Gene Variants Involved in Nonsense-Mediated mRNA Decay Suggest a Role in Autism Spectrum Disorder

Abstract

Autism Spectrum Disorder (ASD) is a heterogeneous neurodevelopmental condition with unclear etiology. Many genes have been associated with ASD risk, but the underlying mechanisms are still poorly understood. An important post-transcriptional regulatory mechanism that plays an essential role during neurodevelopment, the Nonsense-Mediated mRNA Decay (NMD) pathway, may contribute to ASD risk. In this study, we gathered a list of 46 NMD factors and regulators and investigated the role of genetic variants in these genes in ASD. By conducting a comprehensive search for Single Nucleotide Variants (SNVs) in NMD genes using Whole Exome Sequencing data from 1828 ASD patients, we identified 270 SNVs predicted to be damaging in 28.7% of the population. We also analyzed Copy Number Variants (CNVs) from two cohorts of ASD patients (N = 3570) and discovered 38 CNVs in 1% of cases. Importantly, we discovered 136 genetic variants (125 SNVs and 11 CNVs) in 258 ASD patients that were located within protein domains required for NMD. These gene variants are classified as damaging using in silico prediction tools, and therefore may interfere with proper NMD function in ASD. The discovery of NMD genes as candidates for ASD in large patient genomic datasets provides evidence supporting the involvement of the NMD pathway in ASD pathophysiology.

Keywords: autism spectrum disorder; copy number variants; nonsense-mediated mRNA decay; single nucleotide variants.

Figure 1

Simplified representation of the NMD pathway in mammalian cells. (A) The Exon Junction Complex (EJC; composed by eIF4A3, RBM8A and MAGOH) is formed during splicing and deposited 20–24 nucleotides (nts) upstream of exon–exon junctions, remaining associated with the mRNA during its transport to the cytoplasm. (B) In the cytoplasm, the mRNA is translated by the ribosome. If the mRNA carries a premature translation termination codon (PTC), during the pioneer round of translation, i.e., while the mRNA is still bound to the cap binding complex CBP80/20, the ribosome stops at the PTC. If the PTC is located more than 50–54 nts upstream of the last exon–exon junction, the ribosome is not able to displace downstream EJCs. Instead, when the ribosome stops at the PTC, UPF1 can interact with eRF3a, inducing premature translation termination that triggers NMD. For that, the SURF complex is formed by UPF1, eRF1, eRF3a, the SMG1c kinase complex and DHX34 associated with RUVBL1 and RUVBL2. At this stage, both PYM1 and UPF3A can act as NMD repressors. (C) UPF2 and UPF3B, either diffused in the cytoplasm (in case of EJC-independent NMD) or bound to the EJC, interact with UPF1, favoring its phosphorylation by SMG1 and the formation of the DECID complex. (D) Phosphorylated UPF1 recruits SMG6, which cleaves the mRNA near the PTC, and the SMG5-SMG7 dimer, which recruits the decapping complex DPC2-DCP1a through PNRC2 and the deadenylation complex CCR4-NOT through CNOT8. SRSF1 and SMG5-SMG7 independently recruit PP2A, which dephosphorylates UPF1. NMD-targeted mRNAs are further degraded by 5′-to-3′ and 3′-to-5′ exonucleolytic activities of XRN1 and the exosome, respectively. (E) Model of the ER-NMD pathway. The NMD model was adapted from Nogueira et al., 2021 [29]. Legend: NMD, Nonsense-Mediated mRNA Decay; PTC, Premature Termination Codon; EJC, Exon Junction Complex; SURF, SMG1-UPF1-eRF1-eRF3 complex; DECID, Decay Inducing Complex; PP2A, Protein Phosphatase 2A; ER-NMD, NMD response at the Endoplasmic Reticulum; TC, Termination Codon.

Figure 3

Distribution of rare SNVs predicted to be damaging among 38 NMD genes in ASD patients. (A) Variants in NMD genes were identified in 1 to 5% of ASD patients. (B) Detailed distribution of gene variants in <1% of the ASD population sample.

Figure 4

Schematic domain architecture of proteins involved in NMD, and representations of the protein alterations encoded by MISPD and nonsense/frameshift variants identified in ASD subjects. The binding regions for interacting proteins are indicated. Proteins were grouped by NMD groups: (A) EJC, (B) SURF-DECID, (C) mRNA decay, (D) ER-NMD and (E) Regulator. Legend: MISPD, missense variants predicted to be damaging and deleterious; BD, Binding Domain; CH, cysteine and histidine-rich domain; SQ, serine- and glutamine-rich domain; RS, arginine and serine-rich domain; RRM, RNA recognition motif; EJC, Exon Junction Complex; PIN, PilT N-terminus; P-UPF1, phosphorylated UPF1; WH, Winged-Helix; CTD, C-terminal domain; NTD, N-terminal domain.

Figure 4

Schematic domain architecture of proteins involved in NMD, and representations of the protein alterations encoded by MISPD and nonsense/frameshift variants identified in ASD subjects. The binding regions for interacting proteins are indicated. Proteins were grouped by NMD groups: (A) EJC, (B) SURF-DECID, (C) mRNA decay, (D) ER-NMD and (E) Regulator. Legend: MISPD, missense variants predicted to be damaging and deleterious; BD, Binding Domain; CH, cysteine and histidine-rich domain; SQ, serine- and glutamine-rich domain; RS, arginine and serine-rich domain; RRM, RNA recognition motif; EJC, Exon Junction Complex; PIN, PilT N-terminus; P-UPF1, phosphorylated UPF1; WH, Winged-Helix; CTD, C-terminal domain; NTD, N-terminal domain.

Figure 2

Overview of the analysis. The study began with the identification of 46 genes (Table 1) encoding proteins involved in the NMD pathway or in its regulation (yellow) that were used for further analyses. This workflow describes the analysis of SNVs obtained from ASC WES datasets for 1828 ASD patients (1338 BI and 490 BCM, blue) and the analysis of CNVs predicted from SNP genotyping data in 3570 ASD patients from AGP (2446) and SSC (1124) datasets (green). Analysis proceeded separately for CNVs and SNVs to identify variants in NMD genes and ended up with the identification of 270 SNVs (Figure 3) and 38 CNVs (Table 2) in 524 and 38 ASD probands, respectively. Protein domains affected by SNVs or CNVs were then identified (orange), and a total of 136 genetic variants were located within regions required for NMD in 258 ASD patients (Table 3). Legend: SNVs, Single Nucleotide Variants; CNVs, Copy Number Variants; ASC, Autism Sequencing Consortium; AGP, Autism Genome Project; SSC, Simons Simplex Collection; BI, Broad Institute; BCM, Baylor College of Medicine; DGV, Database of Genomic Variant; CDS, coding sequence for protein; MAF, Minor Allele Frequency; NFE, Non-Finnish European; MISPD, missense variants predicted to be damaging and deleterious; LoF, loss-of-function.