Characterisation of the long non-coding RNA ANRIL in human melanoma

Abstract

The long non-coding RNA ANRIL is transcribed from a gene, antisense to the CDKN2B locus, that encompasses multiple disease-associated polymorphisms. Divergent reports have suggested both concordant and discordant expression of ANRIL and CDKN2A or CDKN2B in different tissues and cell types. In view of these conflicting studies, and given the recurrent involvement of the CDKN2A gene (which encodes two tumour suppressor proteins p16INK4 and p14ARF) in melanoma, we examined the relationship of ANRIL and CDKN2A in patient-derived metastatic melanoma cell lines. Gene expression analyses using quantitative RT-PCR revealed that ANRIL and CDKN2A expression are positively correlated in a panel of melanoma cell lines. We confirmed that p16INK4a mRNA is expressed in several cell lines in which p16INK4a protein is not detectable, ruling out transcriptional suppression of p16INK4a by ANRIL. This suggested that an epigenetic mechanism may regulate p16INK4a mRNA function post-transcriptionally. Multiple isoforms of ANRIL have been identified and several functions have been associated with ANRIL. In our judgement, studies on ANRIL functionality are premature pending a more complete appreciation of the complexity of isoforms. Multiple circular forms of ANRIL (circANRIL) were identified along with linear isoforms in melanoma cells using quantitative RT-PCR. Further characterisation of circANRIL in two melanoma cell lines using outwardfacing primers against each exon, followed by cloning and sequencing, revealed the existence of an assortment of circular isoforms with varying lengths, exon-exon junctions and exon combinations. Moreover, in these two cell lines, the complements of circANRIL isoforms were almost completely different. This reveals the dynamic nature of the locus and constitutes a basis for investigating the functions of ANRIL in melanoma. We also found the family of linear ANRIL was enriched in the nucleus, whilst the circular isoforms were enriched in the cytoplasm and they differed markedly in stability. This variability in localization pattern suggests the hypothesis that ANRIL may have dual functions in melanoma. RNAi studies targeting circANRIL revealed negative regulation of linear ANRIL and p16INK4A mRNA. This further is suggestive of ANRIL functioning either at transcriptional or posttranscriptional levels. Based on our findings, we hypothesise that ANRIL is a multiplicity of transcripts that have wholly distinct dual sets of functions in melanoma.