Candida tropicalis Expresses Two Mitochondrial 2-Enoyl Thioester Reductases That Are Able to Form Both Homodimers and Heterodimers*

Here we report on the cloning of a Candida tropicalis gene, ETR2, that is closely related to ETR1. Both genes encode enzymatically active 2-enoyl thioester reductases involved in mitochondrial synthesis of fatty acids (fatty acid synthesis type II) and respiratory competence. The 5′- and 3′-flanking (coding) regions of ETR2 and ETR1 are about 90% (97%) identical, indicating that the genes have evolved via gene duplication. The gene products differ in three amino acid residues: Ile67 (Val), Ala92 (Thr), and Lys251 (Arg) in Etr2p (Etr1p). Quantitative PCR analysis and reverse transcriptase-PCR indicated that both genes were expressed about equally in fermenting and ETR1 predominantly respiring yeast cells. Like the situation with ETR1, expression of ETR2 in respiration-deficient Saccharomyces cerevisiae mutant cells devoid of Ybr026p/Etr1p was able to restore growth on glycerol. Triclosan that is used as an antibacterial agent against fatty acid synthesis type II 2-enoyl thioester reductases inhibited growth of FabI overexpressing mutant yeast cells but was not able to inhibit respiratory growth of the ETR2- or ETR1-complemented mutant yeast cells. Resolving of crystal structures obtained via Etr2p and Etr1p co-crystallization indicated that all possible dimer variants occur in the same asymmetric unit, suggesting that similar dimer formation also takes place in vivo.

of a single gene or copies of transcripts arising from closely related but separate genes occur frequently (3)(4)(5). Consequently, these enzymes are present as an array of isoforms, each of which can be targeted to different subcellular locations and may serve either identical or different metabolic functions. This genetic redundancy might be used as a back-up system to sustain central cellular needs (6,7). For example, peroxisomal acyl-coenzyme A (CoA) 1 oxidase (Pox) is a key enzyme of ␤-oxidation that is present as multiple forms in C. tropicalis (1,8,9). At least three POX genes are present in the C. tropicalis genome. The closely related POX4 and POX5 have been shown to be differentially expressed depending on what lipids are available for growth (1).
Recently, we reported on the identification of a novel mitochondrial 2-enoyl thioester reductase Etr1p from C. tropicalis (10) and also demonstrated that a homologous Saccharomyces cerevisiae protein Ybr026p involved in mitochondrial respiratory function (11) exhibits the same activity. Disruption of the corresponding nuclear gene in S. cerevisiae results in a respiratory-deficient strain unable to grow on nonfermentable carbon sources (11). S. cerevisiae Etr1p was suggested to link the assembly of the respiratory complexes with prokaryotic type fatty acid synthesis (type II) in fungal mitochondria (10,(12)(13)(14)(15)(16).
In the present study we describe a further C. tropicalis mitochondrial 2-enoyl thioester reductase, Etr2p, that is closely related to Etr1p and compare the function of the two enzymes. Quantitative PCR was used to determine the expression pattern of the two reductase genes in C. tropicalis. The ability of ETR2 to rescue the respiratory-deficient phenotype of the ybr026c⌬ strain from S. cerevisiae was also studied. Crystallization experiments on a mixture of purified Etr2p and Etr1p was undertaken to see whether homo-and heterodimeric variants were possible. The results are discussed in terms of genetic redundancy in lipid metabolism and its significance for the maintaining mitochondrial respiratory function.

EXPERIMENTAL PROCEDURES
Cloning of ETR2 and ETR1-The primers used in this study are described in Table I. The CTGEN1 genomic fragment revealing the sequence for ETR1 and other genomic fragments, among them CT-PVU25 and CTHAE10 (see Fig. 1A), were obtained by PCR, ligationmediated PCR, and screening of a C. tropicalis genomic DNA library as described (10). The sequences of the partially overlapping genomic fragments (CTPVU25, CT55PCR, CTPVUB10, and CTHAE10; see Fig. * This work was supported by the European Community Access to Research Infrastructure action of the Improving Human Potential Program and by grants from the Sigrid Jusélius Foundation and the Academy of Finland. 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. The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  Pulsed Field Gel Electrophoresis-Preparation of C. tropicalis chromosomal DNA and pulsed field gel electrophoresis were carried out according to the Bio-Rad instruction manual for CHEF-DR pulsed field electrophoresis systems (Rev. B, Bulletin # 9086), available on the Bio-Rad Website (www.Bio-Rad.com) and the protocol for the preparation of agarose-embedded yeast DNA. The correct amount of cells required for the procedure was estimated by spinning down the cells to obtain a wet pellet corresponding to a volume of 50 l (17). Chromosomes were separated on a 0.8% agarose gel in TAE (40 mM Tris, 20 mM acetate, and 2 mM EDTA, pH 8.5), for 24 h at 150 V and 200 s of switch time at 13°C, followed by 48 h at 100 V and 700 s of switch time at 10°C. After electrophoresis, the DNA was blotted onto a Hybond-Nϩ nylon transfer membrane (Amersham Biosciences) following the Bio-Rad instruction manual procedure for Megabase blotting. The 32 Plabeled probe used to detect ETR2 and ETR1 was generated using the Roche Random primed DNA labeling kit (Roche Applied Science). The template for probe generation was a 553-bp PstI-SacI fragment from the ETR1 coding region. Hybridization was carried out overnight at 65°C as described (18).
Analysis of Gene Expression by 5Ј-RACE and Quantitative PCR-C. tropicalis pK233 cells (American Type Culture Collection, Rockville, MD) propagated overnight in YPD (1% yeast extract, 2% peptone, and 2% D-glucose) were transferred at an A 600 of 0.1 to YP medium containing 2% glucose or 0.2% oleic acid as a sole carbon source and grown for 7 h at 30°C. The cells were frozen in liquid nitrogen and stored at Ϫ70°C until required. RNA was extracted with a QuickPrep total RNA extraction kit (Amersham Biosciences). The PCR amplification of the 5Ј-RACE fragments included the primers CTRED2R (Table I) corresponding to a sequence of the reductase and AP1 corresponding to the ligated adaptor (Marathon cDNA amplification kit, Clontech Laboratories, Inc., Palo Alto, CA). PCR fragments were ligated into pUC18, transformed into Escherichia coli DH5␣ cells, and screened with 32 Plabeled CTRED2R. The positive colonies were isolated and sequenced.
Moloney murine leukemia virus reverse transcriptase and random hexamer primers (MBI Fermentas, St. Leon-Rot, Germany) were used for the first strand cDNA synthesis. The synthesis product was used for the PCR performed with ABI PRISM® 7700 sequence detection system (Applied Biosystems, Foster City, CA) using TaqMan chemistry and a pair of sense and antisense primers specific to ETR2 or ETR1. The products of the quantitative PCR were detected using bifunctional fluorogenic probes ETR2fluoro and ETR1fluoro, respectively. The results were normalized to 18 S RNA quantified from the same samples. The probe 18Sfluoro was used for the 18 S amplicon detection. The p values were determined as described (19).
Protein Overexpression and Purification-pYE352::ETR2 was generated by site-directed mutagenesis (Stratagene, La Jolla, CA) using pYE352::ETR1 (10) as a template DNA and three pairs of different forward and reverse primers in stepwise PCR amplifications. The resulting pYE352::ETR2 was used to transform the BJ1991-based S. cerevisiae ybr026c⌬ strain for Etr2p overexpression (10,20). Etr2p was purified in a manner similar to that described for Etr1p (10). Soluble protein extracts were analyzed on SDS-PAGE (21) and on a Jasco J715 spectropolarimeter (Great Dunmow, Essex, UK). Reductase activities were determined as described (10), in addition to which, the reaction mixture was supplemented with 50 g/ml of essentially fatty acid free albumin (Sigma-Aldrich).
Mass Spectrometry-Purified Etr2p or Etr1p at a concentration of 5 mg/ml was diluted 1:40 in 40% acetonitrile, 0.1% trifluoroacetic acid and mixed with an equal volume of ␣-cyano-4-hydroxy cinnamic acid matrix, prepared as saturated solution in 40% acetonitrile, 0.1% trifluoroacetic acid (v/v). Mass measurements were carried out on an Applied Biosystems Voyager-DE TM STR Biospectrometry work station operating in linear mode. The mass spectra were calibrated with bovine serum albumin as an external standard.
Crystallization of Etr2p and Etr1p, Data Collection, and Structure Determination-Crystals containing the dimers Etr1p-Etr1p, Etr1p-Etr2p, and Etr2p-Etr2p were obtained with the hanging drop method (22) using 20 mg/ml mixture of Etr1p:Etr2p in a 1:1 ratio and 1.8 M (NH 4 ) 2 SO 4 , 0.1 M N-(2-acetamido)-2-iminodiacetic acid/NaOH (pH 7.0) as a precipitant. Crystallization drops were also supplemented with 5 mM NADPH and 1 mM octanoyl-CoA. The data for the crystal structure were collected at the beam line I711 of the MAX-lab synchrotron (Lund, Sweden). The structure was solved with molecular replacement using the program AMoRe (CCP4 suite). Analysis of Etr1p apoenzyme crystal structure without bound NADPH has been reported previously (23) and was used as a starting model in the calculations. Refinement of the structure was initiated by one cycle of simulated annealing using CNS (24) and continued with Refmac_5.1.12 (CCP4 suite) employing TLS (25). Waters were added by Arp/Warp (26) in cycles of 600 molecules followed by successive refinement runs and rebuilding in the program O (27). The ligand could be reliably built into the structure only after all waters were assigned at the final stage of the refinement, when only small gaps were present in the ligand density and with clear density for the phosphates and the ring structures. The crystal characteristics and refinement statistics are shown in Table II. The geometries of the structures were analyzed with the programs O, WHAT IF (28), and PROCHECK (29). All of the structure drawings were created with Swiss-PDBViewer (30) and edited with Adobe Photoshop 7.0.
Respiratory Growth Analysis-To examine growth on nonferment- Vic-CCTGGTGGTGCCCTTCCGTCA-Tamra able carbon sources, S. cerevisiae BJ1991 ybr026c⌬ cells or those transformed with plasmid DNA for overexpression of Ybr026p, enoyl-ACP reductase (FabI) from E. coli, catalase (Cta1p), Etr2p, or Etr1p (10) were cultured on synthetic complete medium supplemented with 3% glycerol. Glycerol medium plus 0.125, 0.5, or 1.0 g/ml of triclosan was prepared to examine the effect of the drug on the reductase function and viability of respiring yeast cells. After autoclaving the culture medium, triclosan was added from a 10 mg/ml (ethanol) stock solution. Triclosan was a gift from Louis Widmer (Dermatologica Widmer, Helsinki, Finland).

C. tropicalis Genomic DNA Contains a Novel Gene ETR2
Related to ETR1-Sequencing of a C. tropicalis genomic fragment (CTGEN2; see "Experimental Procedures") revealed an open reading frame (ORF) of 1158 nucleotides, encoding a polypeptide of 386 amino acid residues that was termed Etr2p (Fig. 1A). CTGEN2 included 958 and 828 bp of the 5Ј-and 3Ј-flanking sequences from ETR2, respectively. The nucleotide sequences of the ETR2 and ETR1 (10) were compared and found to be 96.8% identical. Differences between Etr2p and Etr1p were seen in three residues: Ile 67 in Etr2p is replaced by Val, Ala 92 is replaced by Thr, and Lys 251 is replaced by Arg in Etr1p. The ETR2 nucleotide sequence and the respective product Etr2p in comparison with Etr1p is shown in Fig. 2. The flanking regions of ETR2 and ETR1 revealed identities of 90.2% (664 bp) and 89.2% (814 bp) on the 5Ј and 3Ј sequences, respectively.
Southern blot analysis of genomic DNA from C. tropicalis using a 32 P-labeled fragment consisting of ETR1 (CTGEN1; see "Experimental Procedures") as a probe yielded two hybridizing fragments in each of the genomic DNA samples, which had been digested and/or double-digested with EcoRI, HindIII, and BamHI, as shown in Fig. 1B. Because the probe did not contain any of these restriction sites, the data suggested that CTGEN2 and CTGEN1 are at different loci in the genome. Subsequently, pulsed field gel electrophoresis was applied to separate chromosomal DNA followed by Southern blotting. As shown in Fig.  1C, eight chromosomal bands were separated on the gel. When compared with previously reported data, the chromosomal pattern we obtained correlated well with results of Kamiryo et al. (31) but differed slightly from the data reported by Doi et al. (32). Our probe hybridized with bands VI and VII (Fig. 1C), indicating that ETR2 and ETR1 reside on separate chromosomes in the genome of C. tropicalis.
Both ETR2 and ETR1 Are Expressed in C. tropicalis under Various Growth Conditions-5Ј-RACE was carried out with total RNAs of 2% glucose and 0.2% oleic acid-grown C. tropicalis cells. Twenty five (twenty) 5Ј-RACE clones were obtained from glucose (oleic acid) grown yeast cells, the longest amplified fragments extending to the Ϫ33 (Ϫ19) position in the 5Ј-untranslated region. The 5Ј-RACE analysis showed that thirteen (six) amplified fragments presented ETR2 and twelve (fourteen) fragments presented ETR1 among the 5Ј-RACE clones from glucose (oleic acid)-grown yeast cells.
To quantify ETR2 and ETR1 expression on different media quantitative PCR was carried out. Total RNA was extracted from cells grown on media containing glucose or oleic acid as sole carbon source. RNA was reverse transcribed to cDNA and amplified by PCR for quantification with ETR2-and ETR1specific primers (see "Experimental Procedures"). Relative values for the amount of ETR2 RNA were 1.00 on glucose and 2.17 on oleic acid grown cells and for the amount of ETR1 RNA 1.31 on glucose and 2.95 on oleic acid, indicating that both genes were more strongly expressed on oleic acid than on glucose as a carbon source (Table III).
The ETR2 Gene Product Encodes an Etr1p-like 2-Enoyl Thioester Reductase with a Preference for Short and Medium Chain over Long Chain Enoyl Thioesters-To analyze the function of Etr2p as a 2-enoyl thioester reductase and to characterize its substrate specificity, it was overexpressed in S. cerevisiae and chromatographically purified to apparent homogeneity. As a comparison, Etr1p was also examined. When the purified proteins were analyzed by circular dichroism spectroscopy, the overall spectra of Etr2p and Etr1p were identical, signifying preservation of the secondary structure elements. Mass spectrometry analysis gave molecular masses which corresponded to theoretical values of 39.34 and 39.39 kDa for Etr2p and Etr1p, respectively, excluding the mitochondrial targeting signals, not present in the mature proteins (10). The mass difference was attributed to the difference in three amino acid residues deduced from cDNA sequence (Fig. 2). The specific reductase activities were determined using trans-2-hexenoyl-CoA (C6), trans-2-decenoyl-CoA (C10), and trans-2-hexadecenoyl-CoA (C16) as substrates. The respective values obtained for Etr2p and Etr1p were 17.3 and 18.0 mol/min/mg of protein with trans-2-hexenoyl-CoA, 13.6 and 11.7 mol/min/mg of protein with trans-2-decenoyl-CoA, and 4.3 and 5.8 mol/min/mg of protein with trans-2-hexadecenoyl-CoA. Hence, both proteins showed a similar preference toward short and medium chain 2-enoyl thioesters. Gas chromatographic analysis of the reaction end products demonstrated that, like Etr1p (10), also Etr2p was able to carry out the reduction of the trans-2 double bond both in monounsaturated trans-2-and conjugated trans-2,trans-4-enoyl thioesters. Similarly to Etr1p, Etr2p was specific for NADPH and could not utilize NADH as hydrogen donor in the catalysis.
Etr2p-Etr1p Heterodimerization Enables Generation of Various Native Forms of the Enzyme-The present results indicated that C. tropicalis simultaneously expresses two highly similar enoyl thioester reductases. Because previous data using purified Etr1p indicated that it functions as a dimer (23), this raised the issue of whether the two reductases may form a heterodimer. Purified Etr2p and Etr1p were mixed, and crystals were grown that diffracted at 1.98 Å resolution. Determination of the crystal structure allowed the identification of the locations of Ile 67 (Val), Ala 92 (Thr), and Lys 251 (Arg) in Etr2p (Etr1p). Detection of these different amino acids in the asymmetric unit in the crystal revealed the existence of all possible dimer variants, Etr1p-Etr1p, Etr1p-Etr2p, and Etr2p-Etr2p (Fig. 3, A and B). Fragmented electron density for the NADPH co-factor could be seen in three of six polypeptide chains in the asymmetric unit. Each type of dimer was liganded with one NADPH, and in the heterodimer Etr1p binds the co-factor in this crystal (Fig. 3A). Thus, the structure contains both Etr2p and Etr1p in liganded and unliganded form. Electron density for the octanoyl-CoA, present in the crystallization mixture, was not observed.
When comparing the apo-Etr1p component in the heterodimeric crystal to the structures of Etr1p with and without the bound ligand (Ref. 23; 1GUF and 1GU7 in the Protein Data Bank, respectively), it is noted that apo-structures show identical conformation and the conformational change from the apo-to holoenzyme form upon binding of NADPH (Fig. 3, C and  D) is similar for the two proteins. It is also worth noting that despite this dynamics in the molecule, the apo-and holoenzymes are sufficiently similar as crystallizable species such that they can be incorporated into the same crystal lattice. The three amino acids distinguishing Etr2p and Etr1p were neither involved in crystal contacts nor restricted dimer formation. Of the differing amino acids Ala 92 (Thr) and Lys 251 (Arg) are located on the protein surface facing the solvent; the former in the middle of a large loop between helix ␣2 and strand ␤4 and the latter in the beginning of helix ␣E (the nomenclature of secondary structures according to Airenne et al. (23)). Ile 67 (Val) at the end of strand ␤3, on the other hand, is located deep inside the nucleotide-and substrate-binding pocket but spatially separated from the catalytic Tyr 79 by an intervening helix ␣2 (Fig. 3B). Thus, the differing residues are not in the near vicinity of the active site.
Etr2p Is Able to Restore Respiration to the ybr026c⌬ Strain from S. cerevisiae, a Function Not Restricted by the Anti-bacterial Agent Triclosan-Etr2p was overexpressed in the respiratory-deficient ybr026c⌬ strain from S. cerevisiae. Immunoelectron microscopy with anti-Etr1p antibodies indicated mitochondrial localization of Etr2p (data not shown) similar to Etr1p (10). Like the situation with mitochondrial Etr1p, overexpression of Etr2p was able to restore growth of the deletion strain on glycerol in the complementation experiment (Fig. 4). Etr1p and Etr2p represent members of the medium chain dehydrogenase/reductase superfamily of proteins (10), structurally different from prokaryotic reductases belonging to the short chain dehydrogenase/reductase superfamily (23,(33)(34)(35)(36). Triclosan is an anti-bacterial agent used in toothpaste, soaps, and other consumer products. Its action is transmitted via its ability to inhibit fatty acid synthesis type II 2-enoyl thioester reductases either by forming noncovalent, high affinity ternary complexes with the enzyme and NAD(P) ϩ , as with FabI from E. coli, or by reversibly inhibiting the enzyme, as with FabL from Bacillus subtilis (37)(38)(39). Introduction of triclosan into the culture plates at concentrations that are inhibitory against bacteria (37, 38) did not prevent the growth of the ybr026c⌬ cells transformed with plasmids encoding Etr1p, Etr2p, and Ybr026p. However, the ybr026c⌬ strain expressing a mitochondrially targeted FabI (10) showed retarded growth on the triclosan containing medium in a concentration-dependent manner (Fig. 4). Triclosan was not inhibitory for the growth of yeast cells on rich glucose (data not shown). DISCUSSION This work reports on the characterization of a novel gene ETR2, whose product functions as a mitochondrial 2-enoyl thioester reductase. In the process of cloning ETR1 (10), genomic fragments were isolated that contained an ORF (ETR2) very similar to that of ETR1. The ORFs were 96.8% identical at the nucleotide level, encoding proteins differing in three amino acid residues only. Differences in the 5Ј and 3Ј FIG. 1. ETR2 and ETR1 in the genome of C. tropicalis. A, organization of the genomic fragments of ETR1 and ETR2. CTGEN1, and CTPVU25, CTPVUB10 and CTHAE10 were obtained by ligation-medi-ated PCR. CTGEN2 was obtained by PCR amplification from genomic DNA. The positions of the oligonucleotides CTRED18 and CTRED19 are indicated by arrowheads, and ORFs are shown as boxes. CTGEN2 included an additional ORF (ORF3) ending 202 bp upstream from the ORF-ETR2 start codon. CTGEN1 also contained another ORF (ORF2) downstream from ORF-ETR1, encoding a putative polypeptide of 163 amino acids with a similarity of 66% to the N-terminal 134 amino acids of allantoinase from S. cerevisiae (accession number S48489). The triplets for initiation (ATG) and stop codons (TAG or TGA) are indicated. B, ETR genes as revealed by Southern blot analysis. Yeast genomic DNA (10 g) was digested with EcoRI (1), HindIII (2), BamHI (3), EcoRIϩHindIII (4), EcoRIϩBamHI (5), or HindIIIϩBamHI (6) and loaded onto an agarose gel. After electrophoretic fractionation and blotting, the nitrocellulose filter was hybridized with 32 P-labeled CT-GEN2. The sizes (kb) of marker fragments are indicated on the left. C, chromosomal localization of ETR1 and ETR2 by pulsed field gel electrophoresis. C. tropicalis and Hansenula wingei (DNA size marker) chromosomal DNA were separated on 0.8% TEAE for 24 h at 150 V and 200 s of switch time at 13°C, followed by 48 h at 100 V and 700 s of switch time at 10°C (ethidium bromide stain is shown on the right). A total of eight chromosomal bands (I-VIII) were observed. Separated DNA was probed with a 553-bp ETR1 PstI/SacI fragment of ETR1 (blot shown on the left). The sizes in megabase pairs of the observed H. wingei chromosomal DNA fragments (3.13, 2.70, 2.35, 1.81, 1.66, 1.37, and 1.05) agreed with those reported previously (3.5, 3.1, 2.8, 2.6, 2.3,  2.1, 1.38, and 1.02 and 2.8, 2.6, 2.4, 2.2, 2.1, 1.3, 1.2, and 1.0 (31, 32), respectively).
regions of the two ORFs were more pronounced (90.2 and 89.2%, respectively). Purified Etr2p contained enoyl-CoA reductase activity, and expression of Etr2p in the respirationdeficient S. cerevisiae ybr026c⌬ strain (11) was able to restore growth to the disruption strain on glycerol. These data demonstrated that C. tropicalis contained at least two functional 2-enoyl thioester reductases.
The phenomenon of genetic redundancy and the presence of different protein isoforms is widespread. In addition to being a fuel for evolvement of protein families, the presence of more than one functional copy of a gene can be seen as advantageous, if genes perform more than one specific function or are functioning under different conditions (6). Redundancy is common for all species and can be found even in the most simple genomes such as that of the parasitic Mycoplasma genitalium and extending to more complex multicellular eukaryotic genomes that may rely on redundant genes during development (6).
Although the random sequenced tags (RSTs) available at the Genolévures data base, the data bank of RSTs of a group of hemiascomycetous yeasts, at present cover only about 20% of the genome of C. tropicalis, genomic exploration of sequences as well as combining the data in the literature demonstrate  (n ϭ 8). The PCR amplification of the 5Ј-RACE fragments were carried with total RNA isolated from cells grown with either glucose or oleic acid. The amplified fragments were transformed in E. coli, and the frequencies of positive colonies are given. that a number of fatty acid metabolism (Table IV) enzymes encoded by single genes in S. cerevisiae have more than one copy of the corresponding genes in C. tropicalis (7,40). The data base search and the literature revealed that ⌬ 3,5 -⌬ 2,4 -dienoyl-CoA isomerases, ⌬ 3 -⌬ 2 -enoyl-CoA isomerases, multifunctional enzymes type-2, acetoacetyl-CoA thiolases, carnitine acetyltransferases, and acyl-CoA oxidase (1,8,(41)(42)(43) together with Etr1p and Etr2p, belong to this category (Table III). Because the percentage of coverage of the C. tropicalis genome is still small and many of genes from S. cerevisiae wait for an identification of their orthologs in C. tropicalis, the number of genes in the group presenting genetic redundancy is likely to increase in the future.
Comparison of the amount of RNA between glucose-and oleic acid-grown cells indicated that both ETR2 and ETR1 were simultaneously present under both conditions. Both gene products were present in equal amounts in cells grown on glucose, but their expression levels are increased upon a shift of cells from fermentable (glucose) to nonfermentable (oleic acid) carbon source in a way that ETR1 was the more predominantly expressed form. Aerobic organisms have evolved sensory systems to monitor oxygen availability, and, for instance in S. cerevisiae, there is a subset of genes of which transcription involves sensing oxygen and hypoxia (44). However, no changes were shown in the expression of ETR2 and ETR1 when cells were shifted to grow from respiration to anoxia (data not shown). The increased expression of both ETR2 and ETR1 on nonfermentable carbon source, oleic acid, suggests that the gene products are linked to cellular respiration.
Assays for the enzymatic activity of Etr2p and Etr1p indicated that both proteins were capable of catalyzing with almost identical specific activities the reduction of the double bonds in a series of trans-2-enoyl thioesters with different chain lengths. In this respect the properties of Etr1p and Etr2p are unique among the acyl-CoA metabolizing enzymes. Often the members of this group show preference for certain chain length of fatty acids, although overall substrate preference overlaps. Typical examples are mammalian acyl-CoA dehydrogenases and ␤-keto-acyl-CoA thiolases (45). The similar kinetic properties of Etr2p and Etr1p are also supported by the structural data; the The respiration-dependent growth of different yeast transformants was examined on glycerol with or without the addition of the drug triclosan (0.125, 0.5, and 1.0 g/ml) in the growth medium. The strains examined are BJ1991 ybr026c⌬ (1), BJ1991 ybr026c⌬ transformed with pYE352::MRF1 (2), pYE352::ETR1 (3), pYE352::ETR2 (4), pYE352::FABI (5), or pYE352::CTA1 (6). The plates were incubated for 5 days at 30°C.
of the structure of Etr2p with the residues differentiating it from Etr1p indicated. The C␣ trace of the ␤-strands (blue), ␣-helices (red), and other polypeptide chains (gray) as well as NADPH are also shown. C, Etr1p with a bound NADPH superimposed on Etr2p with a bound NADPH. The conformation of polypeptide chains and co-factors are virtually identical (residual mean and standard deviation on all backbone atoms 0.81 Å). D, Etr2p with a bound NADPH (red) superimposed on Etr2p without NADPH (green). Only C␣ atoms are shown. Conformational change induced by co-factor binding, as seen earlier for Etr1p (23) is apparent in the co-factor binding Rossmann fold domain (residual mean standard deviation on all backbone atoms 1.14 Å).
architectures in the vicinity of the active sites in the both proteins were similar, and the amino acid residue differences Ile 67 (Val), Ala 92 (Thr), and Lys 251 (Arg) in Etr2p (Etr1p) were positioned in the structure in a way that their interference on the catalysis is unlikely (Fig. 3B). Even if the preferences of the Etr2p and Etr1p were toward short and medium chain 2-enoyl thioester substrates, they showed also activities toward long chain substrates, giving credence to the idea of mitochondrial fatty acid synthesis being required for generation of also other chain lengths of fatty acids than octanoic acid, indicated as a precursor in the synthesis of lipoic acid in yeast and plant mitochondria (46 -48).
Generation of crystals containing the three possible dimeric variants of Etr2p and Etr1p indicates that the protein homodimers dissociate and reassociate in the solution containing a mixture of the proteins. Because both ETR2 and ETR1 are transcribed simultaneously and encode enzymatically active reductases, we suggest that the three dimer variants are also found under physiological conditions. In this respect, the situation would be similar to tissue-specific tetrameric lactate dehydrogenase or to dimeric creatine kinase isoforms in mammals (49 -51).
Respiratory growth experiments with triclosan showed that the drug can enter mitochondria from the culture medium. The inhibition of the respiratory growth of ybr026c⌬ cells comple-mented with mitochondrially targeted FabI supports the view that enzymatically active 2-enoyl thioester reductase is essential for the maintenance of respiratory-competent mitochondria in yeast. Failure to inhibit the respiratory growth that depended on Etr1p, Etr2p, or Ybr026p suggested that triclosan does not inhibit reductases of the medium chain dehydrogenases/reductases protein family at the concentration effective against the short chain dehydrogenases/reductase type of bacterial enoyl thioester reductases.
Close similarity of ETR2 and ETR1 indicates the genes have arisen via gene duplication recently in the evolution. The higher identity of the nucleotide in the coding region than of in the 5Ј-and 3Ј-flanking regions of the genes indicates the importance of both functional and structural integrity of the proteins in C. tropicalis. The allowance of mutations at the promoter regions of genes is reflected by the different ratios of the ETR2 and ETR1 expression levels under fermentable and nonfermentable growth conditions. This suggests that at their promoter regions, binding of transcription controlling factors, which remain to be identify in future, have changed.