Expression of linear and novel circular forms of an INK4/ARF-associated non-coding RNA correlates with atherosclerosis risk

Abstract

Human genome-wide association studies have linked single nucleotide polymorphisms (SNPs) on chromosome 9p21.3 near the INK4/ARF (CDKN2a/b) locus with susceptibility to atherosclerotic vascular disease (ASVD). Although this locus encodes three well-characterized tumor suppressors, p16(INK4a), p15(INK4b), and ARF, the SNPs most strongly associated with ASVD are ∼120 kb from the nearest coding gene within a long non-coding RNA (ncRNA) known as ANRIL (CDKN2BAS). While individuals homozygous for the atherosclerotic risk allele show decreased expression of ANRIL and the coding INK4/ARF transcripts, the mechanism by which such distant genetic variants influence INK4/ARF expression is unknown. Here, using rapid amplification of cDNA ends (RACE) and analysis of next-generation RNA sequencing datasets, we determined the structure and abundance of multiple ANRIL species. Each of these species was present at very low copy numbers in primary and cultured cells; however, only the expression of ANRIL isoforms containing exons proximal to the INK4/ARF locus correlated with the ASVD risk alleles. Surprisingly, RACE also identified transcripts containing non-colinear ANRIL exonic sequences, whose expression also correlated with genotype and INK4/ARF expression. These non-polyadenylated RNAs resisted RNAse R digestion and could be PCR amplified using outward-facing primers, suggesting they represent circular RNA structures that could arise from by-products of mRNA splicing. Next-generation DNA sequencing and splice prediction algorithms identified polymorphisms within the ASVD risk interval that may regulate ANRIL splicing and circular ANRIL (cANRIL) production. These results identify novel circular RNA products emanating from the ANRIL locus and suggest causal variants at 9p21.3 regulate INK4/ARF expression and ASVD risk by modulating ANRIL expression and/or structure.

Figure 1. Identification and characterization of ANRIL splice variants.

A, 3′ and 5′ RACE was performed using primers directed against exons 4 and 6 where long stretches of unique sequence were observed (top). The resulting PCR products were cloned and sequenced, revealing several novel exons shown in blue (10a and 13b) and multiple non-colinear species (red). B, Equal quantities of total RNA were harvested from growing cell lines of various tissue types and absolute expression of the indicated transcript was determined. Expression levels are shown in a box-whisker plot on a log₁₀ scale in 11 of 27 analyzed cell lines which did not harbor homozygous 9p21 deletion. Validated Taqman detection strategies for the indicated ANRIL species are shown (top).

Figure 2. RNA Sequencing of ANRIL transcripts.

A, Coverage plots of RNA sequencing reads derived from short read archive study SRP002274 . Top, Read coverage across all ANRIL exons and nearby tumor suppressor genes p16^INK4a, ARF and p15^INK4b is shown. Bottom, The grey regions in the top panel were graphed on a truncated scale to better depict ANRIL coverage. Annotations above the larger peaks show the maximum number of reads mapping to these areas. B, Maximum peak height at each exon (normalized by overall locus coverage) is displayed from three independent samples: SRP002274 (Brain) and two ENCODE RNA-sequencing replicates of the HeLa cell line (HeLa rep 1 and 2). The inset shows all ANRIL exons on y-axis with peak height of 150 reads.

Figure 3. ANRIL 14-5 and 4-6 are circular RNAs.

A, Schematic representation of the ANRIL14-5 Taqman detection strategy wherein the probe (red) spans the exon 5-exon 14 boundary amplified with outward facing primers. B, Expression of the indicated transcripts was quantified in cDNA from Hs68 cells made using the indicate primers (-RT: no reverse transcriptase, H+dT: an equal mix of random hexamers and oligo dT, dT: oligo dT alone, and HEX: random hexamers alone). Error bars represent the standard deviation for three replicates. C, Total RNA harvested from growing Hs68 (top) and IMR90 (bottom) cells was incubated with our without RNase R, purified and reverse transcribed. The indicated transcripts were quantified in ‘B’. D, The average fold enrichment by RNase R for each transcript is shown on a log₁₀ scale.

Figure 4. ANRIL circular RNAs predominantly contain exons 4-14.

A, cDNA generated in the presence (+R) or absence (-R) of RNase R as in Figure 3C was subjected to PCR using outward facing primers within the same exon as depicted (left) and separated by gel electrophoresis. B and C, The PCR products in ‘A’ were purified, cloned and sequenced. The resulting sequences are shown for each exon pair.

Figure 5. ANRIL4-6 and 14-5 correlate with INK4/ARF expression and rs10757278 genotype in human PBTLs.

A, Taqman anlaysis of ANRIL and INK4/ARF transcripts was conducted and normalized as described in Materials and Methods. Correlations between ANRIL and INK4/ARF transcripts in 106 primary peripheral blood T-lymphocytes (PBTLs) were determined using linear regression. The R value for each pair-wise comparison is shown, with those achieving significance (p<0.05) depicted in blue. Due to limitations in sample availability, ANRIL 1-2 and ANRIL 14-5 levels were determined in only a subset of individuals (n = 94 and 98, respectively). B, The relative expression of ANRIL1-2, 14-5 and 18-19 normalized as in ‘A’ is plotted versus rs10757278 genotype. p-values were determined by a two-sided t-test.

Figure 6. Deep sequencing of 9p21 in pools of rs10757278 homozygotes.

A, The region captured using DNA sequence capture technology is shown on the ‘Tiling’ track. The ‘Unique’ track shows the Duke 35 bp Uniqueness information as provided in the UCSC Genome Browser. The bar at the top of the figure represents the 53 kb risk interval previously defined by Broadbent et al. . B, Venn diagrams depicting SNP calls for the AA (left) and GG (right) samples using three different algorithms. C, Using the UCSC Genome Browser, SNPs identified by next-generation DNA sequencing are depicted across the captured region of 9p21. The ‘Discovered’ track shows the polymophisms identified by two or more algorithms in either the pooled AA or GG sample. SNPs unique to each genotype are shown below. The ‘Unique Splice’ track depicts the location of the 4 SNPs, unique to the GG sample, which modify cis-acting splice regulation sites (See also Table S1).

Figure 7. Model showing how 9p21 polymorphisms influence ANRIL isoform production to modulate INK4/ARF gene transcription.

PcG complexes (e.g. PRC-1) are targeted to the coding INK4/ARF locus by ANRIL, and modulate its repression. Nascent ANRIL transcripts are spliced to produce circular ANRIL species (cANRIL). Causal variants in the ASVD risk interval modulate ANRIL transcription or splicing to influence INK4/ARF expression. See discussion for further description.