In-silico identification of biomarkers and vaccine candidates for advancement of lung cancer therapeutics

Abstract

According to the World Health Organization report, around 10 million new cases are diagnosed with cancer in the year 2018 alone. Out of these, nearly 45% of the cancer incidences were reported from the Asian countries, 26% from European Union, 15% from the North American continent, 6% from the African countries, and 7% in the Latin American countries. These statistics highlight that cancer is a global problem, and alone is responsible for millions of premature deaths. Among the cancer types, lung cancer ranks first in terms of new incidences and mortality rates in all around the world population. It is highly heterogeneous, and despite the heterogeneity, several other factors such as infectious viruses, smoking, and drinking also correlate with the development of lung cancer. Globally it shares approximately 19% of all cancer-related deaths and 11.6% of all newly diagnosed cases. In terms of mortality rate, it ranks first in men and second in the case of females after breast cancer. The report suggests that the median survival time among the patients at an advanced stage of lung cancer is reduced to just 4.5 months, provided there is no treatment given. However, the addition of bevacizumab, along with other drugs, improved the life expectancy of the patients, but still, it is far from satisfactory. The use of all the targeted chemo-therapies suffers from several limitations - the occurrence of drug resistance, toxic nature of the drug, treatment failure, relapse among the patients, delayed wound healing, and many more. Thus oncologists and researchers all around the world are in continuous search of the alternative molecule that can advance and guides lung cancer therapeutics. After successful application in many cancer types, immunotherapy is considered as an alternate and most advanced strategy for the treatment of cancer. The FDA already approved several immunotherapeutic agents in the form of the oncolytic virus-based drug, interleukin, checkpoint blockade for the treatment of various cancer types. The World Health Organization report suggests that nearly 60% of the mortality rate among the patients can be prevented by improving diagnostic, screening, and therapeutic strategies. Among the therapeutics, vaccinations seem to be an effective measure to prevent new incidences of lung cancer caused by viruses. However, the identification and screening of a new class of immunotherapeutic molecules with experimental studies require a lot of time and resources. Thus the present thesis focuses on the development of computational tools that can find their way in aiding and guiding lung cancer therapeutics. The emphasis is given on the development of a web-resource providing up to date experimental information on oncolytic viruses used in cancer therapy; identification of subunit vaccine candidates against lung cancer-causing oncogenic viruses to be used in providing prophylactic immunity; prognostic biomarkers identification for the major subclasses of non-small-cell lung cancer that can serve the basis of precision therapy; and developing a machine learning-based prediction algorithm for the identification and designing of interleukin-2 inducing peptides. The developed web-resource on the oncolytic virus is “OvirusTdb” which is freely available to the scientific community at https://webs.iiitd.edu.in/raghava/ovirustdb/. It catalogues 5927 records against 25 fields, which were manually curated from the 166 and 27 research articles and patents, respectively. In addition to this, the web-resource holds extensive experimental information on 24 oncolytic virus species, 300 genetically modified oncolytic virus strains, 124 cancer types, 400 cancer cell lines, and 22 model organisms. The web-resource, which holds information about identified proteomic based subunit vaccine candidates against 09 oncogenic virus species, is “VLCvirus,” which is freely available to the scientific community at https://webs.iiitd.edu.in/raghava/vlcvirus/. The web-resource provides detailed information on 125 identified best antigenic epitopes having MHC class-I, II binding, B-cell, T-cell, and vaccine adjuvant acting potential. Moreover, the study also identified epitope sequences “VMFVSRVPV,” “LRRFMVALI,” that shows binding potential to nearly 15 MHC class-I and 49 class II molecules, respectively. In addition to this, the study also identifies 25 promiscuous epitopes that are present in multiple viral strains/species, with the majority of them related to E1 and E6 envelope genes. Further to capture the heterogeneity of lung cancer and its basis in advancing the therapeutics, non-small cell lung carcinoma subtype-specific biomarkers have been identified using the Univariate Cox regression and prognostic index-based models. The Univariate Cox analysis identifies 1334 and 2129 genes of some survival predicting potential in lung squamous cell carcinoma (LUSC) and lung adenocarcinoma (LUAD) datasets, respectively. Using random forest variable hunting technique and iterative search approach, we came up with a minimum number of the gene set that can be used as a subtype-specific prognostic biomarker; the study identifies 05 “(KIF16B, KLK7, LONRF3, OPLAH, RIPK3)” and 04 “(AHSG, DKK1, MGAT5B, NEMP2)” genes for LUSC and LUAD, respectively. Since literature evidence suggests that the mutant version of interleukin-2 has a high therapeutic index, we developed the machine learning-based prediction algorithm, which is capable of identifying the interleukin-2 inducing potential from a given sequence. The random forest-based hybrid model (dipeptide composition + length) achieved a maximum accuracy of 73.25, with an MCC of 0.46 and AUC of 0.73. The whole prediction algorithm is integrated into the form of a web-server, which is freely available to the scientific community at https://webs.iiitd.edu.in/raghava/il2pred/. Each of the developed web-resource and algorithms will have the potential to guide the therapeutics of lung cancer. For example, the data stored in the “OvirusTdb” have the potential to be utilized by genetic engineers and biotechnologists for the designing of new oncolytic viruses with the improved anti-cancer response. It can also be serving the basis for the design of experimental protocols for further enhancing the drug efficacy of the existing anti-cancer drugs. The identified 125 best antigenic epitopes can be utilized in clinics for providing immunity against lung cancer-causing viruses. The identified promiscuous epitopes can also be used in offering vaccination to a large human population due to their broad coverage of MHC alleles. Moreover, the identified promiscuous epitopes across the virus strain/species can be utilized in providing heterologous immunity against the concerned pathogenic/viral species. The identified gene biomarkers for non-small cell lung carcinoma have the potential to be investigated for therapeutic and diagnostic possibilities in the form of more subtype-specific interventions. The developed “IL2Pred” server would find its way in clinics for the identification and designing of a mutant version of interleukin-2 inducing peptides more economically as the experimental setup is time-consuming and require a lot of resources. Thus, we conclude that the identified epitopes, biomarker, and interleukin-2 inducing peptides from this study have the potential to be utilized in clinics for aiding and advancing lung cancer therapeutics.