Decoding the muscle transcriptome of patients with late-onset Pompe disease reveals markers of disease progression

Abstract

Late-onset Pompe disease (LOPD) is a rare genetic disorder caused by the deficiency of acid alpha-glucosidase leading to progressive cellular dysfunction owing to the accumulation of glycogen in the lysosome. The mechanism of relentless muscle damage (a classic manifestation of the disease) has been studied extensively by analysing the whole-muscle tissue; however, little, if anything, is known about transcriptional heterogeneity among nuclei within the multinucleated skeletal muscle cells. This is the first report of application of single-nucleus RNA sequencing to uncover changes in the gene expression profile in muscle biopsies from eight patients with LOPD and four muscle samples from age- and sex-matched healthy controls. We matched these changes with histological findings using GeoMx spatial transcriptomics to compare the transcriptome of control myofibres from healthy individuals with non-vacuolated (histologically unaffected) and vacuolated (histologically affected) myofibres of LODP patients. We observed an increase in the proportion of slow and regenerative muscle fibres and macrophages in LOPD muscles. The expression of the genes involved in glycolysis was reduced, whereas the expression of the genes involved in the metabolism of lipids and amino acids was increased in non-vacuolated fibres, indicating early metabolic abnormalities. Additionally, we detected upregulation of autophagy genes and downregulation of the genes involved in ribosomal and mitochondrial function leading to defective oxidative phosphorylation. Upregulation of genes associated with inflammation, apoptosis and muscle regeneration was observed only in vacuolated fibres. Notably, enzyme replacement therapy (the only available therapy for the disease) showed a tendency to restore dysregulated metabolism, particularly within slow fibres. A combination of single-nucleus RNA sequencing and spatial transcriptomics revealed the landscape of the normal and diseased muscle and highlighted the early abnormalities associated with disease progression. Thus, the application of these two new cutting-edge technologies provided insight into the molecular pathophysiology of muscle damage in LOPD and identified potential avenues for therapeutic intervention.

Keywords: glycogen storage disease; metabolism; mitochondria abnormality.

Figure 1

Repartition of cell types in control and Pompe muscle samples. (A) UMAP visualization of all the nuclei from control and Pompe samples coloured by cell identity. Control, n = 4; Pompe, n = 8. (B) Violin plots showing the expression of selected marker genes for each cluster of nuclei. (C) Percentage of nuclei for each cell population between control and Pompe samples. Control, n = 4; Pompe, n = 8. Mann–Whitney U-test; *P < 0.05 and **P < 0.01. (D) UMAP showing clusters identified in control (left) and Pompe (right) samples coloured by cell identity. Control, n = 4; Pompe, n = 8. (E) Principal component analysis showing the distribution of individuals based on the proportion each population of myonuclei. (F) Left: UMAP showing only the myofibre clusters, before assigning identity in A, in control and Pompe disease samples. Clusters 5 and 8 (fast and slow fibres, respectively) appear in Pompe samples. Right: Metascape analysis of the most expressed genes in Clusters 5 and 8. Only genes upregulated with a log₂ fold change of >0.5 were analysed. Control, n = 4; Pompe, n = 8. FAPs = fibro-adipogenic progenitor cells; UMAP = uniform manifold approximation and projection.

Figure 2

Cell–cell communication of satellite cells using CellChat. (A) Heat map of the differential number of interactions between cell types in Pompe disease versus controls. Bars indicate outgoing signalling from each cell, and columns indicate incoming signals. Red squares indicate increased signalling in Pompe, and blue squares indicate increased signalling in controls. (B) Circle plots showing the differential number of ligand–receptor interactions between pairs of cell populations in Pompe disease versus controls. The strength of L-R interactions between cell population pairs is visualized, with the number of L-R pairs labelled, and edge width is proportional to the number of L-R pairs. Red indicates increased signalling in Pompe, and blue indicates increased signalling in controls. (C) Chord diagram showing signals received by satellite cells from muscle fibres, regenerative fibres and macrophages in control and Pompe muscles. (D) Heat map of outgoing or incoming communications of satellite cells in control and Pompe muscles. The coloured bar represents the relative signalling strength. C = Control; FAPs = fibro-adipogenic progenitor cells; L-R = ligand–receptor; P = Pompe.

Figure 3

Mitochondrial dysfunction in Pompe muscle samples. (A) Volcano plot of log₂ fold change and −log₁₀(P) of Pompe genes compared with control. Control, n = 4; Pompe, n = 8. (B) Representative images of combined cytochrome c oxidase and succinate dehydrogenase (SDH/COX) staining, markers of mitochondrial dysfunction, from control and Pompe muscle samples, with COX-negative fibres stained in blue (×20 magnification). (C) GSEA plots showing enrichment score (ES) of the significant enriched hallmark gene sets in myonuclei. A positive value of ES indicates enriched in Pompe disease, and a negative value indicates enriched in controls. Control, n = 4; Pompe, n = 8. FDR = false discovery rate; GSEA = gene set enrichment analysis; LOPD = late-onset Pompe disease; NES = normalized enrichment score.

Figure 4

Metabolic changes observed between control and Pompe muscle samples. (A) Box plot of different metabolic scores in slow and fast fibres between control and Pompe muscle samples. Control, n = 4; Pompe, n = 8. Wilcoxon test; *P < 0.05 and ****P < 0.0001. (B) Heat map displaying Z-scores of metabolic pathway activity in slow and fast fibres from control and Pompe muscle samples. Each row represents a specific metabolic pathway. The colour scale reflects the level of pathway activity, with warmer colours indicating higher activity and cooler colours denoting lower activity. Control, n = 4; Pompe, n = 8. (C) Oil Red staining of Pompe disease muscle; arrows indicate lipid accumulation. UMAP = uniform manifold approximation and projection.

Figure 5

Autophagic changes observed between control and Pompe muscle samples. (A) Box plot of autophagy-related pathway score in slow and fast fibres between control and Pompe muscle samples. Control, n = 4; Pompe, n = 8. Wilcoxon test; *P < 0.05 and ***P < 0.0001. (B) Violin plot showing the expression levels of several genes involved in different stages of autophagy. Each point represents a nucleus. Control, n = 4; Pompe, n = 8. The P-values are from the Wilcoxon test.

Figure 6

High-resolution spatial transcriptomics application in late-onset Pompe disease. (A) Representative images of region of interest of control, Pompe non-vacuolated (NV) and Pompe vacuolated (V) fibres. Fibre contour and autophagic vacuoles were stained with anti-laminin (green) and anti-LC3B (red) antibodies, respectively. Scale bars = 100 μm. (B) Volcano plots of log₂ fold change and log₁₀P-value of Pompe non-vacuolated (NV) genes compared with control (left); and volcano plots of log₂ fold change and log₁₀P-value of Pompe vacuolated (V) genes compared with Pompe NV (right). (C) Metascape results for enriched pathways in upregulated genes for Pompe NV versus control and for Pompe V versus Pompe NV. Mann–Whitney U-test. Control, n = 4; Pompe NV, n = 35; Pompe V, n = 32.