- Email: [email protected]
Critical residues for the coenzyme specificity of NAD+-dependent 15-hydroxyprostaglandin dehydrogenase
ABB Archives of Biochemistry and Biophysics 419 (2003) 139–146 www.elsevier.com/locate/yabbi
Critical residues for the coenzyme specificity of NADþ -dependent 15-hydroxyprostaglandin dehydrogenase Hoon Cho, Marcos A. Oliveira, and Hsin-Hsiung Tai* Division of Pharmaceutical Sciences, College of Pharmacy, University of Kentucky, Lexington, KY 40536-0082 USA Received 14 July 2003, and in revised form 4 September 2003
Abstract NADþ -dependent 15-hydroxyprostaglandin dehydrogenase (15-PGDH), a member of the short chain dehydrogenase/reductase (SDR) family, is responsible for the biological inactivation of prostaglandins. Sequence alignment within SDR coupled with molecular modeling analysis has suggested that Gln-15, Asp-36, and Trp-37 of 15-PGDH may determine the coenzyme specificity of this enzyme. Site-directed mutagenesis was used to examine the important roles of these residues. Several single mutants (Q15K, Q15R, W37K, and W37R), double mutants (Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R), and triple mutants (Q15K–D36A–W37R and Q15K–D36S–W37R) were prepared and expressed as glutathione S-transferase (GST) fusion proteins in Escherichia coli and purified by GSH–agarose affinity chromatography. Mutants Q15K, Q15R, W37K, W37R, Q15K–W37K, and Q15R–W37K were found to be inactive or almost inactive with NADPþ but still retained substantial activity with NADþ . Mutant Q15K–W37R and mutant Q15R–W37R showed comparable activity for NADþ and NADPþ with an increase in activity nearly 3fold over that of the wild type. However, approximately 30-fold higher in Km for NADPþ than that of the wild type enzyme for NADþ was found for mutants Q15K–W37R and Q15R–W37R. Similarly, the Km values for PGE2 of mutants were also shown to increase over that of the wild type. Further mutation of Asp-36 to either an alanine or a serine of the double mutant Q15K–W37R (i.e., triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R) rendered the mutants exhibiting exclusive activity with NADPþ but not with NADþ . The triple mutants showed a decrease in Km for NADPþ but an increase in Km for PGE2 . Further mutation at Ala-14 to a serine of a triple mutant (Q15K–D36S–W37R) decreased the Km values for both NADPþ and PGE2 to levels comparable to those of the wild type. These results indicate that the coenzyme specificity of 15-PGDH can be altered from NADþ to NADPþ by changing a few critical residues near the N-terminal end. Ó 2003 Elsevier Inc. All rights reserved. Keywords: Prostaglandins; Enzymes; Dehydrogenases; Coenzyme; Mutagenesis
Prostaglandins are a family of biologically potent fatty acids derived from arachidonic acid through the cyclooxygenase pathway. Prostaglandins are rapidly metabolized by initial oxidation of the 15(S)-hydroxyl catalyzed by 15-hydroxyprostaglandin dehydrogenase (15-PGDH)1 [1]. Two different types of 15-PGDH have been recognized. Type I is NADþ specific, while Type II is NADPþ preferred. Type I is more prostaglandin specific and exhibits low Km for prostaglandins, whereas *
Corresponding author. Fax: 1-859-257-7585. E-mail address: [email protected] (H.-H. Tai). 1 Abbreviations used: 15-PGDH, NADþ -dependent 15-hydroxyprostaglandin dehydrogenase; DTT, dithiothreitol; GSH, glutathione; PGE2 , prostaglandin E2 . 0003-9861/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.abb.2003.09.019
Type II has a much broader substrate specificity and shows high Km for prostaglandins [2]. In fact, Type II was later found to be identical with carbonyl reductase [3]. Therefore, Type I has been considered the key enzyme responsible for the biological inactivation of prostaglandins. Studies on prostaglandin catabolism have so far been focused on the Type I enzyme. Type I 15-PGDH (hereafter referred to as 15-PGDH) is active as a homodimer with a subunit M.W. of 29 kDa [4]. Sequence alignments indicate that it belongs to the short chain dehydrogenase/reductase (SDR) family. A consensus sequence of Tyr-X-X-X-Lys as well as a nearby generally conserved upstream serine residue in the middle part of the enzymes characterize this protein family [5]. Furthermore, a glycine rich motif
140
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
Gly-X-X-X-Gly-X-Gly typical of a Rossmann fold is found near the N-terminus and is thought to be part of the putative coenzyme binding site [6]. We have identified Tyr-151, Lys-155, and Ser-138 in 15-PGDH as being a catalytically important triad by site-directed mutagenesis [7,8]. A similar catalytic role of these three residues has also been suggested for several enzymes in the family [9,10]. We have also identified Thr-11, Cys-182, and Thr188 in 15-PGDH as being involved in interacting with NADþ by site-directed mutagenesis [11–13]. Enzyme in the SDR family is either NADþ or NADPþ specific or utilize both coenzymes. Either nucleotide stereospecifically transfers a hydride ion from or to the carbon atom of the substrates in position 4 of the nicotinamide ring. The structures of these two nucleotides are very similar. NADPþ differs from NADþ only by the presence of an additional phosphate group esterified to the 20 -hydroxyl group of the ribose at the adenine end. Nevertheless, their biochemistry is substantially different. Although NADþ is used exclusively in the oxidative degradations and therefore as an oxidant, NADPþ is confined, with few exceptions, to the reactions of reductive biosynthesis and thus acts as a reductant. Detailed analysis of the binding stereochemistry of each nucleotide to proteins has indicated that the NADþ molecules are discriminated from those of NADPþ primarily by hydrogen bonds formed between an aspartate side chain and the diol group of the ribose near the adenine moiety in the NADþ binding enzymes. In contrast, NADPþ complexes are generally stabilized by an arginine side chain that faces the adenine plane and is hydrogen bonded to the phosphomonoester. Fig. 1 shows a conserved aspartate and an arginine in
NADþ - and NADPþ -linked enzymes, respectively. Furthermore, a basic amino acid upstream near the Nterminal end in NADPþ -linked enzymes is also conserved. These two basic amino acids in NADPþ -linked enzymes are thought to be able to compensate for the two negative charges of the 20 -phosphate group of NADPþ [14]. On the other hand, the presence of an aspartate which is electrostatically unfavorably for NADPþ binding may be a key determining factor in NADþ preference [14]. 15-PGDH is known to be entirely NADþ specific, as expected from the presence of an aspartate, and the lack of a basic residue next to it near the Rossmann fold sequence [2]. Although the 3D structure of 15-PGDH has yet to be determined, crystal structures of the binary and tertiary complexes of other members of the SDR family, such as 7a-hydroxysteroid dehydrogenase from Escherichia coli [15] and tropinone reductases (TRs) [16,17] from Datura have been elucidated. These studies indicate that the e-amino group of lysine and the guanidino group of arginine most probably interact with the oxygen atom of the 20 -phosphate group of NADPþ . In an attempt to redesign coenzyme specificity, we have employed sequence alignments coupled with molecular modeling to identify critical residues in 15-PGDH that can be altered to accommodate NADPþ . In this report, we found that Glu-15, Asp-36, and Trp-37 were required to undergo mutation in order for the enzyme to be totally NADPþ specific. Both positions 15 and 37 require a basic residue and position 36 needs to be replaced by a neutral residue before NADPþ can effectively bind to the enzyme and be stabilized. Furthermore, the affinity of NADPþ for the enzyme appears to be significantly
Fig. 1. Comparison of sequences near the Rossmann fold of the SDR family. Following enzyme sequences are co-listed with that of human 15PGDH in the figure. Pseudomonas 3b-hydroxysteroid dehydrogenase (3b-HSD); E. coli 7a-hydroxysteroid dehydrogenase (7a-HSD); Drosophila alcohol dehydrogenase (ADH); rat dihydropyridine reductase (DHPR); Datura tropinone reductase-I (TR-I); Datura tropinone reductase-II (TRII); human carbonyl reductase (CR); pig 20b-hydroxysteroid dehydrogenase (20b-HSD); human 17b-hydroxysteroid dehydrogenase (17b-HSD); and rat retinol dehydrogenase (RDH).
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
increased by an alteration of position 14 to a serine residue.
Experimental procedures Materials NADþ , NADH, NADPþ , NADPH, GSH–agarose, sodium dodecyl sulfate (SDS), and dithiothreitol (DTT) were obtained from Sigma. PGE2 was from Cayman Chemicals. Cloned pfu DNA polymerase was from Stratagene. Dpn I endonuclease was from New England BioLabs. Protein A-HRP was from Transduction Laboratories. Polyvinyliedene fluoride (PVDF) membrane was obtained from the Millipore GST gene fusion pGEX2T expression vector was from Pharmacia. The QIAprep Spin Plasmid Miniprep Kit was from QIAGEN. ECLþ Plus Western blotting detection system RPN 2132 was from Amersham Life Science. Oligonucleotide primers were synthesized by the Macromolecular Structure and Analysis Facility of the University of Kentucky. Molecular modeling of the 15-PGDH A 3D structural working model for the human 15PGDH was constructed using the HOMOLOGY module of the program Insight II (Acelerys). The template structure used as a basis for modeling 15-PGDH was based on the reference TR-I crystallographic structure [16]. The sequence of human 15-PGDH was first aligned with the reference TR-I sequence. Alignment shows overall sequence identity of 25%. The insertions/deletions found in 15-PGDH map to loop regions of the template structure TR-I. The largest deletion of eight residues occurs in the loop connecting strand bD and helix aE of the nucleotide binding domain. This loop maps to the substrate binding site of TR-I where one would expect to find structural differences. The largest insertion of three residues is located in a loop connecting strand bC and helix aD of the nucleotide binding domain. The insertion occurs in the region of the adenine moiety of NADPþ in the structure of TR-I. None of the structural differences affect the phosphate binding site of NADPþ found in the structure of TR-I. The phosphate binding pocket of NADPþ is made up of two loops one (Loop 1) connecting strand bA and helix aB, containing the glycine rich motif G-X-X-X-G-X motif and a second (Loop 2) connecting strand bB and helix aC. Considering the conservation of the phosphate binding sites we have modeled this site. Briefly, all the structurally conserved regions (SCRs) were used as a basis for the coordinates of the model structure whose amino acids were then added. The structurally variable regions (SVRs) were built by identifying loop structures in the protein database, which are anchored by the ends of the SCRs.
141
Site-directed mutagenesis All site-directed mutagenesis of human placental 15PGDH cDNA was performed by QuikChange site-directed mutagenesis method [18]. Four pairs of PCR primers were used to perform PCR. The oligonucleotides used for mutagenesis had the following sequences: for mutagenesis of Gln-15 to Lys-15 (Q15K), 50 -ACC GGC GCG GCT AAA GGC ATA GGC AGA-30 and 50 -TCT GCC TAT GCC TTT AGC CGC GCC GGT30 ; Gln-15 to Arg-15 (Q15R), 50 -ACC GGC GCG GCT CGT GGC ATA GGC AGA-30 and 50 -TCT GCC TAT GCC ACG AGC CGC GCC GGT-30 ; Trp-37 to Lys-37 (W37K), 50 -GCG CTG GTG GAT AAA AAT CTT GAA GCA-30 and 50 -TGC TTC AAG ATT TTT ATC CAC CAG CGC-30 ; and Trp-37 to ARG-37 (W37R), 50 GCG CTG GTG GAT CGT AAT CTT GAA GCA-30 and 50 -TGC TTC AAG ATT ACG ATC CAC CAG CGC-30 , where the underlined bases indicate the bases that were changed. Double mutants Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R were prepared using template cDNAs of Q15K or Q15R mutant with above W37K or W37R primers. Triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R were prepared using template cDNAs of Q15K–W37R mutant with the following oligonucleotides: for mutagenesis of Asp-36 to Ala-36 (Q15K–D36A–W37R), 50 GTA GCG CTG GTG GCT TGG AAT CTT GAA-30 and 50 -TTC AAG ATT CCA AGC CAC CAG CGC TAC-30 ; Asp-36 to Ser-36 (Q15K–D36S–W37R), 50 GTA GCG CTG GTG TCC TGG AAT CTT GAA-30 and 50 -TTC AAG ATT CCA GGT CAC CAG CGC TAC-30 . Mutants A14S–Q15K–D36S–W37R were prepared using template cDNAs of Q15K–D36S–W37R mutant with the following oligonucleotides: for mutagenesis of Ala-14 to Ser-14 (A14S–Q15K–D36S– W37R), 50 -GTG ACC GGC GCG TCC CAG GGC ATA GGC-30 and 50 -GCC TAT GCC CTG GGA CGC GCC GGT CAC-30 . Mutants A13G–A14S–Q15K– D36S–W37R were prepared using template cDNAs of A14S–Q15K–D36S–W37R mutant with the following oligonucleotides: for mutagenesis of Ala-13 to Gly-13 (A13G–A14S–Q15K–D36S–W37R), 50 -CTG GTG ACC GGC GGT GCT CAG GGC ATA-30 and 50 -TAT GCC CTG AGC ACC GCC GGT CAC CAG-30 . Pfu DNA polymerase was used for PCR. The PCR product was treated with Dpn I endonuclease to digest the parental DNA template. The DNA sequences of the mutants were confirmed by DNA sequencing. Expression and purification of 15-PGDH The cDNA of 15-PGDH contains an EcoRI site near the C-terminus. This site was mutated from GAATTC to GAATCC allowing no change in amino acid residue (Ile-251) while destroying EcoRI site. The altered cDNA
142
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
was then inserted between BamHI and EcoRI sites of the pGEX-2T expression vector. The recombinant plasmid was used to transform E. coli BL-21 LysS. Cells were grown in 500 ml 2X YT medium containing 50 lg/ml ampicillin at 37 °C with shaking (250 rpm) until OD600 reached 0.6. IPTG was added to a final concentration of 1 mM and cells were allowed to grow for additional 12 h at 25 °C. Cells were then harvested by centrifugation at 5000g for 15 min at 4 °C. The cell pellet was resuspended in 20 ml cold cell lysis buffer [1 PBS buffer (pH 7.4) containing 1 mM EDTA and 0.1 mM DTT]. The cells were broken by sonication. The cell lysate was cleared by centrifugation at 10,000g for 20 min. The extract was slowly loaded onto the GSH–agarose column, which was equilibrated at 4 °C with column buffer [1 PBS buffer (pH 7.4) containing 1 mM EDTA and 0.1 mM DTT]. After washing with column buffer until the OD280 reached less than 0.005, the 15-PGDH was eluted from the GSH–agarose column by incubation at room temperature for 5 min with the elution buffer [50 mM Tris– HCl (pH 8.0) containing 10 mM reduced glutathione, 1 mM EDTA, and 0.1 mM DTT]. The concentration of the purified enzyme was determined and the purity of the enzyme was assessed by SDS/PAGE. 15-PGDH assay
crystal structures of NADPþ -dependent TRs have been elucidated [16]. Furthermore, the crystal structures of the ternary complex (enzyme, NADPþ coenzyme, and product) of TR-I and TR-II have been characterized [17]. Lys-31 and Arg-53 of TR-I and Lys-19 and Arg-41 of TR-II of which correspond to Gln-15 and Trp-37 of 15-PGDH were proposed to make electrostatic and hydrogen-bonding interactions with the 20 -phosphate group of the adenine ribose of NADPþ (Fig. 1). Based on this information site-directed mutagenesis of Gln-15 and Trp-37 of 15-PGDH to lysine and arginine (Q15K, Q15R, W37K, W37R, Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R) were carried out. The activities of the mutants Q15R, W37K, and W37R for NADþ were similar to those of the wild type enzyme (Fig. 2). However, mutants Q15K, Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R had higher activity than that of the wild type for NADþ . Especially, the activities of the double mutants Q15K– W37K and Q15K–W37R for NADþ were increased nearly 3-fold (Fig. 2). Other mutants such as Q15K, Q15R–37K, and Q15R–W37R were also more active than the wild type enzyme. The activities of all these mutants for NADPþ were tested. The activities of the wild type and mutants Q15K, Q15R, W37K, W37R, Q15K–W37K, and Q15R–W37K were inactive or almost inactive for NADPþ (Fig. 3). However, mutants
The activity of 15-PGDH was assayed at 37 °C fluorometrically by measuring the formation of NADH or NADPH at 468 nm following excitation at 340 nm using a Shimadzu RF-5301PC Fluorescence Spectrophotometer as described previously [19]. The reaction was carried out in 2 ml of 50 mM Tris–HCl buffer, pH 7.5, containing 0.1 mM DTT and 10–20 lg purified 15PGDH or its mutants and various concentrations of PGE2 and NADþ or NADPþ as indicated in figures and tables. Each combination of PGE2 and coenzyme was assayed in triplicate.
Results We have employed a pGEX-2T expression vector to express 15-PGDH as a GST fusion protein. The recombinant enzyme was purified from the crude extract by a single step of GSH–agarose column chromatography to near homogeneity. This procedure facilitated greatly the isolation of purified enzyme for kinetic characterization of the wild type as well as the mutant enzymes. The recombinant enzyme exhibited an M.W. of approximately 55 kDa. In this study, the putative binding sites in 15-PGDH for the 20 -phosphate group of the adenine ribose of NADPþ were identified if 15-PGDH were to be able to utilize NADPþ as a coenzyme. Although the crystal structure of 15-PGDH is yet to be determined, the
Fig. 2. Comparison of the enzyme activities of wild type 15-PGDH, mutant 15-PGDH Q15K, Q15R, W37K, W37R, Q15K–W37K, Q15K–W37R, Q15R–W37K, Q15R–W37R, Q15K–D36A–W37R, and Q15K–D36S–W37R for NADþ coenzyme. The purified enzymes were assayed for 15-PGDH activity. 15-PGDH activity was determined fluorometrically by measuring the formation of NADH as described in ‘‘Experimental procedures.’’
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
Fig. 3. Comparison of the enzyme activities of wild type 15-PGDH, mutant 15-PGDH Q15K, Q15R, W37K, W37R, Q15K–W37K, Q15K–W37R, Q15R–W37K, Q15R–W37R, Q15K–D36A–W37R, and Q15K–D36S–W37R for NADPþ coenzyme. The purified enzymes were assayed for 15-PGDH activity. 15-PGDH activity was determined fluorometrically by measuring the formation of NADPH as described in ‘‘Experimental procedures.’’
Q15K–W37R and Q15R–W37R not only exhibited high activities for NADPþ but also had comparable activities for both NADþ and NADPþ (Figs. 2 and 3). These results suggest that Gln-15 and Trp-37 are putative sites for the interaction with the 20 -phosphate group of the adenine ribose of NADPþ in 15-PGDH. These results also indicate that mutants Q15R–W37R and Q15R– W37R have comparable activity for NADPþ , whereas mutant Q15K–W37R appears to have more activity for NADþ than does mutant Q15R–W37R. The apparent Km values for PGE2 and NADþ of these single and double mutants were comparable to those of the wild type. However, the apparent Km value for NADPþ for either mutant Q15K–W37R or Q15R–W37R was approximately 30-fold higher than that for NADþ of the wild type enzyme. Similarly, the apparent Km values for PGE2 of either double mutant were significantly increased as compared to that of the wild type enzyme. Although both double mutants Q15K–W37R and Q15R–W37R utilized NADPþ as a coenzyme, they were not entirely NADPþ specific. It was noted that NADPþ specific enzymes in the SDR family appear to have a neutral amino acid instead of an acidic amino acid immediately upstream of the arginine (Fig. 1). Therefore, site-directed mutagenesis of Asp-36 of the double mutant Q15K–W37R to an alanine and a serine (Q15K–
143
D36A–W37R and Q15K–D36S–W37R) was carried out and the activities of these mutants for both NADþ and NADPþ were tested (Fig. 1). Both triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R showed no activity for NADþ , but still retained the activity for NADPþ . The apparent Km values for NADPþ and for PGE2 were found to be decreased and increased, respectively, for these triple mutants as compared to their parent double mutant. The kcat =Km ratios of double mutants Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R based on NADþ as a variable coenzyme were increased as compared to the wild type, whereas the kcat =Km ratios of double mutants Q15K–W37K, Q15R–W37K, and Q15R–W37R based on PGE2 as a variable substrate were either increased or comparable to that of the wild type, but the ratio of double mutant Q15K–W37R was significantly increased as shown in Table 1. The kcat =Km ratios of double mutants Q15K–W37R and Q15R– W37R based on NADPþ as a variable coenzyme were decreased as compared to the wild type. The kcat =Km ratios of triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R were also decreased using PGE2 as a variable substrate or NADPþ as a variable coenzyme. For NADPþ -linked enzymes, the amino acid residue immediately upstream of the basic amino acid near the Rossmann fold is either a serine or an asparagine (see Fig. 1). Site-directed mutagenesis of Ala-14 of the triple mutant Q15K–D36S–W37R to a serine was carried out. Kinetic studies of the mutant A14S–Q15K–D36S– W37R indicated that the apparent Km values for PGE2 and NADPþ were significantly decreased to the levels comparable to those of the wild type enzyme. However, catalytic efficiency was increased only slightly.
Discussion Type I 15-PGDH is a member of the SDR family that contains two highly conserved regions. The first region is the consensus GLY-X-X-X-Gly-X-Gly motif in the fingerprint domain for dinucleotide binding. The second region contains the Tyr-X-X-X-Lys sequence and may constitute part of the catalytic center of the enzymes. Type I 15-PGDH is known to be an entirely NADþ specific dehydrogenase. Structural studies on several NADþ specific dehydrogenases indicated that the amino terminus of these enzymes consists of a bab-fold, a structure that comprises the binding domain for the AMP moiety of the coenzyme [20]. However, the orientation of the coenzyme may differ based on in their tertiary structures. One orientation is that a negatively charged residue at the C-terminus of the second b strand forms a hydrogen bond with the 20 -hydroxyl group of the adenine ribose of the NADþ , such as that found in rat dihydropteridine dehydrogenase and Drosophila
144
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
Table 1 Km and Kcat /Km values for 15-PGDH mutants using NADþ or NADPþ as a coenzyme NADþ
PGE2 Km (lM)
kcat =Km (ml min1 mg1 )
Km (lM)
kcat =Km (ml min1 mg1 )
534.2 236.9 652.8 409.4 437.3 852.4 1365.8 787.0 957.8 – – –
31.5 1.46 79.4 3.24 32.1 1.15 48.2 2.16 28.9 1.22 37.0 2.32 74.0 1.54 48.0 2.73 40.1 1.70 – – –
97.9 70.7 102.0 65.2 95.2 287.6 130.4 117.6 134.6 – – –
– – – – – –
– – – – – – 914.5 12.82 – 892.2 10.25 458.7 11.24 425.4 8.75 27.9 2.40
– – – – – –
þ
Wild type Q15K Q15R W37K W37R Q15K–W37K Q15K–W37R Q15R–W37K Q15R–W37R Q15K–D36A–W37R Q15K–D36S–W37R A14S–Q15K–D36S–W37R
NAD coenzyme 6.3 0.93 26.7 2.47 5.3 0.88 8.1 1.39 6.4 1.08 13.1 1.24 7.7 1.01 7.8 1.10 6.4 1.17 – – –
Wild type Q15K Q15R W37K W37R Q15K–W37K Q15K–W37R Q15R–W37K Q15R–W37R Q15K–D36A–W37R Q15K–D36S–W37R A14S–Q15K–D36S–W37R
NADPþ coenzyme – – – – – – 46.1 2.41 – 27.3 1.13 103.1 4.00 106.8 7.40 3.1 0.60
115.5 – 239.6 10.6 14.2 67.7
7.9 – 6.8 2.4 4.1 7.2
Values show means SD. The Km value of the purified enzymes were determined by Lineweaver–Burk plots. The Kcat =Km ratios were calculated after Vmax was obtained from the Lineweaver–Burk plots.
alcohol dehydrogenase co-crystallized with the coenzyme [21,22]. Our previous studies on the photoaffinity labeling of 15-PGDH with [a-32 P]2-azido-NADþ are consistent with this proposition, since a labeled oligopeptide containing Asp-36 in the second b strand was identified [23]. Further analysis of the structures of several NADþ specific dehydrogenases has also indicated that an acidic residue is frequently found in the second b strand (Fig. 1). Structural analysis of the NADPþ /NADPH-preferring class of enzymes has indicated that two conserved positively charged residues in the fingerprint region may determine the coenzyme specificity [24]. One is usually situated before the second glycine of the Gly-X-X-XGly-X-Gly motif, whereas the other is located at the end of the second b strand in the bab-folds of SDR family enzymes (Fig. 1). These two positively charged residues are thought to be able to compensate for the two negative charges of the 20 -phosphate group of NADPþ . However, SDR family enzymes utilizing both NADþ and NADPþ coenzymes do not appear to require both basic residues to be present for NADPþ utilization. For example, human estrogenic 17b-hydroxysteroid dehydrogenase type I employing both NAD(H) and NADP(H) as coenzymes has a single Arg-37 in the second b strand with serine substituting for the other
one at position 12 (see Fig. 1). Further mutation of Ser12 to a lysine residue increased the enzyme preference for NADP(H) more than 20-fold, indicating that further stabilization of NADP(H) by an upstream basic residue is needed for increased reactivity with this coenzyme [25]. In contrast, substitution of the negatively charged residue, aspartate, into Leu-36 position switched the enzymeÕs preference from NADP(H) to NAD(H) with a 220-fold change in the ratio of coenzyme specificity [25]. Our finding that a single change of an amino acid residue either at position 15 or 37 in 15-PGDH did not alter either its specificity for NADþ or catalytic efficiency is interesting. The mutant W37R in which Asp-36 precedes Arg-37 has a similar structural arrangement to that of the Asp-36 mutant of human 17b-hydroxysteroid dehydrogenase type I as just described above. The lack of NADPþ reactivity in both mutants is in contrast to that found in rat retinol dehydrogenase, in which the enzyme utilizes NADP(H) preferably and has a similar structural arrangement in the fingerprint region, i.e., an acidic Glu-63 preceding a basic Lys-64 (Fig. 1). One possible explanation is that the interaction between the negatively charged side chain of Glu-63 and c-hydroxyl apart may congroup of Thr-62 which are about 5 A tribute to the stabilizing effect of Lys-64 [26]. This orients the negatively charged side chain of Glu-63 away
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
from the positively charged side chain of Lys-64, minimizing any canceling of the coulombic attraction between lysine and NADPþ , as well as minimizing a destabilizing coulombic interaction between glutamic acid and the 20 -phosphate of NADPþ . The fact that neither W37K nor W37R mutant could utilize NADPþ could be due to the possibility that NADPþ may not fit into the binding site properly because Asp-36 is not oriented away from the positively charged Arg-37 or Lys-37 in 15-PGDH mutants, since there is no interaction between the negatively charged side chain of Asp-36 and hydrophobic Val-35 as opposed to the situation in rat retinol dehydrogenase. This is particularly true for the Lys-37 mutant, since further mutation at position 15 to either a lysine or an arginine did not exhibit reactivity with NADPþ , whereas Arg-37 mutant exhibits some reactivity with NADPþ after further mutation at position 15 (Table 1). However, there are significant increases in Km values for both NADPþ and PGE2 as compared to those of single mutants. Removal of the negatively charged Asp-36 by mutation to either an alanine or a serine abolished the reactivity with NADþ totally but lowered the Km for NADPþ with a decrease in catalytic efficiency. This suggests that Asp-36 is absolutely needed for reactivity with NADþ for the double mutants Q15K–W37R and Q15R–W37R although it does not block reactivity with NADPþ as expounded above. Although substitution of Asp-36 by either an alanine or a serine ensures reactivity with NADPþ , it decreased slightly the Km values for NADPþ but increased significantly the Km values for PGE2 as compared to the kinetic constants of the double mutants. Furthermore, it also decreased significantly the catalytic efficiency of the double mutants. The model of the phosphate binding site of 15-PGDH (Fig. 4) shows that the basic residues at positions 15 and 37 poise the basic side chains for direct interactions with the phosphate of NADPþ . A rather interesting result of our modeling relates to the importance of Arg at position 37. The guanidinium group may likely form a stacking interaction with the adenine moiety of NADPþ as has been observed in other proteins [27]. This potential stacking interaction may explain the preference for Arg at position 37 (Figs. 3 and 4). In addition, our model of the NADPþ phosphate binding site reveals that Asp-36 is located underneath the phosphate moiety and its mutation to a Ser favors the main chain hydrogen bonding interaction that could further stabilize the nucleotide binding site, suggesting an explanation for the importance of this site in conferring coenzyme selectivity (Fig. 3). Mutation of Asp-36 to a comparable size of amino acid such as an asparagine may improve catalytic efficiency, although most of the NADPþ preferring enzymes have an alanine or a serine preceding an arginine in the second b strand (Fig. 1). It is of interest to note that the Km values for either NADPþ or PGE2 were decreased
145
Fig. 4. Constructed 3D structural model of 15-PGDH residues 11-40. The models generated from molecular modeling based on the tropinone reductase-I crystallographic structure. (A) Mutant Q15K–D36S– W37R. (B) Wild type.
to the levels comparable to those found in the wild type enzyme when Ala-14 was mutated to a serine commonly found in NADPþ -linked enzymes. Unfortunately, the catalytic efficiency was only slightly increased after such a mutation. Significant increase in catalytic efficiency with NADPþ as an exclusive coenzyme may require further structural changes. In this regard, Chen et al. [28] reported a 60-fold increase in affinity for NADPþ and a 1.5-fold increase in kcat compared to the wild type when Asp-39 was replaced with an asparagine in Drosophila alcohol dehydrogenase. Furthermore, adding a positive charge by mutation of Ala-46 to an arginine lowered the Km for NADPþ about 2.5-fold and increased kcat about
146
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
3-fold with NADPþ as compared to the wild type alcohol dehydrogenase [29]. It will be interesting to see if these corresponding changes may bring about significant increase in affinity for NADPþ and catalytic efficiency for 15-PGDH. Acknowledgment This work was supported in part by a Grant from NIH (HL-46296). References [1] E. Anggard, B. Samuelsson, J. Biol. Chem. 239 (1964) 4097–4102. [2] C.M. Ensor, H.H. Tai, J. Lipid Mediat. Cell. Signal. 12 (1995) 313–319. [3] B. Wermuth, J. Biol. Chem. 256 (1981) 1206–1213. [4] C.M. Ensor, J.Y. Yang, R.T. Okita, H.H. Tai, J. Biol. Chem. 265 (1990) 14888–14891. [5] H. Jornvall, B. Persson, M. Krook, S. Atrian, R. GonzalezDuarte, J. Jeffery, D. Ghosh, Biochemistry 34 (1995) 6003–6013. [6] M.G. Rossmann, A. Liljas, C.I. Branden, L.J. Banaszak, in: P.D. Boyer (Ed.), The Enzymes, vol. 2, third ed, Academic Press, New York, 1975, pp. 61–102. [7] C.M. Ensor, H.H. Tai, Biochim. Biophys. Acta 1208 (1994) 151– 156. [8] C.M. Ensor, H.H. Tai, Biochem. Biophys. Res. Commun. 220 (1996) 330–333. [9] D. Ghosh, V.Z. Pletnev, D.W. Zhu, Z. Wawrzak, W.L. Duax, W. Pangborn, F. Labrie, S.X. Lin, Structure 3 (1995) 503–513. [10] M.W. Sawicki, M. Erman, T. Puranen, P. Vihko, D. Ghosh, Proc. Natl. Acad. Sci. USA 96 (1999) 840–845. [11] H. Cho, H.H. Tai, Prostaglandins Leukot. Essent. Fatty Acids 66 (2002) 505–509.
[12] C.M. Ensor, H.H. Tai, Arch. Biochem. Biophys. 333 (1996) 117– 120. [13] H.P. Zhou, H.H. Tai, Biochem. Biophys. Res. Commun. 257 (1999) 414–417. [14] N. Tanaka, T. Nonaka, M. Nakanishi, Y. Deyashiki, A. Hara, Y. Mitsui, Structure 4 (1996) 33–45. [15] N. Tanaka, T. Nonaka, T. Tanabe, T. Yoshimoto, D. Tsura, Y. Mitsui, Biochemistry 35 (1996) 7715–7730. [16] K. Nakajima, A. Yamashita, H. Akama, T. Nakatsu, H. Kato, T. Hashimoto, J. Oda, Y. Yamada, Biochemistry 95 (1998) 4876– 4881. [17] A. Yamashita, H. Kato, S. Wakatsuki, S. Tomizaki, T. Nakatsu, K. Nakajima, T. Hashimoto, Y. Yamada, J. Oda, J. Biochemistry 38 (1999) 7630–7637. [18] J. Braman, C. Papworth, A. Greener, Methods Mol. Biol. 57 (1996) 5731–5744. [19] D.T. Kung-Chao, H.H. Tai, Biochim. Biophys. Acta 614 (1980) 1–13. [20] R.K. Wierenga, M.C.H. De Mayer, W.G.J. Hol, Biochemistry 24 (1985) 1346–1357. [21] D. Ghosh, Z. Wawrzak, C.M. Weeks, W.L. Duax, M. Erman, Structure 2 (1994) 629–640. [22] K.I. Varughese, N.H. Xuong, P.M. Kiefer, D.A. Matthews, J.M. Whitely, Proc. Natl. Acad. Sci. USA 91 (1994) 5582–5586. [23] A.J. Chavan, C.M. Ensor, P. Wu, B.E. Haley, H.H. Tai, J. Biol. Chem. 268 (1993) 16442–17437. [24] N. Tanaka, T. Nonaka, M. Nakanishi, Y. Deyashiki, A. Hara, Y. Mitsui, Structure 4 (1996) 33–45. [25] Y.W. Huang, I. Pineau, H.J. Chang, A. Azzi, V. Bellemare, S. Laberge, S.X. Lin, Mol. Endocrinol. 15 (2001) 2010–2020. [26] I. Tsigelny, B.E. Baker, Biochem. Biophys. Res. Commun. 226 (1996) 118–127. [27] J.B. Mitchell, C.L. Nandi, I.K. McDonald, J.M. Thornton, S.L. Price, J. Mol. Biol. 239 (1994) 315–331. [28] Z. Chen, Z.G. Lin, W.R. Lee, S.H. Chang, Eur. J. Biochem. 202 (1991) 263–267. [29] Z. Chen, I. Tsigelny, W.R. Lee, M.E. Baker, S.H. Chang, FEBS Lett. 356 (1994) 81–85.
Critical residues for the coenzyme specificity of NADþ -dependent 15-hydroxyprostaglandin dehydrogenase Hoon Cho, Marcos A. Oliveira, and Hsin-Hsiung Tai* Division of Pharmaceutical Sciences, College of Pharmacy, University of Kentucky, Lexington, KY 40536-0082 USA Received 14 July 2003, and in revised form 4 September 2003
Abstract NADþ -dependent 15-hydroxyprostaglandin dehydrogenase (15-PGDH), a member of the short chain dehydrogenase/reductase (SDR) family, is responsible for the biological inactivation of prostaglandins. Sequence alignment within SDR coupled with molecular modeling analysis has suggested that Gln-15, Asp-36, and Trp-37 of 15-PGDH may determine the coenzyme specificity of this enzyme. Site-directed mutagenesis was used to examine the important roles of these residues. Several single mutants (Q15K, Q15R, W37K, and W37R), double mutants (Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R), and triple mutants (Q15K–D36A–W37R and Q15K–D36S–W37R) were prepared and expressed as glutathione S-transferase (GST) fusion proteins in Escherichia coli and purified by GSH–agarose affinity chromatography. Mutants Q15K, Q15R, W37K, W37R, Q15K–W37K, and Q15R–W37K were found to be inactive or almost inactive with NADPþ but still retained substantial activity with NADþ . Mutant Q15K–W37R and mutant Q15R–W37R showed comparable activity for NADþ and NADPþ with an increase in activity nearly 3fold over that of the wild type. However, approximately 30-fold higher in Km for NADPþ than that of the wild type enzyme for NADþ was found for mutants Q15K–W37R and Q15R–W37R. Similarly, the Km values for PGE2 of mutants were also shown to increase over that of the wild type. Further mutation of Asp-36 to either an alanine or a serine of the double mutant Q15K–W37R (i.e., triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R) rendered the mutants exhibiting exclusive activity with NADPþ but not with NADþ . The triple mutants showed a decrease in Km for NADPþ but an increase in Km for PGE2 . Further mutation at Ala-14 to a serine of a triple mutant (Q15K–D36S–W37R) decreased the Km values for both NADPþ and PGE2 to levels comparable to those of the wild type. These results indicate that the coenzyme specificity of 15-PGDH can be altered from NADþ to NADPþ by changing a few critical residues near the N-terminal end. Ó 2003 Elsevier Inc. All rights reserved. Keywords: Prostaglandins; Enzymes; Dehydrogenases; Coenzyme; Mutagenesis
Prostaglandins are a family of biologically potent fatty acids derived from arachidonic acid through the cyclooxygenase pathway. Prostaglandins are rapidly metabolized by initial oxidation of the 15(S)-hydroxyl catalyzed by 15-hydroxyprostaglandin dehydrogenase (15-PGDH)1 [1]. Two different types of 15-PGDH have been recognized. Type I is NADþ specific, while Type II is NADPþ preferred. Type I is more prostaglandin specific and exhibits low Km for prostaglandins, whereas *
Corresponding author. Fax: 1-859-257-7585. E-mail address: [email protected] (H.-H. Tai). 1 Abbreviations used: 15-PGDH, NADþ -dependent 15-hydroxyprostaglandin dehydrogenase; DTT, dithiothreitol; GSH, glutathione; PGE2 , prostaglandin E2 . 0003-9861/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.abb.2003.09.019
Type II has a much broader substrate specificity and shows high Km for prostaglandins [2]. In fact, Type II was later found to be identical with carbonyl reductase [3]. Therefore, Type I has been considered the key enzyme responsible for the biological inactivation of prostaglandins. Studies on prostaglandin catabolism have so far been focused on the Type I enzyme. Type I 15-PGDH (hereafter referred to as 15-PGDH) is active as a homodimer with a subunit M.W. of 29 kDa [4]. Sequence alignments indicate that it belongs to the short chain dehydrogenase/reductase (SDR) family. A consensus sequence of Tyr-X-X-X-Lys as well as a nearby generally conserved upstream serine residue in the middle part of the enzymes characterize this protein family [5]. Furthermore, a glycine rich motif
140
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
Gly-X-X-X-Gly-X-Gly typical of a Rossmann fold is found near the N-terminus and is thought to be part of the putative coenzyme binding site [6]. We have identified Tyr-151, Lys-155, and Ser-138 in 15-PGDH as being a catalytically important triad by site-directed mutagenesis [7,8]. A similar catalytic role of these three residues has also been suggested for several enzymes in the family [9,10]. We have also identified Thr-11, Cys-182, and Thr188 in 15-PGDH as being involved in interacting with NADþ by site-directed mutagenesis [11–13]. Enzyme in the SDR family is either NADþ or NADPþ specific or utilize both coenzymes. Either nucleotide stereospecifically transfers a hydride ion from or to the carbon atom of the substrates in position 4 of the nicotinamide ring. The structures of these two nucleotides are very similar. NADPþ differs from NADþ only by the presence of an additional phosphate group esterified to the 20 -hydroxyl group of the ribose at the adenine end. Nevertheless, their biochemistry is substantially different. Although NADþ is used exclusively in the oxidative degradations and therefore as an oxidant, NADPþ is confined, with few exceptions, to the reactions of reductive biosynthesis and thus acts as a reductant. Detailed analysis of the binding stereochemistry of each nucleotide to proteins has indicated that the NADþ molecules are discriminated from those of NADPþ primarily by hydrogen bonds formed between an aspartate side chain and the diol group of the ribose near the adenine moiety in the NADþ binding enzymes. In contrast, NADPþ complexes are generally stabilized by an arginine side chain that faces the adenine plane and is hydrogen bonded to the phosphomonoester. Fig. 1 shows a conserved aspartate and an arginine in
NADþ - and NADPþ -linked enzymes, respectively. Furthermore, a basic amino acid upstream near the Nterminal end in NADPþ -linked enzymes is also conserved. These two basic amino acids in NADPþ -linked enzymes are thought to be able to compensate for the two negative charges of the 20 -phosphate group of NADPþ [14]. On the other hand, the presence of an aspartate which is electrostatically unfavorably for NADPþ binding may be a key determining factor in NADþ preference [14]. 15-PGDH is known to be entirely NADþ specific, as expected from the presence of an aspartate, and the lack of a basic residue next to it near the Rossmann fold sequence [2]. Although the 3D structure of 15-PGDH has yet to be determined, crystal structures of the binary and tertiary complexes of other members of the SDR family, such as 7a-hydroxysteroid dehydrogenase from Escherichia coli [15] and tropinone reductases (TRs) [16,17] from Datura have been elucidated. These studies indicate that the e-amino group of lysine and the guanidino group of arginine most probably interact with the oxygen atom of the 20 -phosphate group of NADPþ . In an attempt to redesign coenzyme specificity, we have employed sequence alignments coupled with molecular modeling to identify critical residues in 15-PGDH that can be altered to accommodate NADPþ . In this report, we found that Glu-15, Asp-36, and Trp-37 were required to undergo mutation in order for the enzyme to be totally NADPþ specific. Both positions 15 and 37 require a basic residue and position 36 needs to be replaced by a neutral residue before NADPþ can effectively bind to the enzyme and be stabilized. Furthermore, the affinity of NADPþ for the enzyme appears to be significantly
Fig. 1. Comparison of sequences near the Rossmann fold of the SDR family. Following enzyme sequences are co-listed with that of human 15PGDH in the figure. Pseudomonas 3b-hydroxysteroid dehydrogenase (3b-HSD); E. coli 7a-hydroxysteroid dehydrogenase (7a-HSD); Drosophila alcohol dehydrogenase (ADH); rat dihydropyridine reductase (DHPR); Datura tropinone reductase-I (TR-I); Datura tropinone reductase-II (TRII); human carbonyl reductase (CR); pig 20b-hydroxysteroid dehydrogenase (20b-HSD); human 17b-hydroxysteroid dehydrogenase (17b-HSD); and rat retinol dehydrogenase (RDH).
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
increased by an alteration of position 14 to a serine residue.
Experimental procedures Materials NADþ , NADH, NADPþ , NADPH, GSH–agarose, sodium dodecyl sulfate (SDS), and dithiothreitol (DTT) were obtained from Sigma. PGE2 was from Cayman Chemicals. Cloned pfu DNA polymerase was from Stratagene. Dpn I endonuclease was from New England BioLabs. Protein A-HRP was from Transduction Laboratories. Polyvinyliedene fluoride (PVDF) membrane was obtained from the Millipore GST gene fusion pGEX2T expression vector was from Pharmacia. The QIAprep Spin Plasmid Miniprep Kit was from QIAGEN. ECLþ Plus Western blotting detection system RPN 2132 was from Amersham Life Science. Oligonucleotide primers were synthesized by the Macromolecular Structure and Analysis Facility of the University of Kentucky. Molecular modeling of the 15-PGDH A 3D structural working model for the human 15PGDH was constructed using the HOMOLOGY module of the program Insight II (Acelerys). The template structure used as a basis for modeling 15-PGDH was based on the reference TR-I crystallographic structure [16]. The sequence of human 15-PGDH was first aligned with the reference TR-I sequence. Alignment shows overall sequence identity of 25%. The insertions/deletions found in 15-PGDH map to loop regions of the template structure TR-I. The largest deletion of eight residues occurs in the loop connecting strand bD and helix aE of the nucleotide binding domain. This loop maps to the substrate binding site of TR-I where one would expect to find structural differences. The largest insertion of three residues is located in a loop connecting strand bC and helix aD of the nucleotide binding domain. The insertion occurs in the region of the adenine moiety of NADPþ in the structure of TR-I. None of the structural differences affect the phosphate binding site of NADPþ found in the structure of TR-I. The phosphate binding pocket of NADPþ is made up of two loops one (Loop 1) connecting strand bA and helix aB, containing the glycine rich motif G-X-X-X-G-X motif and a second (Loop 2) connecting strand bB and helix aC. Considering the conservation of the phosphate binding sites we have modeled this site. Briefly, all the structurally conserved regions (SCRs) were used as a basis for the coordinates of the model structure whose amino acids were then added. The structurally variable regions (SVRs) were built by identifying loop structures in the protein database, which are anchored by the ends of the SCRs.
141
Site-directed mutagenesis All site-directed mutagenesis of human placental 15PGDH cDNA was performed by QuikChange site-directed mutagenesis method [18]. Four pairs of PCR primers were used to perform PCR. The oligonucleotides used for mutagenesis had the following sequences: for mutagenesis of Gln-15 to Lys-15 (Q15K), 50 -ACC GGC GCG GCT AAA GGC ATA GGC AGA-30 and 50 -TCT GCC TAT GCC TTT AGC CGC GCC GGT30 ; Gln-15 to Arg-15 (Q15R), 50 -ACC GGC GCG GCT CGT GGC ATA GGC AGA-30 and 50 -TCT GCC TAT GCC ACG AGC CGC GCC GGT-30 ; Trp-37 to Lys-37 (W37K), 50 -GCG CTG GTG GAT AAA AAT CTT GAA GCA-30 and 50 -TGC TTC AAG ATT TTT ATC CAC CAG CGC-30 ; and Trp-37 to ARG-37 (W37R), 50 GCG CTG GTG GAT CGT AAT CTT GAA GCA-30 and 50 -TGC TTC AAG ATT ACG ATC CAC CAG CGC-30 , where the underlined bases indicate the bases that were changed. Double mutants Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R were prepared using template cDNAs of Q15K or Q15R mutant with above W37K or W37R primers. Triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R were prepared using template cDNAs of Q15K–W37R mutant with the following oligonucleotides: for mutagenesis of Asp-36 to Ala-36 (Q15K–D36A–W37R), 50 GTA GCG CTG GTG GCT TGG AAT CTT GAA-30 and 50 -TTC AAG ATT CCA AGC CAC CAG CGC TAC-30 ; Asp-36 to Ser-36 (Q15K–D36S–W37R), 50 GTA GCG CTG GTG TCC TGG AAT CTT GAA-30 and 50 -TTC AAG ATT CCA GGT CAC CAG CGC TAC-30 . Mutants A14S–Q15K–D36S–W37R were prepared using template cDNAs of Q15K–D36S–W37R mutant with the following oligonucleotides: for mutagenesis of Ala-14 to Ser-14 (A14S–Q15K–D36S– W37R), 50 -GTG ACC GGC GCG TCC CAG GGC ATA GGC-30 and 50 -GCC TAT GCC CTG GGA CGC GCC GGT CAC-30 . Mutants A13G–A14S–Q15K– D36S–W37R were prepared using template cDNAs of A14S–Q15K–D36S–W37R mutant with the following oligonucleotides: for mutagenesis of Ala-13 to Gly-13 (A13G–A14S–Q15K–D36S–W37R), 50 -CTG GTG ACC GGC GGT GCT CAG GGC ATA-30 and 50 -TAT GCC CTG AGC ACC GCC GGT CAC CAG-30 . Pfu DNA polymerase was used for PCR. The PCR product was treated with Dpn I endonuclease to digest the parental DNA template. The DNA sequences of the mutants were confirmed by DNA sequencing. Expression and purification of 15-PGDH The cDNA of 15-PGDH contains an EcoRI site near the C-terminus. This site was mutated from GAATTC to GAATCC allowing no change in amino acid residue (Ile-251) while destroying EcoRI site. The altered cDNA
142
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
was then inserted between BamHI and EcoRI sites of the pGEX-2T expression vector. The recombinant plasmid was used to transform E. coli BL-21 LysS. Cells were grown in 500 ml 2X YT medium containing 50 lg/ml ampicillin at 37 °C with shaking (250 rpm) until OD600 reached 0.6. IPTG was added to a final concentration of 1 mM and cells were allowed to grow for additional 12 h at 25 °C. Cells were then harvested by centrifugation at 5000g for 15 min at 4 °C. The cell pellet was resuspended in 20 ml cold cell lysis buffer [1 PBS buffer (pH 7.4) containing 1 mM EDTA and 0.1 mM DTT]. The cells were broken by sonication. The cell lysate was cleared by centrifugation at 10,000g for 20 min. The extract was slowly loaded onto the GSH–agarose column, which was equilibrated at 4 °C with column buffer [1 PBS buffer (pH 7.4) containing 1 mM EDTA and 0.1 mM DTT]. After washing with column buffer until the OD280 reached less than 0.005, the 15-PGDH was eluted from the GSH–agarose column by incubation at room temperature for 5 min with the elution buffer [50 mM Tris– HCl (pH 8.0) containing 10 mM reduced glutathione, 1 mM EDTA, and 0.1 mM DTT]. The concentration of the purified enzyme was determined and the purity of the enzyme was assessed by SDS/PAGE. 15-PGDH assay
crystal structures of NADPþ -dependent TRs have been elucidated [16]. Furthermore, the crystal structures of the ternary complex (enzyme, NADPþ coenzyme, and product) of TR-I and TR-II have been characterized [17]. Lys-31 and Arg-53 of TR-I and Lys-19 and Arg-41 of TR-II of which correspond to Gln-15 and Trp-37 of 15-PGDH were proposed to make electrostatic and hydrogen-bonding interactions with the 20 -phosphate group of the adenine ribose of NADPþ (Fig. 1). Based on this information site-directed mutagenesis of Gln-15 and Trp-37 of 15-PGDH to lysine and arginine (Q15K, Q15R, W37K, W37R, Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R) were carried out. The activities of the mutants Q15R, W37K, and W37R for NADþ were similar to those of the wild type enzyme (Fig. 2). However, mutants Q15K, Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R had higher activity than that of the wild type for NADþ . Especially, the activities of the double mutants Q15K– W37K and Q15K–W37R for NADþ were increased nearly 3-fold (Fig. 2). Other mutants such as Q15K, Q15R–37K, and Q15R–W37R were also more active than the wild type enzyme. The activities of all these mutants for NADPþ were tested. The activities of the wild type and mutants Q15K, Q15R, W37K, W37R, Q15K–W37K, and Q15R–W37K were inactive or almost inactive for NADPþ (Fig. 3). However, mutants
The activity of 15-PGDH was assayed at 37 °C fluorometrically by measuring the formation of NADH or NADPH at 468 nm following excitation at 340 nm using a Shimadzu RF-5301PC Fluorescence Spectrophotometer as described previously [19]. The reaction was carried out in 2 ml of 50 mM Tris–HCl buffer, pH 7.5, containing 0.1 mM DTT and 10–20 lg purified 15PGDH or its mutants and various concentrations of PGE2 and NADþ or NADPþ as indicated in figures and tables. Each combination of PGE2 and coenzyme was assayed in triplicate.
Results We have employed a pGEX-2T expression vector to express 15-PGDH as a GST fusion protein. The recombinant enzyme was purified from the crude extract by a single step of GSH–agarose column chromatography to near homogeneity. This procedure facilitated greatly the isolation of purified enzyme for kinetic characterization of the wild type as well as the mutant enzymes. The recombinant enzyme exhibited an M.W. of approximately 55 kDa. In this study, the putative binding sites in 15-PGDH for the 20 -phosphate group of the adenine ribose of NADPþ were identified if 15-PGDH were to be able to utilize NADPþ as a coenzyme. Although the crystal structure of 15-PGDH is yet to be determined, the
Fig. 2. Comparison of the enzyme activities of wild type 15-PGDH, mutant 15-PGDH Q15K, Q15R, W37K, W37R, Q15K–W37K, Q15K–W37R, Q15R–W37K, Q15R–W37R, Q15K–D36A–W37R, and Q15K–D36S–W37R for NADþ coenzyme. The purified enzymes were assayed for 15-PGDH activity. 15-PGDH activity was determined fluorometrically by measuring the formation of NADH as described in ‘‘Experimental procedures.’’
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
Fig. 3. Comparison of the enzyme activities of wild type 15-PGDH, mutant 15-PGDH Q15K, Q15R, W37K, W37R, Q15K–W37K, Q15K–W37R, Q15R–W37K, Q15R–W37R, Q15K–D36A–W37R, and Q15K–D36S–W37R for NADPþ coenzyme. The purified enzymes were assayed for 15-PGDH activity. 15-PGDH activity was determined fluorometrically by measuring the formation of NADPH as described in ‘‘Experimental procedures.’’
Q15K–W37R and Q15R–W37R not only exhibited high activities for NADPþ but also had comparable activities for both NADþ and NADPþ (Figs. 2 and 3). These results suggest that Gln-15 and Trp-37 are putative sites for the interaction with the 20 -phosphate group of the adenine ribose of NADPþ in 15-PGDH. These results also indicate that mutants Q15R–W37R and Q15R– W37R have comparable activity for NADPþ , whereas mutant Q15K–W37R appears to have more activity for NADþ than does mutant Q15R–W37R. The apparent Km values for PGE2 and NADþ of these single and double mutants were comparable to those of the wild type. However, the apparent Km value for NADPþ for either mutant Q15K–W37R or Q15R–W37R was approximately 30-fold higher than that for NADþ of the wild type enzyme. Similarly, the apparent Km values for PGE2 of either double mutant were significantly increased as compared to that of the wild type enzyme. Although both double mutants Q15K–W37R and Q15R–W37R utilized NADPþ as a coenzyme, they were not entirely NADPþ specific. It was noted that NADPþ specific enzymes in the SDR family appear to have a neutral amino acid instead of an acidic amino acid immediately upstream of the arginine (Fig. 1). Therefore, site-directed mutagenesis of Asp-36 of the double mutant Q15K–W37R to an alanine and a serine (Q15K–
143
D36A–W37R and Q15K–D36S–W37R) was carried out and the activities of these mutants for both NADþ and NADPþ were tested (Fig. 1). Both triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R showed no activity for NADþ , but still retained the activity for NADPþ . The apparent Km values for NADPþ and for PGE2 were found to be decreased and increased, respectively, for these triple mutants as compared to their parent double mutant. The kcat =Km ratios of double mutants Q15K–W37K, Q15K–W37R, Q15R–W37K, and Q15R–W37R based on NADþ as a variable coenzyme were increased as compared to the wild type, whereas the kcat =Km ratios of double mutants Q15K–W37K, Q15R–W37K, and Q15R–W37R based on PGE2 as a variable substrate were either increased or comparable to that of the wild type, but the ratio of double mutant Q15K–W37R was significantly increased as shown in Table 1. The kcat =Km ratios of double mutants Q15K–W37R and Q15R– W37R based on NADPþ as a variable coenzyme were decreased as compared to the wild type. The kcat =Km ratios of triple mutants Q15K–D36A–W37R and Q15K–D36S–W37R were also decreased using PGE2 as a variable substrate or NADPþ as a variable coenzyme. For NADPþ -linked enzymes, the amino acid residue immediately upstream of the basic amino acid near the Rossmann fold is either a serine or an asparagine (see Fig. 1). Site-directed mutagenesis of Ala-14 of the triple mutant Q15K–D36S–W37R to a serine was carried out. Kinetic studies of the mutant A14S–Q15K–D36S– W37R indicated that the apparent Km values for PGE2 and NADPþ were significantly decreased to the levels comparable to those of the wild type enzyme. However, catalytic efficiency was increased only slightly.
Discussion Type I 15-PGDH is a member of the SDR family that contains two highly conserved regions. The first region is the consensus GLY-X-X-X-Gly-X-Gly motif in the fingerprint domain for dinucleotide binding. The second region contains the Tyr-X-X-X-Lys sequence and may constitute part of the catalytic center of the enzymes. Type I 15-PGDH is known to be an entirely NADþ specific dehydrogenase. Structural studies on several NADþ specific dehydrogenases indicated that the amino terminus of these enzymes consists of a bab-fold, a structure that comprises the binding domain for the AMP moiety of the coenzyme [20]. However, the orientation of the coenzyme may differ based on in their tertiary structures. One orientation is that a negatively charged residue at the C-terminus of the second b strand forms a hydrogen bond with the 20 -hydroxyl group of the adenine ribose of the NADþ , such as that found in rat dihydropteridine dehydrogenase and Drosophila
144
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
Table 1 Km and Kcat /Km values for 15-PGDH mutants using NADþ or NADPþ as a coenzyme NADþ
PGE2 Km (lM)
kcat =Km (ml min1 mg1 )
Km (lM)
kcat =Km (ml min1 mg1 )
534.2 236.9 652.8 409.4 437.3 852.4 1365.8 787.0 957.8 – – –
31.5 1.46 79.4 3.24 32.1 1.15 48.2 2.16 28.9 1.22 37.0 2.32 74.0 1.54 48.0 2.73 40.1 1.70 – – –
97.9 70.7 102.0 65.2 95.2 287.6 130.4 117.6 134.6 – – –
– – – – – –
– – – – – – 914.5 12.82 – 892.2 10.25 458.7 11.24 425.4 8.75 27.9 2.40
– – – – – –
þ
Wild type Q15K Q15R W37K W37R Q15K–W37K Q15K–W37R Q15R–W37K Q15R–W37R Q15K–D36A–W37R Q15K–D36S–W37R A14S–Q15K–D36S–W37R
NAD coenzyme 6.3 0.93 26.7 2.47 5.3 0.88 8.1 1.39 6.4 1.08 13.1 1.24 7.7 1.01 7.8 1.10 6.4 1.17 – – –
Wild type Q15K Q15R W37K W37R Q15K–W37K Q15K–W37R Q15R–W37K Q15R–W37R Q15K–D36A–W37R Q15K–D36S–W37R A14S–Q15K–D36S–W37R
NADPþ coenzyme – – – – – – 46.1 2.41 – 27.3 1.13 103.1 4.00 106.8 7.40 3.1 0.60
115.5 – 239.6 10.6 14.2 67.7
7.9 – 6.8 2.4 4.1 7.2
Values show means SD. The Km value of the purified enzymes were determined by Lineweaver–Burk plots. The Kcat =Km ratios were calculated after Vmax was obtained from the Lineweaver–Burk plots.
alcohol dehydrogenase co-crystallized with the coenzyme [21,22]. Our previous studies on the photoaffinity labeling of 15-PGDH with [a-32 P]2-azido-NADþ are consistent with this proposition, since a labeled oligopeptide containing Asp-36 in the second b strand was identified [23]. Further analysis of the structures of several NADþ specific dehydrogenases has also indicated that an acidic residue is frequently found in the second b strand (Fig. 1). Structural analysis of the NADPþ /NADPH-preferring class of enzymes has indicated that two conserved positively charged residues in the fingerprint region may determine the coenzyme specificity [24]. One is usually situated before the second glycine of the Gly-X-X-XGly-X-Gly motif, whereas the other is located at the end of the second b strand in the bab-folds of SDR family enzymes (Fig. 1). These two positively charged residues are thought to be able to compensate for the two negative charges of the 20 -phosphate group of NADPþ . However, SDR family enzymes utilizing both NADþ and NADPþ coenzymes do not appear to require both basic residues to be present for NADPþ utilization. For example, human estrogenic 17b-hydroxysteroid dehydrogenase type I employing both NAD(H) and NADP(H) as coenzymes has a single Arg-37 in the second b strand with serine substituting for the other
one at position 12 (see Fig. 1). Further mutation of Ser12 to a lysine residue increased the enzyme preference for NADP(H) more than 20-fold, indicating that further stabilization of NADP(H) by an upstream basic residue is needed for increased reactivity with this coenzyme [25]. In contrast, substitution of the negatively charged residue, aspartate, into Leu-36 position switched the enzymeÕs preference from NADP(H) to NAD(H) with a 220-fold change in the ratio of coenzyme specificity [25]. Our finding that a single change of an amino acid residue either at position 15 or 37 in 15-PGDH did not alter either its specificity for NADþ or catalytic efficiency is interesting. The mutant W37R in which Asp-36 precedes Arg-37 has a similar structural arrangement to that of the Asp-36 mutant of human 17b-hydroxysteroid dehydrogenase type I as just described above. The lack of NADPþ reactivity in both mutants is in contrast to that found in rat retinol dehydrogenase, in which the enzyme utilizes NADP(H) preferably and has a similar structural arrangement in the fingerprint region, i.e., an acidic Glu-63 preceding a basic Lys-64 (Fig. 1). One possible explanation is that the interaction between the negatively charged side chain of Glu-63 and c-hydroxyl apart may congroup of Thr-62 which are about 5 A tribute to the stabilizing effect of Lys-64 [26]. This orients the negatively charged side chain of Glu-63 away
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
from the positively charged side chain of Lys-64, minimizing any canceling of the coulombic attraction between lysine and NADPþ , as well as minimizing a destabilizing coulombic interaction between glutamic acid and the 20 -phosphate of NADPþ . The fact that neither W37K nor W37R mutant could utilize NADPþ could be due to the possibility that NADPþ may not fit into the binding site properly because Asp-36 is not oriented away from the positively charged Arg-37 or Lys-37 in 15-PGDH mutants, since there is no interaction between the negatively charged side chain of Asp-36 and hydrophobic Val-35 as opposed to the situation in rat retinol dehydrogenase. This is particularly true for the Lys-37 mutant, since further mutation at position 15 to either a lysine or an arginine did not exhibit reactivity with NADPþ , whereas Arg-37 mutant exhibits some reactivity with NADPþ after further mutation at position 15 (Table 1). However, there are significant increases in Km values for both NADPþ and PGE2 as compared to those of single mutants. Removal of the negatively charged Asp-36 by mutation to either an alanine or a serine abolished the reactivity with NADþ totally but lowered the Km for NADPþ with a decrease in catalytic efficiency. This suggests that Asp-36 is absolutely needed for reactivity with NADþ for the double mutants Q15K–W37R and Q15R–W37R although it does not block reactivity with NADPþ as expounded above. Although substitution of Asp-36 by either an alanine or a serine ensures reactivity with NADPþ , it decreased slightly the Km values for NADPþ but increased significantly the Km values for PGE2 as compared to the kinetic constants of the double mutants. Furthermore, it also decreased significantly the catalytic efficiency of the double mutants. The model of the phosphate binding site of 15-PGDH (Fig. 4) shows that the basic residues at positions 15 and 37 poise the basic side chains for direct interactions with the phosphate of NADPþ . A rather interesting result of our modeling relates to the importance of Arg at position 37. The guanidinium group may likely form a stacking interaction with the adenine moiety of NADPþ as has been observed in other proteins [27]. This potential stacking interaction may explain the preference for Arg at position 37 (Figs. 3 and 4). In addition, our model of the NADPþ phosphate binding site reveals that Asp-36 is located underneath the phosphate moiety and its mutation to a Ser favors the main chain hydrogen bonding interaction that could further stabilize the nucleotide binding site, suggesting an explanation for the importance of this site in conferring coenzyme selectivity (Fig. 3). Mutation of Asp-36 to a comparable size of amino acid such as an asparagine may improve catalytic efficiency, although most of the NADPþ preferring enzymes have an alanine or a serine preceding an arginine in the second b strand (Fig. 1). It is of interest to note that the Km values for either NADPþ or PGE2 were decreased
145
Fig. 4. Constructed 3D structural model of 15-PGDH residues 11-40. The models generated from molecular modeling based on the tropinone reductase-I crystallographic structure. (A) Mutant Q15K–D36S– W37R. (B) Wild type.
to the levels comparable to those found in the wild type enzyme when Ala-14 was mutated to a serine commonly found in NADPþ -linked enzymes. Unfortunately, the catalytic efficiency was only slightly increased after such a mutation. Significant increase in catalytic efficiency with NADPþ as an exclusive coenzyme may require further structural changes. In this regard, Chen et al. [28] reported a 60-fold increase in affinity for NADPþ and a 1.5-fold increase in kcat compared to the wild type when Asp-39 was replaced with an asparagine in Drosophila alcohol dehydrogenase. Furthermore, adding a positive charge by mutation of Ala-46 to an arginine lowered the Km for NADPþ about 2.5-fold and increased kcat about
146
H. Cho et al. / Archives of Biochemistry and Biophysics 419 (2003) 139–146
3-fold with NADPþ as compared to the wild type alcohol dehydrogenase [29]. It will be interesting to see if these corresponding changes may bring about significant increase in affinity for NADPþ and catalytic efficiency for 15-PGDH. Acknowledgment This work was supported in part by a Grant from NIH (HL-46296). References [1] E. Anggard, B. Samuelsson, J. Biol. Chem. 239 (1964) 4097–4102. [2] C.M. Ensor, H.H. Tai, J. Lipid Mediat. Cell. Signal. 12 (1995) 313–319. [3] B. Wermuth, J. Biol. Chem. 256 (1981) 1206–1213. [4] C.M. Ensor, J.Y. Yang, R.T. Okita, H.H. Tai, J. Biol. Chem. 265 (1990) 14888–14891. [5] H. Jornvall, B. Persson, M. Krook, S. Atrian, R. GonzalezDuarte, J. Jeffery, D. Ghosh, Biochemistry 34 (1995) 6003–6013. [6] M.G. Rossmann, A. Liljas, C.I. Branden, L.J. Banaszak, in: P.D. Boyer (Ed.), The Enzymes, vol. 2, third ed, Academic Press, New York, 1975, pp. 61–102. [7] C.M. Ensor, H.H. Tai, Biochim. Biophys. Acta 1208 (1994) 151– 156. [8] C.M. Ensor, H.H. Tai, Biochem. Biophys. Res. Commun. 220 (1996) 330–333. [9] D. Ghosh, V.Z. Pletnev, D.W. Zhu, Z. Wawrzak, W.L. Duax, W. Pangborn, F. Labrie, S.X. Lin, Structure 3 (1995) 503–513. [10] M.W. Sawicki, M. Erman, T. Puranen, P. Vihko, D. Ghosh, Proc. Natl. Acad. Sci. USA 96 (1999) 840–845. [11] H. Cho, H.H. Tai, Prostaglandins Leukot. Essent. Fatty Acids 66 (2002) 505–509.
[12] C.M. Ensor, H.H. Tai, Arch. Biochem. Biophys. 333 (1996) 117– 120. [13] H.P. Zhou, H.H. Tai, Biochem. Biophys. Res. Commun. 257 (1999) 414–417. [14] N. Tanaka, T. Nonaka, M. Nakanishi, Y. Deyashiki, A. Hara, Y. Mitsui, Structure 4 (1996) 33–45. [15] N. Tanaka, T. Nonaka, T. Tanabe, T. Yoshimoto, D. Tsura, Y. Mitsui, Biochemistry 35 (1996) 7715–7730. [16] K. Nakajima, A. Yamashita, H. Akama, T. Nakatsu, H. Kato, T. Hashimoto, J. Oda, Y. Yamada, Biochemistry 95 (1998) 4876– 4881. [17] A. Yamashita, H. Kato, S. Wakatsuki, S. Tomizaki, T. Nakatsu, K. Nakajima, T. Hashimoto, Y. Yamada, J. Oda, J. Biochemistry 38 (1999) 7630–7637. [18] J. Braman, C. Papworth, A. Greener, Methods Mol. Biol. 57 (1996) 5731–5744. [19] D.T. Kung-Chao, H.H. Tai, Biochim. Biophys. Acta 614 (1980) 1–13. [20] R.K. Wierenga, M.C.H. De Mayer, W.G.J. Hol, Biochemistry 24 (1985) 1346–1357. [21] D. Ghosh, Z. Wawrzak, C.M. Weeks, W.L. Duax, M. Erman, Structure 2 (1994) 629–640. [22] K.I. Varughese, N.H. Xuong, P.M. Kiefer, D.A. Matthews, J.M. Whitely, Proc. Natl. Acad. Sci. USA 91 (1994) 5582–5586. [23] A.J. Chavan, C.M. Ensor, P. Wu, B.E. Haley, H.H. Tai, J. Biol. Chem. 268 (1993) 16442–17437. [24] N. Tanaka, T. Nonaka, M. Nakanishi, Y. Deyashiki, A. Hara, Y. Mitsui, Structure 4 (1996) 33–45. [25] Y.W. Huang, I. Pineau, H.J. Chang, A. Azzi, V. Bellemare, S. Laberge, S.X. Lin, Mol. Endocrinol. 15 (2001) 2010–2020. [26] I. Tsigelny, B.E. Baker, Biochem. Biophys. Res. Commun. 226 (1996) 118–127. [27] J.B. Mitchell, C.L. Nandi, I.K. McDonald, J.M. Thornton, S.L. Price, J. Mol. Biol. 239 (1994) 315–331. [28] Z. Chen, Z.G. Lin, W.R. Lee, S.H. Chang, Eur. J. Biochem. 202 (1991) 263–267. [29] Z. Chen, I. Tsigelny, W.R. Lee, M.E. Baker, S.H. Chang, FEBS Lett. 356 (1994) 81–85.