Evolutionary constrained genes associated with autism spectrum disorder across 2,054 nonhuman primate genomes

Abstract

Background: Significant progress has been made in elucidating the genetic underpinnings of Autism Spectrum Disorder (ASD). However, there are still significant gaps in our understanding of the link between genomics, neurobiology and clinical phenotype in scientific discovery. New models are therefore needed to address these gaps. Rhesus macaques (Macaca mulatta) have been extensively used for preclinical neurobiological research because of remarkable similarities to humans across biology and behaviour that cannot be captured by other experimental animals.

Methods: We used the macaque Genotype and Phenotype (mGAP) resource consisting of 2,054 macaque genomes to examine patterns of evolutionary constraint in known human neurodevelopmental genes. Residual variation intolerance scores (RVIS) were calculated for all annotated autosomal genes (N = 18,168) and Gene Set Enrichment Analysis (GSEA) was used to examine patterns of constraint across ASD genes and related neurodevelopmental genes.

Results: We demonstrated that patterns of constraint across autosomal genes are correlated in humans and macaques, and that ASD-associated genes exhibit significant constraint in macaques (p = 9.4 × 10^{- 27}). Among macaques, many key ASD-implicated genes were observed to harbour predicted damaging mutations. A small number of key ASD-implicated genes that are highly intolerant to mutation in humans, however, showed no evidence of similar intolerance in macaques (CACNA1D, MBD5, AUTS2 and NRXN1). Constraint was also observed across genes associated with intellectual disability (p = 1.1 × 10^{- 46}), epilepsy (p = 2.1 × 10^{- 33}) and schizophrenia (p = 4.2 × 10^{- 45}), and for an overlapping neurodevelopmental gene set (p = 4.0 × 10^{- 10}).

Limitations: The lack of behavioural phenotypes among the macaques whose genotypes were studied means that we are unable to further investigate whether genetic variants have similar phenotypic consequences among nonhuman primates.

Conclusion: The presence of pathological mutations in ASD genes among macaques, along with evidence of similar genetic constraints to those in humans, provides a strong rationale for further investigation of genotype-phenotype relationships in macaques. This highlights the importance of developing primate models of ASD to elucidate the neurobiological underpinnings and advance approaches for precision medicine and therapeutic interventions.

Keywords: Autism spectrum disorder; GSEA; Genetic constraint; Primate model; Whole genome sequencing.

Fig. 1

A. Density histogram of the Residual Variation Intolerance Scores (RVIS) for SFARI ASD genes (blue, N = 729) and non-ASD (red, N = 17,437) genes. The RVIS scores are normalised and hence centre on a mean of 0. Delimiters on x-axis range from − 5.0 to 5.0. B. The descriptions are the same as A for the DISGENET ASD genes (blue, N = 793) and non-ASD (red, N = 17,373) genes. C. Box and whisker plots depicting the distribution of RVIS scores with median (red line) for SFARI ASD and non-ASD genes. Delimiters on x-axis range from − 1.5 to 1.5. *** P < 0.001 (t test). D. Same as C using the DisGenet ASD genes

Fig. 2

Gene Set Enrichment Analysis (GSEA) using ASD SFARI gene set (A) and DisGeNET ASD gene set (B). The x-axis denotes all genes ranked by the RVIS. The black vertical line indicates the ASD set genes. The y-axis is the Enrichment Score (ES) which represents the degree (i.e., the maximum deviation from zero) to which a set of genes is over-represented at the top of the ranked list accordingly to the RVIS. C. The scatter plot of RVIS scores for ASD-associated genes. The blue plots are for the 2% extreme of the distribution i.e., most intolerant in which the annotated plot indicates the DisGeNET ASD set genes. The red plots indicate 98% of the rest of the genes

Fig. 3

A. Density histogram of the Residual Variation Intolerance Scores (RVIS) for DisGeNET schizophrenia (A), epilepsy (B) and intellectual disability (ID) genes (C) (coloured in blue) and non-disorder associated genes (red). The RVIS scores are normalised and hence centre on a mean of 0. Delimiters on x-axis range from − 5.0 to 5.0. B. The descriptions are the same as A. D-F. Box and whisker plots depicting the distribution of RVIS scores with median (red line) for schizophrenia (D), epilepsy (E) and ID (F) DisGeNET genes. Delimiters on x-axis range from − 1.5 to 1.5. *** P < 0.001 (t-test)

Fig. 4

A. GSEA results using the set genes for schizophrenia (A), epilepsy (B) and ID (C). The descriptions are the same as Fig. 2A

Fig. 5

A. Density histogram of the Residual Variation Intolerance Scores (RVIS) for NDD (N = 101, blue) and non-NDD (N = 18,065, red) genes. The RVIS scores are normalised and hence the centre on a mean of 0. B. The box plots for any NDD genes (i.e., ASD, epilepsy, ID, schizophrenia) and non-NDD genes. The descriptions are the same as Fig. 2C and D. Box plots C. GSEA for the overlapped NDD genes