Enzymatic conversion of prostaglandin H2 to prostaglandin F2 alpha by aldehyde reductase from human liver: comparison to the prostaglandin F synthetase from bovine lung.

The primary structure of prostaglandin (PG) F synthetase from bovine lung shows 62% similarity with that of human liver aldehyde reductase (EC 1.1.1.2) (Watanabe, K., Fujii, Y., Nakayama, K., Ohkubo, H., Kuramitsu, S., Kagamiyama, H., Nakanishi, S., and Hayaishi, O. (1988) Proc. Natl. Acad. Sci. U.S.A. 85, 11-15). We therefore purified human liver aldehyde reductase to homogeneity and compared the immunological and catalytic properties of aldehyde reductase and PGF synthetase. Although both enzymes belong to a group of aldoketoreductases and their molecular weights are essentially identical, aldehyde reductase had no cross-reactivity to anti-PGF synthetase antiserum. Furthermore, there was a difference in the substrate specificity for reduction of PGs between the two enzymes. Aldehyde reductase catalyzed the reduction of PGJ2, delta 12-PGJ2, PGH2, or PGA2, but not that of PGB2, PGD2, or PGE2, whereas PGF synthetase reduced PGD2. The optimum pH, Km value for PGH2, and the turnover number were 6.5, 100 microM, and 3.1 min-1, respectively. The PGH2 9,11-endoperoxide reductase activity of aldehyde reductase was not affected in the presence of a substrate such as p-nitrobenzaldehyde, DL-glyceraldehyde, or 9,10-phenanthrenequinone, suggesting that PGH2 9,11-endoperoxide and other substrates are reduced at different active site(s). The reaction product formed from PGH2 by this enzyme was identified as PGF2 alpha by gas chromatography/mass spectrometry. These results suggest that aldehyde reductase is not exactly identical to PGF synthetase in terms of its immunological property and substrate specificity for PGs, but that this enzyme is also involved in the direct conversion of PGH2 to PGF2 alpha similar to PGF synthetase.

The primary structure of prostaglandin (PG) F synthetase from bovine lung shows 62% similarity with that of human liver aldehyde reductase (EC 1.1.1.2) (Watanabe, K., Fujii, Y., Nakayama, K., Ohkubo, H., Kuramitsu, S., Kagamiyama, H., Nakanishi, S., and Hayaishi, 0. (1988) Proc. Natl. Acad. Sei. U. S. A. 85, [11][12][13][14][15]. We therefore purified human liver aldehyde reductase to homogeneity and compared the immunological and catalytic properties of aldehyde reductase and PGF synthetase. Although both enzymes belong to a group of aldoketoreductases and their molecular weights are essentially identical, aldehyde reductase had no cross-reactivity to anti-PGF synthetase antiserum. Furthermore, there was a difference in the substrate specificity for reduction of PGs between the two enzymes. Aldehyde reductase catalyzed the reduction of PGJz, A"-PGJ2, PGHz, or PGAz, but not that of PGB2, PGDz, or PGE2, whereas PGF synthetase reduced PGD2. The optimum pH, K , value for PGHz, and the turnover number were 6.5, 100 MM, and 3.1 min", respectively. The PGHz 9,ll-endoperoxide reductase activity of aldehyde reductase was not affected in the presence of a substrate such as p-nitrobenzaldehyde, DL-glyceraldehyde, or 9,10-phenanthrenequinone, suggesting that PGHz 9,ll-endoperoxide and other substrates are reduced at different active site(s). The reaction product formed from PGHz by this enzyme was identified as PGFz, by gas chromatography/mass spectrometry. These results suggest that aldehyde reductase is not exactly identical to PGF synthetase in terms of its immunological property and substrate specificity for PGs, but that this enzyme is also involved in the direct conversion of PGHz to PGF2, similar to PGF synthetase.
Prostaglandin (PG)' F synthetase has been purified to homogeneity from bovine lung, and its properties have been studied in detail (1). The enzyme exhibits a broad substrate specificity and catalyzes the reduction of PGH, to PGF2, and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To whom correspondence should be addressed Hayaishi Bioinformation Transfer Project, Research Development Corporation of Japan, c/o Osaka Medical College, 2-7 Daigakumachi, Takatsuki 569, Japan.
The abbreviations used are: PG, prostaglandin; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; GC/MS, gas chromatography/mass spectrometry; TMS, trimethylsilyl; HPLC, high performance liquid chromatography. that of PGD, to 9a,11@-PGFz, which is a stereoisomer of PGF2, ( 2 ) , a t different active sites on the same molecule (1,2 ) . Recently, cDNA sequences specific for PGF synthetase have been isolated from a cDNA library of bovine lung mRNA sequences, and the primary structure of this enzyme has been determined (3). Comparison of the amino acid sequence of PGF synthetase revealed 62% similarity to that of human liver aldehyde reductase (4). However, little information is available about the catalytic properties of aldehyde reductase concerning the reduction of PGs such as PGH,, PGD,, and PGE, (5,6). To examine the role of aldehyde reductase in PG metabolism, we purified this enzyme from human liver and characterized its catalytic properties with respect to various PGs. In this paper, we report that aldehyde reductase as well as PGF synthetase catalyzes the reduction of PGH2 to PGF,,, although this enzyme is biochemically and immunologically different from PGF synthetase.
Enzyme Assay-The aldehyde reductase activity was determined according to the method of Wermuth et al. (5). The standard reaction mixture contained the following in a total volume of 1 ml: 0.1 M sodium phosphate buffer (pH 6.5), 0.08 mM NADPH, and enzyme. The reaction was started by the addition of substrate. For measurements during the earlier three steps of purification, the assay mixture contained 10 mM pyrazole to inhibit alcohol dehydrogenase activity. The enzyme activity was measured spectrophotometrically a t 37 "C by following the decrease in absorbance at 340 nm. One unit of enzyme was defined as the amount that caused the oxidation of 1 pmol of NADPH/min. The PGH, 9,11-endoperoxide reductase activity was determined as described previously (1). The standard reaction mixture contained 0.1 M sodium phosphate buffer (pH 6.5), 500 pM NADP, 5 mM glucose 6-phosphate, glucose-6-phosphate dehydrogenase (1 unit), 80 p~ [1-14C]PGHz (0.1 pCi), and enzyme in a total volume of 0.05 ml. Incubation was carried out at 37 "C for 2 min. The reaction was started by the addition of enzyme and terminated by 0.25 ml of diethyl ether, methanol, 0.2 M citric acid (30:4:1). The mixture of PGDZ, PGEZ, and PGFZ, (20 pg each) was added to the solution as authentic markers. The organic phase (0.1 ml) was subjected to thin layer chromatography (TLC) in a solvent system of diethyl ether/ methanol/acetic acid (902:O.l). The positions of the PGs on the silica gel plate were visualized with iodine vapor. Silica gel was scraped off in sections corresponding to PGFZ,, PGEZ, PGD2, PGH;?, and others, and the radioactivity of each section was measured in a Tritontoluene scintillator by a Beckman liquid scintillation spectrometer model LS 2800.
The PGD, 11-ketoreductase and PGEz 9-ketoreductase activities were assayed under the same conditions as those for the PGHz 9,llendoperoxide reductase activity except that 1.5 mM [3H]PGDz or [3H] PGEz (0.14 pCi each) was used as a substrate in place of 80 p~ [l-"C]PGHZ (0.1 pCi) and that the incubation time was 30 min.
Protein concentration was determined according to the method of Lowry et al. (8), using bovine serum albumin as a standard.
Purification of Aldehyde Reductase from Human Liver-Human livers that appeared normal were obtained from legal autopsies and stored at -80 "C. All procedures were carried out at 4 "C. Aldehyde reductase was purified from the liver by the method of Wartburg and Wermuth (6) with minor modifications. Approximately 50 g of human liver were homogenized in 3 volumes (w/v) of 50 mM Na2HPO4 with an Ultra-turrax homogenizer (Janke and Kunkel, West Germany) three times at top speed for 30 s each time. The homogenate was centrifuged a t 15,000 X g for 1 h a t 4 "C, and the supernatant was recovered by decantation. Crude extracts were subjected to ammonium sulfate fractionation between 35 and 70% saturation. We then performed chromatography with Red-Sepharose instead of the Blue-Sepharose used in the original procedure. About 1,400-fold purification of the aldehyde reductase was achieved from the human liver with a yield of 39%. The specific activity of the final preparation was 14.6 units/mg of protein.
Polyacrylamide Gel Electrophoresis-Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed by the method of Laemmli (9). Molecular weight of the enzyme was determined using a Pharmacia Low Molecular Weight Calibration Kit. Protein bands were stained with Coomassie Brilliant Blue R-250. For immunoblots, samples were run on 0.1% SDS, 12.5% PAGE (9), and immunostained as described (10). Preparation of antisera against PGF synthetase from bovine lung was described previously (1).
Identification of the Reaction Product by Gas Chromatography/ Mass Spectrometry (GCIMSI-The purified enzyme (10 pg) was incubated with [l-"C]PGHZ (80 p~, 0.1 pCi) in the presence of the NADPH-generating system a t 37 "C for 2 min, and the organic extract was subjected to TLC in a solvent system of diethyl ether/methanol/ acetic acid (90:Z:O.l). The major product with an RF value of 0.06 was extracted from a silica gel plate with ethyl acetate/acetic acid (99:l). The combined extracts of 10 samples (equivalent to about 3.5 pg of PGFz,) were evaporated under reduced pressure and the residue was dissolved in 20 ml of 15% (v/v) ethanol. The sample was adjusted to pH 3.0 with acetic acid and applied to a SEP-PAK CIS cartridge (11). The ethyl acetate eluate was pooled and evaporated in uacuo. For analysis by GC/MS, the product was converted to a methyl ester using diazomethane prepared from N-methyl-N-nitrosomethylurea.
The methyl ester was further converted to a trimethylsilyl (TMS) derivative and analyzed by a JEOL combined gas chromatography mass spectrometer model JMS DX-300.

RESULTS
Properties of Aldehyde Reductase-The molecular weight of aldehyde reductase calculated from the amino acid sequence (4) is 36,359, which is practically the same as that of PGF synthetase (Mr 36,517). Moreover, the primary structure of aldehyde reductase shows 62% similarity with that of PGF synthetase (3). T o compare the immunological properties of the two enzymes, we first examined the relative mobility on SDS-PAGE and cross-reactivity to rabbit anti-PGF synthetase antiserum. On the Coomassie Brilliant Blue-stained gel shown in Fig. L4, PGF synthetase and aldehyde reductase each showed a single band of protein with a molecular weight of approximately 35,000 and 36,500, respectively. Since PGF synthetase employed here had the N and C termini deduced from a specific cDNA sequence for the molecule (3), underestimation of the molecular weight on SDS-PAGE appeared to be due to its physicochemical properties. Furthermore, aldehyde reductase did not cross-react with anti-PGF synthetase antiserum at all (Fig. lB), indicating that the antigenicity of aldehyde reductase is different from that of PGF synthetase.
The pH dependence of the reaction is shown in Fig. 2 A . p-Nitrobenzaldehyde, DL-glyceraldehyde, and 9,lO-phenanthrenequinone are routinely used as the substrates for monitoring aldehyde, aldose, and carbonyl reductases, respectively. As shown in Fig. 2A, the pH optimum with all three substrates was 6.5. In addition, we found that aldehyde reductase had PGH2 9,ll-endoperoxide reductase activity with optimal pH also a t 6.5 ( Fig. 2 A , inset).
Substrate Specificity-The substrate specificity of the purified aldehyde reductase was examined spectrophotometrically a t 37 "C ( Table I). As demonstrated in a previous report (5,6), the enzyme had a broad substrate specificity for a number of aromatic aldehydes or aldoses in the presence of NADPH. Among them, p-nitrobenzaldehyde was the best substrate. Sodium glucuronate, phenylglyoxal, and methylglyoxsal were reduced, but the K,,, values were 3.7, 2.2, and 1.7 mM, respectively, or about 10-fold higher than that for pnitrobenzaldehyde. The reaction rates for p-carboxybenzaldehyde and succinic semialdehyde were 56 and 50% of that observed with p-nitrobenzaldehyde. Furthermore, aldoses such as D-erythrose and DL-glyceraldehyde were also reduced. In contrast to the PGF synthetase (l), carbonyl compounds (9,lO-phenanthrenequinone and menadione) were relatively poor substrates of this enzyme.

94K
.  Among PGs tested here, PGJ2, A"-PGJz, PGH2, and PGA, served as substrates; the reaction rates were 1.3-0.5% of that observed with p-nitrobenzaldehyde. However, neither PGB2, PGD2, nor PGE2 was acted upon by this enzyme under the present experimental conditions. Reduction of PGHz by the Purified Enzyme-As shown in Fig. 2B, the PGH, reduction was dependent on the enzyme amount. The purified enzyme in the present study was inactivated by boiling for 5 min and required NADPH as a coenzyme, indicating that the reaction is enzymic. The apparent K,,, value for PGH, was calculated to be about 100 p~ (Fig. 3). Under the standard assay conditions for PGHZ 9,11endoperoxide reductase except with 54 p~ [1-"C]PGH2 and 80 p~ NADPH, approximately 0.273 nmol of NADPH was consumed when 0.340 nmol of PGFz, was produced. The stoichiometry of consumed NADPH to the produced PGF,, was estimated to be 0.80. NADPH and NADH were examined for their capacity to serve as a cofactor under the standard assay conditions using various substrates (Table 11). In the case of ketoaldehyde and aldose substrates, the reaction proceeded at less than 3% of the rates observed with NADPH when the same concentration of NADH was used as coenzyme. The purified enzyme exhibited high specificity for NADPH, and the K,,, values for NADPH and NADH were 5 and 50 p~, respectively (Table I). However, there was a difference in cofactor requirement between the reduction of ketoaldehyde and aldose compounds and that of PGH2 9,llendoperoxide and carbonyl compounds. The reductase activities toward PGH2, 9,10-phenanthrenequinone, and menadione with NADH at 80 p~ were 30, 1160, and 393% of those with NADPH at the same concentration.
To obtain information on the active site(s) of this enzyme, we examined the inhibitory effect of three substrates such as p-nitrobenzaldehyde, DL-glyceraldehyde, and 9,lO-phenanthrenequinone on the PGHz 9,ll-endoperoxide reductase activity. These substrates had no effect on the reduction of PGH, (80 p~) up to 0.5 mM p-nitrobenzaldehyde @-fold the K,,, value for this substrate), 12.5 mM DL-glyceraldehyde (3fold the K,,, value), and 0.04 mM 9,lO-phenanthrenequinone (2-fold the K,,, value) concentration ranges. These results taken together suggest that PGH, and the three substrates Substrate specificity of aldehyde reductase The reaction mixture contained 0.1 M sodium phosphate buffer (pH 6.5), 80 p~ NADPH, substrate, and the purified enzyme in a total volume of 1.0 ml. The initial velocity of decrease in the absorbance at 340 nm of NADPH was followed at 37 "C. The p-nitrobenzaldehyde reductase activity (14.6 units/mg of protein) represents 100% activity.  For the determination of K,,, value for NADH, p-nitrobenzaldehyde was held constant at 0.5 mM.

TABLE I1
Cofactor requirement for various substrates of human liver aldehyde reductase The reaction mixture consisted of 0.1 M sodium phosphate buffer (pH 6.5), enzyme, 0.08 mM NADH or NADPH, and substrates at various concentrations as indicated in a total volume of 1.0 ml. The reaction was initiated by the addition of substrate, and the enzyme activity was measured spectrophotometrically at 37 "C as described under "Experimental Procedures." Identification by GCIMS of the Reaction Product Formed from PGH2-The enzymatic product formed from PGHz was initially converted to a methyl ester, sequentially treated with a silylating reagent, and then subjected to GC/MS. The derivatized compound appeared as a single peak on GC with a retention time of 3.5 min and gave a mass spectrum (Fig.  4B) essentially identical to that obtained with the corresponding derivative of authentic PGF,, (Fig. 44) On the basis of GC/MS analysis, the reaction product formed from PGH, was identified as PGF,,.
In 1981, Watanbe et al. (18) in our laboratory found PGD 11-ketoreductase, which converted PGD2 to PGF2, in the cytosol fraction of various rat tissues. The highest specific activity was observed in the lung. They further purified this enzyme from bovine lung to homogeneity and characterized it in detail (1,2). As a result this enzyme was found to catalyze the conversion of PGHz to PGF,, and that of PGD2 to 9a,11p-PGF,, at different active sites on the same molecule, and was later named PGF synthetase (1). The molecular weight and substrate specificity of PGF synthetase are reportedly similar to those of human brain carbonyl reductase (19). On the other hand, Wermuth (19) has reported that carbonyl reductase purified from human brain also reduced PGE2, but PGH, 9,11-endoperoxide and PGD, reductase activities of the carbonyl reductase have not yet been studied. In the present study, we demonstrated that human liver aldehyde reductase, which belongs to the aldoketoreductase family, catalyzed the conversion of PGH, to PGF2, with NADPH as a cofactor (Figs. 2-4) and that the reduction of PGH, 9,ll-endoperoxide proceeded at 30% of the rate observed with NADPH when the same concentration of NADH was used as coenzyme (Table 11). Furthermore, we showed that the active site of PGH, reduction was different from that used for the reduction of other aldoketo compounds. Under standard assay conditions containing 80 PM [1-14C]PGH2, the formation of [1-'4C] PGF,, was not affected by the addition of PGDz or PGE, up to 1.5 mM (data not shown). Therefore, the 9,ll-endoperoxide of PGH, appears to be reduced directly to PGF,,, but not via PGDz or PGE2. These findings are in agreement with the results on PGF synthetase (1).
Aldehyde reductase purified from human liver also reduced PGA,, PGJ2, and A1*-PGJ2, but the reduction of PGB,, PGD2, or PGEz was undetectable up to 1.0-1.5 mM concentration ranges (Table I). These results suggest that the substrate specificity of aldehyde reductase on PG reduction is similar to that of PGF synthetase except for PGD2. Although the K,,, value of aldehyde reductase for PGH, was greater than that of enzymes such as bovine lung PGF synthetase (l), rat brain PGD synthetase (20), bovine vesicular gland PGE synthetase (21), and rabbit aorta PGI synthetase (22), its value was in the same order as that for PGs of human brain PGE synthetase (23), rat spleen PGD synthetase (24), PGD ll-ketoreductase (25), and PGE 9-ketoreductase (15). The specific activity of aldehyde reductase for PGHz (88 milliunits/mg of protein) was about 1.5-fold higher than that of previously purified PGF synthetase (1). As judged from both the specific activity and the K , value for PGH2, it is possible that aldehyde reductase also, at least in part, contributes to the formation of PGFZ, in human liver.
The enzymes responsible to the formation of PGFz, from PGHz have been demonstrated in various tissues (16,17). Several isozymes of glutathione S-transferase purified from rat liver catalyzes the direct reduction of PGH, to PGF,, with reduced glutathione as a specific electron donor (26)(27)(28). On the other hand, Wong (25) reported that PGD, ll-ketoreductase purified from rabbit liver catalyzed the conversion of PGD2 to PGF,,. Recently, PGD 11-ketoreductase activity has also been found in human liver, and the enzymatic product has been identified as 9a,llP-PGF2, which is a stereoisomer of PGF2, (29). Therefore, the formation of PGF2, and 9a,ll/3-PGFz in the liver is considered to be catalyzed by different enzymes, for example, aldehyde reductase, glutathione Stransferase, and PGD 11-ketoreductase.
Recently the primary structure of PGF synthetase has been established. Watanabe et al. (3) found 62% similarity in the amino acid sequences between PGF synthetase and human liver aldehyde reductase, both members of aldoketoreductase family. Moreover, Carper et al. (30) also discovered that rat lens aldose reductase has 50% identity with human liver aldehyde reductase. Thus, the sequence information gives a better understanding of structural relationship among the functionally similar members of the aldoketoreductase family.