Structure-based identification of natural compound inhibitor against M. tuberculosis thioredoxin reductase: insight from molecular docking and dynamics simulation

Abstract Antioxidant systems of M. tuberculosis (Mtb) play an important role in providing resistance in the hostile environment of mononuclear phagocytes. Thioredoxin system is a known antioxidant system that consists of three copies of thioredoxins (Trxs) and a single copy of thioredoxin reductase (TrxR). TrxR has been validated as an essential gene known to be involved in the reduction of peroxides, dinitrobenzenes and hydroperoxides, and is crucial in maintaining the survival of Mtb in macrophages. Recently, it has been demonstrated to be a druggable target. In this study, molecular docking was applied to screen more than 20,000 natural compounds from the Traditional Chinese Medicine database. Theoretical calculation of ΔGbinding by the Molecular Mechanics Poisson-Boltzmann Surface Area (MM-PBSA) methods indicated two top-hit compounds that bind with a high affinity to the allosteric site, consisting of a hinge region, of TrxR. Further, stability and binding analysis of both compounds were carried out with molecular dynamics simulation. An analysis of conformational variation by principal component analysis (PCA) and protein contact network (PCN) uncovered the conformational changes in the compound-bound forms of protein. The NADPH domain formed many new interactions with the FAD domain in the compound-bound form, signifying that the binding may render an effect on the protein structure and function. Our results suggest that these two compounds could potentially be used for structure-based lead inhibitors against TrxR. The inhibitor selected as lead compound will be used further as a scaffold to optimize as novel anti-tuberculosis therapeutic. Communicated by Ramaswamy H. Sarma


Introduction
Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis that accounts for 1.3 million deaths worldwide every year (World Health Organization [WHO], 2018).Although many novel therapeutics have been developed, still the evolving resistance strains such as MDR (multi-drugresistant) and XDR (extensively-drug-resistant) pose a great challenge to counter TB globally (Libardo et al., 2018).Moreover, the current first line of drugs offers a long treatment regimen.Hence, a search for new drugs or drug-like molecules that can augment the effectiveness of the current treatment and reduce the emerging resistance is warranted.
Oxidative stress and accumulation of reactive oxygen species (ROS) provide a hostile environment in the cell, which is an inevitable challenge to invading pathogens (Imlay, 2013).Pathogens such as Mtb resides in the host phagocytes and copes with the hostile environment by expressing a number of antioxidant systems to ensure their survival inside its host (Trivedi et al., 2012).The thioredoxin and glutathione systems are two well-known antioxidant systems that provide reducing environments and regulate many important cellular processes such as antioxidant pathways, DNA and protein repair enzymes, and the activation of redox dependent transcription factors (Fahey, 2001;Lu & Holmgren, 2014).Hence, antioxidant systems have been considered as a potential drug target.
Mtb thioredoxin system consists of two typical thioredoxins (TrxB and TrxC) and a single copy of thioredoxin reductase (TrxR) (Akif, Khare, Tayagi, Mande, & Sardesai, 2008).TrxR reduces the Trxs by typically utilizing the reducing potential from the cellular NADPH.The reduced Trxs function in reducing the peroxides and dinitrobenzenes, and also play an important role in detoxifying hydroperoxides.This signifies the importance of thioredoxin system for the survival of a pathogen in the hostile environment of macrophages (Jaeger et al., 2004;Zhang, Hillas & Montellano, 1999).Interestingly, as reported earlier for E. coli and other species, reduced Trxs are essential for nucleotide biosynthesis as they donate the reducing equivalents to ribonucleotide reductase (Lu & Holmgren, 2014).Transposon-mediated analysis has validated Mtb TrxR as an essential gene.Moreover, it has been reported that deletion of the trxR gene results in a hyper-susceptible strain (Li et al., 2016;Zhang et al., 2012).TrxR is also shown to be essential for the growth of other organisms such as S. aureus (Uziel et al., 2004) and B. subtilis (Scharf et al., 1998).Thus, considering its crucial role, the bacterial TrxR has recently been demonstrated as a promising drug target (Lu et al., 2013).
The success of a structure-based drug design depends on a selective inhibition of the target compared to its counterpart in the human host.Fortunately, Mtb TrxR differs with the human TrxRs in terms of structure, sequence, and the mode of transfer of reducing equivalent.Human and mammal TrxR is a high molecular weight protein which, apart from NADPHand FAD-binding domains, consists of an extra flexible C-terminal extension of the cysteine-selenocysteine-glycine (CUG) motif as a redox center that transfers the reduction potential to Trxs (Lu & Holmgren, 2014).In contrast, Mtb and other prokaryotes TrxR are a low molecular weight protein with distinct NADPH and FAD domains connected through a two-stranded b-sheet known as a hinge region (Akif et al., 2005;Waksman et al., 1994).The buried redox cysteine residues (CXXC) come out on the surface by a 67 � rotation of the NADPH domain with respect to the FAD domain during catalysis to provide reducing potential to Trxs.This unique feature of TrxR has been demonstrated with complex crystal structure of E. coli TrxR-Trx and Mtb TrxR structure (Akif et al., 2005;Lennon et al., 2000).In earlier reports, TrxR-Trx interaction site of Mtb was selected as a druggable target (Koch et al., 2013).But the main obstacle was a low hit rate of protein-protein interaction site and was considered not to be easily druggable.Thus, targeting an allosteric pocket near the NADPH domain, consisting of the interface of NADPH and FAD domains, active site, and hinge region, may bring about structural changes in the protein such that active site residues cannot be available on the surface to interact with Trxs.In recent years, TrxR from various bacterial species has been targeted, which has yielded some compounds demonstrated to inhibit TrxR, includes ebselene (Gustafsson et al., 2016;Lu et al., 2013), auranofin (Harbut et al., 2015;Owings et al., 2016) and gold (I)-alkynyl chromones (Hikisz et al., 2015).Still, an investigation of inhibitors with a strong specificity toward Mtb TrxR is needed.
Natural compounds have been the single most productive source of leads for the development of drugs.Traditional Chinese Medicine (TCM) (Chen, 2011) is a database of natural compounds that follow the Lipinski's rule of five and almost all compounds have a therapeutic effect.The current study aimed for the identification of natural compound suitable for the potential inhibitors of Mtb TrxR.All natural compounds present in the TCM database were used to carry out virtual screening against Mtb TrxR.This led to the identification of several compounds on the basis of docking scores and the top two compounds were selected for further study.A detailed study of docking and molecular dynamics simulation (MDS) along with MM/PBSA of the selected compounds showed that they form a stable protein-ligand complex.Principal component analysis (PCA) and alteration in solvent accessibility studies confirmed the change in conformation of protein.Protein contact network (PCN) graphs of simulated structures of both the apoand compound-bound complexes showed the changes in the degree of connectivity within the hinge region residues among the two structures, which signifies that binding of compounds affect the protein structure and function.This is the first report where natural compounds have been screened as an inhibitor scaffold against Mtb TrxR.

Preparation of the target coordinate
The target crystal structure of Mtb TrxR (PDB ID 2A87) (Akif et al., 2005) was downloaded from the protein database and subjected for preparation by adding/fixing the missing side chains using what-if server (Vriend, 1990).Water and heteroatoms molecules were removed from the coordinates.Energy minimization of 25,000 cycles of the steepest descent and 20,000 cycles of the conjugate gradient were performed through Swiss PDB viewer.

Virtual screening and molecular docking
Crystal structure of Mtb TrxR was used to probe surface cavity predictions by CastP analysis (Tian et al., 2018).One of the well-defined allosteric cavities (area 1301.594A 2 and volume 855.756A 3 ) was selected for the virtual screening of natural compounds using iScreen web server (Tsai, Chang, & Chen, 2011).iScreen is a robust screening and docking server with 1) TCM integrated into CADD (computer aided drug design) services, 2) PLANTs module to evaluate the docking score, and 3) E-LEAD3D for de-novo docking.More than 20,000 compounds present in the TCM database that follow the Lipinski's rule of five were screened to bind on the cavity of TrxR through PLANTS docking module.The docking algorithm of PLANTS is based on ant colony optimization and provides various conditions of docking.Out of the 200 docked compounds, 2 best compounds were sorted out based on the docking scores of binding in the cavity.These two compounds with the highest scores were further re-docked in the binding cavity by a denovo drug design method implemented in LEAD3D software module of the iScreen package.

Molecular dynamics simulation
To study the structural stability of Mtb TrxR-compound complexes, MD simulations were carried out using GROMACS 5.1.4(Hess et al., 2008;Pronk et.al. 2013).The topology files of the protein and compounds were generated using GROMOS96 45a3 force field (Oostenbrink et al., 2004) and the PRODRG2 server (Sch€ uttelkopf & van Aalten, 2004), respectively.The TIP3P water model was implemented for water molecules.The complex system was solvated in a cubic box with water molecules of 1.5 nm to the box wall from the surface of the protein and the system was further neutralized by adding Na þ counter ions.To minimize the short-range bad contacts, energy minimization was carried out using the steepest descent method for 50,000 steps until the largest force acting in the system was smaller than 10,000 kJ/mol/nm.After energy minimization, the temperature was equilibrated first in an NVT ensemble at 300 K for 50 ps using a modified V-rescale Berendsen thermostat with a time constant of 0.1 ps, followed by NPT ensemble to 1 atm using Parrinello-Rahman coupling method with a time constant of 2 ps for 50 ps.After the systems were equilibrated, the production run was performed for 20 ns at 300 K.The equations of motion were integrated with time steps of 2 fs and the coordinates were saved for every 2500 time steps (5 ps), which resulted in total 4000 frames for a 20 ns simulation.The long-range electrostatics was controlled using Particle Mesh Ewald (PME) method with a space cutoff of 10 Å.The hydrogen bonds were constrained by implementing the P-LINCS algorithm (Hess, 2008).Whole analysis was done using the frames from the production run (4000 for 300 K).RMSD, RMSF, and SASA were respectively calculated using g_rms, g_rmsf, and g_sasa functions of GROMACS.The protein structures were visualized and the figures were generated using PyMOL, XMGRACE, GNUPLOT, and VMD.

Binding-free energy calculations of compounds
Molecular mechanics Poisson Boltzmann surface area (MMPBSA) remains the most widely used method for binding-free energy calculations from the snapshots of the MD trajectory (Kollman et al., 2000).MMPBSA calculates free energy interaction of protein-ligand complexes in three steps: 1) calculates potential energy in the vacuum, 2) calculates polar solvation energy, and 3) calculates non-polar solvation energy.The binding-free energies of the complexes between compounds and TrxR were analyzed during equilibrium phase by capturing snapshots from the last frames of 15 to 20 ns MD simulations, using g_mmpbsa tool of GROMACS (Kumari, Kumar, Open Source Drug Discovery Consortium & Lynn, 2014).Particularly, the binding-free energy of protein-ligand complex in the solvent was expressed as: where G complex is the total free energy of the protein-ligand complex, G protein and G ligand, respectively, are total energy of protein and ligand, in the solvent.The free energy for each individual G complex , G protein and G ligand were estimated by where x is the protein, ligand, or complex, E mm is the average molecular mechanics potential energy in vacuum, and G solvation is the free energy of the solvation.The molecular mechanics potential energy was calculated in vacuum as: where E bonded is bonded interaction including bond, angle, dihedral, and improper interactions and E non-bonded is nonbonded interactions consisting of van der Waals (E vdw ) and electrostatic (E elec ) interactions.DE bonded is always taken as zero.
The solvation free energy (G solvation ) was estimated as the sum of electrostatic solvation free energy (G polar ) and apolar solvation free energy (G non-polar ): where G polar was computed using the Poisson-Boltzmann (PB) equation and G non-polar was estimated from the solventaccessible surface area (SASA) as: where c is a coefficient related to the surface tension of the solvent and b is a fitting parameter.The values of the constant are as follows:

Principal component analysis
Principal component analysis (PCA) was performed to obtain a mass-weighted covariance matrix of the protein atom displacement, which is indicative of the dominant and collective modes of the protein.This covariance matrix is diagonalized to extract a set of eigenvectors and eigenvalues that accurately reflect the concerted motion of the molecule.The GROMACS inbuilt tool g_covar was used to yield distinct eigenvalues and eigenvectors by calculating and diagonalizing the covariance matrix, whereas the g_anaeig tool was used to analyze and plot the eigenvectors (David & Jacobs, 2014).The first two eigenvectors represent the highest eigenvalues and are adequate to capture the overall motion of the protein.

Protein contact networks analysis
Protein contact network (PCN) was performed to carefully analyze changes between the interacting residues by the comparing contact maps of the apo and holo complex structures of the TrxR using CMView (Vehlow et al., 2011).The network is a graph where each residue corresponds to a node, and two nodes are connected by an edge if and only if the two residues are in contact.Two residues are considered to be in contact if they are spatially close in the threedimensional structure and are specified by two key parameters: contact type and distance cutoff.The contact type defines a subset of atoms of the residue.The most commonly used cutoff values for C a -based PCNs are 0.7 nm to 0.8 nm.RING 2.0 web server was used for the identification of both covalent and non-covalent bonds in protein structures (Piovesan et al., 2016).The RING output was visualized directly using Pymol and the python script of RING-Viz script.

Results and discussion
Discovery of anti-tuberculosis drugs have been a challenge due to alarming emergence of MDR and XDR strains of Mtb, making tuberculosis a global health threat.Moreover, the existing anti-tuberculosis drugs have been associated with various side effects.Therefore, extended efforts are needed for the identification of a novel target and the discovery of a specific anti-tuberculosis drug.Many computational tools are available for the structure-based drug design and optimization of lead compounds.Mtb TrxR has been shown to be a druggable target, which helps mycobacteria to survive against oxidative killing in the host cell.Earlier inhibitors of Mtb TrxR were designed based on a protein-protein interaction site.It has been known that protein-protein interaction site has a low hit rate in inhibitor designing efforts.Moreover, it is not easily druggable.Allosteric sites on protein are considered to be a druggable target for designing a selective inhibitor.Hence, an allosteric site is chosen for the specific targeting of Mtb TrxR.Mtb TrxR has a unique feature of catalysis compared to the higher eukaryotic TrxR.It requires to rotate the NADPH domain through hinge region to bring about the catalysis.Hence, the cavity around the hinge region is targeted in our study using virtual screening of natural compounds from the TCM database.

Structure-based virtual screening and docking
The virtual screening function in iScreen identifies potential TCM compounds by a docking algorithm based on the protein structure and the binding site information.The predicted cavity with an area of 1301.554A 2 and volume 855.756A 2 located between the NADPH and FAD domains, as shown in Figure 1A, was utilized for the binding analysis of TCM compounds and which in turn generated a docking score of each compound (Table S1, supplementary material).Among all compounds, two promising compounds (called here compound1 and compound2) were selected based on their docking score value and interaction orientation in the complex.Pharmacokinetic study of inhibitor compounds plays a critical role in the development of a drug.Since, these compounds are screened from the TCM database, they have acceptable absorption, distribution, metabolism, and excretion criteria.The two-dimensional chemical structures of the screened compounds are shown in Figure 1B.

Binding interaction analysis of hit compounds
Binding interactions of hit compounds with the binding cavity in TrxR were analyzed with LIGPLOT (Wallace et al., 1995) as well as manually (Figure 2(A,B)).Different types of molecular binding interactions were observed to stabilize the compounds in the binding cavity and the most prominent is the hydrophobic bonding interaction with important residues of near the active site and hinge region as well as the interface residues of NADPH and FAD domains.The binding pattern within the cavity is almost similar in both compounds except that three additional hydrogen bonds were observed between compound2 and the side chains of Arg295, Glu170, and His250.Interestingly, it was also observed that cation pi and pi-pi interactions in both cases are crucial for an enhanced binding stability in the cavity.It is not surprising that hydrophobic interactions also contribute to the stable binding of the inhibitor to the protein target, as it has been reported earlier for the inhibition of MDM2 by polyphenols (Verma, Grover, Tyagi, Goyal, Jamal, & Singh et al., 2016).

Analysis of molecular dynamics simulations
In order to study the stability and dynamic behavior of the two hit compounds in the complex with Mtb TrxR, a molecular dynamics simulation study was performed for 20 ns.The conformational stability and convergence of TrxR and TrxR bound with hit compounds (TrxR comp1 , TrxR comp2 ) were evaluated by calculating the root mean square deviation (RMSD) of the backbone atoms.RMSD of the protein backbone against the time of the simulation is represented as trajectories (Figure 3A).These trajectories show that the system attained adequate stability after 15 ns of the molecular dynamics simulations, explained by the low variation of the TrxR protein backbone after forming a complex with the individual compound.Interestingly, while both compound complex forms showed similar RMSD trends, the apo form of TrxR tend to have a higher RMSD after 15 ns of the simulation.The higher RMSD is due to enhanced motion of one of the loops (region 47-72) from the FAD binding domain.This loop is observed to be stabilized by forming new hydrogen bond interactions with one of the a-helices of the NADPH domain in the compound1 complex form of TrxR.Individual amino acid residue of a protein play a significant role in providing stability to selective inhibitor binding to the protein.Hence, the position and the relative flexibility of each amino acid residue of TrxR were analyzed in the molecular dynamics simulations.The RMSF trajectories give information about each amino acid fluctuation in TrxR as shown in Figure 3B.A movement (>6 Å) can be observed in the loop region (residues 47-72) of the FAD domain of the apo-TrxR protein, which overlaps with the high RMSD.Binding of the compounds altered the RMSF of this region, suggesting that fluctuation of the loop region facilitates the compounds to bind.The binding induces conformational changes in both NADPH and FAD domains.The compounds binding residues in TrxR, such as A51, A124, A126, S144, C145, T147, C148, S166, E169, E170, and F173, are relatively stable and fluctuate little.In fact, no significant fluctuations of amino acid residues were observed during the entire simulation.As reported earlier (Akif et al., 2005;Waksman et al., 1994), the motion of NADPH domain with respect to the FAD domain is functionally important for the catalysis.The hinge region is less flexible but provides a pivotal point for functionally relevant motion of the NADPH domain.However, to test the effect of the compounds in the hinge region, we performed position restrained MDS, in which hinge region residues were position restrained.Simulation of   the restrained apo-TrxR for 20 ns provided an average RMSD of 0.1 nm whereas the average RMSD of the unrestrained TrxR was observed to be more than 0.3 nm.Further, similarity in the RMSD profile of the restrained apo-TrxR and the compound bound complexes suggests the rigidification of the hinge region upon binding of the compounds (Dantu et al., 2017).Binding of compounds with TrxR, mostly through hydrophobic interactions, caused conformational changes and formed a new set of interaction between the NADPH and the FAD domains.This suggests that the new interactions between the two domains would resist functionally relevant motion for the catalysis.In addition, compound1 was also observed to interact with the active site residues of TrxR.Solvent accessible surface area (SASA) is a parameter computed using the gmx_sasa module of GROMACS, which measures the proportion of protein surface exposed to the water solvent.The buried amino acid residues in the hydrophobic core of the protein are the driving force for protein folding.The relative value of SASA can predict the extent of conformational changes in the protein that have occurred during the course of binding (Marsh & Teichmann, 2011).The SASA value for an unbound protein was calculated to be 180 nm 2 whereas the compound-bound protein exhibited a slight decline in the solvent residual exposure.The average SASA value of the bound compounds 1 and 2 was calculated as 160 and 163 nm 2 , respectively (Figure 3C).This indicates that bound forms were least exposed to the water solvent during the 20 ns of the MDS, which could suggest that the binding of compounds changed the conformation of the protein and that it could provide an irreversible inhibitory effect by binding to the interactive surface residues near the protein hydrophobic core.

Evaluation of binding affinity with the protein molecule
The free energy calculations using MM/PBSA method provides an important parameter for estimating the binding affinities of ligands with a protein molecule (Wang, Hurley, & Merouesh, 2013).The MM/PBSA calculations yielded the binding-free energies of compounds 1 and 2 with TrxR, listed in Table 1.
The results indicate that compound1 possessed a higher negative binding free energy value of À 151.385 kJ/mol compared to the compound2, which has the value of À 121.613 kJ/mol.These binding energies suggest a significant potential for the formation of stable molecular interactions with amino acid residues of the TrxR.Interestingly, it was noted that Van der Waals, electrostatic interactions and non-polar solvation energy negatively contributed to the total interaction energy while only polar solvation energy positively contributing to the total free-binding energy.Overall, the high negative value of Van der Waals energy suggests massive hydrophobic interactions are dominant in the formation of a stable protein-ligand complex.

Analyses of protein conformational variation
The MD trajectories of the system were inspected with the principal components to get a better understanding in the conformational changes of Mtb TrxR in the complexes with two compounds.Correlated motion plot shows how atoms move relative to each other.Motions can be correlated (in the same direction), anti-correlated (in the opposite direction), or uncorrelated (Kasahara et al., 2014).Anti-correlated motions were observed to be significantly dominant in the complex forms (Figure 4).A summary of significant motions between two domains/region such as NADPH-FAD, NADPH-Hinge and FAD-Hinge is presented in Table 2.The magnitude (eigenvalue) and direction (eigenvector) of overall atomic motions in the apo-and complex bound forms of TrxR were evaluated using principal components (PC) (Peng & Zhang, 2014).The first few components can be interpreted as directions, which represent the maximum variance in the backbone atoms.The first two eigenvectors of the apo-TrxR and complex-bound form represent a significant number of conformational dynamics (Figure 5).The projections of the motion on the first two eigenvectors imply that the TrxR complex covers a small space and shows a reduced number of conformational motion along PC1 and  Weakly anti-correlated Weakly anti-correlated Weakly anti-correlated PC2 projections as compared to the apo-TrxR.In complexbound forms of TrxR, each point in phase space describes the specific conformation and reduced displacement in phase space highlights lower conformational sampling upon binding with the compounds.Additionally, the conformational sampling of both systems was inspected by tracing the covariance matrix for backbone atom positions.For apo-TrxR, the covariance trace value was observed to be 40.44 nm 2 and lower covariance trace values of 31.82 and 32.98 nm 2 were observed for both compound1 and com-pound2 complex bound forms respectively (Figure 5).Overall, PCA suggests that binding of both compounds with TrxR results in a significant change in the overall motion of TrxR with a compressed conformational space.The simulated structures of both complexes causing the conformational variations in the protein domains were also analyzed using the PCN method that revealed differences in contacts among the nodes (Piovesan et al., 2016).It indicated that the total number of contacts were slightly increased in the active site region as well as in the interface of NADPH and FAD domains in the complex structure as compared the apo-TrxR.While the number of contacts in the hinge region remains unchanged in both cases as compared to the apo-TrxR (Figure S1, supplementary material).It suggests that rearrangement of contacts between the interface domains contributes the conformational changes in the complexes.Moreover, it was also observed that the binding of compounds impacted the local interaction network between the NADPH and FAD interface domains and the number of linked nodes, links, and links mediated by hubs were slightly changed.Furthermore, the center of the mass distance between NADPH and FAD interface region of the compoundbound forms of TrxR was found to change significantly as compared to the apo-TrxR form.The center of the mass moved closer in the compound-bound forms (Figure S2, supplementary material.).This clearly suggests that the binding of the compounds to TrxR restrict the conformational flexibility.
Furthermore, the comparison of the structures of apo-TrxR with the compound-bound forms of TrxR revealed a significant shift of one of the a-helices from the NADPH domain of TrxR-compound1 toward FAD domain and was observed to stabilize by many new hydrogen bonds.The carbonyl oxygen of Gly139 of the NADPH domain forms a H-bond with the side chain of Thr54 of the FAD domain.In addition, Glu134, Glu135 of the NADPH domain formed H-bonds with Ser47, Gly49, 50 of the FAD domain, respectively (Figure 6).The formation of additional contacts in the complexes may restrict the conformational flexibility of the NADPH binding domain, which in turn may inhibit the ability of Mtb TrxR for transferring the electrons to the substrate thioredoxin.

Conclusion
Mtb TrxR has been validated as an essential gene of Mtb and also shown to be a druggable candidate.This study reports screening of natural compounds from the TCM database against Mtb TrxR.This led to the identification of 200 compounds and the top two compounds, compound1 and com-pound2, were selected.Both compounds bound to the hydrophobic groove of the Mtb TrxR and the binding was stable throughout the MDS.Finally, the data obtained in the study indicate that both these natural compounds may have the potential to bind with Mtb TrxR and can be used as lead molecules for the inhibition of Mtb TrxR.This will be reported later using in-vitro inhibition assay with the purified protein.

Figure 2 .
Figure 2. LIGPLOT and a 3-D representation of protein-ligand interactions of the screened compounds in the selected cavity of Mtb TrxR.(A) compound1, (B) compound2.Hydrophobic interactions are shown in brick red arch and hydrogen bonds are shown with green dashed lines.

Figure 3 .
Figure 3. (A) Backbone RMSD vs. time.(B) RMSF vs. residue number.Secondary structure arrangement of TrxR according to residues shown on the top of the RMSF plot.(C) SASA vs. time of apo-TrxR complex with compound1 and compound2 and the trajectories are represented in black, red, and green, respectively.

Figure 4 .
Figure 4. Covariance matrix plot of apo and complex form of Mtb TrxR during 20 ns MD simulation.(a) apo-TrxR.(b) complex with compound1.(c) complex with compound2.The positive and negative limits are shown.

Figure 5 .
Figure 5. PCA plot constructed by eigenvector 1 and eigenvector 2. The conformational sampling of apo state, compound1 and compound2 bound to the TrxR is represented in black, red, and green respectively.

Figure 6 .
Figure 6.Superimposition of structure of the apo-TrxR and compound1 complex form of TrxR and their respective domains are shown in different colors.The NADPH domain of complex and apo-TrxR is represented in dark gray and light gray, respectively.The FAD domain of complex and FAD domain is shown in dark green and light green, respectively.Doted circles show the orientation of the a-helix in both.The H-bonds formed between the residues from NADPH and FAD domains in the complex are shown in the blowup.

Table 1 .
Average MM/PBSA free energies of Mtb TrxR complexes with compounds, calculated from the MD simulation performed at 20 ns.

Table 2 .
Dominant motion of atoms in NADPH and FAD domains and the hinge region of the protein.