Δ3,5,Δ2,4-Dienoyl-CoA Isomerase Is a Multifunctional Isomerase

Δ3,5,Δ2,4-Dienoyl-CoA isomerase (DI), an auxiliary enzyme of unsaturated fatty acid β-oxidation, was purified from rat mitochondria and peroxisomes and subjected to N-terminal sequencing to facilitate a mechanistic study of this enzyme. The mature mitochondrial DI from rat heart was lacking its 34 N-terminal amino acid residues that have the properties of a mitochondrial targeting sequence. The peroxisomal isomerase was identified as a product of the same gene with a truncated and ragged N terminus. Expression of the cDNA coding for the mature mitochondrial DI in Escherichia coli yielded an enzyme preparation that was as active as the native DI. Because the recombinant DI also exhibited Δ3,5,7,Δ2,4,6-trienoyl-CoA isomerase (TI) activity, both isomerases reside on the same protein. Mutations of any of the 3 acidic amino acid residues located at the active site (Modis, Y., Filppula, S. A., Novikov, D. K., Norledge, B., Hiltunen, J. K., and Wierenga, R. K. (1998) Structure 6, 957–970) caused activity losses. In contrast to only a 10-fold decrease in activity upon replacement of Asp176 by Ala, substitutions of Asp204 by Asn and of Glu196 by Gln resulted in 105-fold lower activities. Such activity losses are consistent with the direct involvement of these latter two residues in the proposed proton transfers at carbons 2 and 6 or 8 of the substrates. Probing of the wild-type and mutants forms of the enzyme with 2,5-octadienoyl-CoA as substrate revealed low Δ2,Δ3-enoyl-CoA isomerase and Δ5,Δ4-enoyl-CoA isomerase activities catalyzed by Glu196 and Asp204, respectively. Altogether, these data reveal that positional isomerizations of the diene and triene are facilitated by simultaneous proton transfers involving Glu196 and Asp204, whereas each residue alone can catalyze, albeit less efficiently, a monoene isomerization.

⌬ 3,5 ,⌬ 2,4 -Dienoyl-CoA isomerase (DI), an auxiliary enzyme of unsaturated fatty acid ␤-oxidation, was purified from rat mitochondria and peroxisomes and subjected to N-terminal sequencing to facilitate a mechanistic study of this enzyme. The mature mitochondrial DI from rat heart was lacking its 34 N-terminal amino acid residues that have the properties of a mitochondrial targeting sequence. The peroxisomal isomerase was identified as a product of the same gene with a truncated and ragged N terminus. Expression of the cDNA coding for the mature mitochondrial DI in Escherichia coli yielded an enzyme preparation that was as active as the native DI. Because the recombinant DI also exhibited The degradation of unsaturated fatty acids by ␤-oxidation requires several auxiliary enzymes in addition to the enzymes necessary for the breakdown of saturated fatty acids (for a review see Ref. 1). One of the auxiliary enzymes is ⌬ 3,5 ,⌬ 2,4 -dienoyl-CoA isomerase (dienoyl-CoA isomerase), 1 which catalyzes the isomerization of 3,5-dienoyl-CoA to 2,4-dienoyl-CoA (2,3). During the ␤-oxidation of unsaturated fatty acids with odd-numbered double bonds, 5-enoyl-CoA intermediates can be converted to 3,5-dienoyl-CoA by the sequential actions of acyl-CoA dehydrogenase and ⌬ 3 ,⌬ 2 -enoyl-CoA isomerase (EC 5.3.3.8) (enoyl-CoA isomerase). The further degradation of 3,5dienoyl-CoA requires its isomerization to 2,4-dienoyl-CoA, because the latter compound can be reduced by 2,4-dienoyl-CoA reductase (EC 1.3.1.34) to 3-enoyl-CoA. The isomerization of 3-enoyl-CoA to 2-enoyl-CoA by enoyl-CoA isomerase completes the reductase-dependent sequence of reactions during the ␤-oxidation of unsaturated fatty acids with odd-numbered double bonds.
Dienoyl-CoA isomerase was first identified in rat liver mitochondria but later was also detected in rat liver peroxisomes (4,5). The molecular characterization of this enzyme revealed the amino acid sequence of the unprocessed subunit, which has a peroxisomal targeting signal, type 1, at the C terminus and an N-terminal sequence that is consistent with the targeting of this protein to mitochondria (6). This situation is suggestive of a dual subcellular localization and agrees with the previously observed kinetic and immunological similarities of the mitochondrial and peroxisomal forms of this enzyme (5). The crystal structure of a recombinant form of dienoyl-CoA isomerase consisting of subunits without the 53 N-terminal amino acid residues was obtained at 1.5-Å resolution (7). This study confirmed the proposed hexameric structure of the enzyme (6) and revealed the active site as a deeply buried hydrophobic pocket with three acidic residues, Asp 176 , Glu 196 , and Asp 204 . The latter two of these residues were predicted to catalyze proton transfers at carbons 2 and 6, respectively, of the 3,5-dienoyl-CoA substrate (6). Surprisingly, ⌬ 3,5,7 ,⌬ 2,4,6 -trienoyl-CoA isomerase (trienoyl-CoA isomerase), an enzyme involved in the degradation of unsaturated fatty acids with conjugated double bonds, was found to be a component enzyme of dienoyl-CoA isomerase (8).
The association of dienoyl-CoA isomerase and trienoyl-CoA isomerase with the same protein prompted this study aimed at elucidating the mechanisms of action of these enzymes. This goal necessitated the characterization of the highly active, mature forms of these enzymes present in mitochondria and peroxisomes.
Purification of Rat Mitochondrial and Peroxisomal Dienoyl-CoA Isomerases-Dienoyl-CoA isomerases from rat liver and heart were purified as described previously (3). Adult Harlan Sprague-Dawley rats were used, which had been fed rodent chow containing 2% (w/w) di(ethylhexyl)phthalate. For the purification of rat liver peroxisomal dienoyl-CoA isomerase, a light mitochondrial fraction was prepared from two rat livers as described by de Duve et al. (11). Peroxisomes were prepared by Nycodenz density gradient centrifugation of a light mitochondrial fraction as described previously (5). For this purpose, a 30% (w/v) solution of Nycodenz containing 1 mM EDTA, 5 mM Hepes (pH 7.3), and 0.1% ethanol was prepared, and 21 ml of this solution was placed in a 30-ml ultracentrifuge tube on top of 1.5 ml of a 60% sucrose cushion. A density gradient was generated by centrifugation at 60,000 ϫ g in a T865 small angle rotor on a DuPont RC70 ultracentrifuge at 4°C for 24 h. A light mitochondrial fraction (ϳ45 mg of protein in 1.5 ml) was layered on top of the gradient followed by 1.5 ml of a cover solution of a 3-fold diluted isolation buffer containing 0.25 M sucrose, 1 mM EDTA, 0.1% ethanol, and 10 mM Tris-HCl (pH 7.4). The sample was centrifuged at 76,000 ϫ g for 1 h at 4°C. Fractions of 2.5 ml each were collected from the bottom of the tube. Peroxisomes, microsomes, and mitochondria were localized by assaying the marker enzymes catalase, esterase, and malate dehydrogenase, respectively. Peroxisomal fractions were combined and diluted 5-fold with isolation buffer before they were harvested by centrifugation at 17,500 ϫ g for 20 min. Pellets were suspended in 2 ml of 5 mM potassium P i (pH 6.3) containing 5 mM mercaptoethanol, 1 mM EDTA, 1 mM EGTA, 1 mM benzamidine, and 0.5 mM PMSF (buffer A). The suspension was centrifuged at 100,000 ϫ g for 1 h after sonicating it 10 times for 20 s each at 4°C. The supernatant was applied to a hydroxylapaptite column (1.5 ϫ 22 cm) previously equilibrated with buffer A. The column was washed with buffer A containing 0.5 M KCl and then was developed with a gradient made up of 160 ml of buffer A and 160 ml of buffer A containing 0.8 M potassium P i (pH 6.3). Fractions of 3 ml each were collected, and the fractions containing the dienoyl-CoA isomerase activity were combined and concentrated in an Amicon concentrator with a YM-10 membrane. After dialysis overnight against 25 mM ethanolamine-acetic acid (pH 9.4) containing 5 mM mercaptoethanol, 1 mM EDTA, 1 mM benzamidine, 0.5 mM PMSF, and 20% glycerol (buffer B), the sample was applied to a chromatofocusing column (1 ϫ 15 cm) containing Polybuffer Exchanger 94 equilibrated with buffer B. The column was extensively washed with buffer B and then developed with 12 column volumes of Polybuffer 96 adjusted to pH 6.0 with acetic acid. Fractions of 3 ml each were collected and assayed for dienoyl-CoA isomerase. The active fractions were combined and concentrated with a Millipore centrifugal filter device.
SDS-PAGE and Immunoblotting-Aliquots of purified dienoyl-CoA isomerase were treated with SDS sample buffer and subjected to SDS-PAGE on gradient (4 -20%) gels (12). Proteins were transferred to a polyvinylidene difluoride membrane by semi-dry blotting (13), and proteins remaining on the gel were visualized by staining with Coomassie Blue. The membrane was incubated for 1 h with a 500-fold-diluted rabbit antiserum or with monospecific antibodies (1 g/ml) prepared from the serum raised against rat liver dienoyl-CoA isomerase. After incubating the membrane with goat anti-rabbit IgG conjugated with alkaline phosphatase, it was developed with a staining mixture containing the alkaline phosphatase substrate until the antigen bands were visualized (14).
Analysis of Protein Sequence-N-terminal amino acid sequencing was performed by Stephen Bobin at the Dartmouth College Molecular Biology Core Facility. The N-terminal sequence of the full-length dienoyl-CoA isomerase was analyzed with the program HelicalWheel to draw a helical wheel as described previously (15).
Cloning and Expression of Dienoyl-CoA Isomerase-Rat heart Marathon-Ready cDNA (CLONTECH) was used as the template for cloning the cDNA of the full-length dienoyl-CoA isomerase by touch-down PCR according to the protocol of CLONTECH. The primers were 5Ј-CAGGATC-CCATATGGCTACCGCGATGACAGTTTCCA-3Ј and 5Ј-CAGTAAGCT-TATCAGAGCTTGGAGAAGGTGATGCTT-3Ј. The PCR product (ϳ1 kb) was inserted into vector pGEM-T Easy (Promega) and amplified. Thereafter, it was subcloned into the BamHI-HindIII site of vector pND-1 (a gift from Dr. Didier Negre) to form expression plasmid pNDDI. This plasmid was used to transform Escherichia coli BL21(DE3)pLysS by the method of Chung et al. (16). Because attempts to express the full-length dienoyl-CoA isomerase were unsuccessful, the cDNA of the mature dienoyl-CoA isomerase was generated from plasmid pNDDI by PCR using primer 5Ј-CAGGATCCCATATGAGCTCCTCTGCACAAGAGGCGT-3Ј for introducing a starting methionine followed by three serines and a 3Ј primer 5Ј-CAGTAAGCTTATCAGAGCTTGGAGAAGGTGATGCTT-3Ј. BamHI and HindIII restriction sites were introduced with the 5Ј primer and 3Ј primer, respectively. The PCR product was subcloned into pGEM-T Easy vector and transformed into E. coli JM109 as described by the manufacturer. The plasmid pGEM with the insert was isolated from transformants and digested with BamHI and HindIII. The DNA fragments corresponding to the dienoyl-CoA isomerase was obtained by GeneClean and ligated into expression vector pND-1. The expression construct, designated as pNDdi, was used to transform E. coli strain BL21(DE3)pLysS. The transformants were grown in LB medium to an absorbance of about 1.0 at 600 nm and then induced by 0.6 mM isopropyl-1-thio-␤-D-galactopyranoside for 4 h. Cells were harvested by centrifugation at 3000 ϫ g for 5 min and stored at Ϫ80°C.
Site-directed Mutagenesis of Dienoyl-CoA Isomerase-Site-directed mutagenesis was carried out by use of a Transformer site-directed mutagenesis kit (CLONTECH) following the manufacturer's instruction. A synthetic oligonucleotide, 5Ј-ATGCTTCAATAAGATTGAAAAAGGAAG-3Ј, designed to eliminate a SspI site, was used as the selection primer. The following synthetic oligonucleotides were used as the mutagenic primers. The substituting nucleotide is underlined, and the mutant codon is in The selection primer and one of the mutagenic primers were simultaneously annealed to the template of the denatured double-stranded pNDdi and then incorporated into a new strand of DNA as a result of the elongation catalyzed by T4 DNA polymerase. After digestion with SspI, the mixture of parent and newly synthesized DNA was transformed into E. coli BMH 71-18 mutS. The plasmids were isolated from the transformants and digested again with SspI. The digestion mixture was transformed into E. coli BL21(DE3)pLysS, and the desired mutant was selected for the absence of the SspI restriction site. The point mutation was confirmed by sequencing of the respective mutant strain. Expression of the mutant enzymes was achieved by the procedure used for expressing the wild-type dienoyl-CoA isomerase.
Purification of Recombinant Wild-type and Mutant Dienoyl-CoA Isomerase-The frozen pellet from ϳ350 ml of cell culture was suspended in 10 ml of 10 mM potassium P i (pH 8.8) containing 5 mM mercaptoethanol, 1 mM EDTA, 1 mM benzamidine, and 0.5 mM PMSF (buffer A) and sonicated 12 times for 20 s each. The resultant suspension was centrifuged at 100,000 ϫ g for 30 min. The supernatant was loaded onto a Q-Sepharose column (1.5 ϫ 17 cm) previously equilibrated with buffer A. The column was extensively washed with buffer A and then eluted with a gradient made up of 120 ml of buffer A and 120 ml of buffer A containing 0.4 M KCl. The active fractions were combined and concentrated in an Amicon concentrator with a PX-10 membrane. The concentrate was diluted 10-fold with 10 mM potassium P i (pH 6.0) containing 1 mM EDTA, 5 mM mercaptoethanol, and 20% glycerol (buffer B) and applied to an S-Sepharose column (1.5 ϫ 4 cm) previously equilibrated with buffer B. After washing extensively with buffer B, the column was developed with a gradient made up of 30 ml of buffer B and 30 ml of buffer B containing 0.4 M KCl. The active fractions were combined and concentrated.
CD Spectra of Wild-type and Mutant Dienoyl-CoA Isomerases-Far-UV CD scans were acquired between 190 and 250 nm with an AVIV CD spectrophotometer equipped with temperature control. Two average scans were acquired at 20°C for each sample. The scans were normalized for protein concentration and corrected for the influence of the buffer.

Molecular Characterization of the Mitochondrial and Peroxisomal Forms of Dienoyl-CoA
Isomerase-For a planned mechanistic study of rat dienoyl-CoA isomerase, milligram quantities of highly active enzyme were required. Although a recombinant form of this enzyme, lacking its 53 N-terminal amino acid residues, has been described (6), its activity was much lower than that of the native enzyme and too low for the contemplated mechanistic study. Attempts to express the full-length isomerase-cDNA were unsuccessful. Hence, we embarked on the molecular characterization of the native mitochondrial and peroxisomal dienoyl-CoA isomerases with the aim of producing a recombinant form of the highly active mature enzyme.
N-terminal sequencing of the purified rat liver dienoyl-CoA isomerase revealed the presence of several polypeptides in agreement with the observation of at least three closely spaced bands when the same preparation was subjected to SDS-PAGE (Fig. 1A, lanes 2 and 5). Because this preparation may have been a mixture of the mitochondrial and peroxisomal forms of dienoyl-CoA isomerase, the enzyme was also purified from rat hearts, which contain few peroxisomes. SDS-PAGE and immunoblotting of the heart preparation led to the identification of the dienoyl-CoA isomerase, which seemed to be slightly larger than the liver forms of the enzyme (Fig. 1, A and B, lanes 2 and  3). When the heart dienoyl-CoA isomerase was subjected to N-terminal sequencing, a unique sequence was obtained for the first 20 residues (Fig. 2A). The sequence, beginning with 3 serine residues, perfectly matched the predicted amino acid sequence of rat liver dienoyl-CoA isomerase (6) from residue 35 through residue 54. The missing residues 1 through 34 constitute a polypeptide that has the properties of a mitochondrial targeting sequence (Fig. 2B). This conclusion is supported by the following properties of this polypeptide: It is rich in positively charged and hydroxylated residues (3 arginine, 1 lysine, 4 serine, and 3 threonine residues), and it is devoid of acidic residues and does not have a large stretch of uncharged residues (Fig. 2B). Moreover, its N terminus forms a positively charged amphiphilic helix (Fig. 2B). The mitochondrial precursor protein seems to have only one cleavage site. Another potential cleavage site that is indicated by a hydrophilic residue and a serine at positions 27 and 30, respectively, is not susceptible to proteolysis, due to the absence of an arginine residue from position 25.
In an effort to characterize the peroxisomal dienoyl-CoA isomerase, the enzyme was isolated from purified rat liver peroxisomes. A better than 80-fold purification was achieved by chromatography on hydroxylapatite followed by chromatofocusing. The resultant preparation was composed of at least three proteins (Fig. 1A, lane 4). The major component with a molecular mass of 32 kDa was identified by immunoblotting as dienoyl-CoA isomerase (Fig. 1B, lane 4). N-terminal sequencing of the material corresponding to the 32-kDa band revealed the presence of several forms of dienoyl-CoA isomerase with different N termini. The N termini of the two most prominent polypeptides were located to positions 37 and 39 of the fulllength protein ( Fig. 2A). In an attempt to determine whether the peroxisomal dienoyl-CoA isomerase exists in vivo as a truncated protein, purified peroxisomes were incubated with boiling SDS incubation buffer and subjected to SDS-PAGE followed by immunoblotting with antibodies purified by affinity chromatography on a dienoyl-CoA isomerase-Sepharose column. As shown in Fig. 1C, only one band corresponding to a 32-kDa protein was observed. Thus, it seems that the native peroxisomal dienoyl-CoA isomerase is a truncated protein with a ragged N terminus.
Mechanistic Study of Dienoyl-CoA Isomerase by Site-specific Mutagenesis-The cDNA coding for dienoyl-CoA isomerase was cloned from a rat liver cDNA library by PCR. However, attempts to express the full-length protein in E. coli were unsuccessful. We therefore generated the cDNA for the mature rat heart isomerase from the full-length cDNA by PCR. This mature mitochondrial form of dienoyl-CoA isomerase was successfully expressed in E. coli. A 10-fold purification of dienoyl-CoA isomerase, beginning with a soluble extract of such cells, yielded the pure enzyme in 65% yield. This enzyme preparation exhibited an activity of 960 units/mg (Table I), which is significantly higher than the activity of the enzyme isolated from rat liver (3). The fact that the recombinant dienoyl-CoA isomerase also exhibited trienoyl-CoA isomerase activity, proved that both catalytic properties are associated with the same protein.
For the planned mechanistic study, mutant proteins were  Table I were introduced by site-specific mutagenesis, and the recombinant mutant proteins were purified to apparent or near homogeneity as indicated by SDS-PAGE (results not shown). The near-UV CD spectra of all mutant proteins were virtually indistinguishable from the spectrum of the wild-type enzyme (data not shown). When assayed for 3,532,4 dienoyl-CoA isomerase activity, the replacement of either Asp 204 or Glu 196 by a neutral amino acid residue resulted in an ϳ10 5 -fold lower activity (Table I). This observation agrees with the proposed functions of these two residues in the direct proton transfer to or from the substrate (7). Substitution of Glu 196 by an aspartate residue produced a mutant enzyme that retained ϳ3% of the isomerase activity (Table I). This lower but significant activity demonstrates that the ␤-carboxyl group of Asp196 can facilitate the proton transfer although less efficiently than the ␥-carboxyl group of Glu 196 . The mutation of the third acidic group at the active site, Asp 176 , to Ala caused a 10-fold decrease in activity. The limited effect of this mutation argues against a direct participation of this residue in catalysis. The effects of mutating Asp 204 , Glu 196 , and Asp 176 on the trienoyl-CoA isomerase activity of this enzyme were comparable to the impact on the dienoyl-CoA isomerase except that the activity losses due to the D176A and E196D mutations were more severe (Table I). Over-all these data indicates that the active site of dienoyl-CoA isomerase is identical with the active site of trienoyl-CoA isomerase and that the same acidic residues, Glu 196 and Asp 204 , catalyze the proton transfers that result in the 3,532,4 and 3,5,732,4,6 isomerizations.
In an effort to further explore the mechanism of dienoyl-CoA/ trienoyl-CoA isomerase, its activity with 2-trans,5-cis-octadienoyl-CoA as a substrate was evaluated. With the wild-type enzyme, a small but significant conversion of the 2,5 diene to the 2,4 isomer was observed (Table I). The rate of the 2,532,4 conversion was more than 10 4 times slower than the 3,5 32,4 isomerization. Surprisingly, mutants of Glu 196 were more active than the wild-type enzyme in catalyzing this reaction ( Table I). The E196Q mutant, which was 10 times as active as the wild-type isomerase, permitted a spectroscopic analysis of the 2,532,4 isomerization. The time-dependent spectral changes shown in Fig. 3B are indicative of a direct 2,532,4 isomerization rather than a sequential 2,533,532,4 conversion. These spectra do not provide evidence for the formation of a 3,5 intermediate with an absorbance maximum at 238 nm nor do they reveal a 3,532,4 isomerization as shown in Fig. 3A. Because the 2,532,4 conversion catalyzed by the E196Q mutant was 60 times faster than the 3,532,4 isomerization catalyzed by the same enzyme (Table I), the 2,532,4 isomerization seems to be a one-step conversion.
The Asp 204 mutants also catalyzed a slow but detectable 2,532,4 isomerization. However, as shown in Fig. 4C, the formation of the 2,4 isomer proceeded with a lag. Spectral analyses of the reactions that occurred during (Fig. 4C, period  A) and after the lag phase (Fig. 4C, period B) revealed an initial 2,533,5 isomerization indicated by an increase in the absorbance at 238 nm due to the formation of the 3,5 diene (Fig. 4A) followed by the formation of the 2,4 isomer detected at 300 nm (Fig. 4B). Overall, the product formation occurred by a 2,533,532,4 conversion that showed a pronounced lag in the formation of the 2,4 isomer, because the first reaction proceeded faster than the second reaction (Table I). Because the D204A mutant catalyzed the 2,533,5 conversion, it was expected to catalyze also the isomerization of 3-octenoyl-CoA to 2-octenoyl-CoA. This conversion was in fact observed and found to take place at a rate of 0.034 unit/mg as compared with 0.4 unit/mg for the 2,533,5 isomerization.
The question of whether the differences between the reaction rates observed with various mutants and substrates were due to changes in K m values, V max values, or both were addressed. The kinetic parameters listed in Table II clearly show that the K m values varied little and that differences between V max values were the major cause of rate differences observed at fixed substrate concentrations of 20 M.

DISCUSSION
The main reason for the molecular characterization of the mature dienoyl-CoA isomerases was the need to produce a highly active recombinant form of this enzyme. Such enzyme had not been obtained when recombinant versions of the fulllength protein and of an artificially truncated isomerase were generated (6). However, this study was also prompted by a desire to demonstrate unambiguously the dual location of this enzyme in mitochondria and peroxisomes. The presence of the same dienoyl-CoA isomerase in both organelles had been suspected when antibodies raised against the mitochondrial enzyme were found to cross-react with the peroxisomal isomerase (5). Cloning of dienoyl-CoA isomerase had revealed the presence of a peroxisomal targeting signal and an N terminus with properties of mitochondrial targeting sequence. Moreover, antibodies raised against a synthetic peptide corresponding to the C terminus of the enzyme recognized 32-kDa and 36-kDa pro- teins in mitochondria and peroxisomes, respectively (6).
This study, besides revealing the N termini of the mature forms of dienoyl-CoA isomerase, confirms the dual localization of the enzyme to mitochondria and peroxisomes. However, in contrast to a previous report (6), the peroxisomal dienoyl-CoA isomerase was found to have a molecular mass of 32 kDa and hence to be a truncated form of the full-length protein. The removal of N-terminal sequences similar in size to the mitochondrial targeting sequence could be a consequence of the susceptibility of this region of the protein to proteolysis. This idea has not been tested nor have other explanations been ruled out. However, the detection of only one band corresponding to a 32-kDa protein when intact peroxisomes were solubilized with boiling SDS incubation buffer and then subjected to SDS-PAGE and immunoblotting, supports the conclusion that the truncated form(s) of the peroxisomal dienoyl-CoA isomerase is(are) present in vivo and are not artifacts of the isolation procedure. This study additionally demonstrates that the heart and liver enzymes are identical over a stretch of 20 amino acid residues. Hence, both enzymes are most likely products of the same gene.
The successful expression of the mature cardiac dienoyl-CoA isomerase in E. coli achieved two goals. Foremost, a highly active form of this enzyme became available. In fact, the maximal specific activity of 2450 units/mg for the recombinant isomerase was 6 times higher than the V max of the enzyme isolated from rat liver (3). Although this difference may be due in part to the use of HPLC-purified substrate in this study, it also reflects the preparation of a purer enzyme. In any case, the recombinant enzyme exhibited an activity well suited for the planned mechanistic study. The second achievement was the demonstration that both dienoyl-CoA isomerase and trienoyl-CoA isomerase are associated with the same protein. This result puts to rest any existing suspicion that the two activities may be expressions of distinct but similar proteins.
The successful creation and purification of several mutant forms of dienoyl-CoA isomerase permitted an analysis of its catalytic mechanism. Dramatic activity decreases were observed as the result of replacing Asp 204 and Glu 196 with neutral amino acids. This finding supports the proposed function of   these residues in proton transfers from and to the substrate (7), because it agrees with the general prediction that the mutation of a residue that directly participates in a reaction as a general acid/base would be expected to cause a 10 5 or greater decrease in activity (17).
The use of 2,5-octadienoyl-CoA as a substrate analog revealed slow, but measurable 2,532,4 isomerizations. Because the different positional isomers of octadienoyl-CoA have distinct UV spectra, it was possible to analyze the mechanisms of these isomerizations. The spectral changes observed with mutant E196Q were suggestive of a direct 2,532,4 isomerization without the formation of an intermediate. The only alternative route, via a sequence of isomerizations with 3,5-octadienoyl-CoA as an intermediate, was ruled out because the 3,532,4 isomerization was much slower than the overall 2,532,4 isomerization. Hence the observed 2,532,4 isomerization must be the result of a 534 double-bond shift as shown in Fig. 5. Asp 204 is the obvious candidate to facilitate this monoene isomerization by catalyzing a 1,3-proton shift from carbon 4 to carbon 6. Such mechanism for single double-bond isomerizations has been proposed for cholesterol oxidase (18) and ⌬ 3 ,⌬ 2enoyl-CoA isomerase (19) based on observed intramolecular 1,3-hydrogen shifts. If mechanistically similar, the 1,3-proton shift catalyzed by Asp 204 may not be concerted, as shown in Fig.  5, but rather proceed in two steps by removal of a proton from carbon 4 and formation of a stabilized carbanion followed by addition of a proton to carbon 6. Noteworthy is the observation that the 2,532,4 isomerization catalyzed by mutant E196Q proceeded 11 times faster than the same reaction catalyzed by the wild-type isomerase. The lower rate detected with the wildtype enzyme may be due to a higher pK value of Asp 204 induced by Glu 196 . Such electrostatic effect on Asp 204 would not be effective in the E196Q mutant.
The 2,532,4 isomerization catalyzed by mutant D204A was more complex than the conversion brought about by the E196Q mutant. The progress curve for the D204A-catalyzed 2,532,4 isomerization showed a lag that was shown to correspond to the conversion of 2,5-octadienoyl-CoA to its 3,5 isomer. Because the 2,533,5 isomerization was faster than the subsequent 3,532,4 isomerization, the 3,5 intermediate accumulated and initially was detectable. The formation of 3,5-octadienoyl-CoA was the result of a double-bond shift from carbon 2 to carbon 3. This double-bond isomerization must have been catalyzed by Glu 196 , which is proposed to facilitate a 1,3-proton shift from carbon 4 to carbon 2 (Fig. 5). Again, the proton transfers may not be con-certed as shown in Fig. 5 but may be sequential, resulting in the formation of a carbanion intermediate. An alternative route with a carbocationic intermediate represents an unlikely mechanism.
The analyses of the 2,532,4 isomerizations provide good evidence for Glu 196 being close to carbon 2 and Asp 204 close to carbon 6 as shown in Fig. 6. Both residues are necessary for the 3,532,4 isomerization that proceeds by a simultaneous shift of both double bonds (3). A similar mechanism is envisioned for the triene isomerization except that Asp 204 must be close to carbon 8 (Fig. 6). Such a dual role of Asp 204 suggests a certain flexibility of the residue and/or requires different positioning of the dienoyl-CoA and trienoyl-CoA substrates at the active site. Either way, the function of Asp 204 in the isomerization of the triene comes at a price that is reflected by the almost 50-fold lower activity of trienoyl-CoA isomerase as compared with dienoyl-CoA isomerase.
Altogether, this study demonstrates the need for two acidic residues to facilitate the proton transfers that result in positional isomerizations of dienes and trienes. In contrast, proton transfers that cause monoenes to shift by one carbon only require a single acidic residue as previously documented for other isomerases (18,19).