Biodegradation of Cosmetics Products : A Computational Study of Cytochrome P 450 Metabolism of Phthalates

Cytochrome P450s are a broad class of enzymes in the human body with important functions for human health, which include the metabolism and detoxification of compounds in the liver. Thus, in their catalytic cycle, the P450s form a high-valent iron(IV)-oxo heme cation radical as the active species (called Compound I) that reacts with substrates through oxygen atom transfer. This work discusses the possible degradation mechanisms of phthalates by cytochrome P450s in the liver, through computational modelling, using 2-ethylhexyl-phthalate as a model substrate. Phthalates are a type of compound commonly found in the environment from cosmetics usage, but their biodegradation in the liver may lead to toxic metabolites. Experimental studies revealed a multitude of products and varying product distributions among P450 isozymes. To understand the regioand chemoselectivity of phthalate activation by P450 isozymes, we focus here on the mechanisms of phthalate activation by Compound I leading to O-dealkylation, aliphatic hydroxylation and aromatic hydroxylation processes. We set up model complexes of Compound I with the substrate and investigated the reaction mechanisms for products using the density functional theory on models and did a molecular mechanics study on enzymatic structures. The work shows that several reaction barriers in the gas-phase are close in energy, leading to a mixture of products. However, when we tried to dock the substrate into a P450 isozyme, some of the channels were inaccessible due to unfavorable substrate positions. Product distributions are discussed under various reaction conditions and rationalized with valence bond and thermodynamic models.


Introduction
Phthalates are commonly used chemicals in a variety of products including cosmetics.Although no clear evidence exists, they actually may be harmful compounds to biosystems and, for instance, may lead to breast cancer in humans.It is believed that phthalates can be biodegraded in the human liver and most likely cytochrome P450s are involved in this process.However, as some of these products may be toxic metabolites, and their origin is unclear, we decided to do a computational study on the most likely products obtained from the reaction of a P450 active site model with a phthalate substrate.Cytochrome P450 enzymes are a superfamily of heme-containing monoxygenases found in most biological systems, including mammals, insects and fungi as well as bacteria [1][2][3][4][5][6].It is believed to be the most abundant enzyme family in the plant kingdom [7].As of 2017, there have been over 39,400 sequences assigned across 236 species [8], and with many genomes yet to be sequenced, these numbers are projected to increase significantly over the next few years [6,7].
The P450s display a large range of substrate activation reactions and hence, they are well studied, and applications of these enzymes are being sought in biotechnology and medicine.Structurally, the P450s are heme enzymes that bind molecular oxygen on the heme iron and their most common reaction mechanism with substrates happens through oxygen atom transfer to substrates [3,[9][10][11].Figure 1 displays the active site structure of P450, as taken from the 4L40 protein databank (pdb) file [12].The resting state of P450 enzymes has an iron(III)(heme)(water) with the metal in a six-coordination ligand environment that is linked to the protein backbone through a chemical bond of the iron with the sulfur atom of a cysteinate residue (Cys 365 ).The pdb file also contains the substrate-i.e., a linear fatty acid-that is located in a pocket on the distal side of the heme and held in position through a salt-bridge with an active site Arg residue.
Upon the substrate entering into the binding pocket and approaching the heme, the catalytic cycle (right-hand side of Figure 1) is triggered through the release of the water molecule from iron and resulting in a spin state change that prompts an electron transfer from the P450 reduction partner.Subsequently, molecular oxygen binds to the heme, is reduced and protonated to form an iron(III)-hydroperoxo species, called Compound 0 (Cpd0) [13].A final protonation step leads to an iron(IV)-oxo heme cation radical species called Compound I (CpdI).The latter reacts with substrates, for instance, through an aliphatic hydroxylation reaction via hydrogen atom abstraction followed by OH transfer [14].Experimental evidence for this catalytic cycle comes from spectroscopy (UV-Vis absorption, electron paramagnetic resonance and Mössbauer spectroscopy) studies that have characterized P450 CpdI [15] as well as Cpd0 [16].The sole oxidant of P450 enzymes is CpdI.A range of experimental and computational studies have ruled out Cpd0 as an oxidant [17][18][19] (i.e., iron(III)-hydroperoxo), although recent studies on biomimetic nonheme iron(III)-hydroperoxo models have shown reactivity with arenes and halides [20][21][22].These differences were assigned as resulting from differences in the O-O bond cleavage patterns between heme and nonheme iron(III)-hydroperoxo systems, where the hemes give favorable heterolytic cleavage and the nonhemes homolytic cleavage.Computational modelling has revealed that CpdI has two close-lying spin states, namely doublet and quartet spin, that are close in energy and hence the structure and reactivity is dependent on the abundances of both spin states.
yet to be sequenced, these numbers are projected to increase significantly over the next few years [6,7].
The P450s display a large range of substrate activation reactions and hence, they are well studied, and applications of these enzymes are being sought in biotechnology and medicine.Structurally, the P450s are heme enzymes that bind molecular oxygen on the heme iron and their most common reaction mechanism with substrates happens through oxygen atom transfer to substrates [3,[9][10][11].Figure 1 displays the active site structure of P450, as taken from the 4L40 protein databank (pdb) file [12].The resting state of P450 enzymes has an iron(III)(heme)(water) with the metal in a six-coordination ligand environment that is linked to the protein backbone through a chemical bond of the iron with the sulfur atom of a cysteinate residue (Cys365).The pdb file also contains the substrate-i.e., a linear fatty acid-that is located in a pocket on the distal side of the heme and held in position through a salt-bridge with an active site Arg residue.
Upon the substrate entering into the binding pocket and approaching the heme, the catalytic cycle (right-hand side of Figure 1) is triggered through the release of the water molecule from iron and resulting in a spin state change that prompts an electron transfer from the P450 reduction partner.Subsequently, molecular oxygen binds to the heme, is reduced and protonated to form an iron(III)-hydroperoxo species, called Compound 0 (Cpd0) [13].A final protonation step leads to an iron(IV)-oxo heme cation radical species called Compound I (CpdI).The latter reacts with substrates, for instance, through an aliphatic hydroxylation reaction via hydrogen atom abstraction followed by OH transfer [14].Experimental evidence for this catalytic cycle comes from spectroscopy (UV-Vis absorption, electron paramagnetic resonance and Mössbauer spectroscopy) studies that have characterized P450 CpdI [15] as well as Cpd0 [16].The sole oxidant of P450 enzymes is CpdI.A range of experimental and computational studies have ruled out Cpd0 as an oxidant [17][18][19] (i.e., iron(III)-hydroperoxo), although recent studies on biomimetic nonheme iron(III)-hydroperoxo models have shown reactivity with arenes and halides [20][21][22].These differences were assigned as resulting from differences in the O-O bond cleavage patterns between heme and nonheme iron(III)-hydroperoxo systems, where the hemes give favorable heterolytic cleavage and the nonhemes homolytic cleavage.Computational modelling has revealed that CpdI has two close-lying spin states, namely doublet and quartet spin, that are close in energy and hence the structure and reactivity is dependent on the abundances of both spin states.The mechanism of these oxygen atom transfer reactions has been studied in detail, with a range of computational techniques [14,23].An overview of commonly catalyzed reactions by P450 isozymes is given in Figure 2. The most extensively studied reaction pathway is aliphatic hydroxylation [24][25][26][27], which happens via a stepwise mechanism with an initial hydrogen atom The mechanism of these oxygen atom transfer reactions has been studied in detail, with a range of computational techniques [14,23].An overview of commonly catalyzed reactions by P450 isozymes is given in Figure 2. The most extensively studied reaction pathway is aliphatic hydroxylation [24][25][26][27], which happens via a stepwise mechanism with an initial hydrogen atom abstraction to form an iron(IV)-hydroxo intermediate, followed by OH transfer-also called rebound-to the radical to give the alcohol product complexes.With most substrates, the hydrogen atom abstraction step is rate-determining, while the rebound barrier is much smaller [28].In particular, the doublet spin state mechanism tends to have negligible rebound barriers, while they are generally higher on the quartet spin state surface.This is due to electron transfer into a virtual orbital in the quartet spin-state, whereas the electron can move into a lower lying orbital in the doublet spin state.P450s are also known to convert C=C double bonds into olefins [29][30][31][32], sulfides into sulfoxides [33][34][35] and arenes into phenols [36][37][38][39] (pathways II, III and IV in Figure 2).Apart from sulfoxidation, which is a concerted process, all other oxygen atom transfer reactions are stepwise, via at least one intermediate that is usually a radical.
Inorganics 2017, 5, 77 3 of 18 abstraction to form an iron(IV)-hydroxo intermediate, followed by OH transfer-also called rebound-to the radical to give the alcohol product complexes.With most substrates, the hydrogen atom abstraction step is rate-determining, while the rebound barrier is much smaller [28].In particular, the doublet spin state mechanism tends to have negligible rebound barriers, while they are generally higher on the quartet spin state surface.This is due to electron transfer into a virtual orbital in the quartet spin-state, whereas the electron can move into a lower lying orbital in the doublet spin state.P450s are also known to convert C=C double bonds into olefins [29][30][31][32], sulfides into sulfoxides [33][34][35] and arenes into phenols [36][37][38][39] (pathways II, III and IV in Figure 2).Apart from sulfoxidation, which is a concerted process, all other oxygen atom transfer reactions are stepwise, via at least one intermediate that is usually a radical.In recent years, however, alternative mechanisms have been established, which lead to desaturation or ring-closure processes (pathways V, VI and VII in Figure 2).In particular, ring-closure reactions are part of the natural product biosynthesis of antibiotics and hormone compounds [40].Other forms of desaturation processes refer to the conversion of aliphatic groups to olefins (pathway V) or the decarboxylation of fatty acids to terminal olefins (pathway VI).Thus, desaturation reactions have been observed in the activation of drug molecules, for instance in ethylcarbamate and valproic acid, by P450 isozymes [41,42] but also in the biosynthesis of ergosterol in the human body [43].Computational studies on the mechanism of substrate desaturation showed these desaturation reactions by CpdI to start with a hydrogen atom abstraction, similar to aliphatic hydroxylation.However, in contrast to rebound of the OH group to the radical, as happens in substrate hydroxylation mechanisms, in desaturation processes a second hydrogen atom abstraction by the iron(IV)-hydroxo group gives iron(III)-water and an olefin product [44,45].As such, the mechanism can bifurcate in the radical intermediates, which can lead to both hydroxylation and desaturation products.Recently, a P450 isozyme was discovered (P450OleTJE) that binds long-chain fatty acids and converts them to terminal olefins through a decarboxylation reaction [46,47].A detailed quantum mechanics/molecular mechanics (QM/MM) study on the mechanisms leading to In recent years, however, alternative mechanisms have been established, which lead to desaturation or ring-closure processes (pathways V, VI and VII in Figure 2).In particular, ring-closure reactions are part of the natural product biosynthesis of antibiotics and hormone compounds [40].Other forms of desaturation processes refer to the conversion of aliphatic groups to olefins (pathway V) or the decarboxylation of fatty acids to terminal olefins (pathway VI).Thus, desaturation reactions have been observed in the activation of drug molecules, for instance in ethylcarbamate and valproic acid, by P450 isozymes [41,42] but also in the biosynthesis of ergosterol in the human body [43].Computational studies on the mechanism of substrate desaturation showed these desaturation reactions by CpdI to start with a hydrogen atom abstraction, similar to aliphatic hydroxylation.However, in contrast to rebound of the OH group to the radical, as happens in substrate hydroxylation mechanisms, in desaturation processes a second hydrogen atom abstraction by the iron(IV)-hydroxo group gives iron(III)-water and an olefin product [44,45].As such, the mechanism can bifurcate in the radical intermediates, which can lead to both hydroxylation and desaturation products.
Recently, a P450 isozyme was discovered (P450 OleTJE ) that binds long-chain fatty acids and converts them to terminal olefins through a decarboxylation reaction [46,47].A detailed quantum mechanics/molecular mechanics (QM/MM) study on the mechanisms leading to decarboxylation and αand β-hydroxylation of the linear fatty acid was performed [48] and gave insight into the origins of the bifurcation pathways.
Clearly, P450 enzymes react via a diverse set of chemical reactions with substrates and the origins of the product distributions are not always clear.A group of substrates with great relevance for human health and excessively present in the environment is the phthalates.Experimental studies using a range of P450 isozymes, however, have found a mixture of products, with product ratios dependent on the isozyme [49].Although, C 5 hydroxylation was commonly observed in most P450 isozymes, in some of them significant amounts of O-dealkylation and aromatic hydroxylation products were also obtained.To gain insight into how P450 enzymes metabolize phthalate substrates, we decided to do a computational study to investigate the mechanisms leading to C 5 -hydroxylation, O-dealkylation, and aromatic ring epoxidation and hydroxylation, which are presented in this work.

Results
Following previous benchmark and calibration studies, we started the work with enzyme active site model complexes of P450 CpdI with the substrate [50][51][52].The model (see Figure 3) contained the iron(IV)-oxo group embedded in a porphyrin without side chains, while the axial cysteinate ligand was abbreviated to SH − .As a model substrate, we investigated the ester of phthalate with 2-ethylhexanol (SubH) and studied the mechanisms leading to aliphatic hydroxylation, O-dealkylation, epoxidation and aromatic hydroxylation, which are suggested products obtained for the reaction of phthalates with P450s [49].Specifically, we focused on aliphatic hydroxylation at the C 5 position of the alkyl chain of the substrate and expected it to start with a hydrogen atom abstraction (via transition state TS A ) to give the radical intermediate, IA [24][25][26][27][53][54][55][56].Radical rebound (via transition state TS rebA ) would then give the alcohol product complex (PA).The mechanism tested for O-dealkylation started with a hydrogen atom abstraction from atom C 1 (via transition state TS B ) to give the radical intermediate, IB.Subsequently, radical rebound (via transition state TS rebB ) gave the alcohol product (structure IB2).In the latter, transfer of the alcoholic proton to the carboxylate group, splits substrate into a phthalate and an aldehyde (product PB) via a dealkylation transition state TS dealk .The third mechanism tested was epoxidation of the aromatic ring, where an electrophilic attack of the oxo-group on one of the aromatic carbon atoms takes place (via transition state TS C ) to form an intermediate, IC.A ring-closure transition state (TS rcC ) then leads to the epoxide product complex, PC.The final mechanism was aromatic hydroxylation, which also starts with an electrophilic attack of the oxo-group on one of the aromatic carbon atoms (via the same transition state TS C ) to form an intermediate, IC.However, a bifurcation of the latter leads to the loss of its ipso-proton to the porphyrin ring and generates the protonated porphyrin intermediate, ID, which reshuttles its proton to the oxygen atom via a proton-transfer transition state TS PT , to form phenol products (PD).
In the next few sections, we will discuss each of these mechanisms in detail, but first we will start with a comprehensive analysis of the structure and electronic configuration of the reactant complex, namely CpdI.

CpdI Structure and Electronic Configuration
CpdI has been characterized experimentally as a triradical system with unpaired electrons on the iron(IV)-oxo and heme groups in an antiferromagnetic manner [15].Early density functional theory (DFT) calculations [50,51,57] as well as subsequent QM/MM studies [58][59][60] have predicted this electronic configuration already, but have shown that CpdI has close-lying doublet and quartet spin states of almost equal energy.Figure 4 displays the relevant valence orbitals of P450 CpdI.The left-hand side of Figure 4 gives the metal dominated molecular orbitals.The lowest three orbitals shown are the bonding-type orbitals along the Fe-O bond and include the σz 2 for the overlap of 3dz 2 on iron with 2pz on oxygen and the degenerate pair of πxz and πyz molecular orbitals for the bonding interaction between the 3dxz (or 3dyz) on iron with the 2px (or 2py) atomic orbital on oxygen.The antibonding combinations of this pair of orbitals (π*xz and π*yz) are higher in energy and both are singly occupied.In between these two pairs of orbitals is the doubly occupied δx 2 -y 2 orbital, which is non-bonding and located in the plane of the heme/porphyrin.The two σ* antibonding orbitals are high in energy and virtual-one along the O-Fe-S axis (the z-axis), namely σ*z 2 , and the other one in the plane of the porphyrin ring for the antibonding interactions of the metal with 2px/2py orbitals on the nitrogen atoms of the porphyrin (σ*xy).Note that in this nomenclature, we took the x-and y-axis through an Fe-N bond.If the x-and y-axis instead are drawn in between two Fe-N bonds, the nomenclature will change and the labels xy and x 2 -y 2 are swapped.
On the right-hand side of Figure 4 are two high-lying π-orbitals on the porphyrin manifold, namely a1u and a2u.The a2u orbital has electron density on the porphyrin nitrogen atoms as well as

CpdI Structure and Electronic Configuration
CpdI has been characterized experimentally as a triradical system with unpaired electrons on the iron(IV)-oxo and heme groups in an antiferromagnetic manner [15].Early density functional theory (DFT) calculations [50,51,57] as well as subsequent QM/MM studies [58][59][60] have predicted this electronic configuration already, but have shown that CpdI has close-lying doublet and quartet spin states of almost equal energy.Figure 4 displays the relevant valence orbitals of P450 CpdI.The left-hand side of Figure 4 gives the metal dominated molecular orbitals.The lowest three orbitals shown are the bonding-type orbitals along the Fe-O bond and include the σ z 2 for the overlap of 3d z 2 on iron with 2p z on oxygen and the degenerate pair of π xz and π yz molecular orbitals for the bonding interaction between the 3d xz (or 3d yz ) on iron with the 2p x (or 2p y ) atomic orbital on oxygen.The antibonding combinations of this pair of orbitals (π* xz and π* yz ) are higher in energy and both are singly occupied.
In between these two pairs of orbitals is the doubly occupied δ x 2 -y 2 orbital, which is non-bonding and located in the plane of the heme/porphyrin.The two σ* antibonding orbitals are high in energy and virtual-one along the O-Fe-S axis (the z-axis), namely σ* z 2 , and the other one in the plane of the porphyrin ring for the antibonding interactions of the metal with 2p x /2p y orbitals on the nitrogen atoms of the porphyrin (σ* xy ).Note that in this nomenclature, we took the xand y-axis through an Fe-N bond.If the xand y-axis instead are drawn in between two Fe-N bonds, the nomenclature will change and the labels xy and x 2 -y 2 are swapped.
On the right-hand side of Figure 4 are two high-lying π-orbitals on the porphyrin manifold, namely a 1u and a 2u .The a 2u orbital has electron density on the porphyrin nitrogen atoms as well as the meso-carbon atoms.As a result of this, the π-orbitals can interact with the axial ligand orbitals and in particular, the a 2u mixes with a π-orbital on sulfur (3p z ), which destabilizes the a 2u orbital in energy [61].Consequently, the a 2u is high-lying and is easier to reduce in the absence of an axial ligand, resulting in a low electron affinity for P450 CpdI [62].
Inorganics 2017, 5, 77 6 of 18 the meso-carbon atoms.As a result of this, the π-orbitals can interact with the axial ligand orbitals and in particular, the a2u mixes with a π-orbital on sulfur (3pz), which destabilizes the a2u orbital in energy [61].Consequently, the a2u is high-lying and is easier to reduce in the absence of an axial ligand, resulting in a low electron affinity for P450 CpdI [62].Our optimized geometries of 4,2 CpdI match previous structures well, with Fe-O distances of 1.626 (1.624) Å and Fe-S distances of 2.570 (2.581) Å for 4 1 ( 2 1), respectively.Both structures corresponded to an electronic configuration of πxz 2 πyz 2 δ(x 2 -y 2 ) 2 π*xz 1 π*yz 1 a1u 2 a2u 1 , whereby the unpaired a2u electron is in an up-spin state in the quartet spin state, but a down-spin state in the doublet spin state.Energetically, the two spin states are within 1 kcal•mol −1 but their exact ordering and energy differences are dependent on whether solvent, entropy and external perturbations are taken into account [51,61].

Phthalate Hydroxylation
Next, we investigated the reaction mechanism of aliphatic hydroxylation at the C 5 position and the obtained potential energy profile is given in Figure 5.The reaction is stepwise via a radical intermediate leading to alcohol products in a highly exothermic process.The hydrogen atom abstraction barrier is rate-determining with values of ∆E + ZPE = 19.9(18.2) kcal•mol −1 on the quartet (doublet) spin states.The imaginary frequency for the hydrogen atom abstraction barriers are large (i1750 and i1549 for 4 TSA and 2 TSA, respectively), which is typical for hydrogen atom abstraction barriers [62-65] and usually means they are affected by replacement of hydrogen by deuterium, which gives a large kinetic isotope effect [66].

Phthalate Hydroxylation
Next, we investigated the reaction mechanism of aliphatic hydroxylation at the C 5 position and the obtained potential energy profile is given in Figure 5.The reaction is stepwise via a radical intermediate leading to alcohol products in a highly exothermic process.The hydrogen atom abstraction barrier is rate-determining with values of ∆E + ZPE = 19.9(18.2) kcal•mol −1 on the quartet (doublet) spin states.The imaginary frequency for the hydrogen atom abstraction barriers are large (i1750 and i1549 for 4 TS A and 2 TS A , respectively), which is typical for hydrogen atom abstraction barriers [62-65] and usually means they are affected by replacement of hydrogen by deuterium, which gives a large kinetic isotope effect [66].The optimized geometries of 4,2 TSA put the transferring hydrogen atom close to the acceptor oxygen atom (1.178 and 1.180 Å for 4 TSA and 2 TSA, respectively) indicating that the transition states are late on the potential energy surface.In general, late transition states correspond with higher barriers on the potential energy surface than earlier transition states in agreement with lower barriers [26].In both cases a radical intermediate is formed, and spin density starts to accumulate on the substrate C 5 atom (ρSubH is 0.98 and 0.77 for the quartet and doublet spin states).At the same time the spin density on the FeO group is polarized toward iron.Formation of the radical intermediates is energetically costly and is overall endothermic by 9.2 and 8.5 kcal•mol −1 in the quartet and doublet spin states.In the hydrogen atom abstraction step, in analogy to previous work [14,[24][25][26][67][68][69], the doublet and quartet spin state surfaces are close in energy and virtually degenerate.After the radical intermediate, however, the two pathways bifurcate and a radical rebound barrier of 1.6 kcal•mol −1 needs to be crossed on the quartet spin state surface whereas the mechanism is barrierless on the low-spin surface.Attempts were made to find a rebound transition state on the doublet spin state surface, but all our geometry scans showed facile pathways with barriers less than 1 kcal•mol −1 to form product complexes.The overall reaction leading to alcohol products is highly exothermic by around 50 kcal•mol −1 on both spin state surfaces.

Phthalate O-Dealkylation
Subsequently, we considered the reaction mechanism for O-dealkylation of our phthalate substrate and the obtained potential energy landscape is given in Figure 6.Similar to the aliphatic hydroxylation reaction described in the previous section, the reaction is stepwise with an initial hydrogen atom abstraction via a radical intermediate leading to alcohol products in a highly exothermic process.However, the alcohol is not the final product and with a proton transfer step via transition state TSdealk the alkyl group comes off as an aldehyde.Also for O-dealkylation, the hydrogen atom abstraction barrier is the rate-determining barrier in the reaction process with values energy plus zero-point energy of ∆E + ZPE = 19.3(20.3) kcal•mol −1 on the quartet (doublet) spin states.The imaginary frequency for the hydrogen atom abstraction barriers are large (i1775 and i1726 for 4 TSB and 2 TSB, respectively), which is similar to those reported above for 4,2 TSA.The optimized geometries of 4,2 TS A put the transferring hydrogen atom close to the acceptor oxygen atom (1.178 and 1.180 Å for 4 TS A and 2 TS A , respectively) indicating that the transition states are late on the potential energy surface.In general, late transition states correspond with higher barriers on the potential energy surface than earlier transition states in agreement with lower barriers [26].In both cases a radical intermediate is formed, and spin density starts to accumulate on the substrate C 5 atom (ρ SubH is 0.98 and 0.77 for the quartet and doublet spin states).At the same time the spin density on the FeO group is polarized toward iron.Formation of the radical intermediates is energetically costly and is overall endothermic by 9.2 and 8.5 kcal•mol −1 in the quartet and doublet spin states.In the hydrogen atom abstraction step, in analogy to previous work [14,[24][25][26][67][68][69], the doublet and quartet spin state surfaces are close in energy and virtually degenerate.After the radical intermediate, however, the two pathways bifurcate and a radical rebound barrier of 1.6 kcal•mol −1 needs to be crossed on the quartet spin state surface whereas the mechanism is barrierless on the low-spin surface.Attempts were made to find a rebound transition state on the doublet spin state surface, but all our geometry scans showed facile pathways with barriers less than 1 kcal•mol −1 to form product complexes.The overall reaction leading to alcohol products is highly exothermic by around 50 kcal•mol −1 on both spin state surfaces.

Phthalate O-Dealkylation
Subsequently, we considered the reaction mechanism for O-dealkylation of our phthalate substrate and the obtained potential energy landscape is given in Figure 6.Similar to the aliphatic hydroxylation reaction described in the previous section, the reaction is stepwise with an initial hydrogen atom abstraction via a radical intermediate leading to alcohol products in a highly exothermic process.However, the alcohol is not the final product and with a proton transfer step via transition state TS dealk the alkyl group comes off as an aldehyde.Also for O-dealkylation, the hydrogen atom abstraction barrier is the rate-determining barrier in the reaction process with values energy plus zero-point energy of ∆E + ZPE = 19.3(20.3) kcal•mol −1 on the quartet (doublet) spin states.The imaginary frequency for the hydrogen atom abstraction barriers are large (i1775 and i1726 for 4 TS B and 2 TS B , respectively), which is similar to those reported above for 4,2 TS A .
Inorganics 2017, 5, 77 8 of 18 The hydrogen atom abstraction transition state geometries ( 4,2 TSB) are similar in structure to 4,2 TSA, with long C-H and short O-H distances; C-H bonds of 1.386 and 1.378 Å and O-H bonds of 1.165 and 1.160 Å are found for 4 TSB and 2 TSB, respectively, indicating that the transition states are also late on the potential energy surface.Intermediates (IB1) are radicals with considerable spin density on the substrate C 1 atom (ρSubH is 0.97 for both the quartet and doublet spin states).Similar to the aliphatic hydroxylation pathway from the previous section, the radical intermediates, 4,2 IB1, are above reactants by about 7-8 kcal•mol −1 .On both spin state surfaces we failed to obtain a rebound transition state, but geometry scans point to a facile and fast process with a barrier of less than 1 kcal•mol −1 , leading to the intermediate, 4,2 IB2.The radical disappears upon formation of the alcohol intermediate, 4,2 IB2, en route to the O-dealkylation product, 4,2 PB.The last step of dealkylation happens away from the P450 active site and can be catalyzed by a solvent water molecule.Thus, when we add a bridging water molecule to the substrate alcohol position a dealkylation barrier of 11.5 kcal•mol −1 is obtained for proton relay, from the alcohol group to the carbonyl of the ester bond.This breaks the ester bond and releases ortho-phthalic acid in a process that is exothermic by 5.8 kcal•mol −1 .Subsequent inclusion of water molecules in the model may even bring the barrier further down, as evidenced in previous studies on P450 catalyzed dealkylation [70].

Phthalate Epoxidation and Aromatic Hydroxylation
The final two pathways of phthalate oxidation by P450 CpdI that were considered are substrate epoxidation and aromatic hydroxylation, which both start with an electrophilic attack of the oxo group on one of the carbon atoms of the aromatic ring to form a radical intermediate, 2,4 IC.From this intermediate, the landscape is seen to bifurcate into two directions (Figure 7).Firstly, a ring-closure transition state (TSrcC) leads to epoxide product complexes (PC).Secondly, from IC the ipso-proton can transfer from the benzene ring to one of the nitrogen atoms of porphyrin to form intermediate ID in a highly exothermic reaction step and without a reaction barrier.A small proton-reshuttle barrier TSPT of less than 1 kcal•mol −1 transfers the proton to the oxygen atom to form phenol products (PD).An alternative proton transfer pathway from 4,2 ID leads, via a significantly higher barrier, to ketone products.The hydrogen atom abstraction transition state geometries ( 4,2 TS B ) are similar in structure to 4,2 TS A , with long C-H and short O-H distances; C-H bonds of 1.386 and 1.378 Å and O-H bonds of 1.165 and 1.160 Å are found for 4 TS B and 2 TS B , respectively, indicating that the transition states are also late on the potential energy surface.Intermediates (IB1) are radicals with considerable spin density on the substrate C 1 atom (ρ SubH is 0.97 for both the quartet and doublet spin states).Similar to the aliphatic hydroxylation pathway from the previous section, the radical intermediates, 4,2 IB1, are above reactants by about 7-8 kcal•mol −1 .On both spin state surfaces we failed to obtain a rebound transition state, but geometry scans point to a facile and fast process with a barrier of less than 1 kcal•mol −1 , leading to the intermediate, 4,2 IB2.The radical disappears upon formation of the alcohol intermediate, 4,2 IB2, en route to the O-dealkylation product, 4,2 PB.The last step of dealkylation happens away from the P450 active site and can be catalyzed by a solvent water molecule.Thus, when we add a bridging water molecule to the substrate alcohol position a dealkylation barrier of 11.5 kcal•mol −1 is obtained for proton relay, from the alcohol group to the carbonyl of the ester bond.This breaks the ester bond and releases ortho-phthalic acid in a process that is exothermic by 5.8 kcal•mol −1 .Subsequent inclusion of water molecules in the model may even bring the barrier further down, as evidenced in previous studies on P450 catalyzed dealkylation [70].

Phthalate Epoxidation and Aromatic Hydroxylation
The final two pathways of phthalate oxidation by P450 CpdI that were considered are substrate epoxidation and aromatic hydroxylation, which both start with an electrophilic attack of the oxo group on one of the carbon atoms of the aromatic ring to form a radical intermediate, 2,4 IC.From this intermediate, the landscape is seen to bifurcate into two directions (Figure 7).Firstly, a ring-closure transition state (TS rcC ) leads to epoxide product complexes (PC).Secondly, from IC the ipso-proton can transfer from the benzene ring to one of the nitrogen atoms of porphyrin to form intermediate ID in a highly exothermic reaction step and without a reaction barrier.A small proton-reshuttle barrier TS PT of less than 1 kcal•mol −1 transfers the proton to the oxygen atom to form phenol products (PD).An alternative proton transfer pathway from 4,2 ID leads, via a significantly higher barrier, to ketone products.The electrophilic attack of the oxo group on the ortho carbon atom, with respect to the carboxylic acid substituents of the aromatic ring, passes transition states 4,2 TSC, at a cost of 20.9 (22.8) kcal•mol −1 on the doublet (quartet) spin states.These transition states have elongated Fe-O bond lengths of 1.717 (1.709) Å and long O-C bond lengths of 1.793 (1.828) Å for 4 TSC ( 2 TSC), respectively.The potential energy surface is broad and hence a small imaginary frequency of well below i600 cm −1 is found.Geometrically, the structures are upright, with an Fe-O-C angle of 136° (132°) and match earlier calculated structures of the aromatic hydroxylation mechanism by P450 CpdI using different substrates [36][37][38][39]66,71].
After the transition state, the systems relax to radical intermediates ( 4,2 IC) with spin densities of ρFeO = 2.31 and ρSubH = 1.01 in the quartet spin state.Both spin states remain degenerate along the pathway, from reactants to IC, as seen above for the other mechanisms as well and are higher in energy than reactants by 13.6 (11.8) kcal•mol −1 for 4 IC ( 2 IC).Similar to aliphatic rebound, the ring-closure barriers leading to epoxide products are much higher in energy in the quartet spin state than in the doublet spin state (cf.7.7 versus 2.5 kcal•mol −1 ).This is a result of promotion of an electron from the substrate into the π*xz orbital in the doublet spin state, whereas in the quartet spin state the electron moves to the higher energy σ*z 2 instead.Overall, epoxidation is a thermoneutral process and at our level of theory, it is slightly endothermic in the quartet spin state (by 4.3 kcal•mol −1 ), while it is slightly exothermic in the doublet spin state (by 3.9 kcal•mol −1 ).Structures along the epoxidation pathway are similar to those calculated before on alternative substrates [29][30][31][32][72][73][74][75].
Proton shuttle from the ipso-carbon atom to the porphyrin ring is virtually barrierless and leads, in highly exothermic process, to intermediate 4,2 ID and is followed by another small proton transfer barrier (TSPT) that gives ortho-phenol products (PD).As the barrier height for the conversion of the radical intermediates ( 4,2 IC) into proton-transfer intermediates ( 4,2 ID) is significantly smaller than the barrier for ring-closure, it can be concluded that substrate epoxidation is an unlikely process The electrophilic attack of the oxo group on the ortho carbon atom, with respect to the carboxylic acid substituents of the aromatic ring, passes transition states 4,2 TS C , at a cost of 20.9 (22.8) kcal•mol −1 on the doublet (quartet) spin states.These transition states have elongated Fe-O bond lengths of 1.717 (1.709) Å and long O-C bond lengths of 1.793 (1.828) Å for 4 TS C ( 2 TS C ), respectively.The potential energy surface is broad and hence a small imaginary frequency of well below i600 cm −1 is found.Geometrically, the structures are upright, with an Fe-O-C angle of 136 • (132 • ) and match earlier calculated structures of the aromatic hydroxylation mechanism by P450 CpdI using different substrates [36][37][38][39]66,71].
After the transition state, the systems relax to radical intermediates ( 4,2 IC) with spin densities of ρ FeO = 2.31 and ρ SubH = 1.01 in the quartet spin state.Both spin states remain degenerate along the pathway, from reactants to IC, as seen above for the other mechanisms as well and are higher in energy than reactants by 13.6 (11.8) kcal•mol −1 for 4 IC ( 2 IC).Similar to aliphatic rebound, the ring-closure barriers leading to epoxide products are much higher in energy in the quartet spin state than in the doublet spin state (cf.7.7 versus 2.5 kcal•mol −1 ).This is a result of promotion of an electron from the substrate into the π* xz orbital in the doublet spin state, whereas in the quartet spin state the electron moves to the higher energy σ* z 2 instead.Overall, epoxidation is a thermoneutral process and at our level of theory, it is slightly endothermic in the quartet spin state (by 4.3 kcal•mol −1 ), while it is slightly exothermic in the doublet spin state (by 3.9 kcal•mol −1 ).Structures along the epoxidation pathway are similar to those calculated before on alternative substrates [29][30][31][32][72][73][74][75].
Proton shuttle from the ipso-carbon atom to the porphyrin ring is virtually barrierless and leads, in highly exothermic process, to intermediate 4,2 ID and is followed by another small proton transfer barrier (TS PT ) that gives ortho-phenol products (PD).As the barrier height for the conversion of the radical intermediates ( 4,2 IC) into proton-transfer intermediates ( 4,2 ID) is significantly smaller than the barrier for ring-closure, it can be concluded that substrate epoxidation is an unlikely process and will not be able to compete with aromatic hydroxylation under these conditions.However, it may come into play in a constraint substrate situation, such as an enzyme active site, where the substrate orientation stabilizes this process.

Discussion
To understand the regioselectivity patterns of phthalate activation by P450 CpdI, we calculated the mechanisms leading to aliphatic hydroxylation at the C 5 position of the aliphatic chain, O-dealkylation, aromatic hydroxylation of the ortho-carbon atom and epoxidation of the aromatic ring.In the next few sections, we will discuss the patterns and the consequences of phthalate activation by the P450s.

Regioselectivity of Phthalate Oxidation by P450 CpdI
Table 1 gives a summary of the rate-determining barriers for hydrogen atom abstraction from the C 5 position ( 4,2 TS A ), hydrogen atom abstraction from the C 1 position ( 4,2 TS B ) and electrophilic addition to the ortho-position at the aromatic ring ( 4,2 TS C ).It can be seen that all barriers at the ∆E + ZPE level of theory fall inside a window of 4.6 kcal•mol −1 , with the lowest one through 2 TS A at 18.2 kcal•mol −1 .This implies that the dominant reaction pathway ideally should lead to C 5 activation in the doublet spin state, but the other barriers for C 5 and C 1 activation are within 2 kcal•mol −1 and hence will be competitive.The same trends and transition state ordering is found at the free energy level of theory.The only difference between ∆E + ZPE and ∆G is a raise in the value of the latter, due to the addition of entropy.Nevertheless, the entropy effect appears to be similar for all transition states and pathways covered.Although the energy differences between aliphatic hydrogen atom abstraction at the C 5 and C 1 positions are very small and also close in energy to the electrophilic addition transition state in our model system, this does not necessarily mean that C 5 hydroxylation is the dominant process in the enzyme.Experimental studies on phthalate activation by a range of P450 isozymes showed that the product distributions were strongly dependent on the isozyme [49].Since each P450 isozyme has a characteristic substrate binding pocket [76], the substrate orientation inside P450 isozymes may be different and consequently the activation of the substrate will give other products.Accordingly, the work of Choi et al. [49] showed that in P450 3A1 and P450 2C12 , the dominant product is C 5 hydroxylation, whereas in P450 2D6 mostly O-dealkylation was found.However, in P450 3A4 , P450 2C6 and P450 3A5 , a mixture of products, originating from O-dealkylation and C 5 -hydroxylation, was found.Our computational modelling gives a small energy difference in the rate-determining barrier heights of C 5 hydrogen atom abstraction versus C 1 hydrogen atom abstraction (for O-dealkylation), with a small preference of about 1 kcal•mol −1 in favor of C 5 activation.Consequently, under ideal conditions, the dominant product should be C 5 hydroxylation, but, if that pathway is hindered, due to substrate positioning, alternative mechanisms may become available.To understand the preferences better, we took a P450 structure of a liver P450 isozyme from the protein databank [77],-i.e., the 1TQN pdb file [78], which is a P450 3A4 isozyme.The substrate was docked into the substrate-binding pocket, using SwissDock [79], and thereafter the active site was modified to a CpdI model by addition of an oxo group.Some plausible conformations of the substrate orientation in the P450 3A4 pocket are shown in Figure 8.We found two low-energy conformations as shown in Figure 8; one with the substrate in a pro-C 5 -hydroxylation conformation and one with the substrate in a pro-O-dealkylation conformation.In the pro-C 5 -hydroxylation binding conformation, the docked structure (Figure 8a) is in a conformation with the aliphatic chains in close approach to the iron(IV)-oxo group, in a crown-shaped orientation with the protons of the C 5 and C 3 positions of the hexyl group as well as the terminal hydrogen atom from the ethyl group (C 8 ) pointing towards the oxo group.In particular, distances found were: 2.10 Å for C 5 H-O, 2.06 Å for C 3 H-O and 3.07 Å for C 8 H-O.Clearly, the aliphatic group fits into the cavity nearby the iron(IV)-oxo species well and should lead to aliphatic hydroxylation products.This means that the most likely substrate activation positions in this binding configuration will be on the terminus of the aliphatic chain, i.e., the C 5 position for which we calculated a low-energy barrier height.On the other hand, the aromatic ring is positioned far away from CpdI and as such we do not expect significant aromatic hydroxylation products.
In the alternative conformation in Figure 8b, we positioned the C 1 carbon in close distance to the oxo group, while the aromatic ring and the alkyl chains were pointing away from the heme group.In this orientation, the ester bond is in hydrogen bonding interaction with the positive side chain of Arg212, which may facilitate the O-dealkylation process through charge stabilization.Moreover, in this binding position, O-dealkylation will be the dominant process and no aliphatic hydroxylation on the side chains will take place.
Consequently, substrate activation by the P450s is a subtle balance of thermochemical and kinetical possibilities with substrate binding and orientation that go hand-in-hand.To understand the thermochemistry better, in the next section we discuss aliphatic C-H bond activation in more detail.

Features of the Hydrogen Atom Abstraction Step
To understand the intrinsic properties related to the hydrogen atom abstraction step, we devised valence bond models to explain the features of the reaction pathway following procedures described previously [80][81][82].Figure 9 displays a valence bond (VB) description of the electronic configurations of reactants, hydrogen atom abstraction transition states and radical intermediates.Thus, in the reactants, CpdI has two 3-electron bonds along the Fe-O interaction, due to occupation of the πxz 2 πyz 2 π*xz 1 π*yz 1 with six electrons, which gives two unpaired electrons at the FeO group and a spin density of about 1 on Fe and O. CpdI has a third unpaired electron in a2u and the C-H bond of the substrate is the σCH bond with two electrons.Upon approach of CpdI on the substrate, the C-H bond orbital is broken, and the two electrons revert to atomic orbitals, i.e., 2pC and 1sH.At the same time, the 3-electron bond in the xz-plane breaks back to atomic orbitals, namely 2pO 2 3dxz 1 .One of the electrons from 2pO pairs up with the incoming hydrogen atom to form the O-H bonding orbital σOH.The second electron from 2pO is then promoted into the a2u orbital.We found two low-energy conformations as shown in Figure 8; one with the substrate in a pro-C 5 -hydroxylation conformation and one with the substrate in a pro-O-dealkylation conformation.In the pro-C 5 -hydroxylation binding conformation, the docked structure (Figure 8a) is in a conformation with the aliphatic chains in close approach to the iron(IV)-oxo group, in a crown-shaped orientation with the protons of the C 5 and C 3 positions of the hexyl group as well as the terminal hydrogen atom from the ethyl group (C 8 ) pointing towards the oxo group.In particular, distances found were: 2.10 Å for C 5 H-O, 2.06 Å for C 3 H-O and 3.07 Å for C 8 H-O.Clearly, the aliphatic group fits into the cavity nearby the iron(IV)-oxo species well and should lead to aliphatic hydroxylation products.This means that the most likely substrate activation positions in this binding configuration will be on the terminus of the aliphatic chain, i.e., the C 5 position for which we calculated a low-energy barrier height.On the other hand, the aromatic ring is positioned far away from CpdI and as such we do not expect significant aromatic hydroxylation products.
In the alternative conformation in Figure 8b, we positioned the C 1 carbon in close distance to the oxo group, while the aromatic ring and the alkyl chains were pointing away from the heme group.In this orientation, the ester bond is in hydrogen bonding interaction with the positive side chain of Arg 212 , which may facilitate the O-dealkylation process through charge stabilization.Moreover, in this binding position, O-dealkylation will be the dominant process and no aliphatic hydroxylation on the side chains will take place.
Consequently, substrate activation by the P450s is a subtle balance of thermochemical and kinetical possibilities with substrate binding and orientation that go hand-in-hand.To understand the thermochemistry better, in the next section we discuss aliphatic C-H bond activation in more detail.

Features of the Hydrogen Atom Abstraction Step
To understand the intrinsic properties related to the hydrogen atom abstraction step, we devised valence bond models to explain the features of the reaction pathway following procedures described previously [80][81][82].Figure 9 displays a valence bond (VB) description of the electronic configurations of reactants, hydrogen atom abstraction transition states and radical intermediates.Thus, in the reactants, CpdI has two 3-electron bonds along the Fe-O interaction, due to occupation of the  Therefore, the hydrogen atom abstraction barrier will be related to the strength of the σCH orbital that needs to be broken, the strength of the σOH orbital that is formed, the strength of the 3-electron πxz/π*xz-orbitals that need to be broken and the promotion energy from 2pO to a2u.Indeed, previous studies of ours have shown that the hydrogen atom abstraction barriers for a series of substrates correlates with the bond dissociation energy (BDE) of the C-H bond that was broken [26,63,83,84].To gain insight into the relative C-H bond strengths of the various aliphatic positions of the phthalate substrate, we calculated the BDECH for all aliphatic positions as the diabatic energy between the substrate and the sum of the substrate minus a hydrogen atom and a hydrogen atom, see Figure 10.The data in Figure 10 shows that most C-H bonds of secondary carbon atoms have a strength that ranges from 98.7 kcal•mol −1 (for the C 3 and C 4 positions) to 100.3 kcal•mol −1 (for the C 1 position).By contrast, the primary carbon atoms-i.e., the terminal methyl groups-have a C-H bond strength of 103.3 kcal•mol −1 (at C 8 ) and 103.6 kcal mol −1 (at C 6 ), whereas the tertiary carbon atom C 2 has a C-H bond strength of only 97.6 kcal•mol −1 .As the full set of BDECH values spans a range of only 5 kcal•mol −1 , it is obvious that a large mixture of products can be expected for aliphatic hydroxylation of the alkyl chains.Moreover, substrate positioning will have a major contribution to the product formation and a low-energy binding conformation may block certain C-H bonds from being activated by the enzyme.For instance, the substrate binding position in Figure 9b gave only one specific C-H bond in close distance to the heme-namely C 1 -and consequently will lead to dominant O-dealkylation.On the other hand, the structure shown in Figure 9a will lead to a mixture of aliphatic hydroxylation products, probably mostly at C 3 and C 5 .Therefore, the hydrogen atom abstraction barrier will be related to the strength of the σ CH orbital that needs to be broken, the strength of the σ OH orbital that is formed, the strength of the 3-electron π xz /π* xz -orbitals that need to be broken and the promotion energy from 2p O to a 2u .Indeed, previous studies of ours have shown that the hydrogen atom abstraction barriers for a series of substrates correlates with the bond dissociation energy (BDE) of the C-H bond that was broken [26,63,83,84].To gain insight into the relative C-H bond strengths of the various aliphatic positions of the phthalate substrate, we calculated the BDE CH for all aliphatic positions as the diabatic energy between the substrate and the sum of the substrate minus a hydrogen atom and a hydrogen atom, see Figure 10.Therefore, the hydrogen atom abstraction barrier will be related to the strength of the σCH orbital that needs to be broken, the strength of the σOH orbital that is formed, the strength of the 3-electron πxz/π*xz-orbitals that need to be broken and the promotion energy from 2pO to a2u.Indeed, previous studies of ours have shown that the hydrogen atom abstraction barriers for a series of substrates correlates with the bond dissociation energy (BDE) of the C-H bond that was broken [26,63,83,84].To gain insight into the relative C-H bond strengths of the various aliphatic positions of the phthalate substrate, we calculated the BDECH for all aliphatic positions as the diabatic energy between the substrate and the sum of the substrate minus a hydrogen atom and a hydrogen atom, see Figure 10.The data in Figure 10 shows that most C-H bonds of secondary carbon atoms have a strength that ranges from 98.7 kcal•mol −1 (for the C 3 and C 4 positions) to 100.3 kcal•mol −1 (for the C 1 position).By contrast, the primary carbon atoms-i.e., the terminal methyl groups-have a C-H bond strength of 103.3 kcal•mol −1 (at C 8 ) and 103.6 kcal mol −1 (at C 6 ), whereas the tertiary carbon atom C 2 has a C-H bond strength of only 97.6 kcal•mol −1 .As the full set of BDECH values spans a range of only 5 kcal•mol −1 , it is obvious that a large mixture of products can be expected for aliphatic hydroxylation of the alkyl chains.Moreover, substrate positioning will have a major contribution to the product formation and a low-energy binding conformation may block certain C-H bonds from being activated by the enzyme.For instance, the substrate binding position in Figure 9b gave only one specific C-H bond in close distance to the heme-namely C 1 -and consequently will lead to dominant O-dealkylation.On the other hand, the structure shown in Figure 9a will lead to a mixture of aliphatic hydroxylation products, probably mostly at C 3 and C 5 .The data in Figure 10 shows that most C-H bonds of secondary carbon atoms have a strength that ranges from 98.7 kcal•mol −1 (for the C 3 and C 4 positions) to 100.3 kcal•mol −1 (for the C 1 position).By contrast, the primary carbon atoms-i.e., the terminal methyl groups-have a C-H bond strength of 103.3 kcal•mol −1 (at C 8 ) and 103.6 kcal•mol −1 (at C 6 ), whereas the tertiary carbon atom C 2 has a C-H bond strength of only 97.6 kcal•mol −1 .As the full set of BDE CH values spans a range of only 5 kcal•mol −1 , it is obvious that a large mixture of products can be expected for aliphatic hydroxylation of the alkyl chains.Moreover, substrate positioning will have a major contribution to the product formation and a low-energy binding conformation may block certain C-H bonds from being activated by the enzyme.For instance, the substrate binding position in Figure 9b gave only one specific C-H bond in close distance to the heme-namely C 1 -and consequently will lead to dominant O-dealkylation.On the other hand, the structure shown in Figure 9a will lead to a mixture of aliphatic hydroxylation products, probably mostly at C 3 and C 5 .

Materials and Methods
Active site models were selected in accordance with previous research [85,86] as an iron embedded in protoporphyrin IX without side chains and linked to SH − as a mimic of cysteinate and with oxo in the distal position.The phthalate substrate (2-ethylhexyl-phthalate) was selected in the charged-neutral state and all complexes of CpdI with the substrate were calculated in the lowest lying doublet and quartet spin states.
Calculations were performed in Gaussian-09 [87], using density functional theory methods that have been calibrated and benchmarked against experimental rate constants previously [88][89][90][91].In general, the UB3LYP [92,93] hybrid density functional was used for all calculations.Geometry optimizations, frequencies and reaction coordinate scans were done with a modest LACVP basis set with core potential on iron [94] and 6-31 G* on the rest of the atoms [95]: basis set BS1.More accurate energies were obtained through a single point calculation of the optimized geometry, using an LACV3P+ basis set with core potential on iron and 6-311 + G* on the rest of the atoms: basis set BS2, as well as an implicit solvent model mimicking the dielectric constant of water [96].

Conclusions
A computational study on phthalate binding and activation by cytochrome P450 liver enzymes was presented here.Initial calculations used active site models of P450 Compound I in its reaction with the ester of phthalate and 2-ethylhexanol, and patterns for C 5 hydroxylation, O-dealkylation, aromatic ring epoxidation and ortho-aromatic ring hydroxylation patterns were reported.Aliphatic hydroxylation and O-dealkylation started with hydrogen atom abstraction followed by rebound to give an alcohol product complex.In O-dealkylation, we found the alcohol product to react with an assisted solvent water molecule by O-dealkylation.Epoxidation and aromatic hydroxylation mechanisms are electrophilic and lead to a radical intermediate where the oxo group forms a single bond with one carbon atom of the aromatic ring.This radical can form the epoxide through ring-closure; however, that pathway incurs a significant barrier height.A lower energy pathway leads through proton shuttle from ipso-proton to heme and back to the phenolate to give phenol products.All pathways were analyzed and rationalized with thermochemical, valence bond and kinetic models.

Figure 1 .
Figure 1.Active site structure of the substrate-bound, resting state of P450 as taken from the 4L40 protein databank (pdb) file and an extract of the catalytic cycle of P450 enzymes.

Figure 1 .
Figure 1.Active site structure of the substrate-bound, resting state of P450 as taken from the 4L40 protein databank (pdb) file and an extract of the catalytic cycle of P450 enzymes.

Figure 3 .
Figure 3. P450 model and reaction mechanisms (with nomenclature) studied in this work.

Figure 4 .
Figure 4. Relevant molecular orbitals of P450 CpdI and orbital occupation in the quartet and doublet spin-states.

Figure 4 .
Figure 4. Relevant molecular orbitals of P450 CpdI and orbital occupation in the quartet and doublet spin-states.

Figure 5 .
Figure 5. Potential energy landscape (B3LYP/BS1 optimized) of phthalate hydroxylation by P450 CpdI at the C 5 position.Relative energies contain zero-point corrections and are taken from B3LYP/BS2//B3LYP/BS1 in kcal•mol −1 .Optimized geometries give bond lengths in angstroms, the Fe-O-C angle in degrees and the imaginary frequency in the transition state in cm −1 .

Figure 5 .
Figure 5. Potential energy landscape (B3LYP/BS1 optimized) of phthalate hydroxylation by P450 CpdI at the C 5 position.Relative energies contain zero-point corrections and are taken from B3LYP/BS2//B3LYP/BS1 in kcal•mol −1 .Optimized geometries give bond lengths in angstroms, the Fe-O-C angle in degrees and the imaginary frequency in the transition state in cm −1 .

Figure 6 .
Figure 6.Potential energy landscape (B3LYP/BS1 optimized) of phthalate O-dealkylation by P450 CpdI.Relative energies contain zero-point corrections and are taken from B3LYP/BS2//B3LYP/BS1 in kcal•mol −1 .Optimized geometries give bond lengths in angstroms, the Fe-O-C angle in degrees and the imaginary frequency in the transition state in cm −1 .

Figure 6 .
Figure 6.Potential energy landscape (B3LYP/BS1 optimized) of phthalate O-dealkylation by P450 CpdI.Relative energies contain zero-point corrections and are taken from B3LYP/BS2//B3LYP/BS1 in kcal•mol −1 .Optimized geometries give bond lengths in angstroms, the Fe-O-C angle in degrees and the imaginary frequency in the transition state in cm −1 .

Inorganics 2017, 5 , 77 9 of 18 Figure 7 .
Figure 7. Potential energy landscape (B3LYP/BS1 optimized) of epoxidation and aromatic hydroxylation of phthalate by P450 CpdI.Relative energies contain zero-point corrections and are taken from B3LYP/BS2//B3LYP/BS1 in kcal•mol −1 .Optimized geometries give bond lengths in angstroms, the Fe-O-C angle in degrees and the imaginary frequency in the transition state in cm −1 .

Figure 7 .
Figure 7. Potential energy landscape (B3LYP/BS1 optimized) of epoxidation and aromatic hydroxylation of phthalate by P450 CpdI.Relative energies contain zero-point corrections and are taken from B3LYP/BS2//B3LYP/BS1 in kcal•mol −1 .Optimized geometries give bond lengths in angstroms, the Fe-O-C angle in degrees and the imaginary frequency in the transition state in cm −1 .

Figure 9 .
Figure 9. Valence bond description of hydrogen atom abstraction from the substrate.Dots represent electrons and a line between two dots is a bond with two electrons.

Figure 9 .
Figure 9. Valence bond description of hydrogen atom abstraction from the substrate.Dots represent electrons and a line between two dots is a bond with two electrons.

Figure 9 .
Figure 9. Valence bond description of hydrogen atom abstraction from the substrate.Dots represent electrons and a line between two dots is a bond with two electrons.

Figure 10 .
Figure 10.BDE CH values (in kcal•mol −1 ) of aliphatic C-H positions of the phthalate substrate.

Table 1 .
Calculated rate determining barrier heights for various oxidation reactions.
One of the electrons from 2p O pairs up with the incoming hydrogen atom to form the O-H bonding orbital σ OH .The second electron from 2p O is then promoted into the a 2u orbital.
1 π* yz 1 with six electrons, which gives two unpaired electrons at the FeO group and a spin density of about 1 on Fe and O. CpdI has a third unpaired electron in a 2u and the C-H bond of the substrate is the σ CH bond with two electrons.Upon approach of CpdI on the substrate, the C-H bond orbital is broken, and the two electrons revert to atomic orbitals, i.e., 2p C and 1s H .At the same time, the 3-electron bond in the xz-plane breaks back to atomic orbitals, namely 2p O 2 3d xz 1 .