NEW YORK (GenomeWeb) – Stanford University School of Medicine researchers have identified a molecular network linked to autism spectrum disorders, as they reported in Molecular Systems Biology today.
By studying the human protein interactome and known candidate genes for autism, the researchers homed in on a protein interaction module that is commonly mutated in people with the disorder.
"We have identified a specific module within this interactome that comprises 119 proteins and which shows a very strong enrichment for autism genes," senior author Michael Snyder from Stanford University School of Medicine said in a statement.
That module, they found, could then be divvied into two further groups: one that is expressed throughout the brain and one that is mostly expressed in the corpus callosum, particularly in oligodendrocyte cells. In mice, the researchers found that the corpus callosum-expressed set of genes is involved in the development of oligodendrocyte cells.
With this approach of associating pathways with disease, the researchers said they hoped to uncover protein interactions that contribute to the disorder and identify novel candidate genes for autism.
Snyder and his colleagues turned to the BioGrid human protein interactome to develop a new topological protein interaction network based on more than 13,000 proteins and 69,000 curated interactions. They clustered their new network into more than 800 topological modules.
Two of those modules — module two and module 13 — are enriched for the 383 genes on the SFARI list of genes that have been linked to autism. Module two, the researchers reported, was enriched for genes involved in transcriptional regulation, while module 13 was enriched for genes involved in synaptic transmission. Even after controlling for its preponderance of synaptic genes, module 13 was still linked to autism. Because of its stronger enrichment to ASD genes, the researchers focused their subsequent analyses on module 13.
The researchers sequenced DNA obtained from post-mortem brains of 25 autism patients — 19 underwent exome sequencing and six underwent whole-genome sequencing —and controls.
From this, they found that 153 non-synonymous mutations mapped to module 13. After excluding mutations also found in the controls, the researchers focused on 113 non-synonymous mutation sites, finding 38 genes — 28 of which had not been described previously — that were significantly affected by non-synonymous variants.
For example, they reported that LRP2, an ASD candidate gene linked to an underdeveloped or missing corpus callosum, had seven distinct non-synonymous mutations, four of which were predicted to be deleterious.
Snyder and his colleagues further noted that the affected loci were likely to be both rare within the population and functionally conserved, which they said suggested that they have a functional role. Indeed, in mice, 10 of those 28 novel candidate genes led to abnormal behavioral traits or nervous system defects.
The researchers also validated their results by sequencing the exomes of a further 505 people with autism and 491 controls.
Using the Allen Human Brain Atlas, Snyder and his colleagues gauged the gene expression of the module 13 genes across various brain tissues. Most genes, they found, were expressed throughout the brain, but a subset, dubbed group one, was more highly expressed in the corpus callosum, the region of the brain that mediates the signals between the left and right hemispheres.
"The module we identified, which is enriched in autism genes, had two distinct components," Snyder noted. "One of these components was expressed throughout different regions of the brain. The second component had enhanced molecular expression in the corpus callosum. Both components of the network interacted extensively with each other."
They validated these gene expression findings through RNA sequencing and immunohistochemical analysis of postmortem human brains.
In a mouse system, Snyder and his colleagues examined the role of the module 13 genes and found that group one genes were up regulated in mature myelinating oligodendrocytes, while group two genes were down regulated. This and other data, they said, indicated that group one genes were associated with oligodendrocyte maturation.
Using mouse knockout models of the module 13 genes, the researchers showed that the myelin gene regulatory factor has a role in developing myelination capacity for oligodendrocytes. Mice that lack that factor, they noted, have myelination defects and suffer from seizures.
By deep RNA sequencing of post-mortem brain tissue from matched autism patients and controls, Snyder and his colleagues found that the expression of module 13 genes was significantly different in the corpus callosum of patients as compared to controls.
They further folded their sequencing and expression findings into their protein interaction network to develop a network-level view of genes involved in ASD and to try to tease out patterns emanating from their topological positions.
For instance, they reported that centrally located proteins in the network were more likely to be mutated in people with autism, as were the peripheral proteins, yielding a U-shaped distribution. Combined with additional analyses, the researchers said that this indicated that the central genes have fundamental roles in the corpus callosum and pathogenic ASD mutations are more likely to be in those genes.
"The use of biological networks allowed us to superimpose clinical mutations for autism onto specific disease-related pathways," Snyder said. "This helps [us to find] the needles in the haystack worthy of further investigation and provides a framework to uncover functional models for other diseases."