Computational Design of Biologically Active Anticancer Peptides and Their Interactions with Heterogeneous POPC/POPS Lipid Membranes

Abstract

Over the last few decades, anticancer peptides (ACPs) have turned into potential warheads against cancer. Apart from small molecules and monoclonal antibodies, ACPs have been proven to be effective against cancer cells. ACPs are small cationic peptides that selectively bind to the negatively charged cancer cell membrane and kill them by various mechanisms. In the present study, we prepared a random scrambled library of 1200 peptides from the 100 known ACPs and virtually screened them for their anticancer properties. From in silico-predicted ACPs, 27 peptides were prioritized based on their support vector machine (SVM) score. Based on the SVM score and properties such as hydrophobicity, size, overall net charge, secondary structure, and synthetic feasibility, finally, four peptides were synthesized and screened for their biological activities. Cancer cell membrane-deforming potential of two most active peptides, peptide1 and peptide2 was assessed with molecular dynamics simulation. We found that peptide1 remains adsorbed to the membrane surface, while peptide2 has membrane penetration capability. The present study will be helpful in the computational design of ACPs and understanding their interaction with the cancerous cell’s membrane.

Automatic construction of molecular similarity networks for visual graph mining in chemical space of bioactive peptides: an unsupervised learning approach

“…According to Ref. 50 , these two experimentally verified peptides were the most active peptides, and molecular dynamics simulations suggest that they have different interaction mechanisms with heterogenous POPC/POPS lipid bilayer membranes. Whereas the Peptide1 remains adsorbed to the membrane surface, the Peptide2 has membrane penetration capability 50 .…”

“…It can be observed that Peptide3, which is an active compound too, is lying on a similarity path between Peptide1 and Peptide2. Indeed, Peptide3 is more similar to Peptide1 by the length of their similarity path, and less similar to Peptide2, which is expected since Peptide3 and Peptide1 are analogs of the same peptide 50 . However, despite Peptide2 and Peptide4 are having a common origin, Peptide4 is inactive 50 , and it is not connected to the previous ones in the K-NNG.…”

“…Indeed, Peptide3 is more similar to Peptide1 by the length of their similarity path, and less similar to Peptide2, which is expected since Peptide3 and Peptide1 are analogs of the same peptide 50 . However, despite Peptide2 and Peptide4 are having a common origin, Peptide4 is inactive 50 , and it is not connected to the previous ones in the K-NNG. Therefore, this retrospective analysis points out that the approach presented here may be useful for getting insight into the chemical space of BPs.…”

“…Then, each projected peptide is connected to its k nearest neighbors in the defined metric space for further analysis. To illustrate this process, we have embedded four in-silico designed anticancer peptides 50 in the molecular similarity network under study: Peptide1 (WLFKFLAWKKK), Peptide2 (FPKLLLKFLRLG), Peptide3 (KKFALKLFWWK), and Peptide4 (RLLRRLRIRG). www.nature.com/scientificreports/ As can be seen in Fig.…”

“…Visualizing the (a) subnetwork of anticancer peptides having central but non-redundant structures at 70% of sequence identity, and (b) the projection into this network of four in-silico designed anticancer peptides (Peptide1-4)50 . This figure has been created using the software starPep toolbox (version 0.8), available at http://mobio sd-hub.com/starp ep/, and PowerPoint 2016 available at https ://www.micro soft.com/es-mx/softw are-downl oad/offic e. https://doi.org/10.1038/s41598-020-75029-1…”