Thrombopoietin receptor binds to TPO from its ECD, so the amino acid sequence of ECD region was retrieved from Uniprot database and searched for 3D structure in protein data bank. In the absence of crystal structure for mammalian TPOR, the 3D structure was generated. Amino acid sequence of TPOR was searched for the identification of suitable template through Basic Local Alignment Search Tool (BLAST) (Camacho et al., 2008).¹ Due to unavailability of a template with higher query coverage and with good identity, 3D structure was generated through de novo method using Iterative Threading ASSEmbly Refinement (I-TASSER).² I-TASSER is web-based protein 3D structure prediction tool based on threading approach. It simulates the generated structure and defines its C-score, TM-score, and active sites. C-score refers to a confidence level of predicted structure of a given protein sequences where a high C-score (range from −5 to 2) indicates an absolute precise quality of the predicted structure. TM-score is structural assessment parameter in which, a smaller distance between the structures is weighted high. TM-score defines the topology of the structure, and it shows that a score more than 0.5 ensures a model of absolute topology (Yang et al., 2015).

Generated TPOR 3D structure was further evaluated by Swiss model workspace.³ Swiss server analyzes the various properties viz. solvation energy, torsion angle energy, solvent accessibility, and atom pairwise energy. Estimate of nativeness of these structural properties of the model with comparable features with high-resolution X-ray 3D protein structures is defined by the QMEAN score. Overall quality of the 3D structure was defined by the QMEAN Z-score (Benkert et al., 2008). Furthermore, TPOR structure was evaluated for its stereochemical properties through Ramachandran plot. Rampage server⁴ was used to determine the dihedral angles [phi (Φ) and psi (ψ)], and the number of residues lying in favored, allowed, and outlier regions of the protein structure. It also defines the quality of the structure, planarity of the peptide bond, the main chain hydrogen bond energy; Cα chiralities and non-bonded interactions of structure (Lovell et al., 2003).

A library was formed on the basis of the interaction sites of TPO-TPOR, known mimetic TMP-TPOR, and by designing analogs of known TPO mimetics through the literature survey. Library was in silico matured by changing amino acids at different positions with hydrophobic amino acids and forming various combinations of predicted mimetics. Hydrophobic amino acids were added to the mimetics since they are known to play an important role in TPOR binding and its activation.

Thrombopoietin receptor 3D structure derived through de novo method was prepared for molecular docking analysis using the protein preparation wizard (PrepWiz) in the Schrödinger software program (Friesner et al., 2006). Water molecules, unnecessary ligand, unwanted bound ligand, and duplication of any chain were removed, and hydrogen atoms were added to the TPOR protein structure, and then structure is selected for energy minimization and followed by optimization for molecular docking analysis.

The 3D structures of mimetics in the library were predicted by Pepfold3, a peptide structure prediction server.⁵ Pepfold3 is a new 3D structure generation engine. It predicts the structure of peptides based on the Hidden Markov model conformation sampling approach (Shen et al., 2014). All mimetics were imported into the workspace of Pepfold3, and the project name was saved and allowed to run. Predicted 3D structures were saved in PDB format.

Interaction analyses of mimetics with TPOR were done using ClusPro protein–protein molecular docking program. ClusPro server is a rigid docking program based on fast Fourier transformation to generate low energy interaction conformations of a protein–protein complex using the pairwise docking potentials. ClusPro clusters the interaction complexes with low energy and analyzes with semi-definite programming based underestimation program, which identifies the stability of the interaction clusters using the medium-range optimization algorithm. Stable interaction complex is refined using Monte–Carlo simulations (Kozakov et al., 2010). Resulting top docking score mimetic was selected, and its docked complex was analyzed to determine the molecular interactions using the Dimplot program of Ligplot+ (v.1.4.5) module. Dimplot determines the involved molecular interactions including the hydrogen bonds and hydrophobic interactions, between receptor and ligand interacting complex (Wallace et al., 1996).

Dynamics and stability of the complex between TPOR and shortlisted mimetic were evaluated through the molecular dynamics simulations using MDWeb program. MDWeb is a workspace, which provides standard protocols for molecular dynamics simulations and to scrutinize the interactions trajectories (Hospital et al., 2012). The optimized structure of the complex of TPOR and the mimetic shortlisted from designed mimetic library on the basis of molecular docking was used as input data for molecular simulations. NAMD full MD setup was used for carrying out the simulation. The whole process of simulation includes cleaning of structure, fixing of side chains, addition of hydrogen atoms, neutralization, addition of solvent box, and minimization and equilibration of the system to finally achieve the structure prepared by simulation. The equilibration of the system includes heating the solvent to 300 K and reducing the restraints to just protein backbone. These steps are carried out to provide the structure with the necessary conditions to study its dynamics along a span of time. Once the structure is prepared, the water molecules in it are removed to reduce the size of the system to achieve the dry trajectory for plotting various graphs for the analysis of the simulated complex structure. The structural stability, flexibility, and deviations per residues were assessed by the RMSD value.

Lead mimetic was analyzed for its physicochemical properties and compared with the TMP. Physicochemical properties of mimetic-9 and TMP including molecular formula, molecular weight, theoretical isoelectric point (pI), the number of negative and positive residues, extinction coefficient, aliphatic index, instability index, and grand average hydropathy (GRAVY) were determined using the Protparam server (Gasteiger et al., 2005).⁶

Molecular weight, extinction coefficient, net charge, and other properties are required to identify the chemical nature of the peptide and evaluate the binding efficiency of mimetic to a specific receptor. Binding efficiency of mimetic is assessed by potency efficiency index, which is defined by percent inhibition at a given compound concentration divided by its molecular weight. The extinction coefficient illustrates absorbance of light by a protein at a certain wavelength and is used to quantify the formation and concentration of peptides during in vitro analyses. Instability index value denotes whether the peptide would be stable in vitro or not. The instability index estimates the stability of peptides from statistical analysis of experimentally defined stable and unstable proteins. Aliphatic index predicts the total volume occupied by aliphatic side chains and indicates the thermostability of protein. Another important parameter that implies the stability of the protein is GRAVY value. It determines hydrophobicity and hydrophilicity of a protein by adding the hydropathy values of each amino acid residue and dividing by the number of amino acids in the protein sequence. A positive value of GRAVY indicates higher hydrophobicity of proteins. Furthermore, mimetic subcellular localization was determined by using the Cello prediction server v.2.5.⁷ Cello server predicts the localization in humans, based on the multiple n-peptide compositions through support vector machine (Yu et al., 2006).

Thrombopoietin receptor (human origin) extracellular region amino acids sequence of size 466 amino acids was retrieved from Uniprot database (Uniprot accession number Uniprotkb-P40238). TPOR 3D structure was predicted using I-TASSER. It generated five models of TPOR, which were analyzed for their properties. Among these models, the model-1 was shortlisted with high C-score (−1.03), TM-score (0.58 ± 0.14), and RMSD value of 9.5 ± 4.6 Å. C and TM-score indicated the TPOR structure to be of good quality with high confidence, and RMSD value showed it has optimal stability and flexibility.

Furthermore, TPOR structure was evaluated for its stereochemical and physiochemical properties. Its stereochemical properties were analyzed using Rampage. It showed 60.1% residues of the TPOR structure were in favored region, and 20.9% residues were in the allowed region, and 19% residues were in outlier region. Physiochemical properties were determined by Swiss server and found QMEAN score is 0.362, representing the overall absolute quality of the TPOR structure for the various properties (solvation energy, torsion angle energy, solvent accessibility, and atom pair wise energy). Values for these properties lie in the light red to red region of QMEAN score plot of TPOR structure and indicate good values of physiochemical properties (Figure 1).

The mimetic library was designed on the basis of binding sites of TPO to TPOR, TMP binding sites to TPOR, analogs of reported mimetics (TMP and TMP dimer), and by introducing the hydrophobic residues at specific positions of mimetics to enhance the binding affinity toward the TPOR. Library was designed for peptides only, and it consisted of more than 40 peptides of all similar structures coordinates with standard amino acids only. Some of the mimetics from designed library with their binding docking score have been shown in Table 1.

For library designing and comparative analyses, TMP was docked with TPOR. TPOR receptor was uploaded, and its unstructured terminal residues were removed, and then TMP was uploaded as a ligand. ClusPro runs PIPER rigid body docking program and rotates the mimetic ligand with 70,000 rotations at each grid point in x, y, z-axis about the TPOR receptor 3D grid with spacing 1.0 Å. Then using the clustering techniques to find near-native conformations along with eliminating the non-native clusters, the 1000 best energy conformations with the lowest score were clustered and among them, 30 largest clusters were refined by minimizing the Charmm energy of the complexes. The clustering of the poses starts with the lowest energy pose and grouping all poses within 9 Å.

ClusPro resulted in a binding score (lowest energy) of −798.4 kcal/mol and center binding energy score of −703.4 kcal/mol for TMP. TMP-TPOR molecular interactions were also identified using the Dimplot and it showed; TMP interacts with TPOR with a total of fourteen hydrophobic interactions and forms seven hydrogen bonds. TMP binds to TPOR at three regions: first region (147–174) involving residues Gly147, Pro148, Pro151, Lys152, Thr155, Gly156, Pro157, Pro170, Arg174, second region (220–277) involving residues Pro223, Tyr227, Gln265, Phe267, and third region involving Phe324 of TPOR. Hydrogen bonds were found in between Leu6 of TMP and Gln173 of TPOR with 2.87 Å bond length, Arg13 of TMP and Gly220 of TPOR with 2.89 Å bond length, Ala14 of TMP and Leu221 of TPOR with 2.80 Å bond length, Ala14 of TMP and Gln222 of TPOR with 2.91 Å bond length, Gln8 of TMP and Gly263 of TPOR with 2.75 Å bond length, Gln8 of TMP and Leu264 of TPOR with 3.08 Å bond length and Pro4 of TMP and Gln277 of TPOR with 3.31 Å bond length. Furthermore, TMP dimer peptide was also docked with TPOR, and it resulted in a lowest binding score of −811.9 kcal/mol and center binding score of −685.3 kcal/mol.

After that, mimetic library was docked with TPOR using ClusPro, and mimetic-9 was found to have high-rank conformation with a weighted lowest energy score of −938.8 kcal/mol and center binding score of −792.5 kcal/mol (Figure 2). Molecular interactions involved in mimetic-9 binding to TPOR was determined which showed, mimetic-9 also binds to three regions on TPOR with more residues (18) involved in hydrophobic interactions and seven hydrogen bonds. Mimetic-9 binding regions at TPOR are: first region (146–175) with involved residues Gly147, Pro148, Pro151, Thr155, Pro157, Cys169, Pro170, Arg174, Pro175, second region (220–277) with involved residues Gly220, Gln222, Tyr227, Gly263, Phe267, Gln277, and third region (324–328) with involved residues Phe324, Arg326, and His328 (Figure 3). Hydrogen bonds were found between Leu8 of mimetic and Gln173 of TPOR with 2.84 Å bond length, Arg11 (amino position NH2) of mimetic with Lys152 (oxygen atom) of TPOR with 2.73 Å bond length, Arg11 (amine position NH1) of mimetic with Lys152 (oxygen atom) of TPOR with 2.66 Å bond length, Tyr15 of mimetic with Tyr146 of TPOR with 2.78 Å bond length, Gln13 (OE1 position) of mimetic with Gln265 (NE2 position) of TPOR with 2.90 Å bond length, Gln13 (NE2) of mimetic with Gln265 (CD position) of TPOR with 2.73 Å bond length, and Glu2 of mimetic with Thr275 of TPOR with 2.90 Å bond length. Moreover, Ala12, Leu8, Leu4, Trp14, Ala9, Leu7, Ala10, Phe5, Val6, and Leu3 residues of mimetic-9 were found to be involved in hydrophobic interactions with TPOR. Molecular docking studies clearly indicated that mimetic-9 binds more potently with TPOR than TMP and TMP dimer. And molecular interactions analyses showed that mimetic-9 binds at similar region to TPOR as TMP binds but mimetic-9 binds with more strong hydrophobic interactions and with more number of amino acid residues of TPOR, indicates the possible reason of potent binding of mimetic-9 with TPOR than TMP. Docked complex of TPOR-mimetic-9 was further analyzed with molecular dynamics simulations study and for physicochemical properties.

Dynamic nature of TPOR–mimetic-9 interactions was studied through the molecular dynamics simulation for 10 ns. During the simulation, the frames of structural deviations were captured of the complex. TPOR–mimetic-9 complex ran for three frames and showed the stable conformations of interactions. Next, RMSD value was calculated and found to be 0.091 Å, which suggests the less deviation per residue of the interaction complex and stability till the end of the simulation (Figure 4).

Lead mimetic-9 was assessed for the physicochemical properties and compared with TMP properties (Table 2). TMP is of size 14 amino acids and its molecular weight is 1581.8 g/mol, and extinction coefficient value is 5500 M⁻¹cm⁻¹. TMP was found to be of basic nature with pI 9.60, and an aliphatic index value of 105 and GRAVY score of −0.150 showed that it has a comparable region of aliphatic nature and hydrophilic contents. Moreover its instability index value of 77.0, suggests it to be an unstable protein.

Lead mimetic-9 is of size 16 amino acids and has molecular weight 1847.2 g/mol and an pI 8.75, which showed that it has a small molecular weight and basic nature. Mimetic-9 possesses the higher extinction coefficient value of 6990 M⁻¹cm⁻¹ indicating the presence of Tyr, Trp, and Cys in abundance. Extinction coefficient value illustrates the absorbance of light by mimetic and help in quantifying the formation and concentration of peptides. The aliphatic index value of 183.12 showed volume occupied by aliphatic side chains and GRAVY value of 1.581 suggests more volume of the mimetic-9 structure is hydrophobic. These properties aid to the isolation, physical separation, and other quantitative protein interactions studies of the mimetic. Importantly, mimetic-9 was found to be stable with instability value of 9.38. Mimetic-9 was also assessed for its subcellular localization in the eukaryotes in the physiological environment using Cello prediction server. Cello prediction server showed mimetic-9 localization in mitochondrial, nuclear organelle, and into the extracellular region with a high-reliability score.