Identification of HOX signatures contributing to oral cancer phenotype

The role of evolutionarily conserved homeobox-containing HOX genes as transcriptional regulators in the developmental specification of organisms is well known. The contribution of HOX genes involvement in oral cancer phenotype has yet to be fully ascertained. TCGA-HNSC HTSeq-counts and clinical data were retrieved from the GDC portal for oral cavity neoplasms. GEO datasets (GSE72627, GSE30784, GSE37991) were accessed and analyzed using GEO2R. Differential HOX gene expression was profiled using the DESeq2 R package with a log2 fold change cut-off (- 1 and + 1) and Benjamini-Hochberg p-adjusted value at ≤ 0.01. Gene set over-representation analysis and semantic analysis associated with the disease ontology was performed using the ClusterProfiler R package, and pathway over-representation analysis was performed using IMPaLa. HOX protein interaction network was constructed using the Pathfind R package. HOX phenotype associations were performed using Mammalian Phenotype Ontology, Human Phenotype Ontology, PhenGenI associations, Jensen tissues, and OMIM entries. Drug connectivity mapping was carried out with Dr. Insight R package. HOXA2 was upregulated in oral dysplasia but silenced during tumor progression. Loss of HOXB2 expression was consistent in the potentially malignant oral lesions as well as in the primary tumor. HOXA7, HOXA10, HOXB7, HOXC6, HOXC10, HOXD10, and HOXD11 were consistently upregulated from premalignancy to malignancy and were notably associated with risk factors. Overrepresentation analysis suggested HOXA10 was involved in the transcriptional misregulation contributing to the oral cancer phenotype. HOX genes subnetwork analysis showed crucial interactions with cell cycle regulators, growth responsive elements, and proto-oncogenes. Phenotype associations specific to the oral region involving HOX genes provide intrinsic cues to tumor development. The 5' HOX genes were aberrantly upregulated during oral carcinogenesis reflecting their posterior prevalence.