Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Molecular BioSystems View Article Online
PAPER
Cite this: DOI: 10.1039/c3mb70418a
View Journal
Blocking the interaction between HIV-1 integrase and human LEDGF/p75: mutational studies, virtual screening and molecular dynamics simulations† Karnati Konda Reddy,a Poonam Singhb and Sanjeev Kumar Singh*a HIV-1 integrase (IN) mediates integration of viral cDNA into the host cell genome, an essential step in the retroviral life cycle. The human lens epithelium-derived growth factor (LEDGF/p75) is a co-factor of HIV-1 IN that plays a crucial role in viral integration. Because of its crucial role in early steps of HIV replication, the IN–LEDGF/p75 interaction represents an attractive target for anti-HIV drug discovery. In this study, the IN–LEDGF/p75 interaction was studied by in silico mutational studies and molecular dynamics simulations. The results showed that all of the key residues in the LEDGF/p75 binding pocket of IN protein are important for stabilization of the complex. Structure-based virtual screening against HIV-1 IN using the ChemBridge database was performed through three different protocols of docking simulations with varying precisions and computational intensities. Six compounds based on the docking score, binding affinity and pharmacokinetic parameters were selected and an analysis of the interactions with key amino acid residues of IN was carried out. Subsequently, molecular dynamics simulations of
Received 20th September 2013, Accepted 3rd December 2013
these compounds in the LEDGF/p75 binding site of IN were carried out in order to study the stability of complexes and their hydrogen bonding interactions. IN residues Glu170, His171, and Thr174 in chain A
DOI: 10.1039/c3mb70418a
as well as Gln95 and Thr125 in chain B were discovered to play important roles in the binding of compounds. These findings could be helpful for blocking IN–LEDGF/p75 interaction, and provide a
www.rsc.org/molecularbiosystems
method for avoiding viral resistance and cross-resistance.
Introduction The acquired immunodeficiency syndrome (AIDS) is a major epidemic with more than 33 million people infected worldwide.1 Human immunodeficiency virus type 1 (HIV-1) is the primary causative agent of AIDS.2 The pol gene of HIV-1 encodes three enzymes for replication of HIV which are reverse transcriptase (RT), protease (PR) and integrase (IN).3 HIV-1 IN is a vital enzyme that catalyzes integration of the reverse transcribed viral DNA into the host genome. Integration of HIV-1 DNA into the host genome is promoted by IN and involves two steps: 3 0 -processing and DNA strand transfer. In the 3 0 processing reaction, IN removes two nucleotides from each viral cDNA end adjacent to a conserved 3 0 -CA sequence, generating CA-3 0 hydroxyl recessed ends. In the strand transfer reaction, the newly processed 3 0 -viral DNA ends are inserted into the host
a
Computer-Aided Drug Design and Molecular Modeling Lab, Department of Bioinformatics, Alagappa University, Karaikudi-630 004, Tamil Nadu, India. E-mail: [email protected]; Fax: +91 4565 225202; Tel: +91 4565 223342 b Division of Toxicology, CSIR-Central Drug Research Institute, Lucknow-226 001, India † Electronic supplementary information (ESI) available. See DOI: 10.1039/c3mb70418a
This journal is © The Royal Society of Chemistry 2014
DNA.4,5 HIV-1 IN is composed of three structurally and functionally distinct domains: the amino-terminal domain (NTD, amino acids 1–49) contains a highly conserved zinc binding motif (HHCC) important for oligomerization; the catalytic core domain (CCD, amino acids 50–212) contains highly conserved Asp64, Asp116, and Glu152 (DDE) residues which are directly involved in the catalytic activities of IN; and the carboxy-terminal domain (CTD, amino acids 213–288) contains an SH3-like domain that binds non-specifically to DNA.6–9 HIV-1 IN has emerged as a promising target as this is responsible for integration of the newly synthesized doublestranded viral DNA into host genomic DNA.10,11 HIV-1 IN strand transfer inhibitors, which target the enzyme active site, have witnessed clinical success over the past few years, but the emergence of drug resistance poses challenges.12 HIV relies on host cellular machinery to complete its replication cycle. Cellular cofactors are important for integration function of IN.13 The lens epithelium-derived growth factor (LEDGF/p75), a transcriptional co-activator, was initially identified as an IN co-factor by co-immunoprecipitation from cells overexpressing HIV IN.14 The crucial role of LEDGF/p75 in HIV replication was evidenced via mutagenesis, RNAi-mediated depletion, transdominant overexpression of the integrase binding domain (IBD) of LEDGF/p75
Mol. BioSyst.
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Paper
Molecular BioSystems
Fig. 1 Domain organization of HIV-1 IN and LEDGF/p75 proteins. The key residues between the IN catalytic core domain (CCD) and the LEDGF/p75 integrase binding domain (IBD) are highlighted.
and knockout studies.15 The cellular cofactor LEDGF/p75 has been identified as the dominant binding partner of HIV-1 IN in human cells.14,16 LEDGF/p75 depletion impairs binding of IN to chromatin.17 A chromatin-binding domain (PWWP) is located at the N-terminus of LEDGF/p75, and the C-terminus contains the IBD. The domain organization of HIV-1 IN and LEDGF/p75 is shown in Fig. 1. The interaction between IN and LEDGF/p75 is an optimal target for HIV drug design. The host cell cofactor LEDGF/p75 plays an important role in the integration process by tethering IN to chromatin.18–20 Crystal structure of the dimeric CCD of HIV-1 IN complexed to the IBD of LEDGF/p75 gave a major advance for structure-based drug design targeting the IN–LEDGF/p75 protein–protein interaction (PPI). The structure has pseudo-two fold symmetry; two LEDGF/p75 IBD domains (residues 347–429) bind at nearly identical positions at either end of the IN dimer interface.20,21 The LEDGF/p75 IBD contacts residues from both chains of the IN dimer as shown in Fig. 2. The key interactions between HIV-1 IN and LEDGF/p75 have also been reported earlier.22,23 Specific cloning and quantification of the circular full site integration products attested that the IN–LEDGF/p75 complex catalyses 2 to 10 times more concerted integration events than isolated IN molecules.24 Blocking IN–LEDGF/p75 interaction can seriously inhibit viral replication.21 We used fast and inexpensive docking protocols combined with accurate but more expensive molecular dynamics (MD) simulation techniques to understand the stability of protein– protein and protein–ligand complex structures.25 PPIs are critical components of the machinery within living organisms and represent a large and important class of therapeutic targets. Use of PPI inhibitors will amplify the scope for discovery of drugs that can conquer HIV-resistance.14 The main
Mol. BioSyst.
aim of our study is to find small molecules that target the IN–LEDGF/p75 interaction, and thereby inhibit HIV replication. This is of particular interest in virology that demands new antivirals preferably with a completely different mechanism of action to avoid resistance and cross-resistance. The interaction between viral HIV-1 IN and its cellular cofactor LEDGF/p75 is one of the most promising targets at the moment. Identification of effective small molecule PPI inhibitors as new antiretroviral drugs would considerably boost generic interest in PPI inhibitors.
Materials and methods Protein model preparation The X-ray crystal structure of the dimeric CCD of HIV-1 IN complexed with IBD of LEDGF/p75 (PDB ID: 2B4J) was downloaded from Protein Data Bank.20 Both wild type and mutant forms of IN–LEDGF/p75 protein complexes were prepared by a multi-step process through Protein Preparation Wizard.26 All crystallographic water molecules and other chemical components were omitted, right bond orders as well as charges and atom types were assigned, and hydrogen atoms were added to the crystal structure. Each model was subjected to energy minimization using the Optimized Potentials for Liquid Simulations (OPLS)-2005 force field with implicit solvation. Each model was minimized and the minimization was terminated when root mean square deviation (RMSD) of heavy atoms in the minimized structure relative to the X-ray structure exceeded 0.3 Å. We prepared one wild type, four single mutant forms Q168A, E170A, H171A, and T174A, and three double mutant forms (Q168A and E170A) and (E170A
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Molecular BioSystems
Paper
Fig. 2 Co-crystal structure of HIV-1 IN CCD (green) complexed with LEDGF/p75 IBD (red). LEDGF/p75 IBD domains (residues 347–429) bind at nearly identical positions at either end of the IN dimer interface. Key contacts at the interface are highlighted.
and H171A) in chain A, and (Q95A and T125A) in chain B of IN. In order to perform virtual screening studies of HIV-1 IN, we removed the host protein LEDGF/p75 and prepared the HIV-1 IN CCD as described in the protocol earlier. Ligand preparation A total of 50 000 molecules from ChemBridge DIVERSet database were selected for structure-based virtual screening against the LEDGF/p75 binding site of HIV-1 IN. All the hydrogen atoms were added to the ligand molecules as they only had implicit hydrogen atoms, bond order of these ligands were fixed and ionization states of ligands were generated in the pH range of 5.0–9.0 using Epik during ligand preparation by LigPrep.27 Most probable tautomers and all possible stereo isomers were generated to study the activity of individual stereotypes of each ligand. In the final stage of LigPrep, compounds were minimized using the OPLS-2005 force field.28 After running the ligand preparation a total of 91 169 ligands were generated. Structure-based virtual screening We ran high throughput virtual screening (HTVS), standard precision (SP) and extra precision (XP) protocols using the program Glide.29,30 In docking simulation, we used semiflexible docking protocols. The ligands being docked were kept flexible, in order to explore an arbitrary number of torsional degrees of freedom spanned by the translational and rotational
This journal is © The Royal Society of Chemistry 2014
parameters. The ligand poses generated were passed through a series of hierarchical filters that evaluated ligand interaction with its receptor. The glide score was used as a function of fitness and best-fit pose for a given ligand was determined by the Emodel score. Prime MM-GBSA free energy The top docked poses for each ligand were rescored for calculation of binding free energy by the Prime/MMGBSA method.31 The binding free energy (DGbind) was calculated using the following equation:32 DGbind = DEMM + DGsolv + DGSA where DEMM is the difference in energy between the IN–inhibitor complex and the sum of the energies of the unliganded IN protein and the ligand. DGsolv is the difference in the GBSA solvation energy of the complex and the sum of solvation energies of the unliganded uncomplexed IN protein and the ligand. DGSA is the difference in surface area energy of the complex and the sum of surface area energies of the IN protein and the ligand. ADME properties prediction The identified best six hits were studied for their absorption, distribution, metabolism and excretion (ADME) properties by using QikProp.33 Accurate prediction of ADME properties prior to experimental procedures can eliminate unnecessary testing
Mol. BioSyst.
View Article Online
Paper
of compounds. The percentage of their human oral absorption was also predicted to determine their toxicity levels.
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Density functional theory calculations The molecules selected from the docking study were used as inputs for density functional theory (DFT) calculations. All DFT computations were carried out using Jaguar.34 Complete geometry optimization was carried out using hybrid DFT with Becke’s threeparameter exchange potential and Lee–Yang–Parr correlation functional (B3LYP), using the basis set 3-21G* level.35–38 Highest Occupied Molecular Orbitals (HOMOs) and Lowest Unoccupied Molecular Orbitals (LUMOs) were computed. Molecular dynamics simulations Protein and ligand structures used in Desmond simulation must be complete for all-atom 3D structures with a reasonable geometry. The protein and ligand structures used in simulation
Molecular BioSystems
were prepared using Protein Preparation Wizard and LigPrep. MD simulations of all wild type as well as mutant forms of IN–LEDGF/p75 and IN–ligand complexes were performed using Desmond Molecular Dynamics System39–41 with OPLS all-atom force field 2005.28,42 The complexes were immersed in an orthorhombic box containing TIP3P water molecules and neutralized by adding salt counter-ions. The distance between the box wall and complexes was set to greater than 10 Å to avoid direct interactions with its own periodic image. The energy of prepared systems for MD simulations was minimized up to a maximum of 5000 steps using a steepest descent method until a gradient threshold (25 kcal mol1 Å1) was reached, followed by LBFGS (Low-memory Broyden–Fletcher–Goldfarb–Shano quasi-Newtonian minimizer) until a convergence threshold of 1 kcal mol1 Å1 was met. The NPT ensemble, the Nose–Hoover chain thermostat, and the Martyna–Tobias–Klein barostat were employed for the actual simulation, and the RESPA integrator was used
Fig. 3 RMSD of the backbone of wild type and mutant forms during the simulation. (A) RMSD of the wild type and mutant forms of IN–LEDGF/p75 complexes. (B) RMSD of only LEDGF/p75 in the wild type and mutant forms of IN–LEDGF/p75 complexes through 10 ns MD simulations.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Molecular BioSystems
complexes were simulated for a 10 ns time period using the parameters described above. The atom numbers of wild type and mutant IN–LEDGF/p75 complexes are listed in ESI,† Table S1.
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
with a time step of 0.002 ps. A 9 Å cutoff was used for the Coulombic range interactions, and long-range electrostatics were treated using the particle mesh Ewald method, with a tolerance of 1 e9 Å. All IN–LEDGF/p75 and IN–ligand
Paper
Fig. 4 Binding modes and molecular interactions of three compounds 1–3 with the LEDGF/p75 binding site at the IN dimer interface (chain A – green and chain B – yellow color). The hydrogen bonds are displayed in black dashed lines with distances.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst.
View Article Online
Paper
atoms of compound 2 forming hydrogen bonding interactions with IN residues was calculated and plotted through 10 ns simulations.
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
The RMSDs for IN–LEDGF/p75 and IN–ligand complexes were calculated for the entire simulation trajectory with reference to their respective first frames. The distance between
Molecular BioSystems
Fig. 5 Binding modes and molecular interactions of three compounds 4–6 with a LEDGF/p75 binding site at the IN dimer interface (chain A – green and chain B – yellow color). The hydrogen bonds are displayed in black dashed lines with distances.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Molecular BioSystems
Paper
Results and discussion
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Mutational studies and MD simulations We performed in silico mutational studies on HIV-1 IN complexed with LEDGF/p75 to understand the role of specific residues in stabilizing the complex. The amino acids of HIV-1 IN that are important for the interaction with LEDGF/p75 were mutated to alanine. Single mutant forms Q168A, E170A, H171A, and T174A, double mutant forms (Q168A and E170A) and (E170A and H171A) in chain A and (Q95A and T125A) in chain B of IN were generated. The RMSDs of IN–LEDGF/p75 complexes versus time are given in Fig. 3A. All systems were not stable throughout the simulation time with a few structural changes in the complexes. The LEDGF/ p75 shows less stability in the double mutant (Q168A and E170A) form. The RMSDs of only bound LEDGF/p75 in both wild type and mutant forms are plotted to understand its stability [Fig. 3B]. Overall, the single and double mutants seem to affect the complex stability. Further, the side chain RMSDs of IN–LEDGF/p75 complexes are also given in ESI† (Fig. S1). All the key residues analyzed through mutational studies have an important role in stabilizing LEDGF/p75. All these observations suggest that the interactions between HIV-1 IN and LEDGF/p75 are governed by all reported key amino acid residues. The LEDGF/p75 binding interface is located in a pocket formed by the two subunits of the IN–core dimer. Residues located in the a-4/5 connector (loop) and the hydrophobic pocket of IN engage tightly with LEDGF/p75. Structure-based virtual screening In order to identify compounds targeting HIV-1 IN and block LEDGF/p75 interaction, we probe small-molecule inhibitors effective at disrupting the IN–LEDGF/p75 interaction. We performed structure-based virtual screening by docking the ChemBridge database compounds targeting the LEDGF/p75 binding site of HIV-1 IN. We used 50 000 compounds of the ChemBridge database initially which further generated 91 169 ligands after ligand preparation. Initially, we performed docking using a HTVS docking protocol which is the least computationally intense process for rapid screening of the ligands. Ligands were selected for further steps on the basis of the docking score, the cutoff was set as 5.0 i.e., only 3377 ligands were selected having a docking score below 5.0 kcal mol1. These 3377 ligands were docked in the same receptor using a SP docking protocol in
Table 1
which top 10% compounds were selected based on the docking score. 337 ligands were docked using a XP docking protocol which is a more powerful and discriminative procedure and takes a longer time than SP docking. Top 25 ligands from XP docking, after removing the duplicates were considered for visual inspection. The resulting compounds were further analyzed for their interactions with key amino acid residues in the LEDGF/p75 binding site of IN. The compounds were selected based on their interaction modes at the LEDGF/p75 binding site of HIV-1 IN. The present virtual screening exercise provided six compounds that were suitable starting points for lead optimization and that are expected to block the IN and LEDGF/p75 interaction. The flatness of protein–protein interfaces often hampers the identification of small molecule PPI inhibitors. In this case, the LEDGF/p75 binds to a defined pocket in the dimer interface of HIV-1 IN CCD. Due to the tight and small interface of the LEDGF/p75 binding site on HIV-1 IN, the design and/or selection of small molecules binding to this pocket is a feasible endeavour.43 Binding mode analysis The compounds showed stable interactions with the LEDGF/ p75 binding site of IN. The interactions of top six compounds with the IN obtained in the docking study are shown in Fig. 4 and 5. From the docking results, amino acid residues Glu170, His171, and Thr174 in chain A and Gln95 and Thr125 in chain B of HIV-1 IN CCD were seen as key players in the binding of compounds. All selected six lead hits showed hydrogen bonding interactions with Glu170, His171 and Thr174 in chain A and Thr125 in chain B of IN. They also showed hydrophobic interactions with IN B-chain residues Ala98, Leu102, Ala128, Ala129 and Trp132 and the IN A-chain residue Met178, at the LEDGF/ p75 binding site of the HIV-1 IN CCD dimer interface. Carboxyl groups of compounds 1 and 2 form hydrogen bonds with main chain nitrogens of Glu170 and His171 and the hydroxyl group of Thr174 in chain A of IN. Furthermore, Gln95 and Thr125 residues formed hydrogen bonds with compounds 1 and 2. In the IN–LEDGF/p75 core complex, equivalent hydrogen bonds were formed with the side chain of Asp366 located in the LEDGF/p75 IBD loop. HIV-1 IN residues Thr125, Glu170, His171, and Thr174 formed hydrogen bonds and binding free
Glide docking and Prime-MMGBSA results for the best six compounds
Compound
ChemBridge ID
1
9123953
2
7662210
3
5791465
4
9057093
5
5314403
6
9120780
H bond interactions
Hydrophobic interactions
Glu170, Thr174, Glu170, Thr174, Glu170, Thr174, Glu170, Thr174 Glu170, Thr174, Glu170, Thr174,
Met178, Leu102, Ala128, Ala129, Trp132 Ala169, Met178, Ala98, Leu102, Ala128, Ala129, Trp132 Met178, Leu102, Ala128, Ala129, Trp132 Met178, Ala98, Leu102, Ala128, Ala129, Trp132 Met178, Ala98, Leu102, Ala128, Ala129, Trp132 Met178, Ala98, Leu102, Ala128, Ala129, Trp132
His171, Gln95, Thr125 His171, Thr125 His171, Gln95, Thr125 His171, His171, Gln95 His171, Thr125
This journal is © The Royal Society of Chemistry 2014
Glide docking score (kcal mol1)
Glide Emodel score
Prime-MMGBSA DGbind (kcal mol1)
8.37
43.80
17.04
8.27
53.34
29.07
7.83
45.02
13.08
7.47
45.99
6.70
7.24
45.27
17.48
7.029
47.447
18.969
Mol. BioSyst.
View Article Online
Paper
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Table 2
Molecular BioSystems ADME or pharmacokinetic predictions of the best six compounds
Compound
ChemBridge ID
QP log Pa
QP log Sb
QP PCaco2c
Percentage of human oral absorptiond
1 2 3 4 5 6
9123953 7662210 5791465 9057093 5314403 9120780
1.54 4.025 0.356 3.822 1.892 1.697
2.665 4.346 0.913 4.151 2.507 2.509
57.818 145.341 312.841 1824.883 92.037 85.785
67.503 89.213 73.694 100 73.175 71.488
The permissible ranges are as follows. a Log of the octonal–water partition coefficient (2.0 to 6.5). b Log of the aqueous solubility S (mol L1): (6.5 to 0.5). c Caco2 cell permeability in nm s1: (o25 is poor and >500 is great). d Percentage of human oral absorption: >80% high, o25% low.
Table 3
Frontier orbital energies of the best six compounds
Compound
HOMO (eV)
LUMO (eV)
HLG (eV)a
1 2 3 4 5 6
0.24 0.24 0.21 0.21 0.22 0.23
0.01 0.03 0.06 0.01 0.05 0.04
0.23 0.21 0.15 0.20 0.17 0.19
a
HLG is the HOMO–LUMO energy gap.
energy (29.065 kcal mol1) with compound 2 was high. Compound 3 interacts with Gln95, Thr125, Glu170, His171, and Thr174 forming a total of five hydrogen bonds. In the case of compound 4 the docking results showed that the nitrogen atom of the pyrimidine ring and the oxygen atom of the compound established hydrogen bonding with Glu170, His171 and Thr174 respectively. The carboxyl group and the nitrogen atom of compound 5 formed hydrogen bonds with Gln95, Glu170, His171 and Thr174. From the docking conformation of compound 6 it was observed that the oxygen atom of ether and the carboxyl group of the compound established hydrogen bonding with Thr125, Glu170, His171 and Thr174 respectively. Compounds 4, 5 and 6 showed a similar binding mode establishing hydrophobic interactions with Met178, Ala98, Leu102, Ala128, Ala129 and Trp132 respectively. All the six compounds were rescored using Prime/MM-GBSA. These approaches predict the binding free energy for a set of ligands to receptor. The docking results and binding free energy of the best six compounds are shown in Table 1. ADME prediction The predictions of pharmacologically important properties were computed using QikProp for the selected compounds as shown in Table 2. Different properties consisting of principal descriptors and physicochemical properties were calculated. When QikProp was run for a set of 1700 oral drugs, 95% were predicted to have molecular weights between 130 and 725, log p values between 2 and 6, log S values between 6.0 and 0.5, and Caco2 values greater than 25 nm s1.44 The predicted properties of the six identified hits compared favorably with these ranges and also found a very high human oral absorption. The newly searched compounds appear to be suitable for further development. DFT analysis
Fig. 6 Plots of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of compound 2.
Frontier orbital energies and the HOMO–LUMO gap were calculated for the best six compounds [Table 3]. Both the
Fig. 7 Ca RMSD from the initial structures of protein–ligand complexes over time during MD simulations.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Molecular BioSystems
HOMO and LUMO energies are small, ranging between 0.24 and 0.21 and 0.06 and 0.01 eV, respectively, indicating the fragile nature of bound electrons. Because of small values of HOMO and LUMO, both rapid electron transfer and exchange are equally possible, making these molecules very reactive. HOMO and LUMO sites are plotted onto the surface of compound 2 and shown in Fig. 6. Analysis of the HOMO map of compound 2 illustrates that HOMO molecular orbitals are located on the carboxyl group indicating the existence of reactive sites and electrophilic attack takes place at the residues of the active site of HIV-1 IN. The LUMO map is plotted on the sulphur group and the benzene group. The HOMO–LUMO gap is the potential energy difference between the HOMO and the LUMO. A small HOMO–LUMO gap means small excitation
Fig. 8 Intermolecular hydrogen plot between IN and compound 2 through 10 ns simulations.
Fig. 9
Paper
energies to the manifold of excited states. Therefore, molecules with a small gap, will be more polarizable than the molecules with large gap. This helps in the reaction of these compounds with the surrounding amino acids of the receptor. MD simulations of protein–ligand complexes The docking protocols can be combined with accurate but more costly MD techniques to predict more reliable protein–ligand complexes. In order to gain insight into the stability and dynamic properties of the complexes, 10 ns MD simulations were performed. We assessed the dynamics stability of all the complexes during the MD simulation and the RMSDs of backbone Ca atoms were analyzed and plotted in Fig. 7. MD simulation results revealed that the RMSD of the complexes almost reached equilibrium at B1000 ps. After the initial deviation, the complexes did not deviate further and showed consistent RMSD of around 1 Å throughout the simulation process. All the ligands formed stable hydrogen bonding interactions with Glu170, His171, and Thr174 in chain A and Thr125 in chain B of IN. Hydrophobic interactions with Ala98, Leu102, Ala128, Ala129, Trp132 and Met178 are also stable throughout MD simulations. Stable hydrogen bonds among protein and ligands strengthen the binding affinity of all the complexes. Recently, the small molecule inhibitors identified for the newly explored binding site region and site mutagenesis studies highlighted Gln168, Glu170, His171, and Thr174 in chain A and Glu95, Thr125, and Trp131 in chain B of HIV-1 IN as key amino acid residues.45–48 The first reported potent molecule targeting the LEDGF/p75 binding site of HIV-1 IN is the benzoic acid derivative, 4-[(5-bromo-4-{[2,4-dioxo-3-(-oxo-2-phenylethyl)-1,3thiazololidin-5-ylidene] methyl}-2-ethoxyphenoxy)methyl]benzoic acid (D77), which interacts with residues Gln95, Thr125, Trp131 and Thr174.49 The series of 2-(quinolin-3-yl) acetic acid derivatives
The interaction of compound 2 before and after the MD simulations.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst.
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Paper
Molecular BioSystems
Fig. 10 The atomic distances between the atoms involved in hydrogen bond formation through MD trajectory based analysis of the IN–compound 2 complex. Distance between the hydrogen bonding residues Glu170, His171, Thr174, and Thr125 and compound 2 depicted for the entire 10 ns simulation.
(LEDGINs) were the first genuine allosteric HIV-1 IN inhibitors shown to form hydrogen bonds with residues Glu170, His171, and Thr174 in chain A of HIV-1 IN.50 They have not shown any hydrogen bonding interaction with chain B residues. Recently, De Luca et al. reported the interactions between LEDGINs and the LEDGF/p75 binding site in HIV-1 IN combining docking and molecular dynamics simulations.51 Compared with LEDGINs, these newly identified compounds have significant potential for further development. The reported compounds in this study showed very good interactions with the critical residues of Glu170, His171, and Thr174 in chain A. They also showed interactions with Thr125 and Trp131 in chain B which are critical for LEDGF/p75 binding. We herein present potent small molecules of HIV-1 IN that bind to the LEDGF/p75 binding site, which have groups important for the interaction at the dimeric interface. MD analysis of HIV-1 IN complexed with compound 2 MD analysis of the IN–compound 2 complex stability during simulation was monitored by understanding the RMSD profiles, the number of hydrogen bonds and the distance between atoms during the MD trajectory. The hydrogen bond profile of the IN–compound 2 complex, as displayed in Fig. 8, reveals that the complex is stabilized by 4–5 hydrogen bonds (Glu170, His171, Thr174, Gln95 and Thr125) throughout 10 ns simulations. Fig. 9 clearly shows that stacking interaction with His171 and hydrogen bonding interactions with Glu170, His171, Thr174, and Thr125 are similar before and after the MD simulations. The compound is also stabilized by more hydrophobic interactions with Ala169, Met178, Ala98, Leu102, Ala128, Ala129 and Trp132. The nitrogen atoms of Glu170 and His171, and OG1 of Thr174 form a stable interaction with the carboxyl group of compound 2. The oxygen atom of Thr125 forms stable interaction with the sulfonyl group of compound 2. The atomic distances between atoms of IN and compound 2
Mol. BioSyst.
through 10 ns MD simulations are shown in Fig. 10. We found an additional stable hydrogen bond formed with Gln95 during simulation. The stacking interaction between the benzene ring of compound 2 and His174 is also stable throughout the MD simulation.
Conclusion The interaction between HIV-1 IN and LEDGF/p75 is an attractive novel target and has increasingly gained attention for antiHIV drug discovery. In this study, an integrated computational approach by mutational studies, MD simulations, structurebased virtual screening, MM-GBSA binding free energy and DFT analysis was used to find the inhibitors which can block the PPI between HIV-1 IN and human LEDGF/p75. We identified HIV-1 IN inhibitors that did not target the catalytic site through structure-based virtual screening with docking simulations. Analysis of the docking modes of ligands showed that the most crucial residues in the binding site are Glu170, His171, and Thr174 in the A-chain and Gln95 and Thr125 in the B-chain of IN. Owing to criticality of these inferences, these ligands can prove to be highly effective if incorporated into drug development phases and experimental studies. These inhibitors could block the PPI between a viral protein HIV-1 IN and a cellular host factor LEDGF/p75. The computational methods used and the hits identified in this study provide an excellent guide to disrupt the IN–LEDGF/p75 interaction.
Acknowledgements KKR thanks Council of Scientific and Industrial Research (CSIR), New Delhi, India, for awarding a Senior Research Fellowship. The authors thank M. Ravikumar for the detailed discussion and suggestions.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Molecular BioSystems
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
References 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
UNAIDS, Report on the Global HIV/AIDS Epidemic, 2012. R. A. Weiss, Science, 1993, 260, 1273–1279. D. J. McColl and X. Chen, Antiviral Res., 2009, 85, 101–118. S. A. Chow, K. A. Vincent, V. Ellison and P. O. Brown, Science, 1992, 255, 723–726. G. Goodarzi, G. J. Im, K. Brackmann and D. Grandgenett, J. Virol., 1995, 69, 6090–6097. M. Cai, R. Zheng, M. Caffrey, R. Craigie, G. M. Clore and A. M. Gronenborn, Nat. Struct. Biol., 1997, 4, 567–577. F. Dyda, A. B. Hickman, T. M. Jenkins, A. Engelman, R. Craigie and D. R. Davies, Science, 1994, 266, 1981–1986. A. Engelman and R. Craigie, J. Virol., 1992, 66, 6361–6369. D. Esposito and R. Craigie, Adv. Virus Res., 1999, 52, 319–333. M. H. Nielsen, F. S. Pedersen and J. Kjems, Retrovirology, 2005, 2, 10. K. K. Reddy, S. K. Singh, N. Dessalew, S. K. Tripathi and C. Selvaraj, J. Enzyme Inhib. Med. Chem., 2012, 27, 339–347. A. Engelman, J. J. Kessl and M. Kvaratskhelia, Curr. Opin. Chem. Biol., 2013, 17, 339–345. B. Van Maele, K. Busschots, L. Vandekerckhove, F. Christ and Z. Debyser, Trends Biochem. Sci., 2006, 31, 98–105. P. Cherepanov, G. Maertens, P. Proost, B. Devreese, J. Van Beeumen, Y. Engelborghs, E. De Clercq and Z. Debyser, J. Biol. Chem., 2003, 278, 372–381. F. Christ and Z. Debyser, Virology, 2013, 435, 102–109. M. Tsiang, G. S. Jones, M. Hung, S. Mukund, B. Han, X. Liu, K. Babaoglu, E. Lansdon, X. Chen, J. Todd, T. Cai, N. Pagratis, R. Sakowicz and R. Geleziunas, J. Biol. Chem., 2009, 284, 33580–33599. M. C. Shun, Y. Botbol, X. Li, F. Di Nunzio, J. E. Daigle, N. Yan, J. Lieberman, M. Lavigne and A. Engelman, J. Virol., 2008, 82, 11555–11567. O. Taltynov, B. A. Desimmie, J. Demeulemeester, F. Christ and Z. Debyser, Mol. Biol. Int., 2012, 2012, 863405. E. M. Poeschla, Cell. Mol. Life Sci., 2008, 65, 1403–1424. P. Cherepanov, A. L. Ambrosio, S. Rahman, T. Ellenberger and A. Engelman, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 17308–17313. L. Q. Al-Mawsawi and N. Neamati, Trends Pharmacol. Sci., 2007, 28, 526–535. Y. Zhao, W. Li, J. Zeng, G. Liu and Y. Tang, Proteins, 2008, 72, 635–645. C. Tintori, N. Veljkovic, V. Veljkovic and M. Botta, Proteins, 2010, 78, 3396–3408. B. Maillot, N. Levy, S. Eiler, C. Crucifix, F. Granger, L. Richert, P. Didier, J. Godet, K. Pradeau-Aubreton, S. Emiliani, A. Nazabal, P. Lesbats, V. Parissi, Y. Mely, D. Moras, P. Schultz and M. Ruff, PLoS One, 2013, 8, e60734. M. Karplus and J. A. McCammon, Nat. Struct. Biol., 2002, 9, 646–652. ¨dinger Suite 2011 Protein Preparation Wizard, Epik Schro ¨dinger, LLC, New York, NY, 2011; Impact version 2.2, Schro ¨dinger, LLC, New York, NY, 2011. version 5.7, Schro
This journal is © The Royal Society of Chemistry 2014
Paper
¨dinger, LLC, New York, NY, 2011. 27 LigPrep, version 2.5, Schro 28 W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am. Chem. Soc., 1996, 118, 11225–11236. ¨dinger, LLC, New York, NY, 2011. 29 Glide, version 5.7, Schro 30 R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren, J. J. Klicic, D. T. Mainz, M. P. Repasky, E. H. Knoll, M. Shelley, J. K. Perry, D. E. Shaw, P. Francis and P. S. Shenkin, J. Med. Chem., 2004, 47, 1739–1749. ¨dinger, LLC, New York, NY, 2011. 31 Prime, version 3.0, Schro 32 P. D. Lyne, M. L. Lamb and J. C. Saeh, J. Med. Chem., 2006, 49, 4805–4808. ¨dinger, LLC, New York, NY, 2011. 33 QikProp, version 3.4, Schro ¨dinger, LLC, New York, NY, 2011. 34 Jaguar, version 7.8, Schro 35 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter Mater. Phys., 1988, 37, 785–789. 36 A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652. 37 J. S. Binkley, J. A. Pople and W. J. Hehre, J. Am. Chem. Soc., 1980, 102, 939–947. 38 W. J. Pietro, M. M. Francl, W. J. Hehre, D. J. DeFrees, J. A. Pople and J. S. Binkley, J. Am. Chem. Soc., 1982, 104, 5039–5048. 39 Desmond Molecular Dynamics System, version 3.1, D. E. Shaw Research, New York, NY, 2012; Maestro-Desmond Interoper¨dinger, New York, NY, 2012. ability Tools, version 3.1, Schro 40 Z. Guo, U. Mohanty, J. Noehre, T. K. Sawyer, W. Sherman and G. Krilov, Chem. Biol. Drug Des., 2010, 75, 348–359. 41 A. Grover, S. P. Katiyar, J. Jeyakanthan, V. K. Dubey and D. Sundar, BMC Genomics, 2013, 13(Suppl 7), S20. 42 G. A. Kaminski, R. A. Friesner, J. Tirado-Rives and W. L. Jorgensen, J. Phys. Chem. B, 2001, 105, 6474–6487. 43 D. P. Ryan and J. M. Matthews, Curr. Opin. Struct. Biol., 2005, 15, 441–446. 44 V. V. Thakur, J. T. Kim, A. D. Hamilton, C. M. Bailey, R. A. Domaoal, L. Wang, K. S. Anderson and W. L. Jorgensen, Bioorg. Med. Chem. Lett., 2006, 16, 5664–5667. 45 L. De Luca, S. Ferro, R. Gitto, M. L. Barreca, S. Agnello, F. Christ, Z. Debyser and A. Chimirri, Bioorg. Med. Chem., 2010, 18, 7515–7521. 46 K. Busschots, A. Voet, M. De Maeyer, J. C. Rain, S. Emiliani, R. Benarous, L. Desender, Z. Debyser and F. Christ, J. Mol. Biol., 2007, 365, 1480–1492. 47 G. Hu, X. Li, X. Sun, W. Lu, G. Liu, J. Huang, X. Shen and Y. Tang, J. Mol. Model., 2012, 18, 4995–5003. 48 G. Hu, X. Li, X. Zhang, Y. Li, L. Ma, L. M. Yang, G. Liu, W. Li, J. Huang, X. Shen, L. Hu, Y. T. Zheng and Y. Tang, J. Med. Chem., 2012, 55, 10108–10117. 49 L. Du, Y. Zhao, J. Chen, L. Yang, Y. Zheng, Y. Tang, X. Shen and H. Jiang, Biochem. Biophys. Res. Commun., 2008, 375, 139–144. 50 F. Christ, A. Voet, A. Marchand, S. Nicolet, B. A. Desimmie, D. Marchand, D. Bardiot, N. J. Van der Veken, B. Van Remoortel, S. V. Strelkov, M. De Maeyer, P. Chaltin and Z. Debyser, Nat. Chem. Biol., 2010, 6, 442–448. 51 L. De Luca, F. Morreale and A. Chimirri, J. Chem. Inf. Model., 2012, 52, 3245–3254.
Mol. BioSyst.
Molecular BioSystems View Article Online
PAPER
Cite this: DOI: 10.1039/c3mb70418a
View Journal
Blocking the interaction between HIV-1 integrase and human LEDGF/p75: mutational studies, virtual screening and molecular dynamics simulations† Karnati Konda Reddy,a Poonam Singhb and Sanjeev Kumar Singh*a HIV-1 integrase (IN) mediates integration of viral cDNA into the host cell genome, an essential step in the retroviral life cycle. The human lens epithelium-derived growth factor (LEDGF/p75) is a co-factor of HIV-1 IN that plays a crucial role in viral integration. Because of its crucial role in early steps of HIV replication, the IN–LEDGF/p75 interaction represents an attractive target for anti-HIV drug discovery. In this study, the IN–LEDGF/p75 interaction was studied by in silico mutational studies and molecular dynamics simulations. The results showed that all of the key residues in the LEDGF/p75 binding pocket of IN protein are important for stabilization of the complex. Structure-based virtual screening against HIV-1 IN using the ChemBridge database was performed through three different protocols of docking simulations with varying precisions and computational intensities. Six compounds based on the docking score, binding affinity and pharmacokinetic parameters were selected and an analysis of the interactions with key amino acid residues of IN was carried out. Subsequently, molecular dynamics simulations of
Received 20th September 2013, Accepted 3rd December 2013
these compounds in the LEDGF/p75 binding site of IN were carried out in order to study the stability of complexes and their hydrogen bonding interactions. IN residues Glu170, His171, and Thr174 in chain A
DOI: 10.1039/c3mb70418a
as well as Gln95 and Thr125 in chain B were discovered to play important roles in the binding of compounds. These findings could be helpful for blocking IN–LEDGF/p75 interaction, and provide a
www.rsc.org/molecularbiosystems
method for avoiding viral resistance and cross-resistance.
Introduction The acquired immunodeficiency syndrome (AIDS) is a major epidemic with more than 33 million people infected worldwide.1 Human immunodeficiency virus type 1 (HIV-1) is the primary causative agent of AIDS.2 The pol gene of HIV-1 encodes three enzymes for replication of HIV which are reverse transcriptase (RT), protease (PR) and integrase (IN).3 HIV-1 IN is a vital enzyme that catalyzes integration of the reverse transcribed viral DNA into the host genome. Integration of HIV-1 DNA into the host genome is promoted by IN and involves two steps: 3 0 -processing and DNA strand transfer. In the 3 0 processing reaction, IN removes two nucleotides from each viral cDNA end adjacent to a conserved 3 0 -CA sequence, generating CA-3 0 hydroxyl recessed ends. In the strand transfer reaction, the newly processed 3 0 -viral DNA ends are inserted into the host
a
Computer-Aided Drug Design and Molecular Modeling Lab, Department of Bioinformatics, Alagappa University, Karaikudi-630 004, Tamil Nadu, India. E-mail: [email protected]; Fax: +91 4565 225202; Tel: +91 4565 223342 b Division of Toxicology, CSIR-Central Drug Research Institute, Lucknow-226 001, India † Electronic supplementary information (ESI) available. See DOI: 10.1039/c3mb70418a
This journal is © The Royal Society of Chemistry 2014
DNA.4,5 HIV-1 IN is composed of three structurally and functionally distinct domains: the amino-terminal domain (NTD, amino acids 1–49) contains a highly conserved zinc binding motif (HHCC) important for oligomerization; the catalytic core domain (CCD, amino acids 50–212) contains highly conserved Asp64, Asp116, and Glu152 (DDE) residues which are directly involved in the catalytic activities of IN; and the carboxy-terminal domain (CTD, amino acids 213–288) contains an SH3-like domain that binds non-specifically to DNA.6–9 HIV-1 IN has emerged as a promising target as this is responsible for integration of the newly synthesized doublestranded viral DNA into host genomic DNA.10,11 HIV-1 IN strand transfer inhibitors, which target the enzyme active site, have witnessed clinical success over the past few years, but the emergence of drug resistance poses challenges.12 HIV relies on host cellular machinery to complete its replication cycle. Cellular cofactors are important for integration function of IN.13 The lens epithelium-derived growth factor (LEDGF/p75), a transcriptional co-activator, was initially identified as an IN co-factor by co-immunoprecipitation from cells overexpressing HIV IN.14 The crucial role of LEDGF/p75 in HIV replication was evidenced via mutagenesis, RNAi-mediated depletion, transdominant overexpression of the integrase binding domain (IBD) of LEDGF/p75
Mol. BioSyst.
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Paper
Molecular BioSystems
Fig. 1 Domain organization of HIV-1 IN and LEDGF/p75 proteins. The key residues between the IN catalytic core domain (CCD) and the LEDGF/p75 integrase binding domain (IBD) are highlighted.
and knockout studies.15 The cellular cofactor LEDGF/p75 has been identified as the dominant binding partner of HIV-1 IN in human cells.14,16 LEDGF/p75 depletion impairs binding of IN to chromatin.17 A chromatin-binding domain (PWWP) is located at the N-terminus of LEDGF/p75, and the C-terminus contains the IBD. The domain organization of HIV-1 IN and LEDGF/p75 is shown in Fig. 1. The interaction between IN and LEDGF/p75 is an optimal target for HIV drug design. The host cell cofactor LEDGF/p75 plays an important role in the integration process by tethering IN to chromatin.18–20 Crystal structure of the dimeric CCD of HIV-1 IN complexed to the IBD of LEDGF/p75 gave a major advance for structure-based drug design targeting the IN–LEDGF/p75 protein–protein interaction (PPI). The structure has pseudo-two fold symmetry; two LEDGF/p75 IBD domains (residues 347–429) bind at nearly identical positions at either end of the IN dimer interface.20,21 The LEDGF/p75 IBD contacts residues from both chains of the IN dimer as shown in Fig. 2. The key interactions between HIV-1 IN and LEDGF/p75 have also been reported earlier.22,23 Specific cloning and quantification of the circular full site integration products attested that the IN–LEDGF/p75 complex catalyses 2 to 10 times more concerted integration events than isolated IN molecules.24 Blocking IN–LEDGF/p75 interaction can seriously inhibit viral replication.21 We used fast and inexpensive docking protocols combined with accurate but more expensive molecular dynamics (MD) simulation techniques to understand the stability of protein– protein and protein–ligand complex structures.25 PPIs are critical components of the machinery within living organisms and represent a large and important class of therapeutic targets. Use of PPI inhibitors will amplify the scope for discovery of drugs that can conquer HIV-resistance.14 The main
Mol. BioSyst.
aim of our study is to find small molecules that target the IN–LEDGF/p75 interaction, and thereby inhibit HIV replication. This is of particular interest in virology that demands new antivirals preferably with a completely different mechanism of action to avoid resistance and cross-resistance. The interaction between viral HIV-1 IN and its cellular cofactor LEDGF/p75 is one of the most promising targets at the moment. Identification of effective small molecule PPI inhibitors as new antiretroviral drugs would considerably boost generic interest in PPI inhibitors.
Materials and methods Protein model preparation The X-ray crystal structure of the dimeric CCD of HIV-1 IN complexed with IBD of LEDGF/p75 (PDB ID: 2B4J) was downloaded from Protein Data Bank.20 Both wild type and mutant forms of IN–LEDGF/p75 protein complexes were prepared by a multi-step process through Protein Preparation Wizard.26 All crystallographic water molecules and other chemical components were omitted, right bond orders as well as charges and atom types were assigned, and hydrogen atoms were added to the crystal structure. Each model was subjected to energy minimization using the Optimized Potentials for Liquid Simulations (OPLS)-2005 force field with implicit solvation. Each model was minimized and the minimization was terminated when root mean square deviation (RMSD) of heavy atoms in the minimized structure relative to the X-ray structure exceeded 0.3 Å. We prepared one wild type, four single mutant forms Q168A, E170A, H171A, and T174A, and three double mutant forms (Q168A and E170A) and (E170A
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Molecular BioSystems
Paper
Fig. 2 Co-crystal structure of HIV-1 IN CCD (green) complexed with LEDGF/p75 IBD (red). LEDGF/p75 IBD domains (residues 347–429) bind at nearly identical positions at either end of the IN dimer interface. Key contacts at the interface are highlighted.
and H171A) in chain A, and (Q95A and T125A) in chain B of IN. In order to perform virtual screening studies of HIV-1 IN, we removed the host protein LEDGF/p75 and prepared the HIV-1 IN CCD as described in the protocol earlier. Ligand preparation A total of 50 000 molecules from ChemBridge DIVERSet database were selected for structure-based virtual screening against the LEDGF/p75 binding site of HIV-1 IN. All the hydrogen atoms were added to the ligand molecules as they only had implicit hydrogen atoms, bond order of these ligands were fixed and ionization states of ligands were generated in the pH range of 5.0–9.0 using Epik during ligand preparation by LigPrep.27 Most probable tautomers and all possible stereo isomers were generated to study the activity of individual stereotypes of each ligand. In the final stage of LigPrep, compounds were minimized using the OPLS-2005 force field.28 After running the ligand preparation a total of 91 169 ligands were generated. Structure-based virtual screening We ran high throughput virtual screening (HTVS), standard precision (SP) and extra precision (XP) protocols using the program Glide.29,30 In docking simulation, we used semiflexible docking protocols. The ligands being docked were kept flexible, in order to explore an arbitrary number of torsional degrees of freedom spanned by the translational and rotational
This journal is © The Royal Society of Chemistry 2014
parameters. The ligand poses generated were passed through a series of hierarchical filters that evaluated ligand interaction with its receptor. The glide score was used as a function of fitness and best-fit pose for a given ligand was determined by the Emodel score. Prime MM-GBSA free energy The top docked poses for each ligand were rescored for calculation of binding free energy by the Prime/MMGBSA method.31 The binding free energy (DGbind) was calculated using the following equation:32 DGbind = DEMM + DGsolv + DGSA where DEMM is the difference in energy between the IN–inhibitor complex and the sum of the energies of the unliganded IN protein and the ligand. DGsolv is the difference in the GBSA solvation energy of the complex and the sum of solvation energies of the unliganded uncomplexed IN protein and the ligand. DGSA is the difference in surface area energy of the complex and the sum of surface area energies of the IN protein and the ligand. ADME properties prediction The identified best six hits were studied for their absorption, distribution, metabolism and excretion (ADME) properties by using QikProp.33 Accurate prediction of ADME properties prior to experimental procedures can eliminate unnecessary testing
Mol. BioSyst.
View Article Online
Paper
of compounds. The percentage of their human oral absorption was also predicted to determine their toxicity levels.
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Density functional theory calculations The molecules selected from the docking study were used as inputs for density functional theory (DFT) calculations. All DFT computations were carried out using Jaguar.34 Complete geometry optimization was carried out using hybrid DFT with Becke’s threeparameter exchange potential and Lee–Yang–Parr correlation functional (B3LYP), using the basis set 3-21G* level.35–38 Highest Occupied Molecular Orbitals (HOMOs) and Lowest Unoccupied Molecular Orbitals (LUMOs) were computed. Molecular dynamics simulations Protein and ligand structures used in Desmond simulation must be complete for all-atom 3D structures with a reasonable geometry. The protein and ligand structures used in simulation
Molecular BioSystems
were prepared using Protein Preparation Wizard and LigPrep. MD simulations of all wild type as well as mutant forms of IN–LEDGF/p75 and IN–ligand complexes were performed using Desmond Molecular Dynamics System39–41 with OPLS all-atom force field 2005.28,42 The complexes were immersed in an orthorhombic box containing TIP3P water molecules and neutralized by adding salt counter-ions. The distance between the box wall and complexes was set to greater than 10 Å to avoid direct interactions with its own periodic image. The energy of prepared systems for MD simulations was minimized up to a maximum of 5000 steps using a steepest descent method until a gradient threshold (25 kcal mol1 Å1) was reached, followed by LBFGS (Low-memory Broyden–Fletcher–Goldfarb–Shano quasi-Newtonian minimizer) until a convergence threshold of 1 kcal mol1 Å1 was met. The NPT ensemble, the Nose–Hoover chain thermostat, and the Martyna–Tobias–Klein barostat were employed for the actual simulation, and the RESPA integrator was used
Fig. 3 RMSD of the backbone of wild type and mutant forms during the simulation. (A) RMSD of the wild type and mutant forms of IN–LEDGF/p75 complexes. (B) RMSD of only LEDGF/p75 in the wild type and mutant forms of IN–LEDGF/p75 complexes through 10 ns MD simulations.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Molecular BioSystems
complexes were simulated for a 10 ns time period using the parameters described above. The atom numbers of wild type and mutant IN–LEDGF/p75 complexes are listed in ESI,† Table S1.
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
with a time step of 0.002 ps. A 9 Å cutoff was used for the Coulombic range interactions, and long-range electrostatics were treated using the particle mesh Ewald method, with a tolerance of 1 e9 Å. All IN–LEDGF/p75 and IN–ligand
Paper
Fig. 4 Binding modes and molecular interactions of three compounds 1–3 with the LEDGF/p75 binding site at the IN dimer interface (chain A – green and chain B – yellow color). The hydrogen bonds are displayed in black dashed lines with distances.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst.
View Article Online
Paper
atoms of compound 2 forming hydrogen bonding interactions with IN residues was calculated and plotted through 10 ns simulations.
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
The RMSDs for IN–LEDGF/p75 and IN–ligand complexes were calculated for the entire simulation trajectory with reference to their respective first frames. The distance between
Molecular BioSystems
Fig. 5 Binding modes and molecular interactions of three compounds 4–6 with a LEDGF/p75 binding site at the IN dimer interface (chain A – green and chain B – yellow color). The hydrogen bonds are displayed in black dashed lines with distances.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Molecular BioSystems
Paper
Results and discussion
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Mutational studies and MD simulations We performed in silico mutational studies on HIV-1 IN complexed with LEDGF/p75 to understand the role of specific residues in stabilizing the complex. The amino acids of HIV-1 IN that are important for the interaction with LEDGF/p75 were mutated to alanine. Single mutant forms Q168A, E170A, H171A, and T174A, double mutant forms (Q168A and E170A) and (E170A and H171A) in chain A and (Q95A and T125A) in chain B of IN were generated. The RMSDs of IN–LEDGF/p75 complexes versus time are given in Fig. 3A. All systems were not stable throughout the simulation time with a few structural changes in the complexes. The LEDGF/ p75 shows less stability in the double mutant (Q168A and E170A) form. The RMSDs of only bound LEDGF/p75 in both wild type and mutant forms are plotted to understand its stability [Fig. 3B]. Overall, the single and double mutants seem to affect the complex stability. Further, the side chain RMSDs of IN–LEDGF/p75 complexes are also given in ESI† (Fig. S1). All the key residues analyzed through mutational studies have an important role in stabilizing LEDGF/p75. All these observations suggest that the interactions between HIV-1 IN and LEDGF/p75 are governed by all reported key amino acid residues. The LEDGF/p75 binding interface is located in a pocket formed by the two subunits of the IN–core dimer. Residues located in the a-4/5 connector (loop) and the hydrophobic pocket of IN engage tightly with LEDGF/p75. Structure-based virtual screening In order to identify compounds targeting HIV-1 IN and block LEDGF/p75 interaction, we probe small-molecule inhibitors effective at disrupting the IN–LEDGF/p75 interaction. We performed structure-based virtual screening by docking the ChemBridge database compounds targeting the LEDGF/p75 binding site of HIV-1 IN. We used 50 000 compounds of the ChemBridge database initially which further generated 91 169 ligands after ligand preparation. Initially, we performed docking using a HTVS docking protocol which is the least computationally intense process for rapid screening of the ligands. Ligands were selected for further steps on the basis of the docking score, the cutoff was set as 5.0 i.e., only 3377 ligands were selected having a docking score below 5.0 kcal mol1. These 3377 ligands were docked in the same receptor using a SP docking protocol in
Table 1
which top 10% compounds were selected based on the docking score. 337 ligands were docked using a XP docking protocol which is a more powerful and discriminative procedure and takes a longer time than SP docking. Top 25 ligands from XP docking, after removing the duplicates were considered for visual inspection. The resulting compounds were further analyzed for their interactions with key amino acid residues in the LEDGF/p75 binding site of IN. The compounds were selected based on their interaction modes at the LEDGF/p75 binding site of HIV-1 IN. The present virtual screening exercise provided six compounds that were suitable starting points for lead optimization and that are expected to block the IN and LEDGF/p75 interaction. The flatness of protein–protein interfaces often hampers the identification of small molecule PPI inhibitors. In this case, the LEDGF/p75 binds to a defined pocket in the dimer interface of HIV-1 IN CCD. Due to the tight and small interface of the LEDGF/p75 binding site on HIV-1 IN, the design and/or selection of small molecules binding to this pocket is a feasible endeavour.43 Binding mode analysis The compounds showed stable interactions with the LEDGF/ p75 binding site of IN. The interactions of top six compounds with the IN obtained in the docking study are shown in Fig. 4 and 5. From the docking results, amino acid residues Glu170, His171, and Thr174 in chain A and Gln95 and Thr125 in chain B of HIV-1 IN CCD were seen as key players in the binding of compounds. All selected six lead hits showed hydrogen bonding interactions with Glu170, His171 and Thr174 in chain A and Thr125 in chain B of IN. They also showed hydrophobic interactions with IN B-chain residues Ala98, Leu102, Ala128, Ala129 and Trp132 and the IN A-chain residue Met178, at the LEDGF/ p75 binding site of the HIV-1 IN CCD dimer interface. Carboxyl groups of compounds 1 and 2 form hydrogen bonds with main chain nitrogens of Glu170 and His171 and the hydroxyl group of Thr174 in chain A of IN. Furthermore, Gln95 and Thr125 residues formed hydrogen bonds with compounds 1 and 2. In the IN–LEDGF/p75 core complex, equivalent hydrogen bonds were formed with the side chain of Asp366 located in the LEDGF/p75 IBD loop. HIV-1 IN residues Thr125, Glu170, His171, and Thr174 formed hydrogen bonds and binding free
Glide docking and Prime-MMGBSA results for the best six compounds
Compound
ChemBridge ID
1
9123953
2
7662210
3
5791465
4
9057093
5
5314403
6
9120780
H bond interactions
Hydrophobic interactions
Glu170, Thr174, Glu170, Thr174, Glu170, Thr174, Glu170, Thr174 Glu170, Thr174, Glu170, Thr174,
Met178, Leu102, Ala128, Ala129, Trp132 Ala169, Met178, Ala98, Leu102, Ala128, Ala129, Trp132 Met178, Leu102, Ala128, Ala129, Trp132 Met178, Ala98, Leu102, Ala128, Ala129, Trp132 Met178, Ala98, Leu102, Ala128, Ala129, Trp132 Met178, Ala98, Leu102, Ala128, Ala129, Trp132
His171, Gln95, Thr125 His171, Thr125 His171, Gln95, Thr125 His171, His171, Gln95 His171, Thr125
This journal is © The Royal Society of Chemistry 2014
Glide docking score (kcal mol1)
Glide Emodel score
Prime-MMGBSA DGbind (kcal mol1)
8.37
43.80
17.04
8.27
53.34
29.07
7.83
45.02
13.08
7.47
45.99
6.70
7.24
45.27
17.48
7.029
47.447
18.969
Mol. BioSyst.
View Article Online
Paper
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Table 2
Molecular BioSystems ADME or pharmacokinetic predictions of the best six compounds
Compound
ChemBridge ID
QP log Pa
QP log Sb
QP PCaco2c
Percentage of human oral absorptiond
1 2 3 4 5 6
9123953 7662210 5791465 9057093 5314403 9120780
1.54 4.025 0.356 3.822 1.892 1.697
2.665 4.346 0.913 4.151 2.507 2.509
57.818 145.341 312.841 1824.883 92.037 85.785
67.503 89.213 73.694 100 73.175 71.488
The permissible ranges are as follows. a Log of the octonal–water partition coefficient (2.0 to 6.5). b Log of the aqueous solubility S (mol L1): (6.5 to 0.5). c Caco2 cell permeability in nm s1: (o25 is poor and >500 is great). d Percentage of human oral absorption: >80% high, o25% low.
Table 3
Frontier orbital energies of the best six compounds
Compound
HOMO (eV)
LUMO (eV)
HLG (eV)a
1 2 3 4 5 6
0.24 0.24 0.21 0.21 0.22 0.23
0.01 0.03 0.06 0.01 0.05 0.04
0.23 0.21 0.15 0.20 0.17 0.19
a
HLG is the HOMO–LUMO energy gap.
energy (29.065 kcal mol1) with compound 2 was high. Compound 3 interacts with Gln95, Thr125, Glu170, His171, and Thr174 forming a total of five hydrogen bonds. In the case of compound 4 the docking results showed that the nitrogen atom of the pyrimidine ring and the oxygen atom of the compound established hydrogen bonding with Glu170, His171 and Thr174 respectively. The carboxyl group and the nitrogen atom of compound 5 formed hydrogen bonds with Gln95, Glu170, His171 and Thr174. From the docking conformation of compound 6 it was observed that the oxygen atom of ether and the carboxyl group of the compound established hydrogen bonding with Thr125, Glu170, His171 and Thr174 respectively. Compounds 4, 5 and 6 showed a similar binding mode establishing hydrophobic interactions with Met178, Ala98, Leu102, Ala128, Ala129 and Trp132 respectively. All the six compounds were rescored using Prime/MM-GBSA. These approaches predict the binding free energy for a set of ligands to receptor. The docking results and binding free energy of the best six compounds are shown in Table 1. ADME prediction The predictions of pharmacologically important properties were computed using QikProp for the selected compounds as shown in Table 2. Different properties consisting of principal descriptors and physicochemical properties were calculated. When QikProp was run for a set of 1700 oral drugs, 95% were predicted to have molecular weights between 130 and 725, log p values between 2 and 6, log S values between 6.0 and 0.5, and Caco2 values greater than 25 nm s1.44 The predicted properties of the six identified hits compared favorably with these ranges and also found a very high human oral absorption. The newly searched compounds appear to be suitable for further development. DFT analysis
Fig. 6 Plots of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of compound 2.
Frontier orbital energies and the HOMO–LUMO gap were calculated for the best six compounds [Table 3]. Both the
Fig. 7 Ca RMSD from the initial structures of protein–ligand complexes over time during MD simulations.
Mol. BioSyst.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Molecular BioSystems
HOMO and LUMO energies are small, ranging between 0.24 and 0.21 and 0.06 and 0.01 eV, respectively, indicating the fragile nature of bound electrons. Because of small values of HOMO and LUMO, both rapid electron transfer and exchange are equally possible, making these molecules very reactive. HOMO and LUMO sites are plotted onto the surface of compound 2 and shown in Fig. 6. Analysis of the HOMO map of compound 2 illustrates that HOMO molecular orbitals are located on the carboxyl group indicating the existence of reactive sites and electrophilic attack takes place at the residues of the active site of HIV-1 IN. The LUMO map is plotted on the sulphur group and the benzene group. The HOMO–LUMO gap is the potential energy difference between the HOMO and the LUMO. A small HOMO–LUMO gap means small excitation
Fig. 8 Intermolecular hydrogen plot between IN and compound 2 through 10 ns simulations.
Fig. 9
Paper
energies to the manifold of excited states. Therefore, molecules with a small gap, will be more polarizable than the molecules with large gap. This helps in the reaction of these compounds with the surrounding amino acids of the receptor. MD simulations of protein–ligand complexes The docking protocols can be combined with accurate but more costly MD techniques to predict more reliable protein–ligand complexes. In order to gain insight into the stability and dynamic properties of the complexes, 10 ns MD simulations were performed. We assessed the dynamics stability of all the complexes during the MD simulation and the RMSDs of backbone Ca atoms were analyzed and plotted in Fig. 7. MD simulation results revealed that the RMSD of the complexes almost reached equilibrium at B1000 ps. After the initial deviation, the complexes did not deviate further and showed consistent RMSD of around 1 Å throughout the simulation process. All the ligands formed stable hydrogen bonding interactions with Glu170, His171, and Thr174 in chain A and Thr125 in chain B of IN. Hydrophobic interactions with Ala98, Leu102, Ala128, Ala129, Trp132 and Met178 are also stable throughout MD simulations. Stable hydrogen bonds among protein and ligands strengthen the binding affinity of all the complexes. Recently, the small molecule inhibitors identified for the newly explored binding site region and site mutagenesis studies highlighted Gln168, Glu170, His171, and Thr174 in chain A and Glu95, Thr125, and Trp131 in chain B of HIV-1 IN as key amino acid residues.45–48 The first reported potent molecule targeting the LEDGF/p75 binding site of HIV-1 IN is the benzoic acid derivative, 4-[(5-bromo-4-{[2,4-dioxo-3-(-oxo-2-phenylethyl)-1,3thiazololidin-5-ylidene] methyl}-2-ethoxyphenoxy)methyl]benzoic acid (D77), which interacts with residues Gln95, Thr125, Trp131 and Thr174.49 The series of 2-(quinolin-3-yl) acetic acid derivatives
The interaction of compound 2 before and after the MD simulations.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst.
View Article Online
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
Paper
Molecular BioSystems
Fig. 10 The atomic distances between the atoms involved in hydrogen bond formation through MD trajectory based analysis of the IN–compound 2 complex. Distance between the hydrogen bonding residues Glu170, His171, Thr174, and Thr125 and compound 2 depicted for the entire 10 ns simulation.
(LEDGINs) were the first genuine allosteric HIV-1 IN inhibitors shown to form hydrogen bonds with residues Glu170, His171, and Thr174 in chain A of HIV-1 IN.50 They have not shown any hydrogen bonding interaction with chain B residues. Recently, De Luca et al. reported the interactions between LEDGINs and the LEDGF/p75 binding site in HIV-1 IN combining docking and molecular dynamics simulations.51 Compared with LEDGINs, these newly identified compounds have significant potential for further development. The reported compounds in this study showed very good interactions with the critical residues of Glu170, His171, and Thr174 in chain A. They also showed interactions with Thr125 and Trp131 in chain B which are critical for LEDGF/p75 binding. We herein present potent small molecules of HIV-1 IN that bind to the LEDGF/p75 binding site, which have groups important for the interaction at the dimeric interface. MD analysis of HIV-1 IN complexed with compound 2 MD analysis of the IN–compound 2 complex stability during simulation was monitored by understanding the RMSD profiles, the number of hydrogen bonds and the distance between atoms during the MD trajectory. The hydrogen bond profile of the IN–compound 2 complex, as displayed in Fig. 8, reveals that the complex is stabilized by 4–5 hydrogen bonds (Glu170, His171, Thr174, Gln95 and Thr125) throughout 10 ns simulations. Fig. 9 clearly shows that stacking interaction with His171 and hydrogen bonding interactions with Glu170, His171, Thr174, and Thr125 are similar before and after the MD simulations. The compound is also stabilized by more hydrophobic interactions with Ala169, Met178, Ala98, Leu102, Ala128, Ala129 and Trp132. The nitrogen atoms of Glu170 and His171, and OG1 of Thr174 form a stable interaction with the carboxyl group of compound 2. The oxygen atom of Thr125 forms stable interaction with the sulfonyl group of compound 2. The atomic distances between atoms of IN and compound 2
Mol. BioSyst.
through 10 ns MD simulations are shown in Fig. 10. We found an additional stable hydrogen bond formed with Gln95 during simulation. The stacking interaction between the benzene ring of compound 2 and His174 is also stable throughout the MD simulation.
Conclusion The interaction between HIV-1 IN and LEDGF/p75 is an attractive novel target and has increasingly gained attention for antiHIV drug discovery. In this study, an integrated computational approach by mutational studies, MD simulations, structurebased virtual screening, MM-GBSA binding free energy and DFT analysis was used to find the inhibitors which can block the PPI between HIV-1 IN and human LEDGF/p75. We identified HIV-1 IN inhibitors that did not target the catalytic site through structure-based virtual screening with docking simulations. Analysis of the docking modes of ligands showed that the most crucial residues in the binding site are Glu170, His171, and Thr174 in the A-chain and Gln95 and Thr125 in the B-chain of IN. Owing to criticality of these inferences, these ligands can prove to be highly effective if incorporated into drug development phases and experimental studies. These inhibitors could block the PPI between a viral protein HIV-1 IN and a cellular host factor LEDGF/p75. The computational methods used and the hits identified in this study provide an excellent guide to disrupt the IN–LEDGF/p75 interaction.
Acknowledgements KKR thanks Council of Scientific and Industrial Research (CSIR), New Delhi, India, for awarding a Senior Research Fellowship. The authors thank M. Ravikumar for the detailed discussion and suggestions.
This journal is © The Royal Society of Chemistry 2014
View Article Online
Molecular BioSystems
Published on 03 December 2013. Downloaded by Heinrich Heine University of Duesseldorf on 07/01/2014 15:23:58.
References 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
UNAIDS, Report on the Global HIV/AIDS Epidemic, 2012. R. A. Weiss, Science, 1993, 260, 1273–1279. D. J. McColl and X. Chen, Antiviral Res., 2009, 85, 101–118. S. A. Chow, K. A. Vincent, V. Ellison and P. O. Brown, Science, 1992, 255, 723–726. G. Goodarzi, G. J. Im, K. Brackmann and D. Grandgenett, J. Virol., 1995, 69, 6090–6097. M. Cai, R. Zheng, M. Caffrey, R. Craigie, G. M. Clore and A. M. Gronenborn, Nat. Struct. Biol., 1997, 4, 567–577. F. Dyda, A. B. Hickman, T. M. Jenkins, A. Engelman, R. Craigie and D. R. Davies, Science, 1994, 266, 1981–1986. A. Engelman and R. Craigie, J. Virol., 1992, 66, 6361–6369. D. Esposito and R. Craigie, Adv. Virus Res., 1999, 52, 319–333. M. H. Nielsen, F. S. Pedersen and J. Kjems, Retrovirology, 2005, 2, 10. K. K. Reddy, S. K. Singh, N. Dessalew, S. K. Tripathi and C. Selvaraj, J. Enzyme Inhib. Med. Chem., 2012, 27, 339–347. A. Engelman, J. J. Kessl and M. Kvaratskhelia, Curr. Opin. Chem. Biol., 2013, 17, 339–345. B. Van Maele, K. Busschots, L. Vandekerckhove, F. Christ and Z. Debyser, Trends Biochem. Sci., 2006, 31, 98–105. P. Cherepanov, G. Maertens, P. Proost, B. Devreese, J. Van Beeumen, Y. Engelborghs, E. De Clercq and Z. Debyser, J. Biol. Chem., 2003, 278, 372–381. F. Christ and Z. Debyser, Virology, 2013, 435, 102–109. M. Tsiang, G. S. Jones, M. Hung, S. Mukund, B. Han, X. Liu, K. Babaoglu, E. Lansdon, X. Chen, J. Todd, T. Cai, N. Pagratis, R. Sakowicz and R. Geleziunas, J. Biol. Chem., 2009, 284, 33580–33599. M. C. Shun, Y. Botbol, X. Li, F. Di Nunzio, J. E. Daigle, N. Yan, J. Lieberman, M. Lavigne and A. Engelman, J. Virol., 2008, 82, 11555–11567. O. Taltynov, B. A. Desimmie, J. Demeulemeester, F. Christ and Z. Debyser, Mol. Biol. Int., 2012, 2012, 863405. E. M. Poeschla, Cell. Mol. Life Sci., 2008, 65, 1403–1424. P. Cherepanov, A. L. Ambrosio, S. Rahman, T. Ellenberger and A. Engelman, Proc. Natl. Acad. Sci. U. S. A., 2005, 102, 17308–17313. L. Q. Al-Mawsawi and N. Neamati, Trends Pharmacol. Sci., 2007, 28, 526–535. Y. Zhao, W. Li, J. Zeng, G. Liu and Y. Tang, Proteins, 2008, 72, 635–645. C. Tintori, N. Veljkovic, V. Veljkovic and M. Botta, Proteins, 2010, 78, 3396–3408. B. Maillot, N. Levy, S. Eiler, C. Crucifix, F. Granger, L. Richert, P. Didier, J. Godet, K. Pradeau-Aubreton, S. Emiliani, A. Nazabal, P. Lesbats, V. Parissi, Y. Mely, D. Moras, P. Schultz and M. Ruff, PLoS One, 2013, 8, e60734. M. Karplus and J. A. McCammon, Nat. Struct. Biol., 2002, 9, 646–652. ¨dinger Suite 2011 Protein Preparation Wizard, Epik Schro ¨dinger, LLC, New York, NY, 2011; Impact version 2.2, Schro ¨dinger, LLC, New York, NY, 2011. version 5.7, Schro
This journal is © The Royal Society of Chemistry 2014
Paper
¨dinger, LLC, New York, NY, 2011. 27 LigPrep, version 2.5, Schro 28 W. L. Jorgensen, D. S. Maxwell and J. Tirado-Rives, J. Am. Chem. Soc., 1996, 118, 11225–11236. ¨dinger, LLC, New York, NY, 2011. 29 Glide, version 5.7, Schro 30 R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren, J. J. Klicic, D. T. Mainz, M. P. Repasky, E. H. Knoll, M. Shelley, J. K. Perry, D. E. Shaw, P. Francis and P. S. Shenkin, J. Med. Chem., 2004, 47, 1739–1749. ¨dinger, LLC, New York, NY, 2011. 31 Prime, version 3.0, Schro 32 P. D. Lyne, M. L. Lamb and J. C. Saeh, J. Med. Chem., 2006, 49, 4805–4808. ¨dinger, LLC, New York, NY, 2011. 33 QikProp, version 3.4, Schro ¨dinger, LLC, New York, NY, 2011. 34 Jaguar, version 7.8, Schro 35 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B: Condens. Matter Mater. Phys., 1988, 37, 785–789. 36 A. D. Becke, J. Chem. Phys., 1993, 98, 5648–5652. 37 J. S. Binkley, J. A. Pople and W. J. Hehre, J. Am. Chem. Soc., 1980, 102, 939–947. 38 W. J. Pietro, M. M. Francl, W. J. Hehre, D. J. DeFrees, J. A. Pople and J. S. Binkley, J. Am. Chem. Soc., 1982, 104, 5039–5048. 39 Desmond Molecular Dynamics System, version 3.1, D. E. Shaw Research, New York, NY, 2012; Maestro-Desmond Interoper¨dinger, New York, NY, 2012. ability Tools, version 3.1, Schro 40 Z. Guo, U. Mohanty, J. Noehre, T. K. Sawyer, W. Sherman and G. Krilov, Chem. Biol. Drug Des., 2010, 75, 348–359. 41 A. Grover, S. P. Katiyar, J. Jeyakanthan, V. K. Dubey and D. Sundar, BMC Genomics, 2013, 13(Suppl 7), S20. 42 G. A. Kaminski, R. A. Friesner, J. Tirado-Rives and W. L. Jorgensen, J. Phys. Chem. B, 2001, 105, 6474–6487. 43 D. P. Ryan and J. M. Matthews, Curr. Opin. Struct. Biol., 2005, 15, 441–446. 44 V. V. Thakur, J. T. Kim, A. D. Hamilton, C. M. Bailey, R. A. Domaoal, L. Wang, K. S. Anderson and W. L. Jorgensen, Bioorg. Med. Chem. Lett., 2006, 16, 5664–5667. 45 L. De Luca, S. Ferro, R. Gitto, M. L. Barreca, S. Agnello, F. Christ, Z. Debyser and A. Chimirri, Bioorg. Med. Chem., 2010, 18, 7515–7521. 46 K. Busschots, A. Voet, M. De Maeyer, J. C. Rain, S. Emiliani, R. Benarous, L. Desender, Z. Debyser and F. Christ, J. Mol. Biol., 2007, 365, 1480–1492. 47 G. Hu, X. Li, X. Sun, W. Lu, G. Liu, J. Huang, X. Shen and Y. Tang, J. Mol. Model., 2012, 18, 4995–5003. 48 G. Hu, X. Li, X. Zhang, Y. Li, L. Ma, L. M. Yang, G. Liu, W. Li, J. Huang, X. Shen, L. Hu, Y. T. Zheng and Y. Tang, J. Med. Chem., 2012, 55, 10108–10117. 49 L. Du, Y. Zhao, J. Chen, L. Yang, Y. Zheng, Y. Tang, X. Shen and H. Jiang, Biochem. Biophys. Res. Commun., 2008, 375, 139–144. 50 F. Christ, A. Voet, A. Marchand, S. Nicolet, B. A. Desimmie, D. Marchand, D. Bardiot, N. J. Van der Veken, B. Van Remoortel, S. V. Strelkov, M. De Maeyer, P. Chaltin and Z. Debyser, Nat. Chem. Biol., 2010, 6, 442–448. 51 L. De Luca, F. Morreale and A. Chimirri, J. Chem. Inf. Model., 2012, 52, 3245–3254.
Mol. BioSyst.
