HMGA2 Antisense Long Non-coding RNAs as New Players in the Regulation of HMGA2 Expression and Pancreatic Cancer Promotion.

Background: Natural antisense long non-coding RNAs (lncRNAs) are regulatory RNAs transcribed from the opposite strand of either protein coding or non-coding genes, able to modulate their own sense gene expression. Hence, their dysregulation can lead to pathologic processes. Cancer is a complex class of diseases determined by the aberrant expression of a variety of factors, among them, the oncofetal chromatin architectural proteins High Mobility Group A (HMGA) modulate several cancer hallmarks. Thus, we decided to investigate the presence of natural antisense lncRNAs in HMGA1 and HMGA2 loci, and their possible involvement in gene expression regulation. Methods: We used FANTOM5 data resources, FANTOM-CAT genome browser and Zenbu visualization tool, which employ 1,829 human CAGE and RNA-sequencing libraries, to determine expression, ontology enrichment, and dynamic regulation of natural antisense lncRNAs in HMGA1 and HMGA2 loci. We then performed qRT-PCR in different cancer cell lines to validate the existence of HMGA2-AS1 transcripts. We depleted HMGA2-AS1 transcripts with siRNAs and investigated HMGA2 expression by qRT-PCR and western blot analyses. Moreover, we evaluated cell viability and migration by MTS and transwell assays, and EMT markers by qRT-PCR and immunofluorescence. Furthermore, we used bioinformatics approaches to evaluate HMGA2 and HMGA2-AS1 correlation and overall survival in tumor patients. Results: We found the presence of a promoter-associated lncRNA (CATG00000088127.1) in the HMGA1 gene and three antisense genes (RPSAP52, HMGA2-AS1, and RP11-366L20.3) in the HMGA2 gene. We studied the uncharacterized HMGA2-AS1 transcripts, validating their existence in cancer cell lines and observing a positive correlation between HMGA2 and HMGA2-AS1 expression in a cancer-derived patient dataset. We showed that HMGA2-AS1 transcripts positively modulate HMGA2 expression and migration properties of PANC1 cells through HMGA2. In addition, Kaplan-Meier analysis showed that high level of HMGA2-AS1 is a negative prognostic factor in pancreatic cancer patients. Conclusions: Our results describe novel antisense lncRNAs associated with HMGA1 and HMGA2 genes. In particular, we demonstrate that HMGA2-AS1 is involved in the regulation of its own sense gene expression, mediating tumorigenesis. Thus, we highlight a new layer of complexity in the regulation of HMGA2 expression, providing new potential targets for cancer therapy.