Molecular dynamics simulation based prediction of T-cell epitopes for the production of effector molecules for liver cancer immunotherapy

Abstract

Liver cancer is the sixth most frequent malignancy and the fourth major cause of deaths worldwide. The current treatments are only effective in early stages of cancer. To overcome the therapeutic challenges and exploration of immunotherapeutic options, broad spectral therapeutic vaccines could have significant impact. Based on immunoinformatic and integrated machine learning tools, we predicted the potential therapeutic vaccine candidates of liver cancer. In this study, machine learning and MD simulation-based approach are effectively used to design T-cell epitopes that aid the immune system against liver cancer. Antigenicity, molecular weight, subcellular localization and expression site predictions were used to shortlist liver cancer associated proteins including AMBP, CFB, CDHR5, VTN, APOBR, AFP, SERPINA1 and APOE. We predicted CD8+ T-cell epitopes of these proteins containing LGEGATEAE, LLYIGKDRK, EDIGTEADV, QVDAAMAGR, HLEARKKSK, HLCIRHEMT, LKLSKAVHK, EQGRVRAAT and CD4+ T-cell epitopes of VLGEGATEA, WVTKQLNEI, VEEDTKVNS, FTRINCQGK, WGILGREEA, LQDGEKIMS, VKFNKPFVF, VRAATVGSL. We observed the substantial physicochemical properties of these epitopes with a significant binding affinity with MHC molecules. A polyvalent construct of these epitopes was designed using suitable linkers and adjuvant indicated significant binding energy (>-10.5 kcal/mol) with MHC class-I and II molecule. Based on in silico cloning, we found the considerable compatibility of this polyvalent construct with the E. coli expression system and the efficiency of its translation in host. The system-level and machine learning based cross validations showed the possible effect of these T-cell epitopes as potential vaccine candidates for the treatment of liver cancer.