Autism genetics perturb prenatal neurodevelopment through a hierarchy of broadly-expressed and brain-specific genes

Abstract

Numerous genes are associated with autism spectrum disorder (ASD); however, it remains unclear how most ASD risk genes influence neurodevelopment and result in similar traits. Recent genetic models of complex traits suggest non-tissue-specific genes converge on core disease genes; so we analyzed ASD genetics in this context. We found ASD risk genes partition cleanly into broadly-expressed and brain-specific genes. The two groups show sequential roles during neurodevelopment with broadly-expressed genes modulating chromatin remodeling, proliferation, and cell fate, while brain-specific risk genes are involved in neural maturation and synapse functioning. Broadly-expressed risk genes converge onto brain-specific risk genes and core neurodevelopmental genes through regulatory networks including PI3K/AKT, RAS/ERK, and WNT/ -catenin signaling pathways. Broadly-expressed and brain-specific risk genes show unique properties, wherein the broadly-expressed risk gene network is expressed prenatally and conserved in non-neuronal cells like microglia. However, the brain-specific gene network expression is limited to excitatory and inhibitory neurons, spanning prenatal to adulthood. Furthermore, the two groups are linked differently to comorbidities associated with ASD. Collectively, we describe here the organization of the genetic architecture of ASD as a hierarchy of broadlyexpressed and brain-specific genes that disrupt successive stages of core neurodevelopmental processes.Recent studies present an alternate model of complex traits where the functional impact of genetic aberrations propagates through gene regulatory networks to converge on downstream core processes related to the trait (Boyle et al., 2017;Califano and Alvarez, 2017;Courchesne et al., 2020;Gazestani et al., 2019;Iakoucheva et al., 2019;Liu et al., 2019). In contrast to the single-group rASD network model, if the polygenic and omnigenic models are relevant to ASD, they would suggest that many risk genes are not necessarily part of underlying neural mechanisms in ASD. Rather many rASD genes would be broadly-expressed across many tissues, but regulate core neurodevelopmental processes underlying ASD (Boyle et al., 2017;Califano and Alvarez, 2017;Courchesne et al., 2020;Gazestani et al., 2019). Moreover, by emphasizing the gene regulatory networks, these alternative models suggest that, to the extent that the gene regulatory networks are preserved, the dysregulation could also be reflected in non-neuronal cells . Supporting these ideas, many rASD genes are broadly-expressed across tissues and strongly enriched for gene expression regulators such as chromatin remodelers, transcription factors, and modulators of signaling pathways (Courchesne et al., 2020;Courchesne et al., 2019;Werling et al., 2020). Similarly, genome wide association studies (GWAS) and whole genome sequencing both highlight the important role of broadly functional eQTLs and gene regulatory networks in ASD liability (Brandler et al., 2018;Courchesne et al., 2020;Grove et al., 2019).The fundamen...

Atypical genomic patterning of the cerebral cortex in autism with poor early language outcome

“…6A-C; Table S1). This is compatible with the idea that broadly expressed genes can interact and impact key ASD-risk genes, particularly in prenatal periods (9,10). Downstream targets of highly penetrant genes like FMR1 and CHD8 were also enriched in non-zero modules from SA LV1, CT LV1, and CT LV2.…”

“…Microglia enrichments are present and particularly strong for SA LV1 and CT LV1 non-zero modules. This effect may have implications for programmed cell death and pruning explanations (43) and which may be relevant to ideas behind ASD-relevant broadly expressed genes and their particularly strong effects on nonneuronal cell types such as microglia (10). The results so far suggest that SA and CT non-zero modules are highly prenatally relevant for establishing cortical patterning and regionalization and implicate several cell types that may be of mechanistic importance to different ASD early language outcome subtypes.…”

“…The omnigenic model also proposed that broadly expressed genes should have large impact on complex traits encompassed by neuropsychiatric phenotypes and diagnoses. Indeed, broadly expressed genes manifest in this and other studies (6) and are a key class of ASD genetic risk that operates at early prenatal timepoints (9,10). In future work it will be important to explore how genomic patterning of the cortex can affect other ASD subtypes.…”

“…Microglia enrichments are present and particularly strong for SA LV1 and CT LV1 non-zero modules. This effect may have implications for programmed cell death and pruning explanations (43) and which may be relevant to ideas behind ASD-relevant broadly expressed genes and their particularly strong effects on nonneuronal cell types such as microglia (10). (14).…”

Pre-treatment clinical behavioral and blood leukocyte gene expression patterns predict rate of change in response to early intervention in autism

“…Another revelation is that blood leukocyte genes associated with autism tend to be within a class of broadly expressed genes that are highly expressed in the brain and many other tissues ( 25 ). Broadly expressed genes are one class of important ASD-associated genes that primarily have peak levels of expression during prenatal development ( 3, 28 ). Given the sensitivity of blood leukocyte gene expression activity as a tool for assessing the living biology behind ASD toddlers ( 2 ), we reasoned that there may be pre-treatment gene expression patterns in ASD toddlers that may be predictive of treatment slopes.…”