A large number of putative risk genes for autism spectrum disorder (ASD) have been reported. The functions of most of these susceptibility genes in developing brains remain unknown, and causal relationships between their variation and autism traits have not been established. The aim of this study was to predict putative risk genes at the whole-genome level based on the analysis of gene co-expression with a group of high-confidence ASD risk genes (hcASDs). The results showed that three gene features – gene size, mRNA abundance, and guanine-cytosine content – affect the genome-wide co-expression profiles of hcASDs. To circumvent the interference of these features in gene co-expression analysis, we developed a method to determine whether a gene is significantly co-expressed with hcASDs by statistically comparing the co-expression profile of this gene with hcASDs to that of this gene with permuted gene sets of feature-matched genes. This method is referred to as "matched-gene co-expression analysis" (MGCA). With MGCA, we demonstrated the convergence in developmental expression profiles of hcASDs and improved the efficacy of risk gene prediction. The results of analysis of two recently-reported ASD candidate genes, CDH11 and CDH9, suggested the involvement of CDH11, but not CDH9, in ASD. Consistent with this prediction, behavioral studies showed that Cdh11-null mice, but not Cdh9-null mice, have multiple autism-like behavioral alterations. This study highlights the power of MGCA in revealing ASD-associated genes and the potential role of CDH11 in ASD.

Background: Gene co-expression analysis (GCA) has emerged as an important tool to identify convergent molecular pathways of ASD risk genes. The aim of this study is to identify ASD-relevant genes at the whole-genome level using GCA with consideration of the effect of confounding factors on GCA, including the size, expression level, and guanine-cytosine content of genes. Methods: Pearson’s correlation coefficient was computed to indicate the co-expression of a gene pair based on the BrainSpan human brain transcriptome dataset. Whether a gene is significantly co-expressed with a group of high-confidence ASD risk genes (hcASDs) was determined by statistically comparing the co-expression of this gene with the hcASD gene set to that of this gene with permuted gene sets of matched gene features. This method is referred to as "matched-gene co-expression analysis" (MGCA). Gene ontology (GO) analysis and construction of integrated GO enrichment networks were performed to reveal convergent pathways of co-expressed genes. Behavioral tests were carried out in gene knockout mice. Results: Gene size, mRNA length, mRNA abundance, and guanine-cytosine content were found to affect co-expression profiles of ASD genes. Using the MGCA method, we confirmed the convergence in the developmental expression profiles of hcASDs. MGCA also effectively revealed convergent molecular pathways of ASD risk genes and determined that CDH11, but not CDH9, is associated with ASD. Mouse behavioral studies showed that Cdh11-null mice, but not Cdh9-null mice, have multiple autistic-like behavioral alterations.Limitations: The use of tissue-derived transcriptomes instead of single-cell transcriptomes may have detected coincident expression of some functionally irrelevant genes in different cell types. Some ASD risk genes may have been missed due to the highly stringent statistical standard of MGCA. Another limitation is the relatively small number of animals analyzed in behavioral tests. Conclusions: Results of this study revealed the importance of considering matched gene features in GCA. CDH11 was confirmed to be an important ASD risk gene and Cdh11-null mice were found to be a very useful animal model for investigation of ASD.

Genetic variants of a large group of susceptibility genes have been associated with similar clinical manifestations of autism spectrum disorders (ASD), suggesting convergent function of these genes during brain development. Gene co-expression analysis has emerged as an important tool to identify convergent molecular pathways shared by ASD risk genes. In this study, four gene features, including gene size, mRNA length, mRNA abundance, and guanine-cytosine content, were found to profoundly affect gene co-expression. A new method termed "matched-gene coexpression analysis" (MGCA) was developed to screen for biologically meaningful gene coexpression relationships, taking into the consideration of the effect of these features on gene coexpression. With this method, significant convergence of the expression of high-confidence ASD risk genes (hcASDs) in the brain was demonstrated, and an improved efficacy in the prediction of convergent molecular pathways was achieved. This method also allowed the identification of " homophilic cell adhesion" as one of the convergent pathways of ASD-relevant genes. Analyses of CDH11 and CDH9, two specific genes coding for homophilic cell adhesion molecules, revealed that CDH11, but not CDH9, was significantly co-expressed with hcASDs. In addition, Cdh11-null mice, but not Cdh9-null mice, were found to exhibit multiple autistic-like behavioral alterations.These results suggest that CDH11 is an important ASD risk gene. Results of this study also highlight the importance of considering matched gene features in the analysis of gene coexpression.Conventionally, genome-wide gene co-expression networks are constructed after setting an empirically determined threshold of CC 6 . A major limitation in most of these studies is lack of consideration of effects of confounding factors such as the size, expression level (mRNA abundance), and guanine-cytosine (GC) content of genes on the result of gene co-expression analysis (GCA) 11 . Most hcASDs are large genes with a higher expression level in the brain than in other tissues 12 . It is unclear whether the size or expression level of genes affect the result of GCA. It is also unknown whether convergent expression is specific to ASD risk genes or a common property of genes with similar features, such as large gene size and high mRNA abundance 11 . Some hcASDs code for adhesion molecules, such as members of neurexin and neuroligin families, which mediate pre-and post-synaptic adhesion, respectively, in ASD-related brain circuits 13,14 . Genetic variants of several other adhesion molecules, including classical and nonclassical cadherin family members, are also frequently found to be associated with ASD by GWAS 15 and WES studies [16][17][18][19] . Cadherin family members play important roles in multiple developmental processes, including cell proliferation, polarization, neuronal migration, axon projection, dendrite arborization, and synapse assembly, by mediating homophilic and heterophilic cell-cell interactions [20][21][22][23][24] . It is unclear as...

No abstract