The amino acid sequence (gi: 57116827) of Mtb-SHMT was obtained from National Center for Biotechnology Information (NCBI, http:// www.ncbi.nlm.nih.gov/). Homologous sequence identity was carrier out through the BLASTp server (₂₁), to search for short nearly exact matches and corrected the dataset to remove redundant sequences. The SHMT from Thermus thermophilus SHMT (tSHMT) (2DKJ) (₁₇), Geobacillus Steorothermophillus SHMT (gSHMT) (1KL2) (₁₈), B. Stearo thermophilus SHMT (bSHMT) (1YJS) (₁₉), Human mitochondrial serine SHMT (hSHMT) (2A7V) (₂₀) has sequence aligned with sequences of Mtb-SHMT using the ClustalW server (₂₁. The alignments from BLAST were manually refined by comparison with the alignment of Mtb-SHMT of other species SHMT. (Fig.1) The MODELLER 9v3 program uses the spatial constrains determined from the crystal structures, to build a 3-D model of target protein with unknown tertiary structures. 100 runs were carried out to obtain the most reasonable model (₂₂). Based on sequence alignment as explained in Figure1 and by taking default parameters one hundred models were generated and their quality assessed using the Modeller Objective Function parameter. The built Mtb-SHMT secondary structure characterization were carried out by submitting Mtb-SHMT model to PDBSUM online server (₂₃)

Energy minimization of the model is necessary in order to relieve short contacts and correct bad geometry that may be present in the model. The best initial model obtained from the homology modeling was solvated with solvent water molecule and was roughly energy-minimized in order to make it suitable for performing molecular dynamics (MD) simulation to relax the loop and side chain (₂₄). The MD stimulation of Mtb-SHMT was performed using GROMACS and in particular “ffG43a1” (GROMOS96) force field (₂₅). All bonds were constrained using LINCS (₂₆) First, 100 ps of molecular dynamics at 300 K in the water molecules inside the box to allow for the equilibration of the solvent around the protein residues. In this dynamics, all protein atoms had their positions restrained. Then, a full molecular dynamics simulation of 5000 ps at 300 K with no restrictions using 1 fs of integration time and a cut-off of 14 Å, for long-range interactions was carried out. Because these proteins have only a residual net charge, which was balanced by contra-ions for these simulations, a twin-range cutoff was used for long-range interactions: 1.5 nm for electrostatic interactions and 1.2 nm for vanderwaal interactions. The Shake algorithm was used to constrain hydrogen bond lengths (₂₇). The effects of the electrostatic potential truncation were minor at 14 Å, and the cut-off procedure made the calculations faster than other methods. Considering that the protein and water in the simulation box can achieve about 130,000 atoms, this method can save significant computational time.

Stereochemical quality and structure analysis (backbone ?and dihedral angle values) for the refined Mtb-SHMT model was done with the PROCHECK ()₂₈ WHAT IF (₂₉) and ProSA-Web (₃₀) programs, respectively, through the server: http://biotech.emblebi.ac.uk (Biotech for Validation Suite Protein Structures). The results were compared with the corresponding analysis of the crystallographic structures (1YJS, 2A7V, 2DKJ and 1KL2), whose coordinates were taken directly from the PDB.

After building the structure of Mtb-SHMT apoenzyme, the next step was to build the corresponding holoenzyme by docking PLG and FFO into the respective active sites. The atomic partial charges of PLG and FFO were added by online PRODRG server (₃₁). AutoDock 4.0 (₃₂) was used for the docking study of holoenzyme complex of Mtb-SHMT combined exploits the Lamarckian genetic algorithm.. The docking grid size was prepared with the autogrid utility of autodock setting to 60X60X60 points with a grid spacing of 0.370 A?. The grid center was placed in the active site pocket center. The grid boxes included the entire binding site of the enzyme and provided enough space for the ligand translational and rotational walk. . The consistencies of the maps were ascertained by checking the minimum and maximum values of the vanderwaal energies and electrostatic potentials for each calculated grid map. Flexible ligand docking was performed for all synthesized compounds. Docking was carried out using the empirical free energy function and the Lamarckian genetic algorithm, applying a standard protocol, the energy evaluations were 250,000, the maximum number of iterations 27,000 for an initial population of 150 randomly placed individuals with a mutation rate of 0.02, a crossover rate of 0.80, and an elitism value of 1. The number of docking runs was 100 and, after docking, the 100 solutions were clustered into groups with the RMS deviations lower than 0.5 A?. The clusters were ranked by the lowest-energy representative of each binding mode

In order to perform high frequency computational analysis such as molecular modeling, molecular dynamics and protein ligand interaction studies are performed using, a Hi-end server (Pentium IV 3.4 MHzs, AMD Athlon 64 bit, Dual processor with 1 GB video graphics card). Structural diagrams were prepared using the programs VMD (₃₃), and Other analyses were performed using scripts included with the Gromacs distribution.

Fig.1 explains final alignment, which was modeled as Mtb-SHMT in Modeller 9v3. This alignment was obtained after manual adjustments of the initial alignment from the BLAST server. Four reference proteins, hSHMT (2A7V), tSHMT (2DKJ), sSHMT (1YKJ) and rSHMT (1KL2) were used to model the structure of the Mtb-SHMT and homology scores comparing to target proteins were 60%, 57%, 57% and 45% respectively. High levels of sequence identity could guarantee more accurate alignment between the target sequence and template structure. In order to define structurally conserved regions (SCRs) of the protein family, the multi-dimensional alignment based on the structural identity was used to superimpose four reference structures, and 126 SCRs were determined (Fig.1).

Figure 1

Figure 1: Sequence alignment of Mtb-SHMT with the experimental structures gSHMT, bSHMT, tSHMT, hSHMT show the conserved region in stars (*) and deleted regions with dashes (-)

The Modeller program uses the spatial constraints, determined from the crystal structure of a template protein, to build a 3D model of the target protein with unknown tertiary structures, on the basis of amino acid sequence homology to the sequence alignment. The Ramachandran plots reveal more than 95% of the amino acid residues in the favorable regions of the plot for the whole enzyme. The main structural elements of the optimized Mtb-SHMT homology model are appearing in (Fig.2A) In secondary structure build model consists of three domains, the N-terminal domain, the second N-terminal domain and small domain, The N-terminal domain mediates inter subunit contacts and folds into two ?-helices and one ?-stand. The second N-terminal domain or large domain binds PLP, has most of the active site residues and folds into an ?-?-? sandwich containing nine confirmations, The confirmation clearly appears four beta sheets, three beta-alpha-beta motifs, four beta hairpins, one beta bulge, fifteen stands, twenty two helices, thirty five helix-helix interactions, nineteen beta turns, three gamma turns (Fig.2B).

Figure 2

Figure 2: A) Shows the refined model of Mtb-SHMT, it revels the secondary structure elements alpha helices, beta pleated sheets, loops. B) Secondary structure of Mtb-SHMT

The modeled Mtb-SHMT energy minimized using gromacs force fields. Later the energy minimization, the 3D Mtb-SHMT model and the decent identity crystallographic hSHMT holoenzymes were submitted to molecular dynamics simulations using the GROMOS 96 force field. This procedure had the goal of putting both enzymes in the same physiologic conditions for further superposition and also of providing a way for additional validation and refinement of the structure of Mtb-SHMT. In fact, an overall average structure conserved in time during a molecular dynamics simulation is essential to consider a model acceptable. After the dynamics simulations, Averages and fluctuations of several energies were calculated for the 5000 ps trajectory to examine stability of the Mtb-SHMT the total energy fluctuation, root mean square fluctuation (RMSF), was as small as 0.2%. The kinetic and potential energies showed fluctuations 3.5 and 0.49%, respectively (Fig.3A, B and C). The time evolution of the root mean square deviation (RMSD) of the MD trajectory from the hSHMT X-ray structure was computed to analyze a structure stability of Mtb-SHMT (not shown in Figure).

Figure 3

Figure 3: A) Variation in potential energy during the 5000 ps of MD on the Mtb-SHMT. The potential energy is averaged over 1000 ps interval. Variation in Total energy during 5000ps C) RMS fluctuation peaks in Mtb-SHMT side chain residues (D1, D2 and

The RMSD increased rapidly to 0.15 Å for the main and 0.31 Å for the side chain atoms, respectively, within the first 10 psec of the simulation (Fig.4)

Figure 4

Figure 4: Shows from the MD simulation of Mtb-SHMT, the time evolution of RMS deviation between the back bone, side chain at the 300 K initial temperature.

then, the RMSD values gradually decreased for the main and increased for side chain atoms. The total RMSD values were less than 0.31 Å indicating high structural stability of Mtb-SHMT. From the MD simulations the side chain of Lys227, His201 and Thr224 in Mtb-SHMT is oriented about at 3 to 4 Ao when it compared with the active site amino acids of hSHMT (Fig.5). This orientation make possible for efficient binding of FFO monoglutamate tail and PLP. On the other hand, the corresponding amino acid in the human enzyme, Asp146, is forced by the protein backbone to orient its carboxylate group in opposition to the monoglutamate tail of FFO, thus avoiding any possibility of interaction even if it were a positively charged side chain. Another significant difference between both enzymes is that Thr145 in hSHMT sets its side chain at about 4 to 5 Ao from FFO, with its OH group pointing towards FFO, while Asp198 in Mtb-S HMT have its carboxylate group in an opposite position, i.e., about 8 to 11 Ao from FFO. Finally, Ser193 in hSHMT and Thr224 in Mtb-SHMT set their side chains between PLP and FFO, and, in both cases, they establish hydrogen bonds with FFO, thus behaving practically in the same way. These significant orientation differences lead to development of new drugs for deadly Mycobacterium.

Figure 5

Figure 5: A) shows the active site amino acids of Mtb-SHMT (orange) and hSHMt (cyan) before refinement revels, structural orientation of active site amino acids of Mtb-SHMT after the refinement

In general, the evaluation parameters of the structural model obtained for Mtb-SHMT by WHATIF PROCHECK and ProsA-Web are within the interval of values derived for their homologs and for highly refined structures RMSD Z-score values for bonds and angle parameters for the model are within values typical of highly refined structures. The fact that the RMS Z-score values of bonding distances and angles for the crystal structures are small might indicate that too-strong constrains have been used in the original refinement of 2A7V, 2DKJ, 1YJS and 1KL2. From the analysis of backbone conformations, the Mtb-SHMT interface presents only two residues located in a generous region and the remaining interface residues are in the favorable region of the Ramachandran Plot. One indication that our model is a well-refined structure is the fact that its evaluation criterion of stereo chemical and structural parameters is better than those for 2DKJ, which is the worst of the 2A7V crystal structures. The low overall RMS values for backbone superposition reflect the high structural conservation of this complex through evolution, making it a good system for homology modeling. From ProSA-Web analysis of a Mtb-SHMT protein structure shows the energy graphs having negative values correspond to stable parts of the structure (Fig.6A, B). The ProSA-web analysis of high identical template structure 2DKJand Mtb-SHMT model showed the energy graphs nearly similar to each other and in correlation with the energy pattern between the X-ray structures. The output graphs showed the Z scores,–9.69 for 2DKJ and -9.12 for Mtb-SHMT model, from this data we can consider the structure of the Mtb-SHMT interface as a good representation of the actual system.

Figure 6: (A) ProSA-web Z-score of Mtb-SHMT (black spot) revels the good quality of the model when it compared with crystal structure 2DKJ amino acid sequence position showing negative values like crystal structures2DKJ. (B) Prosa energy plot shows local model quality by plotting energies as a function of amino acid sequence position; positive values correspond to problematic or erroneous parts of the input structure. Where as the negative values corresponds to good quality of the structure. Mtb-SHMT showing 98% amino acids in negative region similar as in 2DKJ.

The active site determination in our model was accomplished based on its alignment to the templates. There were found degrees of identity above 95% for the active site in all the cases when the high identity template model is submitted to the PDBsum server it deliver a catalytical residues or active site amino acids, taking that it compared to our model which exhibit much more similar identities at active site amino acids region (Lys227, His 126, Thr224, Asp198and His 201). At first hand, it seems that it would be very difficult to design selective inhibitor for Mtb-SHMT, however, if this is possible, we expect that those compounds would be more durable as efficient anti-tuberculosis drugs, once the close similarity between the SHMT active sites of the different species is an indication of the low potential for mutation of the enzyme. Furthermore, Fu et al., (₃₄) propose that the canonical SHMT structure supports the inference that the folate interacting site evolved from a type I PLP precursor enzyme through sequence insertions and not by domain swapping from other folate requiring enzymes. This may be the reason why antifolate compounds developed as chemotherapeutic agents are ineffective as inhibitors of SHMT and suggests that effective antifolate inhibitor of SHMT may not inhibit other folate enzymes. Thus, the challenge of developing selective inhibitor for SHMT would bring the additional advantage of achieving selective to the other enzyme of folate cycle.

The computer simulated automated docking studies were performed using the widely distributed molecular docking software, AutoDock 4.0. Energy minimized FFO and PLP from Dundee PRODRG2 server, were docked with Mtb-SHMT. The PLP and FFO specifically bind at active site amino acids Lys227, His 124, His127, Thr224, Asp198 and His201, give different docked energies AutoDock binding affinities of the PLP and FFO, showing binding affinity evaluated by

Table 1: Mtb-SHMT docking results with PLP and FFO

the binding free energies (?Gb, kcal/ mol), inhibition constants (Ki), hydrogen bonds, and RMSD values. The obtained success rates of AutoDock is highly excellent shown in (Table. 1), where the docked PLP binding free energies -28.43 kcal/mol and for FFO binding free enrgies -18.08 Kcal/mol (Fig.7) From the results it will revel that PLP give lowest docked energy when compared with FFO this analysis reveal that the PLP able to bind tightly with Mtb-SHMT then the FFO and showing greater binding energies, on other hand, experimental hSHMT also showing less similar interactions but due to the structural orientation in binding of hSHMT FFO and PLP is distinctive with Mtb-SHMT FFO and PLP holoenzyme complex. This unique feature may be help full to design a novel drug to against the Mycobacterium tuberculosis.

Fig.7 Active groove of Mtb-SHMT. FFO and PLP involve in binding of active site amino acids of Mtb-SHMT and the binding pocket shown in transparent solid surface with labeled amino acids (yellow) and the ligands is shown as a pink.