Oxidation of Zl-Dehydrocorticosteroids to Steroidal 2O-Oxo-21-oic Acids by an Aldehyde Dehydrogenase of Sheep Adrenal*

Abstract Sheep adrenal cortex contains an enzyme which catalyzes the oxidation of steroidal 20-oxo-21-al derivatives to 20-oxo-21-oic acids. The enzyme was purified 60- to 70-fold and retained broad specificity for a variety of aliphatic and aromatic aldehydes. The aldehyde dehydrogenase was localized in the cytosol. It was protected from inactivation by NAD+, which it uses as specific coenzyme, but was otherwise unstable. Reagents that react with —SH groups inactivated the enzyme. Molecular weight was estimated as 164,000. The product of 21-dehydrocorticosterone oxidation was 11β-hydroxy-3,20-diketo-4-pregnen-21-oic acid. For each mole of product made or substrate used, 1 mole of NAD+ was reduced.

The enzyme was purified 60-to 'IO-fold and retained broad specificity for a variety of aliphatic and aromatic aldehydes.
The aldehyde dehydrogenase was localized in the cytosol.
It was protected from inactivation by NADf, which it uses as specific coenzyme, but was otherwise unstable.
Reagents that react with -SH groups inactivated the enzyme.
Molecular weight was estimated as 164,000.
Studies in a number of laboratories have shown that corticosteroids are oxidized to steroidal carboxylic acids by human and other mammalian tissues in vitro and in viva. The three classes of acidic steroid which have been isolated after exposure of corticosteroids directly to tissues or to partially purified enzymes are 17&carboxylic (l-6), 20-hydroxy-21-oic (5,7), and 20-oxo-21-oic (8) acids. The isolation and characterization of radioactive steroid acids from the urine of patients given tracer doses of corticosteroids has been accomplished (9)(10)(11)(12). The extent of conversion of corticosteroids to these various metabolites is quantitatively significant in all cases. A possible pathway leading to some of these end products may involve the following sequence: (R)-20-0x0-21-01 -+ (R)-20.0x0-21-al --f (R)-20-oxo-21-oic acid ---f R-20-oic acid, where (R) represents the fused ring system. This pathway requires that the 21-dehydro- (i.e. steroids containing the 20-oxo-21-al side chain) be metabolic intermediates in the synthesis of the 20oxo-21-oic acids or the 20-oic acids (5,13). In liver, the 21-dehydrosteroids are oxidized by an NAD+-dependent enzyme to keto acids (8). A similar enzymic conversion may in part account for the keto acids found in the urine of patients who were given synthetic corticosteroid alcohols (9). Oxidative decarboxylation of the keto acids would lead to 20-oic (etienic) acids, and thus provide an explanation for reports that corticosteroids perfused through liver are converted to 20-carbon acids (2,5). The adrenal gland has also been found to convert corticosteroids to steroidal carboxylic acids (1,3,4,14,15). It was anticipated, therefore, that the adrenal would have enzyme activity corresponding to that found in the liver.
In this paper we describe the properties of an enzyme of broad specificity obtained from the cortices of sheep adrenal glands, which catalyzes the oxidation of 21-dehydrocorticosteroids to 20-0x0-21-oic acids.

MATERIALS AND METHODS
Sheep adrenals were obtained from a local slaughterhouse and kept frozen until used. Ammonium sulfate, enzyme grade, was bought from Schwarz-Mann.
Sephadex G-25, G-100, G-200, and A-50, purchased from Pharmacia Fine Chemicals, were prepared as recommended by the manufacturer.
All water was twice distilled with a Corning all glass still, model AG-2. NADH, NADPH, NAD+, and NADP+ were obtained from P-L Biochemicals.
Broken cell suspensions of adrenal cortex were prepared and fractionated as described by Sweat et al. (16) using a Teflon pestle fitted into a smooth wall glass tube. Each fraction was reconstituted to the original volume of homogenate with cold 0.25 M sucrose.
Activity was measured in a system containing 0.75 pmole of NADf, 0.29 pmole of 21-dehydrocorticosterone, and 0.1 ml of enzyme in 0.2 M Tricine,2 pH 8.4. It is important that this sequence of adding the components be followed, for in the absence of NAD+, the dilute enzyme was slowly inactivated, and the initial rates were not consistently reproduced.
The 21-dehydro derivatives of other corticosteroids are named in a corresponding manner. * The abbreviation used is : Tricine, N-tris (hydroxymethyl)methylglycine.
8547 ume was 1.0 ml. Rate of reduction of NAD+ was measured continuously at 340 nm with a Gilford model 2000 multiple absorbance spectrophotometer.
Concentrations of other substrates used are indicated under "Results." Steroids were bought from Steraloids, Inc. The 21 -dehydrocorticosteroids were synthesized as described previously (17). Methylglyoxal was purified (18) and the concentration determined by the method of Racker (19). Other aldehydes used in this study were purchased commercially and used without further purification.
Disc gel electrophoresis was performed in a Canalco apparatus.
The running gel consisted of 1 part of an aqueous mixture of 30% acrylamide and 0.8% bis-acrylamide; 2 parts of a solution of 0. 15  activities were combined, and an aliquot of 6.3 ml was applied to a column (35 x 0.9 cm) of Sephadex A-50 previously equilibrated with 0.2 M Tricine, pH 7.5. Fractions of 2 ml were collected. Elution was initiated with 0.2 M Tricine, pH 7.5, followed by a linear gradient of sodium chloride in the same buffer. The reservoirs contained 100 ml of buffer, and 100 ml of 0.5 M sodium chloride in buffer, respectively.
The fractions of highest specific activity (fractions 27 to 36) were combined.
Further attempts at purification with CM-cellulose, DEAE-cellulose, alumina gel C-y, calcium phosphate gel, ethanol, or acetone invariably led to losses in total activity and specific activity. The initial ammonium sulfate fractionation removed inhibitory material since the total activity increased 6.5-fold as a result of this step. Specific activity increased 60-to 70-fold during the purification.
Activity of the enzyme was also followed with 7.6 RIM methylglyoxal.

RESULTS Procedure
Preparation of Enzyme-Sheep adrenals were found to have 5 to 10 times more activity than those of pig, cow, or rat. Initial attempts to purify the enzyme from whole sheep adrenals were frustrated by the irreversible inactivation of some preparations by an unidentified component in the medulla. The inhibitor was not epinephrine. Therefore, before fractionation was started, chilled adrenal glands from sheep (or lambs) were split and the cortical tissue was carefully separated from the medulla and capsule. Subsequent steps in the purification were done at 3". A combined total of 13 g of cortical tissue was homogenized in 25 ml of 0.1 M sodium phosphate, pH 7.5, containing 0.0025 M EDTA.
The suspension was centrifuged at 27,000 x g for 30 min. Solid ammonium sulfate was added to 25 ml of the resulting supernatant fluid to a final concentration of 35% of saturation (1.4 M). The preparation was stirred for 15 min, then centrifuged at 10,000 X g for 10 min. The precipitate was discarded.
Additional ammonium sulfate was slowly added with stirring to the supernatant fluid until 60% of saturation (2.4 M) was reached.
Fifteen minutes later, the suspension was centrifuged at 10,000 X g for 10 min, and the supernatant layer was discarded.
The precipitate was dissolved in 6 ml of 0.1 M sodium phosphate, pH 7.5. The solution was passed through a column (50 x 0.9 cm) of Sephadex G-100 which had previously been equilibrated with 0.2 M Tricine, pH 7.5. Flow rate was 7.5 ml per hour. Each fraction contained 4.5 ml. The enzyme emerged in fractions 12 to 18 were compared with controls containing water.
to 0.005 M EDTA, 0.001 M /Smercaptoethanol, 0.002 M glutathione, 20% glycerol, or suspension in 3.0 M ammonium sulfate. A solution of 0.6% metal-free gelatin stabilized enzyme activity for at least 3 days, but the solutions were viscous and difficult to pipette.
For this reason and because gelatin represented a source of additional protein contamination, this method of preserving the dehydrogenase was discarded. Enzyme was always prepared fresh and used as quickly as possible.
Freezing and thawing the enzyme solution for repeated cycles in a Dry Ice-acetone bath did not affect activity.
Lyophilization completely destroyed the enzyme. The enzyme lost no activity in 3 hours at 30" in 0.1 M sodium phosphate buffer, pH 7.5. Raising or lowering the pH accelerated the destruction of the enzyme, and this was not reversed by incubation at pH 7.5. In 0.1 M sodium acetate, pH 4.0, 55% of the activity was lost in 5 min at 30", and 70% in 30 min. At pH 9.0, the inactivation was much slower.
In 0.1 M carbonate, a decrease of 32y0 was found in 30 min. In all cases, the rates of loss in activity were the same with methylglyoxal and 21-dehydrocorticosterone. Heat Stability- Fig.  1 shows that at 55" and 60" in 0.1 M Tricine, pH 7.5, the dehydrogenase was completely inactivated within 2 min, and at 50", within 10 mm. Addition of 3 x 10m3 M 21.dehydrocorticosterone slowed the inactivation rate to about half that in buffer alone. The enzyme was completely protected from heat inactivation by 1.5 x 10h3 M NAD+.
pH Dependence-The pH activity curves with 21-dehydrocorticosterone as substrate are reproduced in Fig. 2. In barbital and borate buffers the enzyme was inactive over the range of pH 7.2 to 10.0. In Tricine, a maximum occurred at pH 8.5. Further increased activity was seen above pH 8.8. At pH values above 9.0, 21-dehydrocorticosterone slowly turned yellow spontaneously.
The color change was in part due to small amounts of glycine in the Tricine buffer which reacted with the steroid at high pH values, and in part to self polymerization In carbonate buffer, no activity was seen until the pH exceeded 8.5; then a steady rise occurred with no optimum or plateau.  3 shows that the rate of oxidation of either methylglyoxal or 21-dehydrocorticosterone was proportional to enzyme concentration. The lack of curvature indicated that the purified preparations contained no detectable reversible inhibitors or activators.
Cofactor Specificity-With 21-dehydrocortisol or methylglyoxal as substrate, NADP+ was totally inactive up to a concentration of 0.26 pmole per ml final concentration.
The NADPf had no effect on the reduction of NAD+.
Other Enzymes of Steroid Metabolism-No 21-hydroxysteroid dehydrogenase activity was detected in purified enzyme preparations.
Also absent were 2Oor-and BOP-hydroxysteroid dehydro- Substrate Specificity- Table  III shows that steroidal keto aldehydes were readily oxidized by the enzmye.
The corresponding steroidal ketols were found to be totally inactive.
No correlations could be made between the oxidation rates and the steroid structures.
K, Values-The apparent K, values for several substrates were determined at an NAD+ concentration of 0.75 mM.a The double reciprocal plots did not deviate from linearity (Fig. 4). At 2 X lo+ M 21.dehydrocorticosterone, the K, value was 3.7 X lop5 M.
Nonsteroid Substrates-Unlike liver ketoaldehyde dehydrogenase, which specifically oxidized a-ketals (8,18), the substrate specificity of the adrenal enzyme was broad. Table IV shows that both simple aliphatic and aromatic aldehydes were oxidized. Relative rates of oxidation for the various substrates studied did not change significantly during enzyme purification, although relative activity of the enzyme with respect to formaldehyde appeared to increase.
Glyoxylic acid and o-aminobenzaldehyde were inactive.
Disc Gel Electrophoresis-The purified dehydrogenase was subjected to disc gel electrophoresis in order to determine the degree of heterogeneity of the preparation, and to see whether the multiple substrate specificities were due to a single enzyme. The gel was developed with substrate, NAD+, blue tetrazolium, and phenazine methosulfate, and scanned on a Gilford 2000 3 The double reciprocal plots of all substrates, with NAD+ as the constant variable, yielded a series of parallel lines. A similar pattern was obtained with steroid as constant variable.
These data have been examined in conjunction with information derived from product inhibition studies. It has not yet been possible to obtain a meaningful interpretation of the results of these studies. . . 3~~,17ol-Dihydroxy-ll,20-dioxo-5p-pregnan-21-a1.
The protein distribution was heterogeneous. A major peak coincided with the enzyme activity and constituted 75 to 80% of the total protein.
It was concluded that a single enzyme of broad specificity oxidized all of the substrates tested.
MoZecuZur Weight-The apparent molecular weight of the enzyme was determined by gel filtration.
The position at which activity emerged was compared with absorbance peaks at 280 nm of known standard proteins emerging concurrently, according to the method of Andrews (21). From the relationships shown in Fig. 6, the mean molecular weight of the unknown was estimated to be 164,000 f 9,500 for three determinations.
Identi,ficalion of Steroid Product-A mixture containing 5 mg of 21-dehydrocorticosterone and 25 mg of NAD+ in 0.2 M Tri-tine, pH 9.2, was incubated with 2.5 ml of purified adrenal aldehyde dehydrogenase at 27' in a volume of 15 ml. Progress of the reaction was followed spectrophotometrically. At intervals, more enzyme or NAD+ was added as required to stimulate the  After 24 hours, the pH was adjusted to about 4 with glacial acetic acid and the incubation mixture was extracted with ethyl acetate.
The organic extract was washed with water, dried by the addition of anhydrous sodium sulfate, and evaporated to a small volume.
The acid was isolated by preparative layer chromatography on silica gel using ethyl acetate-formic acid (99:1, v/v) as the developing solvent. Steroid was extracted from the silica gel with methanol-chloroform (1: 1) containing a few drops of glacial acetic acid and dried under a stream of nitrogen.
The residue was dissolved in 5% potassium bicarbonate, then acidified with 10% hydrochloric acid. White needles separated out overnight at 3". The crystals were washed with dilute acetic acid and dried in vacua over phosphorus pentoxide.
From 10 mg of 21-dehydrocorticosterone, a total of 6 mg of crystalline product was obtained.
A control incubation containing no NAD+ yielded no acid product.
Both enzymically and chemically synthesized steroids had identical infrared spectra (Fig. 7). In sulfuric acid, maxima at 442, 370, 285 nm, shoulder at 420 nm, and minimum at 382 nm, corresponded to the spectrum of authentic steroid.
CdLsOs Calculated: C 69.9, H 7.8 Found : C 69.7, H 8.0 The free steroid acid was esterified by reaction with diazomethane.
Conversion to ester was quantitative. The spectra of the methyl esters of the enzymically and chemically prepared steroids were identical.
Stoichiometry-In order to establish the quantitative relationships between the components of the reaction, a complete system containing 2.9 pmoles (2.0 X lo5 cpm) of 21-dehydro-[4-14C]corticosterone, 1.5 pmoles of NAD+, 0.2 M Tricine, pH 9.5, and enzyme in a volume of 3.0 ml was prepared.
A parallel control contained no NAD+.
At intervals, 0.3 ml of the incubation mixture was pipetted into 0.2 ml of glacial acetic acid. The mixture was diluted to 1.0 ml with water and extracted repeatedly with ethyl acetate to remove all radioactivity from the aqueous phase. The organic layer was dried over sodium sulfate, concentrated, and transferred to thin layer plates. Steroids were resolved with ethyl acetate-formic acid (99: 1). The radioactivity on developed plates was measured on a Packard model 7201 radiochromatogram scanner.
The incubation was allowed to proceed for 24 hours with periodic additions of enzyme or NAD+.
The final absorbance of the complete system was 1.64, and for the control, 0.083 absorbance units. Fig. 8 shows the profile of radioactivity during the course of the oxidation.
Initially two radioactive peaks were present in incubations performed in Tricine buffer; one of these was highly polar.
Re-examination of the 21-dehydrocorticosterone con-firmed that the steroid was chromatographically homogeneous and contained no impurities.
The polar peak did not appear in carbonate buffer. Tricine was analyzed on an amino acid analyzer and was found to contain 2.1 pmoles of glycine per mmole of buffer. Thii amount of glycine added to the carbonate buffer containing 21-dehydrocorticosterone resulted in the appearance of the polar material on thin layer plates. This is, therefore, most probably a condensation product of steroid aldehyde and amino acid. AS the reaction proceeded, the glycinebound steroid dissociated.
At the end of the incubation, no complex was left. The stoichiometry of the reaction is presented in Table V. It is concluded that all components reacted in an equimolar relationship.
No keto acid was formed in the control incubations.
Alternative Pathways-In an earlier paper (18) it was shown that sheep liver ketoaldehyde dehydrogenase converts methylglyoxal to pyruvic acid by direct oxidation and not via prior isomeriaation to lactic acid. The oxidation of 21-dehydrocorticosterone to keto acid could also theoretically occur through a hydroxy acid intermediate.
The  I  I  I  I  I  I  1  I  I  I oic acid (----). The incubation mixture contained initially 2.9 kmoles (2.0 X 106 cpm) of 21-dehydro-[4-14C]corticosterone, 1.5 pmoles of NAD+, 0.2 M Tricine, pH 9.5, and 0.3 ml of enzyme (1.8 mg of protein per ml) in a final volume of 0.3 ml. At intervals, 0.3-ml samples were anaIyzed as described in the text. Sampling times were: S1, 0 hour; SZ, 0.5 hour; SZ, 1 hour; S4, 2 hours; SS, 3 hours; S'S, 4 hours; S'S, 24 hours. Profile of control incubation was identical with that of S1 over 24hour incubation period (CS). Additional 0.2-ml aliquots of enzyme were added after 1, 2, 3, and 4 hours; an additional 1.5 pmoles of NAD+ were added after 2 and 3 hours of incubation. Peaks a and c are substrate (see text for explanation of dual peak); Peak b is product. the rate of reaction (c) incubation of enzyme, substrate, and glutathione with or without NAD+ led to no formation of a hydroxy acid intermediate; (d) incubation of the steroid keto acid, enzyme, and NADH did not result in the formation of a steroid hydroxy acid; (e) neither llfl,2Ocu-nor llp,20&dihydroxy-3-keto-4-pregnen-21-oic acids were oxidized to the corresponding 20-keto acids in the presence of enzyme and NAD+; (f) with these steroid hydroxy acids, rabbit muscle and bovine heart lactate dehydrogenases were tiompletely inactive.
Reversibility-Enzyme was incubated with 1 l/%hydroxy-3,20diketo-4-pregnen-21.oic acid and NADH for periods up to 24 hours over a range of pH values and substrate concentration. No evidence was found for reversibility of the reaction.

Inhibition
Xtudies- Table  VI summarizes the effects of a number of reagents on enzyme activity.
Compounds that react with enzyme sulfhydryl groups were generally effective inhibitors, although iodoacetate was not. Sodium azide did not inhibit.
That inhibition by sodium bisulfite and flavin adenine dinucleotide was due to their combination with the substrate was supported by chromatographic evidence of altered mobility of the steroid in the presence of these compounds.
The effects of a number of metal ions is shown in Table VII. Inhibition by mercuric and silver ions is consistent with requirement of free sulfhydryl groups by the enzyme. Lead was also effective.
The inhibitory effects of arsenite on activity is shown in Table  VIII.
Arsenite is a reagent which inhibits enzymes whose activity depends on the presence of contiguous sulfhydryl groups (22). Its effects are demonstrated in the presence of exogenous mercaptans.
It is concluded, therefore, that the mechanism of  A number of compounds containing sulfhydryl groups including glutathione, P-mercaptoethanol, ol-lipoic acid, cysteine, coenzyme A, and dithiothreitol up to 1 mM had no effect on enzyme activity with 21-dehydrocorticosterone, methylglyoxal, or acetaldehyde as substrates.
Glutathione partially overcame the inhibition of the enzyme by p-hydroxymercuribenzoate.

DISCUSSION
The enzyme described in this paper was discovered as a result of our attempts $0 explain how corticosteroids may be oxidized to steroidal carboxylic acids by the adrenal gland (1,3,4,14,15). It was postulated that the tranformations require the oxidation of the ketol side chain through intermediate ketoaldehydes which are subsequently oxidized to acidic metabolites. The properties of the enzyme described here are consistent with such a scheme.
Unlike liver ketoaldehyde dehydrogenase (18), the adrenal enzyme has broad substrate specificity and utilizes both unsubstituted and a-substituted aldehydes. The enzyme was present as a single peak when examined by gel electrophoresis on polyacrylamide supports. The activity profile was the same for all of the substrates examined. Single peaks also emerged after gel filtration and ion exchange chromatography.
No evidence was obtained for any other enzymes acting on any substrate.
The ratios of activity for several substrates remained constant during purification, and when the enzyme was treated with inhibitors.
This broad range of substrate action is similar to that of other aldehyde dehydrogenases (23-26), although the specific substrates oxidized and their relative rates differ. It is not yet possible to draw quantitative conclusions about relative rates of substrate oxidation, since this is probably determined by the fraction of substrate present as the unhydrated free aldehyde. The percentage of hydration differs for each substrate and depends on the nature of the substituent next to the carbonyl (27).
The enzyme was not homogeneous with respect to protein. All attempts at further purification led to loss in activity and substantial decrease in stability.
The instability is similar to that of other purified mammalian aldehyde dehydrogenases. The enzyme was stabilized against inactivation by heat or high pH by the addition of NAD+ or aldehyde substrate.
Other aldehyde dehydrogenases are similarly protected (28). Protection of enzyme against sulfhydryl group inhibitors suggests that active site interaction involves thiol groups (29). The effects of arsenite on activity indicate that the activity or stability of the enzyme requires dithiol groups (22, 30). Glutathione and other compounds containing sulfhydryl groups did not affect enzyme activity, although other aldehyde dehydrogenases were affected (31, 32).
The stoichiometric relationships during the reaction are like that of other aldehyde dehydrogenases (33,34), as is the irreversibility of the oxidation (24, 35). The enzyme is found in the cytosol, as are other aldehyde dehydrogenases of the same category (28).
The molecular weight of approximately 165,000 is close to that which has been reported for horse liver dehydrogenase (36, 37).
Whether the ability of the dehydrogenase to utilize a steroid substrate is a unique one, or is a general property of aldehyde dehydrogenases, will require further investigation.
The ability of the enzyme to oxidize steroid aldehydes may be physiologically significant.
This reaction occurs in an organ whose unique ability is directed to the synthesis and metabolic alteration of corticosteroids.
The transformation described in this paper supports the pathway which has been proposed for the biosynthesis of known steroidal carboxylic acids.