MGA loss-of-function variants cause premature ovarian insufficiency

Abstract

Although premature ovarian insufficiency (POI), a common cause of female infertility and subfertility, has a well-established hereditary component, the genetic factors currently implicated in POI account for only a limited proportion of cases. Here, using an exome-wide, gene-based case-control analysis in a discovery cohort comprising 1,027 POI cases and 2,733 ethnically matched women controls from China, we found that heterozygous loss-of-function (LoF) variants of MAX dimerization protein (MGA) were significantly enriched in the discovery cohort, accounting for 2.6% of POI cases, while no MGA LoF variants were found in the matched control females. Further exome screening was conducted in 4 additional POI cohorts (2 from China and 2 from the United States) for replication studies, and we identified heterozygous MGA LoF variants in 1.0%, 1.4%, 1.0%, and 1.0% of POI cases, respectively. Overall, a total of 37 distinct heterozygous MGA LoF variants were discovered in 38 POI cases, accounting for approximately 2.0% of the total 1,910 POI cases analyzed in this study. Accordingly, Mga+/- female mice were subfertile, exhibiting shorter reproductive lifespan and decreased follicle number compared with WT, mimicking the observed phenotype in humans. Our findings highlight the essential role of MGA deficiency for impaired female reproductive ability.

Keywords: Genetic diseases; Genetics; Monogenic diseases; Reproductive biology.

Figure 1. Identification of MGA as a POI-associated gene in the discovery cohort.

(A) Genetic analysis pipeline of exome-wide gene burden tests using WES data in the discovery POI cohort and matched female controls. (B) Gene-based Q-Q plot of P values in the discovery stage based on 2-sided Fisher’s exact tests. Expected P values are obtained from 1,000 permutations. The burden of variants adhering to 3 variant types, namely LoF, damaging missense, and synonymous, was tested for each of the 19,199 genes. Points on the plot are color coded according to the variant types indicated in the legend. The inflation factors (λ) for the 3 variant types are 1.08, 1.03, and 1.00, respectively. The red dashed line represents the Bonferroni’s correction threshold of 2.6 × 10^–6 (0.05/19,199) for each variant type. (C) Manhattan plot illustrating the association between genes with LoF variants and POI from the discovery cohort. The –log₁₀(observed P values) are plotted against the genetic position for each analyzed gene. The red dashed line represents the Bonferroni’s correction threshold, and the blue dashed line indicates a suggestive significance threshold of 0.005. The genes achieving the suggestive significance threshold are labeled.

Figure 2. Validation and schematic representation of LoF variants identified in patients with POI.

(A and B) Splicing results of the MGA c.5504-2A>G (A) and c.7139+1G>A (B) variants by mini-gene assays. Sanger sequencing results of the recombinant vectors are shown. Electrophoresis results showing the transcript PCR products obtained from both 293T and HeLa cell lines. The transcript of the c.5504-2A variant results in skipping of exon 17 and introducing a new premature termination in exon 18. The transcript of the c.7139+1G>A variant retained an 83 bp fragment of intron 18 and results in the introduction of a new premature termination codon. Both the c.5504-2A>C and c.7139+1G>A variants had shorter products compared with WT. (C) Schematic representation of the MGA protein and the LoF variants identified in patients with POI. The depicted domains include the T-box and basic helix-loop-helix (bHLH) domains of MGA.

Figure 3. Variant inheritance in 10 independent pedigrees with heterozygous MGA LoF variants.

Pedigrees of 10 unrelated families with individuals harboring heterozygous LoF variants in MGA. The arrows in families 3 and 4 indicate the probands, i.e., the individuals through whom the genetic variants were initially identified. The genotypes of the individuals are indicated below the symbols, with W representing the WT allele.

Figure 4. Subfertility of Mga^+/− female mice.

(A) Schematic illustration of the targeting strategy used to generate Mga-mutated mice using CRISPR/Cas9 technology. (B) The sgRNA was designed to target exon 3 of mouse Mga, and the protospacer adjacent motif is indicated by the blue highlight. Sanger sequencing results of the WT and heterozygous Mga mutants showed 1 bp deletion plus 3 bp insertion (c.1344delinsGTA) was introduced in mouse Mga, resulting in a frameshift mutation (p.Ser448Argfs*13) of MGA. The mutated nucleotides and amino acids are highlighted in red. The termination codon is indicated by an asterisk. (C) Cumulative number of pups produced per female mouse through continuous breeding. Both WT (n = 8) females and Mga^+/− (n = 8) females were mated with WT male mice. Asterisks indicate statistically significant differences between the 2 groups. (D) Statistical summary of breeding/reproductive features of Mga^+/− and WT female mice, including maternal age at first litter, maternal age at last litter, average number of pups per litter, and average number of litters. Central line of boxplots denotes median, box marks interquartile range and whiskers ×1.5 interquartile range. Statistical significances of C and D were determined using a 2-sided unpaired t test. **P < 0.01; ***P <0.001.

Figure 5. Accelerated depletion of ovarian reserve in Mga^+/− female mice.

(A) Representative images of reconstructed 3D ovaries. Mouse oocytes collected at P0 were stained with TRA98; oocytes collected at 4M and 8M were costained with DDX4 and P63 antibodies. Data analysis was conducted by spot transformation using Imaris software. Spots representing oocytes are colored based on size. Small oocytes with small diameter between 14 and 40 μm, which represent quiescent oocytes in primordial follicles, are indicated as gray spots. Large oocytes with diameter greater than 40 μm, representing growing oocytes in secondary and antral follicles, are indicated as red spots. The scale bars indicate 100 µm at P0 and 300 µm at both the 4M and 8M time points. The enlarged views are magnified to 5 times the original scale. (B) The number of total oocytes per ovary at P0, 4M, and 8M, respectively. Significantly fewer total oocytes were observable in Mga^+/− females compared with WT females at 8M. (C) Statistical summary of small and large oocytes per ovary in 4M and 8M between Mga^+/− and WT females. Data in B and C show means ± SEM for at least n = 6 mice per genotype for each time point and can be found in the Supporting Data Values file. (D) H&E staining of ovarian sections of 8M mice. (E) Box plot showing statistical summary of primordial, primary, secondary, and antral follicles per ovary in WT (n = 9) and Mga^+/− (n = 8) mice age 8M. Central line of box plots denotes median; box marks interquartile range and whiskers 1.5× interquartile range. Statistical significances of B, C, and E were determined using a 2-sided unpaired t test. *P < 0.05; **P < 0.01.

Figure 6. MGA suppresses the transcript expression of meiotic genes in ovary.

(A) Volcano plots of the gene expression profile in the ovaries between WT and Mga^+/− female mice aged P1, P5, 1M, and 5M, respectively. Bulk RNA-Seq was conducted on at least 3 ovaries of each genotype at each time point. Upregulated differential expressed genes with log₂FC > 1 and FDR < 0.001 are denoted in red, while downregulated DEGs with log₂FC < –1 and FDR < 0.001 are denoted in blue. (B) Heatmaps showing expressions of the total of 37 upregulated DEGs, 14 of which are meiotic genes and are denoted in bold. (C) GO enrichment analyses of 37 upregulated DEGs. (D) Expressions of the 14 DEGs related to meiosis in the ovaries of female mice aged P1, P5, 1M, and 5M. Points indicate means.

Figure 7. scRNA-Seq analysis uncovers altered functions in Mga^+/− female mice.

(A) Uniform manifold approximation and projection (UMAP) showing clusters of ovarian cell types for 6 major cell types. (B) PCA plot showing 4 subpopulations of GCs. (C) PCA plot showing the expression of Mga in GC subpopulations. (D) Pathway enrichment analysis using KEGG database of DEGs for 6 major cell types. (E) Bar chart showing KEGG pathway enrichment result in PAF GCs, SAF mGCs, LAF mGCs, and CCs. (F) GSEA showing enrichment of suppressed ovarian steroidogenesis in ovaries from Mga^+/− female mice compared with WT female mice. (G) Heatmap showing the expression of DEGs involved in ovarian steroidogenesis in GCs.