THE JOURNAL OF BIOLOGICAL CHEMISTRY © 1998 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 273, No. 52, Issue of December 25, pp. 35381–35387, 1998 Printed in U.S.A.

Substrate Specificity and Kinetic Mechanism of the Insect Sulfotransferase, Retinol Dehydratase* (Received for publication, May 12, 1998, and in revised form, September 26, 1998)

Efsevia Vakiani, John Gately Luz‡, and Jochen Buck§ From the Department of Pharmacology, Joan and Sanford I. Weill Medical College of Cornell University, New York, New York 10021

Spodoptera frugiperda retinol dehydratase catalyzes the conversion of retinol to the retro-retinoid anhydroretinol. It shares sequence homology with the family of mammalian cytosolic sulfotransferases and provides the first link between sulfotransferases and retinol metabolism. In this study the enzymatic properties of retinol dehydratase were examined using bacterially expressed protein. We show that retinol dehydratase can catalyze the transfer of the sulfonate moiety to small phenolic compounds and exhibits many functional similarities to the mammalian cytosolic sulfotransferases. The bisubstrate reaction that it catalyzes between retinol and the universal sulfonate donor 3*-phosphoadenosine 5*-phosphosulfate seems to involve ternary complex formation and to proceed via a Random Bi Bi mechanism. In addition to the low nanomolar Km value for free retinol, retinol dehydratase is strongly inhibited by retinol metabolites, suggesting a preference for retinoids. Conversely, a number of tested mammalian cytosolic sulfotransferases do not utilize retinol, indicating that retinol is not a general substrate for sulfotransferases.

Vitamin A (retinol) regulates cellular function in multiple ways. Since it is not known to have biological activity itself, it is thought to serve as the parent compound for the biosynthesis of a number of retinoid metabolites. 11-cis-Retinaldehyde, for example, has been established as the essential chromophore for vision (1), and all-trans-retinoic acid (RA)1 has been widely studied for its roles in cellular differentiation (2) and morphogenesis (3). The two retro-retinoids, 14-hydroxy-4,14-retro-retinol (14-HRR) (4) and anhydroretinol (AR) (5), are among the most recently characterized bioactive retinoids. They are physiologically present in a number of insect and mammalian cell types, and evidence to date suggests they play a role in cell survival (reviewed in Ref. 6). 14-HRR appears to be essential in lymphocyte and fibroblast activation and can prevent cell death

* This work was supported in part by National Institutes of Health Grant DK48022. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ Supported by National Eye Institute Training Grant NRSA T32EY07138. § To whom correspondence should be addressed: Dept. of Pharmacology, Joan and Sanford I. Weill Medical College of Cornell University, 1300 York Ave., New York, NY 10021. Tel.: 212-746-6274; Fax: 212746-8835; E-mail: [email protected]. 1 The abbreviations used are: RA, retinoic acid; 14-HRR, 14-hydroxyretro-retinol; AR, anhydroretinol; PAPS, 39-phosphoadenosine 59-phosphosulfate; PAP, 39-phosphoadenosine 59-phosphate; BSA, bovine serum albumin; ST, sulfotransferase; DTT, dithiothreitol; Mops, 4-morpholinepropanesulfonic acid; Ches, 2-(cyclohexylamino)ethanesulfonic acid; Caps, 3-(cyclohexylamino)propanesulfonic acid; Mes, 4-morpholineethanesulfonic acid. This paper is available on line at http://www.jbc.org

in retinol-dependent cell lines grown in serum-free medium. AR can competitively inhibit the growth-supportive effects of retinol and 14-HRR. These retro-retinoids seem to define a novel class of messenger molecules. Progress in the characterization of retinol metabolizing enzymes and isolation of their genes has lagged behind the identification of the various bioactive metabolites. Although most studies have focused on alcohol and aldehyde dehydrogenases and their potential role in the oxidation of retinol and retinaldehyde to RA (reviewed in Ref. 7), other types of enzymes are required for the complete array of retinol metabolism. The cloning of retinol dehydratase (8) raises the possibility that sulfotransferases, a class of enzymes not previously linked to vitamin A metabolism, may play a role in the biosynthesis of certain retinol metabolites. Retinol dehydratase is the enzyme responsible for the conversion of retinol to the retro-retinoid AR (see Fig. 1) and was cloned from the insect cell line Spodoptera frugiperda (Sf)-21. It shares sequence homology to the family of mammalian (reviewed in Refs. 9 –11) and plant (reviewed in Ref. 12) cytosolic sulfotransferases, and it represents the first insect homologue of this growing family of enzymes. Like all sulfotransferases, retinol dehydratase uses 39-phosphoadenosine 59-phosphosulfate (PAPS) as co-substrate (Fig. 1). Cytosolic sulfotransferases catalyze bisubstrate reactions where the sulfonate moiety from the universal sulfonate donor PAPS is transferred to small acceptor molecules, such as steroids and catecholamines. Many of the putative substrates bind with high Km or Ki values, and only a few physiological substrates (i.e. endogenous substances that serve as substrates at their physiological concentrations, e.g. 17b-estradiol) have been described. A single enzymatic mechanism for the sulfonation of these compounds has not been established. A number of kinetic studies point toward a sequential mechanism (13– 19), whereas the first sulfotransferase crystal structure suggested that at least some sulfotransferase reactions may follow a ping-pong mechanism (20). S. frugiperda retinol dehydratase differs from the described cytosolic sulfotransferases in several important ways: its molecular mass (41 kDa) is higher than that of the known cytosolic sulfotransferases (30 –36 kDa), and its end product, AR, is not sulfonated. In addition, retinol dehydratase catalyzes the formation of a putative signaling molecule, and its low Km value for its apparent substrate, free retinol, indicates a specific interaction between enzyme and substrate. We were interested in examining whether retinol dehydratase functions as a sulfotransferase and in establishing the specificity of interaction between retinol dehydratase and retinoids. Our studies of the enzymatic properties of retinol dehydratase are also important in understanding the mechanism of AR formation and the role of sulfotransferases in retinol metabolism.

35381

35382

Retinol Dehydratase

FIG. 1. Retinol dehydratase-catalyzed reaction. * indicates the putative reaction intermediate, retinyl sulfate. EXPERIMENTAL PROCEDURES

Chemicals—[3H]Retinol and [35S]PAPS were purchased from NEN Life Science Products. 14-Hydroxy-4,14-retro-retinol, 13,14-dihydroxyretinol, anhydroretinol, and 4-oxoretinol were kindly provided by Dr. Fadila Derguini (Sloan-Kettering Cancer Center, New York). All other chemicals were purchased from Sigma. Expression and Purification of Retinol Dehydratase—The entire open reading frame of the cDNA coding for S. frugiperda retinol dehydratase was cloned in-frame into the expression vector pET19b (Novagen, Madison, WI), transformed into Escherichia coli strain BL21(DE3), and grown at 37 °C in Luria Broth containing ampicillin (100 mg/ml) until the A595 was 0.5. Isopropyl-1-thio-b-D-galactopyranoside was added (final concentration of 1 mM), and the culture was grown for an additional 3 h. Cells pelleted (4 °C, 10,000 3 g, 10 min) from 4 liters of culture were lysed by sonication in 50 ml of lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM dithiothreitol (DTT), 2 mM EDTA, 10 mg/ml aprotinin, 10 mg/ml leupeptin, and 100 mg/ml phenylmethylsulfonyl fluoride), and cellular debris was removed by centrifugation (4 °C, 10,000 3 g, 10 min). The supernatant was dialyzed at 4 °C against 20 mM Tris-HCl, pH 7.4, 1 mM DTT (2 3 4 liters) and applied to a DE52 column (Waters, Bedford, MA) (2.5 3 20 cm; flow rate, 1.5 ml/min 20 mM Tris-HCl, pH 7.4, 1 mM DTT; linear gradient, 0 – 0.5 M NaCl over 280 min; fraction volume, 9 ml). Fractions containing retinol dehydratase (eluting between 0.18 and 0.23 M NaCl) were pooled, concentrated to 5 ml by Centricon (Amicon, Inc., Beverly, MA) ultrafiltration, and applied to an Ultrogel AcA54 gel filtration column (Amersham Pharmacia Biotech) (2 3 100 cm; flow rate, 1 ml/min 20 mM Tris-HCl, pH 7.4, 1 mM DTT, 50 mM NaCl; fraction volume, 8 ml). Fractions containing retinol dehydratase were pooled and applied to a SourceQ column (Amersham Pharmacia Biotech) (10-ml bed volume; flow rate, 0.65 ml/min 20 mM Tris-HCl, pH 7.4, 1 mM DTT; linear gradient, 0 – 0.5 M NaCl over 160 min; fraction volume, 2.5 ml). Three fractions (eluting between 0.1 and 0.15 M NaCl) containing the majority of retinol dehydratase were pooled. The protein purity was determined by SDS-polyacrylamide gel electrophoresis followed by silver staining. The protein concentration was determined to be 6 mg/ml by Bradford assay referenced to bovine serum albumin (BSA) standard. Protein was stored in 200-ml aliquots at 270 °C and maintained stable enzymatic activity for at least 6 months. Retinol Dehydratase Assay—Recombinant retinol dehydratase (6 mg/ assay) was incubated with [3H]retinol (37.3 mCi/mmol) and PAPS in the presence of delipidated BSA (final BSA concentration was 30 mM unless otherwise indicated) at room temperature in 100 ml of buffer (20 mM Tris-HCl, pH 7.2, 1 mM DTT). The concentration of free retinol in each assay was calculated based on the published value for the dissociation constant of retinol for BSA (2 3 1026 M) (21). The reaction was allowed to proceed 5 min, and the reaction mixtures were delipidated according to the procedure of McClean et al. (22). The synthesis of [3H]AR was determined by on-line liquid scintillation counting after separation by

high pressure liquid chromatography gradient elution (water/methanol/ chloroform) from a C18 reverse-phase 201TP54 column (Vydac, Hesperia, CA) as described previously (5). To examine the pH dependence of retinol dehydratase-catalyzed AR formation different buffers (50 mM Mes at pH 5.6, 6, 6.4; 50 mM Mops at pH 6.8, 7.2, 7.4; 50 mM Hepes at pH 7.2, 7.6, 8; 50 mM bis-tricine propane at pH 8, 8.4; 50 mM Ches at pH 8.6, 8.8, 9.2, 9.6, 9.8; 50 mM Caps at pH 10, 10.4, 10.8, 11.1) were substituted in the reaction mixture. To examine the effects of different cations, different salts (NaCl, KCl at final concentration 100 mM; MgCl2, MnCl2, CaCl2, CoCl2, NiCl2, ZnCl2 at final concentration 10 mM) were included in the assay. Sulfonation Assay—The method of Foldes and Meek (23) was used. Briefly, 50 mg of retinol dehydratase was incubated with 1 ml of [35S]PAPS (1.6 mCi/mmol) and 1 ml of substrate in 1 ml of buffer (20 mM Tris-HCl, pH 7.2) at room temperature for 0.5 h. The reaction was stopped by addition of 200 ml of 0.1 M Ba(OAc)2. Free [35S]PAPS was precipitated twice by addition of 100 ml of 0.1 M ZnSO4 and 100 ml 0.1 M Ba(OH)2. The precipitate was removed by centrifugation (5,000 3 g, 5 min, room temperature). After the second precipitation 35S-sulfonated products in the supernatant were counted by liquid scintillation (Beckman Instruments, LS 6500). Assays were performed in triplicate. Kinetic Studies—Initial velocity measurements were made under conditions where conversion of [3H]retinol did not exceed 10%. Measurements of the kinetic constants for each substrate were performed by varying the concentration of one substrate, while keeping the other substrate at a fixed and near saturating concentration. To determine the effect of an inhibitor, these measurements were repeated at different fixed inhibitor concentrations. The apparent Km, Vmax, and Ki values were determined using two computer programs, k.cat (Biometallics, Inc., Princeton, NJ) and KinetAsyst II (IntelliKinetics, State College, PA). Both programs use nonlinear regression based on the algorithms by Cleland (24). Two-substrate kinetic measurements were performed by varying the concentration of retinol at several fixed concentrations of PAPS. The kinetic data were fit to models of bisubstrate reactions (ping-pong, sequential, and equilibrium ordered) using KinetAsyst II. For all-trans-RA, 17b-estradiol, PAP, and AR the mode of inhibition was determined by fitting the kinetic data to models of competitive, noncompetitive, and uncompetitive inhibition using KinetAsyst II. The choice of fit was based on a combination of visual examination and comparison of parameter values and residuals for all models tested. IC50 values were determined by measuring [3H]AR production at a fixed free retinol concentration (34 nM) in the presence of varying concentrations of inhibitor and plotting the log [inhibitor] versus vi/vo. The IC50 values calculated were based on duplicate measurements. In those cases, the Ki values were calculated from the IC50 values assuming competitive inhibition. The equation used to describe the relationship between Ki and IC50 for a competitive inhibitor was Ki 5 IC50/{11([S]/Km)}.

Retinol Dehydratase

35383

FIG. 2. Purification of recombinant retinol dehydratase and determination of enzyme activity. A, SDS-polyacrylamide gel electrophoresis analysis of retinol dehydratase. Proteins were visualized by silver staining. From left, lane A, protein size markers. Lane B, crude E. coli protein lysates after induction of retinol dehydratase expression. Lane C, purified retinol dehydratase. B, Michaelis-Menten plot and LineweaverBurk plot (inset) of [3H]AR formation as a function of free retinol concentration at a fixed PAPS concentration (20 mM). Km(app) 5 3.9 6 0.4 nM and Vmax(app) 5 494 6 20 pmol/(minzmg of retinol dehydratase). Assays were performed as described under “Experimental Procedures.” C, MichaelisMenten plot and Lineweaver-Burk plot (inset) of [3H]AR formation as a function of PAPS concentration at a fixed free retinol concentration (34 nM). Km(app) 5 0.9 6 0.1 mM and Vmax(app) 5 375 6 11 pmol/(minzmg of retinol dehydratase). Assays were performed as described under “Experimental Procedures.” Expression of Mammalian Sulfotransferases—Oligonucleotide primers were designed to specifically amplify the coding sequences of phenol/ aryl ST (25), hydroxysteroid ST-20 (26), estrogen ST (27), and dopa/ tyrosine ST (ST1B1) (28) and used for polymerase chain reaction from rat liver first-strand cDNA. The products were cloned in-frame into either the expression vector pET23a (estrogen ST and ST1B1) or the expression vector pET28a (aryl ST and hydroxysteroid ST) and expressed in E. coli strain BL21(DE3) similar to retinol dehydratase. Crude protein lysates were prepared as described above for retinol dehydratase. Crude protein lysates were also prepared from cells where sulfotransferase expression was not induced with isopropyl-1-thio-b-Dgalactopyranoside and from cells that were transformed with vector alone (either pET23a or pET28a) and were used as controls. Activity Assays for Rat Sulfotransferases—Sulfonation of p-nitrophenol (1 mM) and all-trans-retinol (52 mM) was determined using 100 ml of crude protein lysates as described above. To test for AR production, crude protein lysates were incubated with [3H]retinol, and reaction mixtures were analyzed as described above. The activity assays for mammalian sulfotransferases were performed at 37 °C. RESULTS

Expression of Recombinant Retinol Dehydratase—Recombinant S. frugiperda retinol dehydratase was overexpressed in E. coli and purified to homogeneity (Fig. 2A) by sequential column chromatography. Enzyme activity was measured by quantitating the reaction product, [3H]AR, after separation by C18 reverse-phase high pressure liquid chromatography. The apparent Km and Vmax values were 3.9 nM and 494 pmol AR/minzmg enzyme for free all-trans-retinol, and 0.95 mM and 375 pmol of AR/minzmg enzyme for PAPS (Fig. 2, B and C). These values are consistent with what was previously reported for the enzyme purified from Sf-21 cells (8) and indicated the recombi-

nant enzyme was representative of the native form. Effects of pH and Divalent Cations—Enzyme activity was assayed in different buffers covering pH values ranging from 5.6 to 11. Maximal activity was obtained at two distinct pH values, pH 7.4 and pH 9.2 (data not shown). Cations have been reported to differentially regulate the activity of mammalian cytosolic sulfotransferases and have been proposed as tools for distinguishing between various sulfotransferases (29). For this reason we tested the effects of different divalent cations on retinol dehydratase activity and found that whereas Mg21, Mn21, Ba21, and Ca21 had no effect, Zn21, Ni21, and Co21 had a significant inhibitory effect (40 – 60% inhibition at 10 mM concentration) (data not shown). In addition, the monovalent cations NaCl and KCl had no effect on enzyme activity. Retinol Dehydratase Is a Sulfotransferase—Retinol dehydratase was originally identified as a sulfotransferase based on sequence homology (8), but its function as a sulfotransferase had not been directly demonstrated. We tested its ability to sulfonate a variety of compounds that have been extensively studied as substrates for mammalian cytosolic sulfotransferases (e.g. p-nitrophenol). Under the conditions used, several compounds, including p-nitrophenol, vanillin, and serotonin, were sulfonated by retinol dehydratase (Table I). The highest level of activity was obtained with vanillin. Surprisingly, we observed ethanol sulfonation with as low as 0.1% ethanol concentration (corresponding to 17 mM). Substrate Specificity of Retinol Dehydratase—To study the substrate specificity of retinol dehydratase, structurally different compounds were tested for their ability to inhibit AR pro-

35384

Retinol Dehydratase

TABLE I Sulfonation activity of retinol dehydratase Activity was measured as described under “Experimental Procedures.” The concentration indicated in parentheses was the highest tested. 2 indicates no increase above background. Substrate

Specific sulfonation activity nmol/minzmg

Retinol (240 mM) 14-Hydroxy-retro-retinol (50 mM) 13,14-Dihydroxyretinol (50 mM) 17b-Estradiol (500 mM) Hydrocortisone (100 mM) Androsterone (100 mM) a-Ecdysone (210 mM) b-Ecdysone (210 mM) Catechin (1 mM) Minoxidil (500 mM) Dopamine (1 mM) p-Nitrophenol (1 mM) Hydroxybenzylhydrazine (1 mM) Serotonin (1 mM) Vanillin (1 mM) Ethanol (170 mM)

2 2 2 2 2 2 2 2 2 2 0.057 6 0.009 0.470 6 0.024 0.197 6 0.013 0.305 6 0.076 2.249 6 0.037 0.315 6 0.022

duction (Table II). The mode of inhibition versus retinol was established for two compounds, all-trans-RA and 17b-estradiol. For both these compounds the best data fit was to a model of competitive inhibition with Ki values in the range of 0.24 mM for all-trans-RA (Fig. 3A) and 17.6 mM for 17-b-estradiol (Fig. 3B). For the other compounds listed in Table II, including retinoids and steroids, IC50 values were determined at a fixed retinol concentration (34 nM), and Ki values were calculated from the measured IC50 assuming competitive inhibition. The retinoid compounds tested were strong inhibitors of retinol dehydratase with Ki values in the low micromolar range similar to all-transRA. Certain steroids were also found to be potent inhibitors of retinol dehydratase activity, in particular hydrocortisone and androsterone, even though these compounds were not sulfonated when presented as substrates (Table I). In contrast, the ecdysteroids tested, which play an important role in the postembryonic development and metamorphosis of insects (30), did not seem to have significant affinity for retinol dehydratase. Other compounds that serve as substrates for mammalian sulfotransferases, including p-nitrophenol, minoxidil, and several catecholamines, were weak inhibitors of retinol dehydratase activity at millimolar concentrations. As might be expected from the sulfonation assays, ethanol inhibited retinol dehydratase activity at ethanol concentrations above 0.1% (17 mM). Two-substrate Kinetics—The crystal structure of mouse estrogen sulfotransferase showed similarities between protein kinases and sulfotransferases and suggested that sulfonation may proceed via a ping-pong mechanism (20). In order to examine whether a ping-pong mechanism is operative in retinol dehydratase, [3H]AR formation was measured as a function of retinol at different fixed concentrations of PAPS. The kinetic data was fit to several models of bisubstrate reactions, and the best fit was obtained in the case of a sequential model. As shown in Fig. 4, the double-reciprocal plots of 1/v versus 1/[retinol] at different fixed PAPS concentrations resulted in an intercepting family of lines, which suggests the reaction proceeds via ternary complex formation. Product Inhibition—We tested the effects of PAP and AR on retinol dehydratase activity in order to see whether there is product inhibition and to get more information about the reaction mechanism. PAP was found to be a very potent inhibitor of retinol dehydratase similar to what has been reported for other mammalian (e.g. Refs. 14 –16) and plant (e.g. Ref. 17) sulfotransferases. It inhibited PAPS binding competitively (Fig. 5A) with a very low Ki value (0.38 mM 6 0.16), whereas it showed a

TABLE II Substrate specificity of retinol dehydratase IC50 values were determined as described under “Experimental Procedures.” Ki values for all compounds except all-trans-retinoic acid and 17b-estradiol were calculated from the IC50 values assuming a competitive mode of inhibition (Ki 5 IC50/{1 1 ([S]/Km)}). Free retinol concentration ([S]) was 34 nM. For weaker inhibitors, the % inhibition values at the highest inhibitor concentrations tested are reported. NI indicates no inhibition. IC50

Ki

mM

Retinoids 14-Hydroxy-retro-retinol 13,14-Dihydroxy-retinol All-trans-retinoic acid All-trans-retinaldehyde 9-cis-Retinaldehyde 4-oxoretinol Steroids 17b-Estradiol Hydrocortisone Androsterone a-Ecdysone (210 mM) b-Ecdysone (210 mM) Other p-Nitrophenol (1 mM) Dopamine (1 mM) Serotonin (1 mM) Hydroxybenzylhydrazine (1 mM) Vanillin (1 mM) Pyridoxal phosphate (1 mM) Catechin (1 mM) Minoxidil (0.5 mM) Ethanol (170 mM) ATP (1 mM) ADP (1 mM)

7 8.3

Inhibition %

5.1 13 14.2

0.72 0.85 0.24 0.52 1.34 1.46

3.6 1

17.4 0.37 0.11 3% 4% 17% NI NI 13% 17% 16% 29% NI 25% NI NI

noncompetitive mode of inhibition versus retinol (Fig. 5B) (Ki 5 9.9 mM 6 1.83). AR inhibited retinol binding competitively (Fig. 5C), whereas inhibition of PAPS binding appeared to be noncompetitive (Fig. 5D). These data taken together suggest that the retinol dehydratase-catalyzed reaction between all-transretinol and PAPS follows a Random Bi Bi mechanism. AR Is Not Produced by Known Mammalian Cytosolic Sulfotransferases—Despite the fact that the substrate specificities of mammalian cytosolic sulfotransferases have been extensively studied, there are no reports indicating utilization of retinol as substrate. To see whether mammalian sulfotransferases could utilize retinol, we expressed four rat sulfotransferases representing the known sulfotransferase subfamilies (shown in Fig. 6; expressed STs are highlighted) and tested them for their ability to produce AR and/or sulfonate retinol. The activity of the sulfotransferases was assayed in crude bacterial lysates, and in the case of phenol ST and dopa/tyrosine ST the results were confirmed with purified protein. All of the sulfotransferases were active in sulfonating p-nitrophenol, confirming we were expressing enzymatically active proteins (data not shown). When retinol was tested as a substrate (at total concentrations as high as 52 mM), none of the expressed mammalian sulfotransferases were able to produce AR or sulfonate retinol (data not shown), indicating that retinol is not a general substrate for cytosolic sulfotransferases. DISCUSSION

Our results show that retinol dehydratase catalyzes the transfer of the sulfonate moiety from PAPS to a variety of acceptor molecules and that it has many other functional similarities to known cytosolic sulfotransferases. Like them, it exhibits product inhibition, has a low micromolar Km value for PAPS, and low micromolar Ki value for PAP. Additionally, dual pH optima and inhibition by various metals have been reported for a number of sulfotransferases (e.g. Refs. 17 and 29).

Retinol Dehydratase

35385

FIG. 3. Competitive inhibition of retinol dehydratase by all-trans-retinoic acid and 17b-estradiol. 1/Velocity versus 1/[retinol] at different fixed concentrations of all-trans-retinoic acid (A) or 17b-estradiol (B). PAPS concentration was fixed at 20 mM. Assays were performed as described under “Experimental Procedures.” A, best data fit was obtained with a competitive inhibition model. Ki 5 0.24 6 0.018 mM. Filled circles, no alltrans-RA; open circles, 0.5 mM RA; filled squares, 1 mM RA. B, best data fit was obtained with a competitive inhibition model. Ki 5 17.6 6 4.3 mM. Filled circles, 2.5 mM estradiol; filled squares, 10 mM estradiol; open squares, 20 mM estradiol.

FIG. 4. Double-reciprocal plots for the retinol dehydratase-catalyzed conversion of retinol to anhydroretinol. 1/Velocity versus 1/[retinol] at different fixed PAPS concentrations. Best fit was to a sequential (ternary complex) mechanism. Filled circles, 80 mM PAPS; open circles, 5 mM; filled squares, 2 mM; open squares, 1 mM. Assays were performed as described under “Experimental Procedures.”

Mammalian cytosolic sulfotransferases are believed to have broad substrate specificities (9), whereas plant sulfotransferases exhibit strict substrate specificity (12). The low nanomolar Km value of retinol dehydratase for free retinol and the low micromolar Ki values for other retinoids suggest a specific interaction between the insect sulfotransferase and retinoids. Among the wide variety of other compounds tested, hydrocortisone and androsterone, and to a smaller extent 17b-estradiol, seemed to bind tightly to retinol dehydratase. However these compounds do not appear to be substrates. We could not detect sulfonated products, and based on their chemical structures, it is not likely that the sulfonated forms of these steroids are converted to more stable products lacking the sulfonate moiety.

It is worth noting the effect of ethanol on retinol dehydratase. As far as we know, sulfonation of ethanol has not been reported, even though many putative substrates for sulfotransferases are often dissolved in ethanol or other alcohols. Our results indicate that any analysis of the sulfonation of such compounds needs to be revisited to take into account the effect of the solvent. The structure of mouse estrogen sulfotransferase suggested that a sulfotransferase-catalyzed reaction may proceed via the formation of a sulfonated enzyme intermediate, and for this reason, we tested whether this is the case for retinol dehydratase. As shown in Fig. 4, the retinol dehydratase-catalyzed reaction appears to proceed via ternary complex formation,

35386

Retinol Dehydratase

FIG. 5. Inhibition of retinol dehydratase by PAP and AR. A, 1/velocity versus 1/[PAPS] at varying concentrations of PAP. Free retinol concentration was fixed at 34 nM. Best data fit was obtained with a competitive inhibition model. Ki 5 0.38 6 0.16 mM. Filled circles, no PAP; filled squares, 1 mM PAP; open squares, 2 mM PAP; open circles, 4 mM PAP. B, 1/velocity versus 1/[retinol] at varying concentrations of PAP. PAPS concentration was fixed at 20 mM. Best data fit was obtained with a noncompetitive inhibition model. Ki 5 14.2 6 2 mM. Filled circles, no PAP; filled squares, 10 mM PAP; open circles, 25 mM PAP. C, 1/velocity versus 1/[retinol] at varying concentrations of AR. PAPS concentration was fixed at 20 mM. Assays were performed as described under “Experimental Procedures” except the final BSA concentration was 5 mM. Best data fit was obtained with a competitive inhibition model. Ki 5 18.6 6 2.9 mM. Filled circles, no AR; open squares, 25 mM AR; filled squares, 50 mM AR; open circles, 100 mM AR. D, 1/velocity versus 1/[PAPS] at varying concentrations of AR. Free retinol concentration was fixed at 17 nM. Assays were performed as described under “Experimental Procedures.” Best data fit was obtained with a noncompetitive inhibition model. Ki 5 46 6 9 mM. Filled circles, no AR; open circles, 25 mM AR; open squares, 75 mM AR.

rather than a ping-pong mechanism, although it is not possible to exclude a mechanism with an unstable enzyme intermediate. Previous kinetic studies with mammalian and plant sulfotransferases (13–19) have provided evidence for several types of sequential mechanisms. For example, studies of human phenol sulfotransferases (15, 16) and flavonol (17) sulfotransferases suggested an ordered Bi Bi mechanism where PAPS binds first, whereas sulfonation of cortisol was reported to follow an ordered mechanism where the steroid binds first (18). In the case of retinol dehydratase, the competitive modes of inhibition of all-trans-RA (Fig. 3A) and AR (Fig. 5C) versus retinol and of PAP versus PAPS (Fig. 5A), as well as the noncompetitive mode of inhibition of PAP versus retinol (Fig. 5B) and AR versus PAPS (Fig. 5D) are consistent with a Random Bi Bi reaction mechanism. Based on the functional similarities between retinol dehydratase and mammalian cytosolic sulfotransferases, it is reasonable to believe that AR formation proceeds via the formation of retinyl sulfate (see Fig. 1). Our inability to detect retinyl sulfate in our sulfonation assay is presumably due to the fact that retinyl sulfate is an unstable and short-lived intermediate. This would also support the assumption inherent in our interpretation of the kinetic data that the formation of retinyl sul-

FIG. 6. Comparison of the insect sulfotransferase retinol dehydratase to mammalian sulfotransferases. Phylogenetic tree analysis depicts the degree of amino acid sequence identity among retinol dehydratase and rat cytosolic sulfotransferases. It was constructed using the program MegAlign (DNASTAR, Inc.). Proteins whose names are highlighted were expressed in E. coli and tested for AR production and/or retinol sulfonation as described under “Experimental Procedures.” HSST1, hydroxysteroid ST-20 (26); HSST2, hydroxysteroid ST-40 (31); HSST3, hydroxysteroid ST-60 (32); SMP2, senescence marker protein (33); ST1B1, dopa/tyrosine ST (28); ST1C1, N-hydroxyl2-acetylaminofluorene ST (34); PST, phenol ST (25); EST, estrogen ST (27).

Retinol Dehydratase fate is the rate-limiting step in AR production. A more rigorous and detailed determination of the mechanism of AR formation awaits the chemical synthesis of retinyl sulfate which has not been reported to date. Anhydroretinol is present in significant amounts in mammalian liver and in smaller amounts in mammalian lung (5). By analogy to the insect enzyme, we speculate that the mammalian enzyme responsible for the production of AR will also be a sulfotransferase. None of the rat sulfotransferases tested (Fig. 6) were able to produce AR, indicating that retinol is not a general substrate for sulfotransferases. Based on this finding we speculate that an as yet unidentified mammalian sulfotransferase will have retinol dehydratase activity. We are currently pursuing the cloning and characterization of additional mammalian sulfotransferases to identify the mammalian retinol dehydratase in the hope that it will provide important information about the metabolism of vitamin A and the synthesis of retro-retinoids. Acknowledgments—We thank Dr. Fadila Derguini (Sloan-Kettering Cancer, New York) for kindly providing us with 14-HRR, 13,14-dihydroxyretinol, AR, and 4-oxoretinol; Dr. Philip Cole (Rockefeller University, New York) for help in using the computer program KinetAsyst II and analyzing our data; Dr. Lonny Levin (Joan and Sanford I. Weill Medical College of Cornell University, New York) for critical reading of the manuscript; and an anonymous reviewer for their helpful insights. REFERENCES 1. Wald, G. (1968) Science 162, 230 –239 2. Love, J. M. & Gudas, L. G. (1994) Curr. Opin. Cell Biol. 6, 825– 831 3. Hoffman, C. & Eichelle, G. (1994) in The Retinoids: Biology, Chemistry, and Medicine (Sporn, M. B., Roberts, A. B. & Goodman, D. S., eds) 2nd Ed., pp. 387– 442, Raven Press, Ltd., New York 4. Buck, J., Derguini, F., Levi, E., Nakanishi, K. & Ha¨mmerling, U. (1991) Science 254, 1654 –1656 5. Buck, J., Gru¨n, F., Derguini, F., Chen, Y., Kimura, S., Noy, N. & Ha¨mmerling, U. (1993) J. Exp. Med. 178, 675– 680 6. Vakiani, E. & Buck J. (1998) in Handbook of Experimental Pharmacology: Retinoids (Nau, H. & Blaner, W. S., eds) Springer-Verlag Inc., New York, in press 7. Napoli, J. L. (1996) FASEB 10, 993–1001

35387

8. Gru¨n, F., Noy, N., Ha¨mmerling, U. & Buck, J. (1996) J. Biol. Chem. 271, 16135–16138 9. Falany, C. N. (1997) FASEB J. 11, 206 –216 10. Weinshilbourn, R. M., Otterness, D. M., Aksoy, I. A., Wood, T. C., Her, C. & Raftogiannis, R. B. (1997) FASEB J. 11, 3–14 11. Runge-Morris, M. A. (1997) FASEB J. 11, 109 –117 12. Varin, L., Marsolais, F., Richard, M. & Rouleau, M. (1997) FASEB J. 11, 517–525 13. Pennings E. J. M., Vrielink, R. & Van Kempen, G. M. J. (1978) Biochem. J. 173, 299 –307 14. Duffel, M. W. & Jakoby, W. B. (1981) J. Biol. Chem. 256, 11123–11127 15. Whittemore, R. M., Pearce, L. B. & Roth, J. A. (1985) Biochemistry 24, 2477–2481 16. Whittemore, R. M., Pearce, L. B. & Roth, J. A. (1986) Arch. Biochem. Biophys. 249, 464 – 471 17. Varin, L. & Ibrahim, R. K. (1992) J. Biol. Chem. 267, 1858 –1863 18. Singer, S. S. & Bruns, L. (1980) Can. J. Biochem. 58, 660 – 666 19. Zhang, H., Varmalova, O., Vargas, F. M., Falany, C. N. & Leyh, T. S. (1998) J. Biol. Chem. 273, 10888 –10892 20. Kakuta, Y., Pedersen, L. G., Carter, C. W., Negishi, M. & Pedersen, L. C. (1997) Nat. Struct. Biol. 4, 904 –908 21. Noy, N. & Xu, Z. J. (1990) Biochemistry 29, 3878 –3883 22. McClean, S. W., Rudel, M. E., Gross, E. G., DiGiovani, J. J. & Pede, G. L. (1982) Clin. Chem. 28, 693– 696 23. Foldes, A. & Meek, J. L. (1973) Biochim. Biophys. Acta 327, 365–374 24. Cleland, W. W. (1967) Adv. Enzymol. 29, 1–32 25. Ozawa, S., Nagata, K., Gong, D., Yamazoe, Y. & Kato, R. (1990) Nucleic Acids Res. 18, 4001– 4005 26. Ogura, K., Kajita, J., Narihata, H., Watabe, T., Ozawa, S., Nagata, K., Yamazoe, Y. & Kato, R. (1989) Biochem. Biophys. Res. Commun. 165, 168 –174 27. Falany, J. L., Krasnykh, V., Mikheeva, G. & Falany, C. N. (1995) J. Steroid Biochem. Mol. Biol. 52, 35– 44 28. Sakakibara, Y., Takami, Y., Zwieb, C., Nakayama, T., Suiko, M., Nakajima, H. & Liu, M. C. (1995) J. Biol. Chem. 270, 30470 –30478 29. Singer, S. S. (1985) in Biochemical Pharmacology and Toxicology (Zakim, D. & Vessey, D. A., eds) Vol. 1, pp. 95–159, John Wiley & Sons, Inc., New York 30. Gilbert, L. I., Bollenbacher, W. E. & Granger, N. A. (1980) Annu. Rev. Physiol. 42, 493–510 31. Ogura, K., Kajita, J., Narihata, H., Watabe, T., Ozawa, S., Nagata, K., Yamazoe, Y. & Kato, R. (1990) Biochem. Biophys. Res. Commun. 166, 1494 –1500 32. Watabe, T., Ogura, K., Satsukawa, M., Okuda, H. & Hiratsuka, A. (1994) Chem. Biol. Interact. 92, 87–105 33. Chatterjee, B., Majumdar, D., Ozbilen, O., Murty, C. V. R. & Roy, A. K. (1987) J. Biol. Chem. 262, 822– 825 34. Nagata, K., Ozawa, S., Miyata, M., Shimada, M., Gong, D. W., Yamazoe, Y. & Koto, R. (1993) J. Biol. Chem. 268, 24720 –24725