Nucleolar FRG2 lncRNAs inhibit rRNA transcription and cytoplasmic translation, linking FSHD to dysregulation of muscle-specific protein synthesis

Abstract

Facioscapulohumeral muscular dystrophy (FSHD) is a hereditary myopathy linked to deletions of the tandemly arrayed D4Z4 macrosatellite at human chromosome 4q35. These deletions cause local chromatin changes and anomalous expression of nearby transcripts FRG2A, DBET, and D4Z4. We discovered that FRG2A is part of a family of long noncoding RNAs (lncRNAs) expressed in skeletal muscle cells, with levels varying among patients. FRG2A localizes in the nucleolus and associates with repetitive DNA at ribosomal DNA (rDNA) loci and centromeres. Elevated FRG2A expression in FSHD cells alters the three-dimensional architecture of heterochromatin at the nucleolar periphery and reduces rDNA transcription and translation rates, resulting in decreased synthesis of skeletal muscle proteins. We also show that myoblasts from FSHD patients display reduced synthesis of skeletal muscle proteins during differentiation. Our results support a disease model in which nucleolar accumulation of D4Z4-driven lncRNA impairs protein synthesis and contributes to muscle wasting.

Graphical Abstract

Figure 1.

The human genome encodes 14 FRG2 paralogs. (A) Chromosome ideograms showing the location of FRG2 paralogs and percentage identity to FRG2A on chromosome 4. The different colors indicate FRG2 paralogs that are complete (2918 bp) with coding potential, (FRG2A, FRG2B, FRG2C,andFRG2C-1), complete without coding potential, (FRG2FPandFRG2EP) or partial and noncoding, (FRG2KP, FRG2DP, FRG2IP, FRG2GP, FRG2HP, FRG2JP, FRG2LP,andFRG2MP) according to T2T-CHM13 annotation. (B) Genomic origin of FRG2s transcripts: NGS of PCR-enriched FRG2 cDNAs obtained from HPFs and human muscular biopsies (HMBs) from FSHD subjects and healthy relatives (four families). The analysis demonstrated muscle-specific expression of FRG2A and its overexpression in FHSD HMBs compared to controls. The statistical significance was assessed using a paired t-test, comparing the mean number of reads in controls (n = 4) and FHSD (n = 4) HMBs (P = 0.042). (C) RNA fractionation in primary myoblasts (HPMs) from CTRL and FSHD subjects showed protein-coding FRG2 transcripts enriched in the chromatin-associated fraction of RNA. GAPDH, NEAT1 [77], and TERRA [45] were used as positive controls for cytoplasmic, nuclear, and chromatin-associated RNA enrichment, respectively. (D) ChRIP in CTRL and FSHD HPMs showed FRG2s transcripts enrichment with heterochromatin marks H3K9me3 and H3K27me3. NEAT1, associated with H3K4me3-marked euchromatin, was used as a negative control [77].

Figure 2.

FRG2A/B are chromatin-associated RNAs preferentially enriched at rDNA and centromeric loci. (A) Schematic diagram of the ChIRP workflow. (B) Pie chart showing relative representation of various genomic regions obtained by the FRG2A/B-t ChIRP-DNA sequencing. A total of 2313 genomic target sites were found, with the enrichment of highly repetitive elements clustered into two main regions: the centromeric α-satellites (47.4%) and the ribosomal DNA arrays (38.7%). Within rDNA arrays, FRG2A/B-t was found to be predominantly associated with the IGS region (82.8%) and the ITS2 (17.2%), as illustrated in the schematic representation of the rDNA unit shown to the right. (C) Chromosome ideograms overlapped with FRG2A/B-t-association track displaying the enrichment of FRG2A/B-t at the centromeres and rDNA arrays. FRG2 genes and rDNA arrays localization are indicated. (D) Pie-chart showing FRG2A/B-t association sites per chromosome, detected via ChIRP-DNA sequencing. Values are in bp and percentages are indicated in the legend. (E) FRG2 paralog transcripts interacting with FRG2A/B-t identified by ChIRP-RNA sequencing. Bar plot (right) shows log₂ fold-change (ChIRP versus input) for each enriched transcript. Transcript IDs follow the CAT LiftOff annotation based on the T2T-CHM13 reference genome, and the numeric suffix (e.g. -201) indicates specific transcript isoforms derived from alternative splicing or promoter usage.

Figure 3.

FRG2A/B RNA defines a previously undescribed nucleolar domain. (A) RT-qPCR of RNA from HeLa cell fractionation confirmed FRG2 paralogs enrichment in the nucleolar fraction, while protein-coding transcripts (GAPDH and FRG1) showed no enrichment (two-way ANOVA, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). (B) RT-qPCR on DNA fractions from HeLa cell fractionation validated assay success. The nucleolar target (rDNA region IGS28) was enriched in the nucleolar fraction, whereas the non-nucleolar region (FRG2A promoter) showed no enrichment (two-way ANOVA, *P< 0.05). (C–N) RNA-FISH with FRG2A/B-specific fluorescent LNA probe (red) in HPMs from CTRL and FSHD subjects. Nucleolar segregation was induced in HPMs (I–N) by actinomycin D (ActD, 1 μg/ml, 4 h); scale bar: 5 μm. (O–I’) Immuno-RNA-FISH with anti-UBF, anti-FBL, and FRG2A/B-specific LNA probe in FSHD-HPMs, untreated (O–U) or ActD-treated (1 μg/ml) for the indicated times (V–I’). Panels P–U, W–B’, and D’–I’ show 3D renderings of respective confocal images (O, V, and C’). Dotted circles indicate nuclei, white squares highlight zoomed regions. (J’–M’) STED confocal images of immuno-RNA-FISH with anti-FBL (blue) and FRG2A/B LNA probe (red) in untreated CTRL and FSHD-HPMs. Panels K’ and M’ display 3D renderings of images (J’ and L’). Circles indicate nuclei, white squares mark zoomed regions. (N’) Schematic representation of nucleolar structure in untreated and ActD-treated cells, showing FRG2A/B-t localization and Froggy2 Bodies formation.

Figure 4.

FRG2A/B-t mediates centromere association with the NP in FSHD cells. (A) Pie chart quantifying FRG2A/B-t-associated regions overlapping with NADs in HeLa cells (from existing data). Statistical analysis in Supplementary Fig. S13. (B) The plot shows the total number of base pairs (bp) associated with FRG2A/B-t across each chromosome, along with the subset overlapping with NADs (FRG2A/B-NADs). (C–J) Immunofluorescence with anti-FBL (magenta) and anti-CENP-B (green) in CTRL and FSHD HPMs showed increased nucleolar-centromere association in FSHD. (K) Quantification of nucleolus-associated centromeres per cell (from C to J), defined as CENP-B foci colocalizing with FBL or within 140 nm from FBL. Two families (CTRL and FSHD) were analyzed using ScanR software. Mann–Whitney test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P< 0.0001). (L–E’) Immuno-RNA-FISH with anti-FBL (magenta), anti-CENP-B (green), and FRG2A/B probe (red) in FSHD HPMs after FRG2A/B siRNA silencing (siFRG2A/B#1, #2, or both). Scrambled siRNA (siCTRL) was used as control. FRG2A/B signal was reduced (R, W, B’), and centromere localization was restored (T, Y, D’). (F’) FRG2A/B mean intensity quantification (from L to E’) confirmed signal reduction (two-way ANOVA, *P < 0.05, ***P < 0.001, ****P< 0.0001). (G’) Quantification of nucleolus-associated centromeres in FRG2A/B-silenced cells showed a significant reduction (two-way ANOVA, *P < 0.05, ***P < 0.001, ****P< 0.0001). (H’–M’) Immuno-RNA-FISH with anti-FBL (magenta) and anti-CENP-B (green), and with FRG2A/B-specific probe (red), in nucleoli purified from HeLa cells. Panels I’–L’ show 3D rendering of (H’ and J’), and (M’) shows a rotated 3D view.

Figure 5.

FRG2A/B-t inhibits rDNA transcription in FSHD myoblasts. (A) RT-qPCR of HPMs cDNA of CTRL and FSHD subjects. Statistical significance was tested by two-way ANOVA analysis. P-value (*P < 0.05, **P < 0.01, ***P < 0.001, ****P< 0.0001). (B) Chromatin-immunoprecipitation (ChIP) performed in CTRL and FSHD-derived HPMs showing H3K9me3 enrichment at IGS and rDNA promoter. Data from two CTRLs and two FSHD patients were averaged (n = 2). FRG2A/B-associated loci are indicated by the pink box (see Supplementary Fig. S8). Statistical significance was tested via two-way ANOVA analysis. (C) RT-qPCR of FSHD HPMs cDNA upon FRG2A/B siRNA-mediated silencing. Statistical significance was tested by using the two-way ANOVA analysis. P-value (*P < 0.05, **P < 0.01, ***P < 0.001, ****P< 0.0001). (D) Via EU incorporation, rRNA transcription was quantified as the ratio between nucleolar/nuclear EU fluorescent signals. Statistical significance was tested by using the Mann–Whitney test. (EandF) RT-qPCR of nascent RNAs captured by EU incorporation in CTRL and FSHD HPMs (E), and upon siFRG2A/B treatments of FSHD-HPMs (F). Statistical analysis was performed by using the two-way ANOVA test. (GandH) The number of nucleoli per cell and the total nucleolar area per cell were measured in FSHD and CTRL HPMs. Statistical significance was tested by using the Mann–Whitney test. Each FSHD sample was compared to a healthy control selected from a related family member (matched control), chosen based on availability and degree of relatedness. This family-based design ensures consistency across comparisons.

Figure 6.

Mass spectrometry revealed impaired synthesis of ribosomal and muscle-specific proteins in FSHD. (AandB) Quantification of nascent proteins in CTRL and FSHD HPMs at basal conditions (A) and after FRG2A/B silencing in FSHD-HPMs (B). HPG incorportion in newly translated proteins was quantified as cytoplasmic HPG mean intensity. Statistical significance was tested with the Mann–Whitney test **P < 0.01, ***P < 0.001, ****P< 0.0001). (CandD) Mass spectrometry analysis in CTRL and FSHD HPMs and myotubes. Histograms show the most significant misregulated GO molecular function terms (−log₁₀ P-value > 5, by using DAVID annotation [49]). Upregulated and downregulated proteins in FSHD are in purple and pink, respectively. (C) Volcano plot comparing FSHD versus CTRL HPMs. (D) Volcano plot of the longitudinal analysis: DEPs identified in myoblast-to-myotube transition in FSHD and controls were used to compare conditions by calculating the delta log₂FC. (E) A hypothesis for the contribution of FRG2 lncRNAs to FSHD pathogenesis. In individuals carrying a normal-sized D4Z4 allele at 4q35, FRG2A is expressed at low levels, and has no effects on genome architecture or rDNA arrays, allowing normal muscle growth and function. In subjects carrying a reduced D4Z4 allele (≤10 repeats), FRG2A is overexpressed, altering nucleolar chromatin interactions and impairing rDNA transcription, reducing protein synthesis and potentially contributing to muscle atrophy over time.