Quantum chemical exploration on the metabolic mechanisms of caffeine by flavin-containing monooxygenase

Tetrahedron 72 (2016) 2858e2867

Contents lists available at ScienceDirect

Tetrahedron journal homepage: www.elsevier.com/locate/tet

Quantum chemical exploration on the metabolic mechanisms of caffeine by flavin-containing monooxygenase Yuan Kang a, y, Jing Tao a, y, Zhiyu Xue a, Yan Zhang a, Zeqin Chen a, *, Ying Xue b a

College of Chemistry and Chemical Engineering, Chemical Synthesis and Pollution Control Key Laboratory of Sichuan Province, China West Normal University, Nanchong 637002, PR China b College of Chemistry, Key Laboratory of Green Chemistry and Technology in Ministry of Education, Sichuan University, Chengdu 610064, PR China

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 December 2015 Received in revised form 17 March 2016 Accepted 28 March 2016 Available online 4 April 2016

Caffeine, a ubiquitous natural product, is widely consumed by humans. The metabolic mechanisms of caffeine by flavin-containing monooxygenase (FMO) were systematically investigated in this study by quantum mechanics calculations. Four main metabolic pathways were characterized, including N-demethylations at N1-, N3-, and N7- sites (paths IeIII) and C-8 oxidation (path IV). N-demethylation proceeds via the concerted homolytic cleavages of CeH and OeO bonds, while C-8 oxidation is an oxygen atom transfer mechanism. It shows that C-8 oxidation predominates over N-demethylations and trimethyluric acid is therefore the optimum metabolite of caffeine by FMO. Additionally, N3-demethylation is more favorable than N1 and N7-demethylations. This study can offer important clues for the biodecaffeination techniques. Ó 2016 Elsevier Ltd. All rights reserved.

Keywords: Flavin-containing monooxygenase Caffeine N-demethylation C-8 oxidation Homolysis

1. Introduction Caffeine, known as 1,3,7-trimethylxanthine (Scheme 1), is a commercially important purine alkaloid existing mainly in the seeds, leaves or fruits of some plants.1,2 It is endowed as the world’s most popular psychoactive substance because of its activity at adenosine receptors,3,4 which makes it widely used as a respiratory stimulant, a diuretic in pharmaceuticals, and an analgesic enhancer in headache, etc.5e11 However, heavy coffee consumption may increase the risk of bladder cancer.12 The implications of caffeine on human health have caused great scientific interests in its biodegradation.13,14 Metabolized by biological oxidation, caffeine in humans can produce at least 12 metabolites which have been detected in the urine.15 The four important primary metabolites of caffeine are theophylline (TP, 1,3-dimethylxanthine), paraxanthine (PX, 1,7dimethylxanthine) and theobromine (TB, 3,7-dimethylxanthine), formed by N-demethylation at each of the three tertiary amine nitrogen atoms, as well as 1,3,7-trimethyluric acid (TMU) formed by C-8 oxidation.16,17 Bioconversion of caffeine into its metabolites is an enzymatic process and occurs primarily in the liver.18 Two main types of enzymes presented in the liver microsomes of human can

* Corresponding author. Tel.: þ86 817 2568081; e-mail address: chenzeqin@ cwnu.edu.cn (Z. Chen). y These authors contributed equally to this work. http://dx.doi.org/10.1016/j.tet.2016.03.091 0040-4020/Ó 2016 Elsevier Ltd. All rights reserved.

potentially catalyze the metabolism of caffeine, namely, flavincontaining monooxygenase (FMO) and cytochrome P450 (CYP).19,20 Both enzymes are monooxygenases and involve in the oxygenation of a wide range of heteroatom-containing compounds, by which the lipophilic compounds are converted into more hydrophilic metabolites for rapid excretion.21 The mechanism by which each monooxygenase operates, however, is quite distinct.22e24 Drug toxicity thus far observed in the clinic is mainly the result of CYP-dependent oxidation because CYP can also oxidize compounds to electrophilic reactive metabolites that can have significant consequences for toxicity. Our previous study has systematically explored the metabolic mechanisms of caffeine by CYP.25 In contrast, FMO generally converts lipophilic nucleophiles into more polar, readily excreted and harmless metabolites and is rarely inhibited. Nevertheless, the oxidation metabolic mechanisms of caffeine by FMO have remained elusive up to now yet. FMO is a class of monooxygenases capable of oxygenating nucleophilic oxygen, nitrogen, halide, selenium, and phosphorous atoms of a wide range of substrates, such as amines, amides, thiols, and sulfides.26e30 As the oxygenant of FMO, the electrophilic C-4aflavinhydroperoxide (FLHOOH, Scheme 1) intermediate is a covalent adduct between the C-4a-carbon of flavin and dioxygen molecule,31 which is a short-lived (t1/2¼2.5 ms) intermediate oxidant and can break down into oxidized flavin (Flox) and HOOH rapidly in the absence of the stabilizing influence.32,33 In the past years, some theoretical attempts have been performed to identify

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

2859

Scheme 1. Metabolic mechanisms of caffeine catalyzed by FMO.

the oxidation mechanism of FMO.34e41 Canepa and his co-workers identified the oxidation mechanism of dimethyl sulfide using a series of bicyclic and tricyclic models of FLHOOH and proposed a SN2like attack mechanism.34 Subsequently, Bach and his co-workers confirmed the SN2-like attack mechanism of the heteroatom on the distal oxygen of the hydroperoxide based on the investigation on the oxidation of N-, S-, P-, Se-containing nucleophiles catalyzed by FMO.35 More recently, Bach formulated another new mechanism for FMO oxidation of N- and S-nucleophiles, namely, a concerted emolytic OeO bond cleavage in concert with hydroxyl radical transfer from the flavin hydroperoxide.36,37 Additionally, the oxidation mechanisms of p-hydroxybenzoate (p-OHB) and its derivatives mediated by p-hydroxybenzoate hydroxylase (PHBH) were also characterized based on the FLHOOH model. Three possible mechanisms have been indicated, including a OH radical transfer mechanism, an electrophilic aromatic substitution mechanism, and an oxygen atom transfer reaction with intramolecular 1,2-proton transfer.38e41 This work aims at uncovering the oxidation metabolic mechanisms of caffeine by FMO. The obtained results show that Ndemethylation proceeds via the concerted homolytic cleavages of CeH and OeO bonds, while C-8 oxidation proceeds via an oxygen atom transfer mechanism. C-8 oxidation is more favorable than Ndemethylations and trimethyluric acid therefore is the optimum metabolite of caffeine by FMO. To date, this is the first time to report the involvement of FMO in the metabolism of caffeine at the theoretical level, which can offer important understandings for the bio-decaffeination techniques. 2. Computational details The computational reaction model adopted in this work consists of the two parts: (a) caffeine and (b) the reactive FLHOOH of FMO enzyme which comprises three segments, the tricyclic isoalloxazine moiety, the C-4a-hydroperoxide functionality, and the bhydroxyethyl group to model the effect of the 20 -OH group of the ribityl side chain of native FADHOOH (Scheme 1).42 The coordinate of FLHOOH was established from the crystallographic structure of FMO enzyme (PDB code: 2GVC).43 Standard procedures within the Gaussview program were used to incorporate the coordinate of

caffeine to build the suitable initial structure to search for transition state geometries at the DFT level. All the quantum chemistry calculations were performed using the Gaussian 09 program.44 Geometric structures for all the stationary points, including the reactant complexes, product complexes, intermediates, and transition states, were optimized in the gas phase using the dispersion corrected hybrid functional of B3LYP-D345,46 in conjunction with the standard 6-31þG(d) basis set. Vibrational frequency calculations were performed to confirm the stationary point as a minimum with all positive frequencies or as a transition state with only one imaginary frequency. The connectivity between the stationary points was established by intrinsic reaction coordinate (IRC) calculations.47,48 Natural Population Atomic (NPA) charges were determined with the Natural Bond Order (NBO) theory.49 The binding energies of caffeine and FLHOOH were corrected using basis set superposition error. Single-point calculations were performed both in the gas phase and in the protein environment using B3LYP-D3 functional with a higher basis set, 6-311þþG(d,p), which has be proved to have consistent results with B3PW91-D3 and PBE1PBE-D3 functionals.50 The weak polarization effect of a protein environment was modeled using conductor-like polarizable continuum model (CPCM)51 with dielectric constant of ε¼5.62 (chlorobenzene). The value (ε¼5.62) was taken as a reasonable compromise for the enzyme active site.25,52,53 All of the single-point energies were corrected by the gas-phase thermodynamic quantities. The thermodynamic data reported in this paper are at 298.15 K and 1 atm. Cartesian coordinates for the optimized geometries and energies of all stationary points along the potential energy profiles are provided in the Supplementary data document. 3. Results and discussion Mediated by FMO, the degradation of caffeine can take place along two classes of mechanisms: demethylation and oxidation, which were divided into four reaction pathways (Scheme 1). Paths IeIII involve the demethylation of caffeine at N1-, N3-, and N7- sites, yielding theophylline, paraxanthine, and theobromine, respectively, whereas path IV concerns the oxidation of caffeine at C8

2860

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

site, yielding 1,3,7-trimethyluric acid. Each type of mechanisms is discussed in detail below. 3.1. N-demethylation (paths IeIII) As depicted in Scheme 1a, the overall N-demethylation process of caffeine occurs in two discrete stages: (i) OeO bond homolysis is accompanied by H-abstraction from a N-methyl group to produce a C-centered radical and a water molecule, which subsequently transfers an H-atom back to the proximal oxygen. Concomitantly, the developing HO fragment is transferred to the newly formed carbon centered radical to produce carbinolamine in a similar

manner like P450 hydroxylation.54 (ii) the carbinolamine formed decomposes to yield the final dimethylxanthine species metabolite and formaldehyde. 3.1.1. N-methyl hydroxylation 3.1.1.1. Geometrical features. The optimized geometries of all the stationary points for N-methyl hydroxylation of caffeine by FLHOOH are shown in Fig. 1. The single imaginary frequencies of ITS1, II-TS1 and III-TS1 (TS1 species) are 316.2, 343.3, and 239.8 i cm1, respectively. It indicates that heavy oxygen atom motion is comprised in the reaction vector, since, in principle, the magnitude of the imaginary frequency is quite high to exceed

Fig. 1. B3LYP/6-31þG(d) optimized structures ( A) along the potential energy surfaces of N-methyl hydroxylations of caffeine (paths IeIII) catalyzed by FMO. (vi represents for the single imaginary frequency of the transition state. Atom and symbol definitions are the same in the following Figs. 3e5.).

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

1000 i cm1 if atom motion in the transition state is largely comprised of light hydrogen atom migration.35 Animation of the single imaginary frequencies of TS1 species shows largely OeO bond stretching of FLHOOH with the apparent transfer of methyl hydrogen Hm atom to the HO fragment. The distance between the distal oxygen Od and proximal oxygen Op atoms (OdeOp) is elongated by 0.653  A for I-TS1, 0.651  A for II-TS1, and 0.706  A for III-TS1 when compared with their corresponding reactant complexes. The internal OpeOdeHd bond angle of 87.1 in TS1 species allows for a modest hydrogen bond of the HO radical to Op atom (2.273e2.327  A). Thus, in each transition state, there is essentially a bound HO radical suspended between Op atom (rOeO¼2.096e2.156  A) and Hm atom (rOeH¼1.078e1.107  A). This  transferring HO radical is also stabilized by its H-bonding interaction with the adjacent C]O group (1.863e1.867  A). Owning to these available alternative sources of intramolecular stabilization, we can not observe a complete somersault rearrangement with a nearly linear Op/Hd/Od structure as shown in the literature.36,55 Besides, the departing Hm atom from the N-methyl group resides between the developing HO radical and the methyl carbon Cm atom (rCeH¼1.430e1.484  A) of methylene. An almost collinear arrangement of the Cm/Hm/Od atoms is displayed in each transition state with the bond angle ranging from 161.0 to 171.1. When the barriers are crossed, the FLHO radicals formed in the transition states are very reactive and readily abstract a H-atom from the water molecule formed late on the reaction coordinate to produce the OH group of FLHOH. The developing HO radical is then rebounded to the newly formed carbon centered radical to produce carbinolamine. In the carbinolamine intermediates, I-2, II-2, and III2, the methyl groups bonded with the tertiary amine nitrogen atoms are hydroxylated and the developing OH group at methyl carbon Cm atom forms strong H-bonding interactions with C]O (1.822e2.063  A) and OH (1.855e1.925  A) groups on FLHOH. The profiles of the changes of bond lengths and electronic energies along the IRCs of TS1 species are depicted in Fig. 2. The IRC results exhibit similar geometrical variations for N-methyl

2861

hydroxylation processes. As shown in Fig. 2aec, the OdeOp and CmeHm bond cleavages occur simultaneously at the coordinate of ca. 1 amu1/2 bohr, previous to the formation of TS1 species. And then, with the continuous breakages of OdeOp and CmeHm bonds, the OdeHm, CmeOd and OpeHm distances keep decreasing. When TS1 species is reached, interestingly, a pronounced water molecule is formed due to the partial formation of OdeHm bond. At the coordinate of ca. 1 amu1/2 bohr after crossing TS1 species, an obvious water molecule is formed wherein a stable single bond is created between Od and Hm atoms. The formed water molecule can survive a long time up to the coordinate of ca. 6 amu1/2 bohr. During this period, the water keeps changing its orientation to facilitate the subsequent proton (Hm) transfer to Op atom and nucleophilic addition of HO fragment at Cm atom for structural stability, which results in the breakage of OdeHm bond and the formation of CmeOd and OpeHm bonds. The CmeOd bond-formation is slightly more advanced than OpeHm bond. 3.1.1.2. Electronic structural features. The preference of Nmethyl hydroxylation can be further analyzed by the NPA charge distribution (Table S1). NBO analysis of reactant complexes shows that the net charge on FLHOOH is slightly negative (0.027w0.030e), while it is correspondingly positive (0.027e0.030e) on caffeine, suggesting a little charge separation in the pre-reaction complexes. In TS1 species, the net charge on the transferring HO fragment (0.255w0.296e) is greatly less negative than OH (1.0e). The slight negative charge on HO radical fragment instead of the familiar zero mainly results from the pronounced characteristic of the water molecule in TS1 species, where OdeHm bond is partial formed and the transferred Hm atom is positively charged rather than zero. This observation indicates the viewpoint of the homolytic cleavage of OeO bond. Obviously, the CA fragment, the remainder of caffeine by removing the methyl hydrogen Hm atom, is positively charged (0.223e0.240e) in TS1 species, while the net charge on FLHO fragment is negative (0.328w0.370e), implying a clear charge separation due to the

Fig. 2. Profiles of electronic energies (pink dotted lines) and the changes of bond lengths (solid lines) along the IRCs of I-TS1 (a), II-TS1 (b), III-TS1 (c), and IV-TS1 (d) at the B3LYP/631þG(d) level.

2862

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

migration of electron from caffeine to FLHOOH. This discovery fits in well with the orbital characteristic of TS1 species (Fig. S1). As depicted in Fig. S1, the highest occupied molecular orbitals (HOMOs) of TS1 species reside mainly on the FLHO fragment while the lowest unoccupied molecular orbitals (LUMOs) delocalize on the CA fragment and the formed water molecule. Due to the charge separation, TS1 species possesses polarity characteristic that reflects in the large dipole moments (11e15 Debye) in Fig. S1. Consequently, environment may exert important effect on the stability of TS1 species. 3.1.1.3. Energy profiles. The changes in electronic energies (DE) for N-methyl hydroxylation of caffeine mediated by FMO are presented in Table 1. As shown in Table 1, the gas-phase activation energies for N1-, N3-, and N7-methyl hydroxylations are 40.4, 37.3 and 39.6 kcal mol1, respectively, which correspond to 33.7, 31.2 and 33.6 kcal mol1 when chlorobenzene solution is incorporated. The dramatic reduction of activation energies (ca. 6 kcal mol1) in chlorobenzene solution can be well explained by the polar characters of TS1 species as shown above. For each N-methyl hydroxylation process, a larger dipole moment is possessed by TS1 species with respect to their corresponding RC species (Table 1). It is expected, therefore, that the chlorobenzene solution can exert a remarkable effect on the stability of TS1 species due to the strong electrostatic interactions with the solvent, which inevitably cause the reduction of the activation energies in chlorobenzene solution. All the N-methyl hydroxylation processes are largely exothermic with reaction energies of 57e62 kcal mol1 both in the gas phase and in chlorobenzene solution.

Table 1 Changes of electronic energy (kcal mol1) in the gas phase (DEgas) and chlorobenzene solution (DEsol) and dipole moments (m, Debye) for the first step of paths IeIV at the B3LYP-D3/6-311þþG(d,p) level

m Path I I-1 I-TS1 I-2 Path III III-1 III-TS1 III-2

DEgas

DEsol

4.38 11.25 2.53

0.0 40.4 61.8

0.0 33.7 61.1

9.52 11.43 3.05

0.0 39.6 57.6

0.0 33.6 58.0

Path II II-1 II-TS1 II-2 Path IV IV-1 IV-TS1 IV-2

m

DEgas

DEsol

6.16 14.22 9.11

0.0 37.3 59.0

0.0 31.2 59.6

8.11 9.19 8.16

0.0 28.5 42.2

0.0 25.6 41.0

3.1.2. Carbinolamine decomposition. By N1-, N3-, and N7-methyl hydroxylations, three types of carbinolamines are formed, including carbinol theobromine (I-2, path I), carbinol paraxanthine (II-2, path II) and carbinol theophylline (III-2, path III). The second stage along N-demethylation mechanism involves the decomposition of carbinolamine via proton transfer, which proceeds in the similar manner like the metabolism of caffeine catalyzed by cytochrome P450. Compared with our previous theoretical study of cytochrome P450-catalyzed metabolism of caffeine,25 both the dispersion effect and the polarization effect of protein environment were taken into account in the present study. Especially, more detail pathways were characterized for the decompositions of carbinol paraxanthine and carbinol theophylline. Consequently, this study can afford a complete depiction of the carbinolamine decomposition process and provide energetics comparison. All the possible decomposition mechanisms of carbinolamine species via proton transfer are depicted in Fig. S2, which can be classified into two categories, the direct proton transfer mechanism and the heteroatom-assisted proton transfer mechanism. The direct proton transfer mechanism involves the direct transfer of proton

from hydroxymethyl to the bonded tertiary amine nitrogen atom, yielding the keto metabolite, whereas the heteroatom-assisted proton transfer mechanism concerns the proton transfer from hydroxymethyl to the adjacent heteroatom, yielding the enol metabolite which can convert into the keto structure via further enoleketo tautomerism. The key geometric information and electronic energies of the transition states involved are displayed in Fig. 3 and in further details for potential energy surfaces and stationary geometries in Figs. S3eS6. 3.1.2.1. Direct proton transfer mechanisms. As depicted in Fig. S2, paths IA, IIA and IIIA are the direct proton transfer mechanisms for the decompositions of carbinol theobromine, carbinol paraxanthine and carbinol theophylline, respectively, forming the keto structures of theobromine (I-3A), paraxanthine (II-3A) and theophylline (III-3A). The located transition states I-TS2A, II-TS2A, and III-TS2A are all tetratomic ring structures with the single imaginary frequencies of 420e490 i cm1 (Fig. 3). Animation of the single imaginary frequencies shows the direct motion of hydroxyl hydrogen to nitrogen atom. The gas-phase activation energies of ITS2A, II-TS2A, and III-TS2A are 58.1, 54.1, and 56.0 kcal mol1, respectively, which correspond to 54.6, 51.4, and 53.1 kcal mol1 when chlorobenzene solution is incorporated. It can be readily seen that large activation energies are required for the direct proton transfer processes due to the high constraint of the tetratomic transition state structures. Considering the presence of one water molecule in the active site of FMO,54 one explicit water molecule is added to act as the proton-transfer bridge. Owning to the assistance of one explicit water molecule, the located transition states I-TS2A-H2O, II-TS2AH2O, and III-TS2A-H2O are expanded to hexatomic ring structures with the single imaginary frequencies of 930e1250 i cm1. The quite higher imaginary frequencies arise from the greater force required by the two light proton migration. The gas-phase activation energies of I-TS2A-H2O, II-TS2A-H2O, and III-TS2A-H2O are 44.0, 37.2 and 39.0 kcal mol1, respectively, which decrease to 41.6, 36.6, and 36.8 kcal mol1 when chlorobenzene solution is incorporated. As expected, the water-assisted activation energies are decreased drastically by 13e17 kcal mol1 than the direct mechanism, attributed to the lower constraint of the hexatomic transition state structures, where the energies required for the bond dissociation and formation are smaller than the tetratomic so much that a large amount of deformation energy is saved. This suggests that the presence of the explicit water is thermodynamically and kinetically beneficial to this proton transfer process. Even so, large activation energy is still required for the water-assisted direct proton transfer. 3.1.2.2. Heteroatom-assisted proton transfer mechanisms. Apart from paths IA, IIA and IIIA, the remainder mechanisms belong to the heteroatom-assisted proton transfer mechanisms. Path IB/IC along path I involves the proton transfer from hydroxymethyl of carbinol theobromine to heteroatom O2/O6 atom via transition states I-TS2B/ I-TS2C, yielding the enol isomer of theobromine (I-3B/I-3C). Similarly, path IIB/IIC along path II involves the proton transfer from hydroxymethyl of carbinol paraxanthine to heteroatom O2/N9 atom via transition state II-TS2B/II-TS2C, yielding the enol isomer of paraxanthine (II-3B/II-3C). However, only one heteroatom-assisted proton transfer mechanism (path IIIB) along path III is considered, which involves the proton transfer from hydroxymethyl of carbinol theophylline to heteroatom O6 atom via transition states III-TS2B, yielding the enol isomer of theophylline (III-3B). As shown in Fig. 3, the located transition states I-TS2B, I-TS2C, II-TS2B, and II-TS2C are all hexatomic ring structures, whereas transition state III-TS2B is a heptatomic ring structure. The gas-phase activation energies of ITS2B, I-TS2C, II-TS2B, II-TS2C, and III-TS2B are 25.1, 24.6, 23.3, 28.3 and 23.9 kcal mol1, respectively, which correspond to 24.3, 23.5,

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

2863

Fig. 3. B3LYP/6-31þG(d) optimized geometries ( A) and activation energies (kcal mol1) for the decompositions of carbinol theobromine in path I (a), carbinol paraxanthine in path II (b) and carbinol theophylline in path III (c) as well as intramolecular proton transfer in path IV (d) (values in parentheses, activation energies DEs-s in chlorobenzene solution; values out of parentheses, activation energies DEs-s in the gas phase).

22.8, 23.3, and 24.8 kcal mol1 when chlorobenzene solution is incorporated. And then, the generated enol isomers, I-3B, I-3C, II-3B, II-3C, and III-3B can convert into their corresponding keto structures via enoleketo tautomerisms. Considering the less ring strain of hexatomic ring compared with tetratomic ring, transition states involving in this proton transfer processes were modeled with one explicit water molecule and should represent a lower energy pathway. The optimized transition state structures and activation energies for the enoleketo tautomerisms of xanthine derivatives are shown in Fig. 4 and Fig. S6. Owing to the assistance of one explicit water molecule, transition states I-TS3B-H2O, I-TS3C-H2O, IITS3B-H2O, and II-TS3C-H2O are hexatomic ring structures, while transition state III-TS3B-H2O is a heptatomic ring structure. The single imaginary frequencies of all the transition states are greater than 1200 i cm1, suggesting the motions of two light hydrogen atoms. The gas-phase activation energies of I-TS3B-H2O, I-TS3CH2O, II-TS3B-H2O, II-TS3C-H2O, and III-TS3B-H2O are 4.6, 5.5, 6.5, 7.3,

and 0.5 kcal mol1, which correspond to 4.7, 5.7, 6.5, 9.8, and 0.5 kcal mol1 when chlorobenzene solution is incorporated. All the enoleketo tautomerism processes are exothermic with the reaction energies of 8e15 kcal mol1 in the gas phase and 9e14 kcal mol1 in chlorobenzene solution. It is obvious that all the ketoeenol tautomerisms are thermodynamically and kinetically favorable. Especially, the activation energy of the heptatomic ring transition state III-TS3B-H2O is negative, indicating that the enol-toketo tautomerism is barrierless. Thus, the enol structure of theophylline can convert into its keto structure rapidly, which confirms the priority of path IIIB over path IIIA. As a result, the heteroatom-assisted proton transfer mechanism is more feasible than the direct proton transfer mechanism. The generated carbinolamines species are prone to decomposition via the heteroatom-assisted proton transfer. The O6-assisted path IC and O2-assisted path IIB are predicted to be the most favorable pathways for the decompositions of carbinol theobromine and carbinol paraxanthine, which agrees well with the P450-catalyzed metabolism

2864

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

Fig. 4. B3LYP/6-31þG(d) optimized transition structures ( A) and activation energies (kcal mol1) for the enoleketo tautomerisms of xanthine derivatives (values in parentheses, activation energies DEs-s in chlorobenzene solution; values out of parentheses, activation energies DEs-s in the gas phase).

of caffeine.25 However, the optimum decomposition pathway of carbinol theophylline should be the O6-assisted path IIIB rather than the direct proton transfer process path IIIA identified in the literature. Generally, the dispersion corrected energetics in chlorobenzene solution obtained in the present study is close to those obtained in the water solution without dispersion correction.25 3.2. C-8 oxidation (path IV) Considering the C-8 oxidation process, two possible oxidation patterns of FLHOOH were taken into account in our initial computations, the hydroxyl transfer mechanism and oxygen atom transfer mechanism. The hydroxyl transfer mechanism, a similar manner like N-methyl hydroxylation,42 involves the synchronous OeO bond homolysis and H-abstraction from C8 atom to produce a C8-centered radical and a water molecule. The formed water molecule concomitantly transfers an H-atom back to the proximal oxygen and the generated HO is transferred to the C8 centered

radical to produce C8-hydroxylated caffeine. However, this mechanism is implausible because imidazole species is more prone to addition rather than displacement. So even we have tried to locate the transition state for the hydroxyl transfer mechanism as Nmethyl hydroxylation does, but still failed. By comparison, the oxygen atom transfer mechanism concerns the OeO bond homolysis and the concomitant hydrogen transfer from the developing HO to the proximal oxygen to form the stable FLHOH and 8-oxocaffeine. Taken together, the oxygen atom transfer mechanism responsible for C-8 oxidation is stepwise (Scheme 1b): (i) the nucleophilic attack of distal oxygen atom of FLHOOH, leading to 8-oxocaffeine formation; (ii) which converts into stable trimethyluric acid through intramolecular proton transfer. 3.2.1. Nucleophilic attack of oxygen atom 3.2.1.1. Geometrical features. The optimized geometries of the stationary points for the nucleophilic attack of oxygen atom of FLHOOH in path IV are shown in Fig. 5. The located transition state

Fig. 5. B3LYP/6-31þG(d) optimized structures ( A) along the potential energy surface of the nucleophilic attack of oxygen atom in C-8 oxidation (path IV) of caffeine catalyzed by FMO.

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

IV-TS1 is characterized by its single imaginary frequency of 526.6 i cm1, implying that the heavy oxygen atom is comprised in the reaction vector. The major contribution of the single imaginary frequency is to promote OeO bond elongation with remarkable transfer of HO fragment to C8 atom of caffeine. The distance of OeO bond in IV-TS1 (1.946  A) is elongated by 0.501  A when compared with the reactant complex IV-1 (1.445  A). The internal OpeOdeHd bond angle in IV-TS1 is 91.7, which allows for a modest hydrogen bond of the HO radical to the proximal oxygen (rOeO¼2.208  A). Thus, there is essentially a bound HO radical suspended between Op and C8 atoms (rOeC¼1.783  A). This transferring HO radical is also stabilized by its H-bonding interaction with the adjacent C]O group (1.968  A). Owning to the available alternative sources of intramolecular stabilization, we still can not observe a complete somersault rearrangement. With the nucleophilic addition of HO radical at C8 atom, the FLHO radical formed in the transition state abstracts the distal hydrogen Hd atom of HO radical to produce the OH group of FLHOH. In the intermediate IV-2, the distal oxygen Od atom is bonded with C8 atom, leading to the partial cleavage of N7eC8 bond (2.771  A). More obvious results of the bond-breaking and bond-forming can be gained though the IRC results shown in Fig. 2d. The OdeOp bond cleavage is the most advanced among all bond-cleavage and bond-formation processes. Similar to N-methyl demethylation, it occurs at the coordinate of ca. 1 amu1/2 bohr, previous to the formation of IV-TS1. Afterwards, with the sharp increase of OdeOp bond, the OdeC8 bond decreases remarkably to form a single bond at the coordinate of ca. 0.8 amu1/2 bohr after crossing IV-TS1. The OdeHd bond, however, keeps constant until the coordinate of ca. 4 amu1/2 bohr. And then the OdeHd bond increases gradually, leading to the concurrent formations of OpeHd and Op]C8 bonds at the coordinate of ca. 6 amu1/2 bohr. 3.2.1.2. Electronic structural features. The preference of the nucleophilic addition of oxygen atom in C-8 oxidation can be further analyzed by the NPA charge distribution (Table S1). There is very little charge separation in the reaction complex IV-1 with the net charges on FLHOOH and caffeine of 0.007e and 0.007e, respectively. In the transition state IV-TS1, the transferring OH group is slightly negatively charged (0.113e) attributed to the partial formation of C8eOd bond (1.783  A), which complies with the idea of the OeO bond homolysis. The net charges on FLHO and caffeine in IV-TS1 are 0.306e and 0.419e, respectively, showing an obvious charge separation due to the intramolecular electron migration from caffeine to FLHOOH. This observation is consistent with the orbital characteristic of IV-TS1 (Fig. S1). The HOMO of IV-TS1 delocalizes mainly on the FLHO, whereas the LUMO resides primary on the caffeine. Due to the charge separation, IV-TS1 also exhibits polarity characteristic that reflected in the large dipole moment of 9.19 Debye. 3.2.1.3. Energy profiles. The changes in electronic energies (DE) for the nucleophilic attack of oxygen atom in C-8 oxidation are presented in Table 1. The gas-phase activation energy for this step is 28.5 kcal mol1, which corresponds to 25.6 kcal mol1 when chlorobenzene solution is incorporated. The reduction of activation energies (ca. 3 kcal mol1) in chlorobenzene solution accords well with the more polar characteristic of IV-TS1 with respect to IV-1 (8.11 Debye). It is expected that the chlorobenzene solution can promote the stability of IV-TS1 and inevitably leading to the reduction of the activation energy. The nucleophilic attack process is largely exothermic with reaction energy of 42.2 kcal mol1 in the gas phase (41.0 kcal mol1 in chlorobenzene solution). 3.2.2. Intramolecular proton transfer. Starting from the oxidized intermediate IV-2, two possible pathways were considered for the

2865

intramolecular proton transfer to yield trimethyluric acid (as shown in Fig. S7). Path IVA is one step reaction, which involves the direct proton transfer of H8 from C8 to N9 atom, yielding the keto isomer of trimethyluric acid (IV-3A). By contrast, path IVB is stepwise, including the proton (H8) transfer from C8 to Od atoms, forming the enol isomer of trimethyluric acid (IV-3B), which then converts into its keto isomer via enol-keto tautomerism. The key geometric information and electronic energies of the transition states involved are displayed in Fig. 3 and in further details for potential energy surfaces and stationary geometries in Figs. S8. 3.2.2.1. Path IVA. The located transition state, IV-TS2A, for the direct reaction process is a triatomic ring structure, which expands to the heptatomic ring structure of IV-TS2A-H2O when two explicit water molecules are comprised (Fig. 3).56 The single imaginary frequencies of IV-TS2A and IV-TS2A-H2O are 1170.7 and 225.4 i cm1, respectively. The gas-phase activation energies of IVTS2A and IV-TS2A-H2O are 32.3 and 25.8 kcal mol1, respectively, which correspond to 30.3 and 23.0 kcal mol1 when chlorobenzene solution is incorporated. It can be readily seen that the waterassisted mechanism is more favorable than the direct reaction mechanism. 3.2.2.2. Path IVB. The optimized geometries of transition state IV-TS2B is still a triatomic ring structure. With the mediation of two explicit water molecules, the located IV-TS2B-H2O is a heptatomic ring structure (Fig. 3). The single imaginary frequencies of IV-TS2B and IV-TS2B-H2O are 1343.5 and 475.4 i cm1, respectively. The gasphase activation energies of IV-TS2B and IV-TS2B-H2O are 41.8 and 21.0 kcal mol1, respectively, which decrease to 40.8 and 18.8 kcal mol1 when chlorobenzene solution is incorporated. It is expected, therefore, that the presence of the explicit water molecules contribute to the decrease of activation energy. And then, the formed enol isomer of trimethyluric acid, IV-3B, can converted to its keto isomer via transition state IV-TS3B-H2O, a hexatomic ring structure due to the mediation of one explicit water molecule (Fig. 4). The gas-phase activation energy for the enol-keto tautomerism is 8.5 kcal mol1, which decrease to 7.7 kcal mol1 when chlorobenzene solution is incorporated. This process is exothermic with the reaction energy of 5.4 kcal mol1 in the gas phase and 6.6 kcal mol1 in chlorobenzene solution, suggesting that the enolketo tautomerism is thermodynamically and kinetically feasible. Taken together, path IVB is more favorable than IVA because of its lower activation, which agrees well with the NBO charge analysis. The NPA charges on Od and N9 atoms are 0.638e and 0.488e, respectively. The stronger electronegativity of Od atom makes it more prone to being protonated. Therefore, water-mediated path IVB is the optimum pathway for this intramolecular proton transfer. However, different orientation of caffeine and FLHOOH in the reactant complexes, I-1, II-1, III-1, and IV-1, required by each reaction pathway leads to the binding energy discrepancies. The binding energies of caffeine to FLHOOH in the reactant complexes are shown in Table S2. The most stable reactant complex is I-1, with the largest binding energies of caffeine to FLHOOH (10.2 kcal mol1), followed by II-2 (8.8 kcal mol1), III-2 (8.1 kcal mol1), and IV-1 (2.5 kcal mol1) in turn. As a result, the relative energies of II-1, III-1, and IV-1 are 1.1, 1.3 and 2.3 kcal mol1 higher than I-1. Taken together, the energy profiles for the optimum reaction pathways of N-demethylations and C-8 oxidation of caffeine are depicted in Fig. 6. It can be readily seen that the ratedetermining steps of N-demethylations and C-8 oxidation involve the N-methyl hydroxylation and nucleophilic attack of oxygen atom, respectively. The activation energies for the rate-determining steps of paths IeIV obey the following order: path IV (27.9 kcal mol1)
2866

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867

Fig. 6. Potential energy profiles for the optimum reaction pathways of N-hydroxylation (paths IeIII) and C-8 oxidation (path IV) of caffeine by FLHOOH in chlorobenzene solution.

solution. The ratio of the reaction rates for paths IeIV is 1:12:1:3294. Consequently, C-8 oxidation is the optimum metabolic pathway and 1,3,7-trimethyluric acid therefore is the most optimum caffeine catabolic product catalyzed by FMO. N3demethylation is the most plausible N-demethylation of caffeine. And then paraxanthine is more favorable than theophylline and theobromine. All these results accords with the metabolism of caffeine by P450 enzyme on the high-spin state.25 4. Conclusions In this study, two types of metabolic mechanisms of caffeine by FMO have been explored, i.e., N-demethylation (paths IeIII) and C-8 oxidation (path IV). Our calculations have shown that the Ndemethylations of caffeine proceed in two stepwise steps, Nmethyl hydroxylation, the rate-determining step on the potential energy surface, and the decomposition of the formed carbinolamine to yield dimethylxanthine species metabolite. C-8 oxidation is an oxygen atom transfer mechanism, comprising of the ratedetermining nucleophilic attack of oxygen atom and the subsequent intramolecular proton transfer to form trimethyluric acid. Either the N-methyl hydroxylation or the nucleophilic attack of oxygen atom each involves the OeO bond homolysis with concomitant incomplete somersault rearrangement of hydroperoxy group. Each carbinolamine is more prone to decomposition through the adjacent heteroatom-assisted proton transfer. As a result, C-8 oxidation (path IV) predominates over N-demethylations (paths IeIII) due to its lowest activation energy (25.6 kcal mol1) and trimethyluric acid is therefore the optimum metabolite of caffeine by FMO in chlorobenzene solution. As for the three Ndemethylation pathways, path II, N3-demethylation, is more favorable than paths I and III, N1- and N7-demethylations. Consequently, paraxanthine is more feasible than theobromine and theophylline in chlorobenzene solution. This study has shed light on the detail metabolic mechanisms of caffeine by FMO at the first time, which can provide more essential insights into the oxidation of C-nucleophiles by FMO and offer important clues for the biodecaffeination techniques and the methylation mechanisms of dimethylxanthines. Acknowledgements

2011JY0136) and Department of Education of Sichuan Province (Grant No. 12ZA174) and China West Normal University (Grant No. 11B002). Supplementary data Supplementary data (Binding energies of caffeine to FLHOOH in the reactant complexes and Cartesian coordinates of all stationary points along the potential energy profiles.) associated with this article can be found in the online version, at http://dx.doi.org/ 10.1016/j.tet.2016.03.091. References and notes 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.

This work was supported by grants from National Natural Science Foundation of China (Grant No. 21203153), Department of Science and Technology of Sichuan Province (Grant No.

29.

Ashihara, H.; Crozier, A. Trends Plant Sci. 2001, 6, 407. Ashihara, H.; Suzuki, T. Front. Biosci. 2004, 9, 1864. Fisone, G.; Borgkvist, A.; Usiello, A. Cell. Mol. Life Sci. 2004, 61, 857. Ribeiro, J. A.; Sebasti~ ao, A. M. J. Alzheimers Dis. 2010, 20, S3. Barry, R. J.; Clarke, A. R.; Johnstone, A. J.; Rushby, J. A. Biol. Psychol. 2008, 77, 304. Maughan, R. J.; Griffin, J. J. Hum. Nutr. Diet. 2003, 16, 411. Migliardi, J. R.; Armellino, J. J.; Friedman, M.; Gillings, D. B.; Beaver, W. T. Clin. Pharmacol. Ther. 1994, 56, 576. Borota, D.; Murray, E.; Keceli, G.; Chang, A.; Watabe, J. M.; Ly, M.; Toscano, J. P.; Yassa, M. A. Nat. Neurosci. 2014, 17, 201. Ding, M.; Bhupathiraju, S. N.; Satija, A.; Van Dam, R. M.; Hu, F. B. Circulation 2014, 129, 643. Johnson, S.; Koh, W. P.; Wang, R.; Govindarajan, S.; Yu, M. C.; Yuan, J. M. Cancer, Causes Control 2011, 22, 503. Muriel, P.; Arauz, J. Fitoterapia 2010, 81, 297. Arab, L. Nutr. Cancer 2010, 62, 271. Gonzalez de Mejia, E.; Ramirez-Mares, M. V. Trends Endocrinol. Metab. 2014, 25, 489. Ludwig, I. A.; Clifford, M. N.; Lean, M. E. J.; Ashihara, H.; Crozier, A. Food Funct. 2014, 5, 1695. pez, S.; Sarria , B.; Go mez-Juaristi, M.; Goya, L.; Mateos, R.; BravoMartínez-Lo Clemente, L. Food Res. Int. 2014, 63, 446. Ibrahim, S.; Shukor, M. Y.; Syed, M. A.; Rahman, N. A. A.; Khalil, K. A.; Ariff Khalid, A.; Ahmad, S. A. Asian J. Plant Biol. 2014, 2, 18. Dash, S. S.; Gummadi, S. N. Biotechnol. Lett. 2006, 28, 1993. Chung, W. G.; Cha, Y. N. Biochem. Biophys. Res. Commun. 1997, 235, 685. Lang, D. H.; Rettie, A. E. Br. J. Clin. Pharmacol. 2000, 50, 311. Kot, M.; Daniel, W. A. Pharmacol. Rep. 2008, 60, 789. Cashman, J. R.; Zhang, J. Annu. Rev. Pharmacol. Toxicol. 2006, 46, 65. Cashman, J. R. Biochem. Biophys. Res. Commun. 2005, 338, 599. Krueger, S. K.; Williams, D. E. Pharmacol. Ther. 2005, 106, 357. Ziegler, D. M. Drug Metab. Rev. 2002, 34, 503. Chen, Z. Q.; Kang, Y.; Zhang, C. H.; Tao, J.; Xue, Y. Theor. Chem. Acc. 2015, 134, 110. Chen, G. P.; Ziegler, D. M. Arch. Biochem. Biophys. 1994, 312, 566. Rooseboom, M.; Commandeur, J. N.; Floor, G. C.; Rettie, A. E.; Vermeulen, N. P. Chem. Res. Toxicol. 2001, 14, 127. Murray, A. T.; Matton, P.; Fairhurst, N. W.; John, M. P.; Carbery, D. R. Org. Lett. 2012, 14, 3656. Henderson, M. C.; Siddens, L. K.; Krueger, S. K.; Stevens, J. F.; Kedzie, K.; Fang, W. K.; Heidelbaugh, T.; Nguyen, P.; Chow, K.; Garst, M.; Gil, D.; Williams, D. E. Toxicol. Appl. Pharmacol. 2014, 278, 91.

Y. Kang et al. / Tetrahedron 72 (2016) 2858e2867 30. Huijbers, M. M. E.; Montersino, S.; Westphal, A. H.; Tischler, D.; Van Berkel, W. J. H. Arch. Biochem. Biophys. 2014, 544, 2. 31. Massey, V. Biochem. Soc. Trans. 2000, 28, 283. 32. Anderson, R. F. In Bray, R. C., Engel, P. C., Mayhew, S. G., Eds.; Walter de Gruyter: Berlin, Germany, 1984, pp 57. 33. Wongnate, T.; Surawatanawong, P.; Visitsatthawong, S.; Sucharitakul, J.; Scrutton, N. S.; Chaiyen, P. J. Am. Chem. Soc. 2014, 136, 241. 34. Canepa, C.; Bach, R. D.; Dmitrenko, O. J. Org. Chem. 2002, 67, 8653. 35. Bach, R. D.; Dmitrenko, O. J. Phys. Chem. B 2003, 107, 12851. 36. Bach, R. D. J. Phys. Chem. A 2011, 115, 11087. 37. Badieyan, S.; Bach, R. D.; Sobrado, P. J. Org. Chem. 2015, 80, 2139. 38. Perakyla, M.; Pakkanen, T. A. J. Am. Chem. Soc. 1993, 115, 10958. 39. Ridder, L.; Mulholland, A. J.; Rietjens, I. M. C. M.; Vervoort, J. J. Am. Chem. Soc. 2000, 122, 8728. 40. Billeter, S. R.; Hanser, C. F. W.; Mordasini, T. Z.; Scholten, M.; Thiel, W.; Van Gunsterenc, W. F. Phys. Chem. Chem. Phys. 2001, 3, 688. 41. Ridder, L.; Harvey, J. N.; Rietjens, I. M. C. M.; Vervoort, J.; Mulholland, A. J. J. Phys. Chem. B 2003, 107, 2118. 42. Massey, V. J. Biol. Chem. 1994, 269, 22459. 43. Eswaramoorthy, S.; Bonanno, J. B.; Burley, S. K.; Swaminathan, S. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 9832. 44. 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.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery,

45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56.

2867

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.; Fox, D. J. Gaussian 09, Revision D01; Gaussian: Wallingford, CT, 2013. Becke, A. D. J. Chem. Phys. 1993, 98, 5648. Stephens, P. J.; Devlin, F. J.; Chabalowski, C. F.; Frisch, M. J. J. Phys. Chem. 1994, 98, 11623. Zhang, Q.; Bell, R.; Truong, T. N. J. Phys. Chem. 1995, 99, 592. Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154. Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899. Tao, J.; Kang, Y.; Xue, Z. Y.; Wang, Y. T.; Zhang, Y.; Chen, Q.; Chen, Z. Q.; Xue, Y. J. Mol. Graphics Modell. 2015, 61, 123. Mennucci, B. WIREs Comput. Mol. Sci. 2012, 2, 386. Schutz, C. N.; Warshel, A. Proteins 2001, 44, 400. Schyman, P.; Lai, W.; Chen, H. Z.; Wang, Y.; Shaik, S. J. Am. Chem. Soc. 2011, 133, 7977. Bach, R. D. J. Phys. Chem. A 2010, 114, 9313. Adamo, C.; Barone, V. J. Chem. Phys. 1999, 110, 6158. Considering the less constraint of heptatomic ring than pentatomic structure, two explicit water molecules were taken into account for the water-mediated paths IVA and IVB.