Members of the Arabidopsis FAE1-like 3-Ketoacyl-CoA Synthase Gene Family Substitute for the Elop Proteins of Saccharomyces cerevisiae*

Several 3-keto-synthases have been studied, including the soluble fatty acid synthases, those involved in polyketide synthesis, and the FAE1-like 3-ketoacyl-CoA synthases. All of these condensing enzymes have a common ancestor and an enzymatic mechanism that involves a catalytic triad consisting of Cys, His, and His/Asn. In contrast to the FAE1-like family of enzymes that mediate plant microsomal fatty acid elongation, the condensation step of elongation in animals and in fungi appears to be mediated by the Elop homologs. Curiously these proteins bear no resemblance to the well characterized 3-keto-synthases. There are three ELO genes in yeast that encode the homologous Elo1p, Elo2p, and Elo3p proteins. Elo2p and Elo3p are required for synthesis of the very long-chain fatty acids, and mutants lacking both Elo2p and Elo3p are inviable confirming that the very long-chain fatty acids are essential for cellular functions. In this study we show that heterologous expression of several Arabidopsis FAE1-like genes rescues the lethality of an elo2Δelo3Δ yeast mutant. We further demonstrate that FAE1 acts in conjunction with the 3-keto and trans-2,3-enoyl reductases of the elongase system. These studies indicate that even though the plant-specific FAE1 family of condensing enzymes evolved independently of the Elop family of condensing enzymes, they utilize the same reductases and presumably dehydratase that the Elop proteins rely upon.

found in the pollen coat and in leaves, and they are also found in the storage triacylglycerols of some seeds. Very long chain polyunsaturated fatty acids, including arachidonic acid, eicosapentaenoic acid, and docosahexaenoic acid, are abundant in the membranes of brain and retina, and they serve as precursors of the biologically active eicosanoids. Despite mounting evidence that the VLCFAs are important in cellular physiology, many aspects of their synthesis and function remain to be resolved.
In yeast the saturated VLCFAs are present in the sphingolipids, which are essential for cell viability. Interestingly, a mutation that leads to the synthesis of C26-containing phosphatidylinositol can bypass the requirement for sphingolipids (though the mutant strains grow poorly) highlighting the importance of the VLCFA (1,2). It is now well established that the Elop proteins are components of the membrane-associated elongase system that mediates VLCFA synthesis. Elo1p is involved in the elongation of medium-chain fatty acids, whereas Elo2p and Elo3p participate in the elongation of C16 to C24, and Elo3p is required for elongating C24 to C26 (3)(4)(5)(6). Yeast transformed with foreign ELO homologs synthesize novel (e.g. mono-and polyunsaturated) elongated fatty acids demonstrating that the Elop proteins are responsible for selection of the acyl-CoA substrates for elongation (7)(8)(9)(10)(11)(12)(13). Although this implicates the Elop proteins as condensing enzymes, conclusive evidence that they directly catalyze the condensation reaction is lacking.
In plants the condensation step of fatty acid elongation is catalyzed by the FAE1-KCSs. For example FAE1, the founding member of the Arabidopsis FAE1-KCS gene family, is involved in the synthesis of the 22-carbon monounsaturated fatty acid (14). Lassner et al. (15) demonstrated that the FAE1 homolog from jojoba encodes the condensing enzyme, 3-ketoacyl-CoA synthase (KCS), and Millar and Kunst (16) showed that KCS regulates the substrate specificity and activity of the elongation process. All 3-keto synthases characterized to date, including the FAE1-KCSs, the soluble fatty acid synthases, and those involved in polyketide synthesis, catalyze a Claissen condensation by essentially the same mechanism (17). In these systems, studies of the mechanism, analysis of essential catalytic residues, sequence homology, and, when available, the three-dimensional structure reveal that all of these condensing enzymes have a common ancestor.
In contrast, the Elop proteins provide an exception to the notion that all condensing enzymes involved in fatty acid elongation arose from a common origin. The hallmarks of all condensing enzymes studied to date are missing from the ELO gene family. For example, all condensing enzymes contain a conserved cysteine, which is the acyl acceptor in the first step of the mechanism. Although there are two conserved cysteines among the Elop homologs, there is no evidence of the catalytic triad that is found in other 3-keto synthases. A highly conserved feature of the Elop sequences is a histidine box, suggestive of a metal binding site, yet no other condensing enzymes contain such a domain. The FAE-KCSs are membrane-bound condensing enzymes with two membrane-spanning domains at the N terminus (18). In contrast, hydropathy analyses of the Elop homologs predict the presence of 5-7 transmembrane domains spread between the N and C termini.
In addition to the Elop proteins, there are several other proteins that are required for fatty acid elongation. Each C 2 elongation is a four-step process consisting of a condensation-reduction-dehydration-reduction cycle (see Fig. 1). The yeast YBR159w and TSC13 genes encode the 3-keto and 2,3-enoyl reductases of the elongation cycle (19 -21). Consistent with early biochemical characterization of fatty acid elongation that suggested the elongase enzymes are organized in a complex (22), the Elop, Ybr159p, and Tsc13p elongase proteins coimmunoprecipitate with one another (20,21). The dehydratase component of the elongase system has yet to be identified, and whether there are other proteins required is also not yet known.
Although yeast have no FAE-KCS homologs, several Arabidopsis FAE-KCSs have been expressed in yeast, and these heterologously expressed enzymes are able to direct the synthesis of novel fatty acids (23). For example, expression of FAE1 in yeast leads to accumulation of monounsaturated C22 erucic acid (18,(23)(24)(25). That the FAE-KCSs in yeast lead to the synthesis of novel elongated fatty acids implies the reconstitution of a heterologous elongase and raises several questions. For example, can the FAE-KCSs substitute for the yeast Elop proteins? Does the conversion of the 3-keto intermediates generated by the FAE-KCSs into elongated fatty acids depend on the other enzymatic activities of the elongase complex? If so, do the FAE-KCSs physically associate with the reductases? In the studies reported here, we have addressed these questions.

EXPERIMENTAL PROCEDURES
Media, Strains, and Genetic Manipulations-Yeast media were prepared, and cells were grown according to standard procedures (26). The yeast strains used in this study are listed in Table 1.

Construction of the FAE1-KCS Expression Plasmid, the HA-FAE-KCS Expression Plasmids, and the Myc-tagged Yeast and Arabidopsis
Ybr159p-FAE1-KCS was PCR-amplified from a cDNA kindly provided by Dr. Ljerka Kunst (15) using primers FAE1-PvuF and FAE1-PvuR. The amplified fragment was restricted with PvuII and ligated into a HindIIIcut pADH plasmid after filling the 5Ј-protruding ends with DNA polymerase I. For HA tagging of the FAE-KCSs, XhoI-ended or SalIended PCR products were generated that extended from codon 2 through to the stop codon for each gene. The primers used are listed in Table 2 with the Arabidopsis At gene designator and either -HAF (for HA-tagging primer, forward) or -HAR (for HA-tagging primer, reverse). The PCR-generated fragments were restricted with either XhoI or Sal1and ligated into the SalI site of plasmid pADH-3X-HA (21) resulting in an in-frame fusion of the N-terminal triple-HA tag to each of the FAE-KCSs. For tagging of the yeast Ybr159p and the Arabidopsis homolog of Ybr159p (encoded by At1g67730 and hereafter referred to as AtYbr159p), a triple-Myc epitope tag was introduced at the N terminus of the pESC-Ybr159 and pESC-AtYbr159 constructs described before (17). For this an AvrII site was first introduced after the start codon in both constructs by QuikChange PCR mutagenesis (Stratagene), using the complementary mutagenic primers 159AvrF and 159AvrR or At159AvrF and At159AvrR, respectively. An SpeI-ended fragment carrying the triple-Myc cassette was generated by PCR using a Bluescript-based 3X-Myc-containing plasmid (19) and ligated into the AvrII site.
Elongase Assays-Microsomes were prepared from exponentially growing cells. Cells were pelleted at 5000 ϫ g, washed with water, and repelleted. The cell pellets were resuspended in TEGM (0.05 M Tris (7.5), 1 mM EGTA, 1 mM ␤-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 1 g/ml pepstatin A, 1 g/ml aprotinin) buffer at 1 ml/50 OD cells, and glass beads (0.5-mm diameter) were added to the meniscus. Cells were disrupted by four cycles of vortexing for 1 min followed by cooling on ice for 1 min, and pelleted at 8,000 ϫ g for 10 min. The supernatant was transferred to a fresh tube and spun at 135,000 ϫ g for 30 min. The resulting pellet was resuspended by Dounce homogenization in at least a 10ϫ volume of TEGM buffer and repelleted at 135,000 ϫ g. The final membrane pellet was resuspended in TEGM buffer containing 33% glycerol and stored at Ϫ80°C.
Total elongase activity was measured in a volume of 200 l containing 50 mM Tris, pH 7.5, 1 mM MgCl 2 , 150 M Triton X-100, 1 mM NADPH, 1 mM NADH, 10 mM ␤-mercaptoethanol, 1 mM dithiothreitol, 40 M acyl-CoA acceptor, and 60 M 2[ 14 C]malonyl-CoA (0.05 Ci/ml) at 37°C. The reaction was initiated by the addition of 0.2 mg of microsomal protein. Protein concentrations were determined using the Bio-Rad protein assay reagent (Bio-Rad). For assays of the condensing activity, the NADPH/NADH was omitted. The reaction was terminated after 10 min by adding 200 l of 5 M KOH/10% MeOH and heating at 80°C for 1 h. Following addition of 200 l of 10 N H 2 SO 4 , free fatty acids were recovered by two 1.5-ml extractions into hexane. Following evaporation of hexane, the fatty acid methyl esters (FAMES) were generated by dissolving in 1 ml of 1 M methanolic HCl and heating at 80°C for 1 h. An equal volume of 0.15 M NaCl was added, and the FAMES were extracted into hexane and resolved by reverse phase C18-silica gel TLC using CHCl 3 :MeOH:H 2 O (5:15:1) as the developing solvent. The radiolabeled fatty acids were detected and quantified using the Cyclone Storage Phosphor System (Packard Instrument Company, Inc., Meriden, CT).
Immunoblotting and Immunoprecipitation-Microsomes were prepared from strains containing HA-tagged FAE1-KCS and the yeast Myc-tagged Tsc13p or the Arabidopsis homolog of Tsc13p (encoded by At3g55360 and hereafter referred to as AtTsc13p) (21,28) or the yeast and the Arabidopsis Myc-Ybr159p (described above). The microsomes were solubilized at 1 mg/ml with 0.1% sucrose monolaurate (Roche Diagnostics) for 10 min, and the high speed (135,000 ϫ g, 30 min) supernatant was collected. The supernatant (100 l) was incubated with 3 l of the precipitating antibody for 1 h and then with 25 l of protein A-Sepharose (125 mg/ml from Sigma) for 1 h. The precipitates were washed three times with 1 ml of TEGM buffer containing 0.1% sucrose monolaurate and resuspended in 100 l of SDS loading buffer, and a 10-l sample was subjected to SDS-PAGE on a 4 -12% gradient gel. Following transfer of the separated proteins to nitrocellulose, the blots were blocked in 0.1 M Tris, 7.5, 0.15 M NaCl, 0.1% Tween 20, 5% dry milk. The Myc-Tsc13p was detected with horseradish peroxidase-conjugated monoclonal anti-Myc antibodies (Sigma) at 1/2500. HA-FAE1-KCS was detected using horseradish peroxidase-conjugated monoclonal anti-HA antibodies (from Roche Molecular Biochemicals) at 1/1000. The bound antibodies were detected by the ECL Western blotting detection system (Amersham Biosciences).
Fatty Acid Analysis-Cells (5 ϫ 10 8 ) in mid-logarithmic growth were harvested and resuspended in 100 l of distilled H 2 O along with C21 and C23 fatty acids as internal standards. Fatty acids were extracted, and the fatty acid methyl esters were prepared as previously described (21). Gas chromatography/mass spectrometry (GCMS) was performed using an HP-6890 Series GC system with a Supelcowax 10 column, coupled to an HP-5973 Mass Selective Detector.
IPC Analysis-Cells were grown in SD medium (0.67% yeast nitrogen base, 2% bacto-agar, 2% glucose) without inositol overnight and harvested during exponential growth phase. The cells were resuspended to 10 A 600 /ml in SD without inositol, and 50 Ci/ml myo-[2-3 H]inositol (16 Ci/mmol, Amersham Biosciences) was added. After a 10-min labeling, the cells were diluted to 2 A 600 /ml with SD media containing 30 M unlabeled inositol and incubated for an additional 90 min. Sodium azide (4 mM final) was added, and the lipids were extracted by bead beating in CHCl 3 :MeOH (1:1). The extraction was repeated, and the extracted material was combined, clarified by centrifugation, dried, and desalted using water saturated butanol. 250,000 cpm were spotted on TLC and developed using CHCl 3 :MeOH:H 2 O:AcOH (16:4:1.6:4). The TLC was dried, sprayed with EN 3 HANCE (PerkinElmer Life Sciences), and visualized using Kodak XAR film.

The elo Mutants Are Affected in the Condensation
Step of Fatty Acid Elongation-A number of studies are consistent with the conclusion that the Elop proteins participate specifically in the condensation step of fatty acid elongation. For example, many previous studies of the Elop family of proteins have shown that they control the substrate specificity of fatty acid elongation (reviewed in Ref. 29). To address this question further, the condensation step of fatty acid elongation was assayed in vitro using microsomes prepared from wild-type and elo1⌬, elo2⌬, and elo3⌬ mutant yeast cells. The ability of the mutant microsomes to direct the elongation of various acyl-CoA acceptors was assessed by measuring incorporation of label from [ 14 C]malonyl-CoA into elongated fatty acids. By omitting NADPH from the assay, the condensation reaction can be assayed (Fig. 1).
As shown in Fig. 2, microsomes prepared from wild-type cells form the 3-keto condensation products by condensing malonyl-CoA with each of the acyl-CoA substrates when NADPH is omitted. For example, with the C14-CoA substrate the C16 -3-keto product accumulates, with the C16-CoA substrate the C18 -3-keto product accumulates, and so forth ( Fig. 2A). In contrast, when NADPH is included in the in vitro assay, the whole series of elongated fatty acids with chain lengths up to C26 are generated ( Fig. 2A). Assays of the condensation reaction using the microsomes from the elo mutants confirm that the Elop proteins participate in the condensation step of elongation. Previous in vivo studies have shown that Elo1p is required for elongation of myristate to palmitate (3,5,6). The in vitro assays reveal that the condensation step is specifically affected, because microsomes prepared from cells lacking Elo1p are unable to form the C16 -3-keto product when C14-CoA is used as a substrate for the condensation reaction (Fig. 2B). With the elo1⌬ mutant microsomes, the longer substrates (C16-CoA to C24-CoA) are converted to the 3-keto products by Elo2p and/or Elo3p. The elo2⌬ mutant microsomes have low condensation activity toward the C16 and C18-CoA substrates (Fig. 2C). Finally, the elo3⌬ mutant microsomes form very little C24 -3-keto product with the C22-CoA substrate and no C26 -3-keto product with the C24-CoA substrate (Fig. 2D). These data provide additional evidence that the Elop proteins are required for the condensation step of elongation.

The Arabidopsis FAE1 Gene and Other FAE-KCS Genes Rescue the
Lethality of an elo2⌬elo3⌬ Mutant-The yeast elo2⌬elo3⌬ double mutant is inviable, reflecting that the VLCFAs have essential functions for cell growth. Whereas the Elop proteins are needed for the condensation step of fatty acid elongation in yeast, members of the FAE1-like family of enzymes, hereafter referred to as the FAE-KCSs, catalyze the condensation step of fatty acid elongation in plants. As discussed above, there is no significant homology between the FAE-KCS proteins and the Elop proteins. Although previous studies have clearly demonstrated that FAE1-KCS expression in yeast leads to the synthesis of novel fatty acids (16,18,19,23,25), whether the FAE-KCSs can substitute for the yeast Elo2p/Elo3p proteins had not been tested. Therefore, it was of interest to test whether the Arabidopsis FAE1-KCS-condensing enzyme could rescue the lethality of the elo2⌬elo3⌬ double mutant.
To address this question, a yeast mutant with chromosomal disruption of both the ELO2 and ELO3 genes that carried a wild-type ELO3 gene on a URA3-marked, and therefore 5-fluoroorotic acid (FOA)-FIGURE 1. Fatty acid elongation in S. cerevisiae. Palmitoyl-CoA is elongated to a C26-VLCFA by the membrane-associated elongase enzymes. Each cycle of elongation involves four successive reactions, the condensation of malonyl-CoA with the acyl-CoA substrate, reduction of the 3-keto acyl-CoA, dehydration of the 3-hydroxy acyl-CoA, and reduction of the trans-2,3-acyl-CoA. Although the Elop proteins have not been shown to catalyze condensation, it is clear that they mediate substrate selection at the condensation step.
Whether other members of the Arabidopsis FAE-KCS gene family are able to restore viability to the elo2⌬elo3⌬ double mutant was investigated by screening an Arabidopsis cDNA library. A yeast expression library was constructed by ligating Arabidopsis cDNAs behind the constitutively expressed yeast pADH1 promoter in a LEU2-marked plasmid ("Experimental Procedures"). The library was transformed into the elo2⌬elo3⌬ double mutant carrying the wild-type ELO3 gene on a URA3-marked plasmid (TDY7005, Table 1), and leucine prototrophic transformants were selected. Those transformants that were able to lose the ELO3-containing URA3-marked plasmid were identified by screening for growth on FOA. The cDNA-containing plasmids were recovered from the FOA-resistant transformants and retested to confirm that they were able to rescue the lethality of the elo2⌬elo3⌬ double mutant. Three additional FAE-KCS genes, At2g16280, At5g43760, and At1g04220, were identified that rescue the elo2⌬elo3⌬ mutant (Fig. 3A). It is likely that other members of the Arabidopsis FAE-KCS gene family can also substitute for the yeast Elo2p/3p proteins but that they are not as highly represented in this cDNA expression library. It is worth noting that five additional FAE-KCSs were PCR-amplified from the expression library, cloned into the 3X-HA expression vector, and confirmed to be expressed. However, they were not able to rescue the elo2⌬elo3⌬ double mutant suggesting that they are unable to elongate the endogenous  (38). The small amount of C16 and C18 that forms in the absence of any added acyl-CoA (shown for the wt, labeled "None" in panel A) results from contaminating fatty acid synthase that purifies with the microsomes, as it is eliminated by including cerulenin in the assays or by deleting the FAS2 gene. A, the elo2⌬elo3⌬ double mutant is unable to lose the URA3-marked plasmid that carries the wild-type ELO3 gene, and is therefore unable to grow on FOA. Introduction of the indicated FAE-KCS genes on a LEU2-marked yeast expression plasmid (but not the "vector" alone) allows the mutant strain to lose the URA3-marked ELO3-containing plasmid, demonstrating that the FAE-KCS genes can substitute for the ELO3 gene. B, the FAE-KCS genes also rescue the elo1⌬elo2⌬elo3⌬ triple mutant. FIGURE 4. The FAE-KCS genes restore VLCFA synthesis to the rescued elo1⌬elo2⌬elo3⌬ triple mutants but do not completely reverse the phenotypes associated with VLCFA synthesis defects. A, fatty acids were extracted from wild-type, elo1⌬, elo2⌬, elo3⌬, and FAE-KCS-rescued elo1⌬elo2⌬elo3⌬ triple mutant cells, converted to methyl esters, and analyzed by GCMS. C21/23 fatty acids were added to the cells as internal standards prior to extraction of the fatty acids and were used for normalization but are not shown in the traces. The majority of the intracellular fatty acids are saturated and monounsaturated C16 and C18 (not shown in these traces) with the elongated (ϾC18) saturated fatty acids representing only ϳ5% of the total fatty acids (21). In wild-type yeast a high proportion (Ͼ50%) of the C26 is ␣-hydroxylated, also not shown in these traces. The insets identify the ␣-OH-C16, C20:0, and the C20:1 species. B, the relative abundance of the unhydroxylated saturated and monounsaturated C20 -C26 and hydroxylated C16 -C26 fatty acids is presented. C, the indicated strains were grown to an A 600 of 0.5 and were diluted (1/5000) into the wells of a microtiter plate. The cells were transferred to plates containing YPD medium (1% yeast extract, 2% bacto-peptone, 2% bacto-agar, 2% glucose), and the plates were incubated at 26°, 30°, 33°, or 37°C for 3 days prior to photographing.
substrates to provide the saturated VLCFA required for yeast viability (see "Discussion").
Because Elo1p is involved in synthesizing C16 and C18 fatty acids, it was assumed that it was not required for the rescue of the elo2⌬elo3⌬ double mutant by the FAE-KCSs. This was confirmed by demonstrating that an elo1⌬elo2⌬elo3⌬ triple mutant (TDY7006, Table 1) harboring the FAE-KCS expression plasmids grew similarly to the rescued elo2⌬elo3⌬ double mutant (Fig. 3B).
Characterization of in Vivo and in Vitro Activities of the FAE-KCSs-The elongated fatty acids that accumulated in the elo2⌬elo3⌬ mutant rescued by the FAE-KCSs were analyzed by GCMS (Fig. 4A). The segment of the GCMS trace from 8 to 20 min profiling the elongated (ϾC18) and ␣-hydroxylated C16 -C22 fatty acids is shown. Although these are the fatty acids relevant to this study, they represent a minor fraction (Ͻ5%) of the total fatty acids, with C16, C18, and C18:1 comprising the majority of the yeast fatty acids. In Fig. 4B the relative abundance of each of the elongated (ϾC18) and ␣-hydroxylated C16 -C22 fatty acid species is indicated. Consistent with previous characterizations (14,19,23), the predominant elongated saturated fatty acid in the FAE1-KCS rescued mutant is C20. In contrast to wild-type yeast, there are also significant levels of C20:1 and C22:1 in the elo2⌬elo3⌬ mutant rescued by FAE1-KCS. This demonstrates that, unlike Elo2p and Elo3p, FAE1-KCS is able to elongate the C18:1 fatty acid that is present in yeast. The fatty acid profiles of the At5g43760-and At1g04220-rescued elo2⌬elo3⌬ mutants are very similar, with the major VLCFA in these strains being C22. Overall, these rescued mutants have an in vivo fatty acid composition that is very similar to that of the elo3⌬ mutant. The previously reported fatty acid profiles of yeast expressing these two FAE-KCSs (23) are somewhat different from the results reported here. In that study the genes were expressed in wild-type yeast, and thus the fatty acid profiles reflect the combined activities of the endogenous yeast Elo1/2/3 proteins and the heterologously expressed FAE-KCSs (see "Discussion"). The elo2⌬elo3⌬ mutant rescued by At2g16280 FAE-KCS accumulates significant levels of C24 VLCFA, making it the most similar to wild-type yeast.
Although they restored viability to the elo2⌬elo3⌬ mutant, none of the heterologously expressed FAE-KCS genes restored wild-type VLCFA profiles. As mentioned above, the fatty acid profiles of the rescued mutants are similar to the fatty acid profiles observed in some of the yeast elongase mutants. For example, ␣-hydroxylated fatty acids with chain lengths of C16, C18, C20, and C22 accumulate, whereas in wild-type yeast only the ␣-hydroxylated fatty acid is the C26 species (Fig. 4B). From our previous studies (19 -21), it is clear that the shorter ␣-hydroxylated fatty acids arise because phytosphingosine is acylated with shorter fatty acids in the VLCFA-deficient mutants, and the resulting C16 -C22 ceramides are substrates for the Scs7p hydroxylase (30) (see "Discussion").
In addition to the altered VLCFA profiles, the FAE-KCS rescued elo2⌬elo3⌬ mutants display other phenotypes that are characteristic of mutants with deficiencies in VLCFA synthesis. Similar to the elo2⌬ and elo3⌬ mutants, and in contrast to wild-type cells (20,21), the rescued elo2⌬elo3⌬ double mutants have high levels of the free long-chain bases (data not shown). These rescued strains also display temperature sensitivity, another common phenotype of the mutants with reduced VLCFA synthesis (Fig. 4C). The temperature-sensitive growth phenotype is most severe for the FAE1-KCS-rescued elo2⌬elo3⌬ double mutant, which is not able to grow at or above 30°C.
The analysis of the fatty acids that accumulate in the rescued mutants provides information about which of the endogenous yeast fatty acids can be elongated by the FAE-KCSs. To further investigate the activities of the FAE-KCSs, in vitro elongase assays were conducted using microsomes prepared from yeast lacking the endogenous Elo1p/2p/3p proteins and rescued by each of the four FAE-KCS genes. Using a variety of different acyl-CoA acceptors, the substrate specificities of the FAE-KCS proteins were addressed. All four of these HA-tagged FAE-KCSs were expressed well in yeast, and similar amounts of the FAE-KCS proteins were present in the microsomal protein preparations used for these in vitro assays (Fig. 5A). The indicated acyl-CoAs were used at 40 M in the standard elongation assay ("Experimental Procedures"). The elongase reactions were initiated by adding [ 14 C]malonyl-CoA along with NADPH, allowing overall elongation to be measured. The products were extracted and converted to the FAMES as previously described, and the FAMES were resolved by reverse-phase TLC (Fig. 5, B-E). In the reactions without added NADPH, substrates that could be elongated showed the expected 3-keto products (data not shown).
FAE1-KCS elongated the saturated C16-and C18-CoA substrates well, and also elongated C20 -CoA to C22, albeit less well. The monounsaturated C16:1-and C18:1-CoA substrates were efficiently elongated, but FAE1-KCS had little or no activity toward the C22-and C24-CoAs or the C24:1-CoA. It also elongated the C18:2 substrate, forming C20:2 and small amounts of C22:2. The At5g43760 FAE-KCS displayed a strong preference for the C18-and C20-CoA substrates and  ). B-E, fatty acid elongation activity was assayed using microsomes prepared from each of the FAE-KCS-rescued elo1⌬elo2⌬elo3⌬ triple mutants. Microsomes were incubated with the indicated acyl-CoAs in the presence of NADPH, and the reaction products were analyzed by reverse phase TLC as described in Fig. 2. The chain lengths of the elongated products are indicated on the right side of each panel.
The FAE-KCSs Coimmunoprecipitate with Ybr159p and Tsc13p, the Reductases of the Elongase Complex-The ability of the FAE-KCS enzymes to direct VLCFA synthesis in yeast raised questions as to whether this elongation pathway is dependent upon the other characterized components of the yeast Elop-containing elongase system. We have shown that Ybr159p is the major 3-keto reductase (19,20) and that Tsc13p is the trans-2,3-enoyl reductase required for fatty acid elongation (21,28). Furthermore, immunoprecipitation experiments have shown that these reductases, along with the Elop proteins, are organized in a complex. Because the FAE-KCSs are not homologous to the Elops, it was an open question as to whether they could associate with the other elongase proteins. However, the activity of FAE1 in yeast is dependent on Ybr159p (19), arguing that this component of the elongase is responsible for reducing the 3-keto intermediate generated by the FAE-KCSs. In addition, expression of the FAE-KCS genes did not bypass the slow growth phenotype of the ybr159⌬ mutant or the lethality of the tsc13⌬ mutant (data not shown).
Whether the FAE-KCSs physically associate with the reductases was tested directly by immunoprecipitation experiments. Membranes were prepared from cells coexpressing HA-tagged FAE1 with either Myctagged yeast or Myc-tagged Arabidopsis Tsc13p (Fig. 6A). Membranes were also prepared from cells co-expressing HA-tagged FAE1 with either Myc-tagged yeast or Myc-tagged Arabidopsis Ybr159p (Fig. 6B). Following solubilization, the microsomal proteins were immunoprecipitated with either HA or Myc antibodies. Samples of the solubilized membranes (Tot), along with the supernatants (Sup) and pellets (Pell) from the immunoprecipitations were immunoblotted with anti-HA or anti-Myc antibodies. The HA antibodies immunoprecipitated the Myctagged yeast Tsc13p and Myc-tagged Arabidopsis Tsc13p proteins along with the HA-tagged FAE1. Conversely, the Myc antibodies pulled down HA-tagged FAE1 along with the Myc-tagged yeast and Arabidopsis Tsc13p. The HA-tagged FAE1 also coimmunoprecipitated with Myc-tagged Ybr159p from Arabidopsis and to a lesser extent with Myctagged yeast Ybr159p (Fig. 6B). The weak interaction between FAE1 and yeast Ybr159p compared with its interaction with the Arabidopsis Ybr159p homolog is discussed further below ("Discussion"). Coimmunoprecipitation experiments using the HA-tagged At2g16280 FAE-KCS and Myc-tagged yeast and Arabidopsis Tsc13p confirmed that this FAE-KCS also associates with Tsc13p (Fig. 6C). These results demonstrate that the FAE-KCSs physically associate with the Tsc13p trans-2,3-enoyl reductase and the Ybr159p 3-keto reductase of the elongase system.
To determine whether FAE-KCSs and Elops can interact with the same elongase complexes, we also tested whether Elo3p coimmunoprecipitated with HA-FAE1 (Fig. 6D). The HA-FAE1 was expressed in wild-type yeast cells, and solubilized microsomal protein was immunoprecipitated with anti-HA-conjugated Sephadex beads. The failure to see any Elo3p (which migrates as three discrete bands due to differential glycosylation) coimmunoprecipitated with the HA-tagged FAE1 suggests that elongase complexes can form with either an Elop protein or a FAE-KCS protein, but not with both. The failure of Elo3p to coimmunoprecipitate with HA-FAE1 also demonstrates the specificity of the interactions of the FAE-KCSs with the reductases. It is important to note that the level of endogenously expressed Elo3p is higher than the levels of the HA-FAE-KCSs. This is evident from Fig. 5A, which shows that HA-Elo3p (expressed from its own promoter and thus reflecting endogenous Elo3p levels) is more abundant than the HA-tagged FAE-KCSs. The specificity of the association of the FAE-KCSs with the reductases is also underscored by the finding that an HA-tagged Arabidopsis Elop-like protein (At3g06460) failed to coimmunoprecipitate with Myc-tagged Arabidopsis or yeast Tsc13p (Fig. 6E). Finally, HAtagged FAE1 also failed to coimmunoprecipitate with the endoplasmic reticulum-localized Lcb1p protein providing additional evidence for the specificity of the coimmunoprecipitations (data not shown). Taken together, these results strongly suggest that the FAE-KCSs associate specifically with the reductases of the elongase complex.
The VLCFA Synthesized by the Heterologously Expressed FAE-KCSs Are Incorporated into Inositolphosphoceramides-The experiments above demonstrate that the FAE-KCSs interact with the yeast elongase complex and direct the synthesis of sufficient VLCFA to rescue the lethality of an elo2⌬elo3⌬ double mutant. In yeast, the majority of the VLCFAs are found in the sphingolipids, primarily the inositolphosphoceramides (IPCs). The IPCs are synthesized from the ceramides, which are in turn generated by joining the VLCFA in amide linkage with phytosphingosine in a reaction catalyzed by ceramide synthase (Fig.  7A). Because sphingolipids are essential for yeast cell viability, it appears that ceramide synthase is capable of using the VLCFA derived from the FAE-KCSs to acylate phytosphingosine, creating ceramides that can be converted to IPCs. The appearance of the ␣-OH C16 -C22 fatty acids in the rescued mutants provides indirect evidence that the VLCFA produced by the FAE-KCSs are incorporated into ceramides, because Scs7p hydroxylates the fatty acid in ceramides. To directly test whether the FAE-KCS-derived VLCFAs are incorporated into sphingolipids, the FAE-KCS rescued elo1⌬elo2⌬elo3⌬ triple mutant cells were labeled with inositol, and the extracted IPCs were analyzed by thin layer chromatography. As shown in Fig. 7B, each of the FAE-KCS-rescued elo1⌬elo2⌬elo3⌬ triple mutants synthesized IPCs. Whereas the IPC in wild-type yeast contains exclusively ␣-OH C26 fatty acid (labeled as "26" Fig. 7B), mutants lacking Elo3p accumulate C22 IPC-C. The reduced hydrophobicity of the C22 IPC-C relative to the C26 PC-C is reflected in its lower mobility in the thin layer chromatographic analysis. Interestingly, the C22 IPC-C that accumulates in the elo3⌬ mutant is not mannosylated; like the csg mutants that fail to convert IPC to MIPC (31,32), the elo3⌬ mutant makes IPC but no MIPC or M(IP) 2 C (Fig. 7B). With the exception of the strain expressing At2g16280, the IPCs that are synthesized by the FAE-KCS-rescued elo1⌬elo2⌬elo3⌬ triple mutants are relatively hydrophilic and, like the C22 IPC-C that accumulates in the elo3⌬ mutant, they are not mannosylated. The data are consistent with the fatty acid analyses (Figs. 4). For example, in the FAE1-rescued mutant, C20 is the major VLCFA in the cells and C20-IPC-C accumulates. Similarly, in the At5g43760-and At1g04420-rescued mutants, C22 is the predominant fatty acid and C22 IPC-C accumulates. The At2g16280-rescued mutant accumulates C24 and thus C24-IPC-C; unlike the IPCs with shorter fatty acids, this IPC species is mannosylated (note the appearance of MIPC). The spots above the indicated IPC-C species (and below PI) are IPC-B species that lack the ␣-hydroxylation on the fatty acids (30). The appearance of these IPC-B species in the mutants that accumulate the C20 and C-22 IPCs suggests that the fatty acid chain length of the IPCs affects their hydroxylation by Scs7p.

DISCUSSION
The large number of different FAE-KCS isozymes (22 in Arabidopsis) is believed to reflect their activities toward different acyl-CoA substrates as well as the fact that some of the genes display tissue-specific expression. There are no FAE-KCS homologs in the yeast genome. Rather, the Elop family of proteins is believed to catalyze the condensation step of fatty acid elongation in yeast, as well as in other eukaryotes, including mammals. The evidence that the Elop proteins are responsible for condensation relies in part upon the finding that heterologously expressed Elop homologs can direct the synthesis of novel fatty acids in yeast. Because the Elop proteins have none of the hallmarks of the well characterized 3-ketoacyl-CoA synthases, they either represent a completely novel type of condensing enzyme or they could be a regulatory subunit that confers acyl-CoA substrate recognition to an as yet unidentified condensing enzyme. In this regard, it should be noted that the dehy-dratase activity of the elongase system has not been identified, so it is entirely possible that additional elongase proteins will be identified.
We have recently reported that Ybr159p is the major 3-ketoreductase, and that Tsc13p is the trans-2,3-enoyl reductase of the yeast elongase system (19 -21). Furthermore, we found that these enzymes coimmunoprecipitate with one another and with the Elop proteins, a result that is consistent with early biochemical characterizations that suggested that the elongase enzymes are organized in a complex (33). Despite the lack of sequence homology between the Elop and FAE-KCS proteins, the data presented here demonstrate that the FAE-KCSs act in conjunction with the reductases of the elongase complex. The coimmunoprecipitation data strongly suggests that the FAE-KCSs physically associate in a complex with the Ybr159p and Tsc13p reductases. This raises the intriguing possibility that the Elop and FAE-KCS proteins evolved independently as distinct 3-keto-CoA synthases and yet are able to interact with the other components of the elongase complex. Although the significance is unclear, we previously observed that all of the known components of the yeast elongase complex (Elo1p/2p/3p, Ybr159p, and Tsc13p) have a high pI (Ͼ 10), and we note that the FAE-KCSs also have high pIs (Ͼ9.2). The coimmunoprecipitation results indicate that FAE1 interacts more strongly with the Arabidopsis Ybr159p homolog than with yeast Ybr159p. Although it is tempting to speculate that this reflects differences between the Ybr159p proteins that affect their association with FAE1, there is currently no information as to which proteins of the elongase complex are in direct contact.
In addition to the large FAE-KCS gene family, there are also several Elop homologs in Arabidopsis, the roles of which have not been defined. On the other hand, there is a single homolog of Tsc13p, the trans-2,3enoyl reductase in Arabidopsis (encoded by At3g55360 that functionally complements a yeast tsc13⌬ mutant (28). Kunst and coworkers (34) recently reported the characterization of an Arabidopsis mutant lacking the Tsc13p homolog, referred to in that study as ECR. The mutant displayed reduced synthesis of VLCFA, and decreased cuticular wax, storage lipid, and VLCFA-containing sphingolipids. Although the plants displayed morphological abnormalities that could be specifically attributed to the sphingolipid deficiency, the plants were viable and accumulated some VLCFA, indicating that there is at least one other reductase that can reduce the trans-2,3-CoA intermediates in Arabidopsis (34). It is unclear whether the residual fatty acid elongation that is occurring in the mutant plants is dependent on the FAE-KCSs, or whether it might be mediated by the Arabidopsis Elop homologs.
There are two genes that encode 3-ketoacyl-CoA reductases with homology to Ybr159p in both Arabidopsis (At1g67730 and At1g24470) and maize (gl8a and gl8b). The gl8a and gl8b genes of maize are 97% identical to one another, and both are required for normal accumulation of cuticular waxes. Double mutant kernels accumulate substantially reduced levels of VLCFAs and ceramides and are inviable suggesting that sphingolipids or other VLCFA-containing compounds are essential for normal embryo development (35). It is important to point out that the Arabidopsis reductase genes are able to substitute for their yeast counterparts (20,21). Because yeast have no FAE-KCSs, these plant reductases clearly interface with the Elop proteins to restore fatty acid elongation in the yeast mutants. The role of the Arabidopsis ELOS has yet to be established, but it is likely that they function analogously to their yeast homologs and that the plant 3-keto and trans-2,3-enoyl reductases act with both FAES and ELOS in plants.
In this study four Arabidopsis FAE-KCSs that are able to substitute for the yeast Elo2p/Elo3p proteins by directing the synthesis of saturated VLCFA is reported. It is likely that additional members of the Arabidopsis FAE-KCSs will also functionally complement the elo2⌬elo3⌬ double mutant, but that they were not recovered, because they are either not present or are poorly represented in the Arabidopsis yeast expression library. For example, despite the fact that FAE1 does complement and that we were able to PCR amplify it from the expression library, it was not found in the complementation screening. The characterization of the complementing FAE-KCSs confirms that yeast require saturated VLCFA with a minimal chain length of C20 for viability. As mentioned under "Results," five additional FAE-KCSs (At2g26250, At1g01120, At2g26640, At1g19440, and At1g25450) were PCR-amplified from the library and expressed in yeast, but were unable to rescue the elo2⌬elo3⌬ double mutant. Thus, they apparently cannot elongate the available substrates (mainly C16 in the elo2⌬elo3⌬ mutant) to produce the required VLCFA. It will be interesting to characterize their substrate specificities using the in vitro elongase assays.
The heterologous expression of several FAE-KCS genes in wild-type yeast was recently reported (23), including several that were also characterized in this study. Three of the six that were found to be enzymatically active in that study (FAE1, At5g43760, and At1g04220) were among those that were found to be capable of complementing the elo2⌬elo3⌬ double mutant. Expression of At1g01120 was reported to result in the accumulation of C22, -24, and -26 VLCFA and expression of At1g25450 in the accumulation of C26, -28, and -30 (23), but we found that neither was able to rescue the elo2⌬elo3⌬ double mutant. This may reflect the difference in expressing the genes in wild-type yeast instead of in the elo2⌬elo3⌬ double mutant. For example, if these FAE-KCSs elongate fatty acids with chain lengths greater than C18, these substrates would have been available (produced by Elo2p/Elo3p) in a wild-type background but not in the elo2⌬elo3⌬ double mutant background. Previous studies of yeast expressing the At1g01120 (KCS1) gene suggested that it does elongate C18, -20, -22, and -24 fatty acids (36). It is possible that the coexpression of some of these noncomplementing FAE-KCSs together with FAE1 in the elo2⌬elo3⌬ mutant will result in the ability to grow at 37°C, a nonpermissive temperature for the FAE1rescued mutant (Fig. 4B). For example, an FAE-KCS that elongates C20 substrates, along with FAE1 that elongates the C16 produced by de novo fatty acid synthesis to C20, would restore synthesis of VLCFA ϾC22 and thereby would likely restore growth at 37°C.