The elucidation of the X-ray crystal structure of TMPKmt was the major initiation factor for the numerous strcucture based drug design investigations (22). The structure of TMPKmt bound with its natural substrate TMP determined at a resolution of 1.95 Å was the first of its kind NMPK structure to be reported in a pathogen. TMPK is a homodimer in solution (monomer molecular mass=24 kDa; 214 residues per monomer), each monomer being composed of nine α-helices surrounding a five-stranded β-sheet core. The TMPKmt structure has three functionally essential regions for its catalytic activity.The P-loop region acts as the controlling centre for the phosphoryl group of the phosphate donor and also features the second essential region, a conserved arginine residue. The third important region is the LID region which functions in a flexible manner to close on the phosphoryl donor upon binding. The presence of arginine residues in both the P-loop and the LID region of TMPKmt is something unlike the other earlier reported TMPK structures. Further, the presence of a Magnesium ion in the active site and the characteristic helical conformation of the LID region are the other features that make the TMPKmt strcuture unique.

Also, noteworthy is that the crystal structure revealed the point of contacts of the natural substrate TMP and the important amino acid residues involved in catalytic activity. Even before the structure determination, these unique characteristics of TMPKmt was predicted by Labesse G and colleagues through their proposed structural model (23). Further, TMPKmt was found to be a stable protein with a melting temperature of 68 °C and similarly resistant to thermal denaturation like other Mtb proteins. The high thermal resistance of TMPKmt is due to its conformational stability as evidenced by its high total standard Gibb’s free energy for dissociation upon urea induced denaturation.

Chemical synthesis of TMPKmt inhibitors were mainly oriented towards structure guided generation of lead compounds due to the availability of its good quality three dimensional structure. Synthesis of nucleotide/nucleoside analogues were the most sought after inhibitors for TMPKmt owing to the simple methods of synthesis and also due to the confidence offered by the success of azidothymidine as an antiviral agent. On the other hand only very few studies explored non-nucleotide/nucleoside based inhibitors against TMPKmt.

Computer aided drug design (CADD) has revolutionized drug discovery pipelines in recent times by playing a key role in accelerating the processes of hit identification and lead optimization. It holds true for antituberculsis drug discovery as well. The two main methods of CADD are structure based drug design (SBDD) and ligand based drug design (LBDD). SBDD also referred to as direct drug designing approach relies on the 3-dimensional structural information of the protein target, in order to identify target specific active compounds. On the other hand LBDD, also referred to as indirect method of drug designing is very useful for protein targets with unknown structural knowledge. In LBDD, new potential hits are designed based on the existing set of ligands known to bind with the target. In the case of TMPKmt, the availability of good quality 3D structure and the knowledge of target specific active ligand sets have led to several studies that successfully integrated SBDD and LBDD approaches.

Manallack and colleagues performed the first CADD study on TMPKmt in 2002, shortly after its structure elucidation (22). In this study compounds showing potential binding affinity towards TMPKmt crystal structure were screened from The Available Chemicals Directory database. Though this study did not result in significant findings, they were able to confirm the hydrogen bond forming ability of Asp100 and Tyr39 of TMPKmt. As years progressed, the advent of chemical synthesis projects of TMPKmt inhibitors and availability of different TMPKmt ligand complex crystal structures provided opportunities for several CADD investigations of TMPKmt inhibitors.

Quantitative structure activity relationship (QSAR) is the major LBDD method that describes a mathematical relatonship between the structural attributes of ligands and their respective biological responses on the target. Several studies have been using this method to discover novel drug candidates against well established Mtb targets (46). Gopalakrishnan and colleagues carried two CADD investigation studies on TMPKmt. In the first series of work published in 2005, the group of Gopalakrishnan and Desiraju derived a pharmacophore model from a list of 47 compounds that were earlier reported in literature as thymidine monophosphate kinase inhibitors (47). The pharmacophore model generated with HypoGen was used as a 3D structural query to virtually screen a combined database of 500,000 molecules, which were derived from several publicly available databases like IBS, MDPI etc., Later using three subsequent filtering schemes, a total of 186 compounds were obtained as hits spanning five different chemical clusters. In their second study in 2006, they employed molecular field analysis (MFA), a type of 3D-QSAR method, which involves the alignment of molecular interaction fields of ligands inorder to extract the aligned features that might be related to biological activity (48). Least square alignment, pharmacophore based alignment and receptor based alignment were the three alignment protocols utilized to develop a 3D-QSAR model using a combined training/test set of 47 compounds as defined in the earlier study. The structures of all compounds were built and energy minimized using the coordinates of the PDB structure, 1G3U. Of the three alignment protocls, the 3D-QSAR model obtained from receptor based alignment was found to be superior in predicting as well as mapping the active site residues involved in ligand binding. These two studies provided early information on the structural and chemical features of TMPKmt inhibitors which were further explored using a variety of CADD methods by different research groups.

Kumar A and colleagues screened compounds available in the Maybridge small molecule database for 3D pharmacophore query in a virtual screening protocol (49). The pharmacophore model was derived from a dataset of 110 compounds collected from literature and defined using the various PDB structures of TMPKmt such as 1N5L (50), 1W2G (25), 1W2H (25) and 1MRS (24). The screening protocol also included molecular docking and receptor based weighting using Structure Interaction Fingerprints to yield 8 hit compounds, which were further evaluated for their in vitro antitubercular activity in Mtb H37Rv. This was the first CADD study on TMPKmt to successfully identify 5 compounds that exhibited antitubercular activity with MIC in the range of 3.12 µg/mL to 12.5 µg/mL.

Andrade.C.H and her colleagues adopted a receptor-independent (RI) 4D-QSAR method to derive a 3D-pharmacophore for a set of 34 5-thiourea-substituted α-thymidine inhibitors (51). The most potent inhibitors obtained from the statistically significant RI-4D QSAR models were further validated by docking to the active site of TMPKmt crystal structure (1G3U). This study led to the identification of new regions of TMPKmt inhibitors such as the sugar-pyrimidine ring structure and the 5’-arylthiourea moiety with good pharmacophore sites. In 2010, they performed a similar RI 4D QSAR investigation and derived a 3D pharmacophore for a larger set of 81 thymidine analogues (52). Thus, using 4D-QSAR methods, which is an extension of 3D-QSAR, these two studies explored an ensemble of different sampling features of the chosen TMPKmt inhibitors. The main drawback of 4D-QSAR analyses is the absence of large and chemically diverse datasets. Keeping this in mind, in their next study Andrade.C.H’s group explored a larger dataset of 97 thymidine based analogues utilizing a fragment based QSAR to generate 2D-QSAR models (53). Specifically they employed hologram QSAR (HQSAR) to identify unique molecular finger print patterns to further support the Quantum Mechanics/Molecular Modeling (QM/MM) docking experiment with the crystal structure of TMPKmt (22). Through this study, the authors were able to modify the existing inhibitors with better drug like properties.

In 2010, a combinatorial screening technique was employed to identify more potent analogues of bicyclic thymidine derivatives as inhibitors of TMPKmt (54). The execution of TMPKmt structure-1MRS (24) based focused screening and with the development of a QSAR model, resulted in a subset of bicyclic thymidine analogues with ADME properties and inhibitory potential that were predicted to be favorable.

Trupti and group developed 2D and 3D QSAR models from the molecular modeling studies on 28 substituted benzyl pyrimidine derivatives that were earlier reported by Gasse et al (55). Later pharmacophore optimization was achieved using the information derived from statistically significant SAR models, which further led to the design of new chemical entities (NCE) using the Combi Lib tool. Further molecular docking studies were performed on the most potent NCEs using the GLIDE tool, which revealed their superiority in docking TMPKmt over AZTMP.

Amidst several studies focusing on SAR studies Zaheer Ul Haq and his team optimized a structure based virtual screening protocol for TMPkmt inhibitors (56). To test the protocol they used a collection of 10,000 randomly selected molecules from the National Cancer Institute database along with 105 known substrates and inhibitors of TMPKmt. All the molecules were subjected to rigid docking algorithm using Fast Rigid Exhaustive Docking (FRED) program and scored using various methods such as enrichment curves, ROC curves and consensus scoring.

In 2013, TMPKmt (1GSI) bound conformations of 15 thymidine inhibitors with known activity and as part of a training set were built and later used to develop 3D QSAR pharmacophore model (57). This model was further validated with a PH4 3D QSAR to screen for TMPKmt inhibitors with potencies in the sub-nanomolar range among a library consisting of literature reported thymidine analogue inhibitors of TMPKmt.

In a recent study, Koseki.Y and colleagues performed an in silico pharmacophore screening of 461,883 moelcules in the ChemBridge database (58). After the three step pharmacophore screening and validation with molecular docking of TMPKmt structure, one compound (KTP3) was identified with significant growth inhibitory effect on Mycobacterium smegmatis, a surrogate model organism of Mtb. A further screening was carried out to identify KTP3 analogues and it resulted in two compounds namely KTPS1 and KTPS2 with half maximal inhibitory concentration values of 8.04 μM and 17.1 μM, respectively. However, since these two molecules exhibited only 18 % and 36 % inhibition of TMPKmt activity, they were suspected to have off target effects in M.smegmatis.

While all studies discussed so far designed small molecule inhibitors against TMPKmt, Kumar M and his colleagues have adopted a rational designing approach to identify small peptide inhibitors for TMPKmt (59). Using in silico structure based approach such as molecular docking and molecular dynamic simulations they had designed a tripeptide Trp-Pro-Asp that had high selectivity towards TMPKmt and also with favourable ADMET properties.