Theoretical studies of the mechanism of N-hydroxylation of primary aromatic amines by cytochrome P450 1A2: radicaloid or anionic?

Subscriber access provided by DUESSELDORF LIBRARIES

Article

Theoretical studies of the mechanism of N-hydroxylation of primary aromatic amines by cytochrome P450 1A2: Radicaloid or anionic? Lena Ripa, Christine Mee, Peter Olof Sjö, and Igor Shamovsky Chem. Res. Toxicol., Just Accepted Manuscript • DOI: 10.1021/tx400376u • Publication Date (Web): 13 Jan 2014 Downloaded from http://pubs.acs.org on January 18, 2014

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Chemical Research in Toxicology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Chemical Research in Toxicology

Theoretical studies of the mechanism of N-hydroxylation of primary aromatic amines by cytochrome P450 1A2: Radicaloid or anionic? Lena Ripa,† Christine Mee,‡,§ Peter Sjö,† and Igor Shamovsky*,† †

Department of Medicinal Chemistry, RIA iMed, AstraZeneca R&D, Pepparedsleden 1, S-431 83 Mölndal, Sweden; Genetic Toxicology, AstraZeneca R&D, Alderley Park, Macclesfield, Cheshire, SK10 4TG, UK; § Current affiliation: Gentronix Ltd., BioHub at Alderley Park, Alderley Edge, Cheshire, SK10 4TG, UK ‡

Table of Contents Graphic ∆G, kcal/mol 15

ArNH-

10 5

ArNH· ArNH2

0 -5

FeIVO

FeIIOO2-

-10

Radicaloid mechanism

Anionic mechanism

ABSTRACT: Primary aromatic and heteroaromatic amines are notoriously known as potential mutagens and carcinogens. The major event of the mechanism of their mutagenicity is N-hydroxylation by P450 enzymes, primarily P450 1A2 (CYP1A2), which leads to formation of nitrenium ions that covalently modify nucleobases of DNA. Energy profiles of the NH bond activation steps of two possible mechanisms of N-hydroxylation of a number of aromatic amines by CYP1A2, radicaloid and anionic, are studied by dispersion-corrected DFT calculations. The classical radicaloid mechanism is mediated by H-atom transfer to the electrophilic ferryl-oxo intermediate of the P450 catalytic cycle (called Compound I or Cpd I), whereas the alternative anionic mechanism involves proton transfer to the preceding nucleophilic ferrous-peroxo species. The key structural features of the catalytic site of human CYP1A2 revealed by X-ray crystallography are maintained in calculations. The obtained DFT reaction profiles and additional calculations that account for nondynamical electron correlation suggest that Cpd I has higher thermodynamic drive to activate aromatic amines than the ferrous-peroxo species. Nevertheless, the anionic mechanism is demonstrated to be consistent with a variety of experimental observations. Thus, energy of the proton transfer from aromatic amines to the ferrous-peroxo dianion splits aromatic amines into two classes with different mutagenicity mechanisms. Favorable or slightly unfavorable barrier-free proton transfer is inherent in compounds that undergo nitrenium ion mediated mutagenicity. Monocyclic electron-rich aromatic amines that do not follow this mutagenicity mechanism show significantly unfavorable proton transfer. Feasibility of the entire anionic mechanism is demonstrated by favorable Gibbs energy profiles of both chemical steps, NH bond activation and NO bond formation. This suggests that the N-hydroxylation of aromatic amines in CYP1A2 undergoes the anionic mechanism. Possible reasons for the apparent inability of Cpd I to activate aromatic amines in CYP1A2 are discussed.

1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

INTRODUCTION Comprehensive genotoxicity testing of new compounds is an important part of the drug development process and is a regulatory requirement before the approval of new drugs.1,2 Primary aromatic and heteroaromatic amines (ArNH2), a class of compounds notoriously known as potential mutagens and carcinogens,3 are common motifs in compounds of pharmaceutical interest.4 To accelerate drug discovery projects it is important to identify potential genotoxic fragments that could be released by the cleavage of covalent bonds in drugs. According to the AstraZeneca corporate database, only 30% of ArNH2 are mutagenic in Ames bacterial mutagenicity assays in strains TA98 and TA100, with or without an enzymatic activation system derived from rat liver microsomes (S9). However, rational design of non-mutagenic ArNH2 is often a challenging task due to poor understanding of the structuremutagenicity relationship (SMR),4-6 most likely because the mutagenic effect of ArNH2 is the result of a sequence of chemical reactions.6,7 A better understanding of the SMR of ArNH2 would greatly facilitate the design of new drug candidates. The mechanism of mutagenicity of ArNH2 in most cases starts with metabolic activation to hydroxylamines (ArNHOH) mainly by monooxygenases of the cytochrome P450 family, primarily by P450 1A2 (CYP1A2) and further to their bioesters (ArNHOR) by a number of Phase II enzymes.7 Hydrolytic dissociation of these forms under slightly acidic conditions leads to nitrenium ions (ArNH+) able to form covalent adducts with nucleobases of DNA, primarily guanines (dG), which may lead to mutations.7-9 Chemical reactivity of ArNHOH and ArNHOR toward dG, which is determined by the rates of formation and quenching of ArNH+ under aqueous conditions, has been shown to represent an important component of the mutagenic potency of ArNH2 and to depend on the stability of ArNH+.10 This result is in line with prior theoretical studies of chemical reactivity of epoxides with dG11-13 that further highlights the fundamental link between reactivity of the ultimate forms of carcinogens with DNA and their mutagenic potency. The nitrenium stabilization concept is found to be very useful in discriminating mutagenic and non-mutagenic ArNH2 in large datasets since the stability of ArNH+ represents the most important descriptor of mutagenicity of ArNH2.4,5 However, stability of ArNH+ clearly correlates with mutagenic potency of ArNH2 only within structural subclasses.6,10,14-17 There are distinct motifs in the SMR that cannot be linked to the stability of ArNH+. For example, published results suggest protective effects of bulky groups regardless of their impact on the stability of ArNH+,10,18,19 and inductive functional groups show reverse effects on the stability of ArNH+ and the mutagenic potency of ArNH2.6,10,19-24 Thus, electronwithdrawing functions like CF3, CHF2, CN, NO2, F, Cl or C(O)NH26,20-24 and alpha pyridine-like nitrogens (α-N)6,16,17 that decrease stability of ArNH+ frequently amplify the mutagenic potency of ArNH2. Studies of aniline (PhNH2) and a number of aniline derivatives with electron-donating groups, such as OH, Me, OMe, NH2 or NMe2, which stabilize ArNH+, showed that these compounds do not follow the classical ArNH+-mediated mechanism of mutagenicity. It has been repeatedly demonstrated that these ArNH2, as opposed to

Page 2 of 14

polycyclic ArNH2, do not cause formation of any DNA adducts in vivo.9 Instead, electron-rich anilines cause their genotoxic or carcinogenic effects by generating oxidative stress conditions through redox cycling of the quinone imine metabolites or by increasing the production of reactive oxygen species.9,25-31 However, when the hydroxylamine or bioesters of aniline, anisidines, toluidines, dimethylanilines and other electron-rich monocyclic ArNH2 are synthesized chemically and tested for mutagenicity, they seem to follow the classical mechanism of mutagenicity through ArNH+.9,32 This indicates that ArNH+ derived from electron-rich anilines are sufficiently stable under aqueous conditions to cause DNA adduct formation,10,25 and the apparent failure of the nitrenium stabilization concept for this class is due to the fact that their metabolic activation pathway does not go through ArNHOH or ArNHOR.9,25 Taken together, experimental observations suggest that there is a gap in our understanding of chemical events reflected in the SMR of ArNH2, and that these gaps are related to the metabolic activation of ArNH2.6,9,19,25 The actual mechanism of metabolic activation of ArNH2 by CYP1A2 is unknown. It has been demonstrated that 2-amino3-methylimidazo[4,5-f]quinoline (IQ), a food-derived carcinogenic heteroaromatic ArNH2, is N-hydroxylated by CYP1A2 under experimental conditions, when the enzyme works as a H2O2-dependent peroxygenase.33 This suggests that the oxidation of ArNH2 undergoes the classical radicaloid mechanism mediated by the H-atom transfer (HAT) from the NH bond of ArNH2 to the electrophilic high-valent iron-oxene species FeIVO (called Compound I, Cpd I),33 which represents the primary oxidant intermediate in the catalytic cycle of P450 enzymes.34,35 Nevertheless, it is still unclear whether the metabolic activation by CYP1A2 undergoes the Cpd I mediated pathway under natural conditions, when the enzyme functions as a monooxygenase. Five mechanistic options for N-hydroxylation of aniline (PhNH2) and three aniline derivatives by the P450 active site, namely HAT, proton transfer (PT), oxygen addition rearrangement (OAR), single-electron transfer (SET) and twoelectron transfer (TET), were recently studied by density functional theory (DFT) calculations using the B3LYP functional.36 The authors concluded that the N-hydroxylation of ArNH2 is mediated by the classical radicaloid mechanism (HAT), because OAR, SET and TET pathways are hindered by higher activation barriers. The recently suggested anionic pathway19 mediated by the proton transfer (PT) from the NH bond of ArNH2 to the dianionic ferrous-peroxo species FeIIOO2- was also rejected mainly since the authors could not locate the transition state (TS) in the PT path for PhNH2.36 However, in our model of the CYP1A2 catalytic site we were able to identify the TS for a number of ArNH2 along the PT reaction coordinate as described below. Thus, there are two putative mechanisms, radicaloid and anionic, that could mediate the N-hydroxylation of primary aromatic amines in CYP1A2, and it is important to compare the feasibility of both mechanisms to gain further understanding of this fundamental process. The feasibility of monooxygenation by P450 enzymes is thought to be coupled to favorable or slightly unfavorable thermochemistry of the rate-determining bond activation step of the bound substrate.19,36-39

2


Page 3 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In this report, the feasibility of the bond activation (BA) step of both the radicaloid and anionic mechanisms of Nhydroxylation by CYP1A2 (Figure 1) is studied for eight ArNH2 by dispersion-corrected DFT calculations. We focus our study on the particular SMR features of ArNH2 that cannot be explained by the nitrenium stabilization concept. Consistent with published recommendations for large molecular systems containing transition metals, in which thermochemistry, kinetics, noncovalent interactions and internuclear distances at TSs are essential,40,41 we used the M06-L density functional. It is demonstrated that mutagenic ArNH2 can be activated in CYP1A2 by both the radicaloid and anionic pathways. However, only the anionic pathway is consistent with the SMR of ArNH2 and only within this pathway non-mutagenic ArNH2 show significantly unfavorable thermodynamics. -H

.

[FeIVO]

ArNH2

- H+ [FeIIOO]2-

ArNH

.

ArNH -

+ OH

by Me groups unless the backbone atoms were explicitly included. The peptide bond between G-316 and A-317 was retained. The backbone motif of the B’ region from S-122 to F125 was also added to provide the anionic side chain of D-313 with the H-bonding partners that exist in CYP1A2. Only the sidechain portions of residues T-321 and L-382 that are exposed to the catalytic site, i.e. CH4 and C3H8, respectively, were included. The side chains of S-122 and D-313 starting with beta-hydrogens, CH3 hydrogens of A-317, as well as CH3 and OH hydrogens of T-124 were allowed to move during energy optimizations. Cartesian coordinates of the rest of the included atoms of the CYP1A2 amino acid residues were kept fixed. Neither geometry nor binding modes of ArNH2 were constrained.

.

[FeIIIOH] + ”OH +” [FeIII OOH]-

HAT ArNHOH PT

BA

Figure 1. Possible mechanisms of N-hydroxylation of ArNH2 by CYP1A2, mediated by hydrogen atom (H·) transfer to Cpd I (HAT) or proton (H+) transfer to the ferrous-peroxo dianion (PT). BA is the rate-determining bond activation step. Subsequent reactions are exothermic, and represent radical recombination of ArNH· with OH radical in the HAT pathway,36 and protoncatalyzed heterolytic cleavage of the hydroperoxyl moiety of [FeIIIOOH]-, which is accompanied by the reaction of the formal OH cation42 with ArNH- in the PT pathway.19

METHODS Fully optimized dispersion-corrected density functional theory calculations at the M06-L/6-31+G* level of theory40 were undertaken to study reaction profiles of the bond activation steps of two putative mechanisms of N-hydroxylation of ArNH2 (HAT and PT) in the catalytic center of human CYP1A2. The doublezeta Chizmadia’s effective core potential CSDZ was used to describe the inner orbitals of Fe.43 Following the two-statereactivity pattern of Cpd I,44,45 the oxidant in the HAT pathway was simulated by both low-spin (triradicaloid doublet) and highspin (quartet) states of Fe4+O-(C20N4H12)2-(SMe)-, where C20N4H12 is porphyrin. The ferrous-peroxo species in the PT pathway was simulated by the dianionic doublet Fe2+O2-(C20N4H12)2-(SMe)-. Molecular systems were treated in the framework of the spinunrestricted formalism. Fragments of residues located in the proximity of the heme in CYP1A2 were added, and geometric constraints were applied to describe molecular environment of the catalytic center of the enzyme revealed in the X-ray structure (Brookhaven PDB database, access code 2HI4).46 The model of the CYP1A2 active site is illustrated in Figure 2. Within this model, the location and geometric constraints of the ironporphyrin heme models with respect to the rest of the enzyme were defined as described earlier.19 The anionic side chain of C458 that coordinates heme Fe in CYP1A2 was represented by methanethiolate, with the observed eclipsed conformation of C-S of C-458 and Fe-N of heme being retained during geometry optimizations. Alpha-carbon of residue G-316, the side chains of S-122, T-124, A-317, T-321, D-313 and L-382 of CYP1A2 were included. Alpha-carbons of the included residues were replaced

Figure 2. The model of CYP1A2 active site with bound PhNH2. Non-polar hydrogens are omitted for clarity. Carbon atoms of PhNH2, porphyrin and CYP1A2 residues are shown in dark green, cyan, and grey, respectively. Oxygens, nitrogens, sulfur and iron are shown in red, dark blue, yellow and purple, respectively. Hbonds are shown by red dotted lines. Fragments of included CYP1A2 residues are denoted. The arrow indicates the alternative conformation of the OH group H-bonded to the α-N of ArNH2, if present. The search for the TS structures was carried out by the mixed Murtagh-Sargent/Powell Hessian-updating protocol.47 The nature of the TS was verified by the number of negative eigenvalues of Hessian during this iterative algorithm. In all identified TS structures, the only eigenvector with the negative eigenvalue corresponded to the H transfer from ArNH2 to the active oxygen of the CYP1A2 model. The nature of each final stationary point was subsequently confirmed by performing calculations of vibrational frequencies of a simplified molecular system at the same level of theory (M06-L/6-31+G*). In these calculations fragments of CYP1A2 amino acids that were kept fixed during geometry optimizations were removed, and only three residues, T124, D-313 and C-458, were modeled. It was shown that the reaction coordinate of the identified TS structures corresponded to the negative eigenvalue of the true Hessian in all cases. Singlepoint calculations at the M06-L/6-311++G** level were carried out to determine the final energy. In these calculations the triplezeta effective core potential LACV3P was used to describe the inner orbitals of Fe.48 This level of theory was also utilized to calculate Gibbs free energy profiles of the key reactions of the CYP1A2 catalytic cycle, including zero-point and thermal corrections at T=298.15K, to confirm the feasibility of the entire PT mechanism for N-hydroxylation of ArNH2. These single-point calculations were performed with symplified molecular systems, in which amino-acid residues forming the CYP1A2 active site were removed from the energy-optimized structures of the full

3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

complexes. The same approximation was used to obtain Gibbs energy profiles of the BA step of the HAT pathway. Current DFT methods are able to cover dynamical electron correlation effects, including pair and three-electron correlation.49 However, they are typically unable to accurately account for nondynamical correlation effects, which is required to properly describe a multiconfigurational nature of iron-porphyrin systems, especially in the low spin states.50-52 The intrinsic propensities of the two oxidants to activate NH bonds of ArNH2 predicted by the M06-L functional were verified by performing single-point calculations of reaction energies of the oxidants using hydrid DFT (B3LYP),53 ab initio multiconfigurational second order perturbation (CASPT2)54 and B3LYP with localized orbital correction (B3LYP-LOC).55-57 In the last two methods, the nondynamical electron correlation was taken into account explicitly. The all-electron CASPT2 calculations were undertaken using the ROHF starting molecular orbitals by the program Gaussian 09,58 with the two-electron integral accuracy being raised from the standard value to 10-12 au. All DFT calculations were performed with Jaguar (version 8.1, Schrödinger, LLC, New York, NY, 2013).59 Statistical analysis of the significance of various properties of ArNH2 for the observed SMR was performed by a cross-validated Partial Least Squares Discriminant Analysis (PLS-DA) method60 using the program Simca P+ (version 12.0.1.0, Umetrics AB, Umeå, Sweden, 2013). Compound lipophilicity at pH=7.4 was predicted by the LogD module of ACD/Labs (release 12.0, Advanced Chemistry Development, Inc., Toronto, Ontario, Canada, 2009). Basicity (pKa) of NH2 groups of ArNH2 were predicted by ab initio quantum mechanical calculations in water61 using the pKa prediction module of Jaguar.

RESULTS AND DISCUSSION The focused set of studied ArNH2 is presented in Figure 3. These compounds exemplify particular features of the SMR of ArNH2 that cannot be linked to the stability of ArNH+. Three compounds, aniline 1 (PhNH2), 4-aminobiphenyl 7 (4-ABP) and IQ 6, were included in the set because there is a considerable weight of evidence of their metabolism in P450 enzymes. Aniline is non-mutagenic by itself and is not readily N-hydroxylated by P450 enzymes,9,32,62,63 whereas carcinogens 4-ABP and IQ are among the most studied ArNH2, which have been established to exert their toxic effects through the ArNH+-mediated mechanism triggered at least partially by the initial N-hydroxylation by CYP1A2.64,65 The pro-mutagenic effects of electron-withdrawing groups were exemplified by mutagenic p-CN-aniline 8. Compounds 2-5 were included to investigate the origin of the intrinsic propensity of α-N to trigger mutagenicity in ArNH2. Typical binding modes of ArNH2 prior to the BA in the model of the catalytic site of CYP1A2 obtained by geometry optimizations at the M06-L/6-31+G* level are illustrated in Figure 4. In both pathways, aromatic rings of all considered ArNH2 remain to be nearly co-planar with the active site cavities identified in the X-ray structures of the entire CYP1 family (see Figure 5A).66 In all reactant complexes (RC), the active oxygen of the oxidant is H-bonded to the NH group of bound ArNH2, and the OH group of T-124 is involved in Hbonding with the edge of the aromatic system. Because of the particular arrangement of the CYP1A2 active site, the bound ArNH2 in the RC of the PT pathway has the most stable H-

Page 4 of 14

bonding pattern with the enzyme, which consists of two strong H-bonds (pattern I in Figure 4).19 In this pattern the NH bond of ArNH2 is H-bonded to the distal oxygen of the ferrousperoxo dianion, while the α-N is H-bonded to the OH group of T-124, in line with predictions.46 In the lowest energy PT reactions of 4, 5 and 6 that have α-N, this H-bonding pattern is maintained all the way through the reaction (Figures 4 and 5B, Tables S1 and S2). However, this H-bonding slows down the BA reaction of these compounds in both the low- and highspin HAT pathways (Table S1), for which the low-barrier path corresponds to the H-bonding pattern II (Figures 4 and 5A, Tables S1 and S2). The pattern II is also maintained by ArNH2 without α-N in both pathways. N N

N N

NH2

NH2

1

NH2

2

NH2

3

4

PhNH2 N

N

N N NH2

N

N NH2

NH2

5

6

NH2

7 4-ABP

IQ

8

Figure 3. The set of studied ArNH2. Three non-mutagenic and five mutagenic compounds in the standard bacterial Ames mutagenicity assays in strains TA98 and TA100, with or without S9, are shown in green and red, respectively.

I

II N

N

D-313

D-313 N

O

H

N

N H

O

O

N H

H

N

T-124

O

H

N H

T-124

D-313 H O

H

N

H O

H

T-124

Figure 4. The sketch of alternative H-bonding patterns of ArNH2 in the active site of CYP1A2 prior to the BA steps. Strong and weak H-bonds are shown by red and blue dashed lines, respectively. The OH group of T-124 is H-bonded to the α-N of ArNH2 and to the side-chain O of D-313 in the patterns I and II, respectively. If there is no α-N in ArNH2, the NH and α-CH bonds are weakly H-bonded to the OH of T-124. The arrow from the active oxygen indicates bonding to FeIV of Cpd I or to the proximal oxygen of the ferrous-peroxo dianion.

4


Page 5 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Natural Bond Orbital (NBO) analysis67 was carried out using the electron density distributions obtained in all identified stationary points of the reactions at the M06-L/6-31+G* level of theory. The spin densities and partial charges, as well as the orbital occupancies were used to characterize the mechanisms of the BA reactions along the low-spin HAT and PT pathways. Results are presented in Figures 6 and 7. The obtained geometries and electronic structures of both oxidants as well as their reaction products are in line with published results of DFTonly calculations, which did not include the NH···S H-bond or the external electric field and utilized the hybrid B3LYP functional.35,50-52 The low-spin Cpd I appeared in a triradicaloid doublet ground state with the mixed 2(A2u – П/ΣS) electronic structure. In line with previous theoretical studies, in this structure two unpaired electrons occupy two orthogonal antibonding π* orbitals of the oxyferryl (FeO) moiety and couple to triplet configuration, whereas the third electron occupies the mixed (π/σS - a2u) orbital and couples antiferromagnetically with the triplet (Figure 6C).50-52 Because of geometric distortions in the ground state of Cpd I, in which the methanethiolate ligand deviates from the exact axial coordination of the porphyrin ring, the tilted hybrid orbital on the sulfur includes both πS and σS components.50,51,68 The partial cation-radicaloid structure of the porphyrin ring is inherent in Cpd I, in line with prior theoretical investigations performed by DFT and CASPT2 approaches, as well as with spectroscopic experiments.35,50-52,68 Although the contribution of the sulfur-centered radical in the electronic structure of Cpd I seems quite significant (Figure 6B), it is consistent with previous studies by DFT-only and CASPT2 calculations.35,50-52,68 Besides, this contribution has been shown to be strongly dependent on the particular environment of the P450 active site.35 During the course of the H-atom transfer from ArNH2 to the low-spin Cpd I, the πyz and π*yz orbitals of the FeO moiety are destroyed by the forming OH bond, and the negative spin of the (π/σS - a2u) orbital disappears and appears in the πHOMO of ArNH2. Simultaneously, the π and σ components of the hybrid sulfur-centered orbital get separated, which results in formation of the pure πyz and π*yz orbitals of the FeS moiety and lowering the pure (σS - a2u) orbital. Correspondingly, in the resulting hydroxo intermediate (HYD) the methanethiolate ligand takes the exact axial position. The orbital occupancy in the energy optimized geometry of HYD suggests the triradicaloid doublet ground state, in which the unpaired electrons of the oxidant are coupled in the triplet configuration (Figure 6C). Thus, the BA electronic state of the HAT pathway is formed by the electron shift from the NH bond to the porphyrin-based a2u orbital. This shift generates the FeIV type HYD intermediate, in line with previous studies of the Cpd I mediated C-H bond hydroxylation.45,69 The alternative electron shift to the π*yz(FeO) orbital (Figure 6C)45,69 in our model resulted in a much less stable triradicaloid doublet electromer with the diradicaloid singlet configuration of the oxidant (+9 kcal/mol). The π-type of the singly occupied HOMO orbital of the ligand in the BA state (Figure 6A) suggests that the HAT pathway may have proton-coupled electron transfer character (PCET), in which movements of proton and electron are concerted but asynchronous.70 In this pathway the proton and electron are indeed transferred to different locations, proton to the oxene, and electron to the mixed S/porphyrin based orbital (Figure 6C). However, remarkable spin density of the oxene in

Cpd I in the RC with ArNH2 and small negative charge of the ligand in the TS (Figure 6B) suggest that the TS of the reaction of Cpd I with ArNH2 has a dominant HAT character, which is in line with the established electrophilic behavior of Cpd I.71 The doublet and quartet spin RC of the HAT pathway are within 1 kcal/mol, and the potential energy surfaces of the BA steps in these two spin states are very close in energy (Tables S1 and S2), in line with previous theoretical studies of HAT reactions in other systems.39,51,52,72 The ground state of the high-spin Cpd I was shown to have the mixed 4(A2u – П/ΣS) electronic structure, in which three unpaired electrons occupy the same orbitals as in the low-spin state but couple ferromagnetically.39,51,52 Because of the involvement of the same orbitals in the electron shifts, the HAT energy profiles of the BA steps in both spin states are very similar.

Figure 5. The lowest energy TS structures of the BA steps of IQ in the CYP1A2 model along the low-spin HAT (A) and PT (B) pathways. The plane of IQ nearly coincides with that of αnaphthoflavone (shown by thin lines), which is present in the Xray structure of CYP1A2 as illustrated in (A). Non-polar H and fragments of L-382 and T-321 are omitted for clarity. H-bonds are illustrated by red dotted lines. The H transfer routes are shown by red arrows. Carbons of IQ are shown in green.

The changes in spin density and orbital occupancies along the PT pathway given in Figure 7 agree well with previous theoretical studies of the ferrous-peroxo intermediate and Compound 0 (Cpd 0).35,73 Formation of the OH bond destroys

5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 14

Figure 6. Mechanism of BA of 2-aminopyridine in the CYP1A2 active site in the low-spin HAT pathway. (A) Spin densities in the reactant complex (RC), transition state (TS) and bond activated state (BA). Positive and negative spin densities are shown in red and blue, respectively. (B) Spin and charge populations obtained by the NBO analysis. The negative spin population rising in the ArNH moiety is highlighted in blue. (C) Occupancy diagram of the reaction. The residues defining the active site are not shown. The arrows indicate the major electron shifts that accompany the concerted NH bond activation mechanism. The alternative electron shift shown by the dashed arrow illustrates another mechanism that results in the higher energy electromer with the diradicaloid singlet configuration of the oxidant.

Figure 7. Mechanism of BA of 2-aminopyridine in the CYP1A2 active site in the PT pathway. (A) Spin densities in the reactant complex (RC), transition state (TS) and bond activated state (BA). Spin densities are shown in red. (B) Spin and charge populations of obtained by the NBO analysis. The negative charge rising in the ArNH moiety is highlighted in red. (C) Occupancy diagram of the reaction.

6


Page 7 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Chemical Research in Toxicology difference of bond dissociation energies of the bond of the substrate undergoing abstraction and the OH bond in the enzyme being made as expressed in eq. (1),37-39 provided the interaction energy of the enzyme and substrate is small and virtually constant throughout the reaction.38 Consistent with this concept, the barrier heights and reaction energies in both HAT and PT pathways linearly increase with increasing NH bond dissociation energies (BDENH) of ArNH2 in most cases. ∆E = BDENH – BDEOH

Figure 8. The barrier heights and reaction energies of the bond activation steps of compounds 1-8 along the low-spin HAT (A) and PT (B) pathways obtained at the M06-L/6-311++G**//631+G* level of theory in the CYP1A2 model plotted against BDENH of ArNH2 (M06-L/6-311++G**, in kcal/mol). The reactant complexes were used as references. Reaction energies and activation barriers are illustrated as circles and triangles, respectively; red for mutagenic, green for non-mutagenic. The feasibility boundary of the BA step of the PT pathway is illustrated as a horizontal red dotted line of 6.5 kcal/mol. Two pro-mutagenic effects that facilitate the BA of ArNH2 along the PT pathway are shown by black arrows: (a) resonance stabilization of the anionic form (ArNH-), and (b) the H-bonding of the α-N to the OH group of T-124 of CYP1A2. The red arrows point to the intrinsic propensities of the respective oxidants for abstraction of hydrogen atom (A) or proton (B).

the πyz and π*yz orbitals of the bound superoxide anion-radical. Correspondingly, the unpaired electron, which is shared by the antibonding π*yz orbital of OO with the dyz orbital of Fe all the way through the reaction, condenses on the py orbital of the proximal O. Since this py orbital is stabilized in the course of the reaction, the electron pair and the unpaired electron swap their places, thereby changing the effective oxidation state ofFe from FeII to FeIII and forming the bound hydroperoxyanion HOO-. It should be noted that in the product of the PT reaction, the π-HOMO in the anionic form of ArNH2 remains doubly occupied. The activation barriers and energies of the BA reactions of the compounds under study in the model of the CYP1A2 active site are presented in Figure 8 and Table S1. The reaction energy (∆E) is thought to be determined by the

(1)

Both BDENH and BDEOH values refer to the homolytic and heterolytic processes within the HAT and PT pathways, respectively. Accordingly, the obtained data suggest that the BA reactions in these pathways are facilitated by the resonance stabilization of the radicaloid or anionic forms of ArNH2. The data shown in Figure 8B indicate that there are several outliers from the major linear trends of the activation barriers and reaction energies on BDENH. The low barriers of 6 and 8 deviate from linearity in the PT pathway simply because they cannot be smaller than zero. The PT reaction profiles of 4 and 5, however, represent remarkable exceptions. These profiles are much less endothermic and hindered by a much lower barriers than expected from the corresponding BDENH. This suggests that the anionic forms of 4 and 5 are significantlymore stabilized in the catalytic site of CYP1A2 than ArNH2 in the RC. These compounds have α-N in the resonance positions of the anionic forms, therefore the additional stabilization is caused by charge-reinforceing of the H-bonding between α-N and the OH group of T-124 in the BA state, when the H-bond acceptor becomes negatively charged. This stabilization of the BA product is independent of BDENH and is an intrinsic feature of the PT pathway in CYP1A2. The stabilizing effect of the charge-reinforcing of the α-N···HO Hbonding in the BA state seems insignificant for IQ, in which the excess negative charge is delocalized over three aromatic rings. It should be noted that physical significance of small activation barrier heights of mutagenic compounds as well as shallow energy minima of BA reaction intermediates along the PT pathway have to be verified by Gibbs energy calculations. It is important to compare the intrinsic propensities of Cpd I and ferrous-peroxo dianion to activate NH bonds of ArNH2 in CYP1A2. The oxidant with higher propensity is expected to activate NH bonds with higher thermodynamic drive, in line with eq. (1). The key parameter of the intrinsic propensity of the oxidant is BDEOH, which can be calculated in two ways, i.e. from the linear regressions of eq. (1) obtained for a series of substrates or by the direct QM calculations of the isolated oxidant. The values of BDEOH obtained by the first method are shown in Figure 8. The red arrows and the accompanying red values in Figures 8A and 8B illustrate the location of BDEOH of Cpd I and ferrous-peroxo species, respectively, obtained from eq. (1) with ∆E = 0. Compounds 4 and 5 were excluded from the PT regressions. It is important to keep in mind that the intrinsic propensity of the ferrous-peroxo dianion to activate NH bonds of ArNH2 with α-N in the CYP1A2 active site is significantly higher than the value shown in Figure 8B. The BDEOH values of the oxidants inherent in both HAT and PT pathways obtained by the direct QM calculations at different levels of theory are presented in Table 1. Both triplet and diradicaloid singlet states of the hydroxo intermediate are


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

included. The electronic structure of the iron-porphyrin systems suggests that they can be properly treated only by the multireference wavefunction, which accounts for both dynamical and nondynamical electron correlation.35,73 The recently developed DFT methods like M06-L are able to implicitly account for these effects since they are parametrized using the observed data on kinetics and thermochemistry.40,41,55 Nevertheless, the explicit description of both the dynamical and nondynamical electron correlation is required to estimate possible errors in the BDEOH values caused by an incorrect type of the wavefunctions. The tabulated values of BDEOH obtained at the M06-L/6311++G**//6-31+G* level of theory (see Table 1) are in line with published DFT predictions for the low-spin HAT and PT pathways (95.738 and 422.674 kcal/mol, respectively). As is seen, different levels of theory reproduce these values reasonably well. As expected, a better account for the nondynamical correlation by B3LYP-LOC and CASPT2 methods tends to decrease the singlet-triplet gap in the diradicaloid hydroxo intermediate, as the stabilizing effect of this type of electron correlation is more pronounced in the low spin states.35 The H-atom abstracting power of Cpd I given by the CASPT2(11,9) approximation appears too high with respect to the other levels of theory. As it has been noted, multireference ab initio calculations of iron-porphyrin systems that use large active spaces are not readily applicable to predictions of reaction energies or activation barriers.73 Hence, the intrinsic propensities of the oxidants for H-atom or proton abstraction given by the B3LYP-LOC approach, which is built on description of the effects of the nondynamical electron correlation on reaction energies, seem most reliable. As is seen, this approximation is in agreement with the M06-L/6311++G** level of theory in predicting the values of BDEOH. This confirms results of prior investigations that suggested that molecular systems of the multideterminant nature are reasonably well described by modern DFT methods, despite the incorrect description of their electronic structure by a single Kohn-Sham determinant.35,52,55-57,73,75 The intrinsic propensity of the low-spin Cpd I for H-atom abstraction predicted by the direct QM calculations of BDEOH (95.9 kcal/mol, Table 1) matches the average propensity obtained from the low-spin HAT reaction energies when using the same M06-L functional (95.6 kcal/mol, Figure 8A). However, this is not the case for the two estimates of BDEOH of Cpd 0 (358.9 kcal/mol in Figure 8B versus 424.8 kcal/mol in Table 1). Both these values reflect the proton abstracting power of the ferrous-peroxo dianion, but the BDEOH estimate of 358.9 kcal/mol is directly relevant to the proton-abstracting propensity in the enzyme, where the total charge of the molecular system does not change. On the other hand, the proton affinity of the isolated ferrous-peroxo dianion (424.8 kcal/mol) is consistent with values computed at higher levels of theory. As is seen, the direct QM calculations of the absolute values of BDEOH of the two oxidants at the M06-L/6311++G**//6-31+G* level of theory give virtually the same result as the highly correlated levels of theory. This also validates the positions of the regression lines of BA reaction energies of these oxidants with ArNH2 (Figure 8), which are obtained at the same DFT level. The intrinsic propensity of the high-spin Cpd I to activate NH bonds of ArNH2 is nearly the same as its low-spin electromer (Tables 1 and S1). The attained barriers and reaction energies of mutagenic ArNH2 in the HAT and PT pathways shown in Figure 8 suggest that both reactions are possible. The effective affinity

Page 8 of 14

of Cpd I for H-atom (95.6 kcal/mol) exceeds BDENH of seven out of eight considered ArNH2. Consistent with eq. (1), the BA reactions undergoing the HAT pathway are exothermic for these seven ArNH2. The high intrinsic propensity of Cpd I for H-atom abstraction from ArNH2 is in line with recent theoretical study, in which the HAT pathway for PhNH2 has been shown to be hindered by a small activation barrier (4.3 and 3.8 kcal/mol in the low-spin and high-spin reactions, respectively).36 The apparent ease of this reaction in both doublet and quartet spin states of the oxidant contradicts experimental data, which demonstrated that PhNH2 is not mutagenic and is likely not N-hydroxylated by P450s.9,32,62,63 The effective propensity of the ferrous-peroxo species for proton abstraction (358.9 kcal/mol) is higher than the heterolytic BDENH of only two out of eight considered ArNH2. This suggests that the intrinsic propensity of the ferrousperoxo dianion to activate NH bonds of ArNH2 is lower than that of Cpd I. As a result, the BA reactions of the PT pathway become unlikely for ArNH2 with strong NH bonds. Figure 8B indicates that non-mutagenic compounds 1-3 are particularly difficult to activate by the PT mechanism (Figure 8B). The BA states of these compounds along the PT pathway are shallow and the least stable. As is seen, the endothermicity boundary of 6.5 kcal/mol splits the BA reaction energies of mutagenic and non-mutagenic ArNH2, including PhNH2. This split reflects the previously noticed electon affinity of ArNH2 as an important factor for their mutagenic potency.21 The promutagenic electron-withdrawing groups, which increase electron affinity of ArNH2, also increase acidity of their NH2 groups,76 which facilitates the BA step of the PT pathway. Likewise, the pro-mutagenic effects of α-N of ArNH2 are consistent with the PT pathway, where these nitrogens remain H-bonded to the OH group of T-124 and facilitate the BA reaction (Figure 8B), but not with the HAT pathway, where the H-bonded route is not feasible (Table S1). As is seen in Figure 8B, the effect of the H-bonding in facilitating the PT reactions of 4 and 5 relative to 3 and 4 is independent of the resonance stabilization of the anionic forms. It should be noted that the effect of this H-bonding can also be caused by the increase of residence time of ArNH2 in CYP1A2.19,46 The HAT pathway is also inconsistent with the observed effects of electron-donating groups placed in resonance positions of PhNH2. These groups raise the π-HOMO of ArNH2, thereby stabilizing their radicaloid forms and thus are anticipated to facilitate the H-atom abstraction from ArNH2 by Cpd I regardless of the spin state of the oxidant (Figure 6C), in line with results of recent theoretical studies.36 For the same reason these groups also stabilize ArNH+, thereby increasing the chemical reactivity of ArNHOH and ArNHOAc toward dG.10 However, according to experiments electron rich anilines do not follow the classical mechanism of mutagenicity through ArNH+.9,25-31 Within the PT pathway, on the other hand, the occupancy of the HOMO of ArNH2 does not change (Figure 7C). In fact, electron-donating groups added to PhNH2 increase heterolytic BDENH and shift the molecule further to the right in Figure 8B, thus making the PT event impossible. Data suggest that endothermicity of the PT reaction is linked to hindering the classical ArNH+-mediated mechanism of mutagenicity of ArNH2. Indeed, the move along the regression line to the top right corner in Figure 8B starts with the exothermic reaction profile of IQ (6), which follows only the classical mechanism of mutagenicity,77 then passes a slightly endothermic reaction profile of 4-ABP (7), which is thought to exert mutagenicity both by the classical mechanism and by


8

Page 9 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


oxidative damage of DNA,64,65,78,79 and finally reaches the area of electron-rich anilines (1 and above) that exhibit high endothermicity of the PT reaction and do not follow the classical mechanism.9,25-31 This being the case, the feasibility of the BA step of the PT mechanism in the CYP1A2 catalytic site likely represents a requirement for the ArNH+-mediated mechanism for the majority of ArNH2. Table 1. The values of BDEOH in the hydroxo intermediate and Cpd 0 obtained by direct QM calculations at different levels of theory.a Level of theory

HAT

PT

Theory

Basis

M06-L

6-311++G**

95.9/96.3b

85.8

424.8

B3LYP

6-311G*

92.1/92.2

86.3

422.6

B3LYP

6-311++G**

95.8/96.8

82.9

429.0

[FeIVO-H]

[FeIIIO-H]

[FeIIIOO-H]-

B3LYP-LOC

6-311G*

94.1

89.1

417.1

CASPT2(3,3)

6-311G*

NC

98.7

424.6

CASPT2(3,4)

6-311G*

103.1

101.4

429.0

CASPT2(7,6)

6-311G*

91.1

87.0

430.1

CASPT2(11,9)

6-311G*

114.5

112.2

NC

predisposed for formation of two reactive oxygen species, either OH radical (reaction a) or OH cation (reaction b), whereas the route c remains impossible. Results suggest that the O-O bond cleavage that undergoes the release of the OH cation (route b) shows the most favorable thermodynamics if the substrates are negatively charged. It should be noted that formation of OH cation is possible only in the concerted fashion, in which proximal protonation of the ferric hydrogen peroxide complex goes along with the charge recombination of OH cation and a bound anionic species. The OH cationic route is the only mechanism of the O-O bond cleavage in the ferric hydrogen peroxide complex that results in the resting state of the P450 catalytic cycle.

a All values are in kcal/mol. Homolytic and heterolytic dissociation of the indicated O-H bonds are studied in HAT and PT pathways, respectively. Data are obtained by single-point calculations using the M06-L/6-31+G* optimized geometries. bValues before and after a slash denote dissociation routes to the triradicaloid doublet and quartet states of Cpd I, respectively. Two numbers within the parentheses in the CASPT2 method signify the number of electrons and the number of molecular orbitals included in the active space. One more electron was augmented to the active space of the hydroxo intermediate. Situations when CASSCF does not converge are denoted by NC. Absolute energies are given in Table S3.

To demonstrate that the entire catalytic cycle of CYP1A2 is consistent with the PT mechanism, it is important to confirm that formation of ArNHOH from the complex of the anionic doublet state of FeIIIOOH (Cpd 0) with ArNH- is physically reasonable and returns the P450 catalytic cycle to the initial ferric resting state.35 Two protonation events are needed to close the P450 catalytic cycle back to the resting state. The first protonation of the proximal oxygen atom of the ferrichydroperoxide complex results in the neutral FeIIIO(H)OH species, i.e. hydrogen peroxide bound to the doublet ferric state of Fe.42 The second proximal protonation cleaves the peroxide bond of the isolated complex and releases OH radical with no barrier. We have to demonstrate that in the presence of arylamine anion, the preferred O-O bond cleavage undergoes the heterolytic mechanism with formal formation of OH cation that reacts with arylamine anion to form ArNHOH. There are three possibilities for the O-O bond cleavage of bound hydrogen peroxide, one homolytic and two heterolytic (Figure 9A). Gibbs free energies (∆G) of these reactions with 13 substrate models (M) were obtained by single-point calculations at the M06-L/6-311++G** level of theory using the fully M06-L/6-31+G* energy optimized structures of the involved species (Figure 9B, Table S4). To make the model reactions relevant to the PT mechanism, the set of substrate models consisted of three neutral and ten negatively charged closed-shell species including the anionic forms of all considered ArNH2. As is seen, Gibbs free energies of the model reactions with no bound substrate leave no chance for formation of either OH anion or OH cation. On the other hand, the doubly protonated complex of Cpd 0 with a substate is

Figure 9. Gibbs free energies of three possible mechanisms of the peroxide bond cleavage of hydrogen peroxide bound to the methanethiolate-ligated ferric-porphyrin triggered by proximal oxygen protonation with different model substrates (M). (A) Alternative mechanisms of the cleavage include one homolytic (a) and two heterolytic (b and c) routes. Ellipses represent porphyrin rings. Charge and multiplicity of the ground states are shown in the top right corner of each intermediate. (B) A plot of free energies of the reactions with indicated M against free energy of reaction with OH cation calculated at the M06-L/6-311++G**//631+G* level. Ethyl-guanidine was used to study reaction profiles with neutral arginine (Arg). Gibbs energies of reactions a, b and c are shown in green, red and blue, respectively. Dashed lines represent reaction energies without substrate. The heterolytic cleavage mechanism (b), which is mediated by formal release of the OH cation, results in the classical ferric resting state and represents the preferred route for hydroxylation of negatively charged species in the concerted fashion, is outlined in (B).


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 14

A N

O Fe S

B

O

H N

H -2

2+

Fe

RC

S

N

H

O

Fe

BA

3+

S

O O

H+ Fe

Peroxide

H

OH 0 H

H 3+

S

O Fe S

H 3+

RS

O2 ArNH2, 2e-

∆G, kcal/mol

∆G, kcal/mol

15

100

BA

10 5

RC

2

0

1 3 7 8

Peroxide

-100

RS -200

-5 -10

N

H 0

H

H

O

H+

3+

N H+

0 H

ArNHOH H2O

0

H

-1

O O

H

-300

6 Step 1: NH bond activation by H+ transfer

Step 2: HO+ rebound

Figure 10. Catalytic cycle of cytochrome P450 1A2, which N-hydroxylates aromatic amines through the PT mechanism. (A) Schematic representation of the major events. Ellipses represent porphyrin rings. Charge and multiplicity of the ground state of each intermediate of the catalytic cycle are shown in the top right corner. The rate-limiting step that splits aromatic amines into mutagenic and non-mutagenic classes is framed. (B) Free energes of both steps of the N-hydroxylation mechanism, calculated at the M06-L/6-311++G**//6-31+G* level of theory for indicated ArNH2. All amino-acid residues of the CYP1A2 substrate binding site were removed from the energy optimized structures. Free energy profiles of mutagenic and non-mutagenic ArNH2 are shown in red and green, respectively. Free energies of the HO+ rebound step were calculated according to the route (b) in Figure 9A. RC is reactant complex, BA is bond activated complex, Peroxide is hydrogen peroxide bound to the ferric state of the catalytic center, RS is resting state. Details are given in Tables S4-S6.

The entire catalytic cycle of P450 1A2 that executes the Nhydroxylation of ArNH2 by the PT mechanism is presented in Figure 10A. Figure 10B gives free energy profiles of the two steps of this mechanism, NH bond activation and OH cation rebound. Since the intrinsic propensity of the ferrous-peroxo species for proton abstraction is comparable with typical heterolytic BDENH of ArNH2 (Figure 8B), the first step is the rate-determining event of the overall mechanism. The OH cation rebound step is highly exothermic for all ArNH2 regardless of their mutagenicity. Gibbs free energies of the RC, TS and BA states of the NH bond activation steps along both HAT and PT pathways were obtained by single-point calculations of the simplified molecular systems at the M06-L/6-311++G**//6-31+G* level of theory (Tables S5 and S6). The BA free energy profiles of 4 and 5 along the PT pathway were not studied because correct descriptions of these profiles require inclusion of residues comprising the CYP1A2 active site. Gibbs energy profiles of the

BA reactions of the simplified complexes along the PT

pathway (Figure 10B, Table S5) differ from the DFT-energy profiles obtained for the full molecular systems at the same level of theory but without the zero-point and thermal corrections (Figure 8B, Table S1). Reaction Gibbs energies (∆G) and Gibbs barrier heights (∆G#) for proton transfer are considerably more favorable than the corresponding DFTvalues (∆E and ∆E#), namely by 2.2±0.5 and 4.8±0.7 kcal/mol, respectively. This results in vanishing of the activation barriers along the PT reaction coordinate for all considered ArNH2. The BA states of the non-mutagenic compounds 1, 2 and 3 do not represent local minima on the Gibbs energy surface. This effect is in line with prior B3LYP calculations of the PT event of 1, which showed that the TS structure between the RC and BA states does not exist.36 On the other hand, the BA steps of the mutagenic compounds 6, 7 and 8 entail favorable barrierfree proton transfer to the ferrous-peroxo intermediate. Thus, the BA states of considered mutagenic compounds represent true reaction intermediates along the PT pathway both with and without the required corrections to the Gibbs free energy.


10

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


The Gibbs free energies of the HAT reactions obtained for the simplified molecular systems (∆G, Table S5) do not show any significant shift with respect to the DFT-energy profiles of the full complexes (∆E, Figure 8A, Table S1). The activation barriers in Gibbs energy remain in the HAT pathway, which are about 40% lower than ∆E#. The noticeable stabilization of the BA states of the free energy profiles detected for the simplified molecular systems in the PT pathway is caused by the removal of the ionic repulsion between anionic forms of ArNH2 and the negatively charged residue D-313 in these systems. Since the radicaloid forms of ArNH2 are neutral, this drift is not detected in the HAT pathway. As a result, the primary difference between the BA profiles of ArNH2 along the HAT and PT pathways in the CYP1A2 active site remains in Gibbs free energy, namely the split of mutagenic and nonmutagenic ArNH2 is linked to the thermodynamics of the NH bond activation step of the PT pathway (Figure 10B). Stabilization of the anionic form of ArNH2 plays the central role in the rate-determining NH bond activation event of the entire mechanism of N-hydroxylation of aromatic amines. The exothermic or slightly endothermic thermodynamics of the NH bond activation is likely required for mutagenicity of ArNH2. Results suggest that N-hydroxylation of ArNH2 in CYP1A2 undergoes the PT pathway mediated by a nucleophilic ferrousperoxo dianion. This conclusion is not based on an insufficient propensity of Cpd I to metabolize ArNH2 in the catalytic cavity of CYP1A2, but rather on its apparent ability to Nhydroxylate non-mutagenic compounds. This means that ArNH2 likely do not bind Cpd I in the CYP1A2 substrate cavity in the productive binding mode suitable for the Nhydroxylation (Figure 5A). In other words, binding of ArNH2 to the catalytic site of CYP1A2 in the productive binding mode required for the N-hydroxylation inhibits the formation of Cpd I. According to the X-ray structure of this enzyme, the nearly planar substrate binding cavity includes the proton delivery route that passes through D-313 and T-124 to the distal oxygen of the ferrous-peroxo species. In the productive binding mode of ArNH2, the NH2 group occupies the position of the key water molecule within the proton delivery route (see Figures 4 and 5B),19 which may interrupt the formation of Cpd I from the preceding ferrous-peroxo dianion. This could be one explanation for the fact that the modest thermodynamic drive of the ferrous-peroxo species, but not of the more powerful Cpd I, is reflected in the SMR of ArNH2. Perhaps in other P450 enzymes, which have been shown to Nhydroxylate a number of particular monocyclic or polycyclic ArNH2, e.g. 1B1, 2B1, 2B2, 2C11, 2D6, 2E1 and 3A4,9,65 binding of ArNH2 to the catalytic site does not prevent the formation of Cpd I. In these cases Cpd I could contribute to the metabolic activation of ArNH2. However, the anionic mechanism of metabolic activation of ArNH2 in CYP1A2 makes the deepest impact on the SMR. Thus, our results suggest that the observed SMR of ArNH2 including noted deviations from the nitrenium stabilization concept can be rationalized if the metabiolic activation of ArNH2 by CYP1A2 undergoes the anionic pathway. This does not immediately imply that the anionic mechanism does occur in reality. Perhaps there could be other causes for the contradictory motifs in the SMR of ArNH2, not necessarily related to the mechanism of N-hydroxylation. Electronwithdrawing functions and pyridine-like α-N could accelerate ArNH+-mediated mutagenesis by facilitating yet unknown chemical reactions paving the mutagenicity pathway. To

explore this possibility we investigate the statistical significance of a number of properties of ArNH2 for the observed split into mutagenic and non-mutagenic compounds. A set of 14 variables that could be relevant to the actual mechanism of mutagenicity by ArNH2 was calculated for each compound (Table S7). Apart from reaction energy (∆E) and activation barrier (∆E#) of the BA steps of the low-spin HAT and PT pathways in CYP1A2 obtained for the full complexes of ArNH2 in the CYP1A2 active site, this set included proton affinity, electron affinity, lipophilicity, energies of HOMO and LUMO, basic pKa of NH2, homolytic and heterolytic BDENH, and formation energy of nitrenium and cation-radicaloid forms. Since Gibbs energies (∆G and ∆G#) obtained for the simplified molecular systems correlated well with the DFTenergies (∆E and ∆E#), Gibbs energies were not included. Partial Least Squares Discriminant Analysis (PLS-DA) guided us to the subset of parameters, which are most likely responsible for the observed SMR of the considered ArNH2. The application of the PLS-DA algorithm for the initial set of variables did not result in a statistically significant linear regression (Q2 = 0.19) but allowed us to identify and remove insignificant variables based on their importance in projection scores. This multivariate technique reduces the noise in the data by finding a smaller number of components that explain the variance best. After unimportant variables were removed, the PLS-DA method resulted in the statistically significant model (Q2 = 0.86). The coefficients of the optimal discriminant regression inherent in the non-mutagenic class are presented in Figure 12. As is seen, the subset of variables, which are most relevant to the SMR of the focused set of ArNH2 are dominated by parameters that are linked to the efficiency of the anionic pathway, namely reaction energy (∆E) and barrier height (∆E#) of the bond activation along the PT path, energy of LUMO, heterolytic BDENH and electron affinity of ArNH2.19 The basicity of NH2 group is also present in the final PLS-DA model, but its influence on the observed split is statistically insignificant. Therefore, the results of PLSDA suggest that non-mutagenic ArNH2 are mainly described by more endothermic thermodynamics of the BA step of the PT pathway.

PLS-DA regression coefficients

Page 11 of 14

1.0 0.5 0.0 -0.5 -1.0 -1.5 PT reaction energy

Electron affinity (ArNH2)

PT barrier height

Energy of LUMO

Heterolytic BDENH

pKa (N in ArNH2)

Figure 11. Coefficients of the optimal PLS-DA regression that discriminates mutagenic and non-mutagenic compounds in the focused set of ArNH2. The illustrated signs of the coefficients describe the non-mutagenic class. Error bars correspond to the confidence level of 95%.


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

CONCLUSIONS The energy profiles of the bond activation steps of two previously suggested mechanisms of N-hydroxylation of primary aromatic amines in the catalytic cavity of CYP1A2, radicaloid and anionic, are studied by DFT calculations at the M06-L/6-311++G**//6-31+G* level. The purpose was to find a mechanism-based explanation for the distinct features of the structure-mutagenicity relationships of ArNH2 that go beyond the nitrenium stabilization concept. The major focus was made on the thermodynamics of the NH bond activation of ArNH2. Both mechanisms are shown to be feasible for the activation of NH bonds of considered mutagenic ArNH2, but the radicaloid pathway does not explain the key features of the strucuremutagenicity relationship. Specifically, the lack of the ArNH+mediated mutagenicity of electron-rich monocyclic ArNH2, such as aniline, p-toluidine and o-anisidine, cannot be linked to an insufficient reactivity or low thermodynamic drive of Cpd I in the framework of the radicaloid mechanism. Besides, the mechanistic origin of the pro-mutagenic effects of electron-withdrawing groups or alpha pyridine-like nitrogens remains unclear. In fact, relative stability of both radicaloid and nitrenium forms of ArNH2 depends on the energy of the πHOMO of the parent ArNH2, therefore the ease of the H-atom abstraction from ArNH2 by Cpd I is parallel to the nitrenium ion stabilization. Conversely, feasibility of the proton transfer from ArNH2 to the ferrous-peroxo species within the anionic pathway is clearly reflected in the structure-mutagenicity relationship of ArNH2. Thus, only non-mutagenic ArNH2 exhibit distinctly endothermic thermodynamics of the proton transfer. Further, electron-withdrawing groups make NH2 of ArNH2 more acidic, thereby facilitating the proton transfer event. The alpha pyridine-like nitrogens are H-bonded to the OH group of T124 of CYP1A2, which facilitates the proton transfer event by reinforcing the H-bond in the product of the reaction. This likely makes the anionic pathway the major mechanism for Nhydroxylation of ArNH2 in CYP1A2. Involvement of the ferrous-peroxo species in N-hydroxylation of ArNH2 is due to the efficient barrier-free bond activation step characteristic of the PT mechanism. The NH2 group of ArNH2 bound in the productive binding mode to the CYP1A2 active site probably interrupts the proton delivery events of the catalytic cycle required for the formation of Cpd I in this enzyme, thereby preventing metabolic activation of electron-rich ArNH2 by the HAT mechanism. Thus, both the classical Cpd I intermediate and the ferrous-peroxo dianion are capable of N-hydroxylation of ArNH2, but the former, more robust oxidant, does not seem available in the CYP1A2 enzyme with bound ArNH2, whereas the latter being more reactive has a limited thermodynamic drive to N-hydroxylate electron-rich monocyclic ArNH2. Hereby we suggest that the chemical event in the mechanism of mutagenicity of ArNH2 that complicates the observed structure-mutagenicity relationship of ArNH2 by causing many exceptions from the nitrenium stabilization concept is the rate-determining proton abstraction from the NH bond by the dianionic ferrous-peroxo intermediate in the CYP1A2 catalytic site prior to the N-hydroxylation. Further chemical events of the CYP1A2 catalytic cycle that finilize the N-hydroxylation include proton-catalyzed heterolytic cleavage of the peroxide bond of the ferric-hydroperoxide intermediate complex that goes along with the reaction of formal OH cation with arylamine anion in a concerted fashion, which is highly exothermic for all aromatic amines. Results of Partial Least

Page 12 of 14

Squares Discriminant Analysis demonstrate that unfavorable thermodynamics of the NH bond activation along the anionic pathway appears to be the most significant feature of nongenotoxic ArNH2. This statistical analysis validates the primary role of the ferrous-peroxo dianion of the CYP1A2 catalytic cycle in metabolic activation of ArNH2.

AUTHOR INFORMATION Corresponding author *E-mail: [email protected]. Telephone: +4631-7064347.

Notes The authors declare no competing financial interest.

ACKNOWLEDGEMENTS Inspirational comments of Dr. Niklas Blomberg are deeply appreciated. Dr. Art Bochevarov is acknowledged for both discussions and reviewing the manuscript.

ASSOCIATED CONTENT Supporting Information Available (i) Computational details; (ii) absolute and relative energies, and energy-optimized atomic Cartesian coordinates of the structures discussed in the paper; (iii) information for the statistical analysis. This information is available free of charge via the Interner at http://pubs. acs.org.

ABBREVIATIONS 4-ABP, 4-aminobiphenyl; BA, bond activation; BDE, bond dissociation energy; CASPT2, complete active space second order perturbation; Cpd 0, Compound 0; Cpd I, Compound I; CYP1A2, cytochrome P450 1A2; 1B1, 2B1, 2B2, 2C11, 2D6, 2E1 and 3A4, other enzymes of the cytochrome P450 family; DFT, density functional theory; dG, 2’-deoxyguanosine; HAT, H-atom transfer; HOMO, highest occupied molecular orbital; HYD, hydroxo species; IQ, 2-amino-3-methylimidazo[4,5f]quinoline; LUMO, lowest unoccupied molecular orbital; PCET, proton-coupled electron transfer; PLS-DA, Partial Least Squares Discriminant Analysis; PT, proton transfer; RC, reactant complex; RS, resting state; SMR, structuremutagenicity relationship; TS, transition state; α-N, pyridinelike nitrogen in the α position

REFERENCES (1) Benigni, R., and Passerini, L. (2002) Carcinogenicity of the aromatic amines: From structure-activity relationships to mechanisms of action and risk assessment. Mutat. Res. 511, 191-206. (2) Inami, K., Okazawa, M., and Mochizuki, M. (2009) Mutagenicity of aromatic amines and amides with chemical models for cytochrome P450 in Ames assay. Toxiol. In Vitro 23, 986-991. (3) Josephy, P. D., and Mannervik, B. (2006) Molecular Toxicology, 2nd Edition, Oxford University Press, New York, USA. (4) McCarren, P., Springer, C., and Whitehead, L. J. (2011) An investigation into pharmaceutically relevant mutagenicity data and the influence on Ames predictive potential. Cheminformatics 3:51. (5) Bentzien, J., Hickey, E. R., Kemper, R. A., Brewer, M. L., Dyekjær, J. D., East, S. P., and Whittaker, M. (2010) Predicting Ames


12

Page 13 of 14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


activities of primary aromatic amines by calculating the stabilities of nitrenium ions. J. Chem. Inf. Model. 50, 274-297. (6) Hatch, F. T., Knize, M. G., and Colvin, M. E. (2001) Extended quantitative structure-activity relationships for 80 aromatic and heterocyclic amines: Structural, electronic, and hydropathic factors affecting mutagenic potency. Env. Molec. Mutagenesis 38, 268-291. (7) Turesky, R. J. (2010) Aromatic amines and heterocyclic aromatic amines: From tobacco smoke to food mutagens. In The Chemical Biology of DNA Damage (Geacintov, N. E., and Broyde S., Eds.) pp 157-183, WILEY-VCH, Weinheim, Germany. (8) Cho, B. (2010) Structure-function characteristics of aromatic amine-DNA adducts. In The Chemical Biology of DNA Damage (Geacintov, N. E., and Broyde S., Eds.) pp 217-238, WILEY-VCH, Weinheim, Germany. (9) Skipper, P. L., Kim, M. Y., Sun, H.-L. P., Wogan, G. N., and Tannenbaum, S. R. (2010) Monocyclic aromatic amines as potential human carcinogens: Old is new again. Carcinogenesis 31, 50-58. (10) Shamovsky, I., Ripa, L., Blomberg, N., Eriksson, L. A., Hansen, P., Mee, C., Tyrchan, C., O’Donovan, M., and Sjö, P. (2012) Theoretical studies of chemical reactivity of metabolically activated forms of aromatic amines toward DNA. Chem. Res. Toxicol. 25, 2236-2252. (11) Bren, U., Zupan, M., Guengerich, F. P., and Marvi, J. (2006) Chemical reactivity as a tool to study carcinogenicity: Reaction between chloroethylene oxide and guanine. J. Org. Chem. 71, 4078-4084. (12) Bren, U., Guengerich, F. P., and Marvi, J. (2007) Guanine alkylation by the potent carcinogen aflatoxin B1: Quantum chemical calculations. Chem. Res. Toxicol.20, 1134-1140. (13) Galeša, K., Bren, U., Kranjc, A., and Marvi, J. (2008) Carcinogenicity of acrylamide: A computational study. J. Agric. Food Chem. 56, 8720-8727. (14) Ford, G. P., and Herman, P. S. (1992) Relative stabilities of nitrenium ions derived from polycyclic aromatic amines. Relationship to mutagenicity. Chem. Biol. Interact. 81, 1-18. (15) Ford, G. P., and Griffin, G. R. (1992) Relative stabilities of nitrenium ions derived from heterocyclic amine food carcinogens: relationship to mutagenicity. Chem. Biol. Interact. 81, 19-33. (16) Borosky, G. L. (2007) Ultimate carcinogenic metabolites from aromatic and heterocyclic aromatic amines: A computational study in relation to their mutagenic potency. Chem. Res. Toxicol. 20, 171-180. (17) Borosky, G. L. (2013) Quantum-chemical studies on mutagenicity of aromatic and heteroaromatic amines. Frontiers Biosci., Scholar Edition 5, 600-610. (18) Glende, C., Klein, M., Schmitt, H., Erdinger, L., and Boche, G. (2002) Transformation of mutagenic aromatic amines into non-mutagenic species by alkyl substituents. Part II: alkylation far away from the amino function. Mutat. Res. 515, 15-38. (19) Shamovsky, I., Ripa, L., Börjesson, L., Mee, C., Nordén, B., Hansen, P., Hasselgren, C., O’Donovan, M., and Sjö, P. (2011) Explanation for main features of structure-genotoxicity relationships of aromatic amines by theoretical studies of their activation pathways in CYP1A2. J. Am. Chem. Soc. 133, 16168-16185. (20) Hatch, F. T., Knize, M. G., and Felton, J. S. (1991) Quantitative structure-activity relationships of heterocyclic amine mutagens formed during the cooking of food. Env. Molec. Mutagenesis 17, 4-19. (21) Felton, J. S., Knize, M. G., Wu, R. W., Colvin, M. E., Hatch, F. T., and Malfatti, M. A. (2007) Mutagenic potency of food-derived heterocyclic amines. Mutat. Res. 616, 90-94. (22) Restrepo, G., Basak, S. C., and Mills, D. (2011) Comparison of SARs and characterization of structural basis of bioactivity using partial order theory and formal concept analysis: a case study with mutagenicity. Curr. Comput.-Aided Drug Des. 7, 109-121. (23) Vance, W. A., Wang, Y. Y., and Okamoto, H. S. (1987) Distributed amino-, nitroso-, and nitrofluorenes: A physicochemical basis for structure-activity relationships in salmonella typhimurium. Environ. Mutagenesis 9, 123-141. (24) Lau, E. Y., Felton, J. S., and Lightstone, F. C. (2006) Insights into the O-acetylation reaction of hydroxylated heterocyclic amines by human arylamine N-acetyltransferases: A computational study. Chem. Res. Toxicol. 19, 1182-1190.

(25) Chao, M.-W., Kim, M. Y., Ye, W., Ge, J., Trudel, L. J., Belanger, C. L., Skipper, P. L., Engelward, B. P., Tannenbaum, S. R., and Wogan, G. N. (2012) Genotoxicity of 2,6- and 3,5-dimethylaniline in cultured mammalian cells: The role of reactive oxygen species. Toxicol. Sci. 130, 48-59. (26) Ma, H., Wang, J., Abdel-Rahman, S. Z., Boor, P. J., and Khan, M. F. (2008) Oxidative DNA damage and its repair in rat spleen following subchronic aniline exposure. Toxicol. Appl. Pharmacol. 233, 247-253. (27) Floreani, M., and Carpenedo, F. (1990) Inhibition of rat liver monooxygenase activities by 2-methyl-1,4-naphthoquinone (menadione). Toxicol. Appl. Pharmacol. 105, 333-339. (28) Brennan, R. J., and Schiestl, R. H. (1997) Aniline and its metabolites generate free radicals in yeast. Mutagenesis 12, 215-220. (29) Brennan, R. J., and Schiestl, R. H. (1999) The aromatic amine carcinogens o-toluidine and o-anisidine induce free radicals and intrachromosomal recombination in Saccharomyces cereviciae. Mutat. Res. 430, 3745. (30) Martínez, A., Urios, A., and Blanco, M. (2000) Mutagenicity of 80 chemicals in Escherichia coli tester strains IC203, deficient in OxyR, and its oxyR+ parent WP2 uvrA/pKM101: detection of 31 oxidative mutagens. Mutat. Res. 467, 41-53. (31) Ben Mansour, H., Barillier, D., Corroler, D., Ghedira, K., GhekirGhedira, L., and Mosrati, R. (2009) In vitro study of DNA damage induced by acid orange 52 and its biodegradation derivatives. Environ. Toxicol. Chem. 28, 489-495. (32) Famulok, M., and Boche, G. (1989) Formation of N(deoxyguanosin-8-yl)aniline in the in vitro reaction of N-acetoxyaniline with deoxyguanosine and DNA. Angew. Chem. Int. Ed. 28, 468−469. (33) Anari, M. R., Josephy, P. D., Henry, T., and O’Brien, P. J. (1997) Hydrogen peroxide supports human and rat cytochrome P450 1A2catalyzed 2-amino-3-methylimidazo[4,5-f]quinoline bioactivation to mutagenic metabolites: Signiﬁcance of cytochrome P450 peroxygenase. Chem. Res. Toxicol. 10, 582-588. (34) Ortiz de Montellano, P. R. (2010) Hydrocarbon hydroxylation by cytochrome P450 enzymes. Chem. Rev. 110, 932-948. (35) Shaik, S., Cohen, S., Wang, Y., Chen, H., Kumar, D., and Thiel, W. (2010) P450 enzymes: Their structure, reactivity, and selectivity – modelled by QM/MM calculations. Chem. Rev. 110, 949-1017. (36) Ji, L., and Schüürmann, G. (2013) Model and mechanism: Nhydroxylation of primary aromatic amines by cytochrome P450. Angew. Chem. Int. Ed. 52, 744-748. (37) Mayer, J. M. (1998) Hydrogen atom abstraction by metal-oxo complexes: Understanding the analogy with organic radical reactions. Acc.Chem. Res. 31, 441-450. (38) Shaik, S., Kumar, D., and de Visser, S. P. (2008) A valence bond modeling of trends in hydrogen abstraction barriers and transition states of hydroxylation reactions catalyzed by cytochrome P450 enzymes. J. Am. Chem. Soc. 130, 10128-10140. (39) de Visser, S. P. (2010) Trends in substrate hydroxylation reactions by heme and nonheme iron(IV)-oxo oxidants give correlations between intrinsic properties of the oxidant with barrier height. J. Am. Chem. Soc. 132, 1087-1097. (40) Zhao, Y., and Truhlar, D. G. (2008) Density functionals with broad applicability in chemistry. Acc. Chem. Res. 41, 157-167. (41) Xu, X., Alecu, I. M., and Truhlar, D. G. (2011) How well can

modern density functionals predict intermolecular distances at transition state. J. Chem. Theory Comput. 7, 1667-1676. (42) Newcomb, M., Hollenberg, P. F, and Coon, M. J. (2003) Multiple mechanisms and multiple oxidants in P450-catalyzed hydroxylations. Arch. Biochem. Biophys. 409, 72-79. (43) Cundari, T. R., and Stevens, W. J. (1993) Effective core potential methods for the lanthanides. J. Chem. Phys. 98, 5555-5565. (44) Shaik, S., de Visser, S. P., Ogliaro, F., Schwarz, H., and Schröder, D. (2002) Two-state reactivity mechanisms of hydroxylation and epoxidation by cytochrome P-450 revealed by theory. Curr. Opin. Chem. Biol. 6, 556-567. (45) Shaik, S., Kumar, D., de Visser, S. P., Altun, A., and Thiel, W. (2005) Theoretical perspective on the structure and mechanism of cytochrome P450 enzymes. Chem. Rev. 105, 2279-2328.


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(46) Sansen, S., Yano, J. K., Reynald, R. L., Schoch, G. A., Griffin, K. J., Stout, C. D., and Johnson, E. F. (2007) Adaptations for the oxidation of polycyclic aromatic hydrocarbons exhibited by the structure of human P450 1A2. J. Biol. Chem. 19, 14348-14355. (47) Bofill, J. M. (1994) Updated Hessian matrix and the restricted step method for locating transition structures. J. Comput. Chem. 15, 1-11. (48) Hay, P. J., and Wadt, W. R. (1985) Ab initio effective core potentials for molecular calculations. Potentials for potassium to gold including the outermost core orbitals. J. Phys. Chem. 82, 299-310. (49) He, Y., Gräfenstein, J., and Kraka, E., Cremer, D. (2000) What correlation effects are covered by density functional theory? Molec. Phys. 98, 1639-1658. (50) Green, M. T. (1999) Evidence for sulfur-based radicals in thiolate Compound I intermediates. J. Am. Chem. Soc. 121, 7939-7940. (51) Ogliaro, F., de Visser, S. P., Groves, J. T., and Shaik, S. (2001) Chameleon states: High-valent metal-oxo species of cytochrome P450 and its Ruthenium analogue. Angew. Chem. Int. Ed. 50, 2874-2878. (52) Radoń, M., and Broclawik, E. (2007) Peculiarities of the electronic structure of cytochrome P450 compound I: CASPT2 and DFT modeling. J. Chem. Theory Comput. 3, 728-734. (53) Becke, A. D. (1993) Density-functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 98, 5648-5652. (54) Andersson, K., Malmqvist, P.-Å., and Roos, B. O. (1992) 2Ndorder perturbation-theory with a complete active space self-consistent field reference function. J. Phys. Chem. 96, 1218-1226. (55) Goldfeld, D. A., Bochevarov, A. D., and Friesner, R. A. (2008) Localized orbital corrections applied to thermochemical errors in density functional theory: The role of basis set and application to molecular reactions. J. Chem. Phys. 129, 214105/1-214105/13. (56) Hughes, T. F., and Friesner, R. A. (2011) Correcting systematic errors in DFT spin-splitting energetics for transition metal complexes. J. Chem. Theory Comput. 7, 19-32. (57) Hughes, T. F., Harvey, J. N., and Friesner, R. A. (2012) A B3LYPDBLOC empirical correction scheme for ligand removal enthalpies of transition metal complexes: parameterization against experimental and CCSD(T)-F12 heats of formation. Phys. Chem. Chem. Phys. 14, 77247738. (58) Gaussian 09, Revision D.01. Frisch, M. J., Trucks, G. W., Schlegel, H. B., Scuseria, G. E., Robb, M. A., Cheeseman, J. R., Scalmani, G., Barone, V., Mennucci, B., Petersson, G. A., Nakatsuji, H., Caricato, M., Li, X., Hratchian, H. P., Izmaylov, A. F., Bloino, J., Zheng, G., Sonnenberg, J. L., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., T. Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Vreven, T., Montgomery, Jr., J. A., Peralta, J. E., Ogliaro, F., Bearpark, M., Heyd, J. J., Brothers, E., Kudin, K. N., Staroverov, V. N., Keith, T., Kobayashi, R., Normand, J., Raghavachari, K., Rendell, A., Burant, J. C., Iyengar, S. S., Tomasi, J., Cossi, M., Rega, N., Millam, J. M., Klene, M., Knox, J. E., Cross, J. B., Bakken, V., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R. E., Yazyev, O., Austin, A. J., Cammi, R., Pomelli, C., Ochterski, J. W., Martin, R. L., Morokuma, K., Zakrzewski, V. G., Voth, G. A., Salvador, P., Dannenberg, J. J., Dapprich, S., Daniels, A. D., Farkas, O., Foresman, J. B., Ortiz, J. V., Cioslowski, J., and Fox, D. J. Gaussian, Inc., Wallingford CT, 2013. (59) Bochevarov, A. D., Harder, E., Hughes, T. F., Greenwood, J. R., Braden, D. A., Philipp, D. M., Rinaldo, D., Halls, M. D., Zhang, J., and Friesner, R. A. (2013) Jaguar: A high-performance quantum chemistry software program with strengths in life and materials sciences. Int. J. Quant. Chem. 113, 2110-2142. (60) Afzelius, L., Masimirembwa, C. M., Karlén, A., Andersson, T. B., and Zamora, I. (2002) Discriminant and quantitative PLS analysis of competitive CYP2C9 inhibitors versus noninhibitors using alignment independent GRIND descriptors. J. Comput.-Aided Mol. Des. 16, 443– 458. (61) Klicić, J. J., Friesner, R. A., Liu, S.-Y., and Gilda, W. C. (2002) Accurate prediction of acidity constants in aqueous solution via Density Functional Theory as self-consistent reaction field methods. J. Phys. Chem. A 106, 1327-1335. (62) Masumura, K.-I., Totsuka, Y., Wakabayashi, K., and Nohmi, T. (2003) Potent genotoxicity of aminophenylnorharman, formed from non-

Page 14 of 14

mutagenic norharman and aniline, in the liver or gpt delta transgenic mouse. Carcinogenesis 24, 1985-1993. (63) Ohnishi, S., Murata, M., Oikawa, S., Totsuka, Y., Takamura, T., Wakabayashi, K., and Kawanishi, S. (2001) Oxidative DNA damage by an N-hydroxy metabolite of the mutagenic compound formed from norharman and aniline. Mutat. Res. 494, 63−72. (64) Nauwelaers, G., Bessette, E. E., Gu, D., Tang, Y.-J., Rageul, J., Fessard, V., Yuan, J.-M., Yu, M. C., Langouet, S., and Turesky, R. J. (2011) DNA adduct formation of 4-aminobiphenyl and heterocyclic aromatic amines in human hepatocytes. Chem,. Res. Toxicol. 24, 913-925. (65) Kim, D., and Guengerich, F. P. (2005) Cytochrome P450 activation of arylamines and heterocyclic amines. Annu. Rev. Pharmacol. Toxicol. 45, 27-49. (66) Walsh, A., Szklarz, G. D., and Scott, E. E. (2013) Human cytochrome P450 1A1 structure and utility in understanding drug and xenobiotic metabolism. J. Biol. Chem. 288, 12932-12943. (67) Weinhold, F., and Landis, C. R. (2005) Valency and Bonding: A Natural Bond Orbital Donor-acceptor Perspective, Cambridge University Press. (68) Isobe, H., Yamaguchi, K., Okumura, M., and Shimada, J. (2012) Role of perferryl-oxo oxidant in alkane hydroxylation catalyzed by cytochrome P450: A hybrid density functional study. J. Phys. Chem. B 116, 4713-4730. (69) Shaik, S., Hirao, H., and Kumar, D. (2007) Reactivity of highvalent iron-oxo species in enzymes and synthetic reagents: a tale of many states. Acc. Chem. Res. 40, 532-542. (70) Usharani, D., Lacy, D. C., Borovik, A. S., and Shaik, S. (2013) Dichotomous hydrogen atom transfer vs proton-coupled electron transfer during activation of X-H bonds (X = C, N, O) by nonheme iron-oxo complexes of variable basicity. J. Am. Chem. Soc. 135, 17090-17104. (71) Coon, M. J., Vaz, A. D. N., McGinnity, D. F., and Peng, H.-M. (1998) Multiple activated oxygen species in P450 catalysis. Contribution to specificity in drug metabolism. Drug Metab.Disp. 26, 1190-1193. (72) de Visser, S. P., and Tan, L. S. (2010) Is the bound substrate in Nitric Oxide Synthase protonated or neutral and what is the active oxidant that performs substrate hydroxylation? J. Am. Chem. Soc. 130, 1296112974. (73) Altun, A., Kumar, D., Neese, F., and Thiel, W. (2008) Multireference ab initio quantum mechanics/molecular mechanics study on intermediates in the catalytic cycle of cytochrome P450cam. J. Phys. Chem. B 112, 12904-12910. (74) Harris, D. L., and Loew, D. H. (1998) Theoretical investigation of the proton assisted pathway to formation of cytochrome P450 compound I. J. Am. Chem. Soc. 120, 8941-8948. (75) Vancoillie, S., Zhao, H., Radoń, M., and Pierloot, K. (2010) Performance of CASPT2 and DFT for relative spin-state energetic of heme models. J. Chem. Theory Comput. 6, 576-582. (76) Lipping, L., Kütt, A., Kaupmees, K., Koppel, I., Burk, P., Leito, I., and Koppel, I. A. (2011) Acidity of anilines: Calculations vs experiment. J. Phys. Chem. A 115, 10335-10344. (77) Möller, P., Wallin, H., Vogel, U., Autrup, H., Risom, L., Hald, M. T., Daneshvar, B., Dragsted, L. O., Poulsen, H. E., and Loft, S. (2002) Mutagencity of 2-amino-3methylimidazo[4,5-f]quinoline in colon and liver of Big Blue rats: Role of DNA adducts, strand breaks, DNA repair and oxidative stress. Carcinogenesis 23, 1379-1385. (78) Makena, P., and Chung, K.-T. (2007) Evidence that 4aminobiphenyl, benzidine, and benzidine congeners produce genotoxicity through reactive oxygen species. Environ. Molec. Mutagenesis 48, 404413. (79) Siraki, A. G., Chan, T. S., Galati, G., Teng, S., and O’Brien, P. J. (2002) N-oxidation of aromatic amines by intracellular oxidases. Drug Metab. Rev. 34, 549-564.


14

Mechanism of the N-hydroxylation of primary and secondary amines by cytochrome P450.

Caffeine Cytochrome P450 1A2 Metabolic Phenotype Does Not Predict the Metabolism of Heterocyclic Aromatic Amines in Humans.

Theoretical study on the N-demethylation mechanism of theobromine catalyzed by P450 isoenzyme 1A2.

Bioactivation of Aromatic Amines by Human CYP2W1, An Orphan Cytochrome P450 Enzyme.

Modification of cytochrome P450 1A2 enzymes by the mechanism-based inactivator 2-ethynylnaphthalene and the photoaffinity label 4-azidobiphenyl.

Genetic polymorphisms of cytochrome P450-1A2 (CYP1A2) among Emiratis.

Co-expression of active human cytochrome P450 1A2 and cytochrome P450 reductase on the cell surface of Escherichia coli.

Functional Significance of Cytochrome P450 1A2 Allelic Variants, P450 1A28, 15, and *16 (R456H, P42R, and R377Q).

Cytochrome P450 1A2 Metabolizes 17β-Estradiol to Suppress Hepatocellular Carcinoma.

Mechanism-based inactivation of cytochrome P450 2B6 by isoimperatorin.

Mechanism-Based Inactivation of Human Cytochrome P450 2B6 by Chlorpyrifos.

Influence of cytochrome P450 mixed-function oxidase induction on the acute toxicity to rainbow trout (Salmo gairdneri) of primary aromatic amines.

Investigation of the major cytochrome P450 1A2 genetic variant in a healthy Tibetan population in China.

Lipid molecules can induce an opening of membrane-facing tunnels in cytochrome P450 1A2.

Linear Interaction Energy Based Prediction of Cytochrome P450 1A2 Binding Affinities with Reliability Estimation.

Inhibitory effects of celastrol on rat liver cytochrome P450 1A2, 2C11, 2D6, 2E1 and 3A2 activity.

Inhibitory mechanisms of celastrol on human liver cytochrome P450 1A2, 2C19, 2D6, 2E1 and 3A4.

In vitro effect of important herbal active constituents on human cytochrome P450 1A2 (CYP1A2) activity.

Caffeine Concentration Ratios in Hair: An Alternative for Plasma-Based Phenotyping of Cytochrome P450 1A2?

High-level expression of functional human cytochrome P450 1A2 in Escherichia coli.

Effective catalysis of imine metathesis by means of fast transiminations between aromatic-aromatic or aromatic-aliphatic amines.

Formation of nitric oxide by cytochrome P450-catalyzed oxidation of aromatic amidoximes.

Use of P450 cytochrome inhibitors in studies of enokipodin biosynthesis.

Cytochrome P450 1B1 and Primary Congenital Glaucoma.