steroidal anti-inflammatory drugs

Subscriber access provided by McMaster University Library

Article

Molecular mechanisms in the selectivity of non-steroidal anti-inflammatory drugs Yasmin Shamsudin Khan, Hugo Gutiérrez de Terán, and Johan Åqvist Biochemistry, Just Accepted Manuscript • Publication Date (Web): 18 Jan 2018 Downloaded from http://pubs.acs.org on January 18, 2018

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.

Biochemistry 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 35 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

Biochemistry

Molecular mechanisms in the selectivity of nonsteroidal anti-inflammatory drugs Yasmin Shamsudin Khan, Hugo Gutiérrez-de-Terán, Johan Åqvist* Department of Cell and Molecular Biology, Box 596, Uppsala University, BMC, SE-751 24 Uppsala, Sweden

*

Corresponding author: Phone: +46 18 471 4109, Fax: +46 18 53 69 71, E-mail:

[email protected]

Keywords: Non-steroidal anti-inflammatory drugs, cyclooxygenase, molecular dynamics simulations, binding free energy, molecular docking, coxibs, slow tight-binding

Abbreviations: NSAID, non-steroidal anti-inflammatory drug; MD, molecular dynamics; FEP, free energy perturbation; LIE, linear interaction energy; COX, cyclooxygenase; PMF, potential of mean force

ACS Paragon Plus Environment

Biochemistry 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 Nonsteroidal anti-inflammatory drugs (NSAIDs) inhibit cyclooxygenase (COX) 1 and 2 with varying degrees of selectivity. A group of COX-2 selective inhibitors – coxibs – bind in a time-dependent manner through a three-step mechanism, utilizing a side-pocket in the binding site. Coxibs have been extensively probed to identify the structural features regulating the slow tight-binding mechanism responsible for COX-2 selectivity. In this study, we further probe a structurally and kinetically diverse data set of COX inhibitors in COX-2 by molecular dynamics and free energy simulations. We find that the features regulating the high affinities associated with time-dependency in COX depend on the inhibitor kinetics. In particular, most time-dependent inhibitors share a common structural binding mechanism, involving an induced-fit rotation of the side-chain of Leu531 in the main binding pocket. The high affinities of two-step slow tight-binding inhibitors and some slow reversible inhibitors can thus be explained by the increased space in the main binding pocket after this rotation. Coxibs that belong to a separate class of slow tight-binding inhibitors benefit more from the displacement of the neighboring side-chain of Arg513, exclusive to the COX-2 side-pocket. This displacement further stabilizes the aforementioned rotation of Leu531, and can explain the selectivity of coxibs for COX-2.

2 Environment ACS Paragon Plus

Page 2 of 35

Page 3 of 35 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

Biochemistry

INTRODUCTION Pain-relieving and fever-reducing non-steroidal anti-inflammatory drugs (NSAIDs) inhibit both isoforms of cyclooxygenase (COX). COX-1 is constitutively expressed1 while COX-2 is induced in response to inflammation.2 Most traditional NSAIDs inhibit both isoforms in a non-selective way, but it is believed that it is the inhibition of COX-2 that results in pain relief.3 Moreover, equally or more potent inhibition of COX-1 is responsible for the common side-effects of NSAIDs, such as peptic ulcers, gastrointestinal bleeding,4 and kidney problems.5,6 Consequently, there is a strong focus on the development of pain-killer drugs that selectively target COX-2, which resulted in the release of a new class of NSAIDs, often referred to as “coxibs”. However, some of these were withdrawn from the market due to their increased risk of myocardiac infarctions.7 Still, coxibs are commonly used to treat arthritis, chronic pain8 and colon cancers.9 In the cancer area, the selective inhibition of COX-2 is a promising drug target, since it was found that this enzyme isoform is overexpressed in many different human tumors cells,10,11 but not the surrounding healthy cells.12

However, the underlying structural reasons for COX selectivity have remained elusive. The two isoforms share the same tertiary structure with 58% sequence identity. To facilitate comparison between the two isoforms the labelling of residue numbers in hCOX-2 in this work will follow that of the crystal structures of COX-1. Several points of variability are located within the binding pocket and the adjacent side-pocket (Figure 1), with some having been explicitly singled out as selectivity hotspots. One of these is Ile523 in the main binding pocket of COX-1, which is Val523 in COX-2. Since this residue is positioned between the main and side pockets (Figure 1), it was originally believed that the bulkier Ile523 in COX-1 would block access to the side-pocket. The selectivity of coxibs towards COX-2 would thus be caused by the increased facility to access the side-pocket of COX-2.13 However, it has since been demonstrated that coxibs can bind in the corresponding pocket in COX-1.14 An alternative explanation for selectivity is based on the kinetic profile of COX inhibitors.15 NSAIDs can be divided into five kinetic classes, which differ in the reversibility and number of steps involved in the binding mechanism.16


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

Page 4 of 35

Figure 1. Detailed view of the binding sites and side-pockets of the generated homology models of human COX-1 (gray) and COX-2 (magenta). Identical residues in both isoforms are colored green. Leu531 is shown in both the open (blue) and closed (yellow) rotational states.

Rapidly reversible inhibitors, such as ibuprofen (Figure 2), bind according to the simple one-step mechanism

E + I ⇄ EI

(1)

The only known truly irreversible NSAID is acetylsalicylic acid (aspirin) (Figure 2). It binds irreversibly in a time-dependent manner E + I ⇄ EI → EI*

(2)

In the first step the ligand binds to the reactive site and this binding event is reversible. In the second step Ser530 in the binding pocket is acetylated, which results in a non-covalently bound side-product of salicylic acid.17 In between these two extreme cases lie the three classes of time-dependent NSAIDs. These include the slowly reversible, which bind according to E + I ⇄ EI ⇄ EI*


(3)

Page 5 of 35 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

Biochemistry

where both steps are reversible. The next class is the slow tight-binding inhibitors where, despite the fact that they are technically reversible, the final step is only very slowly reversible, making them behave like irreversible inhibitors under physiological conditions. Thus, in practice slow tight-binding inhibitors bind in a two-step mechanism according to eq. 2, with the only difference that the second step is time-dependent and slowly reversible. Finally, the fifth class are also slow tight-binding inhibitors, but bind through a three-step mechanism E + I ⇄ EI ⇄ EI* → EX

(4)

where the third step is time-dependent and only slowly reversible. These inhibitors, which include the coxib drug celecoxib (Figure 2), are time-dependent in COX-2 but not in COX-1.15 Because of this different kinetic profile for each COX isoform, it has been suggested that the time-dependence causes selectivity. However, the structural features of this three-step mechanism have yet to be identified, which is the focus of the present study.

Figure 2. Chemical structures of the inhibitor subset reported in Table 1 and relevant ones from Table 2. The names of the structures are color-coded according to kinetic class: rapidly reversible (orange),


Biochemistry 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

irreversible (green), slowly reversible (blue), two-step slow tight-binding (purple), three-step slow tight binding (red), and unknown (gray).

Crystal structures of ovine COX-1 and murine COX-2 show that COX has at least two stable conformations, mainly identified by the rotational conformation of the Leu531 side-chain.18,19 The most common conformation observed in crystal structures, referred to as the “closed” conformation, presents the Leu531 side-chain pointing into the binding pocket (Figure 1). Through conformational analyses of COX-1 we have previously demonstrated that this is the preferred conformation in apo human COX-1 (hCOX-1) and also in complex with the rapidly reversible inhibitor ibuprofen (Figure 2).

20,21

However, two-step slow tight-binding inhibitors (eq. 2) stabilize a secondary “open”

conformation, which in turn induces higher binding affinities of the inhibitors. Herein, we extend our analysis to the computational characterization of ligand binding in COX-2. Our approach includes homology modelling, molecular dynamics (MD), binding free energy calculations and potential of mean force (PMF) simulations of the COX enzymes.20,22 The molecular mechanism behind different kinetic profiles of COX inhibitors is thus elucidated, including the three-step mechanism of slow tightbinding inhibition of coxibs, which is exclusive to COX-2. Furthermore, to provide a deeper exploration of the selectivity hotspots, we calculated the binding affinity change upon the V523I mutation for different classes of inhibitors, through free energy perturbation (FEP) calculations. The structural rationalization of COX selectivity provided here should thus be useful in the design of new COX-2 inhibitors.

METHODS

Homology Modeling and Molecular Docking

Human cyclooxygenase structures were created using default homology modeling routines in Modeller23, using an ovine COX-124 and a murine COX-2 structure25 as the templates for hCOX-1 and hCOX-2 (PDB codes 1Q4G and 3NT1 respectively). The two structures were superimposed onto the


Page 6 of 35

Page 7 of 35 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

Biochemistry

ovine COX-1 structure and the heme group was imported from the template crystal structures by structural superposition. Hydrogen atoms were added and the structure was energy minimized using Macromodel with the OPLS_2005 force field.26 The conformation of these original structures, similar to the template crystal structures, are hereafter referred to as the closed conformation. The open conformations, which are the induced secondary conformations, were created by rotating the Leu531 χ1 dihedral angles.18,20 This was done in the apo structure using the methodology described in the PMF section below. The average structure of the final window was subjected to a short energy minimization run and cleared of water molecules. The resulting structure is thus very similar to the open conformation structure form the PMF calculations, and also to the open conformation crystal structure of murine COX-2.18

All of the 37 COX inhibitors probed were built using the software Maestro included in the Schrödinger Suite 2011 package.27 Carboxylated ligands were modeled as negatively charged, and for chiral compounds only the S-enantiomers were modeled. To obtain initial poses for the molecular dynamics (MD) simulations all ligands were docked to both the open and closed conformations using 28

GLIDE XP.

Ligands were docked inside a box with the dimensions 20 × 10 × 20 Å centered on the

Val523 residue. One to three poses were generated per inhibitor and all were retained for MD simulations. An additional pose was retrieved by superimposing the energetically optimal structures of identical ligands previously docked (automated or manually) in hCOX-1. Average structures of these ligands after docking and MD simulations have been shown to be in good agreement with available crystal structures,22 which are very similar for COX-1 and COX-2 complexes. The force field parameters needed for the molecular dynamics (MD) simulations of the ligands were retrieved from automatic parameterization performed with Macromodel.26

MD Simulations Docking was followed by solvation using TIP3P29 water and MD simulations in both open and closed conformations. All MD simulations were performed using spherical boundary conditions and the


Biochemistry 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

OPLS-AA force field30 with the MD program Q.31 A sphere of 30 Å radius centered on Cβ atom of Ser530 was solvated with TIP3P water molecules and subject to polarization and radial constraints according to the surface constrained all-atom solvent (SCAAS) model31,32 at the sphere surface to mimic the properties of bulk water. Protein atoms outside the simulation sphere were restrained to their initial positions and only interacted with the system through bonds, angles and torsions. All titratable residues within 20 Å of the sphere center were treated as ionized while the remaining residues inside the simulation sphere were manually assessed with the following residues treated as ionized: Arg433, Glu465, Lys468, Lys473, Glu480, Glu502, Lys511. Titratable residues close to the sphere boundary were modeled in their neutral form to account for dielectric screening. With this setup the simulation sphere was overall neutral, except for the charge of ligands bearing a carboxylate group. Thus, the protein and water (reference) simulation systems have the same net charge and the consideration of additional Born terms31 in the calculation of free energies is avoided.

All MD simulations started with a heating and equilibration phase followed by subsequent data collection. The system was thus gradually heated to a target temperature of 298 K and positional restraints on all solute heavy atoms were gradually released. An MD time step of 2 fs was used for the data collection phase, and water bonds and angles, as well as solute bonds, were constrained using the SHAKE33 algorithm. In the water (reference state) simulations a weak harmonic restraint was applied to the center of mass of the ligands to keep them centered in the water droplet. Non-bonded interactions were calculated explicitly up to a 10 Å cutoff, except for the ligand atoms, for which no cutoff was used. Beyond the cutoff, long-range electrostatics were treated with the local reaction field 34

multipole expansion method.

Non-bonded pair lists were updated every 25 steps and the same

interval was used for the sampling of the ligand–surrounding interaction energies.

Each ligand was subjected to three MD simulations in the reference state where each heating and equilibration scheme lasted 2 ns, followed by 5 ns of data collection. In an initial screening every ligand-enzyme complex pose was subjected to one MD simulation using the same heating and equilibration scheme as in the reference state simulations, followed by 10 ns of data collection. Each


Page 8 of 35

Page 9 of 35 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

Biochemistry

pose was then , scored using the linear interaction energy (LIE) method. The initial pose of each ligand resulting in the lowest relative LIE binding free energy from this screening was selected as the final starting pose and was subsequently subjected to five independent MD simulations using a scheme consisting of 3 ns of heating, 2 ns unrestrained equilibration, and 10 ns of data collection. These simulation replicas were initiated with different random velocities, but otherwise set up with identical conditions. The convergence of the calculated binding free energies was judged both on the basis of the convergence of the energy components from each individual simulation replica and in terms of the final standard errors or the mean (s.e.m.) for all replicas. The individual simulations were thus considered converged if the average polar and non-polar interaction energies had an error bar < ± 1 kcal/mol based on splitting the trajectory in two halves. For the final calculated binding free energies the average s.e.m. over all 37 inhibitors is 0.51 kcal/mol and for the subset of 13 inhibitors studied in more detail the corresponding value is 0.42 kcal/mol. Although there are a few low affinity ligands with an s.e.m. < 1.5 kcal/mol, the general convergence of the calculated binding free energies can thus be considered quite satisfactory.

Potential of mean force calculations

Potential of mean force (PMF) calculations were performed using the Cγ−Cβ−Cα−N dihedral angle (χ1) of Leu531 as variable. Following 10 ns of MD simulations in the closed conformation according to the procedure described above and an additional 10 ps equilibration phase where the MD step size was reset to 1 fs, the dihedral angle was pushed from 90 to 190 degrees in 100 discrete windows, each of 30 ps duration.20,35 Here, the transformation from the closed to the open state utilizes the biased force field potentials = − ( ) + ( )

( ) = − + ( )

(5)

for the open and closed states, respectively. ( ) is the torsion term for χ1 in the regular force field

potential ( ) and and are biasing potentials for the open and closed states, respectively.


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

Page 10 of 35

These are quadratic, = ( − ), with a 75 kcal mol-1 rad-2 force constant (k) and have their respective minima ( ) coinciding with the open and closed conformations. The transformation is then performed via a free energy perturbation mapping potential = (1 − ) + , where the mapping parameter is varied between 0 and 1 in 100 discrete simulations windows. The free energy change on the true force field potential surface is then obtained by removing the bias according to the umbrella sampling formula ∆( ) = ∑⊂'( ∆( ) − 〈" #

%$&& (' )#% (' )*/,( ) (

〉 / / ∑⊂'(

(6)

where Xn is our discretized reaction coordinate, defined as = − , and the relevant energy values ( and ) are binned with respect to Xn. The first term on the right hand side of eq. 5 is the free energy associated with moving on the mapping potential and is computed as the average of forward and reverse application of the free energy perturbation equation #(%(12 #%( )/,- 〉 ∆( ) = − ∑#3 4 〈"

(7)

Since a given value of reaction coordinate Xn is sampled by multiple windows, their contribution is weighted in eq. 6 with the number of data points from each by the factor / ∑ . The PMF simulations were repeated five times for each system and the reported free energy profile is the average over these five replicas. To judge convergence, standard errors of the mean (s.e.m.) were calculated over all pooled data points in each bin and the maximum s.e.m. was 0.75 kcal mol-1. The average hysteresis error obtained from forward and reverse application of eq. 7 for the entire transformation is < 0.03 kcal mol-1.

Binding Affinity Calculations Binding free energies were calculated using the linear interaction energy (LIE) method. 36,37 7897 ∆〈9#; +@


(8)

Page 11 of 35 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

Biochemistry

7897 Binding affinities (∆ 56 ) are thus estimated on the basis of the difference (∆) in average ligand-

surrounding interaction energies 〈9#; 〉 extracted from the MD simulations of the ligand in the two states referred to above, i.e. in water and embedded in the active site of the solvated protein. The non −4.5 −9.7 ± 0.2 Meclofenamate −7.1 ± 0.1 −9.5 ± 0.0

−6.7 ± 0.4 −10.9 ± 0.3 −7.8 ± 0.5 −8.8 ± 0.5 −5.3 ± 0.4 −9.1 ± 0.5 −8.6 ± 0.4 −9.8 ± 0.5

−9.0 ± 0.2

−9.4 ± 1.0 −7.2 ± 0.2 − −8.9 ± 0.6

Slow tight-binding inhibitors: E + I ⇄ EI ⇄ EI* → EX Celecoxib Table 3 −9.2 ± 0.1 −9.2 ± n.d. -1 All energies are in kcal mol , with error bars ± 1 s.e.m. for calculated values. γ = −4.9. aBinding free energies from experimental instant inhibition assays ± 1 s.d.52 bAverage binding free energies from pre-incubated assays ± 1 s.e.m.47-51,62 n.d., not determined or reported. Ligands for which the open conformation is unstable are denoted by ‘−‘.


Page 33 of 35 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

Biochemistry

Table 2. Calculated and experimental binding free energies for inhibitors in complex with hCOX-2. Ligand

`Ra S[ `Ra S[ 〈_[#V 〉bScSbSQYS 〈_[#V 〉bScSbSQYS 〈_[#V 〉Y\dU[ST 〈_[#V 〉Y\dU[ST

YX[Y c ∆NOPQR

Lowest energy conformation Closed Open Closed Closed Open Closed Open Open Closed Open Open Closed Open Closed Closed Open Open Open Closed Closed Closed Open Open Open

STU (PQY.)b

∆NOPQR

−19.3 ± 0.0 −21.6 ± 0.0 −33.2 ± 0.7 −18.4 ± 1.0 Ampyrone −6.0 ± 0.5 −5.8 ± n.d. −27.0 ± 0.0 −36.9 ± 0.0 −46.3 ± 0.7 −33.9 ± 0.3 DFP −7.1 ± 0.2 −9.6 ± n.d. −12.2 ± 0.0 −160.8 ± 0.1 −22.6 ± 0.6 −159.6 ± 1.2 Diflunisal −6.8 ± 0.5 −5.5 ± n.d. −17.0 ± 0.0 −191.7 ± 0.3 −29.7 ± 2.3 −193.2 ± 0.6 Etodolac −8.5 ± 0.5 −8.3 ± 0.2 −14.7 ± 0.0 −173.7 ± 0.1 −28.2 ± 2.3 −174.4 ± 1.7 Fenoprofen −8.3 ± 1.1 −7.2 ± 0.3 −15.8 ± 0.0 −175.7 ± 0.2 −29.9 ± 0.1 −177.1 ± 2.8 Ketoprofen −8.7 ± 1.2 −9.1 ± 0.3 −15.7 ± 0.0 −177.0 ± 0.1 −31.3 ± 0.5 −179.8 ± 0.0 Ketorolac −9.8 ± 0.1 −9.8 ± 0.2 −29.0 ± 0.0 −31.6 ± 0.0 −45.4 ± 0.2 −28.4 ± 0.9 L745 −6.5 ± 0.4 −8.2 ± 0.1 −23.8 ± 0.0 −17.7 ± 0.0 −38.2 ± 1.3 −12.0 ± 0.5 Nabumetone −5.0 ± 0.3 −5.8 ± 0.8 −14.2 ± 0.0 −164.8 ± 0.1 −26.7 ± 0.4 −163.0 ± 4.8 Niflumic acid −6.9 ± 2.1 −7.0 ± n.d. −28.4 ± 0.0 −26.9 ± 0.2 −43.7 ± 0.6 −25.3 ± 1.2 NS398 −7.0 ± 0.5 −9.7 ± 0.6 −12.8 ± 0.0 −34.1 ± 0.0 −23.4 ± 0.5 −29.2 ± 0.5 Paracetamol −5.0 ± 0.2 −6.5 ± 0.6 −27.0 ± 0.0 −37.8 ± 0.1 −45.2 ± 0.7 −37.1 ± 0.1 Rofecoxib −7.9 ± 0.1 −9.2 ± n.d. −11.1 ± 0.0 −18.0 ± 0.0 −19.6 ± 0.1 −16.7 ± 0.7 Salicylaldehyde −6.0 ± 0.3 −4.7 ± n.d. −4.2 ± 0.0 −159.2 ± 0.1 −10.3 ± 0.3 −162.5 ± 1.0 Salicylic acid −8.3 ± 0.4 −6.9 ± n.d. −28.2 ± 0.0 −35.4 ± 0.2 −47.1 ± 1.2 −35.4 ± 0.8 SC58125 −8.3 ± 0.4 −9.0 ± n.d. −21.9 ± 0.1 −187.3 ± 0.2 −36.4 ± 1.0 −185.7 ± 1.9 Sulindac −7.4 ± 0.9 −6.3 ± 0.3 −22.2 ± 0.0 −173.0 ± 0.1 −43.0 ± 1.2 −172.5 ± 1.2 Sulindac sulfide −9.0 ± 0.6 −8.2 ± 0.2 −16.0 ± 0.0 −176.6 ± 0.2 −30.5 ± 0.3 −176.2 ± 1.2 Suprofen −8.0 ± 0.5 −7.2 ± n.d. −26.8 ± 0.0 −35.4 ± 0.1 −41.8 ± 1.5 −35.0 ± 4.1 Tenidap −7.4 ± 1.8 −7.8 ± n.d. −16.1 ± 0.0 −178.8 ± 0.1 −30.4 ± 0.2 −180.3 ± 0.5 Tolmetin −8.9 ± 0.2 −8.2 ± 0.1 −31.6 ± 0.0 −26.2 ± 0.0 −49.7 ± 0.2 −24.3 ± 0.2 Tomoxiprol −7.4 ± 0.1 −9.2 ± n.d. −13.0 ± 0.0 −172.9 ± 0.0 −23.9 ± 0.6 −177.8 ± 0.2 Valerylsalicylate −9.9 ± 0.0 −10.5 ± n.d. −17.3 ± 0.0 −175.9 ± 0.2 −32.0 ± 0.8 −175.3 ± 0.7 Zomepirac −7.9 ± 0.4 −10.0 ± n.d. 52 All energies are in kcal mol-1, with error bars ± 1 s.e.m. for calculated values. aBinding free energies from experimental instant inhibition assays ± 1 s.d. 47-51,62 b Average binding free energies from pre-incubated assays ± 1 s.e.m. n.d., not experimentally determined or reported.


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

Page 34 of 35

Table 3. Average energies from MD simulations in different binding modes Ligand

Celecoxib

Conformation Water Main pocket (closed) Side pocket (closed) Side pocket (open)

`Ra 〈_[#V 〉 −27.7 ± 0.0 −44.8 ± 0.9 −48.6 ± 1.2 −46.6 ± 0.2

S[ 〈_[#V 〉 −43.6 ± 0.1 −33.4 ± 1.1 −33.6 ± 1.0 −44.1 ± 0.3

YX[Y ∆NOPQR

−3.5 ± 0.5 −4.3 ± 0.5 −9.2 ± 0.1

All energies are in kcal mol-1, with error bars ± 1 s.e.m. for calculated values.


Page 35 of 35 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

Biochemistry

Table of Contents image


steroidal anti-inflammatory drugs

Recommend Documents