Peroxisomal Bifunctional Protein from Rat Liver Is a Trifunctional Enzyme Possessing 2-Enoyl-CoA Hydratase, 3-Hydroxyacyl-CoA Dehydrogenase, and A3,A2-Enoyl-CoA Isomerase Activities*

Peroxisomal A3,A2-enoyl-CoA isomerase (EC 5.3.3.8) was studied in the liver of rats treated with clofibrate. The mitochondrial and peroxisomal isoen- zymes were separated chromatographically and the peroxisomal isomerase purified to apparent homoge- neity. In addition to the isomerization of 3-enoyl-CoA esters, the purified protein also catalyzed hydration of trans-2-enoyl-CoA and oxidation of L.-3-hydroxyacyl-CoA. Incubation of the purified protein with trans-3- decenoyl-CoA, NAD+, and Mg+ resulted in an increase in absorbance at 303 nm, indicating the formation of 3-ketoacyl-CoA. The protein purified was monomeric, with an esti- mated molecular weight of 78,000. In immunoblotting it was recognized by the antibody to peroxisomal bi- functional protein from rat liver. Comparison of the amino acid sequences of cyanogen bromide

The mitochondrial and peroxisomal isoenzymes were separated chromatographically and the peroxisomal isomerase purified to apparent homogeneity. In addition to the isomerization of 3-enoyl-CoA esters, the purified protein also catalyzed hydration of trans-2-enoyl-CoA and oxidation of L. - Multifunctional proteins metabolizing fatty acyl-CoA esters are typical of extramitochondrial P-oxidation systems (1). In contrast to mitochondrial @oxidation pathways, which are characteristic of animals, extramitochondrial P-oxidation pathways are widely distributed among living organisms (2). In eukaryotes these systems are located in microbodies such as peroxisomes or glyoxisomes, whereas in prokaryotes they have a cytosolic location.
The bifunctional protein having 2-enoyl-CoA hydratase and 3-hydroxyacyl-CoA dehydrogenase activities is one of the best known mammalian peroxisomal proteins and has been purified from many mammalians (3,4) including man (5). The amino acid sequence of the rat enzyme is known (6), and the relevant gene in the rat has been cloned (7). The bifunctional protein, its antibody, and cDNA probes have been commonly used when studying the biogenesis of peroxisomes, protein targeting to these organelles (8-lo), and also the pathogenesis of human inherited peroxisomal disorders (11)(12)(13). Even though mitochondrial and peroxisomal P-oxidation are similar in terms of their chemical reactions, their physiological roles in the mammalian metabolism appear to be quite different.
Peroxisomal P-oxidation is considered to be physiologically relevant to a large spectrum of fatty acids or fatty acid derivatives which are poor substrates for mitochondrial P-oxidation (for a review, see Ref. 14). Experiments with polyunsaturated fatty acids and isolated peroxisomes from rat liver have provided evidence that peroxisomes contain enzymatic activity for the A3,A2-isomerization of acyl-CoA (15), the later investigations into the distribution of the enzyme in liver subcellular organelles have demonstrated location of isomerase activity in both peroxisomes and mitochondria (16). The peroxisomal isoenzyme has never been characterized, however.
The aim of this work was to purify and characterize peroxisomal A3,A2-enoyl-CoA isomerase. In addition to its isomerization activity, the purified peroxisomal enzyme was found to catalyze hydration of trans-2-enoyl-CoA and dehydrogenation of L-3-hydroxyacyl-CoA, indicating that it is a multifunctional protein. The physical, immunological, and kinetic properties of the protein purified here and identified it as the same polypeptide which is known as the peroxisomal bifunctional protein. Thus, the data indicate that the bifuncational protein from rat liver can act as a trifunctional hydratasedehydrogenase-isomerase enzyme.

RESULTS
Purification of the Peroxisomal Protein Catalyzing Isomeri-z&ion-When studying A3,A2-enoyl-CoA isomerases in rat liver, we found recently that mitochondrial and peroxisomal activities can be separated using general dye ligand chromatography on Matrex gel red A and succeeded in demonstrating that the isomerase activity originating from peroxisomes elutes at 0.4-0.9 M KC1 at pH 7.0, whereas the mitochondrial isomerase activity elutes at a lower salt concentration.' In the same study we observed that the treatment of rats with clofibrate caused a remarkable increase in the activity of both isoenzymes.
Taking advantage of the above-mentioned finding, liver extracts from clofibrate-treated rats were applied to Matrex gel red A, and the isomerase peak eluting at a high salt concentration was taken for further purification. When ammonium sulfate precipitation and chromatographies on carboxymethylcellulose (CM32) and hydroxylapatite were carried out, sodium dodecyl sulfate-polyacrylamide gel electrophoresis of the isomerase preparation showed only one band with the molecular weight of 78,000. This and the appearance of one symmetrical peak in reverse phase column chromatography using an acetonitrile gradient indicated that the protein had been purified to apparent homogeneity. The specific activity of the purified enzyme preparation was 5.5 pmol x min-' X mg protein-' when 60 ELM trans-3-hexenoyl-CoA was used as the substrate. Gel filtration on S-200 HR gave a molecular weight of 83,000 for the native protein, revealing that the purified protein was monomeric. Zsomerase Activity-Enzyme activity during purification was measured with trans-3-hexenoyl-CoA. The ratio of the reaction velocity of trans-3-hexenoyl-CoA to that of trans-3decenoyl-CoA was 2.7. If cls-3-hexenoyl-CoA was taken as a substrate and similar incubations were carried out (with crotonase, L-3-hydroxyacyl-CoA dehydrogenase and NAD+ present), NADH was generated. This demonstrated that the protein functioned as a cis,trans-3,trans-2-enoyl-CoA isomerase. In further experiments, the purified protein was incubated with trans-3-decenoyl-CoA and NAD'. Surprisingly, these factors were enough to generate NADH, a process which requires 2-enoyl-CoA hydratase and 3-hydroxyacyl-CoA dehydrogenase activities in addition to A3,A2-enoyl-CoA isomerase. If pyruvate and lactate dehydrogenase were added to remove the NADH formed and the same reaction was followed at 303 nm in the presence of Mg2+, an increase in absorbance was observed. The fact that Mg2+ forms a complex with 3ketoacyl compounds absorbed in the near-UV region indicates that trans-3-decenoyl-CoA was metabolized to a 3-keto derivative during the incubation. In control experiments, crotonase, 3-hydroxyacyl-CoA dehydrogenase, and the mitochondrial isomerase purified from rat heart were alone insufficient to catalyze this conversion, but all of them were required simultaneously (Fig. 1, Miniprint).
The above data provide evidence that the purified protein is a multifunctional enzyme. This being the case, and assuming that the results were not caused by possible trace contaminants in the preparation, the enzyme activities and polypeptide should be chromatographically inseparable. To test this further, an aliquot from the hydroxylapatite eluate was dialyzed and applied to an agarose-hexane-CoA affinity column to which all the activities were bound. Elution of the bifunctional activity (2-enoyl-CoA hydratase and 3-hydroxyacyl-CoA dehydrogenase) in a KC1 gradient was parallel to the isomerase activity (data not shown). An aliquot from the pooled activity fraction of the agarose-hexane-CoA column was again applied to a cation exchanger Mono S column of a fast protein liquid chromatography apparatus, and a linear sodium chloride gradient was developed. One sharp symmetrical protein peak was observed, and the A3,A2-enoyl-CoA isomerase, 2-enoyl-CoA hydratase, and 3-hydroxyacyl-CoA dehydrogenase activities were all eluted together with this protein (Fig. 2, Miniprint). We have also tested a 5'-AMP-Sepharose affinity column, which it has been proposed will bind NAD+-linked enzymes, and found that all three activities were bound to this material and showed parallel elution with a KC1 gradient.

Identification of the Novel Trifunctionul Protein a-s the Earlier Bifunctional
Protein-Because the molecular weight of the enzyme purified here was the same as that of the peroxisomal bifunctional protein and it possessed both known catalytic activities of the bifunctional protein, the question arose as to whether they were identical molecules. To test this possibility further, we purified peroxisomal bifunctional protein from livers of clofibrate-treated rats following the procedure described by Osumi and Hashimoto (3). After purification the enzyme preparation showed only one band on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, indicating that the protein was purified to apparent homogeneity. When the catalytic activities of this protein were measured with either trans-3-decenoyl-CoA or trans-2-decenoyl-CoA in the presence of purified enzymes as auxiliary catalytes, the enzyme possessed separately measurable 2-enoyl-CoA hydratase, 3-hydroxyacyl-CoA dehydrogenase, and A3,A2-enoyl-CoA isomerase activities and also had bifunctional and trifunctional properties when measured with trans-2-and trans-3-decenoyl-CoA, respectively, as substrates (Table I, Miniprint). The purified isomerase, bifunctional protein, and liver homogenate from clofibrate-treated rats were immunoblotted with antibody to the bifunctional protein from rat liver, and a band of the same molecular size was detectable in all of them.
The published amino acid sequence for the bifunctional protein shows 12 methionines. This and the known property of cyanogen bromide of cleaving peptides at methionine at neutral or acidic pH (17) provide a means to test whether the peroxisomal bifunctional protein from rat liver and the isomerization catalyzing enzyme purified here were identical. The protein purified according to the present procedure was therefore cleaved with cyanogen bromide and the peptides separated on a reverse phase column (Fig. 3, Miniprint). The amino acid sequences of two of the peptides were partially determined by means of an automatic protein sequencer, allowing comparison with the known sequence of the bifunctional protein. Two identical sequences were found in the bifunctional protein. In both cases methionine was the amino acid preceding the sequence observed, demonstrating the reliability of the method used. No amino acid sequence was obtained from the first peptide emerging in the acetonitrile gradient. One explanation for this may be that the peptide represents the amino terminal part of the bifunctional protein, which is known to be blocked (18). DISCUSSION The discovery of A3,A2-enoyl-CoA isomerase activity in the peroxisomal bifunctional protein was based on the following findings: 1) the purified peroxisomal protein which catalyzed the isomerization of 3-enoyl-CoA esters was observed to catalyze the formation of 3-ketoacyl-CoA and the reduction of NAD+ in the presence of A3-enoyl-CoA substrates. 2) Edman degradation analysis of peptides obtained by cyanogen bromide cleavage of the enzyme showed the amino acid sequence to be identical to the published sequence for the rat liver bifunctional protein over the regions determined.
3) The bifunctional protein purified according to the published method and the isomerization catalyzing protein purified here showed similar catalytic properties with different acyl-CoA intermediates of P-oxidation. In contrast to the present work, there is a short note in the literature that the bifunctional protein from liver peroxisomes is devoid of the cis-3-trans-2enoyl-CoA isomerase activity (19). The reason for this discrepancy remains unclear.
The rat liver peroxisomal bifunctional protein contains homologous regions with mitochondrial enoyl-CoA hydratase close to its amino terminus, whereas the carboxyl-terminal end shows homology with mitochondrial 3-hydroxyacyl-CoA dehydrogenase. This has been taken to indicate that the bifunctional protein contains two functional domains (6). It has been demonstrated by kinetic studies with the purified bifunctional protein that intermediates are channelled from the active site of the enoyl-CoA hydratase domain to the 3hydroxyacyl-CoA dehydrogenase domain without release into the bulk phase (20). The present finding that additions of 2enoyl-CoA hydratase (crotonase) and 3-hydroxyacyl-CoA dehydrogenase to the incubation as auxiliary catalytes do not increase the observed rate of isomerization of A3-enoyl-CoA esters can be interpreted as indicating that channelling also occurred between the active centers catalyzing isomerization and hydration. Experiments with liver peroxisomes isolated from rats treated with clofibrate have demonstrated chain shortening of [ 1-Wlarachidonic acid by three acetyl groups, indicating that peroxisomal P-oxidation in vitro can proceed beyond a double bond positioned at an odd-numbered carbon atom in fatty acids (15). Furthermore, infusion of truns-3dodecenoic acid in isolated rat liver stimulated H202 production (16), which is a sequence requiring isomerization of the double bond. The present data demonstrated isomerase activity in the bifunctional protein, which has a well documented peroxisomal location (3,4). Taken together, these findings are in line with the proposal that one of the physiological functions of peroxisomal P-oxidation is the degradation of polyunsaturated fatty acids (14, 15), a process which utilizes A3,A2-enoyl-CoA isomerase as an auxiliary catalyte.
In filamentous fungi and yeasts, peroxisomal A3,A2-enoyl-CoA isomerase is separate from the multifunctional protein (2), which catalyzes hydration, dehydrogenation, and epimerization reactions in P-oxidation (21), but in rat liver the multifunctional protein of peroxisomal P-oxidation catalyzes isomerization, hydration, and dehydrogenation of acyl-CoA metabolites, although it is not capable of epimerization of 3hydroxyacyl-CoA esters (22). The hydratase-dehydrogenaseepimerase enzyme in Candida tropicalis, which is also a monomeric protein with an estimated molecular weight of 99,535, shows only limited amino acid homology over a short distance with the peroxisomal bifunctional protein in the rat (23), which was identified here as an isomerase-hydratase-dehydrogenase enzyme. This and the difference in enzyme pattern between the multifunctional proteins involved in fungal and rat peroxisomal p-oxidations indicate distant phylogenic divergence of these proteins in different phyla.