Pathogenicity Prediction of GABAA Receptor Missense Variants

Abstract

Variants in the genes encoding the subunits of gamma-aminobutyric acid type A (GABA_A) receptors are associated with epilepsy. To date, over 1000 clinical variants have been identified in these genes. However, the majority of these variants lack functional studies and their clinical significance is uncertain although accumulating evidence indicates that proteostasis deficiency is the major disease-causing mechanism for GABA_A receptor variants. Here, we apply two state-of-the-art modeling tools, namely AlphaMissense, which uses an artificial intelligence-based approach based on AlphaFold structures, and Rhapsody, which integrates sequence evolution and known structure-based data, to predict the pathogenicity of saturating missense variants in genes that encode the major subunits of GABA_A receptors in the central nervous system, including GABRA1, GABRB2, GABRB3, and GABRG2. Our results demonstrate that the predicted pathogenicity correlates well between AlphaMissense and Rhapsody although AlphaMissense tends to generate higher pathogenic probability. Furthermore, almost all annotated pathogenic variants in the ClinVar clinical database are successfully identified from the prediction, whereas uncertain variants from ClinVar partially due to the lack of experimental data are differentiated into different pathogenicity groups. The pathogenicity prediction of GABA_A receptor missense variants provides a resource to the community as well as guidance for future experimental and clinical investigations.

Keywords: Epilepsy; GABAA receptors; Ion channels; pathogenicity; proteostasis.

Figure 1.. Structures and sequence alignment of GABA_A receptors.

(A) Cartoon representation of pentameric α1βγ2 receptors, built from 6X3S.pdb. The principal side of one subunit is denoted as “+”, whereas the complementary side of one subunit is denoted as “−”. (B) The schematic of the primary protein sequence of a GABA_A receptor subunit. NTD, N-terminal domain; M1-M4, transmembrane helices 1 to 4. (C) The secondary structures of a GABA_A receptor subunit. The two cysteines in the signature Cys-loop are colored in yellow. (D) The sequence alignment of major human GABA_A receptor subunits, including α1, β2, β3, and γ2. The residue positions that harbor clinical missense variants are highlighted. According to ClinVar annotation, pathogenic variants are colored in red, uncertain in yellow, and benign in green.

Figure 2.. Pathogenicity prediction of GABA_A receptor α1 subunit, GABRA1.

(A) Positions of clinical missense variants are displayed in the three-dimensional structure of α1 subunit, built from 6X3S.pdb using PyMOL. According to ClinVar annotation, Class I pathogenic variants are colored in red, Class II uncertain variants in yellow, and Class III benign variants in green. (B) Pathogenicity prediction of saturating substitutions of GABRA1. Heat maps display the pathogenic probability of selected key regions. The top picture is from Rhapsody (RS) prediction, whereas the bottom picture is from AlphaMissense (AM) prediction. Wild type amino acids were colored in yellow for RS, and black for AM. (C) Correlation between Rhapsody and AlphaMissense prediction. Red dots represent pathogenic variants, yellow dots represent uncertain variants, and green dots represent benign variants according to ClinVar annotation. If no predictions are available in Rhapsody, such variants are artificially assigned a value of zero for Rhapsody probability.

Figure 3.. Pathogenicity prediction of GABA_A receptor β2 subunit, GABRB2.

(A) Positions of clinical missense variants are displayed in the three-dimensional structure of β2 subunit, built from 6X3S.pdb using PyMOL. According to ClinVar annotation, Class I pathogenic variants are colored in red, Class II uncertain variants in yellow, and Class III benign variants in green. (B) Pathogenicity prediction of saturating substitutions of GABRB2. Heat maps display the pathogenic probability of selected key regions. The top picture is from Rhapsody (RS) prediction, whereas the bottom picture is from AlphaMissense (AM) prediction. Wild type amino acids were colored in yellow for RS, and black for AM. (C) Correlation between Rhapsody and AlphaMissense prediction. Red dots represent pathogenic variants, yellow dots represent uncertain variants, and green dots represent benign variants according to ClinVar annotation. If no predictions are available in Rhapsody, such variants are artificially assigned a value of zero for Rhapsody probability.

Figure 4.. Pathogenicity prediction of GABA_A receptor β3 subunit, GABRB3.

(A) Positions of clinical missense variants are displayed in the three-dimensional structure of β3 subunit, built from 6HUK.pdb using PyMOL. According to ClinVar annotation, Class I pathogenic variants are colored in red, Class II uncertain variants in yellow, and Class III benign variants in green. (B) Pathogenicity prediction of saturating substitutions of GABRB3. Heat maps display the pathogenic probability of selected key regions. The top picture is from Rhapsody (RS) prediction, whereas the bottom picture is from AlphaMissense (AM) prediction. Wild type amino acids were colored in yellow for RS, and black for AM. (C) Correlation between Rhapsody and AlphaMissense prediction. Red dots represent pathogenic variants, yellow dots represent uncertain variants, and green dots represent benign variants according to ClinVar annotation. If no predictions are available in Rhapsody, such variants are artificially assigned a value of zero for Rhapsody probability.

Figure 5.. Pathogenicity prediction of GABA_A receptor γ2 subunit, GABRG2.

(A) Positions of clinical missense variants are displayed in the three-dimensional structure of γ2 subunit, built from 6X3S.pdb using PyMOL. According to ClinVar annotation, Class I pathogenic variants are colored in red, Class II uncertain variants in yellow, and Class III benign variants in green. (B) Pathogenicity prediction of saturating substitutions of GABRG2. Heat maps display the pathogenic probability of selected key regions. The top picture is from Rhapsody (RS) prediction, whereas the bottom picture is from AlphaMissense (AM) prediction. Wild type amino acids were colored in yellow for RS, and black for AM. (C) Correlation between Rhapsody and AlphaMissense prediction. Red dots represent pathogenic variants, yellow dots represent uncertain variants, and green dots represent benign variants according to ClinVar annotation. If no predictions are available in Rhapsody, such variants are artificially assigned a value of zero for Rhapsody probability.