Nuclei clusters show differential expression of DUX4 target genes detected by MERFISH. (A) Two-dimensional projections (UMAP) of gene expression profiles of all identified nuclei. (Left) UMAPs overlaid with the corresponding cluster identities identified by scrattch.hicat algorithm; (right) proportion of each cluster in each genotype. (B) Heatmap illustrating DUX4 target gene expression level of the nuclei clusters. The expression profile for each nucleus cluster is the averaged, area-normalized, log-transformed expression across different samples and genotypes. (C) Volcano plot illustrates the differential expression between the two identified nuclei clusters across all samples. About 88% of DUX4 target genes show upregulation in cluster 1 nuclei, whereas one DUX target gene (ZNF596) shows upregulation in cluster 0 nuclei (Log2FC threshold: 0.5; P-value threshold: 0.001). (D) Examples of DAPI and cell boundary staining overlaid with nuclei, colored with their corresponding cluster identities. Annotated myotubes are circled in cyan. (E) Total number of intra-myotube and intra-MNC cluster 1 nuclei for each FSHD and DEL5 sample. No significant difference was noted in counts between intra-myotube and intra-MNC cluster 1 nuclei (intra-myotube: 551.0 ± 240.7; intra-MNC: 206.0 ± 66.1; mean ± SE; P = 0.3939, Mann–Whitney U test; N = 6 samples). (F) Proportion of intra-myotube and intra-MNC nuclei to be cluster 1 for each FSHD and DEL5 sample. Intra-myotube nuclei contain a higher percentage of cluster 1 nuclei compared with intra-MNC nuclei for all samples (intra-myotube: 0.3570 ± 0.1130; intra-MNC: 0.0316 ± 0.0098; mean ± SE; P = 0.0087, Mann–Whitney U test; N = 6 samples). (G) UMAP of gene expression profiles of intra-myotube nuclei. UMAPs are overlaid with the clusters identified by scrattch.hicat algorithm. (H) Distribution of intra-myotube nuclei counts for different genotypes. Overall, control myotubes show a significantly higher nuclei count compared with FSHD, whereas DEL5 myotubes do not show significantly different nuclei counts compared with FSHD (370 FSHD myotubes: 15.47 ± 0.60; 213 DEL5 myotubes: 24.60 ± 1.29; 404 control myotubes: 27.42 ± 1.07; mean ± SE; control vs. FSHD: P = 0.0092; DEL5 vs. FSHD: P = 0.0996; linear mixed effects model). (I) Distribution of myotube volume by genotype. Overall, myotubes from control and DEL5 samples show higher volume than those in FSHD (370 FSHD myotubes: 11499 ± 381.9 µm2; 213 DEL5 myotubes: 14652 ± 686 µm2; 404 control myotubes: 15768.0 ± 483.9 µm2; mean ± SE; DEL5 vs. FSHD: P = 0. 0174; control vs. FSHD: P = 2.4655 × 10−6; linear mixed effects model). (J) Comparison of intra-myotube cluster 1 nuclei between FSHD and DEL5 genotype myotubes. Distribution of counts of intra-myotube cluster 1 nuclei per myotube. Only myotubes with cluster 1 nuclei are included. No significant differences are noted between the two genotypes (234 FSHD myotubes: 12.54 ± 0.61; 29 DEL5 myotubes: 11.31 ± 1.38; mean ± SE; FSHD vs. DEL5: P = 0.7145, linear mixed effects model). (K) Proportion of cluster 1 nuclei per myotube. The percentage is calculated as the ratio of cluster 1 nuclei counts to total nuclei counts. Only myotubes with cluster 1 nuclei are included. FSHD myotubes show a significantly higher proportion of cluster 1 intra-myotube nuclei compared with DEL5 (234 FSHD myotubes: 0.7470 ± 0.0150; 29 DEL5 myotubes: 0.5705 ± 0.0535; mean ± SE; FSHD vs. DEL5: P = 0.0011, linear mixed effects model).
