The binding of aristolochic acid I to the active site of human cytochromes P450 1A1 and 1A2 explains their potential to reductively activate this human carcinogen

Cancer Letters 229 (2005) 193–204 www.elsevier.com/locate/canlet

The binding of aristolochic acid I to the active site of human cytochromes P450 1A1 and 1A2 explains their potential to reductively activate this human carcinogen Marie Stiborova´a,*, Bruno Sopkoa, Petr Hodeka, Eva Freib, Heinz H. Schmeiserb, Jirˇ´ı Hudecˇeka a

Department of Biochemistry, Faculty of Science, Charles University, Albertov 2030, 128 40 Prague 2, The Czech Republic Division of Molecular Toxicology, German Cancer Research Center, In Neunheimer Feld 280, 69120 Heidelberg, Germany

b

Received 1 May 2005; received in revised form 25 June 2005; accepted 29 June 2005

Abstract Aristolochic acid (AA), a naturally occurring nephrotoxin and carcinogen, has been associated with the development of urothelial cancer in humans. Using the 32P-postlabeling assay we showed that AAI is activated by human recombinant cytochrome P450 (CYP) 1A1, CYP1A2 and NADPH:CYP reductase to species generating DNA adduct patterns reproducing those found in renal tissues from humans exposed to AA. 7-(Deoxyadenosin-N6-yl)aristolactam I, 7-(deoxyguanosin-N2-yl) aristolactam I and 7-(deoxyadenosin-N6-yl)aristolactam II were identified as AA-DNA adducts formed from AAI by the enzymes. The formation of these AA-derived DNA adducts indicates that all the human enzymes reduce the nitro group of AAI to the putative reactive cyclic nitrenium ion responsible for adduct formation. The concentrations of AAI required for its halfmaximum DNA binding were 38, 65 and 126 mM AAI for reductive activation by human CYP1A2, CYP1A1 and NADPH:CYP reductase, respectively. CYP1A1 and 1A2 homology modeling followed by docking of AAI to the CYP1A1 and 1A2 active centers was utilized to explain the potential of these enzymes to reduce AAI. Models of human CYP1A1 and 1A2 were constructed on the basis of the crystallographic structure of truncated mammalian CYP enzymes, CYP2B4, 2C5, 2C8, 2C9 and 3A4. The in silico docking of AAI to the active sites of CYP1A1 and 1A2 indicates that AAI binds as an axial ligand of the heme iron and that the nitro group of AAI is in close vicinity to the heme iron of CYP1A2 in an orientation allowing the efficient reduction of this group observed experimentally. The orientation of AAI in the active centre of CYP1A1 however causes an interaction of the heme iron with both the nitro- and the carboxylic groups of AAI. This observation explains the lower reductive potential of CYP1A1 for AAI than CYP1A2, detected experimentally. q 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Aristolochic acid; Carcinogen; Urothelial cancer; Cytochromes P450 1A1 and 1A2, reductive activation; DNA adduct; Computer modeling; Docking

* Corresponding author. Tel.: C420 2 2195 1285; fax: C420 2 2195 1283. E-mail address: [email protected] (M. Stiborova´).

0304-3835/$ - see front matter q 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.canlet.2005.06.038

194

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

1. Introduction The so-called Chinese herbs nephropathy (CHN), a unique type of rapidly progressive renal fibrosis associated with the prolonged intake of Chinese herbs during a slimming regimen, was observed for the first time in Belgium in 1991 [1,2]. About 100 CHN cases have been identified so far in Belgium, half of which needed renal replacement therapy, mostly including renal transplantation [3–5]. The observed nephrotoxicity has been traced to the ingestion of Aristolochia fangchi containing carcinogenic and nephrotoxic aristolochic acid (AA) inadvertently included in slimming pills [2]. CHN patients (about 170 cases) [6] have been identified in other European countries, in Asia and in the USA, who were exposed to Aristolochia species containing AA and had no relationship with the Belgian slimming clinic. Therefore, it has been proposed to designate the interstitial nephropathy in which the unequivocal role of AA has been fully documented as aristolochic acid nephropathy (AAN) [7,8]. Recently, a high prevalence of urothelial cancer was found in a large cohort of AAN patients in Belgium [9,10] and a case with urothelial cancer has also been described in the UK [11]. It is also noteworthy that AA consumption may be a cause for the development of a similar type of fibrosis of the kidneys with malignant transformation of the urothelium, the Balkan endemic nephropathy (BEN) [12–15], which is widely found in certain areas of Rumania, Croatia, Bosnia, Serbia and Bulgaria along the Danube river basin [12–15]. This highlights the carcinogenic potential of AA in human beings. Of note, herbal remedies containing species of the genus Aristolochia were recently classified as carcinogenic to humans by the International Agency for Research on Cancer [16]. AA is a mixture of structurally related nitrophenanthrene carboxylic acids, with 8-methoxy-6-nitro-phenanthro-(3,4-d)-1,3-dioxolo-5-carboxylic acid (AAI) and 6-nitro-phenanthro-(3,4-d)-1,3-dioxolo-5-carboxylic acid (AAII), being the major components (Fig. 1). Since the demonstration that AA forms covalent DNA adducts in rodents [17–19] as well as in AAN patients [6,20–23], AA-DNA adducts have been used

as biomarkers of exposure to AA and to investigate the mutagenic and carcinogenic potential of AA. The predominant AA-DNA adduct in vivo, 7(deoxyadenosin-N 6-yl)aristolactam I (dA-AAI), which is the most persistent of the adducts in the target tissue, is a mutagenic lesion leading to A/T transversions in vitro [24,25]. This transversion mutation is found at high frequency in codon 61 of the H-ras oncogene in tumors of rodents induced by AAI, suggesting that dA-AAI might be the critical lesion in the carcinogenic process in rodents. DNA binding studies confirmed that both AAs bind to the adenines of codon 61 in the H-ras mouse gene [24,25] and preferentially to purines in the human p53 gene [5,6,26]. In DNA isolated from the urothelial tumor of one AAN patient the dA-AAI adduct and an AAG to TAG mutation in codon 139 (Lys/Stop) of exon 5 were detected [27]. Interestingly, to date only 5% of the patients treated with the slimming regimen in Belgium have suffered from nephropathy. One of the possible causes for the different responses of patients may be individual differences in the activities of the enzymes catalyzing biotransformation (detoxication and/or activation) of AA. Many genes of enzymes metabolizing carcinogens are known to exist in variant forms or show polymorphisms resulting in differing activities of the gene products. These genetic variations appear to be important determinants of cancer risk [28]. Therefore, screening AAN patients as well as healthy persons treated with the slimming regimen, for genetic variations in the genes of the enzymes involved in AA metabolism should lead to possible relationships between genotypes and nephropathy. Thus, the identification of the enzymes principally involved in the activation of AA in humans and a detailed knowledge of their catalytic specificities is of major importance. Recently we found that in vitro, both microsomal and cytosolic enzymes activate AAI and AAII to form the same DNA adducts found in vivo in rodents [29–32] and in humans suffering from AAN [6,20–23]. Several animal reductases such as rat hepatic microsomal cytochromes P450 (CYP) 1A1 and 1A2, rabbit NADPH:CYP reductase [31,32], rat hepatic and renal NAD(P)H:quinone oxidoreductase (NQO1, DT-diaphorase) [33], butter milk xanthine oxidase [29], peroxidases such as ovine prostaglandin

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

195

COOH

O

NO2

O

aristolochic acid I (AAI): R=OCH3 aristolochic acid II (AAII): R=H proximale carcinogen

R

O O N

OH

O

ultimate carcinogen O

O O

O N

R

O

NH O

+

N-hydroxyaristolactam

R

Bamberger Umlagerung

R

aristolactam I, II aristolactam-nitriumion DNA

O O

O

O

NH

R N

O

N

OH

HO

7-hydroxyaristolactam I, II

N

R

O N H

O

R

O

NH

O

N NH

NH

H

H

OH

H

H

N

C H

NH

C O

O N

dG-AAI, dG-AAII

N

HO

dA-AAI, dA-AAII

O H

H

OH

H

H

H

Fig. 1. Metabolic activation and DNA adduct formation of AA. (Umlagerung Z rearrangement).

H synthase (cyclooxygenase, COX) [30] and milk lactoperoxidase [29] were found to be able to activate AAI to form DNA adducts. Beside these animal enzymes, human CYP1A1/2 and NQO1 were identified as the enzymes activating AAI in human liver, while NADPH:CYP reductase, NQO1 and COX are active in human kidney [34,35]. Interestingly, whereas CYP enzymes are known to metabolize xenobiotics mainly by oxidation (hydroxylation) reactions, we clearly demonstrated the efficiencies of rat and human CYP1A1 and 1A2 to generate the AA-DNA adducts which are formed if AAI is reduced [7-(deoxyadenosin-N6-yl)aristolactam I (dA-AAI), 7-(deoxyguanosin-N2-yl) aristolactam

I dG-AAI), and 7-(deoxyadenosin-N6-yl)aristolactam II (dA-AAII)] [31,35]. The present study was undertaken to investigate the molecular basis of such an AAI reduction by CYP1A enzymes.

2. Materials and methods 2.1. Chemicals and enzymes Chemicals were obtained from the following sources: NADPH and nuclease P1, from Sigma Chemical Co. (St. Louis, MO), and calf thymus DNA (CT-DNA) from Roche Diagnostics Mannheim

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

196

(Germany). Enzymes and chemicals for the 32P-postlabeling assay were obtained commercially from sources described previously [19–22]. The natural mixture consisting of 65% AAI and 34% AAII was a gift from Madaus (Cologne, Germany). AAI was isolated from the mixture by preparative HPLC; its purity was 99.7% as estimated by HPLC [36]. All other chemicals were of analytical purity or better. Supersomese, microsomes isolated from insect cells transfected with baculovirus constructs containing the cDNA of CYP1A1 or 1A2 and expressing NADPH:CYP reductase were obtained from Gentest Corp. Human hepatic microsomes were of the source described in our previous work [35]. 2.2. Incubations of AAI with Supersomese The de-aerated and argon-purged incubation mixtures, in a final volume of 750 ml, consisted of 50 mM potassium phosphate buffer (pH 7.4), 1 mM NADPH, microsomes containing human recombinant CYPs and NADPH:CYP reductase (Supersomese) (10–50 pmol of CYPs) or human hepatic microsomes (sample H8 used previously [35]) (100 pmol of CYP), 0.01–0.5 mM AAI and 0.5 mg of CT-DNA (4 mM). The reaction was initiated by adding AAI. Incubations were carried out at 37 8C for 20–90 min. Control incubations were carried out either (i) without activating system (Supersomese, microsomes) or (ii) with activating system and AAI, but without DNA or (iii) with activating system and DNA but without AAI. Supersomese containing human recombinant NADPH:CYP reductase alone were used for comparison. After the incubation, DNA was isolated from the residual water phase after ethyl acetate extraction by the phenol/chloroform extraction method as described [32]. 2.3.

32

P-Postlabeling analysis

The nuclease P1 enrichment version [37] of the assay, known to be the most suitable version of the 32 P-postlabeling technique for AA-DNA adduct detection and quantitation was used and performed exactly as described [6,17,20–23,29–35]. Enzymatic synthesis of reference compounds, dAp-AAI, dGpAAI and dAp-AAII and their 32P-postlabeling were carried out as described earlier [19].

2.4. CYP1A1 and 1A2 homology modeling and AAI docking The 3D structures of human CYP1A1 and 1A2 were built based on the crystal structure of the following CYPs (PDB code): CYP2C5 (1n6b) [38], CYP2C8 (1pq2) [39], CYP2C9 (1r9o) [40], CYP2B4 (1suo) [41] and CYP3A4 (1w0 g) [42] found by NHINCBI Blast Iteractive Service using two iteration runs [43]. The sequences of CYP1A1 and 1A2 were obtained from SwissProt database. The sequence aligment was done in two steps by Modeller 6.2. [44,45] and Clustal X program [46]. In the first step, the known structures were spatially aligned by Modeller 6.2. (using Malign 3D function). Finally the CYP1A1 and 1A2 were aligned from residue 45 to the created alignment among all CYP enzymes used for the modeling by Clustal X (command ‘align sequence to the profile’). Based on this alignment the model structures were generated by the program Modeller 6.2. The resulting structures were corrected manually (e.g. amino acid clashes) using the SwissProt Viewer 3.7 [47]. In each of the built structures, the heme group was docked using AutoDock 3.05 sofware [48] and all of them were subjected to the energy minimization by CHARMM 29 program [49]. AutoDock parameters ˚ , the heme were set as follows: grid spacing 0.15 A positioned in the grid center, and used maximum (120) of allowed points. The heme iron charge as well as the length of the Cys thiolate-Fe bond were computed using Sybyl program and adjusted based on CYP crystal base structure. Solvation parameters for the heme were default values in AutoDock AddSol program module. Parameters of algorithm were chosen to be 270,000,000 and 250 and 0.2 for the number of generations, population size and mutation rate, respectively. After 100 steps of the steepest descent, the resulting model structures were checked using the PROCHECK sofware [50]. The structures with lowest energy and the best PROCHECK score were used for further energy minimization by CHARMM 29 (100 steps of the conjugated gradient followed by 100 steps of the steepest descent). Due to the quite large number of resulting structures produced by the Modeller 6.2., no further molecular dynamics computing was performed [44,45]. The CYP1A1 and 1A2 models

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

were finally analyzed by PROCHECK. AAI structure was built with fragment libraries supplied with the modeling software. The initial structure was first energy minimized to a root-mean-square force of less than 0.001 with the consistent valence force field [51]. The energy-minimized AAI molecule structure was then docked into the CYP1A1 and 1A2 models. The docking of AAI to the CYP1A1 and 1A2 active sites was performed with the program Sybyl 6.6.5 and Autodock 3.0.3 (Tripos GmbH, Germany) [51], using the genetic algorithm method, with 27,000,000 generations and 200 populations, with 20 runs for AAI [51]. The conformation with the lowest energy was chosen as the result.

CYP1A1, 1A2 and NADPH:CYP reductase or only NADPH:CYP reductase alone were used in incubations with AAI and DNA. Supersomese containing both NADPH:CYP reductase and CYP1A1 or 1A2 were capable of reducing AAI; AAI-derived DNA adducts (assayed by 32P-postlabeling) were generated in these enzymatic systems (Fig. 2). As described in our previous works [31,35], the adducts shown in Fig. 2 as spots 1,2 and 3 are generated also in human hepatic microsomes and represent the adducts 7-(deoxyadenosin-N6-yl)aristolactam I, 7-(deoxyguanosin-N2-yl)aristolactam I and 7-(deoxyadenosin-N6yl)aristolactam II, respectively. The CYP1A1/2- and NADPH:CYP reductase-catalyzed DNA binding of AAI was shown to be dependent on the time of incubation, being linear up to 70 min (not shown), and on the concentrations of the enzymes (Fig. 3). A 7- and 10-fold increase in AAI-DNA adduct levels was found when Supersomese contained human CYP1A1 and 1A2 besides the NADPH:CYP reductase, respectively (Fig. 3). AAI-DNA adduct 25

AAI-DNA adducts [RAL per 107nucleotides]

Fig. 2. Autoradiographic profiles of AAI-DNA adducts obtained from calf thymus DNA after activation by Supersomese containing human recombinant CYP1A2 in the presence of NADPH. The nuclease P1-enrichment procedure was used for analysis. Origins, in the bottom left-hand corner were cut off before exposure. Screen enhanced autoradiography was at K80 8C for 1 h. Chromatographic conditions: D1, 1 M sodium phosphate, pH 6.8; D3, 3.5 M lithium formate, 8.5 M urea, pH 4.0; D4, 0.8 M LiCl, 0.5 M Tris–HCl, 8.5 M urea, pH 9.1; D5, 1.7 M NaH2PO4, pH 6.0. Spot 1, dG-AAI; spot 2, dA-AAI; spot 3, dA-AAII.

197

20

15

10

5

3. Results 0

3.1. Reductive activation of AAI by human CYP1A1, 1A2 and NADPH:CYP reductase To investigate the kinetics of the reductive activation of AAI to species binding to DNA, microsomes from baculovirus-transfected insect cells (Supersomese) containing recombinant human

0

10

20 30 40 Enzyme [pmol]

50

60

Fig. 3. DNA binding of AAI after activation with Supersomese containing different concentrations of human recombinant CYP1A2 (:), CYP1A1 (&) and NADPH:CYP reductase or NADPH:CYP reductase alone (C). The nuclease P1-enrichment procedure was used for analysis. Values represent mean S.E.M. (nZ4) of two separate incubations each determined by two post-labeled analyses.

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

198

AAI-DNA adducts [RAL per 107 nucleotides]

25

20

15

10

5

0 0.0

0.1

0.2

0.3 AAI [mM]

0.4

0.5

0.6

Fig. 4. AAI concentration dependence of DNA-adduct formation by human recombinant CYP1A2 (:), CYP1A1 (&) and NADPH: CYP reductase or NADPH:CYP reductase alone (C). The nuclease P1-enrichment procedure was used for analysis. Numbers are averages GS.E.M. (nZ4) of two separate incubations each determined by two post-labeled analyses.

formation was also dependent on AAI concentrations using both Supersomese (Fig. 4) and human hepatic microsomes (not shown). The concentrations required for half-maximum DNA binding were 65, 38, 126 and 40 mM AAI for its activation by human CYP1A1, CYP1A2, NADPH:CYP reductase and human hepatic microsomes, respectively. 3.2. Molecular modeling of AAI binding to CYP1A1 and 1A2 active sites To investigate the molecular basis of the reductive activation of AAI by human CYP enzymes, interaction of AAI with the active centers of both enzymes was examined. Since the crystal structures of human CYP1A1 and 1A2 have not yet been resolved, we constructed molecular models of CYP1A1 and 1A2. Homology modeling of 3D CYP structures based on crystal structures of microbial CYPs has been shown to be an important tool to study structure-function relationship on mammalian CYPs. In recent years, the crystal structure of the first solubilized mammalian CYP, CYP2C5, proved a more reliable template to model mammalian CYPs [52,53]. In the meantime the crystal structures of additional mammalian CYPs,

CYP2B4, 2C8, 2C9 and 3A4, were resolved. To construct the most appropriate CYP1A1 and 1A2 models, a combination of data from all these structurally characterized mammalian CYP enzymes was employed. Multiple sequence alignment used for the CYP1A1 and 1A2 model construction is shown in Fig. 5. To achieve correct heme docking, heme and Cys sulfur charges were computed first using Sybyl and compared with CYP crystal base structures (PDB), and then solvation parameters were added by AddSol software. The heme placement found was the final best position followed from the assessment of the binding distance of Cys sulfur and heme iron, their relative orientation and symmetry, as well as from the results of a further energy minimization, showing the final energy minimum. Moreover, the G-factor of the amino acids surrounding the heme in our model structures agreed well with that of known CYP structures. The resulting CYP1A1 and 1A2 models were finally analyzed by PROCHECK. As it is clear from Fig. 6, there were no significant differences in structure quality parameters between the models and the starting structures used for modeling. The final overall model energies were as low as K17012.4 and K17304.4 kJ/mol for CYP1A1 and 1A2, respectively. By docking AAI into the model structure of CYP1A1 and 1A2 optimized by molecular-dynamics we found the orientation of AAI binding to both enzymes characterized by the highest ranking computed by AutoDock runs. The calculated model structures for the human CYP1A1-AAI- and CYP1A2-AAI-complexes are shown in Fig. 7. The AAI molecule fits well into the active sites of CYP1A1 and 1A2, being bound approximately perpendicularly to the heme, as an axial ligand of iron of the enzymes. This orientation allows an electron transfer during the AAI reduction by both enzymes. The values of the theoretical apparent dissociation constants for the CYP1A1(AAI- and CYP1A2-AAI-complexes were calculated to be 0.024 and 4.670 mM, respectively. The position of the AAI molecule in the active centre of CYP1A2 indicates a single interaction of only the nitro group of AAI with heme iron (Fig. 7). The distance between one oxygen atom of the nitro ˚ makes its group and heme iron as low as 2.3 A

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

199

Fig. 5. Multiple sequence alignment among crystallized CYP templates (PDB code): CYP2C5 (1n6b) [38], CYP2C8 (1pq2) [39], CYP2C9 (1r9o) [40], CYP2B4 (1suo) [41] and CYP3A4 (1w0 g) [42] and primary structures of human CYP1A1 (A) and CYP1A2 (B).

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

200

Ramachandran plot quality assessment

(a)

Standard deviation (kcal/mol)

%-tage of residues in most favoured regions

2.0

80 60 40 20

1

2

3

4

5

6

1.2 1.0 0.8 0.5 0.2 1

Chi-1 pooled standard deviation

(d)

2

3

4

5

6

7

Standard deviation of Chi-2 trans angle 50

Standard deviation (degrees)

Standard deviation (degrees)

1.5

7

50 40 30 20 10 0

40 30 20 10 0

1 (e)

2

3

4

5

6

7

1

2

3

4

5

6

7

Dihedral angles G-factor 1.0

CYP structure

0.5

G-factor

1.8

0.0

0

(c)

Hydrogen bond energies

(b)

100

1

1n6b

0.0

2

1pq2

-0.5

3

1r90

-1.0

4

1suo

-1.5

5

1w0g

-2.0

6

1A1

7

1A2

-2.5 1

2

3

4

5

6

7

Fig. 6. Analysis of structural parameters of CYP1A1 (6) and CYP1A2 (7) models in comparison with known structures of CYP2C5 (1), CYP2C8 (2), CYP2C9 (3), CYP2B4 (4) and CYP3A4 (5).

reduction feasible. In the case of CYP1A1, interactions of the oxygen atoms of both the nitro- and carboxylic groups of AAI are observable (Fig. 7). Therefore, even though the distance between an

oxygen of the AAI nitro group and heme iron in ˚ ), CYP1A1 is very similar to that in CYP1A2 (2.27 A an electron transfer to the nitro group is hampered by the close vicinity of oxygen of the carboxylic group of

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

Fig. 7. AAI is shown docked to the active sites of human CYP1A1 (A) and CYP1A2 (B) indicating the position of the AAI molecule to a heme prosthetic group.

˚ ). This corresponds to AAI (a distance of 2.22 A experimental data. CYP1A1 is less efficient in reductive activation of AAI than CYP1A2 (Figs. 3 and 4).

4. Discussion By correlation analysis, we found that AAI is bioactivated by CYP1A1 and 1A2 enzymes (predominantly CYP1A2) in human liver microsomes to dG and dA adducts identical to those found in humans

201

exposed to AA [35]. Based on the structures of the AAI-adducts identified, we suggested that nitroreduction of AAI to the corresponding aristolactam I is the main activating pathway in animals [6,16–19] and humans [6,10,11,20–22]. An intermediate cyclic nitrenium ion with a delocalized positive charge was postulated by us [18,19] as the ultimate electrophilic species leading to binding via C7 of the phenanthrene ring to the exocyclic amino group of dG and dA, as shown in Fig. 1. It is noteworthy that CYP1A1/2 are more efficient enzymes reducing AAI than NADPH:CYP reductase, both in human hepatic microsomes [31,35] and in Supersomese. This finding indicates a higher affinity of AAI to the active centre of CYP1A than to that of NADPH:CYP reductase. The explanation of this feature remains, however, to be resolved. The reduction of the nitro-aromatic, AAI, by human CYP1A enzymes is consistent with results of Chae et al. [54]. These authors showed that human recombinant CYP1A2 reconstituted with NADPH:CYP reductase was able to reduce 4-nitropyrene and to a lower extent also 1- and 2-nitropyrene. Moreover, we recently showed that human recombinant CYP1A1 and 1A2 reductively activate another aromatic nitro-compound, 3-nitrobenzanthrone [55]. In the present work, we studied the reduction of AAI by human CYP1A1/2 on a molecular level to explain the nature of this reaction catalyzed by both enzymes. The interaction of AAI with human CYP1A1 and 1A2 enzymes can be studied by two methods; (i) spectroscopy, based on the analysis of difference substrate-binding spectra of AAI with both CYPs and (ii) molecular modeling. The analysis of the binding spectra of CYP1A1 and 1A2 has some limitations, because the two enzymes occur in different spin forms; the high spin form is typical for CYP1A1, while CYP1A2 is found mainly in the low spin form [56,57]. Hence, both enzymes behave differently in the resting states, which can, in addition, influence the substrate binding spectra of the enzymes and subsequently the interpretation of the binding of substrates to the heme iron in the porphyrin, as a consequence binding is not very well visible in the spectra. This was the case in preliminary experiments, in which the binding spectra of human CYP1A1 and 1A2 with AAI were determined. The binding titration of AAI gave ligand type II with CYP1A1 and reverse type I spectra with CYP1A2 [58]. Therefore,

202

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

molecular modeling was employed to evaluate the nature of AAI binding to the active centers of CYP1A1 and 1A2, and to explain the observed reduction of the substrate, AAI. Homology models of human CYP1A1 and 1A2 were constructed on the basis of the crystallographic structure of truncated mammalian CYP enzymes, CYP2B4, 2C5, 2C8, 2C9 and 3A4. The docking of AAI to the active sites of CYP1A1 and 1A2 indicates that AAI binds as a ligand instead of oxygen, interacting with heme iron. Even though the AAI molecule is bound more tightly to CYP1A1 than to CYP1A2 (the value of the theoretical apparent dissociation constant for the human CYP1A1-AAI-complex is two orders of magnitude lower than that for the CYP1A2-AAI-complex), AAI is more efficiently reduced by CYP1A2. The orientation of the AAI molecule in the active sites of CYP1A1 and 1A2 explains this observation. In silico docking revealed binding of the nitro group of AAI to heme iron of CYP1A2 in an orientation allowing the efficient reduction of this group observed experimentally. The binding orientation of AAI in the active centre of CYP1A1 allows the interaction of the heme iron not only with the nitro group of AAI but also with the carboxylic group of this carcinogen. This orientation might explain the lower capacity of CYP1A1 to reduce AAI than CYP1A2 as found experimentally.

Acknowledgements This work was supported by Grant Agency of Charles University (Grant 432/2004/04/B-CH/PrF). Dedicated to Prof. Dr. Sylva Leblova´.

References [1] J.L. Vanherweghem, M. Depierreux, C. Tielemans, D. Abramowicz, M. Dratwa, M. Jadoul, et al., Rapidly progressive interstitial renal fibrosis in young women: association with slimming regimen including Chinese herbs, Lancet 341 (1993) 387–391. [2] M. Vanhaelen, R. Vanhaelen-Fastre, P. But, J.L. Vanherweghem, Identification of aristolochic acid in Chinese herbs, Lancet 343 (1993) 174.

[3] M. Vanhaelen, J.J. Cuykens, Ph. Vandenberg, K.P. Bouman, Y. Hagens, Valvular heart disease and Chinese-herbs nephropathy, Lancet 351 (1998) 991. [4] M.-C. Martinez, J. Nortier, P. Vereestaeten, J.L. Vanherweghem, Progression rate of Chinese herb nephropathy: impact of Aristolochia fangchi ingested dose, Nephrol. Dial. Transpl. 17 (2002) 408–412. [5] J.-P. Cosyns, Aristolochic acid and ‘Chinese Herbs nephtopathy’. A review if the evidence to date, Drug Saf. 26 (2003) 33–48. [6] V.M. Arlt, M. Stiborova´, H.H. Schmeiser, Aristolochic acid as a probable cancer hazard in herbal remedie: a review, Mutagenesis 17 (2002) 265–277. [7] G. Gillerot, M. Jadoul, V.M. Arlt, C. van Ypersele de Strihou, H.H. Schmeiser, P.H.H. But, et al., Aristolochic acid nephropathy in a Chinese patient: time to abandon the term ‘Chinese herbs nephropathy’?, Am. J. Kidney Dis. 38 (2001) E26. [8] K. Solez, J. Daugirdes, M.C. Gregory, P.P. Frohnert, D.M. Blownik, V. Iha, et al., Is ‘Chinese herbs nephropathy’ a prejudical term?, Am. J. Kidney Dis. 38 (2001) 1141–1142. [9] J.-P. Cosyns, M. Jadoul, J.P. Squifflet, F.X. Wese, C. van Ypersele de Strihou, Urothelial lesions in Chinese-herb nephropathy, Am. J. Kidney Dis. 33 (1999) 1011–1017. [10] J.L. Nortier, M. Muniz Martinez, H.H. Schmeiser, V.M. Arlt, C.A. Bieler, M. Petein, et al., Urothelial carcinoma associated with the use of a Chinese herb ({i}Aristolochia fangchi{/i}), N. Engl. J. Med. 342 (2000) 1686–1692. [11] G.M. Lord, T. Cook, V.M. Arlt, H.H. Schmeiser, G. Williams, C.D. Pusey, Urothelial malignancy and Chinese herbal nephropathy, Lancet 358 (2001) 1515–1516. [12] M. Ivic, The problem of etiology of endemic nephropathy, Acta. Fac. Med. Naiss. 1 (1970) 29–37. [13] C.A. Tatu, W.H. Oren, R.B. Finkelman, G.L. Feder, The etiology of Balkan endemic nephropathy: still more questions than answers, Environ. Health Perspect. 106 (1998) 689–700. [14] V.M. Arlt, D. Ferluga, M. Stiborova´, A. Pfohl-Leskowicz, M. Vukelic, S. Ceovic, et al., Is aristolochic acid a risk for Balkan endemic nephropathy-associated urothelial cancer?, Int. J. Cancer 101 (2002) 500–502. [15] A. Pfohl-Leskowicz, T. Petkova-Bocharova, I.N. Chernozemsky, M. Castegnaro, Balkan endemic nephropathy and associated urinary tract tumours: a review on aetiological causes and the potential role of mycotoxins, Food Addit. Contam. 19 (2002) 282–302. [16] IARC, Some Traditional Herbal Medicines, Some Mycotoxins, Naphthalene and Styrene, vol. 82, International Agency for Research on Cancer, Lyon, 2002. [17] H.H. Schmeiser, K.-B. Schoepe, M. Wiessler, D.N.A. adduct, formation of aristolochic acid I and II in vitro and in vivo, Carcinogenesis 9 (1988) 297–303. [18] W. Pfau, H.H. Schmeiser, M. Wiessler, 32P-postlabelling analysis of the DNA adducts formed by aristolochic acid I and II, Carcinogenesis 11 (1990) 1627–1633. [19] M. Stiborova´, R.C. Fernando, H.H. Schmeiser, E. Frei, W. Pfau, M. Wiessler, Characterization of DNA adducts

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

formed by aristolochic acids in the target organ (forestomach) of rats by 32P-postlabeling analysis using different chromatographic procedures, Carcinogenesis 15 (1994) 1187–1192. H.H. Schmeiser, C.A. Bieler, M. Wiessler, C. van Ypersele de Strihou, J.-P. Cosyns, Detection of DNA adducts formed by aristolochic acid in renal tissue from patients with Chinese herbs nephropathy, Cancer Res. 56 (1996) 2025–2028. C.A. Bieler, M. Stiborova´, M. Wiessler, J-P. Cosyns, C. van Ypersele de Strihou, H.H Schmeiser, 32P-post-labeling analysis of DNA adducts formed by aristolochic acid in tissues from patients with Chinese herbs nephropathy, Carcinogenesis 18 (1997) 1063–1067. M. Stiborova´, E. Frei, A. Breuer, C.A. Bieler, H.H. Schmeiser, Aristolactam I a metabolite of aristolochic acid I upon activation forms an adduct found in DNA of patients with Chinese herbs nephropathy, Exp. Toxic. Pathol. 51 (1999) 421–427. V.M. Arlt, A. Pfohl-Leszkowicz, J-P. Cosyns, H.H. Schmeiser, Analyses of DNA adducts formed by ochratoxin A and aristolochic acid in patients with Chinese herbs nephropathy, Mutat. Res. 494 (2001) 143–150. H.H. Schmeiser, J.W.G. Janssen, J. Lyons, H.R. Scherf, W. Pfau, A. Buchmann, et al., Aristolochic acid activates ras genes in rat tumors at deoxyadenosine residues, Cancer Res. 50 (1990) 5464–5469. V.M. Arlt, M. Wiessler, H.H. Schmeiser, Using polymerase arrest to detect DNA binding specificity of aristolochic acid in the mouse H-ras gene, Carcinogenesis 21 (2000) 235–242. V.M. Arlt, H.H. Schmeiser, G.P. Pfeifer, Sequence-specific detection of aristolochic acid-DNA adducts in the human p53 gene by terminal transferase-dependent PCR, Carcinogenesis 22 (2001) 133–140. G.M. Lord, M. Hollstein, V.M. Arlt, C. Roufosse, C.D. Pusey, T. Cook, H.H. Schmeiser, DNA adducts and p53 mutations in a patient with aristolochic acid-associated nephropathy, Am. J. Kidney Dis. 43 (2004) Art.No.e11. G. Smith, L.A. Stanley, E. Sim, R.C. Strange, R. Wolf, Metabolic polymorphism and cancer susceptibility, Cancer Surv. 25 (1995) 27–65. H.H. Schmeiser, E. Frei, M. Wiessler, M. Stiborova´, Comparison of DNA adduct formation by aristolochic acids in various in vitro activation systems by 32P-post-labeling: evidence for reductive activation by peroxidases, Carcinogenesis 18 (1997) 1055–1062. M. Stiborova´, E. Frei, A. Breuer, M. Wiessler, H.H. Schmeiser, Evidence for reductive activation of carcinogenic aristolochic acids by prostaglandin H synthase - 32P-postlabeling analysis of DNA adduct formation, Mutat. Res. 493 (2001) 153–164. M. Stiborova´, E. Frei, M. Wiessler, H.H. Schmeiser, Human enzymes involved in the metabolic activation of carcinogenic aristolochic acids: evidence for reductive activation by cytochrome P450 1A1 and 1A2, Chem. Res. Toxicol. 14 (2001) 1128–1137. M. Stiborova´, E. Frei, M. Ha´jek, H.H. Schmeiser, Carcinogenic and nephrotoxic aristolochic acids upon activation by

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44]

203

NADPH: cytochrome P450 reductase form adducts found in DNA of patients with Chinese herbs nephropathy, Gen. Physiol. Biophys. 20 (2001) 375–392. M. Stiborova´, E. Frei, B. Sopko, M. Wiessler, H.H. Schmeiser, Carcinogenic aristolochic acids upon activation by DTdiaphorase form adducts found in DNA of patients with Chinese herbs nephropathy, Carcinogenesis 23 (2002) 617– 625. M. Stiborova´, E. Frei, B. Sopko, K. Sopkova´, V. Markova´, M. Lanˇkova´, et al., Human cytosolic enzymes involved in the metabolic activation of carcinogenic aristolochic acid: evidence for reductive activation by human NAD(P)H: quinone oxidoreductase, Carcinogenesis 24 (2003) 1695– 1703. M. Stiborova´, E. Frei, P. Hodek, M. Wiessler, H.H. Schmeiser, Human hepatic and renal microsomes, cytochromes P450 1A1/2, NADPH: CYP reductase and prostaglandin H synthase mediate the formation of aristolochic acid DNA-adducts found in patients with urothelial cancer, Int. J. Cancer 113 (2005) 189–197. H.H. Schmeiser, B.L. Pool, M. Wiessler, Mutagenicity of the two main components of commercially available carcinogenic aristolochic acids in Salmonella typhimurium, Cancer Lett. 23 (1984) 107–111. V. Reddy, K. Randerath, Nuclease P1-mediated enhancement of sensitivity of 32P-postlabeling test for structurally diverse DNA adducts, Carcinogenesis 7 (1986) 1543–1551. M.R. Wester, E.F. Johnson, C. Marques-Soares, S. Dijols, P.M. Dansette, D. Mansuy, et al., Structure of mammalian cytochrome P4502C5 complexed with diclofenac at 2.1 angstrom resolution: evidence for an induced fit model of substrate binding, Biochemistry 42 (2003) 9335–9345. G.A. Schoch, J.K. Yano, M.R. Wester, K.J. Griffin, C.D. Stout, E.F. Johnson, Structure of human microsomal cytochrome P4502C8 - evidence for a peripheral fatty acid binding site, J. Biol. Chem. 279 (2004) 9497–9503. M.R. Wester, J.K. Yano, G.A. Schoch, C. Yang, K.J. Griffin, C.D. Stout, et al., The structure of human cytochrome P4502C9 complexed with flurbiprofen at 2.0-angstrom resolution, J. Biol. Chem. 279 (2004) 35630–35637. E.E. Scott, M.A. White, Y.A. He, E.F. Johnson, C.D. Stout, J.R. Halpert, Structure of mammalian cytochrome P4502B4 complexed with 4-(4-chlorophenyl) imidazole at 1.9-angstrom resolution - insight into the range of P450 conformations and the coordination of redox partner binding, J. Biol. Chem. 279 (2004) 27294–27301. P.A. Williams, J. Cosme, D.M. Vinkovic, A. Ward, H.C. Angove, P.J. Day, Crystal structures of human cytochrome P450 3A4 bound to metyrapone and progesterone, Science 305 (2004) 683–686. S.F. Altschul, W. Gish, W. Miller, E.W. Myers, D.J. Lipman, Basic local alignment search tool, J. Mol. Biol. 215 (1990) 403–410. A. Sali, T.L. Blundell, Comparative protein modelling by satisfaction of spatial restrains, J. Mol. Biol. 234 (1999) 779– 815.

204

M. Stiborova´ et al. / Cancer Letters 229 (2005) 193–204

[45] A. Fiser, R.K. Do, A. Sali, Modeling of loops in protein structures, Protein Sci. 9 (2000) 1753–1773. [46] J.D. Thompson, T.J. Gibson, F. Plewniak, F. Jeanmougin, D.G. Hiffins, The CLUSTAL X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools, Nucleic Acids Res. 25 (1997) 4876–4882. [47] N. Guex, M.C. Peitsch, SWISS-MODEL and the SwissPdbViewer: an environment for comparative protein modeling, Electrophoresis 18 (1997) 2714–2723. [48] G.M. Morris, D.S. Goodsell, R.S. Halliday, R. Huey, W.E. Hart, R.K. Belew, Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function, J. Comp. Chem. 19 (1998) 1639–1662. [49] B.R. Brooks, R.E. Bruccoleri, B.D. Olafson, D.J. States, S. Swaminathan, M. Karplus, CHARMM: a program for macromolecular energy, minimization, and dynamics calculations, J. Comput. Chem. 4 (1983) 187–217. [50] R.A. Laskowski, M.W. MacArthur, D.S. Moss, J.M. Thornton, PROCHECK - a program to check the stereochemical quality of protein structure, J. Appl. Cryst. 26 (1993) 183–191. [51] P. Dauber-Osguthorpe, V.A. Roberts, D.J. Osguthorpe, J. Wolff, M. Genest, A.T. Hagier, Structure and energetics of ligand binding to protein: Escherichia coli dihydrofolate reductase-trimethoprim, a drug receptor system, Proteins 4 (1988) 31–47.

[52] A. Williams, J. Cosme, V. Sridhar, E.F. Johson, D.E. McRee, Mammalian microsomal cytochrome P450 monooxygenase: structural adaptations for membrane binding and functional diversity, Mol. Cell 5 (2000) 121–131. [53] D.F. Lewis, Homology modelling of human CYP2 family enzymes based on the CYP2C5 crystal structure, Xenobiotica 32 (2002) 305–323. [54] Y.-H. Chae, T. Thomas, F.P. Guengerich, P.P. Fu, K. ElBayoumy, Comparative metabolism of 1-, 2-, and 4-nitropyrene by human hepatic and pulmonary microsomes, Cancer Res. 59 (1999) 1473–1480. [55] V.M. Arlt, M. Stiborova, A. Hewer, H.H. Schmeiser, D.H. Phillips, Human enzymes involved in the metabolic activation of the environmental contaminant 3-nitrobenzanthrone: evidence for reductive activation by human NADPH: cytochrome P450 reductase, Cancer Res. 63 (2003) 2752–2761. [56] F.P. Guengerich, Oxidation of toxic and carcinogenic chemicals by human cytochrome P450 enzymes, Chem. Res. Toxicol. 4 (1991) 391–407. [57] S. Rendic, F.J. DiCarlo, Human cytochrome P450 enzymes: a status report summarizing their reactions, substrates, inducers, and inhibitors, Drug Metab. Rev. 29 (1997) 413–480. [58] M. Stiborova´, E. Frei, H.H. Schmeiser, Identification of human enzymatic systems activating carcinogenic alkaloid aristolochic acid in vitro, Chem. Listy 93 (1999) 723–724.