Mendelian randomization and colocalization analysis reveal novel drug targets for myasthenia gravis

Abstract

Objective: Myasthenia gravis (MG) is a complex autoimmune disease affecting the neuromuscular junction with limited drug options, but the field of MG treatment recently benefits from novel biological agents. We performed a drug-targeted Mendelian randomization (MR) study to identify novel therapeutic targets of MG.

Methods: Cis-expression quantitative loci (cis-eQTL), which proxy expression levels for 2176 druggable genes, were used for MR analysis. Causal relationships between genes and disease, identified by eQTL MR analysis, were verified by comprehensive sensitivity, colocalization, and protein quantitative loci (pQTL) MR analyses. The protein-protein interaction (PPI) analysis was also performed to extend targets, followed by enzyme-linked immunosorbent assay (ELISA) to explore the serum level of drug targets in MG patients. A phenome-wide MR analysis was then performed to assess side effects with a clinical trial review assessing druggability.

Results: The eQTL MR analysis has identified eight potential targets for MG, one for early-onset MG and seven for late-onset MG. Further colocalization analyses indicated that CD226, CDC42BPB, PRSS36, and TNFSF12 possess evidence for colocalization with MG or late-onset MG. pQTL MR analyses identified the causal relations of TNFSF12 and CD226 with MG and late-onset MG. Furthermore, PPI analysis has revealed the protein interaction between TNFSF12-TNFSF13(APRIL) and TNFSF12-TNFSF13B(BLyS). Elevated TNFSF13 serum level of MG patients was also identified by ELISA experiments. This study has ultimately proposed three promising therapeutic targets (TNFSF12, TNFSF13, TNFSF13B) of MG.

Conclusions: Three drug targets associated with the BLyS/APRIL pathway have been identified. Multiple biological agents, including telitacicept and belimumab, are promising for MG therapy.

Keywords: BLyS/APRIL pathway; Colocalization analysis; Drug target; Mendelian randomization study; Myasthenia gravis.

Fig. 1

Overview of the study design. Promising drug targets are identified by multiple steps including eQTL MR, sensitivity, colocalization, pQTL MR analysis, ELISA, clinical trial review, and safety evaluation. eQTL, expression quantitative trait loci; pQTL, protein quantitative trait loci; MR, mendelian randomization; MG, myasthenia gravis; IMGGC, International Myasthenia Gravis Genomics Consortium; PPI, protein-protein interaction

Fig. 2

eQTL MR results of each gene-disease association. (A-C) MR results of druggable genes with (A) MG, (B) Early-onset MG, and (C) Late-onset MG. Circle size indicates the p-value and the color of circles suggests the beta value of MR results of each gene-disease pair. (D) Sensitivity analysis result of 16 gene-disease associations

Fig. 3

Regional Manhattan plot of associations of SNPs with CD226, CDC42BPB, and PRSS36. (A) rs763362 is used to proxy the serum level of CD226. (B) rs763362 with its flanking 400 kb region in MG. (C) rs56249713 is used to proxy the serum level of CD226. (D) rs56249713 with its flanking 400 kb region in late-onset MG. (E) rs34681566 is utilized to proxy the serum level of CDC42BPB. (F) rs34681566 with its flanking 400 kb region in MG. (G) rs59735493 is utilized to proxy the serum level of PRSS36. (H) rs59735493 with its flanking region in MG

Fig. 4

Protein-level drug target identification. (A) The forest plot exhibits MR results of all protein-disease pairs with available pQTL data. The center of the error bars represents the odds ratio of MG, and per 1-standard-deviation increase in protein levels, which is calculated by Wald ratio (1 SNP) or IVW (> 1 SNP). MR analyses are based on three pQTL datasets (Zheng et al., Ferkingstad et al., Sun et al.), which can be distinguished by background color. (B) Protein-protein interaction (PPI) results of (i) TNFSF12, (ii) PRSS36, and (iii) CD226. (C) Serum levels of TNFSF12, TNFSF13, and TNFSF13B detected by ELISA experiments. (D) Schematic illustration of promising drug targets with its targeted medications and signal pathways. PPI, protein-protein interaction; ELISA, enzyme-linked immunosorbent assay; OR, odds ratio; 95% CI, 95% confidential interval Supporting information