Pleiotropic effects of MORC2 derive from its epigenetic signature

Abstract

Heterozygous missense mutations in MORC2 have been implicated in various clinical entities, ranging from early-onset neurodevelopmental disorders to late-onset neuropathies. The mechanism underlying the phenotypic heterogeneity and pleiotropic effects of MORC2 has remained elusive. Here, we analysed blood and fibroblast DNA methylation, transcriptomes, proteomes and phenotypes of 53 MORC2 patients. We identified a MORC2-specific DNA methylation episignature that is universal across all MORC2-associated phenotypes and conserved across different tissues. The MORC2 episignature consists mainly of DNA hypermethylation in promoter regions, leading to transcriptional repression of target genes resulting in a MORC2-specific RNA signature. Concomitant downregulation of three disease-associated genes-ERCC8, NDUFAF2 and FKTN-at different levels mirrors the variable biochemical defects and clinical manifestations observed in MORC2 patients. Silencing of NDUFAF2 accounts for the Leigh syndrome manifestation, whereas dysmorphic features are due to the repression of ERCC8. Overall, we showed that pathogenic MORC2 variants cause specific episignature, whereby methylation level variability and its repression impact on target genes explains the pleiotropy and predicts phenotypic heterogeneity in MORC2-related disorders. We predict that epigenetic variation may underlie pleiotropy in other Mendelian disorders.

Keywords: CMT; Leigh syndrome; MORC2; episignature; multi-omics; pleiotropy.

Figure 1

Phenotypic heterogeneity in MORC2-realted disorders. (A) Main phenotypic abnormalities observed in MORC2-related disorders are shown. (B) Donut chart showing the distribution of different disorders associated with MORC2 pathogenic variants in a total of 112 patients, including both previously reported patients (n = 73) and new patients reported in this paper (n = 39). (C) Lollipop plot showing the distribution of pathogenic variants in MORC2 and their associated disorders. Variants reported more than five times or associated with more than one disorder are labelled in the plot. The data include 112 patients described in B.

Figure 2

Tissue-independent MORC2-specific episignatures are mainly localized in gene promoters. (A) Volcano plot of differential methylation analysis comparing MORC2 patients (n = 10) to healthy controls (n = 46). Shown in blue and red are the 760 significantly differentially methylated CpGs with a Bonferroni-adjusted P-value threshold of < 0.05. (B) The mean methylation Z-score of 220 CpGs (MORC2 episignature) (y-axis) was calculated for each individual with blood DNA methylation. Each filled circle represents one individual. The controls (grey-filled circles) consist of healthy individuals (n = 58) and patients with mutations in other genes than MORC2 (n = 424). The yellow-filled circle shows the individual with a benign stopgain variant in MORC2 (p.Arg90Ter), and the pink-filled circles represent the MORC2 patients. (C) Results of the support vector machine classifier trained on blood DNA methylation of 220 CpGs to predict MORC2 patients. The y-axis shows the probability that the DNA methylation profile matches the MORC2-specific episignature. Each dot represents one individual, and the x-axis indicates the causal genes or groups. ‘Mitochondrial’: Patients with pathogenic variants in mitochondrial disease genes (n = 301). ‘MORC2_Benign’: Two patients with benign missense variants in MORC2. ‘MORC2_LOF’: One patient with a benign loss-of-function variant in MORC2. Training set: Blood DNA methylation of 10 MORC2 patients and 46 controls. Test set: Blood DNA methylation of three MORC2 patients, 12 controls, one MORC2_Benign, 367 patients with other genetic disorders, and fibroblast DNA methylation of six MORC2 patients, four controls and 14 patients with mitochondrial disease. Validation set: Blood DNA methylation of 22 MORC2 patients, 20 patients with other genetic disorders, and fibroblast DNA methylation of six MORC2 patients and one patient with another genetic disorder. (D) Localization of the 760 CpGs identified in A relative to gene structure and known MORC2 target sites. ‘Body’ refers to CpG sites located within gene bodies and 3′-untranslated regions (3′UTRs). ‘TSS’ denotes the transcription start site. ‘TSS200’ represents CpG sites located 0–200 bases upstream of the TSS, and ‘TSS1500’ represents CpG sites located 200–1500 bases upstream of the TSS. Fisher’s exact test showed significant enrichment of differentially methylated CpG sites. (E) Venn diagram illustrating the overlap among CpG sites located in promoter regions, MORC2 binding sites and HUSH target sites. Of the 760 differentially methylated CpG sites, 463 that belong to at least one of these categories are included in the diagram. LOF = loss-of-function.

Figure 3

MORC2-specific RNA-signature in fibroblast cell lines. (A) Volcano plot of differential expression analysis in fibroblast cell lines comparing MORC2 patients (n = 10) to healthy or diseased controls (n = 42). The 106 significant differentially expressed genes at a Benjamin–Hochberg-adjusted P-value threshold of <0.05 are shown in blue and red. (B) Results of the support vector machine classifier trained on the expression profile of the 106 genes identified in A. The y-axis shows the probability that the transcriptome profile matches the MORC2-specific RNA signature. Training set: 10 MORC2 patients, 13 healthy controls, 29 diseased controls. Test set: Two MORC2 patients, five healthy controls, five diseased controls and 168 genetically undiagnosed patients.

Figure 4

The downstream effect of aberrant DNA methylation on gene expression. (A) Mirror Manhattan plot displaying the results of differential methylation analysis (top, red-filled circles) and differential expression analysis (bottom, blue-filled circles). The y-axis represents −log₁₀(P-value) for both analyses, and the x-axis denotes the genomic positions of CpG sites or genes. Threshold P-values were set at 7.93 × 10⁻⁸ (corresponding to Bonferroni-adjusted P-value < 0.05) for differential methylation and 2.7 × 10⁻⁴ (corresponding to Benjamin–Hochberg-adjusted P-value < 0.05) for differential expression analyses (indicated by dashed lines). Differentially methylated CpG sites mapped to C2H2-ZNF genes and differentially expressed C2H2-ZNF genes are highlighted in green. Genes that are both differentially methylated and differentially expressed are shown in bold. OMIM disease genes that are both differentially methylated and expressed are enclosed in rectangles and labelled. The Venn diagram at the bottom illustrates the overlap between C2H2-ZNFs, HUSH target genes, differentially expressed genes and differentially methylated genes. (B) Enrichment of C2H2-ZNFs among both differentially methylated CpGs (top) and differentially expressed genes (bottom) with Fisher’s exact test P-values of 1.35 × 10⁻¹⁴⁶ and 3.18 × 10⁻⁴⁰, respectively. OMIM = Online Mendelian Inheritance in Man; ZNFs = zinc finger genes.

Figure 5

Hypermethylation and downregulation of disease-associated genes by MORC2 gain-of-function variants. (A) LocusZoom plot for ERCC8 and NDUFAF2 genes based on the differential methylation analysis (from Fig. 2A). Each dot represents a CpG site, and the y-axis indicates the −log₁₀(P-value) of the differential methylation analysis. (B) Promoter methylation and expression levels of ERCC8 and NDUFAF2 in MORC2 patients and controls. Left: Mean methylation (y-axis) represents the average of the methylation level (beta value) across 14 CpGs located in the ERCC8-NDUFAF2 promoter region. Right: RNA Z-scores derived from OUTRIDER results for ERCC8 and NDUFAF2 expression levels, respectively. (C) NDUFAF2 protein levels in MORC2 patients and controls. Left: Protein Z-scores of NDUFAF2 derived from PROTRIDER results. Middle: Mitochondrial respiratory chain complex I with significant recurrent protein outliers (protein Z-score < −2) in MORC2 patients highlighted and named in red (one-sided Fisher’s exact test P-value < 0.05). The number of MORC2 patients in whom the protein was identified as an outlier is written in brackets. Proteins shown in grey were either not detected or not significant. Right: Mitochondrial respiration rate in control and MORC2 patients with Charcot–Marie–Tooth disease (CMT) or Leigh syndrome (LS). P-values were computed using an independent two-sample t-test. (D) LocusZoom plot for the FKTN gene based on the differential methylation analysis (from Fig. 2A). Promoter methylation and expression levels of FKTN in MORC2 patients and controls are shown, respectively. The mean methylation represents the average of the methylation level (beta value) across 7 CpG sites located in the FKTN promoter region. (E) LocusZoom plot for SDHAF4 gene-based based on the differential methylation analysis (from Fig. 2A). The promoter methylation level of SDHAF4 in MORC2 patients with measured mitochondrial respiratory chain complex II activity and controls are shown (middle). MORC2 patients with complex II deficiency are indicated by rectangles. The mean methylation represents the average of the methylation level (beta value) across 8 CpGs located in the SDHAF4 promoter. The correlation between the level of promoter methylation and the expression (RNA Z-score) is shown (right). The dashed line in the locusZoom plots represents the P-value threshold of 7.93 × 10⁻⁸, corresponding to a Bonferroni-adjusted P-value < 0.05. MORC2 binding sites and HUSH target sites are shown as green rectangles and double-sided arrows in the locusZoom plots, respectively.

Figure 6

The phenotypic spectrum of MORC2-related disorder can be explained by DNA methylation. (A) Stacked bar plot showing the distribution of HPO terms in our in-house MORC2 patients with clinical diagnoses of Charcot–Marie–Tooth disease (CMT) or mitochondrial disease (MD). The x-axis represents the fraction of patients with a clinical diagnosis of CMT or mitochondrial disease exhibiting specific phenotypic abnormalities. Only HPO terms reported in at least 40% of patients within each category are displayed. (B) Multidimensional scaling plot based on blood DNA methylation of 80 phenotype-specific differentially methylated positions (DMPs). The size of each point reflects the Cockayne syndrome radiological score, with larger points indicating a greater resemblance of patients’ phenotype to Cockayne syndrome. The dataset includes MORC2 patients with CMT phenotype (n = 26), MORC2 patients with mitochondrial disease (n = 8) and controls (n = 58). (C) The methylation level of ERCC8 and NDUFAF2 promoter in controls and MORC2 patients with a clinical diagnosis of CMT or mitochondrial disease in blood and fibroblast. The y-axis represents the average methylation level (beta value) of 14 CpGs mapped to the ERCC8-NDUFAF2 promoter. P-values were computed using an independent two-sample t-test. Patients with Leigh syndrome are included in the category of mitochondrial disease. The variants associated with different phenotypes are labelled in B and C. (D) Scaled density plots of the mean DNA methylation Z-scores of hypermethylated CpGs identified in the EWAS, selected as MORC2 episignature, identified as phenotype specific and located in the ERCC8-NDUFAF2 promoter. The dataset includes blood or fibroblast DNA methylation of 466 healthy or diseased controls, 27 MORC2 patients with CMT and 19 MORC2 patients with mitochondrial disease. P-values were computed to compare the level of DNA methylation between the MORC2 patients with CMT and mitochondrial disease using an independent two-sample t-test. EWAS = epigenome-wide association study; HPO = Human Phenotype Ontology.