Computational Study of Engineered Cytochrome P450-Catalyzed CâH

Subscriber access provided by READING UNIV

Article

Computational Study of Engineered Cytochrome P450-Catalyzed C-H Amination: the Origin of the Regio- and Stereoselectivity Zhe Li, D. Jean Burnell, and Russell Jaye Boyd J. Phys. Chem. B, Just Accepted Manuscript • DOI: 10.1021/acs.jpcb.7b10256 • Publication Date (Web): 13 Nov 2017 Downloaded from http://pubs.acs.org on November 18, 2017

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

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

Page 1 of 27

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

The Journal of Physical Chemistry

Computational Study of Engineered Cytochrome P450catalyzed C-H Amination: the Origin of the Regio- and Stereoselectivity

Zhe Li,* D. Jean Burnell, and Russell J. Boyd*

Department of Chemistry, Dalhousie University, P. O. Box 15000, Halifax, Nova Scotia B3H 4R2, Canada;

Email address: [email protected], [email protected]

1

ACS Paragon Plus Environment


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

Abstract: Cytochrome P450 enzymes were recently engineered to catalyze the C-H amination reaction of aryl sulfonyl azides with excellent regio- and stereoselectivity (Arnold and coworkers JACS, 2014, 136, 15505). The mechanism of this reaction was studied by quantum mechanical (QM)/molecular mechanical (MM) calculations in this work. The C-H activation is found to be a stepwise process consisting of hydrogen abstraction (H-abstraction) of the reactive C-H bond by an iron nitrenoid cofactor to produce the biradical intermediate and the subsequent radical rebinding to form the final product. The rate of the rotation of the carbon radical center was estimated to be much faster than that of radical rebinding, which implies that the H-abstraction does not determine the stereoselectivity. For mutant A, the H-abstraction step has a barrier of 16.7 kcal/mol, which is 3.0 kcal/mol higher than that of the following radical rebinding step. The H-abstraction step determines the regioselectivity but the radical rebinding step determines the stereoselectivity. Barriers of these two steps are 16.1 and 27.5 kcal/mol, respectively, for mutant B. It is different from mutant A in that the radical rebinding step has the higher barrier and determines both the regio- and stereoselectivity. The initial distances between the hydrogens of reactive C-H bonds and the iron nitrenoid were found to not correlate with their reactivities. The calculated barriers are qualitatively consistent with the experimentally observed regio- and stereoselectivity with the exception of the stereoselectivity of mutant B. The lower barriers of mutant A presumably come from the stabilization effect of the H-bond between G265 and the sulfone O. This H-bond does not exist in mutant B. The conformation of the protein backbone, with the exception of the active site, does not change much (RMSD < 0.05) along the reaction pathway.

2


Page 2 of 27

Page 3 of 27

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


Introduction C-H bond functionalization is a powerful tool for the synthesis of complex organic compounds due to the ubiquity of C-H bonds. One important type of C-H functionalization is the transformation of C-H bonds to C-N bonds because many pharmaceuticals and natural compounds contain a nitrogen moiety. Enzyme catalysts are much desired for this type of reaction considering that enzymatic reactions have high efficiency, excellent selectivity, and react under mild conditions.1 A major challenge in enzyme-catalyzed C-H functionalization is how to control the regio- and stereoselectivity. Cytochrome P450 is a superfamily of enzymes that catalyze oxidation reactions of organic compounds. The wild type P450 can react with O2 to form the highly reactive iron-oxo intermediate known as the compound I.2 Recently, C-H amination reactions catalyzed by engineered variants of P450 enzymes have been reported.3-5 Arnold and workers disclosed the P450-catalyzed intramolecular amination of aryl sulfonyl azide 1 to produce sultam 2 or 3, depending on the mutation of the enzyme (Scheme 1).5 This enzyme is supposed to form an iron nitrenoid active species with the substrate to catalyze the C-H amination. This reaction provides a new direction for developing novel and efficient C-H amination reactions.6 In order to improve and extend P450-catalyzed amination of azides, the reaction mechanism and the origin of the regio- and stereoselectivity requires elucidation.

3



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

Scheme 1. C-H Amination Catalyzed by Engineered P450

Preliminary experimental results provided clues to the mechanism of the P450catalyzed reaction (Scheme 2).7 The initialization of the catalytic cycle is the reduction of the Fe(III) of the active site to Fe(II) by NADPH (dihydronicotinamide adenine dinucleotide phosphate), which is a cofactor required for the reaction to proceed. The reduced Fe(II) heme could react with azide 1 to produce Fe(IV) nitrenoid after the release of dinitrogen. The Fe(IV) nitrenoid is the active species for the hydrogen abstraction (Habstraction) reaction occurring at the carbon α or β to the phenyl ring of the substrate. The H-abstraction step generates the carbon radical intermediate. The rebinding of the radical produces the new C-N bond of the final product. In the original experimental report, Arnold and coworkers hypothesized that the regioselectivity is controlled by preorganization of the substrate such that a single C-H bond is close to the iron nitrenoid. This would mean that the H-abstraction step should determine the regioselectivity. However, the authors conceded that a significant influence of the radical rebinding step on the regioselectivity could not be excluded. The radical rebinding can determine the regioselectivity if the activation Gibbs energy of the radical rebinding step is higher than the H-abstraction step, and the formation of the radical intermediate is reversible. 4


Page 4 of 27

Page 5 of 27

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


Although the H-abstraction step was suggested to be rate determining due to the KIE (kinetic isotope effect) values (kH/kD) of 2.8-3.0, this value is much lower than the typical KIE of stepwise C-H abstraction reactions. For example, the KIE values of C-H amination reactions catalyzed by Ru-prophyrins are 6-12.8,9 The low value of the KIE indicates that the H-abstraction step has a similar activation barrier to another step in the catalytic cycle.10,11 The mechanism needs to be elucidated computationally in order to explain how the regioselectivity is determined. Contemporary computational methods are a powerful tool to study the mechanism of enzymatic reactions. Due to the large size of the enzymatic system, the state-of-art calculation method is the combination of quantum mechanical (QM) and molecular mechanical (MM) methods.12-14 In a typical QM/MM calculation, the active site and the substrate involved in bond breaking and forming is calculated by an accurate QM method. The QM region is recommended to include at least atoms that are three bonds away from the atoms that have bonds breaking or forming.12 The remainder of the enzyme and the solvents are calculated by a lower level, but much faster, MM method. There have been a number of QM/MM studies on the selectivity of P450-catalyzed hydroxylation reactions. For example, Shaik’s group studied the regio- and enantioselectivity of P450-catalyzed hydroxylation of fatty acids.15,16 Harvey and coworkers studied the regioselectivity of drug metabolism in P450 2C9 that is one of the major drug-metabolizing P450 isoforms.17 The hydroxylation of unactivated methylene catalyzed by P450 PikC that is engineered from pikromycin natural biosynthetic pathway has been examined.18 However, computational studies on P450 with an iron nitrenoid cofactor that can functionalize C-H bonds are lacking, although there were studies on iron nitrenoid mediated C-H activations

5



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

without P450.19-21 In this work, the mechanism of P450-catalyzed C-H amination was studied by QM/MM calculations. Based on this study, the regio- and stereoselectivity of this reaction can be explained.

Scheme 2. Proposed Mechanism of P450-catalyzed C-H Amination7

Computational Methodology Structure Preparation The initial structure of the engineered P450 enzyme was taken from the X-ray structure of mutant A (PDB code: 4WG2).5 The structure is a trimer with three identical chains A, B and C. Only chain A was kept for further modeling. Chains B and C, crystal 6


Page 6 of 27

Page 7 of 27

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


waters and sulfate anions were removed. Substrate 1 was added to the active site by using Pymol.22 Hydrogen atoms were added to the structure of the PDB file by using the H++ server to predict the protonation states of acidic and basic amino acid residues.23 AMBER-type molecular mechanics parameters were generated for the heme, the axial serine residue and the substrate in the active site by using the MCPB tools.24 Force constants and RESP charges were calculated for the heme, the axial serine and the substrate using Gaussian 09 at the level of B3LYP/6-31G(d).25 Parameters for the amino acid residues of the enzyme were from the AMBER 14 force field (ff14sb). The enzyme was solvated in a pre-equilibrated cubic TIP3P water box with a side length of 10.0 Å by using the tleap tool in AMBER. The system was neutralized by adding explicit counter ions (Na+ and Cl-). Molecular Dynamics Simulation The whole system of the enzyme with water was first equilibrated through a series of minimization and short MD runs: 1) the positions of water and ions were minimized while the enzyme and the substrate were restrained by the harmonic force constant of 500 kcal·mol·Å-2, which was enough to fix the position of these atoms; 2) all atoms of the system were fully minimized; 3) the system was heated rapidly from 0K to 300K in a 20 ps canonical ensemble (NVT) MD simulation in which the number of atoms (N), the volume (V) and the temperature (T) were conserved. The short heating time was chosen to avoid the generation of vacuum bubbles because water molecules added by the tleap tool were loose. Here, isothermal-isobaric ensemble (NPT) was not used because the control of pressure (barostat) is inaccurate at low temperature; 4) the enzyme and substrate were restrained with a weak harmonic potential of 2.0 kcal·mol·Å-2 in a 100 ps

7



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

NPT MD; 5) the whole system was equilibrated in 500 ps NPT MD to achieve a stable density (see Supporting Information). After the system had equilibrated, a 190 ns productive NVT MD was performed. Here NVT was chosen because it is more efficiently implemented in AMBER and both NVT and NPT can give us a reliable sampling of conformations for the system in solution. In all MD simulations, the time step was 2 fs and the SHAKE algorithm was effected to constrain all bond lengths involving hydrogens.26 All MD simulations were performed using the AMBER 14 software package.27 QM/MM calculation A snapshot was selected from conformations that had close distances between the reactive hydrogens on the substrate and the N atom of the iron nitrenoid of the MD trajectory. This is because it was found that QM/MM energy profiles calculated from MD structures that have large initial distances between reacting species give unrealistically high activation energies.17 The structure of this snapshot was initially optimized using the conjugate gradient method with AMBER software to avoid large forces in the subsequent optimization by Gaussian 09. After the initial optimization, water molecules that were within 3 Å of the protein or within 12 Å of the core residues, i.e., the heme, the substrate, and the axial serine residue of the heme, were retained. Other waters and neutralizing ions were removed to reduce the computational cost of the QM/MM calculations. All QM/MM calculations were performed using the ONIOM method in the Gaussian 09 software package.25 The QM region included the substrate, the heme, and the –CH2O- part of the axial serine ligand. Atoms in the QM layer were calculated by B3LYP density functional theory (DFT) method with the 6-31G(d) basis set for nonmetal

8


Page 8 of 27

Page 9 of 27

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


atoms and the Lanl2dz basis set for Fe; all other atoms were calculated by AMBER force field in the process of geometry optimization. Per Seigbahn’s group found that the B3LYP with reduced Hartree-Fock (HF) exchange of 15% calculated the binding energies of CO, NO and O2 to heme more accurately than the normal B3LYP.28 The performance of the modified B3LYP method was not examined because the mechanism in this paper does not include association or disassociation of the substrate to the heme. The TAO package was used to prepare the input file for Gaussian.29 In the QM/MM optimization, only atoms within 6 Å of core residues were allowed to move, other atoms were kept frozen. Frequency calculations were performed at the same level as the geometry optimization. Transition states were confirmed according to the vibrational mode of their one and only imaginary negative frequency. Other stable structures do not have an imaginary frequency. The thermodynamic correction for the gas-phase Gibbs energy (TCG) was obtained in the frequency calculation. Single point energies were calculated by B3LYP/6-311+G(2d,p) method for the QM layer and AMBER force field for all the other atoms. Only mechanical embedding was used in all ONIOM calculations. Single point energies using electronic embedding were found to have small impact on the activation barriers (within 3 kcal/mol) and did not change the relative barrier trends between different pathways (see supporting information). Empirical dispersion correction with Becke-Johnson damping was added to single point energies.30 Gibbs energies were calculated by adding TCG to single point energies. All the energies reported in this paper are Gibbs energies. Results and Discussion QM calculation of the model reaction

9



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

In order to obtain the starting geometries and favored electronic spin states for the intermediates and transition states for the QM/MM calculations, a simple model system was studied in gas phase by the same QM method as the high layer in the QM/MM calculation. The porphyrin in P450 was simplified to porphin, and the axial serine residue was simplified to a methoxide ligand (Figure 1). The ground state of the model iron nitrenoid compound is triplet with an energy that is 3.0 kcal/mol lower than that of the quintet state. The subsequent C-H activation process could be either stepwise or concerted. Concerted C-H insertion to rhodium nitrenoid has been proposed in a Rh-catalyzed C-H amination reaction.31 A concerted CH activation could also be a possible explanation for the low experimental KIE values (2.8 and 3.0) of P450 mutant A and mutant B.5 However, searches for concerted C-H insertion transition states were unsuccessful. The stepwise C-H functionalization consists of two steps: H-abstraction and radical rebinding. The triplet transition state of α-C-H Habstraction (3MTS1-2) has a barrier of 16.1 kcal/mol. The following C-N rebinding step has a barrier of 25.5 kcal/mol relative to the radical intermediate 3M2. For the β-C-H functionalization pathway, the barrier of H-abstraction (3MTS1-4) is 4.3 kcal/mol lower than that of α-C-H functionalization, and the barrier of C-N rebinding is 3.0 kcal/mol lower (3MTS4-3).

10


Page 10 of 27

Page 11 of 27

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


Figure 1. Gibbs energy profile of the model amination reaction catalyzed by iron porphin (kcal/mol).

Compared to the triplet pathway, the transition state of α- and β-C-H hydrogen abstractions in the quintet pathway (5MTS1-2 and 5MTS1-4) have barriers of 29.2 and 25.1 kcal/mol relative to the ground state 3M1. These high barriers indicate that the quintet pathway is not favored compared to the triplet pathway. As a result, only the triplet was considered in the QM/MM calculations of the full enzyme systems.

11



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

For both the α and β-C-H functionalization of the triplet pathways, the radical rebinding steps have higher barriers than the first H-abstraction step. This indicates that the regioselectivity is determined by the radical rebinding step, at least for the model system. The following QM/MM calculations for the full enzymes revealed whether this is the case for the real reactions. QM/MM calculation of the full enzyme In QM/MM calculations, the full substrate with the n-propyl substituent on the phenyl ring was calculated because this substituent influences the reactivity and regioselectivity of the enzymes in experiments.5 The active sites of the optimized reactant complexes for mutant A and B, i.e. A1 and B1, are shown in Figure 2. There is a hydrogen bond between G265 and one sulfone O atom of the substrate in mutant A. The average length of this H-bond during the MD simulation is 2.61 Å. In contrast to mutant A, the similar H-bond does not exist in mutant B. This is true not only for the selected snapshot, but also for most conformations generated in the whole MD simulations. The average distance between the H atom of G265 and the sulfone O atom is 6.94 Å for mutant B. The electronic energies of the quintet states of A1 and B1 were calculated after geometry optimization. The energies of 5A1 and 5B1 are +13.5 kcal/mol and +12.7 kcal/mol relative to triplet A1 and B1, respectively. These results indicate that triplet reaction pathways are still favored in full enzymes for mutant A and B. Many snapshots in the MD simulation have close distances between the active α or β hydrogens and the iron nitrenoid. The average H-N distances from the MD simulation are shown in Figure 3. The two hydrogens of the α and β positions are no longer equivalent because of the chiral environment of the enzyme. The H-N distances

12


Page 12 of 27

Page 13 of 27

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


for α-HR and α-HS (subscripts R and S denote pro-R and pro-S configurations, respectively) are 3.54 Å and 2.87 Å, respectively, for mutant A. These values are shorter than the H-N distances for β-HR and β-HS, which are observed to be functionalized by mutant A. This shows that the H-N distances of the hydrogens in the reactant complexes do not correlate directly with their reactivity. Just because the initial positions of the βhydrogens are relatively distant to the iron nitrenoid on average does not mean the activation barriers of breaking these β-C-H bonds are high. The activation barriers of different reaction pathways should be calculated to explain the different selectivity between mutants A and B.

Figure 2. The structures of the reactant complexes of mutant A, A1 (left) and mutant B, B1 (right). Atoms in QM region are shown as sticks, and amino acid residues within 3.5 Å of the substrate, but part of the MM region, are shown as wires (unit of length: Å).

13



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

The Gibbs energy profile of mutant A-catalyzed C-H amination is shown in Figure 4. In the pathway of α-H activation, the barrier of H-abstraction of α-HS is 17.5 kcal/mol, which is 1.1 kcal/mol lower than that of α-HR. The transferring hydrogen is almost in line between the C and N atoms in both ATSA1-2 and ATSB1-2 with C-H-N angles of 157.1o and 155.6o, respectively (Figure 5). The intermediate A2 is formed after the H-abstraction step with a slightly lower energy of -0.7 kcal/mol than the reactant complex A1. In the pathway of β-H activation, the barriers of H-abstraction of β-HR and β-HS are 16.7 kcal/mol (ATSA1-4) and 28.5 kcal/mol (ATSB1-4), respectively. The positions of the transferring H are more linear than those of α-H activation, with C-H-N angles of 170.3o and 161.4o, respectively. The H-bond between G265 and the sulfone O exists in all of these four H-abstraction transition states. The Gibbs energy of the intermediate A4 is +2.8 kcal/mol, which is higher than that of A2. This is expected because the α carbon radical is more stabilized by the phenyl ring than the β radical. This is also the same as the case of the model system. ATSA1-4 has the lowest barrier among the four H-abstraction transition states. Although this supports the observed β selectivity of mutant A, the barriers of the radical rebinding step need to be compared to draw the conclusion.

Figure 3. Distances between the N atom of the iron nitrene and hydrogens ([Fe] = P450, DH-N, unit: Å)

14


Page 14 of 27

Page 15 of 27

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


Figure 4. Gibbs energy profile of mutant A-catalyzed C-H amination (kcal/mol).

Intermediates A2 or A4 can rebind to the iron nitrenoid with either side of the carbon radical via two diastereoisomeric transition states, each of which leads to an enantiomer of the product. The barriers of the rotation of the α- or β-carbon radicals of the model system were calculated to be 10.9 and 1.8 kcal/mol, respectively. The barrier of the α-carbon radical is close to the reported rotational barrier of α-benzyl radical, which is 11.0 kcal/mol calculated by G4 method.32 We assume that rotational barriers of the model system are good estimates of the full enzymes. These barriers are lower than the corresponding barriers of radical rebinding of both mutant A and B (see below). As a

15



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

result, the rotation of the α- or β-carbon radicals is much faster than the rate of subsequent radical rebinding, which indicates that the hydrogen abstraction step does not influence the stereochemistry of the final product. In order to explain the experimentally observed stereoselectivity, transition states of the radical rebinding step were located for α- or β-C-H-activations. For the α-C-Hactivation pathway, the barriers are 16.7 kcal/mol for ATSA2-3 and 30.4 kcal/mol for ATSB2-3. For the β-H-activation pathway, the transition state ATSA4-3 that leads to the observed product 2 has a barrier of +13.7 kcal/mol. The incipient C-N bond length of ATSA4-3 has the longest length of 2.24 Å of the four transition states in Figure 6, indicating that it is an early transition state. The barrier of the radical rebinding step is lower than that of the first H-abstraction step in either the α or the β C-H amination pathway. This is different from the model system. The favored α C-H amination pathway is consistent with the observed regioselectivity of the reaction. Here the H-abstraction step determines the regioselectivity, but the radical rebinding step determines the stereoselectivity. The above calculations show that the rate determining step of the reaction of mutant A is the H-abstraction step. However, this is not the case of mutant B. The Gibbs energy profile of mutant B-catalyzed C-H amination is shown in Figure 7. The barrier of H-abstraction of α-HS is 19.9 kcal/mol (BTSA1-2), which is 3.8 kcal/mol higher than that of α-HR (BTSB1-2). The Gibbs energy of the biradical intermediate B2 is -6.3 kcal/mol. The barrier of the transition state BTSA2-3 that leads to the observed product is +27.5 kcal/mol, which is 2.1 kcal/mol higher than that of its diastereomer, BTSB2-3. This is at the limit of our computational method because the error of DFT calculations could be

16


Page 16 of 27

Page 17 of 27

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


several kcal/mol. In the β-C-H-activation pathway, the H-abstraction step of β-HR is favored and the barrier is +21.5 kcal/mol (BTSA1-4). The two transition states of the radical rebinding step have barriers of 30.5 (BTSA4-3) and 32.6 (BTSB4-3) kcal/mol, respectively. Here the barriers of BTSA4-3 and BTSB4-3 were counted relative to B2 not B1 because B2 has a lower Gibbs energy.33 For either α- or β-C-H activation, the barrier of radical rebinding is higher than that of H-abstraction. In addition, the rotational barrier of the radical center of B2 is estimated to be about 11 kcal/mol from the calculation of the model system. This is much lower than the barrier of the radical rebinding step. This indicates that both regio- and stereoselectivity are determined by the radical rebinding step of mutant B. The favored pathway is the α-C-H amination, which is consistent with experimental observation.

17



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

Figure 5. The geometry of H-abstraction transition states. Atoms in QM region are shown as sticks, and amino acid residues within 3.5 Å of the substrate, but part of the MM region, are shown as wires (unit of length: Å).

18


Page 18 of 27

Page 19 of 27

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


Figure 6. The geometry of radical rebinding transition states. Atoms in QM region are shown as sticks, and amino acid residues within 3.5 Å of the substrate, but part of the MM region, are shown as wires (unit of length: Å).

19



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

Figure 7. Gibbs energy profile of mutant B-catalyzed C-H amination (kcal/mol).

The high barriers of the radical rebinding step of mutant B are worth further investigation. They are 27.5 and 30.5 kcal/mol for α- and β-C-H amination, respectively, which are higher than their counterparts of either mutant A (16.7 and 13.7 kcal/mol) or the model system (25.5 and 22.5 kcal/mol). In other words, mutant B achieves αselectivity by elevating the barriers of the radical rebinding step and favors the radical rebinding transition states that form a five-membered ring. The structural difference between the radical rebinding transition states of mutant A and mutant B gives some clue

20


Page 20 of 27

Page 21 of 27

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


to this phenomenon. There is a H-bond between G265 and the sulfone O of the substrate in mutant A (Figure 6). This H-bond does not exist in mutant B because G265 is not close to the substrate. This means that the transition states of mutant B are less restricted than those of mutant A and are more like the model system in which the α-C-H amination is favored and the radical rebinding step has a higher barrier than the H-abstraction step. The geometries of the protein excluding the QM region of both mutant A and mutant B only have small changes throughout the reaction pathways. The RMSDs of the intermediates and transition states relative to the reactant complex are smaller than 0.05 Å of both mutant A and mutant B (Table S1 in the supporting information). This indicates that the catalytic ability of enzymes A and B does not depend on large conformational changes. In the experimental study, the KIE values of substrate 1 with mutant A and B were measured to be 2.8 and 3.0, respectively.5 These KIE values were interpreted by Arnold and coworkers to support the idea that the H-abstraction is the rate determining step. This is consistent with our calculated mechanism of mutant A but not mutant B. However, these KIE values around 3 are lower than reported KIE values of stepwise hydrogen transfer reactions. For instance, the KIE of the oxidation of benzyl alcohol with [FeIV=NTs](N4Py)]2+ (Ts = p-toluenesulfonyl, N4Py = N,N-bis(2-pyridyl-methyl)-Nbis(2-pyridyl) methylamine), which is a model compound for an iron enzyme, was measured to be 7.34 The KIE values of the oxidation of N,N-dimethylnitrosamine and of 7-methoxylcoumarin by P450 enzyme were measured to be 8.135 and 9.436, respectively. In order to study this further, the KIE of the H-abstraction step of the model reaction were calculated. The KIE values are 6.0 for the α-H and 6.7 for the β-H. Based on the

21



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

above data, the relatively low experimental KIE values of mutant A and B support that the H-abstraction step is partially rate determining with an activation barrier that is close to the radical rebinding step and the high KIE values of the H-abstraction step are masked by the low KIE values of the radical rebinding step.7,11 This indicates that the computational method has somewhat overestimated the barrier of BTSA2-3 of mutant B. Conclusions The P450-catalyzed intramolecular C-H amination of aryl sulfonyl azides is a versatile method to synthesize sultam compounds in a regio- and stereospecific way. In this reaction, the ion nitrenoid in the active site undergoes an intramolecular Habstraction step to produce the diradical intermediate, which transforms into the final product via the radical rebinding step. The face of the carbon radical intermediate can rotate along the C-C bond with low barriers that were estimated to be 10.9 and 1.8 kcal/mol for the α- or β-carbon radicals, respectively. The rate of the rotation is much faster than that of the subsequent radical rebinding step, which indicates that the Habstraction step does not influence the stereoselectivity. The barriers of the H-abstraction and the radical rebinding of mutant A are 16.7 and 13.7 kcal/mol, respectively. The Habstraction step determines the regioselectivity but the radical rebinding step determines the stereoselectivity. For mutant B, the barriers of the H-abstraction and the radical rebinding are 16.1 and 27.5 kcal/mol, respectively. With this mutant, the radical rebinding step determines both the regio- and stereoselectivity. The mean distances between the hydrogens of the reactive C-H bonds and the iron nitrenoid in the reactant complex do not correlate with the regio- and stereoselectivity to this reaction.

22


Page 22 of 27

Page 23 of 27

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


The stabilization effect of the H-bond between G265 and the sulfone O of the substrate presumably contributes to the relative low barriers of mutant A. This H-bond does not exist in mutant B, which leads to less restricted transition states like those without the enzyme in the model system. The average distances between the α or β hydrogens and the iron nitrenoid do not correlate with the regioselectivity. The conformation of the enzyme backbone, except the active site, does not change much along the reaction pathway. Clearly, future study and development of C-H amination reactions catalyzed by engineered P450 should be mindful of the previously underestimated significance of the radical rebinding step in determining the regio- and stereoselectivity.

ACKNOWLEDGEMENT: We are grateful to the Natural Sciences and Engineering Research Council of Canada (NSERC) for financial support of this research. Author D. Jean Burnell received funding from NSERC RGPIN/1480-2012. Authors Zhe Li and Russell J. Boyd received funding from NSERC RGPIN/9-2013. High performance computational facilities were provided by Compute Canada.

ASSOCIATED CONTENT: Cartesian coordinates of optimized structures of the model system, pdb files of optimized structures in QM/MM calculations, and detailed thermodynamic data. This material is available free of charge via the Internet at http://pubs.acs.org.

References and Notes:

23



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

(1) Upp, D. M.; Lewis, J. C. Selective C-H Bond Functionalization Using Repurposed or Artificial Metalloenzymes. Curr. Opin. Chem. Biol. 2017, 37, 48-55. (2) Ortiz de Montellano, P. R. Hydrocarbon Hydroxylation by Cytochrome P450 Enzymes. Chem. Rev. 2010, 110, 932-948. (3) Dydio, P.; Key, H. M.; Hayashi, H.; Clark, D. S.; Hartwig, J. F. Chemoselective, Enzymatic C-H Bond Amination Catalyzed by a Cytochrome P450 Containing an Ir(Me)-PIX Cofactor. J. Am. Chem. Soc. 2017, 139, 1750-1753. (4) McIntosh, J. A.; Heel, T.; Buller, A. R.; Chio, L.; Arnold, F. H. Structural Adaptability Facilitates Histidine Heme Ligation in a Cytochrome P450. J. Am. Chem. Soc. 2015, 137, 13861-13865. (5) Hyster, T. K.; Farwell, C. C.; Buller, A. R.; McIntosh, J. A.; Arnold, F. H. Enzyme-Controlled Nitrogen-Atom Transfer Enables Regiodivergent C-H Amination. J. Am. Chem. Soc. 2014, 136, 15505-15508. (6) Prier, C. K.; Zhang, R. K.; Buller, A. R.; Brinkmann-Chen, S.; Arnold, F. H. Enantioselective, Intermolecular Benzylic C-H Amination Catalysed by an Engineered Iron-haem Enzyme. Nat. Chem. 2017, 9, 629-634. (7) Singh, R.; Kolev, J. N.; Sutera, P. A.; Fasan, R. Enzymatic C(sp3)-H Amination: P450-Catalyzed Conversion of Carbonazidates into Oxazolidinones. ACS Catal. 2015, 5, 1685-1691. (8) Au, S. M.; Huang, J. S.; Yu, W. Y.; Fung, W. H.; Che, C. M. Aziridination of Alkenes and Amidation of Alkanes by Bis(tosylimido)ruthenium(VI) Porphyrins. A Mechanistic Study. J. Am. Chem. Soc. 1999, 121, 9120-9132. (9) Leung, S. K.; Tsui, W. M.; Huang, J. S.; Che, C. M.; Liang, J. L.; Zhu, N. Imido Transfer from Bis(imido)ruthenium(VI) Porphyrins to Hydrocarbons: Effect of Imido Substituents, C-H Bond Dissociation Energies, and RuVI/V Reduction Potentials. J. Am. Chem. Soc. 2005, 127, 16629-16640. (10) Kim, D.; Cha, G. S.; Nagy, L. D.; Yun, C. H.; Guengerich, F. P. Kinetic Analysis of Lauric Acid Hydroxylation by Human Cytochrome P450 4A11. Biochemistry 2014, 53, 6161-6172. (11) Northrop, D. B. Steady-State Analysis of Kinetic Isotope Effects in Enzymic Reactions. Biochemistry 1975, 14, 2644-2651. (12) Chung, L. W.; Sameera, W. M.; Ramozzi, R.; Page, A. J.; Hatanaka, M.; Petrova, G. P.; Harris, T. V.; Li, X.; Ke, Z.; Liu, F.; Li, H. B.; Ding, L.; Morokuma, K. The ONIOM Method and Its Applications. Chem. Rev. 2015, 115, 5678-5796. (13) Shaik, S.; Cohen, S.; Wang, Y.; Chen, H.; Kumar, D.; Thiel, W. P450 Enzymes: Their Structure, Reactivity, and Selectivity-Modeled by QM/MM Calculations. Chem. Rev. 2010, 110, 949-1017. (14) Senn, H. M.; Thiel, W. QM/MM Methods for Biomolecular Systems. Angew. Chem. Int. Ed. 2009, 48, 1198-1229. (15) Ramanan, R.; Dubey, K. D.; Wang, B.; Mandal, D.; Shaik, S. Emergence of Function in P450-Proteins: A Combined Quantum Mechanical/Molecular Mechanical and Molecular Dynamics Study of the Reactive Species in the H2O2-Dependent Cytochrome P450SPα and Its Regio- and Enantioselective Hydroxylation of Fatty Acids. J. Am. Chem. Soc. 2016, 138, 6786-6797. (16) Dubey, K. D.; Wang, B.; Shaik, S. Molecular Dynamics and QM/MM Calculations Predict the Substrate-Induced Gating of Cytochrome P450 BM3 and the

24


Page 24 of 27

Page 25 of 27

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


Regio- and Stereoselectivity of Fatty Acid Hydroxylation. J. Am. Chem. Soc. 2016, 138, 837-845. (17) Lonsdale, R.; Houghton, K. T.; Zurek, J.; Bathelt, C. M.; Foloppe, N.; de Groot, M. J.; Harvey, J. N.; Mulholland, A. J. Quantum Mechanics/Molecular Mechanics Modeling of Regioselectivity of Drug Metabolism in Cytochrome P450 2C9. J. Am. Chem. Soc. 2013, 135, 8001-8015. (18) Narayan, A. R.; Jimenez-Oses, G.; Liu, P.; Negretti, S.; Zhao, W.; Gilbert, M. M.; Ramabhadran, R. O.; Yang, Y. F.; Furan, L. R.; Li, Z.; Podust, L. M.; Montgomery, J.; Houk, K. N.; Sherman, D. H. Enzymatic Hydroxylation of an Unactivated Methylene C-H Bond Guided by Molecular Dynamics Simulations. Nat. Chem. 2015, 7, 653-660. (19) Kong, C.; Jana, N.; Jones, C.; Driver, T. G. Control of the Chemoselectivity of Metal N-Aryl Nitrene Reactivity: C-H Bond Amination versus Electrocyclization. J. Am. Chem. Soc. 2016, 138, 13271-13280. (20) Ren, Q. H.; Shen, X. Y.; Wan, J. Y.; Fang, J. H. Density Functional Theory Study of the Mechanisms of Iron-Catalyzed Intramolecular C-H Amination [1,2]Shift Tandem Reactions of Aryl Azides. Organometallics 2015, 34, 1129-1136. (21) Liu, Y.; Guan, X.; Wong, E. L.; Liu, P.; Huang, J. S.; Che, C. M. Nonheme Iron-Mediated Amination of C(sp3)-H Bonds. Quinquepyridine-Supported Iron-Imide/Nitrene Intermediates by Experimental Studies and DFT Calculations. J. Am. Chem. Soc. 2013, 135, 7194-7204. (22) Schrodinger, LLC: The PyMOL Molecular Graphics System, Version 1.8. 2015. (23) Anandakrishnan, R.; Aguilar, B.; Onufriev, A. V. H++ 3.0: Automating pK Prediction and the Preparation of Biomolecular Structures for Atomistic Molecular Modeling and Simulations. Nucleic Acids Res. 2012, 40, W537-541. (24) Li, P.; Merz, K. M., Jr. MCPB.py: A Python Based Metal Center Parameter Builder. J. Chem. Inf. Model. 2016, 56, 599-604. (25) Frisch, M. J. et. al. Gaussian 09 (Revision D.01); Gaussian, Inc.: Wallingford, CT, 2009. See supporting information for the full citation. (26) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. Numerical Integration of the Cartesian Equations of Motion of a System with Constraints: Molecular Dynamics of N-Alkanes. J. Comput. Phys. 1977, 23, 327-341. (27) Case, D. A.; Babin, V.; Berryman, J. T.; Betz, R. M.; Cai, Q.; Cerutti, D. S.; Cheatham, T. E.; Darden, T. A.; Duke, R. E.; Gohlke, H.; et al.: Amber 14, 2014. (28) Siegbahn, P. E. M.; Blomberg, M. R. A.; Chen, S.-L. Significant van der Waals Effects in Transition Metal Complexes. J. Chem. Theor. Comput. 2010, 6, 20402044. (29) Tao, P.; Fisher, J. F.; Shi, Q.; Vreven, T.; Mobashery, S.; Schlegel, H. B. Matrix Metalloproteinase 2 Inhibition: Combined Quantum Mechanics and Molecular Mechanics Studies of the Inhibition Mechanism of (4phenoxyphenylsulfonyl)methylthiirane and Its Oxirane Analogue. Biochemistry 2009, 48, 9839-9847. (30) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the Damping Function in Dispersion Corrected Density Functional Theory. J. Comput. Chem. 2011, 32, 1456-1465.

25



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

(31) Fiori, K. W.; Espino, C. G.; Brodsky, B. H.; Du Bois, J. A Mechanistic Analysis of the Rh-Catalyzed Intramolecular C–H Amination Reaction. Tetrahedron 2009, 65, 3042-3051. (32) Li, Z.; Bally, T.; Houk, K. N.; Borden, W. T. Variations in Rotational Barriers of Allyl and Benzyl Cations, Anions, and Radicals. J. Org. Chem. 2016, 81, 9576-9584. (33) Kozuch, S.; Shaik, S. How to Conceptualize Catalytic Cycles? The Energetic Span Model. Acc. Chem. Res. 2011, 44, 101-110. (34) Kumar, S.; Faponle, A. S.; Barman, P.; Vardhaman, A. K.; Sastri, C. V.; Kumar, D.; de Visser, S. P. Long-Range Electron Transfer Triggers Mechanistic Differences between Iron(IV)-Oxo and Iron(IV)-Imido Oxidants. J. Am. Chem. Soc. 2014, 136, 17102-17115. (35) Chowdhury, G.; Calcutt, M. W.; Nagy, L. D.; Guengerich, F. P. Oxidation of Methyl and Ethyl Nitrosamines by Cytochrome P450 2E1 and 2B1. Biochemistry 2012, 51, 9995-10007. (36) Kim, K. H.; Isin, E. M.; Yun, C. H.; Kim, D. H.; Guengerich, F. P. Kinetic Deuterium Isotope Effects for 7-Alkoxycoumarin O-Dealkylation Reactions Catalyzed by Human Cytochromes P450 and in Liver Microsomes. Rate-Limiting C-H Bond Breaking in Cytochrome P450 1A2 Substrate Oxidation. FEBS J 2006, 273, 2223-2231.

26


Page 26 of 27

Page 27 of 27

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


Table of Contents Image

27


Computational Study of Engineered Cytochrome P450-Catalyzed CâH

Recommend Documents

Computational Study of Engineered Cytochrome P450-Catalyzed CâH