Affinity Labeling of Steroid Binding Sites

To further study the steroid binding site of 20&hydroxysteroid dehydrogenase (EC 1.1.1.53) by affinity labeling, 21[2’JH]bromoacetylaminoprogesterone was synthesized. The steroid inactivates the enzyme in a time-dependent and irreversible manner which follows pseudo-first order kinetics. Incorporation of 2 mol of 21-[2’-3Hlbromoacetylaminoprogesterone accompanies inactivation of 1 mol of enzyme. Amino acid analysis of the 6 N HCl hydrolysate of the radiolabeled enzyme identified dicarboxymethylhistidine as the main alkylation product. The stability of the amide bond in 21-bromoacetylaminoprogesterone excludes the possibility that dialkylation of the active site histidine residue occurs by a transacylation mechanism. When 2Ophydroxysteroid dehydrogenase is inactivated with 16a-[2’3Hlbromoacetoxyprogesterone in the presence of excess [214Cliodoacetic acid, amino acid analysis of the acid hydrolysate of the radiolabeled enzyme shows that 3H and 14C are present in dicarboxymethylhistidine in a ratio of 1.03:1. A two-step mechanism involving a specific, site-directed, steroid-mediated alkylation followed by a rapid nonspecific alkylation step is proposed to account for dialkylation of an active site histidine residue resulting from affinity labeling of 20/Shydroxysteroid dehydrogenase by 16a-bromoacetoxyprogesterone and 21-bromoacetylaminoprogesterone. 21-Bromoacetylaminoprogesterone, 21-acetylaminoprogesterone, and 16a-carboxyamidoprogesterone are not reduced in the presence of NADH when tested as substrates for the enzyme. Kinetic studies show that all three progesterone amide analogs are competitive inhibitors of cortisone reduction. However, 21-azidoprogesterone which is electronically and sterically similar to 21-acetylaminoprogesterone is a substrate with a V,,, value of 10.5 nmol min-’ pg-’ and an apparent K, value of 7.1 X 10m5 M. 16aZyanoprogesterone, which is structurally analogous to 16a+arboxyamidoprogesterone, has a V,,, value of 9.25 nmol min-’ pg-’ and an apparent K, value of 6.7 x 10e4 M. A mechanism involving unfavorable hydrogen bonding interactions between the

To further study the steroid binding site of 20&hydroxysteroid dehydrogenase (EC 1.1.1.53) by affinity labeling, 21- [2'JH]bromoacetylaminoprogesterone was synthesized. The steroid inactivates the enzyme in a time-dependent and irreversible manner which follows pseudo-first order kinetics. Incorporation of 2 mol of 21-[2'-3Hlbromoacetylaminoprogesterone accompanies inactivation of 1 mol of enzyme. Amino acid analysis of the 6 N HCl hydrolysate of the radiolabeled enzyme identified dicarboxymethylhistidine as the main alkylation product. The stability of the amide bond in 21-bromoacetylaminoprogesterone excludes the possibility that dialkylation of the active site histidine residue occurs by a transacylation mechanism. When 2Ophydroxysteroid dehydrogenase is inactivated with 16a-  in the presence of excess [2-14Cliodoacetic acid, amino acid analysis of the acid hydrolysate of the radiolabeled enzyme shows that 3H and 14C are present in dicarboxymethylhistidine in a ratio of 1.03:1. A two-step mechanism involving a specific, site-directed, steroid-mediated alkylation followed by a rapid nonspecific alkylation step is proposed to account for dialkylation of an active site histidine residue resulting from affinity labeling of 20/Shydroxysteroid dehydrogenase by 16a-bromoacetoxyprogesterone and 21-bromoacetylaminoprogesterone. 21-Bromoacetylaminoprogesterone, 21-acetylaminoprogesterone, and 16a-carboxyamidoprogesterone are not reduced in the presence of NADH when tested as substrates for the enzyme. Kinetic studies show that all three progesterone amide analogs are competitive inhibitors of cortisone reduction. However, 21-azidoprogesterone which is electronically and sterically similar to 21-acetylaminoprogesterone is a substrate with a V,,, value of 10.5 nmol min-' pg-' and an apparent K, value of 7.1 X 10m5 M. 16aZyanoprogesterone, which is structurally analogous to 16a+arboxyamidoprogesterone, has a V,,, value of 9.25 nmol min-' pg-' and an apparent K, value of 6.7 x 10e4 M steroidal amides and amino acid residues at the enzyme active site is proposed to account for the inability of 2Ophydroxysteroid dehydrogenase to effect hydrogen transfer between NADH and these steroids.
Earlier, we proposed two mechanisms to explain the dialkylation of a single histidine residue. Transacylation to an active site neighboring group of the initially formed steroid-histidine conjugate would displace hydroxyprogesterone and allow a second steroid bromoacetate molecule to enter the active site to react with the remaining imidazole ring nitrogen atom. Alternatively, dialkylation could occur by 2 molecules of steroid bromoacetate reacting with the active site histidine to form N,N-bis(steroid-carboxymethyl)-histidine conjugate (2).
This report describes the synthesis of 21-bromoacetylaminoprogesterone, which does not undergo hydrolysis and therefore cannot undergo a transacylation reaction during affinity labeling of BOP-hydroxysteroid dehydrogenase. The progesterone analog was used to affinity label the enzyme in order to investigate the mechanism by which the active site histidine residue is dicarboxymethylated and also to gain further insight into the interaction between steroids and macromolecular steroid binding sites.

MATERIALS AND METHODS
Affinity labels were synthesized as described in the appendix.l  [2-3H]bromoacetic acid (3.75 x 10e3 M; containing 5 mCi of isotope) in 40 ml of 0.05 M phosphate buffer, pH 7.0, at 25", exhibited nearly identical inactivation kinetics as in the absence of bromoacetic acid (2). The respective molar ratios of bromoacetic acid to enzyme and steroid to enzyme were 15,OOO:l and 15O:l. The control incubation contained the same molar ratios of [2-3H]bromoacetic acid, enzyme, and 16a-acetoxyprogesterone as above. After the enzyme was 90% inactivated, 2-mercaptoethanol was added to the reaction and the control mixtures. The resulting mixtures were separately dialyzed, lyophilized, digested with 6 N HCl at llo", and prepared for amino acid analysis as described above. Authentic 1,3-dicarboxymethylhistidine was added to each sample prior to analysis. Tritium quantitation of the fractionated effluent from the analyzer column revealed that 40% of the total emergent radioactivity (4500 dpm) was coincident with the dicarboxymethylhistidine fraction (the elution profile was similar to that in Fig. 4). The remaining radioactivity, constituting less than 10% of the total, did not correspond to either l-or 3-carboxymethylhistidine or indeed to any known carboxymethyl amino acids and was unevenly distributed across the amino acid profile. By contrast, analysis of the control sample showed no significant peak of radioactivity associated with dicarboxymethylhistidine or any other known carboxymethyl amino acids. and 20jShydroxysteroid dehydrogenase were incubated in a solution containing the same molar ratios and under similar conditions to those described for the single isotope labeling experiment above. Following enzyme inactivation of 22%, 47%, 59%, 75%, and 81% of initial activity a 15 molar excess of 2-mercaptoethanol relative to the steroid was added to each solution and they were exhaustively dialyzed against water to remove radioactivity which was not covalently bound to the protein. The enzyme inactivation kinetic data from this experiment, which are plotted in Fig. 5 A doubly radiolabeled and 75% inactivated enzyme sample was prepared for amino acid analysis according to the procedure described above for the single isotope labeling experi-Affinity Labeling ment. The fractionated effluent from the amino acid analyzer was quantitated for 3H and 14C activity. The resulting elution profiles are shown in Fig. 7. The fractions which contained dicarboxymethylhistidine possessed a constant ratio of 3H to W. There was no correspondence between 3H and WY activity throughout the rest of the elution profile and the total radioactivity in all other fractions was less than 10% of that found in the dicarboxymethylhistidine fractions. Assays conducted under standard conditions (see "Materials and Methods") with varying amounts of C-21 or C-16 substituted progesterone derivatives, and at two different enzyme concentrations, produced the kinetic results summarized in Tables I and II. The presence of 5 x 1OP M acylaminosteroid in the assay mixture did not produce a measurable consumption of NADH, even after 1 h at 25". Kinetic studies of these steroids with cortisone as a substrate showed 21-acetylaminoprogesterone to be a competitive inhibitor as described above;
Each of these steroids produced incorporation of 2 mol of 3H-carboxymethyl groups per mol of inactivated enzyme, and an active site histidine residue was shown to be dicarboxymethylated. These results suggested that 2 reactive steroid molecules sequentially alkylate the histidine residue. Another equally plausible mechanism was that following steroid-directed reaction of the first molecule with the histidyl imidazole ring, a hydroxysteroid moiety is released by hydrolysis, and a second affinity labeling steroid molecule directs the alkylation of the remaining imidazole nitrogen (2). The second mechanism is based on the fact that base-catalyzed hydrolysis of bromoacetoxyprogesterone-amino acid conjugates readily occurs (8). Moreover, it was recently shown that following affinity labeling of 20P-hydroxysteroid dehydrogenase with Gp-bromoacetoxyprogesterone, reactivation of the enzyme can be accomplished by adjusting the incubation mixture to pH 9.0, conditions which hydrolyze the steroid ester (11). 21-Bromoacetylaminoprogesterone was synthesized to unambiguously establish the mechanism of dialkylation of the active site histidine residue. This steroid does not undergo hydrolysis under the incubation conditions, and alkylation of an active site histidine residue would exclude a mechanism based on steroid ester hydrolysis.
Synthesis of 21-bromoacetylaminoprogesterone and its tritium-labeled analog required six steps (Fig. 1). 21-Aminoprogesterone was a logical precursor for the title compound and its two-step synthesis has been reported (4). However, isolation of the steroid in pure form could not be effected due to the tendency of the compound to polymerize by reactions involving the 21-amino, C-3, and C-20 keto groups. Therefore, 21-amino-5-pregnene-3,20-dione diethylene ketal (IV, Fig. 1) was synthesized to serve as starting material for the preparation of 21-bromoacetylaminoprogesterone (VI) because the blocked C-3 and C-20 keto groups permit isolation of this intermediate.
The groundwork for establishing the structure of VI was recently reported (5), and the structure of V was confirmed on a similar basis. 21-Acetylaminoprogesterone and 21-bromoacetylaminoprogesterone are stable in strongly acidic solutions (the latter steroid is synthesized in methanolic hydrogen bromide). The C-21 amide linkage also resists basecatalyzed hydrolysis even under strongly alkaline conditions.
One of our earlier stated criteria for evaluating the capability of an affinity-labeling steroid to react with an amino acid residue at the enzyme active site is the demonstration that the steroids are substrates for the enzyme (1,3,8,9). Under optimally adjusted conditions of maximum steroid concentration, enzyme and cofactor concentrations, and prolonged periods of incubation, substrate activity of 21-bromoacetylaminoprogesterone could not be detected. Similarly, 21-acetylaminoprogesterone is not a substrate for 20/%hydroxysteroid dehydrogenase.
Therefore, both C-21 steroid amides were studied as inhibitors of cortisone reduction by the enzyme. Dixon plots of the kinetic results show that 21-bromoacetylaminoprogestercne and 21-acetylaminoprogesterone are competitive inhibitors with Kj values of 4.5 x 10m5 M and 7.7 x 10m4 M, respectively. Therefore, the C-21 acetylamino steroids are capable of binding at the enzyme active site.
Incubations of 21-bromoacetylaminoprogesterone with cysteine, methionine, and histidine, under conditions similar to those reported for bromoacetoxyprogesterone isomers (2, 7-91, showed that 21-bromoacetylaminoprogesterone alkylates nucleophilic amino acids much more slowly than do any of the steroid bromoacetates. This attenuation in alkylating activity may explain the longer period required for inactivation of BOP-hydroxysteroid dehydrogenase by 21-bromoacetylaminoprogesterone (Fig. 2) compared with that of the structurally analogous cortisone 21-iodoacetate or 16a-bromoacetoxyprogesterone. The t1,2 of enzyme inactivation with 21-bromoacetylaminoprogesterone is 48 h, while under similar experimental conditions inactivation with the structurally analogous cortisone 21-iodoacetate occurs with a t1,2 of 4 h (1).
Similar results obtained earlier with analogous haloacetoxy steroids were rationalized on the basis that the ester linkage of these affinity labeling steroids is susceptible to hydrolysis and, therefore, a transacylation reaction could possibly be driving dialkylation to rapid completion (2). Evidence that this type of transacylation occurs during affinity labeling of chymotrypsin with substrates that contain an ester bond has been reported by Lawson and Schramm (12,13). In the present case the amide linkage of 21-bromoacetylaminoprogesterone is not hydrolyzed under acidic or basic conditions. Moreover, 20P-hydroxysteroid dehydrogenase does not have esterase or amidase activity. The steroid-histidine conjugate (Fig. 6) resulting from alkylation by 21-bromoacetylaminoprogesterone is expected to be stable. Thus, our earlier proposed transacylation mechanism can be ruled out by analogy with the experimental results reported by Schramm and Lawson. They did not obtain the chymotrypsin-induced transacylation reaction with the N-substituted a-bromoacetamide substrates (14) as they had earlier obtained when the analogous esters were used for affinity labeling (12, 13).
The mechanism of dialkylation of a histidine residue at the active site of 20/S-hydroxysteroid dehydrogenase by 21-bromoacetylaminoprogesterone, 16a-bromoacetoxyprogesterone, or cortisone 21-iodoacetate occurs in a two-step sequence represented by the equations in Fig. 8. Incubation of the enzyme for 12 h at 25" with [2JHlbromoacetic acid alone does not produce loss in catalytic activity or formation of 1,3-[2'-3H]dicarboxymethylhistidine, but the presence of 16a-bromoacetoxyprogesterone promotes both of these reactions. When a mixture of 16cu-[2'-3H]bromoacetoxyprogesterone and [2-%]iodoacetic acid is incubated with 20p-hydroxysteroid dehydrogenase the kinetic data from enzyme inactivation (Fig.  6) are nearly identical with those reported earlier (2). Both 3H and 14C are incorporated in the enzyme protein. A 1:l ratio of 3H-carboxymethyl and 'Gcarboxymethyl groups is present in 1,3-dicarboxymethylhistidine derived from amino acid analysis of the doubly radiolabeled enzyme hydrolysate, and no monocarboxymethylhistidine can be detected. These results suggest that a slow steroid-directed alkylation step ( Kj, Fig.  8) is followed by a rapid and nonsteroid-directed alkylation reaction (k, or kB, Fig. 8).
The fact that the steroid amides are not substrates for 2Ophydroxysteroid dehydrogenase deserves further consideration. 21-Azidoprogesterone (II, Fig. 1) was used as an isosteric and isoelectronic model of 21-acetylaminoprogesterone to gain insight into the reason for the lack of substrate behavior of the C-21 acetylaminosteroids.
The similarity in chemical shifts of the C-21 proton magnetic resonance signal (7, 6.10) of II to the corresponding signal (T, 5.92) due to the C-21 protons of 21-acetylaminoprogesterone (4, 5) suggests that contributions by the azido and acetylamino groups to the electronic environment surrounding the C-20 carbonyl group are similar. Comparison of molecular models shows that the azido and acetylamino groups are spatially similar. 21-Azidoprogesterone was found to have a V,,, value of 10.5 nmol min-' pg-' and an apparent K, value of 7.1 x 10m5 M, which are characteristic of a good substrate for 20@hydroxysteroid dehydrogenase (see Table II).
A possible connection between the hydrogen-bonding capac-ity of C-21 amide substituents and the apparent inability of the enzyme to effect transfer of a hydrogen atom between NADH and the C-20 carbonyl group is suggested by the infrared spectra of the C-21 acylamino steroids, which exhibit absorption bands at 3340 cm-' characteristic of strong intermolecular hydrogen bonding. Therefore, 16a-carboxyamidoprogesterone (VIII, Fig. 1) was synthesized in order to further explore the effect that a hydrogen-bonding substituent has on a potential steroid substrate. Preparation of VIII from 16acyanoprogesterone (VII) was accomplished by hydrolysis of the cyano group with hydrochloric acid (Panel B, Fig. 1). Although VII exhibits good substrate characteristics with 20P-hydroxysteroid dehydrogenase (Table II) Mechanism of dialkylation of a histidine residue by 16a-bromoacetoxyprogesterone at the active site of 20@hydroxysteroid dehydrogenase. The first reaction (k,) involves a slow steroid-directed alkylation step. The second step (k, or k,) involves a rapid nondirected alkylation of the remaining imidazole ring nitrogen atom. The structures shown are arbitrarily drawn since the specific nitrogen atoms involved in each step are not known. inhibitor, similar to 21-acetylaminoprogesterone. Both 16cyand 21-haloacetoxy groups on the steroid molecule can alkylate the same histidine residue because the reaction groups have access to a common region of space (2). By analogy, the corresponding 21-acetylamino and 16or-carboxyamido groups can assume similar conformations which enable the amide groups to interact with the active site histidine residue (Fig.  9) and inhibit hydrogen transfer between NADH and the C-20 carbonyl function.