Peroxisomal Δ3-cis-Δ2-trans-Enoyl-CoA Isomerase Encoded by ECI1 Is Required for Growth of the Yeast Saccharomyces cerevisiae on Unsaturated Fatty Acids*

We have identified the Saccharomyces cerevisiae gene ECI1 encoding Δ3-cis-Δ2-trans-enoyl-CoA isomerase that acts as an auxiliary enzyme in the β-oxidation of (poly)unsaturated fatty acids. A mutant devoid of Eci1p was unable to grow on media containing unsaturated fatty acids such as oleic acid but was proficient for growth when a saturated fatty acid such as palmitic acid was the sole carbon source. Levels of ECI1transcript were elevated in cells grown on oleic acid medium due to the presence in the ECI1 promoter of an oleate response element that bound the transcription factors Pip2p and Oaf1p. Eci1p was heterologously expressed in Escherichia coli and purified to homogeneity. It was found to be a hexameric protein with a subunit of molecular mass 32,000 Da that converted 3-hexenoyl-CoA totrans-2-hexenoyl-CoA. Eci1p is the only known member of the hydratase/isomerase protein family with isomerase and/or 2-enoyl-CoA hydratase 1 activities that does not contain a conserved glutamate at its active site. Using a green fluorescent protein fusion, Eci1p was shown to be located in peroxisomes of wild-type yeast cells. Rat peroxisomal multifunctional enzyme type I containing Δ3-cis-Δ2-trans-enoyl-CoA isomerase activity was expressed in ECI1-deleted yeast cells, and this restored growth on oleic acid.

We have identified the Saccharomyces cerevisiae gene ECI1 encoding ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase that acts as an auxiliary enzyme in the ␤-oxidation of (poly)unsaturated fatty acids. A mutant devoid of Eci1p was unable to grow on media containing unsaturated fatty acids such as oleic acid but was proficient for growth when a saturated fatty acid such as palmitic acid was the sole carbon source. Levels of ECI1 transcript were elevated in cells grown on oleic acid medium due to the presence in the ECI1 promoter of an oleate response element that bound the transcription factors Pip2p and Oaf1p. Eci1p was heterologously expressed in Escherichia coli and purified to homogeneity. It was found to be a hexameric protein with a subunit of molecular mass 32,000 Da that converted 3-hexenoyl-CoA to trans-2-hexenoyl-CoA. Eci1p is the only known member of the hydratase/isomerase protein family with isomerase and/or 2-enoyl-CoA hydratase 1 activities that does not contain a conserved glutamate at its active site. Using a green fluorescent protein fusion, Eci1p was shown to be located in peroxisomes of wild-type yeast cells. Rat peroxisomal multifunctional enzyme type I containing ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity was expressed in ECI1-deleted yeast cells, and this restored growth on oleic acid.
S. cerevisiae is an excellent model organism for examining whether there is an in vivo requirement for ⌬ 3 -cis-⌬ 2 -transenoyl-CoA isomerase, since this lower eukaryote is capable of growing on unsaturated fatty acids (20). ⌬ 3 -cis-⌬ 2 -trans-Enoyl-CoA isomerase activity has previously been reported in S. cerevisiae; however, the gene encoding it was not identified (21). In this study, we report the identification of the S. cerevisiae ECI1 gene encoding peroxisomal ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase. Expression of the gene was up-regulated under fatty acid medium conditions via a promoter sequence termed the oleate response element (ORE; Refs. 22 and 23). Eci1p was purified chromatographically, and its subcellular location was determined using fluorescent microscopy. We have constructed a yeast strain devoid of this enzyme activity and examined the * This work was supported in part by Fonds zur Förderung der wissenschaftlichen Forschung (FWF) (Vienna, Austria) Grants P10604 and P12061 (to B. H.), Jubilä umsfonds der Ö sterreichischen Nationalbank, Austria, Grant 6517 (to H. R.), and grants from the Sigrid Juselius Foundation, Finland, and the Academy of Finland (to J. K. H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ ability of the null mutant to grow on various fatty acids. It was additionally used to examine the physiological role of rat peroxisomal MFE type I previously reported to contain ⌬ 3 -cis-⌬ 2trans-enoyl-CoA isomerase activity in vitro (10). The results are discussed in terms of the function of conserved amino acid residues within the active site of members of the low homology hydratase/isomerase family of proteins.

EXPERIMENTAL PROCEDURES
Strains and Plasmids-The S. cerevisiae strains and plasmids used are listed in Table I. Escherichia coli strains DH1OB, HB101, and DH5␣ were used for all plasmid amplifications and isolations.
Yeast Media and Growth Conditions-For RNA isolations, logarithmic cultures of the strain BJ1991 were grown for 16 h in YP medium (1% (w/v) yeast extract, 2% (w/v) bactopeptone) containing 2% (w/v) D-glucose, 2% (v/v) ethanol, or 0.2% (w/v) oleic acid and 0.02% (w/v) Tween 80 (adjusted to pH 7.0 with NaOH) to an A 600 ϭ 1.0. Plates used to assess utilization of fatty acids were prepared by pouring a thin layer of medium at 55°C that consisted of 0.67% (w/v) yeast nitrogen base with amino acids, 0.1% (w/v) yeast extract, 0.5% (w/v) potassium phosphate at pH 6.0, and 2% (w/v) agar. The medium was autoclaved with 0.125% (w/v) palmitic acid (cis-C 16:0 ), oleic acid, cis-12-octadecenoic acid, or arachidonic acid (cis-C 20:4 (5,8,11,14) ) and with 0.5% (w/v) Tween 80 to solubilize the fatty acid used. In oleic acid liquid medium used for the same purpose, the agar was omitted. Growth assays on solid media were performed by propagating cells overnight in liquid YPD (YP with 2% (w/v) D-glucose) medium at 30°C with shaking at 170 rpm. Following a further 5 h in fresh YPD, cells from logarithmic cultures were serially diluted, and aliquots of 2 l were spotted on plates. For complementation of eci1⌬ cells, transformants overexpressing Eci1p, rat peroxisomal MFE type I, or those harboring the vector were grown as above in liquid dropout media consisting of 0.67% yeast nitrogen base, 2% (w/v) D-glucose, and amino acids (20 mg ϫ l Ϫ1 ) as required, and spotted on fatty acid plates. For enzyme assays, yAG826 and yAG827 cells from stationary phase cultures were grown for 18 h in medium containing oleic acid as described for RNA isolation that additionally contained 0.05% (w/v) D-glucose and 75 g ϫ ml Ϫ1 ampicillin. YPD-G418 plates used to select for yeast transformants carrying the kanMX4 deletion  The numbers in superscript following the strains' designations refer to their parental genotypes, e.g. BJ1991eci1⌬ 1 was derived from 1) BJ1991.
cassette contained 200 g ϫ ml Ϫ1 G418 (Geneticin) and 2% w/v agar (24). For fluorescence microscopy, strains yAG872 and yAG873 expressing GFP-Eci1p were grown for 7 h to an A 600 ϭ 1.0 in YP medium containing 2% (v/v) ethanol. Plasmid Constructions-All oligonucleotides used for constructing plasmids and verifying knockout strains are listed in Table II. The plasmid pSK::YLR284c containing ECI1 was constructed by inserting the XbaI-and PstI-digested polymerase chain reaction (PCR) product generated with oligonucleotides H0390 and H0392 using genomic yeast DNA as template into a similarly digested pBluescript ® SK(ϩ) vector (Stratagene, La Jolla, CA). The yeast shuttle vector overexpressing ECI1 (pAG766) was generated by ligating the amplified product of oligonucleotides YLR284C-A1 and YLR284C-A4 to a SmaI-digested YEplac181 (25). For electrophoretic mobility shift assays (EMSAs), the annealed SalI-delimited oligonucleotides defining the ORE in the promoter of ECI1 (YLR284-ORE1 and -ORE2) were ligated to a SalIdigested pBluescript ® SK(ϩ) vector to yield pAG832.
For subcellular localization, Eci1p was fused to the carboxyl terminus of green fluorescent protein (GFP) as follows. Oligonucleotides NOTI-284C-F and HINDIII-284C-R were used with Taq DNA polymerase and chromosomal DNA as template in a PCR to amplify a NotIand HindIII-delimited ECI1 product that was ligated to pGEM ® -T, resulting in pGEM::ECI1. Following NotI and HindIII digestion, the amplification product was ligated to a similarly digested YIplac204based pADH2-OAF1 containing a NotI site (26) from which OAF1 had been removed, yielding pADH2-NOT1-ECI1. Following NotI digestion and dephosphorylation, a NotI-delimited GFP gene (27) was inserted in frame, resulting in pADH2-GFP-ECI1. The EcoRV-linearized plasmid was integrated into the trp1 locus of BJ1991 wild-type and BJ1991pex6⌬ strains, and integrants were designated yAG872 and yAG873, respectively.
To express Eci1p in E. coli, ECI1 was amplified with Pfu high fidelity DNA polymerase (Stratagene) using oligonucleotides YLR-FWD and YLR-REV and genomic DNA as template. The amplification product was ligated to a SmaI-digested pUC18 (SureClone™ Ligation Kit, Amersham Pharmacia Biotech, Uppsala, Sweden), resulting in pUC18::ECI1. The nucleotide sequence of the open reading frame in pUC18::ECI1 was determined, and the derived amino acid sequence was found to be identical to that in the data bases, with the exception that Met-25 had changed to Ile. An NdeI-BamHI fragment of pUC18::ECI1 containing ECI1 was inserted into an appropriately digested pET3a (Novagen, Madison, WI), resulting in pET3a::ECI1.
Expression of Eci1p-Heterologous expression of Eci1p was performed by propagating BL21(DE3)pLysS E. coli cells harboring pET3a::ECI1 according to the manufacturer's instructions. Induction was carried out for 2.5 h at 30°C. Cells (1.5 g, wet weight) were collected by centrifugation, washed once in phosphate-buffered saline (10 mM sodium phosphate, 2 mM potassium phosphate, 140 mM NaCl, 3 mM KCl, 5 mM ␤-mercaptoethanol, pH 7.4), suspended in 15 ml of lysis buffer (20 mM potassium phosphate buffer at pH 7.6 and 100 mM KCl), and stored at Ϫ70°C until needed. Following thawing, the lysed cells were incubated with RNase (2 g ϫ ml Ϫ1 ), DNase (20 g ϫ ml Ϫ1 ), and lysozyme (100 g ϫ ml Ϫ1 ) in the presence of 0.1 mM phenylmethylsulfonyl fluoride at 35°C for 30 min. EDTA (3 mM) was added to the cell lysate, and cell debris was removed by centrifugation at 30,000 ϫ g for 45 min at 4°C.
Purification of Eci1p-A volume of 15 ml of the supernatant described above was diluted in 2 volumes of 20 mM potassium phosphate buffer at pH 7.6 and applied to a DEAE-Sephacel column (2.5 ϫ 9.5 cm; Amersham Pharmacia Biotech) in equilibrium with the same buffer that additionally consisted of 30 mM KCl, 1 mM EDTA, 1 mM EGTA, and 0.5 mM benzamide hydrochloride. The column was washed with 120 ml of the same buffer, and bound ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity was eluted with a 50-ml linear gradient of 0 -0.25 M potassium chloride. Following their adjustment to pH 7.2, the pooled DEAE-Sephacel fractions containing isomerase activity were applied to a 1-ml Resource S column (Amersham Pharmacia Biotech) that was equilibrated with a buffer that consisted of 20 mM potassium phosphate (pH 7.2), 1 mM EDTA, 1 mM EGTA, and 0.5 mM benzamide hydrochloride. After washing of the column with 5 ml of the same buffer, the isomerase activity was eluted with a 25-ml linear gradient of 0 -0.35 M KCl. The purification of Eci1p was completed by applying a sample containing 0.92 mol ϫ min Ϫ1 of isomerase activity to a Superdex 200 HR 10/30 column (Pharmacia Biotech) in equilibrium with a buffer consisting of 200 mM potassium phosphate (pH 7.2) and 3 mM EDTA.
Disruption of ECI1-Construction of ECI1-disrupted strains was performed according to the European Functional Analysis Network guidelines for the short flanking homology method, based on published protocols (24). The product obtained following PCR performed on pFA6a-kanMX4 template DNA using the 58-and 59-mer oligonucleotides YLR284C-S1 and YLR284C-S2 was used to transform S. cerevisiae strains BJ1991 and NKY857. Geneticin-resistant transformants that grew on YPD-G418 plates were verified for correct integration of the eci1::kanMX4 disruption fragment by PCR performed on genomic DNA using the oligonucleotide pairs YLR284C-A1/K2 and K3/YLR284C-A4 (24).
Enzyme Assays-Soluble protein extracts were prepared by breaking oleic acid-induced cells (approximately 100 -400 mg, wet weight) in 400 l of a modified, ice-cold, glass bead disruption buffer (28) that consisted of 20 mM Tris-HCl (pH 7.9), 10 mM MgCl 2 , 1 mM EDTA, 5% (w/v) glycerol, 0.3 M ammonium sulfate, 0.1 mM phenylmethylsulfonyl fluoride, 0.1 mM benzamide hydrochloride, and 300 l of acid-washed glass beads. Cell debris was removed by centrifugation for 10 min. The assay for ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase was conducted according to published methods (10) using 60 M trans-3-hexenoyl-CoA as substrate. The reaction was monitored by following the generation of a

5Ј-TCGACTTCGGAATAATATTTCTTATGCCCGTCG-3Ј
This study SPS19ORE1 a S1 and S2 extend from the 3Ј-ends of the respective oligonucleotides that include them in their designation, e.g. S1 extends from YLR284C-S1.
Fluorescence Microscopy-Fluorescence of cells was recorded using a Zeiss Axioskop fluorescence microscope (Jena, Germany) as described (27). Ethanol-derepressed yeast cells were fixed by adding formaldehyde to a final concentration of 3.7% (v/v) and shaken for an additional 30 min. The fixed cells were washed twice with phosphate-buffered saline and were stained for DNA with 4Ј,6-diamidino-2-phenylindole at a final concentration of 2 g ϫ ml Ϫ1 . A volume of 2 l of fixed cells was spotted on glass slides for microscopic examination.
Miscellaneous-The following procedures were performed according to published methods: nucleic acid manipulations, formaldehyde gel electrophoresis, blotting and hybridization (30), DNA fragment isolation (31), yeast transformation (32,33), yeast RNA preparation for Northern analysis (34), and determination of protein concentration (35). The DNA fragments consisting of ECI1 (1.5 kb; XbaI-PstI), SPS19 (189 base pairs; ClaI-SphI), POX1 (2.4 kb; BglII-BglII), or ACT1 (3.5 kb; BamHI-EcoRI) that were used to probe the Northern blot were obtained from pSK::YLR284c, pAG113 (36), pAD17 (2), and pYA301 (37), respectively. The [␣-32 P]dATP-labeled probes were generated with a random primer labeling kit (Prime-a-Gene; Promega, Madison, WI) according to the manufacturer's instructions. Molecular sizing of native Eci1p was performed by applying the dynamic light scattering method with a DynaPro-MSTC Temperature-Controlled MicroSampler (Protein Solutions, Charlottesville, VA) according to the manufacturer's instructions. SDS-polyacrylamide gel electrophoresis was performed as described (38). For predicting the secondary structure of Eci1p, the related sequences were collected from data banks using the BLAST algorithm and were aligned using the CLUSTAL W program package. Prediction of the secondary structure was done using the PHD prediction server and combined with the three-dimensional structure of 2-enoyl-CoA hydratase 1 (39,40).

Identification of a Novel S. cerevisiae Open Reading Frame
Encoding a Protein with Similarities to Other ␤-Oxidation Enzymes-A search of the data bases for novel genes that contained in their promoters sequences resembling OREs with the consensus CGGN [15][16][17][18]23) and that ended with nucleotides similar to those coding for a peroxisomal targeting signal type 1 (Ref. 41) revealed a homologous pair of open reading frames YLR284c (Fig. 2) and YOR180c that shared a moderate degree of amino acid sequence similarity to other hydratase/ isomerase proteins. In the course of this work, the two genes appeared in searches conducted by other groups following a similar strategy and were reserved in the Saccharomyces Genome Data Base under designations reflecting homology to known enoyl-CoA hydratases type 1. 2 As a consequence of findings presented here, YLR284c was named ECI1. YOR180c is the subject of current investigation and is not addressed further in this study.  (Eci1p), ECIHUMAN, and ECIRAT represent S. cerevisiae, human, and rat ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase, ECHM_RAT; rat mitochondrial 2-enoyl-CoA hydratase 1. The sequences of ECI1, ECIHUMAN, ECIRAT, and ECHM_RAT were aligned using the CLUSTAL X program and were refined and shaded using the GeneDoc program, which allows the use of a gray scale to visualize the residue similarity over the aligned sequences. The darkest tone indicates the highest degree of similarity. The secondary structure elements for ECHM_RAT were taken from the published three-dimensional structure. e and h indicate the location of the ␤-strands and ␣-helices, respectively. The predicted ␤-strands (E) and ␣-helices (H) for ECI1 are marked above the amino acid sequence. The number of amino acid residues, taking the initial methionines as the first residue, are given on the right. The polar amino acid (Glu-164 of ECHM_RAT and Glu-165 of ECIRAT) conserved in enzymes with hydratase or isomerase activities is indicated by an arrow. Tyr-148 is indicated by a filled circle.
an ORE-S. cerevisiae genes regulated by OREs are repressed in rich glucose medium, derepressed in media containing nonfermentable carbon sources such as ethanol, and are induced by the ORE-binding transcription factors Pip2p and Oaf1p (29,43) when cells are grown in a medium containing oleic acid (20,29). A DNA fragment consisting of ECI1 was hybridized with immobilized RNA that was obtained from cells grown in medium conditions supporting either repression, derepression, or induction (Fig. 3A). A strong signal was detected in the lane containing RNA from cells grown in oleic acid, and hence, the transcriptional profile of ECI1 was similar to that of the oleic acid-inducible genes POX1, encoding acyl-CoA oxidase (2), and SPS19, the yeast 2,4-dienoyl-CoA reductase (36). The observation that, like the transcripts of POX1 and SPS19, those of ECI1 did not accumulate in a strain devoid of Oaf1p and Pip2p indicated that up-regulation of ECI1 may be directed by the ORE-like sequence identified in the data base search. To determine whether this element is bound by Oaf1p and Pip2p, an EMSA was performed.
The putative ECI1 ORE was assayed for possible protein-DNA interactions with Pip2p and Oaf1p using crude extracts from wild-type, pip2⌬, and oaf1⌬ yeast cells propagated in oleic acid medium (Fig. 3B). The retarded complex formed with ECI1 ORE (indicated by an arrow) using protein extracts from wildtype cells was absent from the pip2⌬ or oaf1⌬ strains (26,29). In addition to self competition, the signal of the retarded complex was reduced by excess unlabeled oligonucleotides containing the OREs of SPS19, POX1, and CTA1 but not by a mutant version of the ORE of the latter gene (CTA1mut; Ref. 29). Lower and higher mobility complexes appearing in the vicinity of the indicated one were competed by ECI1 ORE and, to varying degrees, by the remaining OREs. However, since they also appeared in the lanes containing extracts from the pip2⌬ or oaf1⌬ mutants, their significance is not clear. From the Northern and EMSA analyses, it is concluded that the induction of ECI1 under oleic acid medium conditions was mediated by Pip2p and Oaf1p interacting with the ORE in the gene's promoter. The fatty acid induction of ECI1 due to this ORE indicated that Eci1p may be involved in the breakdown of fatty acids. This could be examined by constructing a null mutant and determining its growth on a range of carbon sources.
An eci1⌬ Yeast Strain Is Defective for Growth on Unsaturated Fatty Acids-According to current understanding (Fig. 1), ⌬ 3cis-⌬ 2 -trans-enoyl-CoA isomerase is required for degrading fatty acids with double bonds at both odd-and even-numbered positions in the carbon chain but is dispensable for breaking down saturated fatty acids. To elucidate the potential role of Eci1p in the metabolism of unsaturated fatty acids, wild-type and eci1⌬ strains were grown on plates containing palmitic, cis-12-octadecenoic, oleic, or arachidonic acid as the sole carbon source. In these solid media, an emulsion is formed by adding Tween 80, which also acts as a poor carbon source, and therefore cells could grow on these plates, but zones of clearing indicate utilization of the additional fatty acid substrate. A pox1⌬ strain was added to the plates to demonstrate lack of clear-zone formation on the fatty acids tested, and an sps19⌬ disruptant acted as a control for utilization of unsaturated fatty acids with even-numbered double bonds (cis-12-octadecenoic acid and arachidonic acid with the cis-double bond at the ⌬ 8 -position). The ECI1-deleted strain was able to form a clearing zone when grown on palmitic acid but its growth was impaired on all three unsaturated fatty acids (Fig. 4). The

FIG. 3. ECI1 is induced in oleic acid via an ORE.
A, carbon source-dependent transcriptional activation of ECI1 was examined by probing a Northern blot containing RNA from a wild-type BJ1991 strain and a pip2⌬oaf1⌬ mutant grown in the indicated media with labeled ECI1, SPS19, POX1, and the constitutively transcribed ACT1. The weak signal normally observed with RNA from derepressed cells was not obtained during the routine exposures used for recording the oleic acid signal. B, formation of a protein-DNA complex with ECI1 ORE is mediated by Pip2p and Oaf1p. Labeled ECI1 ORE was incubated with crude extracts from oleic acid-induced cells (lanes 1-3), and bound DNA was resolved from free probe using a 4% polyacrylamide gel for 1.5 h. The retarded complex representing the Pip2p-Oaf1p heterodimer is indicated by an arrow. For competition of the heterodimer seen in lane 4, a 25-fold excess of unlabeled double-stranded oligonucleotides comprising the OREs of ECI1, SPS19, POX1, and CTA1 and CTA1mut was added to the reaction mixture prior to the wild-type crude extracts (lanes 5-9).
FIG. 4. An ECI1-deleted strain fails to utilize unsaturated fatty acids. Formation of clearing zones by the mutant strain BJ1991eci1⌬ on palmitic acid was compared with that on solid media containing oleic acid, cis-12-octadecenoic acid, or arachidonic acid. An sps19⌬ strain devoid of 2,4-dienoyl-CoA reductase (yAG141) and one disrupted at the locus for acyl-CoA oxidase (BJ1991pox1⌬) were applied to the plates in order to demonstrate failure to specifically utilize cis-12-octadecenoic acid or any fatty acid, respectively. moderate clearing zones that had developed in the oleic acid medium did not indicate wild-type utilization levels, since subsequent liquid assays demonstrated that the eci1⌬ strain grew slower on this fatty acid (33% of the wild-type growth level on oleic acid compared with 91% on palmitic acid following 62 h). This indicated that Eci1p may be involved in degrading double bonds at both odd-and even-numbered positions and could possess an activity that was germane with a ⌬ 3 -cis-⌬ 2 -transenoyl-CoA isomerase. ECI1 Encodes ⌬ 3 -cis-⌬ 2 -trans-Enoyl-CoA Isomerase-Extracts from oleic acid-induced wild-type cells were examined for isomerase activity; however, this was below the detection limit of the assay used (Ͻ1 nmol ϫ min Ϫ1 ϫ ml Ϫ1 sample). Eci1p was therefore overexpressed in the eci1⌬ mutant background from the multicopy plasmid pAG766 that contains the ECI1 gene under the control of the native promoter (yAG826), and this resulted in an isomerase activity of 60 nmol ϫ min Ϫ1 ϫ mg Ϫ1 protein. Extract from similarly propagated mutant cells harboring the plasmid vector (yAG827) did not contain detectable levels of ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity (Fig. 5A). The ability of strain yAG826 (eci1⌬ [ECI1]) to grow on arachidonic acid is shown in Fig. 5B.
Purification and Characterization of Eci1p-To demonstrate that the yeast ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity was solely due to Eci1p, it was expressed in bacterial cells. Isomerase activity was chromatographically purified using DEAE-Sephacel (anion exchanger), Resource S (cation exchanger), and Superdex 200 HR 10/30 (size exclusion) columns (Fig. 6A). Analysis of SDS-polyacrylamide gels following electrophoresis revealed a single protein band with an apparent molecular mass of 32,000 Da (Fig. 6B), and this indicated that the protein was purified to apparent homogeneity. The polypeptide size obtained in this way agreed with the molecular mass of 31,700 Da calculated from the deduced amino acid sequence. Eci1p eluted from a Superdex column at the same volume as rat ⌬ 3,5 -⌬ 2,4 -dienoyl-CoA isomerase, which is a hexameric hydratase/isomerase protein with a native molecular mass of 170,000 Da (44). This value agrees with that obtained by dynamic light scattering analysis, which yielded only one signal peak that corresponded to an approximate molecular mass of 151,000 Da. In the final preparation, an overall purification of about 30-fold was obtained with a specific activity of 11.2 mol ϫ min Ϫ1 ϫ mg Ϫ1 protein (Table III). The isomerase had a K m value of 21.5 M for trans-3-hexenoyl-CoA.
Eci1p Is Located in Peroxisomes-Eci1p does not contain an obvious N-terminal peroxisomal targeting signal type 2 sequence (45) but instead terminates with a peroxisomal targeting signal type 1-like tripeptide HRL (41, 46) that does not fully conform to the consensus sequence, (S/A/C)(K/H/R)(L/M). Hence, the protein's location in cells was examined by tagging it with a fluorescent moiety. GFP has been amply used before both for demonstrating the subcellular location of proteins and for marking peroxisomes (47,48), and therefore an N-terminal GFP fusion with Eci1p was expressed from the ADH2 promoter. The GFP moiety of the fusion protein was monitored using fluorescent microscopy in S. cerevisiae wild-type and pex6⌬ cells (49), which lack detectable peroxisomes that were grown under ethanol-medium conditions supporting ADH2 ac-FIG. 5. Eci1p contains ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity. A, demonstration of ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity for Eci1p. Crude extracts from oleic acid-induced strains were reacted with 60 M trans-3-hexenoyl-CoA as described under "Experimental Procedures." The addition of protein from eci1⌬ strain yAG827 harboring the YEplac181 vector (arrow 1) did not result in the formation of a Mg 2ϩ 3-ketohexanoyl-CoA complex; however, protein from yAG826 overexpressing Eci1p from plasmid pAG766 (arrow 2) yielded a specific ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity. B, ⌬ 3cis-⌬ 2 -trans-enoyl-CoA isomerase activity encoded by pAG766 was sufficient to restore the growth of the eci1⌬ strain to wild-type levels on arachidonic acid, as demonstrated by the formation of a clearing zone in the medium. The wild-type and mutant strains containing the plasmidborne ECI1 gene or the vector were yAG162, yAG826 (eci1⌬ [ECI1]), and yAG827 (eci1⌬ [Vector]), respectively. tivation (Fig. 7). GFP-Eci1p was clearly detected as punctate fluorescence in wild-type cells, indicating that it was compartmentalized. However, in pex6⌬ cells, which grow normally under carbon-source derepression conditions, this fluorescence was diffuse due to the cytoplasmic location of the fusion protein. To exclude compartmentalization of GFP-Eci1p in mitochondria, both strains were stained for DNA using 4Ј,6-diamidino-2-phenylindole, and since the pex6⌬ mutants showed a mitochondrial staining pattern similar to that seen in the wildtype cells, it was concluded that the GFP punctation was solely due to the presence of peroxisomes.
Heterologous Complementation of the eci1⌬ Mutant Growth Phenotype Using a Rat Peroxisomal ⌬ 3 -cis-⌬ 2 -trans-Enoyl-CoA Isomerase-S. cerevisiae cells devoid of ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity could serve as a tool for examining the in vivo function of previously characterized mammalian isomerases. The rat peroxisomal MFE type I contains 2-enoyl-CoA hydratase 1, L-specific 3-hydroxyacyl-CoA dehydrogenase, and ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activities (50). The in vivo action of the former two activities has previously been shown using a fox2 mutant of S. cerevisiae (51); however, that of the ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity hitherto could not be examined. Mutant cells devoid of Eci1p were supplied with rat peroxisomal MFE type I, and this restored their ability to grow on oleic acid (Fig. 8), demonstrating for the first time that the in vitro ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity previously measured in this protein (50) was functional in vivo.

DISCUSSION
Here we describe the identification and properties of the product of the S. cerevisiae ECI1 gene encoding ⌬ 3 -cis-⌬ 2 -transenoyl-CoA isomerase and show that it is required for the breakdown of fatty acids with double bonds at both odd-and evennumbered positions. This is the first monofunctional peroxisomal isomerase identified at the molecular level. The previously characterized mammalian peroxisomal isomerases are an integral part of the peroxisomal multifunctional isomerase/hydratase/dehydrogenase enzyme (10). Strains disrupted at ECI1 were unable to grow on solid media containing arachidonic acid, oleic acid, or cis-12-octadecenoic acid but were unaffected on palmitic acid. This is in agreement with current understanding of ␤-oxidation of fatty acids (6,7).
For the degradation of unsaturated fatty acids with double bonds at odd-numbered positions, there is evidence in higher eukaryotes for two ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase-requiring pathways. In the direct route, ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase is sufficient to move double bonds to the position appropriate for the hydratase reaction, whereas in the alternative reduction pathway (52,53), an additional ⌬ 3,5 -⌬ 2,4dienoyl-CoA isomerase acts in conjunction with 2,4-dienoyl-CoA reductase in order to achieve this (Fig. 1). It is not yet clear whether a reduction pathway actually exists in fungi. Preliminary investigations into the alternative pathway in yeast using a mutant defective in NADPH-dependent 2,4-dienoyl-CoA reductase (36) revealed that the sps19⌬ strain was proficient for growth on solid medium containing oleic acid. Similarly, a mutant defective in regenerating NADP ϩ (idp3⌬) required for this process was also shown to be able to grow on oleic acid plates, although in liquid medium it was less efficient at utilizing this carbon source (21).
All of the known S. cerevisiae genes encoding peroxisomally targeted ␤-oxidation enzymes, including POX1, FOX2, POT1/ FOX3, and SPS19, are regulated via OREs (2-5, 22, 23, 36, 54). ECI1 was identified through a primary data base search for novel genes that could be inducible by oleic acid. The transcriptional profile of ECI1 was shown to be similar to that of other ORE-containing genes, and this depended on the presence of Pip2p and Oaf1p. The complete ECI1 ORE was found to contain one half-site conforming to the consensus 5Ј-CG-GNNNTNA-3Ј (22, 23) that bound the transcription factors Oaf1p and Pip2p (26,29,43,55). The genomic organization of the ECI1 (YLR284c) locus resembles that of the ORE-containing SPS18/19 pair (36), since the divergent gene YLR285w is not up-regulated in oleic acid medium (data not shown). Although a comprehensive analysis of the ECI1/YLR285w intergenic region has not yet been completed, it would be reasonable to assume that there occurs an overlap of the ECI1 ORE by elements important for the regulation of YLR285w. Such elements could give rise to the complex protein-DNA interaction pattern seen in Fig. 5.
The deduced amino acid sequence of Eci1p is shown here to be moderately homologous (25-27% identity) to other members of the hydratase/isomerase family (56). One common structural feature of the hydratase/isomerase proteins revealed by x-ray crystallography (2-enoyl-CoA hydratase 1, 4-chlorobenzoyl-CoA dehalogenase, and ⌬ 3,5 -⌬ 2,4 -dienoyl-CoA isomerase) is that for substrate binding and catalysis their subunits have a spiral core domain composed of four repetitive right-handed turns each consisting of two ␤strands and an ␣-helix (39,57,58). The predicted secondary structure elements of Eci1p are highly conserved when compared with those of 2-enoyl-CoA hydratase 1 (Fig. 2). This indicates that the subunits of Eci1p may contain core domains (amino acid residues 20 -169) that are similar to those of the other members of the hydratase/isomerase protein family. The predicted helical structures that correspond to trimerization domains 1 (amino acid residues 171-179 and 186 -191 in Eci1p) and 2 (residues 225-237 and 244 -252), and the connective loop (residues 204 -218; Ref. 39) are also conserved, indicating that Eci1p could form a trimer. In line with this idea, size exclusion chromatography and dynamic light scattering gave for Eci1p a native molecular mass of 170,000 Da and 151,000 Da, respectively. When taking into account the polypeptide size (32,000 Da), this indicates that in the native state Eci1p is an oligomer, which we propose is a hexamer similar to rat 2-enoyl-CoA hydratase 1 and ⌬ 3,5 -⌬ 2,4 -dienoyl-CoA isomerase that are dimers of two trimers (39,58).
Amino acid sequence alignments of previously characterized hydratases and isomerases have shown that glutamate at the position equivalent to 164 in 2-enoyl-CoA hydratase 1 is conserved in all of them (18). Furthermore, site-directed mutagenesis studies have indicated that this glutamate participates in catalysis in both hydratase and isomerase reactions by transferring the proton at the C-2 carbon of the substrates (56,59). Crystallographic studies of 2-enoyl-CoA hydratase 1 subsequently confirmed that Glu-164 is in the active site, acting as a catalytic acid in the hydration reaction (39,40). Surprisingly, the amino acid sequence alignment revealed that Glu-164 of 2-enoyl-CoA hydratase 1 was replaced by Phe-150 in Eci1p (Fig. 2, arrow). This phenylalanine was confirmed by sequencing the nucleotides of the plasmid-borne gene used to express and purify active ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase. The only protic amino acid residue in Eci1p that could be aligned to this region is Tyr-148 (Fig. 2, filled circle), which corresponds to Gln-162 in rat 2-enoyl-CoA hydratase 1. Since the side chain of this glutamine has previously been shown to face the activesite pocket where it is hydrogen-bonded to both Glu-144 and Glu-164 (Refs. 39 and 40), it is likely that the side chain at Tyr-148 in Eci1p also points to the active site where it may be in the immediate vicinity of substrate atoms that are susceptible to catalysis. Therefore, this tyrosine could potentially act as a catalytic amino acid residue.
Another insight gained from comparing Eci1p with hydratase/isomerase proteins is the importance of various amino acid residues in predicting whether members of this family are either hydratases or isomerases. Leu-130 of Eci1p is conserved in human and rat monofunctional mitochondrial isomerases ( Fig. 2) and also in rat peroxisomal MFE type I (50). On the other hand, the corresponding amino acid residue in rat 2-enoyl-CoA hydratase 1 is Glu-144, which, according to x-ray crystallographic data, could act as the prerequisite catalytic base in the hydratase reaction (39,40). It is also worth noting that the K m value of Eci1p is in the micromolar range similar to isomerases from bovine liver (32 M) and from rat liver (37 M; Refs. 14 and 56). The V max value of 11.2 mol ϫ min Ϫ1 ϫ mg Ϫ1 protein is in the order of magnitude of that determined for rat mitochondrial monofunctional ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase (35-40 mol ϫ min Ϫ1 ϫ mg Ϫ1 protein; Ref. 10). Thus, despite the low amino acid sequence similarity, the kinetic properties of Eci1p are comparable with those of its monofunctional mammalian counterparts.
The yeast ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase mutant generated here represents the first organism devoid of this activity and has served to underpin the exact point of entry into ␤-oxidation of unsaturated fatty acids in a eukaryote. It confirmed the previous observation that the end product of yeast 2,4dienoyl-CoA reductase (Sps19p), like that of its higher eukaryote counterpart, is 3-enoyl-CoA, since ECI1-deleted cells could not convert it to trans-2-enoyl-CoA (the substrate for Fox2p) and, therefore, could not grow on cis-12-octadecenoic acid. Finally, the eci1⌬ strain provided proof that the ⌬ 3 -cis-⌬ 2 -trans-enoyl-CoA isomerase activity contained in rat peroxisomal MFE type I was functional in vivo, since it was demonstrated that growth on unsaturated fatty acids could be restored using this heterologous enzyme.