Analyses of Human Genetic Data to Identify Clinically Relevant Domains of Neuroligins

Abstract

Background/Objectives: Neuroligins (NLGNs) are postsynaptic adhesion molecules critical for neuronal development that are highly associated with autism spectrum disorder (ASD). Here, we provide an overview of the literature on NLGN rare variants. In addition, we introduce a new approach to analyze human variation within NLGN genes to identify sensitive regions that have an increased frequency of ASD-associated variants to better understand NLGN function. Methods: To identify critical protein subdomains within the NLGN gene family, we developed an algorithm that assesses tolerance to missense mutations in human genetic variation by comparing clinical variants from ClinVar to reference variants from gnomAD. This approach provides tolerance values to subdomains within the protein. Results: Our algorithm identified several critical regions that were conserved across multiple NLGN isoforms. Importantly, this approach also identified a previously reported cluster of pathogenic variants in NLGN4X (also conserved in NLGN1 and NLGN3) as well as a region around the highly characterized NLGN3 R451C ASD-associated mutation. Additionally, we highlighted other, as of yet, uncharacterized regions enriched with mutations. Conclusions: The systematic analysis of NLGN ASD-associated variants compared to variants identified in the unaffected population (gnomAD) reveals conserved domains in NLGN isoforms that are tolerant to variation or are enriched in clinically relevant variants. Examination of databases also allows for predictions of the presumed tolerance to loss of an allele. The use of the algorithm we developed effectively allowed the evaluation of subdomains of NLGNs and can be used to examine other ASD-associated genes.

Keywords: autism spectrum disorder; haploinsufficiency; human genetic variation; missense variation; neurodevelopmental disorder; neuroligin.

Figure 1

Early truncation variants are depicted on the exon maps of (A) NLGN1, (B) NLGN2, (C) NLGN3, and (D) NLGN4X, marked in black on top for gnomAD truncation variants, and in red on bottom for ClinVar early truncation variants. Atop the exon maps are the respective protein domain maps with gross structural features of NLGN1-4X, including signal peptide (black), extracellular domain (light blue), transmembrane domain (gray), and intracellular domain (dark blue). Intronic non-coded for regions of the protein alignment are depicted in blue stripes. The ending nucleotide and amino acid number are provided at the center of each exon. Due to varying intron size between the NLGN genes, introns were scaled down at a (A) 1:600, (B) 1:5, (C) 1:10, and (D) 1:70 ratio, respectively.

Figure 2

Diagrammatic explanation for determining clusters of sensitive residues along the span of a protein. (A) A rolling average is performed along the span of an example protein (in pink), with each instance of a missense variant having its location marked in black. For each residue within the protein, a tolerance value is calculated by summating nearby instances of missense variants multiplied by where they fall within the normal curve. (B) For residues near the far ends of the protein, the window of the normal curve will extend beyond the length of the protein, contributing false zeros. To correct for this, we multiply the calculated tolerance value by the percentage of the normal curve that falls outside the span of the protein. (C) We run this convolution algorithm on variants in both gnomAD and ClinVar and subtract the two curves to determine regions that are sparse in gnomAD variants while relatively more abundant in ClinVar variants.

Figure 3

Tolerance curves of (A) NLGN1, (B) NLGN2, (C) NLGN3, and (D) NLGN4X. The top panel includes the 2 overlayed curves, ClinVar in red and gnomAD in blue, with the lower chart representing the delta of the 2 curves. The protein domain maps for the various NLGNs are provided at the bottom to assist correlating regional sensitivities to functional domains. (Sky blue is signal peptide, blue is A1 and A2 splice inserts, light green is NRXN binding interface, burgundy is the dimerization interfaces, dark gray is transmembrane domain, yellow is phosphorylation sites, pink is gephyrin-binding domain, and mustard is PDZ domain.) (A) The chart for NLGN1 does not include gnomAD variants from residue 165–205, the site of the 2 splice inserts. This is annotated with a pink shading over this missing region of unimpacted gnomAD variants. (D) The sensitive region from residue 50–100 of NLGN4X is shaded in red.

Figure 4

An N-terminal critical region of (A) NLGN1, (B) NLGN3, and (C) NLGN4X, spanning from ~50–110 a.a. Analogous residue alignments are depicted between NLGN1, NLGN3, and NLGN4X, with gnomAD and ClinVar missense variants in blue and red, respectively. Missense variants that are common to both are highlighted in yellow. (D) The crystal structure of NLGN4X reveals the 64th asparagine, found within the first dip of this critical region, interacts with methionine 501 which is near another sensitive region of NLGN4X. (E) The second dip within this critical region is near an identified N-linked glycosylation site at N102.

Figure 5

The second observed critical cluster in (A) NLGN3 and (B) NLGN4X. Analogous residue alignments are depicted between NLGN3 and NLGN4X, with gnomAD and ClinVar missense variants in blue and red, respectively. Missense variants that are common to both are highlighted in yellow.

Figure 6

The third observed critical cluster in (A) NLGN3 and (B) NLGN4X, around the ~450 a.a. region of NLGN3. This is the region in which the NLGN3 R471C (R451C in most of the literature due to alternative splicing) is found. Analogous residue alignments are depicted between NLGN3 and NLGN4X, with gnomAD and ClinVar missense variants in blue and red, respectively. Missense variants that are common to both are highlighted in yellow.