunoprecipitation of Fat1p and Faa1p or Faa4p-To identify a protein complex containing Fat1p and Faa1p or Faa4p, plasmids were constructed expressing each protein fused to a peptide tag, which is recognized by a commercially available antibody. Full-length Fat1p tagged with a T7 epitope was constructed in the vector YEpGALSET983 to generate plasmid YEpDB204. The coding sequence of FAT1 was amplified using the upstream primer 5Ј-GCGGAGCTCATGTCTC-CCATACAGGTTGTTG-3Ј and the downstream primer 5Ј-CGCGGTAC-CATGCTCTAATGGAAAGGTAC-3Ј. The amplified DNA fragment was cleaved with SacI and KpnI and ligated into YEpGALSET983 cleaved with the same restriction enzymes. Expression clones encoding fulllength Faa1p or Faa4p tagged at the C terminus with a V5 epitope were obtained from Invitrogen (GeneStorm TM clones pYES2/YOR317w and pYES2/YMR246w, respectively).
The plasmid pair encoding the proteins of interest (e.g. T7 Fat1p and V5 Faa1p or V5 Faa4p) was transformed into the fat1⌬ faa1⌬ strain to test Fat1p-Faa1p interaction or the fat1⌬ faa4⌬ strain to test Fat1p-Faa4p interaction. The cells were pre-grown in YNBD without leucine and uracil to maintain both plasmids; cells were subsequently subcultured to 0.1 A 600 in 75 ml of YNB containing 2% galactose and 2% raffinose (without leucine and uracil) to induce expression of the epitope-tagged target proteins. When the cell density reached 1.0 A 600 , the cells were harvested, washed once with PBS, and resuspended in 1.5 ml of lysis buffer containing 50 mM Tris, pH 7.5, and 150 mM NaCl. The cells were lysed by vortexing with glass beads on ice as detailed above. The glass beads were pelleted by centrifugation (2,000 rpm, 2 min, 4°C). The supernatant was removed to a new tube, and Triton X-100 was added to a final concentration of 1%, and the mixture was incubated on ice for 45 min. The sample was clarified by centrifugation (4,000 rpm, 15 min, 4°C). The resultant supernatant was split into three 0.5-ml aliquots (ϳ0.7 mg/ml); 2 g of anti-T7 or 2 g of anti-V5 antibodies was added to the first two, and an equal volume of lysis buffer was added to the third as a control (protein A-Sepharose bead control). The samples were incubated with gentle rotation overnight at 4°C. Protein A-Sepharose beads (50 l of 50% slurry) were added to each sample, which were then incubated for 2 h with gentle rotation at 4°C. The protein A-Sepharose beads (containing the antigen-antibody complex) were pelleted by centrifugation (1,000 rpm, 1 min, 4°C) and subsequently washed 5 times in 50 mM Tris, pH 7.5, 150 mM NaCl, 1% Triton X-100. The final pellets containing the protein A-Sepharose beads/antigen-antibody complex were resuspended in 70 l of SDS sample buffer. Samples were boiled 5 min, and the proteins from 5 l of the cell lysate or 15 l of the immunoprecipitated sample were separated by electrophoresis on a 12.5% SDS-polyacrylamide gel. After electrophoresis the proteins were transferred to nitrocellulose for immunoblotting. Tagged proteins were detected using the appropriate antibody (anti-V5, anti-T7, or anti-Pma1p) as detailed in the figure legends.

Identification of Multicopy Suppressors of Synthetic Lethality
Imposed by Cerulenin on faa1⌬ faa4⌬ or faa1⌬ fat1⌬ Strains-When yeast cells are grown on media containing the fatty-acid synthase inhibitor cerulenin, they become auxotrophic for longchain (C 14 -C 18 ) fatty acids. Supplementation of the media with 100 M oleate is sufficient to restore growth to wild-type strains. However, cells carrying deletions in FAT1 or FAA1 and FAA4 are not viable on media containing cerulenin despite the addition of fatty acids. For fat1⌬ strains, we have shown previously (13) this phenotype is due to a defect in the ability to import fatty acids and not due to depressed levels of long-chain fatty acyl-CoA synthetase activities. Strains carrying deletions in the genes encoding the fatty acyl-CoA synthetases Faa1p and Faa4p have a similar phenotype, which we hypothesize is due to a specific coupling between Fat1p-mediated fatty acid transport and Faa1p/Faa4p-mediated fatty acid activation (11). The esterification of the exogenous fatty acid to coenzyme A is required for all subsequent metabolic processes.
In an attempt to identify genes that could functionally replace FAT1 or FAA1 and FAA4, we screened a yeast genomic multicopy library for clones, which suppressed the cerulenininduced lethality of an faa1⌬ faa4⌬ strain (deficient in longchain fatty acyl-CoA synthetase activity) and an faa1⌬ fat1⌬ strain (deficient in fatty acid import and with reduced longchain fatty acyl-CoA synthetase activity). Primary transformants were selected on YNBD plates lacking uracil and subsequently were replica-plated to YNBD plates containing cerulenin and oleate. Plasmids were isolated from colonies that grew on the selective media, and the individual plasmid encoded suppressors verified by retransformation. The identities of the inserts were determined by restriction enzyme analysis and by sequencing using plasmid-specific primers flanking the site of insertion. In both screens, multiple isolates of each plasmid-borne suppressor were identified indicating all possible suppressing clones available in this genomic library had been identified (Table II). As expected, because both strains carried a deletion in FAA1, most of the plasmids identified in either strain encoded the fatty acyl-CoA synthetase Faa1p. A surprising result was that FAT1 was identified at high frequency, whereas FAA4 was identified in only two cases in the screen using the faa1⌬ faa4⌬ strain. FAA4 was not identified as a multicopy suppressor in the faa1⌬ fat1⌬ strain. Subsequent analyses of the faa1⌬ fat1⌬ strain transformed with a YEp24 plasmid derivative encoding Faa4p (YEpDB133) verified this fatty acyl-CoA synthetase could not substitute for FAA1 and FAT1 in this strain. Three plasmids identified as multicopy suppressors using these screens were chosen for further characterization. They were YEpDB02 encoding Faa1p, YEpDB133 encoding Faa4p, and YEpDB17 encoding Fat1p (Table II; Fig. 1).
We noted that several plasmids isolated from the colonies listed in the "other" category in Table II did not confer the suppressor phenotype upon re-transformation. Therefore, we presume the phenotype was associated with an undefined chromosomally encoded suppressor.
The Multicopy Suppressors Alleviate Fatty Acid Import Defects-In an effort to determine whether fatty acid import was restored by the plasmid-encoded suppressors, we monitored the accumulation of the fluorescent fatty acid analogue C 1 -BODIPY-C 12 using confocal laser scanning microscopy in wildtype strains and transformants of the faa1⌬ faa4⌬ and faa1⌬ fat1⌬ strains. Wild-type cells import C 1 -BODIPY-C 12 quickly (within 30 s) by a process that is essentially irreversible, which we suspect reflects metabolic activation (14). Previous work from our laboratory (13,14) has shown a deletion within FAT1 severely restricts C 1 -BODIPY-C 12 accumulation. Deletion of FAA1 alone appears to decrease, but not completely eliminate, accumulation of C 1 -BODIPY-C 12 . Deletion of FAA4 has essentially no effect. In contrast, when both FAA1 and FAA4 are deleted, the accumulation of C 1 -BODIPY-C 12 is restricted in a manner similar to that observed in FAT1 mutants (11). As illustrated in Fig. 2, the accumulation of C 1 -BODIPY-C 12 was restored in the faa1⌬ faa4⌬ strain harboring FAT1, FAA1, or FAA4 on a multicopy plasmid. These results point out that FAT1 is a true multicopy suppressor. Only when expressed from a 2 plasmid can FAT1 compensate for deletions in FAA1 and FAA4. These data support the notion that Fat1p and Faa1p or Faa4p form a functional network facilitating the import and activation of exogenous fatty acids, and in wild-type cells each functions in a distinct yet coordinate manner. It is important to note that FAA4 on a multicopy plasmid (YEpDB133) did not restore in C 1 -BODIPY-C 12 accumulation in the faa1⌬ fat1⌬ strain, whereas both FAA1 and FAT1 did (Fig. 2). These results indicated that Faa1p, Faa4p, and Fat1p have overlapping, yet distinct roles. Of particular importance was the finding that Faa1p and Fat1p appeared to be functionally linked. Table III summarizes both the phenotypes and fatty acid transport profiles of the mutant strains alone and transformed with the selected multicopy suppressor plasmids. These findings are consistent with data obtained on the mmFATP1 and fatty acyl-CoA synthetase, which are proposed to form a functional complex (15).
Deficiencies in Long-chain Acyl-CoA Synthetase Activity in faa1⌬ faa4⌬ and faa1⌬ fat1⌬ Strains Can Be Compensated by Multicopy FAT1-The identification of FAT1 as a multicopy suppressor in experiments using the faa1⌬ faa4⌬ and faa1⌬ fat1⌬ strains demonstrated that in high copy FAT1 alone as well FAA1 alone could compensate for the defects with regard to importing exogenous long-chain fatty acids. These data imply that under these conditions Fat1p contributed an enzymatic activity to promote the unidirectional transport of exogenous long-chain fatty acids. Previously, we have shown that deletion of FAT1 does not reduce long-chain fatty acyl-CoA synthetase activities measured using whole cell extracts, whereas deletion of FAA1 and FAA4 reduced these activities ϳ95% (Table IV) (11,13). Likewise, when FAT1 is cloned into a centromeric plasmid (a pRS316 derivative designated pDB102) (14) and transformed into the faa1⌬ faa4⌬ strain, long-chain fatty acyl-CoA synthetase activities are not substan- tially elevated, and no complementation was observed on YPD-CER-OLE plates (data not shown). Our laboratory and others (14,24) have shown that Fat1p has intrinsic very long-chain (C 22 -C 26 ) fatty acyl-CoA synthetase activity. Indeed, when we first characterized the FAT1 gene, we noted Fat1p shared similarities to the adenylateforming family of enzymes, which includes the fatty acyl-CoA synthetases (13). We reasoned that when expressed from a high copy number plasmid, FAT1 would result in sufficient longchain fatty acyl-CoA synthetase activity to promote growth of the faa1⌬ faa4⌬ and faa1⌬ fat1⌬ strains under the synthetic lethal conditions used in this study. To test this idea, we measured fatty acyl-CoA synthetase activities in total cell extracts from the parental strain and strains harboring the multicopy suppressor plasmids using oleate (C 18:1 ) as a substrate (Table IV). Extracts prepared from the faa1⌬ faa4⌬ strain harboring YEpDB17 (encoding Fat1p) had ϳ4-fold higher oleoyl-CoA synthetase activity compared with the strain carrying the vector YEp24, which was 30% of the level obtained for the wild-type strain. This modest increase in oleoyl-CoA synthetase activity correlated with a 3-fold increase in protein level estimated using Western blot analysis of cellular extracts employing a Fat1p-specific antibody and analyzed using NIH Image analysis software. The same strain transformed with YEpDB02 and YEpDB133 (encoding Faa1p and Faa4p, respectively) had 10-and 2-fold oleoyl-CoA synthetase activities, respectively, compared with the same control cells. It is unclear why increased dosage of FAA4 had such a limited impact on total oleoyl-CoA synthetase activity. This may be due to protein instability as noted for the purified enzyme (26) or due to regulatory parameters poorly defined at the present time. In the case of the faa1⌬ fat1⌬ strain, we noted similar results. Most notable among these was the finding that YEpDB17 (FAT1) resulted in oleoyl-CoA synthetase activities, which were increased 6-fold over the same strain harboring the plasmid vector (Table IV).
As noted above, Fat1p has been shown to confer very longchain fatty acyl-CoA synthetase activity. Therefore, we also measured fatty acyl-CoA synthetase activities in the same cell extracts from above using the very long-chain fatty acid, lignocerate (C 24:0 ) as substrate (Table IV). As expected, expression of FAT1 from YEpDB17 increased these activities just over 4-fold, whereas expression of FAA1 from YEpDB02 or FAA4 from YEpDB133 had no significant effect on total cellular very longchain fatty acyl-CoA synthetase activities. Similar results were obtained for the fat1⌬ faa1⌬ and fat1⌬ host strains (Table IV).
Fat1p and Fatty Acyl-CoA Synthetase Form a Physical Complex-In previous work, we provided independent evidence that Fat1p (13) and Faa1p or Faa4p (11) are each required for fatty acid import in yeast. The results of the multicopy suppressor analyses detailed above extended these results to include a functional dependence of fatty acid transport on both Fat1p and fatty acyl-CoA synthetase (Faa1p or Faa4p). Indeed, these data provided evidence suggesting Fat1p and Faa1p or Fat1p and Faa4p interact to coordinately facilitate fatty acid transport. The results from the multicopy suppressor screen are consistent with the notion that, at least in yeast, no other proteins participate in this process. Yet this experimental approach did not address whether Fat1p and fatty acyl-CoA synthetase form a physical complex. To address this question, we employed three different experimental strategies as follows: 1) negative dominance of mutant Fat1p over fatty acyl-CoA synthetase; 2) yeast two-hybrid analyses to investigate the hypothesized physical linkage between Fat1p and Faa1p or Faa4p; and 3) co-immunoprecipitation of Fat1p and a cognate fatty acyl-CoA synthetase.
Often when two proteins physically interact to form a functional complex, inactivation of one protein due to a mutation will result in a reduction in activity for the partner protein. This phenomenon is called negative dominance. Long-chain acyl-CoA synthetase activity in yeast is primarily contributed by Faa1p (Ն95%). Therefore, we reasoned that the overexpression of nonfunctional Fat1p would result in a reduction of long-chain acyl-CoA synthetase activity if the proteins physically interact to facilitate vectorial acylation. For these experiments, we expressed a peptide derived from Fat1p made up of the C-terminal 125 amino acids (residues 545-669; T7 Fat1p 125C ). FIG. 1. FAA1, FAA4, and FAT1  This peptide derived from Fat1p was non-functional in transport and activation, yet when analyzed using SDS-PAGE it formed a dimer, which was stable to boiling, suggesting it might contain a protein-protein interaction domain (data not shown). The expression of T7 Fat1p 125C significantly reduced oleoyl-CoA synthetase activity (compared with vector control) (Fig. 3A). The reduction in activity was correlated with expression of the T7 Fat1p 125C peptide, detected using a Western blot following expression with anti-T7 antibodies (Fig. 3B). Under these conditions, the T7 Fat1p 125C peptide is expressed at levels nearly 10-fold higher when compared with native Fat1p (data not shown). These data are consistent with the proposal that Fat1p and fatty acyl-CoA synthetase form a functional complex.
Another method, which has become standard to evaluate protein-protein interactions, is the yeast two-hybrid system. These experiments employed plasmids encoding a bait protein, which consisted of a fusion between full-length Faa1p and the DNA binding domain of bacterial LexA and several trap proteins, which consisted of protein fusions containing full-length Fat1p and peptides derived from Fatp1 and the Gal-driven activation domain. In this system a third reporter plasmid contains the DNA-binding site of LexA in the promoter region driving expression of lacZ (encoding ␤-galactosidase), which is dependent on specific protein-protein interactions between the bait (bound to the DNA binding site) and the trap (fused to activation domain, which interacts with yeast RNA polymerase II). As shown in Table V, positive interactions between fulllength Faa1p and either full-length Fat1p or two peptides carrying C-terminal fragments of Fat1p (Fat1p 500C and Fat1p 125C ) were found when compared with the trap vector control alone. The peptide, which conferred negative dominance to fatty acyl-CoA synthetase activity ( T7 Fat1p 125C ) detailed above, also results in a positive interaction with Faa1p using the yeast two-hybrid system.
Additional evidence for specific protein-protein interactions between Fat1p and Faa1p or Faa4p was obtained using coimmunoprecipitation. As detailed under "Experimental Procedures," Fat1p was tagged with a T7 epitope ( T7 Fat1p), and the fatty acyl-CoA synthetases were tagged with a V5 epitope ( V5 Faa1p and V5 Faa4p). Following growth, extracts were prepared from cells expressing T7 Fat1p and V5 Faa1p or V5 Faa4p and immunoprecipitated using anti-T7 or anti-V5 antibodies.
1 ϩϩϩ a Growth was scored by comparison to the wild-type strain. 1, positive growth after 24 h; 2, growth between 24 and 48 h; 3, growth between 48 and 96 h; 4, no growth after 96 h at 30°C.
b Fatty acid accumulation monitored using C 1 -BODIPY-C 12 , visualized using confocal microscopy and scored relative to the wild type: wild-type, ϩϩϩ; ϩϩ, slightly reduced from wild type; ϩ, visible accumulation but down dramatically; Ϫ, no visible accumulation. The presence of the second protein in the complex was detected by Western blot analyses using the reciprocal antibody. The data presented in Fig. 4 showed that V5 Faa1p and V5 Faa4p are co-immunoprecipitated with T7 Fat1p whether the precipitating antibody was anti-T7 directed against Fat1p or anti-V5 directed against one of the Faa proteins. In our control experiments using protein A-Sepharose beads alone (Fig. 4) or using an unrelated antibody (c-Myc) (not shown), we did not pull down the Fat1p-fatty acyl-CoA synthetase complex. Additionally, to test for nonspecific protein-protein interactions, we probed the immunocomplex using an antibody against Pma1p, an unrelated plasma membrane protein (25). No co-immunoprecipitation of Pma1p with Fat1p, Faa1p, or Faa4p was detected (Fig. 4). These data are in agreement with the results of the multicopy suppressor analysis, negative dominance, and yeast two-hybrid data presented above and fully support the notion that Fat1p and Faa1p or Faa4p form a physical complex, which we suggest is crucial to the process of vectorial esterification of exogenous long-chain fatty acids.

DISCUSSION
When long-chain fatty acids are supplied in the growth media, S. cerevisiae transports these compounds into the cell by a process, which requires Fat1p and the fatty acyl-CoA synthetase Faa1p. Even though Faa1p and Faa4p have been suggested to be functionally redundant, previous results and those presented here show that Faa1p, rather than Faa4p, plays a more distinct role in fatty acid import (11,26). Importantly, the experiments reported here provide substantial genetic and biochemical evidence that Fat1p and fatty acyl-CoA synthetase (Faa1p or Faa4p) form a physical complex required to facilitate fatty acid import. These data are consistent with the hypothesis that the fundamental mechanism driving the accumulation of exogenous fatty acids within the cell is vectorial acylation whereby exogenous fatty acids are metabolically trapped as acyl-CoA thioesters.
Until this time, the physical and functional association of FATP and fatty acyl-CoA synthetase has been inferential (3,11,15). The present studies indicate that in the natural environment when fatty acids are limiting as, for example, occurs during hypoxia, Fat1p and Faa1p are each required for fatty acid import. Whereas each protein fulfills a separate function, the activities are coordinated and facilitated by a physical interaction. The former conclusion is based on the observation that in single copy neither gene can substitute for the other. The distinct functions for Fat1p and Faa1p were apparent in enzymatic analyses of acyl-CoA synthetase specificity and activity and in our fatty acid transport studies. In multicopy, Faa1p can substitute for Fat1p, and in turn, Fat1p can substitute for Faa1p in potentiating fatty acid import. Thus the apparent increase in accumulation of C 1 -BODIPY-C 12 when either of these genes is overexpressed appears related to the A, anti-T7 antibody (␣-T7) was used to pull down full-length T7 Fat1p in extracts prepared from cells co-expressing T7 Fat1p and V5 Faa1p or V5 Faa4p as indicated. The proteins were separated by SDS-PAGE, and subsequent Western blots were probed with anti-V5 (␣-V5) antibody to detect V5 Faa1p or V5 Faa4p as shown. B, similarly, anti-V5 was used as the precipitating antibody to pull down V5 Faa1p or V5 Faa4p following co-expression of T7 Fat1p and V5 Faa1p or V5 Faa4p, and the resultant blot was probed with anti-T7. IB, antibody used in the immunoblot; T, total cell extract; IP, samples immunoprecipitated with the indicated antibody; Beads, protein A-Sepharose alone without an immunoprecipitating antibody. Anti-Pma1p was used as a control protein specific to a yeast plasma membrane protein but unrelated to Fat1p, Faa1p, or Faa4p.

FIG. 3. Negative dominance of mutant Fat1p over fatty acyl-CoA synthetase.
Extracts were prepared from wild-type cells carrying a plasmid encoding T7 Fat1p 125C under the control of a galactose-inducible promoter and used to assess oleoyl-CoA synthetase activity (A) and peptide expression (B) using a Western blot probed with anti-T7 antibody. essential role of long-chain fatty acyl-CoA synthetase activity in import and utilization rather than to a transport function per se. Thus utilization creates a diffusional gradient dependent upon the acyl-CoA synthetase Faa1p (and to a more limited extent Faa4p) but not Fat1p. The role of Fat1p in fatty acid import appears to be distinct from Faa1p and essential only at limiting fatty acid concentrations (Յ500 M) such as might occur when cells are growing under hypoxic conditions in the natural environment. We suggest this mechanism of fatty acid transport by vectorial acylation exemplifies a system common to eukaryotes including mammalian cells that functions through FATP and a cognate fatty acyl-CoA synthetase.
There is substantial data showing Fat1p plays a role in long-chain fatty acid import yet has intrinsic very long-chain (C 22 -C 26 ) fatty acyl-CoA synthetase activity (14,24). This presents somewhat of a dilemma. Our results are consistent with the notion that the specificity of the fatty acid import system in yeast is for long-chain fatty acids as opposed to very long-chain fatty acids. Addition of very long-chain fatty acids to the growth media of yeast strains defective in very long-chain fatty acid synthesis does not alleviate the growth defect, suggesting the very long-chain fatty acids cannot be trafficked from an exogenous source to the site of metabolic utilization (27). Yet Fat1p is a central component of the long-chain fatty acid import system in yeast, being required both under anaerobic conditions and under cerulenin-induced conditional lethality, where exogenous long-chain fatty acids are required for growth (14). We have provided evidence recently (28) that the very longchain fatty acyl-CoA synthetase activity intrinsic to Fat1p can be distinguished from fatty acid import in specific mutant alleles of FAT1 with single amino acid substitutions. Additionally, the specificity of Fat1p-dependent import is for long-chain fatty acid substrates, whereasFat1p-dependent fatty acyl-CoA synthetase activity is for very long-chain substrates (14,24). The deletion of FAA1 encoding the major long-chain fatty acyl-CoA synthetase decreases fatty acid import nearly 3-fold; therefore, we suggest this enzyme is primarily responsible for activating fatty acids from an exogenous source and therefore contributes to the specificity of the import system (11).
The ability of FAT1 encoded within a high copy number episome to suppress the phenotype on YNBD containing oleate and cerulenin of a faa1⌬ faa4⌬ strain and the corresponding ability of plasmid-encoded Faa1p to suppress the same phenotype of the faa1⌬ fat1⌬ strain is consistent with a functional interrelationship between Fat1p and Faa1p in long-chain fatty acid import. However, multicopy suppression might also result from alterations in intracellular metabolism and regulation distinct from the coupled transport/activation process when fatty acid import or fatty acyl-CoA synthetase activity is highly elevated by comparison to activities contributed by a single copy of the native gene. As detailed in these studies, we did not observe oleoyl-CoA synthetase activities comparable with or exceeding wild-type levels for either strain expressing FAT1 or FAA4 in high copy. In the case of the faa1⌬ faa4⌬ strain (wild type for FAT1) transformed with YEpDB17 (FAT1), there was sufficient oleoyl-CoA synthetase activity (albeit only ϳ30% wild-type), which appeared to drive the coupled import/activation process. On the other hand in the faa1⌬ fat1⌬ strain transformed with YEpDB133 (FAA4), there was detectable oleoyl-CoA synthetase activity (ϳ10% wild type), but this was not sufficient to overcome the block as a consequence of a deletion in FAT1. By comparison, the faa1⌬ fat1⌬ strain transformed with YEpDB02 (FAA1) had robust oleoyl-CoA synthetase activity (ϳ121% wild type), which was sufficient to overcome the block due to the fat1⌬ deletion. Therefore, we believe the sup-pression is caused by overexpression of one of the partners in the import process, Fat1p or Faa1p.
Although we did not identify new partners in the fatty acid trafficking pathway by selecting multicopy suppressors, these results are of particular significance because they confirmed by using a powerful genetic approach the importance of an interaction between Fat1p and Faa1p in fatty acid import. Indeed, with one note of caution based on the suppressors presumed to be chromosomally encoded, these studies indicate these two proteins may be the only components mediating this process in yeast. Our present results parallel the previous work of Schaffer and Lodish (3) that identified independent clones encoding murine FATP1 and a fatty acyl-CoA synthetase using a functional cloning strategy. Functional cloning requires, in essence, overexpression of the protein target in a manner analogous to our studies using multicopy suppression. The murine FATP1 and fatty acyl-CoA synthetase each were identified and shown to function to promote the accumulation of C 1 -BODIPY-C 12 (3). By analogy, we have shown that Fat1p and Faa1p, when expressed from a 2-m plasmid also function to promote the accumulation of C 1 -BODIPY-C 12 . The murine FATP1 also has intrinsic very long-chain acyl-CoA synthetase activity (29). Likewise, we and others (14,24) have shown yeast Fat1p is a very long-chain acyl-CoA synthetase. Previously, we have shown (14) murine FATP1 complements the biochemical phenotypes associated with the fat1⌬ strain in yeast indicating that the yeast and the mouse proteins are functionally equivalent. Collectively, these data support the notion that the fatty acid import mechanism working through Fat1p (or FATP) and fatty acyl-CoA synthetase is primarily through the esterification of the fatty acid with CoA, which results in metabolic trapping. Our working hypothesis is that Fat1p functions to increase fatty acid binding to the membrane, which in turn potentiates diffusion across the membrane. The fatty acid is subsequently metabolically activated concomitant with abstraction from the membrane by the Faa1p-Fat1p complex thereby generating a concentration gradient, which further drives the import process.
The present work demonstrates for the first time a physical interaction between Fat1p and Faa1p or Faa4p. In each series of experiments (i.e. negative dominance, two-hybrid analyses, and co-immunoprecipitation), the full-length proteins and Cterminal peptides of Fat1p resulted in positive interactions. One outcome from these experiments suggests the proteinprotein interaction domain of Fat1p is localized at least in part to the C-terminal 125 residues. At present, we have no data localizing an interaction domain within Faa1p or Faa4p. Those experiments are currently underway.
Fatty acid transport in S. cerevisiae is tightly coupled to utilization and is primarily dependent upon the products of two genes, Fat1p and Faa1p. These proteins function in concert to couple fatty acid import to fatty acid activation and metabolic utilization, a process first described in bacteria as vectorial acylation. Due to the functional conservation of these proteins in higher eukaryotes, yeast provides a valuable, genetically tractable model system useful to further elucidate the mechanisms that underpin fatty acid import in eukaryotic systems.