Exon-Level Expression Profiling: Evaluation of Global Alternative Splicing in Autism, a Proof of Concept Study

Background: Autism is a complex, heterogeneous neurobehavioral disorder with many causes and varying degrees of severity. To date, the main technique used to screen autism candidate genes has been direct sequencing of exons using genomic DNA. It is surprising that for strong candidate genes such as NLGN3 and NLGN4, few functional mutations have been identified in association with autism. This suggests that implications may be at a higher level than their genomic sequence, including alternative splicing. Previously, we reported novel alternative splicing variants in the NLGN3 and NLGN4 genes using RT-PCR and DNA sequencing. Alternative splicing of primary transcripts (pre-mRNAs) is a regulatory mechanism contributing to the increased complexity of higher eukaryotic organisms resulting in exon skipping, alternative splice site usage, and intron retention, thus leading to structurally and functionally distinct transcript isoforms and protein variants. Alternative splicing is a gene expression regulatory process that allows a single gene to generate multiple transcripts increasing protein diversity. The dysregulation of this process has been associated with some cancers and neurological diseases such as Parkinson's and Alzheimer's. Objectives: Here we report the feasibility of exon-level global gene expression profiling in autism using lymphoblastoid cell lines (LCLs). Methods: Global alternative splicing was evaluated using the Affymetrix Human Exon 1.0 ST array in LCLs from five autism subjects and five unrelated, age and gender matched controls obtained from the Autism Genetics Resource Exchange. Exon array data was analyzed using Agilent's GeneSpring GX 10.0. Gene Ontology and pathway analysis were conducted using the functional annotation tool, DAVID. In addition to DAVID, the Pearson correlation coefficient was used to measure gene relations (acquired from the Human Protein Reference Database and TRANSFAC database) in differentially expressed gene list identified by exon array profiling of autism compared to control samples. Results: ANOVA and Bonferroni multi-test correction identified 57 genes that exhibit differential expression at the exon- level, suggesting potentially different alternatively spliced transcripts between the two study groups. This list includes genes associated with neurological diseases such as Alzheimer's (SLC30A7). Several of the genes identified have been linked to nervous system development or autism including CYFIP1, ROBO1, and CDKL5. Biological processes overrepresented in this gene list included nervous system development, post-translational protein modification, neuronal activities, protein modification and amino acid biosynthesis. Upon RT-PCR validation of gene exons identified as differentially expressed between the two groups, we discovered multiple unreported splice variants of a zinc-finger protein, ZMYM6. Each new ZMYM6 spliced isoform has been confirmed using exon-junction-specific PCR primers and DNA sequencing. Conclusions: This is the first study to evaluate global alternative splicing in autism. Our results indicate that changes at the alternative splicing level should be considered in the etiology of autism. This pilot study demonstrates that even though LCLs are not neuronal cells, they are a viable cellular model for human neurological diseases and can provide informative alternative splicing data regarding complex neurodevelopmental disorders such as autism.