Analysis of the substrate-binding site of human carbonyl reductases CBR1 and CBR3 by site-directed mutagenesis

Chemico-Biological Interactions 178 (2009) 234–241

Contents lists available at ScienceDirect

Chemico-Biological Interactions journal homepage: www.elsevier.com/locate/chembioint

Analysis of the substrate-binding site of human carbonyl reductases CBR1 and CBR3 by site-directed mutagenesis Yasser El-Hawari a , Angelo D. Favia b , Ewa S. Pilka c , Michael Kisiela a , Udo Oppermann c , Hans-Jörg Martin a , Edmund Maser a,∗ a b c

Institute of Toxicology and Pharmacology for Natural Scientists, University Medical School Schleswig-Holstein, Campus Kiel, Brunswikerstr. 10, D-24105 Kiel, Germany European Molecular Biology Laboratory-European Bioinformatics Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge CB10 1SD, UK University of Oxford, Structural Genomics Consortium, Botnar Research Centre, Oxford OX3 7LD, UK

a r t i c l e

i n f o

Article history: Available online 18 November 2008 Keywords: Carbonyl reduction CBR1 CBR3 Mutagenesis Active site

a b s t r a c t Human carbonyl reductase is a member of the short-chain dehydrogenase/reductase (SDR) protein superfamily and is known to play an important role in the detoxification of xenobiotics bearing a carbonyl group. The two monomeric NADPH-dependent human isoforms of cytosolic carbonyl reductase CBR1 and CBR3 show a sequence similarity of 85% on the amino acid level, which is definitely high if compared to the low similarities usually observed among other members of the SDR superfamily (15–30%). Despite the sequence similarity and the similar features found in the available crystal structures of the two enzymes, CBR3 shows only low or no activity towards substrates that are metabolised by CBR1. This surprising substrate specificity is still not fully understood. In the present study, we introduced several point mutations and changed sequences of up to 17 amino acids of CBR3 to the corresponding amino acids of CBR1, to gather insight into the catalytic mechanism of both enzymes. Proteins were expressed in Escherichia coli and purified by Ni-affinity chromatography. Their catalytic properties were then compared using isatin and 9,10-phenanthrenequinone as model substrates. Towards isatin, wild-type CBR3 showed a catalytic efficiency of 0.018 ␮M−1 min−1 , whereas wild-type CBR1 showed a catalytic efficiency of 13.5 ␮M−1 min−1 . In particular, when nine residues (236–244) in the vicinity of the catalytic center and a proline (P230) in CBR3 were mutated to the corresponding residues of CBR1 a much higher kcat /Km value (5.7 ␮M−1 min−1 ) towards isatin was observed. To gain further insight into the protein-ligand binding process, docking simulations were perfomed on this mutant and on both wild-type enzymes (CBR1 and CBR3). The theoretical model of the mutant was ad hoc built by means of standard comparative modelling. © 2008 Elsevier Ireland Ltd. All rights reserved.

1. Introduction The short-chain dehydrogenases/reductases (SDR) represent the largest protein superfamily catalyzing oxidoreductions, with more than 70 genes identified in the human genome. Carbonyl reductase 1 and 3 (CBR1, CBR3) are monomeric NADPH-dependent representatives of the predominantly dimeric or tetrameric SDRs. Both enzymes consist of 277 amino acids. They show an amino acid identity of 72% and a similarity of 85%, which is remarkably high

Abbreviations: CBR1, carbonyl reductase 1; CBR3, carbonyl reductase 3; DMSO, dimethylsulfoxide; IPTG, isopropyl-␤-thiogalactopyranoside; PAGE, polyacrylamide gel electrophoresis; wwPDB, worldwide protein data bank; NNK, nicotine-derived N-nitrosamine ketone; SDR, short-chain dehydrogenases/reductases; SDS, sodium dodecyl sulfate. ∗ Corresponding author. Tel.: +49 431 597 3540; fax: +49 431 597 3558. E-mail address: [email protected] (E. Maser). 0009-2797/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cbi.2008.11.004

compared to the usually observed low identity levels of 15–30% among other SDRs [1]. CBR1 is ubiquitously distributed with highest concentrations found in liver and brain [2,3], and is involved in a wide range of metabolic pathways. It was first identified in 1973 [4] and since then has been the subject of many examinations (see [2,5–7] and literature cited therein). Reductase activity towards prostaglandins and lipid peroxidation products has been described for the enzyme [8,9], and a neuroprotective role was predicted [9,10]. Isatin, an endogenous compound, whose functions include, among others, the inhibition of monoaminooxidase and atrial natriuretic peptide receptor binding activities [11,12] was found to be reduced by CBR1. Moreover, xenobiotics reduced by CBR1 are the cytostatic drugs doxorubicin [13] and daunorubicin [14], quinones [15], the antihistaminic dolasetron and the carcinogenic nitrosamine of tobacco smoke NNK [16,17]. The regulation of CBR1 is described in Ref. [18]. In contrast, CBR3 was discovered only recently, its gene has been identified in 1998 [19] and its physiological role is still unknown.

Y. El-Hawari et al. / Chemico-Biological Interactions 178 (2009) 234–241

Thanks to the availability of experimentally determined structures of the enzymes, the comparison of the catalytic clefts could shed light on the different substrate specificity. Therefore, we constructed 10 mutants of CBR3 by single amino acid substitution and/or replacement of short fragments comprising up to 15 amino acids of the corresponding CBR1 sequence. Kinetic values for isatin and 9,10-phenanthrenequinone reduction were determined for the mutants and both wild-types. Furthermore, in order to better interpret the experimental data and in the hope of unveiling the role of some key residues during binding, isatin and phenanthrenequinone were docked to CBR1 and CBR3 wild-type proteins and to a selected mutant (residues 230 and 236–244 of CBR3 mutated to CBR1). The analysis of the docking poses along with their associated estimates of binding affinities highlighted the importance of an aromatic side-chain at position 230 in binding the substrate. Our results indicate that the different substrate specificity and activity of CBR1 and CBR3 are not depending on single residues.

Table 2 Details on construction and cloning of wild-type enzymes and mutants. Clone

All primers were synthesized by MWG-Biotech (Ebersberg, Germany). Pfu polymerase and the “Rapid Ligation Kit” were purchased from Fermentas (St. Leon-Rot, Germany). Taq polymerase, HotStar Taq polymerase and the QIAquick Gel extraction kit were from Qiagen (Hilden, Germany). Enzymes from New England Biolabs (Frankfurt am Main, Germany) were used for DNA restriction. Preparative agarose gels were made with agarose from Biozym (Hessisch Oldendorf, Germany) and stained with SybrSafe from Invitrogen (California, USA). Ampicillin was obtained from AGS (Heidelberg, Germany), IPTG, NADPH and agarose NEEO ultra for analytical agarose gels were purchased from Carl Roth (Karlsruhe, Germany). DMSO, menadione, glycerol, imidazol and cell strains were obtained from Merck (Darmstadt, Germany), Isatin, 9,10-phenanthrenequinone, Tris, agar, tryptone, yeast extract and lysozyme from Sigma–Aldrich (Munich, Germany). 2.2. Mutagenesis and cloning The mutations were introduced using the splicing by overlap extension method [20]. Details are displayed in Tables 1 and 2. Table 1 PCR primers (5 to 3 ) used for construction and cloning of wild-type enzymes and mutants. #

Sequence

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

GGACTCGAGTTACCAGTTTTGCACAAC GCACATATGTCGTCCTGCAGCCGCGTG GACGACCTCGAGTTACCAGTTTTGCACAAC GTGAGAGCACATATGTCGTCCTGCAGCCGCGTG ACAGCATCAGGACTATGGAGGAGGGGGCT AGCCCCCTCCTCCATAGTCCTGATGCTGT TGGGCTCTTGGTGGCCTTGGGTCCCGCCATGTCTGTCTTCACTGGT GCGGGACCCAAGGCCACCAAGAGCCCAGAGGAGGGGGCTGAGACCC GAGGAGGGGGCTGAGACCCCTGTGTACTTG GCGTGCTGCCCAGGATGGGTGAAGACAGACATG CATGTCTGTCTTCACCCATCCTGGGCAGCACGC GCCGTCGCCTTCAAGGTTGCTGATCCAATGCCCTTT AAAGGGCATTGGATCAGCAACCTTGAAGGCGACGGC GACGACCTCGAGTTACCAGTTTTGCACAACTTTGTCACTGACCAA GACGACCTCGAGTCACCACTGTTCAAC GTGAATATCAGTAGTTTGATGAGCTTAAGGGCTTTTGAA TTCAAAAGCCCTTAAGCTCATCAAACTACTGATATTCAC ACGACGCATATGTCGTCCGGCATCCATGTAGCGCTGGTGACT TCACCACTGTTCAACTCTCTTCTC ATGTCGTCCGGCATCCAT

PCR #

Template

Primers

CBR3

1

IRAUp969E0122D6

1, 2

A

1 2 3

CBR3 CBR3 PCR1, PCR2

4, 6 3, 5 3, 4

B

1 2 3

CBR3 CBR3 PCR1, PCR2

4, 7 3, 8 3, 4

C

1 2

CBR1 CBR3, PCR1

9, 15 4, 15

D

1 2

CBR1 B, PCR1

9, 15 4, 15

E

1 2 3

CBR3 CBR3 PCR1, PCR2

4, 11 3, 10 3, 4

F

1 2 3

B B PCR1, PCR2

4, 11 3, 10 3, 4

G

1 2 3

F F PCR1, PCR3

4, 13 4, 12 3, 4

H

1 2 3

F F PCR1, PCR2

4, 17 3, 16 3, 4

I J

1 1

F H

4, 16 4, 16

CBR1

1 2

cDNA library PCR1

19, 20 15, 18

2. Materials and methods 2.1. Chemicals and kits

235

CBR3 wild-type DNA was purchased from RZPD (Berlin, Germany). Clones I and J: Taq polymerase was used instead of Pfu polymerase because its proof-reading activity prevents mutation, Taq polymerase PCR conditions according to Pfu polymerase conditions but 25 cycles, annealing temperature 45 ◦ C, elongation time 1 min. CBR1 DNA was amplified with Qiagen HotStar Taq polymerase from human liver cDNA with an annealing temperature of 55 ◦ C and 40 cycles.

PCR conditions for assays with Pfu polymerase: initial denaturation for 3 min at 94 ◦ C, 1 min denaturation at 94 ◦ C, 1 min annealing from 44 to 60 ◦ C, elongation for 2 min at 72 ◦ C, 25–36 cycles and a final elongation step of 10 min. Conditions for Taq polymerase were identical to those of Pfu polymerase but 25 cycles were used with an annealing temperature of 45 ◦ C and 1 min of elongation. PCR reactions were analysed via agarose gel electrophoresis. Appropriate DNA bands were excised and centrifuged through glass wool to remove the agarose. A flow-through volume of 1 ␮l was used as a template for consecutive reactions. All constructs were purified via preparative agarose gel electrophoresis and the Qiagen gel extraction kit. Purified DNA was ligated into Novagen’s pET-15b vector using the NdeI/XhoI restriction sites and transformed into calcium chloride competent Novagen XL1Blue cells using the heat shock method. Transformed cells were screened with colony PCR and purified DNA was sequenced (Seqlab, Göttingen, Germany) for verification. 2.3. Expression and purification of proteins All constructs were transformed into calcium chloride competent Novagen OrigamiTM (DE3) cells using the heat shock method. At 37 ◦ C, 200 ml of Lysogeny Broth medium containing 100 ␮g/ml of ampicillin were inoculated with 2 ml of a preculture and cultivated with shaking at 180 rpm. Protein expression was induced at an OD600 of 0.6 by adding IPTG to a concentration of 1 mM. After 3–4 h, cultures were chilled on an ice bath. All following steps were performed at 4 ◦ C or by cooling on ice. The cells were harvested by

236

Y. El-Hawari et al. / Chemico-Biological Interactions 178 (2009) 234–241

Fig. 1. Chemical structures of the substrates used in this study.

centrifugation at 3000 × g for 15 min. After resuspending the pellet in 10 ml of 20 mM sodium dihydrogenphosphate, pH 7.4, containing 10 mM imidazole, 500 mM sodium chloride and 10% (v/v) glycerol, 10 mg lysozyme were added and the mixture was incubated for 30 min. The lysed cells were disrupted by sonication (Sonoplus, Bandelin, Berlin, Germany) using six bursts of 10 s at 200 W with 20 s breaks in between. After centrifugation at 100,000 × g for 1 h, the supernatant was purified by Ni-affinity chromatography with the Äkta purifier and HisTrap HP columns (1 ml, GE Healthcare, New Jersey, USA). A step gradient with the resuspension buffer containing 500 mM imidazole was used to purify and elute the protein. Two consecutive washing steps with 10% and 20% (v/v) of 500 mM imidazole buffer were performed to remove unspecific bound proteins. Overexpressed proteins were eluted with 30% (v/v) of the 500 mM imidazole buffer. The protein concentration was determined using the Bradford method with BSA as standard. The purity of the eluted fractions was controlled by SDS gel electrophoresis and Coomassie Blue staining. Assessed samples containing pure protein were either directly used for further analysis or stored at

4 ◦ C. Under these conditions no loss of activity was observed for at least 1 month. 2.4. Kinetics Catalytic properties were determined by measuring the decrease in absorbance at 340 nm (Cary 100 scan photometer, Varian, California, USA). A reaction mixture consisted of substrate, 500 ␮M NADPH, 100 mM Tris–HCl, pH 7.4, and enzyme. The enzyme solution was diluted in the corresponding elution buffer, a 7:3 mixture of 10:500 (mM) imidazole buffer, to ensure that substrate consumption was linear over time. Unless otherwise stated, for each of two enzyme preparations a minimum of six concentrations of substrate was used. Each concentration was measured at least three times. The reaction temperature was held constant at 25 ◦ C. After a preincubation time of 2 min 10 ␮l of enzyme solution were added to 790 ␮l of reaction mixture. A reference cuvette contained the reaction solution without enzyme. Isatin, menadione and 9,10-

Table 3 Kinetic constants for wild-type enzymes and generated mutants for isatin and 9,10-phenanthrenequinone. #

Construct scheme

Residues exchanged

Isatin Km (␮M)

9,10-Phenanthrene-quinone −1

kcat /Km (min

−1

␮M

)

Km (␮M)

kcat /Km (min−1 ␮M−1 )

CBR3

Wild-type CBR3

>4000

0.018b

>78

0.12b

A

V244M variant

>4000

0.012a

>78

0.14b

B

236–244

0.240 ± 0.0053

>78

0.54b

C

262–277

>4000

0.016b

>78

D

236–244, 262–277

>4000

0.160a

>78

0.07b

E

230

>4000

0.006b

>78

0.07b

F

230, 236–244

47.1 ± 2.1

5.3 ± 0.074

G

97–98, 230, 236–244

55 ± 4.0

5.7 ± 0.18

H

142–143, 230, 236–244

31.5 ± 2.0

15.5 ± 0.34

I

230, 236–244, 270

32.7 ± 2.8

6.6 ± 0.18

8.9 ± 0.83

J

142–143, 230, 236–244, 270

22.5 ± 1.7

18.9 ± 0.45

4.3 ± 0.37

CBR1

Wild-type CBR1

7.8 ± 1.7

13.5 ± 0.81

1600 ± 83

56.4 ± 8.8 9.2 ± 0.91 39.3 ± 1.3

35.4 ± 5.3

–c

5.7 ± 0.5 25.1 ± 0.70 22.0 ± 0.35

24.0 ± 0.58 107.9 ± 2.2 8.9 ± 0.6

In the construct schemes, gray triangles symbolize single residues, gray bars symbolize larger regions mutated to those found in CBR1 amino acids (except mutant A, an allelic variant of CBR3 wild-type). Where Km and kcat /Km are reported as the mean ± asymptotic standard errors, data result from two different protein expressions, a minimum of six substrate concentrations and n ≥ 6. a The solubility of the highest measurable substrate concentration was lower than the calculated Km values, only kcat /Km is reported. b The data result from the slope of linear regression of the relation between activity and substrate concentration as no Michaelis–Menten kinetic could be observed (≥4 concentrations, n ≥ 3). c Activity at the highest substrate concentration was hardly detectable.

Y. El-Hawari et al. / Chemico-Biological Interactions 178 (2009) 234–241

237

Fig. 2. ClustalX alignment of CBR3 and CBR1 with consensus bar. Identical, strongly conservative, and weakly conservative residues are indicated by asterisks, colons, and dots. Mutated residues are highlighted in gray, the amino acids of the catalytic tetrad are framed black.

phenanthrenequinone (Fig. 1) stock solutions were prepared in DMSO. The final concentration of DMSO in the reaction mixture was 10% (v/v). A maximum of 4000 ␮M isatin was used in the kinetic measurement as the change in absorbance of this concentration still follows Lambert–Beer’s law and no precipitation of isatin occurred. The maximum soluble concentration of 9,10-phenanthrenequinone was 78 ␮M. Menadione stock solutions were prepared up to a concentration of 120 ␮M. The kinetic constants were calculated by nonlinear regression (Gnuplot 4.2) with a molar extinction coefficient for NADPH of 6.22 × 103 M−1 cm−1 . 2.5. Docking calculations For the docking calculations we took into account the crystal structures of human carbonyl reductases 1 and 3 available

at the Worldwide Protein Data Bank (wwPDB) [21] (PDB codes 1wma [22] and 2hrb [23], respectively). The protein structures were treated with the protein preparation wizard in Maestro (Version 8.5, Schrödinger, LLC, New York, NY, 2008), whilst the simulations were performed using Glide (Version 5.0, Schrödinger, LLC, New York, NY, 2008). Isatin and 9,10-phenanthrenequinone were manually prepared in Maestro. During the protein preparation step, all water molecules were removed, hydrogen atoms were added and the structures were energy minimised using the recommended settings. The mutant F (see Table 3) model was prepared as explained in the comparative modelling section. The enclosing docking boxes, whose dimensions in each direction were 46 Å, were centered on the nicotinamide C4 atom of each protein. The docking runs were performed in Standard Precision (SP) mode and in order to soften the potential for nonpolar parts of the ligands, the van der Waals

238

Y. El-Hawari et al. / Chemico-Biological Interactions 178 (2009) 234–241

radii of ligand atoms with partial atomic charge lower than 0.15 were scaled by a factor of 0.8. Only the top five energy-ranked poses were retained and energy minimised for each studied compound. The top-scoring conformer for each ligand was identified on the basis of the Emodel score whilst the subsequent comparison between different compounds was performed taking into account their corresponding Gscore values (this is a recommended procedure in Glide). 2.6. Comparative modelling The 3D model of mutant F (see Table 3) was built using the spatial features extracted from the crystallographic data of human CBR3 available at the wwPDB. The only modifications were carried out on the residues 230 and 236–244 which were mutated to the corresponding residues of human CBR1. The model was built using the default settings in the Structure Prediction tool available in Prime (Version 2.0, Schrödinger, LLC, New York, NY, 2008). Once obtained, the model was further refined in order to drive the structure to its closer local energy-minimum by removing possible steric clashes. Several models were generated and ranked on the basis of the internal Prime scoring function. The top-scoring one was eventually used to perform docking simulations. 3. Results and discussion As shown in Fig. 2, the alignment of the protein sequences of human CBR1 and human CBR3 reveals a high similarity between the two isoforms. To understand the unexpected differences in substrate specificity, we generated ten mutant CBR3 proteins with amino acids exchanged to the respective CBR1 sequence. The selec-

Fig. 3. Crystal structure of CBR3 (wwPDB code 2hrb) with a selection of mutated residues shown in green, NADPH in brown, C4 in black, amino acids from the catalytic tetrad in blue. The surface of amino acids surrounding the cofactor at a distance of 4 Å is displayed in gray. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)

tion of the exchanged amino acids was based on the sequence alignment and on the study of the known crystal structures of CBR1 [22,24] and CBR3 [23] available at the wwPDB. Residues of the catalytic cleft with different physico-chemical properties are shown in Fig. 3. As it can be seen from the consensus bar in Fig. 2, several regions that significantly differ between the two enzymes, were identified. All mutants were successfully overexpressed in E. coli and purified in a soluble and active form. The histidine-tagged proteins

Fig. 4. Plots resulting from the nonlinear fitting (Gnuplot v. 4.2) to the Michaelis–Menten equation.

Y. El-Hawari et al. / Chemico-Biological Interactions 178 (2009) 234–241

239

Fig. 5. Top scoring docking poses of 9,10-phenanthrenequinone at CBR1, CBR3 and Clone F active sites (a, b and c, respectively). Residues at positions 96, 142, 194, 230, NADPH and 9,10-phenanthrenequinone are depicted as stick models. To help interpretation, atoms are coloured according to the atom-code (C atoms in light blue for the protein residues, in yellow for the cofactor and in blue for the substrate). Dashed lines are drawn between atoms likely involved in H-bond interactions. Van der Waals surfaces are highlighted in transparent red, when favourable interactions occur. Isatin (not shown) docked in a similar fashion to the three enzymes. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)

displayed single bands on Coomassie-stained SDS gels. Catalytic properties obtained from 9,10-phenanthrenequinone and isatin are listed in Table 3 and a selection of Michaelis–Menten plots is shown in Fig. 4. 3.1. Comparison of CBR1 and CBR3 wild-type enzymes Of the two isoforms of human carbonyl reductase, CBR1 is well characterized in terms of substrate specificity, tissue distribution and regulation [3,15,16,18]. While CBR1 plays an important role in the reduction and detoxification of many endobiotic and xenobiotic carbonyl compounds, only very limited data are available for CBR3. Although the amino acid sequence identity between the two enzymes is greater than 70% (sequence similarity = 85%), menadione was the only substrate reported for CBR3 [25]. However, wild-type CBR3 and a natural allelic variant carrying a point mutation (V244M) were inactive towards menadione in our study (Table 3, construct A). This is in contrast to the study published by Lakhman et al. [25] who reportet Km values in the low micromolar range and a catalytic efficiency of 54 min−1 ␮M−1 . When changing the E. coli expression strains, cell lysis methods or purification conditions in our system, CBR3 did not yield any activity with menadione. Very recently, Miura et al. [26] reported a similar finding to ours. Their recombinant wild-type human CBR3 showed very low activity towards menadione, roughly 100-fold less than that reported by Lakhman et al. [25]. The reason for this discrepancy is unknown at present. It should be noted here that the activity of CBR1 with menadione served as a positive control for the assay, and activities observed in our studies correspond to those observed previously [27]. On the other hand, we found that wild-type CBR3 and the V244M variant exhibited activity towards isatin and 9,10phenanthrenequinone as substrates. Isatin could be dissolved to concentrations higher than 4000 ␮M but the loss of linearity in absorption was the limiting factor at higher concentrations. In the case of 9,10-phenanthrenequinone, its limited solubility and low conversion rates prevented us from determining Km and kcat for CBR3 wild-type. Values for kcat /Km though were available (Table 3). These two substances however, being the only substrates for CBR3 and the allelic variant (Table 3, construct A) we know to date, served as model substrates for further characterization of the wild-type and mutant enzymes. 3.2. Kinetic properties of constructs B–E An obvious candidate that might explain the striking differences of substrate specificity between the two isoforms is tryptophan 230 of CBR1 that is substituted for a proline in CBR3. Tryptophan 230, a highly conserved residue in most CBR proteins, is located

close to the cofactor pyridine ring. Moreover, when aligning the nucleotide sequences of human CBR1 and CBR3 it turned out that this is one of only two cases where the entire triplet encoding amino acids of CBR1 is completely different in all three bases from the CBR3 gene. However, the enzymatic activity towards isatin and 9,10-phenanthrenequinone of the CBR3-P230W mutant was not significantly different from the wild-type (Table 3, construct E). Clearly, the mutation of this single residue alone cannot explain the observed differences between the two enzymes. The sequence alignment of human CBR1 and CBR3 (Fig. 2) reveals a stretch of nine amino acids at positions 236–244 with low sequence similarity. This site is located close to the catalytic cleft and is sterically detached from the otherwise globular protein. Changing these residues resulted in the chimeric CBR3 mutant B (Table 3) which reduced isatin with a Km of 1600 ␮M and a kcat /Km value of 0.24 min−1 ␮M−1 . The C-terminal part of CBR3, which may play a role in substrate specificity, is also poorly conserved. Changing amino acids 262–277 of CBR3 to the corresponding CBR1 residues resulted in construct C (Table 3). Neither with construct C nor with the double mutant carrying the 236–244 and the 262–277 exchanges (Table 3, construct D) significant increases of activity could be observed. 3.3. Docking experiments with CBR1, CBR3 and construct F Interestingly, the mutational combination of constructs B (amino acids 236–244) and E (P230W) to construct F resulted in an enzyme with significantly higher catalytic efficiency with isatin and 9,10-phenanthrenequinone in comparison to the CBR3 wild-type and constructs A–E (Table 3). In silico obtained poses showed that 9,10-phenanthrenequinone and isatin dock similarly to CBR1 and CBR3 (Fig. 5, for clarity, only 9,10-phenanthrenequinone-binding modes are shown). In CBR1, the substrates interact by means of an H-bond with a conserved tyrosine (Y194) and lie in the catalytic cleft between tryptophan 230 and methionine 142, where favourable hydrophobic/aromatic interactions can be achieved (see Fig. 5a). The spatial proximity to the cofactor’s nicotinamide ring supports the hypothesis of those being productive poses (i.e. leading to a catalytic event). On the other hand, in CBR3 the presence of a proline at position 230 and of a glutamine at position 142 does not seem to make the binding in that area as favourable as in CBR1 (see Fig. 5b). For this reason, the substrates docked in such a way that only the H-bonds with tyrosine 194 were kept. Differences in binding modes usually correspond to differences in binding energy estimates. In this case, binding of phenanthrenequinone to CBR1 was estimated to be tighter than binding to CBR3 (GlideScores: −6.18 and −3.84 kcal/mol, respectively) [28].

240

Y. El-Hawari et al. / Chemico-Biological Interactions 178 (2009) 234–241

Furthermore, docking simulations gave interesting results when carried out on the homology model of mutant F (Table 3). Experimental data showed that this mutant had significantly higher catalytic efficiency towards isatin and 9,10-phenanthrenequinone compared to CBR3. In Fig. 5c the top-scoring pose of 9,10phenanthrenequinone at the mutant active site is shown. The substrate is stabilised within the cleft by an H-bond interaction with lysine 96 and by non-specific aromatic interactions with tryptophan 230. The estimated energy of binding of this complex is intermediate between that of the two wild-type complexes (GlideScore: −4.67 kcal/mol). The analysis of those binding modes clearly seems to highlight the key role of residue 230 in ligand binding. It is less clear how the other mutated residues (236–244) affect the binding, since they seem to be further away from the docked substrates. As a speculation, it is appealing to tag those residues as important in the ligand recognition process via loop motion. Even if molecular mechanics methods (such as standard docking) do not provide information about conversion rates in molecular reactions (during the simulation, no bonds are broken or formed) it is likely that a tighter binding corresponds to a better complementarity which in turn can lead to a more efficient catalysis [29–35]. Overall docking outcomes were found in agreement with experimental data. 3.4. Kinetic properties of constructs G–J The constructs G–J are derived from construct F. Additional mutation of residues 97 and 98 (Table 3, constructs G), located on the opposite “wall” of the catalytic cleft, did not further improve catalytic efficiency for isatin but kinetic constants for 9,10-phenanthrenequinone could now be calculated. It may be speculated that the mutation to small hydrophobic amino acids (Table 3, serine and aspartic acid are exchanged to alanine and valine, respectively) favours the binding of hydrophobic 9,10-phenanthrenequinone into the active site. In contrast, the additional conversion of residues 142 and 143 (Table 3, construct H) nearly tripled the kcat /Km -value for isatin from 5.7 to 15.5 min−1 ␮M−1 , whereas the catalytic efficiency for 9,10phenanthrenequinone did not change. While the influence of the C143S substitution is less clear, the docking results suggest that the Q142M mutation may be advantageous to allow hydrophobic interactions between the methionine and the aromatic rings of the substrates (Fig. 5a). In construct I (Table 3) the hydrophilic and sterically demanding histidine in position 270 was substituted with the smaller and less hydrophilic serine resulting in an increase of the kcat /Km value towards 9,10-phenanthrenequinone. These data are comparable to the effects observed with mutant G where serine and aspartic acid were replaced by valine and alanine (positions 97 and 98, respectively). These results indicate the influence of residues that are localized remote to the cofactor and the catalytic tetrad. Nevertheless, the highest catalytic efficiency for both substrates was observed with construct J (Table 3). This mutant was derived from construct H and beared the additional point mutation H270S. Its catalytic efficiency towards both isatin and 9,10phenanthrenequinone could be further improved. This observation corroborates the results found with mutants H and I (Table 3). The effects of the residues 142, 143 and 270 simply seem to be additive. The data obtained with mutants F, G, H, I and J regarding 9,10-phenanthrenequinone indicate that the essential amino acids for binding this and comparable substrates might be reduced to residues 97, 98, 142, 143, 230 and 236–244. Interestingly, the Km value measured for isatin reduction of CBR1 is still significantly lower, though values for kcat /Km are in the same range of constructs H and J (Table 3).

The complexity of interactions regarding substrate specificity becomes even more apparent when considering the results observed for menadione. This molecule is often used as the standard substrate to determine CBR1 activity. With wild-type CBR1 we could observe a Km of 16.0 ± 3.6 ␮M and a kcat /Km of 2.9 ± 0.18 min−1 ␮M−1 , whereas constructs A–E were completely devoid of menadione reducing activity, and mutants F to J had only low and hardly detectable conversion rates. Looking at the crystal structures, the catalytic cleft of CBR3 has to be regarded as being more hydrophilic in comparison to CBR1. The spectrum of possible substrates for CBR3 might comprise polar substances like sugars or polyols, but this has yet to be tested. In cases like this, where high sequence similarity does not imply similar substrates specificity, the combined study of the catalytic clefts, docking outcomes and in vitro assays can represent an invaluable source of information towards the understanding of enzyme mechanisms. Insofar, the data of this study help to understand the unexpected differences of the two carbonyl human reductase isoforms. Conflict of interest The authors declare that there are no conflicts of interest. Acknowledgement This project was supported by a grant to E.M. from the Deutsche Forschungsgemeinschaft (MA 1704/5-1). References [1] U. Oppermann, C. Filling, M. Hult, N. Shafqat, X. Wu, M. Lindh, J. Shafqat, E. Nordling, Y. Kallberg, B. Persson, H. Jörnvall, Short-chain dehydrogenases/reductases (SDR): the 2002 update, Chem. Biol. Interact. 143–144 (2003) 247–253. [2] F. Hoffmann, E. Maser, Carbonyl reductases and pluripotent hydroxysteroid dehydrogenases of the short-chain dehydrogenase/reductase superfamily, Drug Metab. Rev. 39 (2007) 87–144. [3] H. Wirth, B. Wermuth, Immunohistochemical localization of carbonyl reductase in human tissues, J. Histochem. Cytochem. 40 (1992) 1857–1863. [4] M.M. Ris, J.P. von Wartburg, Heterogeneity of NADPH-dependent aldehyde reductase from human and rat brain, Eur. J. Biochem. 37 (1973) 69–77. [5] G.L. Forrest, B. Gonzalez, Carbonyl reductase, Chem. Biol. Interct. 129 (2000) 21–40. [6] U. Oppermann, Carbonyl reductases: the complex relationships of mammalian carbonyl- and quinone-reducing enzymes and their role in physiology, Annu. Rev. Pharmacol. Toxicol. 47 (2007) 293–322. [7] T. Matsunaga, S. Shintani, A. Hara, Multiplicity of mammalian reductases for xenobiotic carbonyl compounds, Drug Metab. Pharmacokinet. 21 (2006) 1–18. [8] B. Wermuth, Purification and properties of an NADPH-dependent carbonyl reductase from human brain. Relationship to prostaglandin 9-ketoreductase and xenobiotic ketone reductase, J. Biol. Chem. 256 (1981) 1206–1213. [9] J.A. Doorn, E. Maser, A. Blum, D.J. Claffey, D.R. Petersen, Human carbonyl reductase catalyzes reduction of 4-oxonon-2-enal, Biochemistry 43 (2004) 13106–13114. [10] E. Maser, Neuroprotective role for carbonyl reductase? Biochem. Biophys. Res. Commun. 340 (2006) 1019–1022. [11] A.E. Medvedev, A. Clow, M. Sandler, V. Glover, Isatin: a link between natriuretic peptides and monoamines? Biochem. Pharmacol. 52 (1996) 385–391. [12] V. Glover, J.M. Halket, P.J. Watkins, A. Clow, B.L. Goodwin, M. Sandler, Isatin: identity with the purified endogenous monoamine oxidase inhibitor tribulin, J. Neurochem. 51 (1988) 656–659. [13] N. Kassner, K. Huse, H. Martin, U. Gödtel-Armbrust, A. Metzger, I. Meineke, J. Brockmöller, K. Klein, U.M. Zanger, E. Maser, L. Wojnowski, Carbonyl reductase 1 is a predominant doxorubicin reductase in the human liver, Drug Metab. Dispos. 36 (2008) 2113–2120. [14] G.L. Forrest, S. Akman, J. Doroshow, H. Rivera, W.D. Kaplan, Genomic sequence and expression of a cloned human carbonyl reductase gene with daunorubicin reductase activity, Mol. Pharmacol. 40 (1991) 502–507. [15] B. Wermuth, K.L. Platts, A. Seidel, F. Oesch, Carbonyl reductase provides the enzymatic basis of quinone detoxication in man, Biochem. Pharmacol. 35 (1986) 1277–1282. [16] U. Breyer-Pfaff, H. Martin, M. Ernst, E. Maser, Enantioselectivity of carbonyl reduction of 4-methylnitrosamino-1-(3-pyridyl)-1-butanone by tissue fractions from human and rat and by enzymes isolated from human liver, Drug Metab. Dispos. 32 (2004) 915–922.

Y. El-Hawari et al. / Chemico-Biological Interactions 178 (2009) 234–241 [17] A. Atalla, U. Breyer-Pfaff, E. Maser, Purification and characterization of oxidoreductases-catalyzing carbonyl reduction of the tobacco-specific nitrosamine 4-methylnitrosamino-1-(3-pyridyl)-1-butanone (NNK) in human liver cytosol, Xenobiotica 30 (2000) 755–769. [18] S.S. Lakhman, X. Chen, V. Gonzalez-Covarrubias, E.G. Schuetz, J.G. Blanco, Functional characterization of the promoter of human carbonyl reductase 1 (CBR1). Role of XRE elements in mediating the induction of CBR1 by ligands of the aryl hydrocarbon receptor, Mol. Pharmacol. 72 (2007) 734–743. [19] K. Watanabe, C. Sugawara, A. Ono, Y. Fukuzumi, S. Itakura, M. Yamazaki, H. Tashiro, K. Osoegawa, E. Soeda, T. Nomura, Mapping of a novel human carbonyl reductase, CBR3, and ribosomal pseudogenes to human chromosome 21q22.2, Genomics 52 (1998) 95–100. [20] A.N. Vallejo, R.J. Pogulis, L.R. Pease, In vitro synthesis of novel genes: mutagenesis and recombination by PCR, PCR Methods Appl. 4 (1994) S123–130. [21] H. Berman, K. Henrick, H. Nakamura, Announcing the worldwide Protein Data Bank, Nat. Struct. Biol. 10 (2003) 980. [22] M. Tanaka, R. Bateman, D. Rauh, E. Vaisberg, S. Ramachandani, C. Zhang, K.C. Hansen, A.L. Burlingame, J.K. Trautman, K.M. Shokat, C.L. Adams, An unbiased cell morphology-based screen for new, biologically active small molecules, PLoS Biol. 3 (2005) e128. [23] E.S. Pilka, F.H. Niesen, Y. El-Hawari, W. Lee, J.E. Dunford, G. Kochan, V. Wsol, H. Martin, E. Maser, U. Oppermann, Structural basis for substrate specificity in human monomeric carbonyl reductases, submitted for publication. [24] K.M. Bohren, B. Wermuth, D. Harrison, D. Ringe, G.A. Petsko, K.H. Gabbay, Expression, crystallization and preliminary crystallographic analysis of human carbonyl reductase, J. Mol. Biol. 244 (1994) 659–664. [25] S.S. Lakhman, D. Ghosh, J.G. Blanco, Functional significance of a natural allelic variant of human carbonyl reductase 3 (CBR3), Drug Metab. Dispos. 33 (2005) 254–257.

241

[26] T. Miura, T. Nishinaka, T. Terada, Different functions between human monomeric carbonyl reductase 3 and carbonyl reductase 1, Mol. Cell. Biochem. 315 (2008) 113–121. [27] N. Usami, K. Kitahara, S. Ishikura, M. Nagano, S. Sakai, A. Hara, Characterization of a major form of human isatin reductase and the reduced metabolite, Eur. J. Biochem. 268 (2001) 5755–5763. [28] R.A. Friesner, R.B. Murphy, M.P. Repasky, L.L. Frye, J.R. Greenwood, T.A. Halgren, P.C. Sanschagrin, D.T. Mainz, Extra precision glide: docking and scoring incorporating a model of hydrophobic enclosure for protein–ligand complexes, J. Med. Chem. 49 (2006) 6177–6196. [29] A.D. Favia, I. Nobeli, F. Glaser, J.M. Thornton, Molecular docking for substrate identification: the short-chain dehydrogenases/reductases, J. Mol. Biol. 375 (2008) 855–874. [30] J.C. Hermann, E. Ghanem, Y. Li, F.M. Raushel, J.J. Irwin, B.K. Shoichet, Predicting substrates by docking high-energy intermediates to enzyme structures, J. Am. Chem. Soc. 128 (2006) 15882–15891. [31] J.C. Hermann, R. Marti-Arbona, A.A. Fedorov, E. Fedorov, S.C. Almo, B.K. Shoichet, F.M. Raushel, Structure-based activity prediction for an enzyme of unknown function, Nature 448 (2007) 775–779. [32] J.J. Irwin, F.M. Raushel, B.K. Shoichet, Virtual screening against metalloenzymes for inhibitors and substrates, Biochemistry 44 (2005) 12316–12328. [33] C. Kalyanaraman, K. Bernacki, M.P. Jacobson, Virtual screening against highly charged active sites: identifying substrates of alpha-beta barrel enzymes, Biochemistry 44 (2005) 2059–2071. [34] A. Macchiarulo, I. Nobeli, J.M. Thornton, Ligand selectivity and competition between enzymes in silico, Nat. Biotechnol. 22 (2004) 1039–1045. [35] S. Tyagi, J. Pleiss, Biochemical profiling in silico—predicting substrate specificities of large enzyme families, J. Biotechnol. 124 (2006) 108–116.