Further Evidence That Non-Coding RNAs Contribute to ASD Risk

Background: We recently reported (Kerin et al, 2012, Sci Transl Med) that the non-coding RNA MSNP1AS (moesin pseudogene 1 anti-sense) is a functional element revealed by genome-wide significant association of the chromosome 5p14.1 marker rs4307059 with ASD risk (Wang et al, 2009, Nature). MSNP1AS expression was increased 12-fold in postmortem temporal cortex of individuals with ASD and increased MSNP1AS expression was correlated with the ASD-associated rs4307059 T/T genotype. These data indicated that the non-coding RNA MSNP1AS, rather than the flanking protein-coding genes CDH9 and CDH10, was the functional element implicated by the chromosome 5p14.1 ASD genome-wide significant association signal. Shortly after the first ASD GWAS report was published, another ASD GWAS report (Anney et al, 2010, Hum Mol Gen) indicated genome-wide significant association of the chromosome 20p12.1 marker rs4141463. 

Objectives: To determine the functional element revealed by the genome-wide significant association of the chromosome 20p12.1 marker rs4141463 with ASD risk.

Methods: Bioinformatics approaches were used to determine transcripts in the chromosomal region of the ASD-associated rs4141463 marker. Northern blot was used to confirm expression and transcript orientation in the human neuronal cell line SK-N-SH. Quantitative PCR (qPCR) was used to determine the expression levels of the identified RNA transcripts in a panel of human tissues as well as postmortem temporal cortex of individuals with ASD and matched controls.

Results: Although the rs4141463 marker lies within an intron of the protein-coding MACROD2 (MACRO domain containing 2) gene, expression of MACROD2 is neither altered in postmortem temporal cortex of individuals with ASD nor correlated with rs4141463 genotype. Our bioinformatics approaches revealed a transcript just 10 kb from rs4141463. Northern blot and qPCR confirmed expression of the 294-nucleotide non-coding RNA RPS10P2AS (ribosomal protein S10 pseudogene 2 anti-sense) in the human neuronal cell line SK-N-SH. In a panel of 15 human tissues, qPCR revealed that RPS10P2AS was expressed at higher levels than MACROD2 in fetal temporal cortex, adult cerebellum, and adult peripheral blood. In postmortem temporal cortex, expression of RPS10P2AS was increased 5-fold in individuals with ASD (P=0.002) and increased 7-fold in individuals with the ASD-associated rs4141463 C/C genotype (P=0.001). RPS10P2AS has high homology, in the reverse complement, to a large number of candidate target transcripts. Ongoing experiments seek to determine the transcriptome-wide impact of RPS10P2AS over-expression on neuronal gene expression.

Conclusions: These data indicate that multiple genome-wide significant associations with ASD implicate long non-coding RNAs. More specifically, the chromosome 5p14.1 and chromosome 20p12.1 GWAS hits both implicate long non-coding RNAs that are anti-sense transcripts of pseudogenes. Subsequent reports of attempts to replicate the original genetic finding did not provide additional support for an association of rs4141463 with ASD risk. However, the original genome-wide significant association of rs4141463 with ASD led to identification of a non-coding RNA with significantly altered expression in postmortem ASD temporal cortex. Because long non-coding RNAs are more abundant in human brain than protein-coding RNAs (Kapranov et al, 2010, BMC Biol), this class of molecules is likely to contribute to ASD risk.