An Alternative Pathway of Oleate (cid:1) -Oxidation in Escherichia coli Involving the Hydrolysis of a Dead End Intermediate by a Thioesterase*

The degradation of 2- trans ,5- cis -tetradecadienoyl-CoA, a metabolite of oleic acid, by the purified complex of fatty acid oxidation from Escherichia coli was studied to determine how much of the metabolite is converted to 3,5- cis -tetradecadienoyl-CoA and thereby diverted from the classical, isomerase-dependent pathway of oleate (cid:1) -oxidation. Approximately 10% of the 2,5-intermediate was converted to the 3,5-isomer. When the latter compound was allowed to accumulate, it strongly inhibited the flux through the main pathway. Since (cid:2) 3,5 , (cid:2) 2,4 -dien-oyl-CoA isomerase was not detected in E. coli cells grown on oleate, the 3,5-intermediate cannot be metabolized via the reductase-dependent pathway. However, it was hydrolyzed by a thioesterase, which was most active with 3,5- cis -tetradecadienoyl-CoA as substrate and which was induced by growth of E. coli on oleate. An analysis of fatty acids present in the medium after growth of E. coli on oleate revealed the presence of 3,5-tetradecadienoate, which was not detected after cells were grown on palmitate or glucose. Altogether, these data prompt the conclusion that oleate is mostly degraded via the classical, isomerase-dependent pathway in E. coli but that a small amount of 2- trans ,5- cis -tetradecadienoyl-CoA (7), recombinant pig liver L -3-hydroxyacyl- CoA dehydrogenase (8), pig heart 3-ketoacyl-CoA thiolase (9), recombinant human peroxisomal enoyl-CoA isomerase (10), rat liver enoyl-CoA isomerase (11), recombinant rat liver dienoyl-CoA isomerase (12), and E. coli FAO complex (13) were purified by published procedures. Syntheses of Substrates and Metabolites— 2- trans -Dodecenoic acid and 2- trans -tetradecenoic acid were synthesized by reacting malonic acid with n -decanal and n -dodecanal, respectively, as described in prin-ciple by Linestead et al (14). Oleoyl-CoA, 5- cis -tetradecenoyl-CoA, 2- trans -tetradecenoyl-CoA, and 2- trans -dodecenoyl-CoA were synthe-* KP E. units dienoyl-CoA isomerase. absorbance an extinction coefficient 28,000 M (cid:5) When the time-dependent formation of metabolites (cid:2) M 2- trans -5- cis -tetradecadienoyl-CoA incubated M KP (cid:2) g E. coli FAO complex, m M , m M CoASH absence dienoyl-CoA isomerase units/ml). Reactions were terminated by the pH to 1.5 with 6 N HCl. The pH readjusted 4.5 with 4 N KOH reaction mixtures clarified by filtration through 0.22- (cid:2) m pore size HPLC. Metabolites quantified multienzyme complex provides the opportunity to study in vitro the entry of 2- trans ,5- cis -tetradecadienoyl-CoA into the classical and alternative pathways of (cid:1) -oxidation, because the key reactions of both pathways, the isomerization of the 2,5- to the 3,5-isomer (III to XI) and completion of the (cid:1) -oxidation cycle (III to VI), are practically irreversible (4). Rates of the flux through the classical, isomerase-dependent pathway (III to X) were determined by incubating 2- trans ,5- cis -tetradecadienoyl-CoA NAD (cid:2) and CoASH in the presence of purified FAO complex (13) and measuring the formation of NADH spectrophotometrically at 340 nm. Rates of the entry into the alternative pathway were obtained by measuring at 300 nm the formation 2,4-tetradecadienoyl-CoA, was generated from 3,5-tetradecadienoyl-CoA (XI) dienoyl-CoA isomerase. the conversion of 2- trans ,5- cis -tetradecadienoyl-CoA 3,5-tetradecadienoyl-CoA is practically irreversible (4), the rate of this reaction is a measure of the flux through the alternate pathway. The results demonstrate that (cid:4) 90%

␤-Oxidation of oleic acid in mammalian mitochondria proceeds by two pathways. One is the classical or isomerase-dependent pathway that involves only one auxiliary enzyme, ⌬ 3 ,⌬ 2 -enoyl-CoA isomerase (enoyl-CoA isomerase) 1 (EC 5.3.3.8), in addition to the enzymes required for the degradation of saturated fatty acids (for a review, see Ref. 1). The other is the alternative or reductase-dependent pathway that requires three auxiliary enzymes, namely enoyl-CoA isomerase, 2,4dienoyl-CoA reductase (EC 1.3.1.34), and ⌬ 3,5 ,⌬ 2,4 -dienoyl-CoA isomerase (dienoyl-CoA isomerase), to reductively remove the preexisting double bond of oleic acid (2,3). A recent study of the two pathways reached the conclusion that in rat heart mitochondria more than 80% of oleate is degraded via the classical pathway, whereas the alternative pathway accounts for the remainder of oleate ␤-oxidation (4). That study relied on the use of a mitochondrial extract that permitted the analysis of intermediates but did not maintain the supramolecular organization of enzymes as they exist in intact mitochondria. Since the organization of these enzymes may affect the flux through one pathway relative to the other, the use of an organized ␤-oxidation system uncompromised by its isolation would be advantageous. Such system is present in Escherichia coli, where a multienzyme complex of fatty acid oxidation (FAO complex) is highly expressed when cells are grown on longchain fatty acids as the sole carbon source (5). The purified complex contains the cellular activities of enoyl-CoA hydratase, 3-hydroxyacyl-CoA dehydrogenase, 3-ketoacyl-CoA thiolase, and enoyl-CoA isomerase (6). A study of fatty acid oxidation in E. coli would also reveal whether the alternative pathway of oleate ␤-oxidation, which requires dienoyl-CoA isomerase, exists in prokaryotes. If yes, it may be a ubiquitous process that is operative in all organisms capable of oxidizing fatty acids. These considerations prompted the following study of oleate ␤-oxidation in E. coli. sized from oleic acid, 5-cis-tetradecenoic acid, 2-trans-tetradecenoic acid, and 2-trans-dodecenoic acid, respectively, by the mixed anhydride method as described by Fong and Schulz (15). 2-trans-Tetradecenoyl-CoA was partially converted to L-3-hydroxytetradecanoyl-CoA by hydration in the presence of crotonase in 0.1 M KP i (pH 8.0). The resultant L-3-hydroxytetradecanoyl-CoA was purified by HPLC. 3-cis-Tetradecenoyl-CoA (16) and 3-ketohexadecanoyl-CoA (17) were synthesized as described. 2-trans-5-cis-Tetradecadienoyl-CoA, L-3-hydroxy-5-cis-tetradecenoyl-CoA, 3-keto-5-cis-tetradecenoyl-CoA, L-3-hydroxydodecanoyl-CoA, 3-ketododecanoyl-CoA, and 2-trans,4-trans-tetradecadienoyl-CoA were prepared as published (4). 3,5-cis-Tetradecadienoyl-CoA was synthesized by incubating 5 mol of 5-cis-tetradecadienoyl-CoA in 15 ml of 0.1 M KP i (pH 8.0) with 12 units of acyl-CoA oxidase at room temperature. The progress of the reaction was monitored by HPLC. After completion of the reaction, NAD ϩ , CoASH, 0.2 units of enoyl-CoA hydratase, 0.4 units of 3-hydroxyacyl-CoA dehydrogenase, and 0.4 units of 3-ketoacyl-CoA thiolase were added to remove traces of 2-trans-5-cistetradecadienoyl-CoA by converting it to 3-cis-dodecenoyl-CoA and acetyl-CoA, which were removed by HPLC. All products were purified by HPLC. The pH values of the acyl-CoA preparations were adjusted to ϳ3-4, and the thioester concentrations of these solutions were determined spectrophotometrically by quantification of CoASH with Ellman's reagent (18) after quantitatively cleaving the thioester bond with NH 2 OH at pH 7.0 (15). The concentrations of 2-trans-5-cis-tetradecadienoyl-CoA and 3,5-cis-tetradecadienoyl-CoA were also determined by converting them enzymatically to 2-trans,4-trans-tetradecadienoyl-CoA (4) and calculating the concentration of the latter compound based on its absorbance at 300 nm by using an extinction coefficient of 28,000 M Ϫ1 cm Ϫ1 (19).
Bacterial Growth Conditions-E. coli cells (strain B) were grown on LB medium from single colonies. The initial culture was diluted 5-fold into M9 minimal medium containing 1% (w/v) Tryptone, 2 mM MgSO 4 , 10 M CaCl 2 , 1 M FeCl 3 , and additionally either glucose (0.2-0.5%, w/v), oleic acid (0.1-0.2%, v/v), or palmitic acid (0.1%) in the presence of 0.4% Triton X-100. The cultures were grown at 37°C in a shaker incubator to an absorbance of 1 when they were diluted 20 times into the same growth medium but without Tryptone. The final culture was harvested at an absorbance of 1 by centrifugation at 3,500 ϫ g for 30 min at 4°C. Cell pellets were washed twice with M9 minimal medium and stored at Ϫ80°C.
Preparation and Fractionation of Bacterial Extracts-Seven g of E. coli cell paste were suspended in 14 ml of 0.1 M KP i (pH 7.0) containing 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 10% glycerol; sonicated for a total of 2 min (10 s ϫ 12) at 0°C; and centrifuged at 100,000 ϫ g for 1 h at 4°C. The resultant supernatant was collected for enzyme assays and protein purification. The precipitate was resuspended in 0.1 M KP i (pH 7.0) containing 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, 10% glycerol, and 1% Triton X-100, or Tween 40, or Tween 80, or ␤-D-glucopyranoside and incubated for 30 min on ice. The mixture was centrifuged at 100,000 ϫ g for 1 h at 4°C, and the supernatant was used for enzyme assays. The soluble E. coli extract was dialyzed overnight against 0.02 M Tris-HCl (pH 7.8), containing 10% glycerol and 1 mM benzamidine. The dialyzed supernatant was applied to a DEAE-cellulose column (13.5 ϫ 2.5 cm) equilibrated with dialysis buffer. The column was then developed with a linear gradient made up of 500 ml each of 0.02 M Tris-HCl (pH 7.8) containing either 50 mM NaCl or 500 mM NaCl. Fractions were assayed for thioesterase activity, and those with high activities were pooled and stored at Ϫ80°C.
Metabolic and Enzyme Assays-Rates of degradation of 2-trans-5-cistetradecadienoyl-CoA via the isomerase-dependent pathway were determined by incubating various amounts of 2-trans-5-cis-tetradecadienoyl-CoA in 0.2 M KP i (pH 8.0) with 0.6 g of E. coli FAO complex in the presence of bovine serum albumin (0.1 mg/ml), 1 mM NAD ϩ , and 0.3 mM CoASH and measuring the rate of NADH formation spectrophotometrically at 340 nm. An extinction coefficient of 6,220 M Ϫ1 cm Ϫ1 was used to calculate rates. The conversion of 2-trans-5-cis-tetradecadienoyl-CoA to 2,4-tetradecadienoyl-CoA was measured by incubating the substrate in 0.2 M KP i (pH 8.0) with 0.6 g of E. coli FAO complex in the presence of 0.04 units of dienoyl-CoA isomerase. The absorbance change at 300 nm was recorded, and an extinction coefficient of 28,000 M Ϫ1 cm Ϫ1 was used to calculate rates. When the time-dependent formation of metabolites was studied, 20 M 2-trans-5-cis-tetradecadienoyl-CoA was incubated in 0.2 M KP i (pH 8.0) with 0.6 g of E. coli FAO complex, 1 mM NAD ϩ , 0.3 mM CoASH and in the presence or absence of dienoyl-CoA isomerase (0.04 units/ml). Reactions were terminated by adjusting the pH to 1.5 with 6 N HCl. The pH was readjusted to 4.5 with 4 N KOH before the reaction mixtures were clarified by filtration through 0.22-m pore size membranes and analyzed by HPLC. Metabolites were quantified by the following procedure. Areas under the peaks were obtained by integration with Millennium software from Waters Corp. Peak areas were normalized by use of extinction coefficients of 15,000, 19,650, and 28,800 M Ϫ1 cm Ϫ1 that had been determined for acyl-CoA thioesters with a saturated ␣ carbon, one double bond, and two double bonds in conjugation with the thioester function, respectively, at 254 nm (4). The sum of all normalized peak areas remained fairly constant throughout the experiment. Hence, the sum of all metabolites was 20 M, the concentration of the substrate that was added to the incubation mixture. The normalized area of one peak relative to the sum of all normalized areas gives the percentage of substrate converted to the indicated metabolite. These values are plotted in Fig. 3 and are labeled Metabolites (%). Activities of dienoyl-CoA isomerase were determined by incubating 20 M 3,5-tetradecadienoyl-CoA in 0.2 M KP i (pH 8.0) with various amounts of soluble cell extract or membranes solubilized with detergents (see "Preparation and Fractionation of Bacterial Extracts") from oleate-grown or glucose-grown cells. After recording the absorbance at 300 nm for 2 min, mammalian dienoyl-CoA isomerase (18 milliunits) was added to the assay to determine whether or not the substrate was still present. An extinction coefficient of 28,000 M Ϫ1 cm Ϫ1 was used to calculate rates. Thioesterase was assayed by measuring the release of CoASH from acyl-CoAs with Ellman's reagent (18). A standard assay mixture contained 0.175 M KP i (pH 8), 0.2 mM 5,5Ј-dithiobis(2-nitrobenzoic acid) (Ellman's reagent), and 20 M acyl-CoA. The progress of the reaction was determined spectrophotometrically at 412 nm, and rates were calculated using an extinction coefficient of 13,600 M Ϫ1 cm Ϫ1 .
Isolation and Analysis of Fatty Acids Present in the Growth Medium-E. coli cells were grown to early exponential or to an absorbance of 1 in M9 medium containing oleate (0.1%) and Triton X-100 (0.4%) or were grown to an absorbance of 1 on M9 medium containing glucose (0.5%, w/v) plus Triton X-100 or palmitate (0.1%, w/v) plus Triton X-100. Cells were separated from the growth medium by centrifugation at 2,300 ϫ g for 30 min at 4°C. The supernatant was acidified (pH 1-2) with 2 N H 2 SO 4 and then extracted four times with 100 ml of ether each. The organic phase was extracted with aqueous sodium bicarbonate. After acidifying the aqueous phase with 2 N H 2 SO 4 , it was extracted three times with 8 ml of ether each. The combined ether extracts were dried over anhydrous sodium sulfate, and the residual material, after removal of drying agent by filtration and ether by evaporation under a stream of N 2 , was methylated by reacting it with 2 ml of BCl 3 -methanol (12%, w/w) for 10 min at 60°C. After allowing the reaction mixture to cool down, 1 ml of H 2 O and 1 ml of hexane were added. The organic layer was carefully removed and dried over anhydrous sodium sulfate. The residue after the removal of sodium sulfate and evaporation of ether was dissolved in a minimal volume of anhydrous ethanol. This fraction, which contained the methyl esters of fatty acids that were present in the growth medium, was analyzed by gas chromatography in combination with mass spectrometry (GC/MS). For the purpose of identifying methyl 3,5-tetradecadienoate, a sample containing 20 nmol of 3,5-tetradecadienoyl-CoA and 20 nmol of n-pentadecanoyl-CoA (internal standard) was hydrolyzed by reacting it with 4 N KOH at 25°C for 1 h. The reaction mixture was acidified (pH 1-2) with 2 N H 2 SO 4 and extracted three times with 8 ml of ether each. The extracted fatty acids were converted to their methyl esters as described above. Aliquots of 1 l of the fatty acid methyl esters were injected at 250°C into a GC/MS instrument (Shimadzu Scientific Instruments) consisting of a gas chromatograph (model GC-17A) interphased with a mass spectrometer (QP-5000) and equipped with a capillary column (30 m; inner diameter, 0.25 mm; film thickness, 0.25 m; EC-5; Alltech Associates Inc., Deerfield, IL). The oven temperature was raised from 100 to 230°C at 5°C/min, to 300°C at 20°C/min, and then held constant for 6 min. The mass spectrometer served as a detector and was operated at 280°C.
Purification and Analyses of Acyl-CoA Thioesters by HPLC-Acyl-CoA thioesters were purified, and metabolites were analyzed by reverse-phase HPLC on a Waters Bondapak C 18 column (30 cm ϫ 3.9 mm) attached to a Waters gradient HPLC system. The absorbance of the eluate was monitored at 254 nm. Separation of substrates and metabolites was achieved by washing the Bondapak C 18 column with 50 mM ammonium phosphate (pH 5.5) containing 40% acetonitrile/ water (9:1, v/v) for 20 min and then eluting acyl-CoAs by linearly increasing the organic phase from 40 to 70% in 20 min at a flow rate of 2 ml/min. All samples were cleared of particulate matter by passing them through a 0.22-m pore size membrane before they were injected into the HPLC system. Diluted samples were concentrated by passing them through Sep-Pak C 18 cartridges and eluting them with small amounts of methanol, which subsequently were removed by evaporation under reduced pressure.
␤-Oxidation of Oleic Acid in E. coli multienzyme complex provides the opportunity to study in vitro the entry of 2-trans,5-cis-tetradecadienoyl-CoA into the classical and alternative pathways of ␤-oxidation, because the key reactions of both pathways, the isomerization of the 2,5-to the 3,5-isomer (III to XI) and completion of the ␤-oxidation cycle (III to VI), are practically irreversible (4). Rates of the flux through the classical, isomerase-dependent pathway (III to X) were determined by incubating 2-trans,5-cis-tetradecadienoyl-CoA with NAD ϩ and CoASH in the presence of purified FAO complex (13) and measuring the formation of NADH spectrophotometrically at 340 nm. Rates of the entry into the alternative pathway were obtained by measuring at 300 nm the formation of 2,4-tetradecadienoyl-CoA, which was generated from 3,5-tetradecadienoyl-CoA (XI) by added dienoyl-CoA isomerase. Since the conversion of 2-trans,5-cis-tetradecadienoyl-CoA to 3,5-tetradecadienoyl-CoA is practically irreversible (4), the rate of this reaction is a measure of the flux through the alternate pathway. The results demonstrate that ϳ90% of 2-trans,5-cis-tetradecadienoyl-CoA was degraded via the classical pathway (Fig. 1, curve 1), whereas a small but significant amount (ϳ10%) of this intermediate was converted to 3,5tetradecadienoyl-CoA as measured by conversion to 2,4-tetradecadienoyl-CoA (Fig. 1, curve 2) and thereby diverted from the main pathway. The fraction of 2-trans,5-cis-tetradecadienoyl-CoA entering either of the two pathways was relatively constant over a wide concentration range of this compound. Hence, subsequent studies were carried out at a fixed concentration of 20 M 2-trans,5-cis-tetradecadienoyl-CoA.
The time course for the degradation of 2-trans,5-cis-tetradecadienoyl-CoA by the FAO complex in the presence of NAD ϩ and CoASH was determined by analyzing and quantifying metabolites by HPLC. Since 3,5-cis-tetradecadienoyl-CoA was coeluted with its 2,5-isomer, dienoyl-CoA isomerase was added to the incubation mixture to convert the 3,5-isomer to the 2,4isomer, which is well separated from the starting material. Shown in Fig. 2 is the spectrum of metabolites that were formed during the first minute of the incubation period. All expected metabolites of 2-trans,5-cis-tetradecadienoyl-CoA (see Scheme 1) were detected and identified by use of authentic compounds that were synthesized by chemical and enzymatic reactions. Changes of metabolite concentrations as a function of the incubation time are shown in Fig. 3A. The rapid hydration of the starting material, 2-trans,5-cis-tetradecadienoyl-CoA (III), led to a built-up of 3-hydroxy-5-cis-tetradecenoyl-CoA (IV), because the latter compound seemed to be more slowly dehydrogenated than formed. Since the product of the dehydrogenation, 3-keto-5-cis-tetradecenoyl-CoA (V), only accumulated to a limited extent (see Fig. 2), it seems that initially the dehydrogenation step limited the flux through the pathway. The subsequent build-up of 3-dodecenoyl-CoA (VI) at 1 min followed by a more pronounced accumulation of 2-dodecenoyl-CoA (VII) at 2 min suggests that a competition between these compounds for the hydratase/isomerase active site (20) may restrict the flux through the pathway during the later part of the incubation period. The time course for the degradation of 2-trans,5-cis-tetradecadienoyl-CoA (III) by the FAO complex was quite different when dienoyl-CoA isomerase was omitted from the incubation mixture (see Fig. 3B). Most dramatic was the much slower progress of the degradation process as illustrated by the conversion of ϳ10% of the starting material to decanoyl-CoA (X) during the first 5 min of the incubation, whereas close to 80% of the substrate was converted to decanoyl-CoA (X) during the same time period when dienoyl-CoA isomerase was present (Fig. 3, compare A and B). All metabolites were more slowly formed and degraded in the absence of dienoyl-CoA isomerase. This observation prompts the suggestion that 3,5-tetradecadienoyl-CoA inhibits the FAO complex and hence may interfere with the efficient ␤-oxidation of oleic acid in intact E. coli cells unless this intermediate is further metabolized.
Search for Dienoyl-CoA Isomerase in Extracts of E. coli Cells-An extract of soluble proteins from E. coli cells grown on oleate as the sole carbon source was assayed for dienoyl-CoA isomerase. As shown in Table I, no activity was detected in the extract prepared from cells that express the enzymes of ␤-oxidation at high levels. Dienoyl-CoA isomerase activity was also not observed in an extract of membrane-bound proteins that were solubilized with Triton X-100, Tween 40, Tween 80, or ␤-D-glucopyranoside. As expected, cell extracts prepared from glucose-grown cells also were devoid of dienoyl-CoA isomerase activity. The lower limit of detecting the activity of this enzyme in a soluble cell extract was 0.2 milliunits/mg of protein under conditions used in this study. Since the specific activities of other ␤-oxidation enzymes are 2-3 orders of magnitude higher (5,6), it is unlikely that dienoyl-CoA isomerase is expressed in E. coli at a functional level. This conclusion raises the question of how 3,5-tetradecadienoyl-CoA is metabolized in the absence of dienoyl-CoA isomerase. A clue to an answer was the observation that the apparent activity of mammalian dienoyl-CoA isomerase was lower than expected when it was added to the assay mixture 2 min after the addition of the E. coli cell extract (see Table I). Furthermore, the observed activity of dienoyl-CoA isomerase was lower as the amount of extract was increased (see Table I) or the preincubation time was extended (data not shown). Together, these observations prompted the idea that the substrate of dienoyl-CoA isomerase, 3,5-cis-tetradecadienoyl-CoA, was used up in another reaction catalyzed by the E. coli extract. The most likely reaction that would cause the disappearance of 3,5-cis-tetradecadienoyl-CoA is its hydrolysis catalyzed by a thioesterase, which might be present in the E. coli extract.
Identification of an E. coli Thioesterase Activity with 3,5-Tetradecadienoyl-CoA-The following experiments were prompted by the idea that a thioesterase may be expressed in E. coli cells for the purpose of hydrolyzing metabolites of ␤-oxidation when an accumulation of such intermediates would block the flux through the pathway. If such thioesterase exists, it should be highly expressed in oleate-grown E. coli cells that have a high capacity to oxidize fatty acids but should not be expressed or poorly expressed in glucose-grown cells with a repressed ␤-oxidation system. The search for such thioesterase was initiated by preparing extracts from oleate-grown and glucose-grown E. coli cells and by fractionating them on DEAEcellulose. Fractions were assayed for thioesterase with tetradecanoyl-CoA as the substrate instead of 3,5-tetradecadienoyl-CoA, because the latter compound is difficult to synthesize. This experiment revealed the presence of at least two thioesterases that correspond to peaks I and II (see Fig. 4, A and B). The same two enzymes may be present in either extract, because similar elution patterns were obtained with both extracts. However, an important difference between the two extracts is the 6 times higher specific activity of thioesterase II in oleate-grown cells as compared with glucose-grown cells (compare peaks II of Fig. 4, A and B). In contrast, the specific activities of thioesterase I were nearly the same in both extracts. Thus, growth on oleate seems to induce the expression of thioesterase II but not that of the type I enzyme. An evaluation  b After allowing the reaction to proceed for 2 min in the presence of the indicated amount of E. coli protein, 18 milliunits of rat dienoyl-CoA isomerase (DI) were added, and the assay was continued. ␤-Oxidation of Oleic Acid in E. coli of substrate specificities revealed thioesterase I to be most active with saturated long-chain fatty acyl-CoAs like palmitoyl-CoA, myristoyl-CoA, and stearoyl-CoA (see Fig 5A). The enzyme exhibited significant activity with 3,5-cis-tetradecadienoyl-CoA but was less active with other long-chain intermediates of ␤-oxidation. Thioesterase II also had a preference for long-chain fatty acyl-CoAs. Its highest activity, however, was observed with 3,5-cis-tetradecadienoyl-CoA as substrate. It is noteworthy that thioesterase II effectively hydrolyzed a number of long-chain ␤-oxidation intermediates besides 3,5-cis-tetradecadienoyl-CoA. Good substrates of this enzyme were ␤-oxidation intermediates that have no double bond at the ␣-carbon, like fatty acyl-CoAs, 3-enoyl-CoA, 3-hydroxyacyl-CoA, and 3-ketoacyl-CoA. Thus, it seems that thioesterase II would be well suited to remove a block in ␤-oxidation by hydrolyzing one or several intermediates that might reduce the flux through the pathway because they would either inhibit certain reactions of ␤-oxidation, tie up free CoA, or do both.
Identification of 3,5-Tetradecadienoic Acid in the Growth Medium of E. coli Cells Grown on Oleate-If 3,5-tetradecadienoyl-CoA is a metabolite of oleate in E. coli and is hydrolyzed to CoASH and 3,5-tetradecadienoic acid, the latter compound is expected to exit from cells and accumulate in the growth medium. To test for the presence of 3,5-tetradecadienoic acid in the growth medium, cells were separated from the medium by centrifugation, and the medium, after acidification, was extracted with ether. Neutral material present in the ether extract was removed by extraction with ether under alkaline conditions. The remaining acidic compounds were converted to methyl esters and analyzed by GC/MS. The analysis of this material by GC (see Fig. 6A) demonstrated the presence of many acidic compounds in the medium after growth of E. coli cells on oleate. Fortunately, the region of the chromatogram between 17.5 and 19.5 min, which is important for the identification of methyl tetradecanoate, methyl tetradecenoate, and methyl tetradecadienoate, was relatively uncongested. Shown in Fig. 6B is an expanded view of the region between 18 and 19.5 min where approximately 10 peaks are visible. Mass spectra corresponding to these peaks were analyzed for ions with mass/charge (m/z) ratios of 238 that might be due to the molecular ion of methyl 3,5-tetradecadienoate. Spectra related to four of the peaks that are marked as 1-4 in Fig. 6B were similar to each other and had apparent molecular ions at m/z ϭ 238. Hence, these peaks may correspond to methyl tetradecadienoates. This view was supported by the absence of these peaks from chromatograms that were obtained with extracts of control cultures grown on either glucose or palmitate to midlogarithmic phase or grown on oleate to early exponential phase. For the purpose of identifying the materials that gave rise to the four peaks, methyl 3,5-tetradecadienoate was prepared by hydrolyzing 3,5-cis-tetradecadienoyl-CoA and converting the resultant acid to the methyl ester. The material obtained by this procedure was analyzed by GC/MS. The gas chromatogram showed six significant peaks in the region between 18 and 19.5 min (see Fig. 6C). The mass spectra corresponding to four of the six peaks (see Fig. 6C) had apparent molecular ions at m/z ϭ 238. Moreover, the positions of these four peaks were virtually identical with the positions of the four peaks tentatively attributed to methyl 3,5-tetradecadienoate in the chromatogram of the material extracted from the medium after the growth of E. coli cells on oleate (Fig. 6, compare B and C). In an effort to identify the compounds that gave rise to peaks 1-4 in Fig. 6B, their mass spectra were compared with the mass spectra of methyl 3,5-tetradecadienoates corresponding to peaks 1-4 in Fig. 6C. Since the mass spectrum related to the major peak (peak 3) in Fig. 6B was virtually identical with the spectrum corresponding to peak 3 in Fig. 6C (shown in Fig. 6D is the mass spectrum of the material corresponding to peak 3 in Fig. 6C), the compound that was isolated from the medium after growth of E. coli cells on oleate was most likely 3,5-tetradecadienoate. The stereochemistry of this 3,5-tetradecadienoate is not certain, but it seems reasonable to assume that it may have a 5-cis double bond, because the authentic methyl 3,5-tetradecadienoate was prepared from 3,5-cis-tetradecadienoyl-CoA. The spectra corresponding to peaks 1 and 4 in the chromatograms of Fig. 6, B and C, were very similar, and only the spectra corresponding to peaks 2 of chromatograms B and C showed significant differences. Contaminations may account for some differences between mass spectra. The presence of different stereoisomers of methyl 3,5-tetradecadienoate may cause spectral variations, which additionally may reflect the existence of positional isomers. Overall, the results of these experiments suggest that growth on oleate gives rise to 3,5-tetradecadienoate in the growth medium.

DISCUSSION
The observation that a double bond at position 5 of an unsaturated fatty acid can be reduced by rat mitochondria in the presence of NADPH (21) led to an investigation that resulted in the characterization of an alternate pathway of ␤-oxidation for unsaturated fatty acids with odd-numbered double bonds (2). The degradation of oleic acid via the classical or isomerase-dependent pathway is outlined in Scheme 1A. The alternate pathway diverts from the classical pathway because of the conversion of 2-trans,5-cis-tetradecadienoyl-CoA (compound III in Scheme 1) to 3,5-cis-tetradecadienoyl-CoA (compound XI) catalyzed by enoyl-CoA isomerase. In rats, the 3,5-intermediate is converted by an auxiliary enzyme, named ⌬ 3,5 ,⌬ 2,4 -dienoyl-CoA isomerase (3), to 2-trans,4-trans-tetradecadienoyl-CoA that is reduced in an NADPH-dependent reaction catalyzed by 2,4-dienoyl-CoA reductase to 3-trans-tetradecenoyl-CoA. The latter intermediate is converted by enoyl-CoA isomerase to 2-trans-tetradecenoyl-CoA, which is a substrate of ␤-oxidation

␤-Oxidation of Oleic Acid in E. coli
and is completely degraded by this process. The reduction of the double bond by 2,4-dienoyl-CoA reductase was the reason for naming the alternate pathway the reductase-dependent pathway. The existence of two pathways for the ␤-oxidation of unsaturated fatty acids with odd-numbered double bonds raised the question as to the flux through each of the two branches. Experiments with extracts from rat mitochondria led to the conclusion that the isomerase-dependent pathway accounted for more than 80% of the total flux (4,22). In contrast, an earlier study suggested that the reductase-dependent pathway might be the major pathway (23). Different experimental approaches and interpretations may account for the contradictory conclusions. A concern regarding the use of mitochondrial extracts was the loss of enzyme organization upon extracting or FIG. 6. Identification of 3,5-tetradecadienoic acid in the medium after growth of E. coli on oleate as the sole carbon source. A, gas chromatogram of the methyl esters of the acidic fraction extracted from the growth medium. B, region of the gas chromatogram where methyl 3,5-tetradecadienoate would be eluted. Peaks marked 1-4 have molecular ions with mass/charge ratios (m/z) of 238. C, gas chromatogram of methyl 3,5-tetradecadienoate prepared from 3,5-cis-tetradecadienoyl-CoA. Peaks 1-4 are due to 3,5-tetradecadienoates or isomers with molecular ions at m/z ϭ 238. D, mass spectrum of the material that gave rise to peak 3 of C.
␤-Oxidation of Oleic Acid in E. coli solubilizing mitochondria, a change that may affect the flux pattern. The use of the FAO complex from E. coli alleviates this problem, because the enzymes, which catalyze the reactions that determine the entry of 2-trans,5-cis-tetradecadienoyl-CoA into the two pathways, remain associated during the isolation and purification of the complex. The demonstration that only a small amount of 2-trans,5-cis-tetradecadienoyl-CoA is diverted from the isomerase-dependent pathway in E. coli agrees with the hypothesis that the classical, isomerase-dependent pathway accommodates most of the flux through ␤-oxidation.
It has been suspected that an accumulation of fatty acid metabolites like 3,5-tetradecadienoyl-CoA might cause an inhibition of ␤-oxidation, because the pool of free CoA would be reduced and/or enzymes of ␤-oxidation would be inhibited by metabolites (4). This study proves this prediction to be correct. The kinetics of the degradation of 2-trans,5-cis-tetradecadienoyl-CoA by the FAO complex reveal a severe inhibition of the classical, isomerase-dependent pathway in the absence of dienoyl-CoA isomerase. This inhibition occurs although free CoA is available. Consequently, the inhibition of at least one enzyme of the FAO complex is most likely responsible for the reduced flux through ␤-oxidation. Since this inhibition is observed when dienoyl-CoA isomerase is omitted from the incubation mixture, 3,5-tetradecadienoyl-CoA accumulates and most likely inhibits the FAO complex. It is reasonable to suggest that this compound binds to the hydratase/isomerase active site of the FAO complex (20) and thereby inhibits hydration or isomerization of enoyl-CoA intermediates.
The conclusion that dienoyl-CoA isomerase is not present in E. coli posed the question of how 3,5-tetradecadienoyl-CoA is metabolized to prevent a severe inhibition of oleate ␤-oxidation. The surprising answer was that 3,5-tetradecadienoyl-CoA is hydrolyzed by a thioesterase. This solution of a metabolic problem is simple, and the cost to the organism is only the incomplete oxidation of a small percentage of oleate that is passing through ␤-oxidation. The thioesterase assumed to be responsible for the hydrolysis of 3,5-tetradecadienoyl-CoA was more active with 3,5-tetradecadienoyl-CoA than with any other acyl-CoA tested as substrate, and the activity of this enzyme was induced when E. coli was grown on oleate. Separation of an E. coli extract by chromatography on DEAE-cellulose yielded two thioesterase fractions that were named thioesterase I and II according to the order of their elution from the column. Barnes et al. (24) introduced this nomenclature. Thioesterase I was purified by Barnes and Wakil (25) and later shown to be a periplasmic protein (26). This enzyme should not be able to hydrolyze fatty acyl-CoAs located in the cytoplasm. Thioesterase II also was purified (27) and could be the thioesterase highly active with 3,5-tetradecadienoyl-CoA as substrate. However, it remains to be determined whether the thioesterase fraction used in this study, referred to as thioesterase II, contained only thioesterase II (27) or perhaps contained more than one thioesterase. Several attempts have been made to elucidate the function of thioesterase II (26,28,29). So far, no specific function has been assigned to this enzyme, because the growth properties of E. coli cells seem to be unaffected when thioesterase II is overexpressed or its gene (tesB) is silenced (29). The assumption underlying most of these studies was that thioesterase II might have a function in controlling or editing fatty acid synthesis that takes place with the growing acyl chain esterified to acyl carrier protein. Since thioesterase II exhibits little or no activity with fatty acyl-acyl carrier proteins (28), it is unlikely to function in fatty acid synthesis, but it could be involved in fatty acid oxidation because the substrates and intermediates of the latter process are fatty acyl-CoA thioesters.
The results presented here establish that in E. coli 3,5tetradecadienoyl-CoA is not metabolized by the reductase-dependent pathway as in mammals but is hydrolyzed so that the resultant carboxylic acid can be excreted (see Scheme 1B). This is a simple but not energy-efficient solution for disposing of this metabolite. The emergence of this metabolic shortcut does not support the idea that an alternative pathway of ␤-oxidation is required to accommodate an increased flux through ␤-oxidation. It seems that in E. coli fatty acids with odd-numbered double bonds are efficiently degraded via the classical, isomerase-dependent pathway. It is likely that in mammals too the isomerase-dependent pathway alone assures a flux of fatty acids through ␤-oxidation that is sufficient to meet the energy needs of the organism. If this assumption is correct, the main function of the alternative pathway is the removal of the 3,5intermediate. A unicellular organism like E. coli can easily dispose of fatty acyl-CoAs by hydrolyzing them and excreting the resultant fatty acids from the cell. In a mammal, in contrast, fatty acids that move out of cells enter the circulation, where they bind to serum albumin and are retained as long as they are hydrophobic enough and not further metabolized. The removal of fatty acids that are resistant to ␤-oxidation from an animal would require their partial degradation and conversion to a water-soluble product that could be excreted in the urine. Since such conversion of fatty acids is a multistep process that takes place inside of cells, the easiest disposal of a dead end fatty acyl-CoA may be its complete ␤-oxidation even if an additional enzyme like dienoyl-CoA isomerase is required to facilitate degradation by an alternative pathway.
A comparison of oleate ␤-oxidation in E. coli and mammals prompts the idea that the classical pathway accommodates a sufficient flux through ␤-oxidation when unsaturated fatty acids with odd-numbered double bonds serve as substrates. The formation of 3,5-cis-tetradecadienoyl-CoA may be an unavoidable side reaction due to the presence of enoyl-CoA isomerase, which catalyzes the conversion of 2,5-tetradecadienoyl-CoA to the more stable 3,5-isomer. Since 3,5-cis-tetradecadienoyl-CoA is an effective inhibitor of ␤-oxidation, it must be removed. This is achieved in E. coli by hydrolysis of 3,5-cis-tetradecadienoyl-CoA and excretion of 3,5-tetradecadienoate, whereas in mammals and perhaps in all multicellular organisms, a pathway has evolved for the ␤-oxidation of 3,5-cis-tetradecadienoyl-CoA as the most efficient way for its disposal.
The work described herein represents the first example of a thioesterase serving an essential role in fatty acid ␤-oxidation coupled to oxidative phosphorylation. The enzyme hydrolyzes an intermediate that would inhibit the pathway if allowed to accumulate. Such function of thioesterases has been discussed in the past but not yet demonstrated (30). It is likely that other acyl-CoA intermediates will be identified, which are formed during fatty acid ␤-oxidation in mitochondria and peroxisomes and which are hydrolyzed by thioesterases to maintain a rapid flux through the pathways.