2,4.-Dienoyl Coenzyme A Reductases from Bovine Liver and Escherichia coli COMPARISON OF PROPERTIES*

2,4-Dienoyl-CoA reductases, enzymes of the beta-oxidation of unsaturated fatty acids which were purified from bovine liver and oleate-induced cells of Escherichia coli, revealed very similar substrate specificities but distinctly different molecular properties. The subunit molecular weights, estimated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis were 32,000 and 73,000 for the mammalian and the bacterial enzyme, respectively. The native molecular weights, calculated from sedimentation coefficients and Stokes radii yielded 124,000 for the bovine liver and 70,000 for the bacterial enzyme. Thus, bovine liver 2,4-dienoyl-CoA reductase is a tetramer consisting of four identical subunits. The E. coli 2,4-dienoyl-CoA reductase, however, possesses a monomeric structure. The latter enzyme contains 1 mol of FAD/mol of enzyme, whereas the former reductase is not a flavoprotein. The bovine liver reductase reduced 2-trans, 4-cis- and 2-trans,4-trans-decadienoyl-CoA to 3-trans-decenoyl-CoA. The E. coli reductase catalyzed the reduction of the same two substrates but in contrast yielded 2-trans-decenoyl-CoA as reaction product. Certain other properties of the two 2,4-dienoyl-CoA reductases are also presented. The localization of the reductase step within the degradation pathway of 4-cis-decenoyl-CoA, a metabolite of linoleic acid, is discussed.

2,4-Dienoyl-CoA reductase catalyzes the reduction of 2trans,4-cis-dienoyl-CoA esters (1). These compounds are assumed to arise as intermediates in the course of the degrada- ' H. Schulz, personal communication. tion of unsaturated fatty acids possessing cis-double bonds at even-numbered carbon atoms (1,3,4). The degradation of 2trans,4-cis-dienoyl-CoA esters according to the epimerase pathway as originally proposed by Stoffel and Caesar (13) has recently been questioned (1-3). Difficulties encountered during attempts to purify 2,4-dienoyl-CoA reductase from various sources have prevented studies concerning their molecular characterization. Very recently, we succeeded in purifying the 2,4-dienoyl-CoA reductase from E. coli'and from beef liver mitochondria by means of affinity chromatography techniques ( 10). The present paper describes molecular and catalytic properties of these isolated reductases. The results demonstrate distinct differences in the properties of the bacterial and the mammalian enzyme.
Other unsaturated carboxylic acids were prepared as reported previously (1). For the synthesis of acyl-CoA thioesters, the mixed anhydride method was used (16), and for their purification the Pullman method was employed (17).
From Bovine Liver-Bovine liver was obtained from the local slaughterhouse. The mitochondria were isolated according to Brosnan et al. (19). Acetone dry powder was prepared as described by Dahlen and Porter (20) and stored a t -18 "C. Crude extracts of soluble mitochondrial proteins were prepared as described previously (1). 20 g of acetone dry powder were extracted with 100 ml of buffer B and centrifuged a t 100,000 X g. The supernatant contained about 100 units of 2,4-dienoyl-CoA reductase activity. It was applied to a blue Sepharose CL-GB column (5.0 X 8.0 cm) which was equilibrated with buffer B. After washing, the enzyme was eluted with 1 M KC1 in buffer B. The desalted eluate (Sephadex G-50) was applied to a Matrex gel red A column (2.6 X 20.0 cm) and washed with buffer B containing 330 mM KCI, and the 2,4-dienoyl-CoA reductase was eluted in a sharp peak with 5 mM NADP and 330 mM KC1 in buffer B. The fractions containing the activity were pooled and separated from NADP and KC1 by chromatography on Sephadex G-50. Then the sample solution was applied to a 2',5'ADP-Sepharose 4B column (2.2 X 3.0 cm) and eluted with 500 ml of 5 HM 2-trans,4-transdecadienoyl-CoA. Concentration was achieved by coupling a small Matrex gel red A column (0.9 X 4.0 cm) directly to the 2',5'ADP-Sepharose 4B column. The enzyme eluted from the 2',5'-ADP-Sepharose and subsequently bound to the small Matrex gel red A column, was eluted with 5 ml of 2 M KC1 in buffer B, desalted on a small Sephadex G-25 column, and stored at -18 "C. This purification procedure yielded a homogeneous enzyme as demonstrated previously (10).

Enzyme Assays
Determinations of 2,4-dienoyl-CoA reductase as well as of the usual @-oxidation enzymes were performed as described previously (1). Activities of catalase, fumarase, malate, and lactate dehydrogenases were measured according to established procedures described in Ref. 21.

Absorption Spectra and Flavin Determination
Absorption spectra of 2,4-dienoyl-CoA reductases were recorded with a Shimadzu spectrophotometer, Model UV 300, at a scan speed of 150 nm/min.
The nature of the flavin moiety dissociated from the enzyme protein by heating at 100 "C for 3 min was analyzed by thin layer chromatography on precoated cellulose plates (0.1-mm thickness). The solvent system used was 1-butano1:acetic acidwater (4:3:3). The FAD content was quantitatively calculated from the spectrum of the supernatant after heating using a molar extinction coefficient a t 445 nm of 11,300 M" cm" (22).

Protein Determination
Protein was assayed using two procedures. Bovine serum albumin served as standard. The spectrophotometric method of Murphy and Kies (23) was used routinely. As an additional proof, the method of Bradford (24) was applied.

Molecular Wezght Determination
Sodium dodecyl sulfate-polyacrylamide slab gel electrophoresis was carried out essentially by the method of Laemmli (25). The running gel contained 10% (w/v) acrylamide and the stocking gel was 5%.
Gel filtration was performed according to Andrews (27) on a Sephadex G-200 column (1.6 x 63.0 cm) equilibrated with buffer B. The column was calibrated with protein markers. Stokes radii of the standard proteins were taken from the literature (28). For sucrose density gradient centrifugation, the enzymes (0.2 ml) were layered on a linear gradient (11.2 ml) of 5-20% sucrose in buffer B and centrifuged a t 4 "C in an SW 41 rotor (Beckman Ultracentrifuge, Model L 5-75) a t 40,000 rpm for 15 h. The tubes were eluted from the bottom and 10-drop fractions were collected and assayed for enzyme activities. Standard proteins were included in each run. Migration distances (r) were calculated from the elution volumes.
Sedimentation velocity experiments were performed using a Beckman Spinco analytical centrifuge, Model E, equipped with an electronic speed control, an ultraviolet scanner, a rotor temperature control unit, and a double-sector cell. The optical density a t 280 nm as a function of distance in the cell was scanned at 8-min intervals. Prior to sedimentation, samples were dialyzed for 20 h a t 4 "C against 50 mM potassium phosphate, pH 7.4, containing 1 mM EDTA and 0.1 mM dithioerythritol. The sedimentation velocity runs were carried out at 56,000 rpm a t 5 "C. Temperature corrections were made according to Ref. 29. The sedimentation coefficients (sz0,+,) were calculated according to the method of Schachman (30).

Incubations and Analyses of Products
Incubations and the analyses of the formed acyl-CoA esters by means of radioactive gas chromatography were performed as described elsewhere (1). Since methyl 3 4 s -and 3-trans-decenoate could not be distinguished by the gas chromatographic procedure employed, the separation of the two isomers was carried out by thin layer chromatography according to Stoffel and Ecker (31).

Molecular
Weights-The native molecular weights of the 2,4-dienoyl-CoA reductases were determined by gel filtration (Fig. 1A) according to Andrews (27) and sucrose density gradient centrifugation (Fig. 2) following the procedure of Martin and Ames (32). Since the values obtained by the two methods did not agree closely, the molecular weights were also calculated by the method of Siegel and Monty (28)   Sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed one protein band for each of the two purified reductases with an apparent M , = 32,000 and 73,000 for the mammalian and bacterial enzymes, respectively (Fig. 3).
Mizugaki et al. (12) recently published a native molecular weight for the E. coli enzyme of 50,000 (determined by gel filtration of Sephacryl S-200 superfine) but did not report a molecular weight determined by electrophoresis under dena- turing conditions. This value for the bacterial enzyme is even smaller than the one of 60,000 found in the present study by gel filtration (Fig. 1A).
Absorption Characteristics and Flavin Content-The absorption spectra for the 2,4-dienoyl-CoA reductases from E. coli and from bovine liver are shown in Fig. 4. water (2 X loG cpm/pmol) (type 3). When the consumption of NADPH had slowed down markedly, the incubations were stopped by 0.2 ml of 2 N NaOH. Fatty acid CoA esters were converted into methyl esters, and aliquots were subjected to gas chromatographic analyses as described previously (1) tase. However, thin layer chromatography (Fig. 6) employing silica gel plates impregnated with silver nitrate (20%) and the solvent system hexane:diethyl ether (95:5) allowed complete separation of the 3-cis-and 3-trans-isomers. The results of Fig. 6 unequivocally demonstrate that, irrespective of whether 2-trans,4-cisor 2-trans,4-trans-decadienoyl-CoA was reduced in the presence of (4S)-[4-3H]NADPH, the product of the bovine reductase was 3-trans-decenoyl-CoA. Kinetic Properties and Substrate Specificities-The dependence of the activities of 2,4-dienoyl-CoA reductases purified from bovine liver and E. coli on substrate and coenzyme concentrations showed for both enzymes a strong inhibition. This is shown in Fig. 7 for 2-trans,4-trans-decadienoyl-CoA as substrate. This implies that there is for both enzymes only a small concentration range of the substrate as well as the coenzyme in which one can expect linearity of the initial reaction rates with time. As a practical consequence, the concentrations of substrate and coenzyme used for the enzyme assays are critical.
NADH is a noncompetitive inhibitor of E. coli 2,4-d1enoyl-CoA reductase with an inhibitor constant (K,) of 1.1 x M (Fig. 7 0 ) , but it does not inhibit the bovine liver enzyme. The effect of 2-trans-decen-4-ynoyl-CoA on the activities of the reductases was tested. Lineweaver-Burk plots (Fig. 7  and B ) in the presence and absence of this inhibitor revealed competitive inhibition (KI = 6.7 X M) with respect to 2trans,4-trans-decadienoyl-CoA for the bacterial enzyme and a mixed inhibiton type for the bovine liver reductase. This particular unsaturated acyl-CoA ester was originally chosen by us as a potential substrate to study the influence of a triple bond in position 4. However, neither the bacterial nor the mammalian enzyme reduced 2-trans-decen-4-ynoyl-CoA (data not shown). The effect of three different sulfhydryl inhibitors on the activity of the two reductases is shown in Table 111.
Induction of E. coli 2,4-Dienoyl-CoA reductase-To determine whether and to which extent 2,4-dienoyl-CoA reductase from E. coli is being induced, cell-free extracts from cells grown on different carbon sources were assayed for reductase activity. As shown in Table IV, activity of 2,4-dienoyl-CoA reductase prepared from cells grown on oleate was highest, about 10 times higher than those obtained from cells grown on acetate medium. The addition of glucose to the synthetic medium either as sole carbon source or in addition to oleate yielded cells whose extracts had depressed levels of 2,4-dienoyl-CoA reductase activity.

DISCUSSION
Not until very recently has it been possible to purify 2,4dienoyl-CoA reductases from two different sources to appar-  (1 ml) contained 100 WM NADPH and 2,4dienoyl-CoA reductase from bovine liver or E. coli in 50 mM potassium phosphate buffer, pH 7.4, containing 1 mM EDTA, and 1% bovine serum albumin. Inhibitors were dissolved in the same buffer and added as indicated. After 3 min of preincubation, the reactions were started by adding 40 nmol of 2-trans,4-trans-decadienoyl-CoA.  IV Specific and relatiue activities of 2,4-dienoyl-CoA reductase in cellfree extracts of E. coli grown on various carbon sources The media used for growing E.coli consisted of a synthetic medium without a carbon source according to Lengeler (18) and the indicated carbon sources. NB-medium is an optimal medium with 8 g of Nutrient Broth (Difco)/liter. The cells were grown to the late exponential phase. Cell-free extracts were prepared as published previously (10). Oleate + glucose (1 1.6 3.9 g/liter + 5 g/liter) NB-medium ent homogenity (10). In the present studies, these two 2,4dienoyl-CoA reductases, from E. coli and bovine liver, have been characterized with respect to their molecular and kinetic properties. Although some similarities have been noted between the two enzymes, comparison of their molecular properties (Table V) reveals considerable differences.

Activity of 2,4-dienoyl-CoA
First, the two 2,4-dienoyl-CoA reductases possess entirely different structures. The native molecular weights calculated from the sedimentation coefficients and the Stokes radii were in very good agreement with four times and one time the subunit molecular weights of bovine liver and E. coli reductase, respectively. Therefore, the mammalian enzyme has a tetrameric structure with four identical subunits whereas the bacterial enzyme consists of one polypeptide chain. Second, the reductase from bovine liver contains no flavin, while the enzyme purified from E. coli is a flavoprotein containing 1 mol of FAD/mol of enzyme. As it is known that hydrogen from reduced FAD readily exchanges with that of water, the absence of flavin might explain why the mammalian enzyme catalyzes the transfer of tritium from tritiated NADPH to the substrate, while the flavin-containing bacterial reductase does not.
A third major difference, and perhaps the most surprising aspect of this comparative characterization, concerns the reaction products. Bovine liver 2,4-dienoyl-CoA reductase catalyzes the reduction of 2-trans,4-cis-decadienoyl-CoA and 2trans,4-trans-decadienoyl-CoA to 3-trans-decenoyl-CoA (Table I). Thus, the overall reaction is a 1,4-addition of hydrogen across the 2,4-diene system. In contrast, the bacterial enzyme catalyzes a 1,2-addition of hydrogen to the double bond in position 4 of 2,4-dienoyl-CoA esters, yielding 2-trans-enoyl-CoA esters as reaction products ( Table I). The latter result is in agreement with a recent report of Mizugaki et al. (12) who described 2-trans-decenoyl-CoA as the reduction product of the E. coli reductase reaction. This property of the bacterial reductase seems to be a unique feature among the ubiquitously occurring 2,4-dienoyl-CoA reductases. This is because, in addition to the bovine liver enzyme described in this study, the peroxisomal 2,4-dienoyl-CoA reductase from rat liver (4) as well as the peroxisomal 2,4-dienoyl-CoA reductase from C. tropicalis (3), all form 3-enoyl-CoA esters. These results seem to indicate that eucaryotic 2,4-dienoyl-CoA reductases form 3-enoyl-CoA esters. Therefore, it is unlikely that the reaction product of the mitochondrial 2,4-dienoyl-CoA reductase from rat liver is the 2-trans-enoyl-CoA ester as assumed by others (7,8,34). The different mechanisms of reduction might be reflected in the different types of inhibition evoked by 2trans-decen-4-ynoyl-CoA on the reactions of E. coli and bovine liver 2,4-dienoyl-CoA reductases ( Table 11). This interesting point clearly needs further investigations to be established.
In contrast to the differences described above, the substrate specificities of both 2,4-dienoyl-CoA reductases are very similar (Table 11). They are strictly NADPH-dependent and require a 2,4-dienoyl-CoA structure as part of their substrate molecules. The corresponding monounsaturated acyl-CoA esters with a cis-double bond in either position 4 or 2 are not reduced. 2-trans-Decen-4-ynoyl-CoA is not a substrate but inhibits both enzymes (Fig. 7, A and B ) . Both reductases are inhibited by substrate and coenzyme at higher concentrations (Fig. 7). The chain length of the substrate seems less critical. Both reductases reduced 2,4-dienoyl-CoA esters with 10 and 6 carbon atoms although at different rates. Borrebaek et al. (6) as well as Hiltunen and Davis (7) showed indirectly that 2,4-dienoyl-CoA reductase(s) from rat liver reduces 2,4-pen- tadienoyl-CoA by measuring in crude extracts NADPH consumption in the presence of this acyl-CoA ester. The reduction of the same compound has also been observed by Schulz in extracts from E. coli' and by Schulz (9) and Hiltunen and Davis (7) in rat heart mitochondria. Previous experiments from this laboratory (35)(36)(37)(38) on partial degradation of highly unsaturated fatty acids with their first double bonds in position 4 (e.g. 4,7,10,13,16-docosapentaenoic acid) and data recently published by Osmundsen et al. (34) suggest that 2,4dienoyl-CoA reductase also participates in the metabolism of polyunsaturated fatty acids with chain lengths of more than 18 carbon atoms.
The 2,4-dienoyl-CoA reductase from E. coli is induced when the cells are grown on oleate as sole carbon source and depressed by glucose in the growth medium (Table IV). The extent of induction was essentially the same as reported by Overath et al. (39,40) and Weeks et al. (41) for the four "classical" @-oxidation enzymes. Since oleate, a fatty acid without a double bond at an even-numbered position, does not require 2,4-dienoyl-CoA reductase for its degradation, but induces its activity together with those of the other @-oxidation enzymes, some form of unit control is suggested. The same induction pattern has been found for the P-oxidation enzymes of C. tropicalis ( 3 ) .
The determination of the reaction products allows us to complete the reaction sequence which we have previously published (1) for the degradation of 4-cis-decenoyl-CoA, a key metabolite of linoleic acid, in bovine liver, and to compare it to the catabolic route in E. coli (Fig. 8). In bovine liver, 2,4decadienoyl-CoA is reduced to 3-trans-decenoyl-CoA which in turn is isomerized to 2-trans-decenoyl-CoA. Thus, two reactions are necessary between the first and the second step of the P-oxidation cycle in order to metabolize the double bond in position 4 (Fig. 8A). In E. coli, in contrast, the isomerization is not necessary as 2-trans-decenoyl-CoA is the direct reduction product of 2-trans,4-cis-decadienoyl-CoA (Fig. 88).
There is growing evidence (1)(2)(3)(4)34) that these reaction sequences of the reductase pathway are obligatory for removal of all cis-double bonds located a t even-numbered carbon atoms of polyunsaturated fatty acids. The molecular reason seems to be that 2-trans,4-cis-dienoyl-CoA esters cannot serve as substrates for enoyl-CoA hydratase (2,3, 7 , 8) required for the epimerase pathway proposed by Stoffel and Caesar (13). As, however, 3-hydroxyacyl-CoA epimerase activity in E. coli is located on one of the two multifunctional polypeptides comprising five @-oxidation activities (42)(43)(44), the interesting question is raised as to the physiological function of the 3hydroxyacyl-CoA epimerase in the degradation of the various naturally occurring fatty acids. At present no relevant experimental data are available.
In conclusion, although 2,4-dienoyl-CoA ieductases from bovine liver and E. coli reduce the same substrates and serve the same metabolic function, their molecular structures and the reactions they catalyze are very different. That makes these enzymes not only interesting as parts of the P-oxidation systems but also from the point of view of reaction mechanisms.