20233
Dynamic Gene Network Analysis of Neuronal Differentiation Identifies Novel Gene-Network Clusters Specifically Enriched for Autism Risk Genes
Autism Spectrum Disorders (ASD), and ASD related disorders such as Fragile-X Mental Retardation Syndrome (FXS) or other syndromic disorders with intellectual disability (ID) are hallmarked by aberrant neuronal function and connectivity. Several studies support the hypothesis of a disturbed neuronal maturation underlying these disorders. Genetic studies report an etiological overlap, however, the clinical phenotypes are specific. Thus, we tested if genes associated with ASD, FXS or ID are activated and or regulated in distinct gene-expression modules during ND, and identified the pathways specifically impaired.
Objectives:
We used a neuronal cellular model (SHSY5Y) to explore at systems biological level the role of ASD, FXS and ID implicated genes during neuronal differentiation (ND). We aimed at identifying transcriptional modules that are enriched for risk genes, and that are specifically impaired in the disorders. Identified modules were characterized with respect to their association with morphological changes. This allows us to estimate the functional effect of impaired expression of cluster associated genes.
Methods:
To analyze ND we differentiated SH-SY5Y cells by simultaneous application of retinoic acid and brain derived neurotrophic factor over a time course of two weeks. Whole transcriptome (Illumina HumanHT-12 v4 Expression BeadChips) and morphological development (Sholl parameters of GFP transfected co-cultivated cells) were analyzed at 7 time-points during this process. Significantly regulated genes were identified via t-test, dynamic time warping (DTW) and parallel independent component analysis (pICA) and subjected to weighted gene co-expression network analysis (WGCNA). Identified modules were analyzed for association with morphological development. GO-term enrichment analyses and gene-set enrichment analyses were performed in parallel.
Results:
Most of the mRNA regulation occurred within 168h of differentiation. Comparing transcriptomic patterns in our in-vitro model with published in-vivo Data (Kang et al., 2011) using the CoNTExT Algorithm (Stein et al., 2014) shows cortical identity of differentiated cells as well as a maturity stage corresponding to pcw 16. Identified modules of regulated genes showed specific overlap with ASD, FXS and ID risk genes. Interestingly, also disorder specific enrichment was observed. One module, associated with Retinoic acid metabolism, was enriched for ASD-genes only, whereas the translation implicated module was associated with both, FXS and ASD implicated genes. ID implicated genes were mainly enriched among down-regulated genes. In addition, we show that modules enriched for ASD genes are associated with neurite outgrowth.
Conclusions:
Our results show that risk genes for ASD, FXS or ID do affect similar but distinct modules and allow thus to hypothesize on the differences in the clinical presentation of the disorders. Furthermore we support the hypothesis that genetic variants identified in ASD patients are targeting on the one hand mechanisms implicated in FXS or ID, and on the other hand ASD-specific mechanisms. The latter processes may thus merit further attention when developing strategies to differentiate ASD from other disorders with overlapping symptomatology.