A Long Noncoding RNA, MSNP1AS, Contributes to ASD Risk

Background: Twin concordance and sibling recurrence rates suggest a strong contribution of genetic factors to autism spectrum disorder (ASD) risk. However, the genetics of ASD have proven to be complex. Genome-wide association studies (GWASs) are designed to identify novel genes and pathways that contribute to complex disorder risk. Application of GWAS techniques to ASD identified genetic markers with genome-wide significant association on chromosomes 5p14.1, 5p15.2, and 20p12.1. Association of the chromosome 5p14.1 marker rs4307059 with ASD diagnosis was replicated in an independent sample. Further, rs4307059 was associated with social communication phenotypes in a general population sample, providing additional evidence that this chromosome 5p14.1 genetic signal contributes to ASD-related phenotypes. Although widely interpreted as implicating the nearest protein-coding genes, cadherin 9 (CDH9) and cadherin 10 (CDH10), the original GWAS publication reported a lack of correlation between rs4307059 genotype and brain expression of the cadherins.

Objectives: To determine the biological basis of the chromosome 5p14.1 GWAS peak.

Methods: Bioinformatics approaches identified a single noncoding RNA directly under the chromosome 5p14.1 GWAS peak. Northern hybridization confirmed expression of the ~4 kb noncoding RNA, MSNP1AS (moesin pseudogene 1, antisense), and indicated that MSNP1AS binds the transcript of the X chromosome protein-coding gene moesin (MSN). Quantitative PCR (qPCR) was used to determine expression levels of MSNP1AS, MSN, CDH9 and CDH10 in 10 pairs of autism-control postmortem temporal cortex samples. Transfection of a MSNP1ASover-expression construct into human neuronal cells was followed by Western blot analysis of moesin protein.

Results: Expression of the noncoding RNA MSNP1AS was increased 12.7-fold (P=0.004) in postmortem temporal cortex of individuals with ASD compared to controls. Further, increased expression of MSNP1AS in postmortem temporal cortex was correlated with the ASD-associated rs4307059 genotype. Consistent with previous microarray reports, expression levels of the nearest protein-coding genes, CDH9 and CDH10, were not altered in ASD temporal cortex. Confirming results in the original GWAS report, we also found that expression levels of CDH9 and CDH10 were not correlated with rs4307059 genotype, suggesting that the ASD genetic association signal does not implicate the cadherins. Expression of the X chromosome gene MSN was increased 2.4-fold (P=0.029). Despite the significantly increased MSN RNA, moesin protein levels were not increased in postmortem temporal cortex of individuals with ASD, suggesting that the noncoding RNA MSNP1AS may play a role in reducing moesin protein. To test this hypothesis, we over-expressed MSNP1AS in a human neuronal cell line. Western blot analysis indicated a significant 40% decrease in moesin protein, establishing that MSNP1ASnegatively regulates moesin protein expression.

Conclusions: We identified a previously uncharacterized noncoding RNA, MPSNP1AS, which represents an ASD candidate gene with genome-wide significant association, functional correlation with the ASD-associated genetic allele, and a large increase in expression in postmortem ASD temporal cortex. MSNP1AS binds MSN, and over-expression of MSNP1AS causes a decrease in moesin protein. This ongoing work represents the critical post-GWAS translation of genetic findings to an understanding of their biological consequences and highlights the potential contributions of noncoding RNAs to ASD risk.