UNDERSTANDING PEPTIDE STRUCTURE-FUNCTION USING MODELLING, MD SIMULATION & AI

Abstract

The study of proteins and their conformation study began in the last century. Understanding the mechanism of action of enzymes plays a key role in deciphering the functional aspects of it. Thus, experimental methods such as X-ray Crystallography, NMR, or Cryo-EM were developed, and it has helped in determining the three-dimensional structure of proteins. Nearly, more than 1.7 lakh protein structures are available in the protein structural data bank. However, determining the structure of protein has taken a huge interest after increase in the computational power. Peptide is a subsequence or a substructure of protein having amino acids less than 50. Peptides are required for various metabolic activities and play an important role in fundamental physiological functions. Peptides have antibacterial, antifungal, antiviral, and anticancer effects that are beneficial to human health. Hence, exploring the substructure of protein, which is a peptide, is the rational approach. A similar approach is used in the model building of protein using threading approach. But there are unexplored structural parameters that need to be evaluated. Some of them are finding stretches of a sequence that are part of three-dimensional structures, identifying variation in the dihedral angles and others. Thus, the present study takes account of peptides as an indicator for structural orientations and builds a foundation for this futuristic approach. As this is an initial phase of such research, we have mapped the peptides to the protein structure database i.e., RCSB PDB. It was interesting to see how many peptides are embedded in the proteins. This data was used to analyse the dipeptide to decapeptide study. Surprisingly, after tetrapeptides, many peptides were observed that are not part of any structure. To facilitate this analysis, a Python based script was developed to calculate the peptide occurrence in PDB database. Further, the free form of peptides has different orientations than that of protein embedded part. To study peptides as a small molecule, their structure needs to be modelled and evaluated for dynamical information. Hence, to investigate it, the peptide’s three-dimensional structure was developed using an automated pipeline that uses existing peptide-building tools. An automated script was created that can run MD simulations on arbitrary number of peptide structures. It was put to the test by simulating 400 dipeptides. Furthermore, as previously indicated, peptides with a high frequency of length 2-10 were chosen as simulation targets. The simulation analysis was also carried out automatically. Peptides has also a great therapeutic value and it includes hormones, growth factors, neurotransmitters, ion channel ligands, and anti-infective medicines. Hence, peptides use in medicinal applications have piqued the interest of drug developers because they are thought to be highly selective, efficacious, safe, and tolerable, with appealing pharmacological characteristics. Such peptides can be obtained from natural sources and are named as food derived peptides. These bioactive peptides have been used for hypertension and the present study aims to study their role as an antihypertensive agent. For hypertension, Angiotensin converting enzyme-I (ACE-I) is a vital target used for development of medications. There have been studies that found peptides that inhibit ACE-I effectively are coming from food sources. We gathered all such peptides derived from various sources of food and made an annotated database for 1587 unique peptides. These peptides were then docked with target ACE-I by using in-house automation script of docking. Our database contains sequence, length, source, peptide preparation and isolation method, ACE inhibition Assay, in-vitro/in-vivo study, IC50 values, binding energies, ADMET profiling, references and three-dimensional structures of peptide. The interaction profiling of these peptides with ACE-I was carried out. It has resulted in the creation of an antihypertensive peptide database where experimentally validated peptides have been deposited. Food-derived peptides have potential drug value in hypertension therapeutic. Additionally, finding functional food requires identifying food proteins that have antihypertensive peptides. Hence, we developed an in-silico gastrointestinal tract to facilitate the digestion of food through various proteases. After the generation of peptides from food proteins, the tool also matches them against the previously developed database and identifies peptides having antihypertensive tendencies. For the remaining peptides, the antihypertensive peptide predictor has been developed. This whole package has been launched by the webserver to facilitate research and the industrial community. This webserver uses “FASTA” file or Uniprot ID as input for food protein and carried out each step, mentioned before, with user-friendly interface. Webserver URL: http://hazralab.iitr.ac.in/ahpp/index.php. Further, to investigate the dynamical behavior of peptides and compare it with small molecules, an Angiotensin Converting Enzyme Inhibitor synthetic compounds and peptides was carried out. For this, purpose hypertensive drugs were collected from the literature and their molecular docking was performed similarly to the peptides. The Molecular Dynamics of small molecule and peptide-bound complexes were carried out. It showed that the peptides behave as similarly to small molecules and bound well with the target enzyme ACE-I. In the final chapter of thesis, the study of keratin-derived peptides was carried to investigate the efficacy of keratinase obtained from a novel keratolytic bacteria. Since this bacterium strain can degrade feather keratin, the keratin degradation mechanism at bacterium's genomic level and structural investigations has not been investigated yet. Therefore, in this study, we sequenced, assembled, and analyzed this bacterium's genome to understand the keratin degradation mechanism with the help of in-silico structural approach. We assembled 5.27 Mb size of B.