br The Cancer Genome Atlas based
3.3. The Cancer Genome Atlas based Validation
3.3.1. Clinical characteristics of the oral cancer cohort
3.3.2. mRNA Clozapine-N-oxide and alteration statistics of markers
The selected set of 221 cancer stem cell markers were cross com-pared within the database (http://www.cbioportal.org) to assess the mutation, copy number and mRNA expression status in the treatment naïve, oral cancer cohort subset (n = 313) of the provisional head and neck cancer cohort (n = 530 samples). Assessment of the overall al-teration status indicated that some markers were altered in ≥15% of patients at the transcript level (Fas Associated Death Domain; 31%, N-Myc downstream regulated1; 21%, Tumor Protein p63; 19%, SNAIL family Repressor 2; 17%, Yes Associated Protein1; 17%, Epidermal growth factor receptor and HOXA1; 15%) (Supplementary Table-4).
In accordance with the objective of identifying markers of prog-nosis, the mRNA Z scores were compared between the patients who recurred and did not recur. In addition to the 221 gene cohort, this analysis was extended to the known cancer stem cell (CD44 and CD133) and pluripotent markers (OCT4 and SOX2) of oral cancer. A comparison of the z scores at the fold level changes (≥ 2 fold; up/down regulation) and statistical significance (p < 0.05) identified a subset of 54 genes that were fulfilled either criteria. This was considered the primary cancer stem cell-associated marker list for tumor recurrence in oral cancer (Table 3). Eight genes (NQO1, HOXA1, RIT1, FKBP4, UBE2C, EDNRB, STAT3, AURKA) (p < 0.05, fold change ≥2) fulfilled both the selection criteria and were identified as the priority list which was as-sociated with recurrence.
Fig. 1. Concordant genes of Agilent, Affymetrix and CSC database (CSCdb) and their Gene Ontology analysis-.
A. Venn diagrams indicating the gene panel comparisons; 809 common between Affymetrix and CSC database. B. 703 genes between Agilent and CSC database. C. Venn Diagram indicating the 364 concordant genes between the two analyses (809 genes of A and 703 genes of B) D. Gene ontology analysis identified protein-containing complex binding, signal receptor binding and enzyme binding as the first three categories in molecular function. E. Gene ontology analysis identified regulation of cell proliferation, locomotion, regulation of cell death as the first three categories in biological process.
Assessment of the interaction network of the markers identified that the major nodes were CDK1, STAT3, ERBB2, Albumin and PHF5 A. Interaction between Maternal Embryonic Leucine Zipper Kinase-CDK1-STAT3-ERBB2-Basigin, STAT3-Vascular Endothelial Growth Factor-Connective Tissue Growth Factor, Transcription Factor 4-NOTCH3-ERBB2 and Casein kinase- ALY/REF export factor-PHF5 A were ex-perimentally validated. Albumin, STAT3, Vascular Endothelial Growth
Factor and CDK1 interact with maximum number of proteins (Supplementary Figure 1). Inclusion of additional interactors from the database (n = 50) emphasized the clusters already identified as well as pointed out to new interactions. The CDK1 cluster was expanded in-cluding other molecules, GINS2, Cyclin B1and Cyclin B2. The second major cluster included Vascular Endothelial Growth Factor A, ERBB2, STAT3, Cadherin 1, Catenin β1, Catenin delta1, Kappa Delta Rho with Aly/REF Export factor (ALYREF)-PHF5 A-RIT1 forming a minor cluster
Pathways identified in Kyoto encyclopedia genes and genomes.
Category ID Name p- Hit Count in Query List Hit Count in Genome Percentage representation
Assessment of the interaction using an independent prediction server, GeneMANIA and subsequent visualization in Cytoscape, iden-tified 204 genes that were co-expressed, physical interactions existed between 23 molecules, co-localizations happened between 14 mole-cules and 99 interactions among the selected 54 genes. CDK1, Vascular Endothelial Growth Factor A, STAT3, ERBB2, AURKA, Cadherin1, Maternal Embryonic Leucine Zipper Kinase and UBE2C were the major nodes of interactions with more than 20 genes (Fig. 2B). A comparison with The Search Tool for the Retrieval of Interacting Genes/Proteins database indicated that GeneMANIA corroborated the two major clus-ters identified; CDK1, AURKA, Maternal Embryonic Leucine Zipper Kinase, GINS2, UBE2C and STAT3, ERBB2, Cadherin1, NOTCH3 and Vascular Endothelial Growth Factor A.
In The Cancer Genome Atlas, these 54 genes were evaluated for the co-occurrence and mutual exclusivity; RIT1-Thioredoxin Reductase 1, NQO1-Thioredoxin Reductase 1, Single Strand Selective Monofunctional Uracil DNA Glycolase 1 (SMUG1)-Basigin, NQO1-Ferrochelatase, AURKA-UBE2C, Myeloid leukemia factor 2-FKBP4, NQO1-RIT1, -Single Strand Selective Mono-functional Uracil DNA Glycolase 1-Myeloid leukemia factor 2, RIT1-Ferrochelatase showed co-occurrence (p < 0.05), the mutual exclusivity of these genes was not significant. NQO1 and RIT1 were co-expressed (Pearson’s co-relation: 0.59) (Supplementary Table-6).