Changing stereochemistry for a metabolic pathway in vivo. Experiments with the peroxisomal beta-oxidation in yeast.

The biosphere is inherently built of chiral molecules, and once their metabolism is established, the stereochemical course of the reactions involved is seen to remain highly conserved. However, by replacing the yeast peroxisomal multifunctional enzyme (MFE), which catalyzes the second and third reactions of beta-oxidation of fatty acids via D-3-hydroxyacyl-CoA intermediates, with rat peroxisomal MFE, which catalyzes the same reactions via L-3-hydroxy intermediates, it was possible to change the chiralities of the intermediates in a major metabolic pathway in vivo. Both stereochemical alternatives allowed the yeast cells to grow on oleic acid, implying that when the beta-oxidation pathways evolved, the overall function was the determining factor for the acquisition of MFEs and not the stereospecificities of the reactions themselves.

The biosphere is inherently built of chiral molecules, and once their metabolism is established, the stereochemical course of the reactions involved is seen to remain highly conserved. However, by replacing the yeast peroxisomal multifunctional enzyme (MFE), which catalyzes the second and third reactions of ␤-oxidation of fatty acids via D-3-hydroxyacyl-CoA intermediates, with rat peroxisomal MFE, which catalyzes the same reactions via L-3-hydroxy intermediates, it was possible to change the chiralities of the intermediates in a major metabolic pathway in vivo. Both stereochemical alternatives allowed the yeast cells to grow on oleic acid, implying that when the ␤-oxidation pathways evolved, the overall function was the determining factor for the acquisition of MFEs and not the stereospecificities of the reactions themselves.
Fatty acids are universal constituents of living cells, and they are used as essential components of biomembranes and as a store for combustion energy (1). Their degradative pathway (␤-oxidation) was found to be confined to peroxisomes in all eukaryotic cells (2)(3)(4) and additionally to mitochondria in animal cells (5,6). Until recently, the second and third reactions of ␤-oxidation, which are catalyzed by multifunctional enzymes (MFEs) 1 in extramitochondrial systems (7)(8)(9) and with very long-chain substrates also in mitochondria (10), were assumed to proceed in all organisms via L-3-hydroxyacyl-CoA esters when degrading fatty acids (11)(12)(13) and via D-3-hydroxy-intermediates in the de novo synthesis of fatty acids (14 -17). However, the observation that ␤-oxidation in the peroxisomes of Saccharomyces cerevisiae occurs via D-3-hydroxy metabolites (18) opened up a new perspective on the evolution of ␤-oxidation systems (Fig. 1). The peroxisomal MFEs of higher (mammals and plants) and lower eukaryotes (fungi) not only catalyze two reactions of opposite chiral specificity, but they also have different native molecular sizes and distinct amino acid sequences, indicating different phylogenetic origins (18,19).
Using yeast cells carrying a deletion of the MFE gene (YFOX-2) (18), we studied whether the stereochemistry plays a role in the evolution of ␤-oxidation systems. As fungi possess only a peroxisomal ␤-oxidation (20,21), this yeast mutant does not degrade fatty acids and is thus unable to grow on fatty acids as a carbon source. We replaced the yeast endogenous MFE, comprising 2-enoyl-CoA hydratase 2 and D-specific 3-hydroxyacyl-CoA dehydrogenase activities (18), with its peroxisomal counterpart in the rat, possessing enoyl-CoA hydratase 1 and L-specific 3-hydroxyacyl-CoA dehydrogenase activities (7,22). This replacement resulted in functional complementation giving the YFOX-2 cells back their capability to grow on oleic acid.

EXPERIMENTAL PROCEDURES
Strains and Media-The S. cerevisiae strains and plasmids used are listed in Table I. Escherichia coli strain DH5␣ was used for all transformations and plasmid isolations.
Plasmid Construction-Into a YEp352 vector containing the catalase A sequence from S. cerevisiae (CTA1) (23) and modified as described by Kragler et al. (24), an additional unique restriction site was introduced by cutting with HindIII, dephosphorylating, and ligating again in the presence of a polynucleotide kinase-treated oligonucleotide representing the XhoI site (5Ј-AGCTTCTCGAGA-3Ј). The resulting plasmid pYE352-CTA1 contains the whole CTA1 gene with modifications allowing the replacement of the open reading frame by cutting with XbaI or SacI and HindIII or XhoI ( Fig. 2A).
The coding region of cDNA for rat MFE (25) was obtained from total RNA isolated from clofibrate-treated rat liver by reverse transcription with moloney murine leukemia virus reverse transcriptase and amplification by polymerase chain reaction using rat MFE-specific primers and cloned into pUC18 vector using the Sure Clone cloning kit (Pharmacia, Uppsala, Sweden) This pUC18-rMFE was used as template for polymerase chain reaction with specific oligonucleotide primers containing unique restriction sites in their 5Ј ends. The resulting rMFE cDNA had an extra SacI restriction site in the 5Ј end and an XhoI cutting site in the 3Ј end and was cloned in pUC18 (pUC18-rM-FE(SacXho)) in order to allow the amplification of the fragment (Fig.  2B).
pYE352-CTA1 and pUC18-rMFE(SacXho) were digested with SacI and XhoI, and the fragments containing the rat MFE and pYE352 with catalase promoter were isolated from 0.7% agarose gel by Geneclean II kit (Bio 101 Inc., Vista, CA) and ligated overnight. The resulting construct contains the rat MFE open reading frame under the control of oleic acid-inducible catalase promoter, and the YFOX-2 cells were transformed by lithium acetate method (26). The transformants were selected on ura Ϫ plates and replica plated on oleic acid plates. Colonies growing on both plates were chosen for complementation studies. UTL-7A, YFOX-2, and YFOX-2-CTA1 strains were used as controls.
Northern Blot Analysis-The strains were grown in liquid cultures, and the REX total RNA isolation kit from USB (Cleveland, OH) was used for isolating total RNA from 0.5 g of yeast cells from each culture. 10-g samples of each RNA were fractionated by electrophoresis in formaldehyde-agarose (0.8%) and blotted onto a nitrocellulose filter. The filter was hybridized with random prime labeled rat MFE probe.
Immunoblot Analysis-For immunoblot analysis the harvested cells were broken with glass beads, and the lysates were subjected to SDSpolyacrylamide gel electrophoresis followed by electroblotting on a nitrocellulose filter. The rat MFE protein was immunodetected with rabbit anti-rMFE and using affinity-purified goat anti-rabbit IgG with horseradish peroxidase conjugate (Bio-Rad) as second antibody.
Enzyme Activity Assays and Studies on Stereospecificities of 2-Enoyl-CoA Hydratases and 3-Hydroxyacyl-CoA Dehydrogenases from Wildtype and Transformed Yeast Cells-2-Enoyl-CoA hydratase 1 activity was measured in the direction of hydration of trans-2-decenoyl-CoA to L-3-hydroxydecanoyl-CoA. The incubation mixture consisted of 60 nmol of trans-2-decenoyl-CoA in 50 M Tris/HCl, pH 9.0, 50 g of bovine serum albumin, 50 mol of potassium chloride, 1 mol of NAD ϩ , 5 g of porcine L-3-hydroxyacyl-CoA dehydrogenase (Sigma) in a volume of 1 ml at 23°C. The reaction was started by adding a sample into the cuvette and monitored by following the NADH production spectrophotometrically. 2-Enoyl-CoA hydratase 2 activity was measured in the direction of hydration of trans-2-decenoyl-CoA to D-3-hydroxydecanoyl-CoA under the conditions described above except that the L-3-hydroxyacyl-CoA dehydrogenase was replaced with 0.5 mol ϫ min Ϫ1 of D-3-hydroxyacyl-CoA dehydrogenase activity (determined with D-3-hydroxydecanoyl-CoA).
For measuring the L-specific 3-hydroxyacyl-CoA dehydrogenase activity, 60 mol of a racemic mixture of D,L-3-hydroxydecanoyl-CoA was incubated in the Tris/HCL/bovine serum albumin/potassium chloride mixture (see above) containing 25 M magnesium chloride, 1 M sodium pyruvate, 10 g of lactate dehydrogenase from rabbit muscle (Boehringer Mannheim) in the presence of recombinant D-specific 3-hydroxyacyl-CoA dehydrogenase. After the D-isoform was removed, the reaction was initiated by adding the sample. When assaying the Dspecific 3-hydroxyacyl-CoA dehydrogenase, the L-isoform of the substrate was first removed from the reaction mixture with the L-specific 3-hydroxyacyl-CoA dehydrogenase followed by adding the sample. The oxidation of 3-hydroxydecanoyl-CoA was followed by monitoring the formation of magnesium-3-ketodecanoyl complex at 303 nm (18,28,29).
Recombinant D-specific 3-hydroxyacyl-CoA dehydrogenase was a truncated version of the peroxisomal MFE from Candida tropicalis, and it was produced as described by Hiltunen et al. (18). Stereospecificities of L-and D-hydroxyacyl-CoA dehydrogenases have been tested separately using synthetized L-and D-isoforms of 3-hydroxydecanoyl-CoA, which were obtained as described by Malila et al. (29).
Immunoelectron Microscopy-UTL-7A and FOX-2-rMFE cells were grown on oleic acid, and the pelleted cells were fixed in 8% paraformaldehyde, processed, and immunolabeled as described by Sormunen et al. (30). The rabbit anti-rMFE IgG bound on cell sections was localized with protein A-colloid gold labeling technique.

Growth of Yeast Cells on Different
Media-All yeast strains tested (UTL-7A, YFOX-2, YFOX-2-CTA1, and YFOX-2-rMFE) were viable (Fig. 3) on glucose-rich YPD plates, whereas on oleic acid plates only YFOX-2-rMFE and wild-type UTL-7A cells were able to grow. In a control experiment on ura Ϫ plates, only the strains transformed with pYE352-rMFE or pYE352-CTA1 were viable (data not shown). Altogether, these results indicate that apparent complementation of fox-2 mutation by the rat MFE encoded by the plasmid pYE352-rMFE had occurred.
Northern Blot and Immunoblot Analyses-To investigate further the expression of rat MFE in YFOX-2-rMFE, Northern blot analysis was carried out. The rat MFE cDNA probe hybridized only with the RNA isolated from YFOX-2-rMFE cells grown in oleic acid medium, the size of the signal (2.2 kb) corresponding well with the estimated size of the rat MFE RNA (25) (Fig. 4A). This signal was not seen in lane 3 of Fig. 4A (YFOX-2-rMFE grown in glucose medium), which shows that the CTA1 promoter is a strong glucose-repressible and oleic acid-inducible promoter, which is in agreement with earlier observations (31,32).
In immununoblot analysis of soluble proteins from the four yeast strains using antibodies against rat MFE, a signal of 78 kDa was obtained only with proteins from YFOX-2-rMFE cells grown in oleic acid medium (Fig. 4B), which corresponds well with the signal obtained with the rat wild-type MFE purified from rat liver (22).
Enzyme Activity Assays-The yeast MFE and the rat peroxisomal MFE are known to catalyze the conversion of trans-2enoyl-CoA to 3-ketoacyl-CoA via hydration and dehydrogenation steps, but the stereochemical courses of the reactions are different; in yeast the events occur via D-3-hydroxyacyl-CoA, whereas in rat they occur via the corresponding L-form (18,28). This difference was used to design kinetic experiments to distinguish between the two enzymes. As shown in Fig. 5A, when the protein sample from oleic acid-grown UTL-7A cells was added to a racemic mixture of the D-and L-forms of 3-hydroxydecanoyl-CoA, only one half of the substrate was converted to 3-ketodecanoyl-CoA. The subsequent addition of L-specific 3-hydroxyacyl-CoA dehydrogenase from porcine heart resulted in consumption of the second half of the substrate. This indicated that the yeast extract used contained only D-specific hydroxyacyl-CoA dehydrogenase. This conclusion was further confirmed by the observation that the addition of the recombinant D-3-hydroxyacyl-CoA dehydrogenase (18) did not result in any additional production of 3-ketoacyl-CoA ester. When the corresponding experiment was carried out with YFOX-2-rMFE rather than wild-type cells, the addition of the cell lysate again caused the removal of only half of the substrate (Fig. 5B), but this time separate additions of the auxiliary enzymes, L-specific and D-specific hydroxyacyl-CoA dehydrogenases, yielded opposite effects; the L-specific enzyme caused no reaction, whereas the D-specific one metabolized the remaining half of the substrate. These results clearly demonstrate that the transformed YFOX-2 strain was devoid of its endogenous D-3-hydroxyacyl-CoA dehydrogenase activity but possessed the expressed L-3hydroxyacyl-CoA dehydrogenase of plasmid origin.
When soluble proteins were extracted from UTL-7A and YFOX-2-rMFE yeast strains grown on glucose and oleic acid media and the samples were tested for combined activity (i.e. capability to convert trans-2-decenoyl-CoA to 3-ketoacyl-CoA), activity was observed within the detection limits only in oleic acid-grown cells (Table II).
When enzyme activities participating in 3-hydroxyacyl-CoA metabolism were studied in more detail, activities of 2-enoyl-CoA hydratase 2 and D-3-hydroxyacyl-CoA dehydrogenase could be detected in extracts of UTL-7A cells grown on oleic  2. Construction of expression plasmid carrying rat peroxisomal MFE behind yeast oleic acid-inducible catalase A promoter. pYE352-CTA1 was constructed from YEp352-CTA1 by adding a linker sequence containing an extra XhoI restriction site (A), and the pYE352-rMFE was created by joining the rat MFE cDNA behind the yeast catalase promoter by digesting pYE352-CTA1 and pUC18-rMFE with SacI and XhoI and ligating them together (B). PCR, polymerase chain reaction. acid (Table II). Similarly, the combined activity (rate of metabolism of trans-2-decenoyl-CoA to 3-ketoacyl-CoA) was also present. Furthermore, in UTL-7A extracts the activities of 2-enoyl-CoA hydratase 1 and L-3-hydroxyacyl-CoA dehydrogenase were below the detection limits in the assay system used.
The extracts from YFOX-2-rMFE cells contained both 2-enoyl-CoA hydratase 1 and L-3-hydroxyacyl-CoA dehydrogenase activities but were lacking D-3-hydroxyacyl-CoA dehydrogenase activity. Unexpectedly, about 42% of hydratase 2 activity was still detectable when compared with the wild-type cells. Because it has been shown earlier that the mRNA for FOX-2p (yeast peroxisomal MFE) is not produced in YFOX-2 cells, this result indicates that in addition to FOX-2p there exists another protein capable of catalyzing hydratase 2 reactions in yeast. Interestingly, recent data have shown that in rat liver hydratase 2 activity can be found in two subcellular compartments: peroxisomes and microsomes (29).
Subcellular Localization-To study the subcellular location of rat MFE expressed in YFOX-2 cells, immunoelectron microscopy was carried out applying protein A colloid gold technique and using anti-rMFE antibodies. As shown in Fig. 6, most of the gold particles were found within peroxisomes in YFOX-2-rMFE cells grown on oleic acid in contrast to wild-type cells, for which no labeling was found with the same procedure. This implies that the targeting signal for rMFE is recognized by the yeast import mechanism. DISCUSSION Present work provides several lines of evidence that the rat peroxisomal MFE can be heterologously expressed in an active form in S. cerevisiae and that it complements the lack of the corresponding endogenous yeast peroxisomal MFE. (i) Both the mRNA and the rat protein were detected only in cells transformed with plasmid pYE352-rMFE. (ii) The transformants had gained 2-enoyl-CoA hydratase 1 and L-3-hydroxyacyl-CoA dehydrogenase activities, which were undetectable in wild-type yeast cells. (iii) Immunoelectron microscopy indicated that the rat MFE was targeted into the peroxisomes. (iv) Finally, the expression of rat peroxisomal MFE allows the YFOX-2 mutant cells to utilize oleic acid as a carbon source.
The observed functional complementation, however, results in the change of stereochemistry of the second (hydration) and the third (oxidation) reactions of ␤-oxidation (Fig. 1), namely from D-3-hydroxyacyl-CoA-dependent pathway to L-3-hydroxy intermediate-dependent one. Consequently, the results indicate that the functioning of the ␤-oxidation pathway in S. cerevisiae is independent of the stereochemistry of its second  and third reactions. This gives rise to the question of why two chirally different ␤-oxidation pathways are found in higher and lower eukaryotes. The simplest explanation, that both were present in a phylogenetic ancestor, is contradicted by the fact that only one type of MFE has been found in eubacteria (33)(34)(35) and mammalian mitochondria (36) belonging to the same protein family as those of animal (25,37) and plant peroxisomes (38). It is more likely that this original MFE was replaced by a new one (hydratase 1/L-specific 3-hydoxyacyl-CoA dehydrogenase by hydratase 2/D-specific 3-hydroxyacyl-CoA dehydrogenase) in the fungi after the bifurcation of animals and fungi.
Thus the results suggest that the acquisition of enzymes for a metabolic pathway was based on the executing of a certain function and not on the stereochemical course of the events. Following this principle, the evolution of ␤-oxidation pathways has resulted in a special type of functional convergence in that the MFEs from yeast and mammalian peroxisomes catalyze the same two sequential reactions but via reciprocal stereochemistry. ␤-Oxidation systems are usually visualized as forming an organized structure in order to provide efficient flux through the many sequential reactions of their pathways (39,40). It therefore seems surprising at first sight that a phylogenetically unrelated heterologous protein can efficiently replace one endogenous component of this metabolon. It cannot be excluded, however, that the two multifunctional proteins of rat and fungal peroxisomes may be related in terms of their three-dimensional structure. If this turns out to be true in the future, it will mean that in addition to functional convergence, structural convergence must also have occurred.