A decline in PABPN1 induces progressive muscle weakness in Oculopharyngeal muscle dystrophy and in muscle aging

Transcriptome changes in OPMD are associated with initiation of symptoms

Symptoms in OPMD often initiate with eye lid muscle weakness (ptosis) or and swallowing difficulties (dysphagia) [7], but muscle weakness in upper limbs was also reported [21]. In our patient cohort limb muscle weakness was often highly abundant (Table 1). Muscle strength in 33 genetically confirmed OPMD patients (Supplementary Table 1) was determined with the Medical Research Council (MRC) scale, where MRC <5 represents clinical symptoms. An association between muscle strength and age was statistically assessed from 11 muscle groups, from which significant association was found in 8 muscle groups (Table 1). Using the same muscle strength measures, in carriers of expPABPN1 under 42 years-old muscle weakness was not observed (Supplementary Table 1). This suggests that age is major confounder in muscle weakness in OPMD.

In OPMD VL muscles we identified genome-wide changes in gene expression [14]. Here we assessed whether these changes in OPMD are associated with with the expression of expPABPN1 and/or with age. We first compared the transcriptome of OPMD patients with that of expPABPN1 carriers at a pre-symptomatic stage. Expression profiles for each group were generated after subtraction to an age-matched control group: for the pre-symptomatic group mean age was 36 ± 3.5 N=6 and 9 (carriers and controls, respectively) and in the symptomatic group mean age was 56 ± 5 years N=9 and 13 (carriers and controls, respectively) (Supplementary Table 1). The genes found to be affected in OPMD (N=1572 genes; p-value < 0.01) were clustered into 73 functional Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathways and were highly similar to those identified in OPMD animal models (Supplementary Table 2). With the same p-value cut-off only 473 genes were identified in the pre-symptomatic group and those did not enrich in any KEGG pathway. This suggests that gene deregulation in pre-symptomatic have little functional impact. After applying the false discovery rate (FDR)-correction, 149 genes were found to be deregulated in the symptomatic group while in the pre-symptomatic group no gene passed those stringent criteria. This analysis thus shows that in OPMD muscle transcriptome changes are associated with symptoms, while changes caused by the presence of expPABPN1 are only minor with limited functional impact. Although more confidence of gene deregulation is obtained after FDR correction, since our dataset is relatively small we selected the deregulated genes with p-value 0.01 for further analysis.

Expression profiles in OPMD muscles and in muscles of elderly are highly similar

We first investigated the possibility of molecular similarities between OPMD muscle and muscle aging using functional groups. The affected KEGG pathways in OPMD were highly similar to those found in an independent study of normal muscle aging [22]. From the 75 KEGG pathways that are significantly affected in this muscle aging study only 3 (96% similarity) were not identified in OPMD (Supplementary Table 2).

If molecular changes in OPMD are similar to these in muscle aging similarity in dysregulated genes and expression trends is expected. We compared OPMD expression profiles (mean age 56 ± 5) with those of elderly (>85 years) (Supplementary Table 1). It is relevant to note that OPMD patients were under 70 years-old and therefore may not considered as elderly. VL muscles from healthy controls were sampled same as the OPMD biopsies and expression profiles were generated on the same platform. Expression profiles in OPMD and elderly were generated with the same control group (mean age 56 ± 5). Close to one third of the genes affected in elderly overlapped with those affected in OPMD (Figure 1A), and the same directionality in fold-change was found for 84% of these genes. The majority of the overlapping genes (77%) were down regulated (Figure 1A). This observation is consistent with other studies that found a prominent transcriptional down-regulation in neurons from the elderly [23]. Gene Ontology (GO) analysis of OPMD- or elderly- regulated genes also revealed high similarities between significantly affected cellular and molecular functions (Table 2).

Figure 1. Similar expression profiles in OPMD and elderly muscles.

(A) Venn diagram shows the number of affected genes (p ≤0.01) in OPMD, elderly overlapping genes. The percentage of similar gene direction and the direction (up or down) are listed. For each group the mean age ± standard deviation are denoted. (B) Cumulative distribution function (CDF) plots show the distribution of normalized literature association-weights of commonly deregulated genes between OPMD and elderly with the concepts: Aging, Muscle contraction, Oxidative phospho-rylation, insulin signalling, TGFβsignalling, and the ubiquitin-proteasome system (UPS). Arrowheads indicate the maximum association weights. The further the curve shifts to the right, the higher the associations of the affected genes with the indicated concepts are. (C) Bar-chart shows fold-change of selected known age-regulated genes in elderly and in OPMD. P-values are indicated: * P≤0.01, ** P<0.005, *** P<0.0005.

We then investigated whether aging- or OPMD-regulated genes are correspondingly enriched in recognized aging concepts. We employed the literature association method [24] with the concepts: aging, muscle contraction, oxidative phosphorylation, insulin signalling, tumour growth factor (TGFβ) signalling and the ubiquitin proteasome system (UPS). The literature association of every gene with each of the 6 concepts was determined and normalized association-weights were assigned. Association enrichment for the significantly regulated genes in each concept was evaluated with the Cumulative Distribution Function (CDF). The distribution of association-weights for the aging concept had a maximum score indicating that the common genes between OPMD and elderly indeed represent well-known aging genes (Figure 1B). As expected, significant scores were also found for oxidative phosphorylation and muscle contraction. Next to the aging concept, the UPS also rated with a maximum score, suggesting that UPS genes have highest impact for both OPMD and elderly. The high significant score indicates that the association is not random (P = 4.3×10⁻³⁹, 8.1×10⁻²⁵ and 2.4×10⁻²⁶, for UPS, oxidative phosphorylation and muscle contraction respectively). In contrast, the distributions of weight-associations for genes in the insulin or TGF-β signalling pathways were not strong (Figure 1B), and P-values did not differ from a theoretical random distribution. This analysis suggests that similar genes involved in the UPS, mitochondria and muscle contraction affect muscle weakness in both elderly and in OPMD. The genes affecting the insulin or TGF-β signalling pathways, however, differ between OPMD and elderly. To visualize the molecular similarity between OPMD and elderly we plotted the mean fold change for well-known aging-regulated genes [25] (Figure 1C). Together our analysis revealed a significant consistencyin molecular signatures that characterize VL muscles in OPMD patients and in elderly.

Transcriptome of an OPMD mouse model clusters with muscle aging transcriptome

We further studied the association of the OPMD transcriptome with aging using the transcriptome of the A17.1 mouse at 6-weeks old, as muscle weakness is not obvious at that age (Trollet et al., 2010). In the A17.1 mouse expPABPN1 is overexpressed under a muscle-specific promoter resulting in severe muscle weakness (Davies et al., 2005) and muscle atrophy at an early age [15]. A literature-aided association study of the differentially expressed genes in the A17.1 muscles of 6 weeks-old mice showed that a large subset of these genes strongly associates with the aging concept (Figure 2A). The fold-change for the genes with a high association-weight score was large (Figure 2A), indicating that well-known aging genes are highly dysregulated in this mouse model.

Figure 2. The transcriptome of the OPMD mouse model is highly associated with aging.

(A) Volcano plot shows the distribution of significantly deregulated genes (P = 0.05; indicated with a dashed line) in 6 week-old A17.1 mice against fold-change. Smaller P-value and higher fold-change suggests a higher impact in OPMD. For every gene a literature-associated weight with the ‘Aging' concept is assigned, and a normalized association-weight is presented with a circle on a linear scale between 0.05 and 1, where 1 equals the highest association. (B) Hierarchical clustering arrangements of 104 datasets in a literature-aided meta-analysis. Shades of blue indicate degree of similarities: from weak (white) to strong (dark blue). Three skeletal muscle aging-related datasets cluster together with the OPMD dataset of 6 week-old mice (highlighted in red). The clusters associated with muscular dystrophies and other myopathies are highlighted in green and blue, respectively.

For an unbiased approach we performed an unsupervised meta-analysis with 104 microarray studies of muscle development and muscle disorders [26], including the transcriptome from muscle of 6-weeks old A17.1 mice. With this analysis similar transcriptional changes were grouped into three main clusters (Figure 2B). The transcriptome of the 6 week-old A17.1 mouse strictly grouped together with those related to skeletal muscle aging [22, 27], but not with datasets from other muscular dystrophies or myopathies (Figure 2B). This suggests that the high overexpression of expPABPN1 in the OPMD mouse model could induce similar molecular changes as in normal muscle aging.

Expression trends are accelerated in OPMD

Close to 30 years separate the OPMD patients in our study from the elderly, but the remarkable overlap in expression profiles suggests that changes in OPMD initiate earlier or progress faster compared with normal aging. To identify expression trends k-means clustering was applied on the overlapping probes between OPMD and elderly. We compared trends per probe in a dataset derived from the healthy cohort age 17-89 years to those derived from expPABPN1 carriers age 31-74, and subsequently, unique genes were identified using Entrez ID. Two clusters of significant age-regulated genes showing accelerated (N=35) or decelerated (N=39) trends in OPMD were identified (Figure 3A, Supplementary Table 3). Examples of trends for selected genes are presented in Figure 3B. Among those are well-known regulators of aging such as the cell cycle regulators CDKN1A (p21) and OSBP, regulators of muscle cells and energy metabolism like LMOD1 and CHRNA1 or FAT1 and PRODH. These genes with accelerated expression changes in OPMD could contribute to disease progression and muscle weakness.

Figure 3. Age regulated gene expression trends change faster in OPMD.

(A) Expression trends in controls (grey) and expPABPN1 carriers (red). Similar expression trends were identified with k-means clustering using probe ID. (B) Examples of expression trends of 8 genes from clusters in A, in healthy controls (grey) or in exPABPN1 carriers at pre-symptomatic and symptomatic stages (red). Pre-symptomatic and symptomatic stages of OPMD are indicated.

PABPN1 levels significantly decreases with age and in OPMD

The mRNA processing and spliceosome GO-terms were the most significantly affected in both OPMD and elderly datasets (Table 2). Noticeably, the majority of the dysregulated genes in this category were down regulated (Figure 4A). Network analysis suggests that central hubs are affected in both OPMD and elderly, including PABPN1 (Figure 4B).

Figure 4. Deregulation of the Spliceosome in both OPMD and elderly.

(A) Bar-chart shows fold-change of spliceosome genes in OPMD and elderly from the transcriptome studies. P-values are indicated: * P<0.05, ** P<0.005, *** P<0.0005. In bold are gene-hubs in the OPMD-affected spliceosome network (B). A schematic gene network presentation of OPMD-regulated genes that are grouped in the spliceosome GO category. PABPN1 is highlighted

Since mutations in PABPN1 are the genetic cause for OPMD, we next validated the decline in PABPN1 mRNA levels using RT-qPCR using an extended cohort of skeletal muscle biopsies. A significant and pronounced decrease in expression was found in OPMD compared with age-matching controls, while at the pre-symptomatic stage PABPN1 levels did not significantly change (Figure 5A). This suggests that symptoms in OPMD patients are associated with a decrease in PABPN1 expression.

Figure 5. RT-qPCR analysis of PABPN1 expression trends in OPMD and during muscle aging.

(A) Box plot shows PABPN1expression in Vastus lateralis from expPABPN1 carriers at a pre-symptomatic (pre-symp) or symptomatic (OPMD) stages, and age-matching control groups. (B) Box plot showsPABPN1expression in blood form OPMD patients and controls. (C) Scatter plot shows PABPN1expression in 78 healthy controls age 17-89 years. Male and female samples are indicated in black and grey, respectively. A quadratic fit is shown with a black line (a), and linear fits are for the age groups: 17-42 years (b) or 43-89 years (c) are denoted in grey. The table summarizes p-values and Beta ± standard errors, which were calculated after gender correction. Fold-changes were calculated after normalization of GUSB housekeeping gene and to control group age 17-22 years (A, C) or age matching controls (B). N denotes the number of samples in each group.

Next, levels of PABPN1 mRNA during aging were determined in VL from 74 healthy controls age 17-89 using RT-qPCR. Fold-change was calculated from GUSB housekeeping gene and young control group at age 17-22 years. A significant decrease in PABPN1 expression with age was observed with a quadratic model (Figure 5C). Using two sequential linear models the decline in this dataset was found from 43 years onwards, whereas between 17-42 years the change in expression levels was insignificant (Figure 5C). Together, our analysis reveals that PABPN1 expression declines with aging from the fourth decade onwards and that this decline is accelerated in OPMD.

PABPN1 is expressed in every cell whilst symptoms in OPMD are predominantly restricted to a subset of skeletal muscles. To investigate whether the decrease in PABPN1 expression can explain the tissue specificity in OPMD, PABPN1 levels in blood samples of OPMD patients and controls was determined using RT-qPCR. PABPN1 levels in blood were unchanged between OPMD patients and age-matching controls (Figure 5B), indicating that the decrease in PABPN1 expression in OPMD is more prominent in skeletal muscles. An age-associated decrease in PABPN1 levels was also not found in cross-sectional microarray studies from blood, parotid glands, kidney cortex and medulla (Table 3). In frontal brain cortex we identified a slight decrease in PABPN1 expression (Table 3), which started only after the age of 70 years (Table 3). Also in Rectus abdominis muscles, where muscle aging is not so prominent [28], PABPN1 expression did not change with age (Table 3). Together, this suggests that the age-associated decrease in PABPN1 expression marks aging of human skeletal muscles that are typically affected in OPMD.

PABPN1 down-regulation (-DR) induces myogenic defects and muscle cell aging

Knockdown of PABPN1 in mouse muscle cells causes myogenic defects [18]. Since in vivo we found a decrease in PABPN1 expression with age in both normal aging and in OPMD, we studied whether at with a moderate down-regulation of PABPN1 myogenic defects can also be observed. Three PABPN1 shRNA clones were selected for functional studies in immortalized human myoblast cultures using the lentivirus expression system for efficient delivery. Stable cultures were generated and down-regulation was evaluated with RT-qPCR and western blot analyses. Compared with controls (H1 empty vector and non- transduced cells), the three PABPN1 shRNA clones, 121, 122 and 123, led to a 70%, 40% and 20% decrease in PABPN1 expression, respectively (Figure 6Ai). This decrease was also reflected in the residual protein levels (Figure 6Aii). Immunofluorescence microscopy with antibody to PABPN1 revealed a decrease in the accumulation of nuclear PABPN1 in the sh121-transduced cells (Figure 6Aiii).

Figure 6. PABPN1-DR in human myotubes causes myogenic defects.

Human myotubes were transduced with shRNA specific to PABPN1 (sh121, sh122, or sh123) or H1 empty vector. Non-transduced (NT) cells were used as controls. (A) i Bar-chart shows PABPN1 mRNA expression in stably-transduced myoblasts. Fold change was normalized to GapDH housekeeping gene and to a non-transduced culture. Averages are of 6 biological replicates. ii Western blot analysis of PABPN1, MHC1 and muscle actin (MSA), as a loading control in sh121, sh122 or H1 myotube cultures. iii Immunofluorescence of PABPN1 (labelled with Alexa-594) and MHC1 (labelled with Alexa-488) in sh121 or H1 myotube cultures. Scale bar 10 μm. (B) Bar chart shows fold change of MHC1, DMD, and CAV3 in 121-, 122-, 123-, and H1- myoblast cultures. Fold change was normalized to GapDH and to a non-transduced culture. Averages are of 3 biological replicates. Significant down-regulation (P<0.05) is indicated with asterisks. (C) i- images of H1 controls and the PABPN1 down regulation sh121 fused cultures. Scale bar is 50 μm. ii- Chart bar shows the percentage of nuclei in fused myotubes that express MHC1 (cell fusion) in H1 sh123 sh122 or sh121 myotube cultures. Averages and SD are from six replicates and the number of nuclei that were quantified per sample is indicated within each bar. Significant effect in PABPN1-DR cultures from control cultures (P<0.05 or P<0.005) is indicated with one or two asterisks, respectively.

Next, we investigated whether PABPN1-DR specifically affect expression levels of myogenic genes. We selected MYH1, DMD and CAV3 for RT-qPCR analysis, as their expression levels significantly decreased in both OPMD and elderly. The strongest decrease in expression of all three genes was found in the sh121 cultures with highest PABPN1-DR (Figure 6B). Only a moderate decrease across all three genes was found in the sh122 culture, and expression levels did not differ from controls in the sh123 culture (Figure 6B). The decrease in MHY1 on mRNA level was confirmed on the protein level (Figure 6A). This suggests that levels of PABPN1-DR can induce a decrease in cell fusion. An apparent decrease in cell fusion was found in the strong PABPN1-DR, sh121, cultures (Figure 6Ci). We then determined rate of cell fusion across PABPN1-DR fused cultures index using a semi-automated and robust imaging procedure, where the percentage of nuclei within cells expressing MHC1 was quantified. The rate of cell fusion was not affected in the mild PABPN1 down-regulated culture, sh123, but in sh122 or sh121 cultures cell fusion significantly reduced (Figure 6Cii). Together the results here reveal that a decrease in PABPN1 expression induces myogenic defects that are proportional to levels of PABPN1-down-regulation.

Aging cells undergo substantial cellular changes [29]. To investigate whether a decrease in PABPN1 expression induces cellular characteristics of aging cells, we applied four different cellular assays. Reduced mitochondrial metabolic rate is often found in aged skeletal muscles [30]. Using JC-1 labeling, we found a decrease in mitochondrial metabolic rate in both myoblast and myotube cultures with PABPN1 down-regulation compared with control cultures (Figure 7Ai). The effect of PABPN1-DR on the mitochondrial metabolic rate was more pronounced and more significant in myotube cultures compared with myoblasts cultures (Figure 7Ai). Importantly, a decrease in mitochondrial metabolic rate was restored in PABPN1-DR cultures after transduction with CFP-PABPN1 lentivirus particles (Figure 7Ai). Noticeably, the decrease in metabolic rate, levels of myogenic genes and fusion index correlated with levels of PABPN1-DR. As levels of mitochondrial metabolic rate are decreased due to aging-regulated increase in oxidative stress [31], next we investigated whether PABPN1 affect cell sensitivity to oxidative stress. Muscle cell cultures were treated with H₂O₂ and mitochondrial metabolic rate was determined after JC-1 labeling. PABPN1-DR muscle cells are 1.5-fold more sensitive to H₂O₂ treatment compared with control cells (Figure 7Aii). In addition, Levels of PABPN1 also affect rates of cell growth. PABPN1-DR cells have a slower growth rate compared with control, and restored after transduction with CFP-PABPN1 lentivirus particles (Figure 7Bi). In PABPN1-DR cells heterochromatic foci, which mark senescent cell [32], were also detected (Figure 7Bii). Last, we also found that PABPN1-DR induces formation of fat droplets in muscle cell cultures (Figure 7C). Importantly, restoring PABPN1 levels with CFP-PABPN1 decreased the proportion of cells with fat droplets (Figure 7Ci). Together, these studies indicate that reduced levels of PABPN1 induce muscle cell aging.

Figure 7. PABPN1-DR induces cellular aging in muscle cells.

(A) PABPN1-down regulation induces a decrease in mitochondrial metabolic rate. (i) Bar chart shows the ratio of J-aggregates to monomers in JC-1-labeled myoblasts or myotube cultures. Cell cultures are of control (7304) or PABPN1-DR (sh122 and sh121) cultures, before (mock) or after transducion with CFP-PABPN1 lentivirus particles. Averages and SD are from six replicates, and the number of cells per sample is indicated within each bar. Significant effect in PABPN1-DR from control cultures (P<0.05 or P<0.005) is indicated with one or two asterisks, respectively. (ii) Plot shows the ratio of J-aggregates to monomers in JC-1-labeled cultures after 1 hour H2O2, in concentrations as indicated in the chart. Averages and SD are from three replicates. (B) A decrease in cell growth is caused by PABPN1-down regulation. (i) Bar chart shows cell growth (24 hours) in control (7304) and PABPN1-DR (sh121) cultures, before (mock) or after transduction with CFP-PABPN1 lentivirus particles. Averages and SD are from three replicates. (ii) Heterochromatic foci are formed in PABPN1-DR cells. Heterochromatic foci are visualized with DAPI (indicated with arrowheads), cells are visualized with Desmin. Scale bar is 20 μm.

Pre-symptomatic carriers, patients and controls and clinical investigation

We invited 33 patients from a cohort of 10 genetically confirmed Dutch OPMD families. All symptomatic and pre-symptomatic carriers were confirmed by genetic analysis. A standard neurological examination including MRC score was performed as described previously in part of this cohort [39]. Muscle biopsies were obtained from 22 patients and 74 healthy controls (Supplementary Table 1). All biopsies are from VL musclesand were collected using the Bergstrom needle procedure. The biopsies froze immediately in liquid nitrogen. Blood was taken from all patients for genetic confirmation. Asymptomatic descendants of these patients were invited to participate without disclosure of their individual histological and genetic results. Eighteen of them underwent a muscle biopsy and genetic testing. Twelve descendants who did not carry the mutation were classified as controls, whereas 6 descendants who did carry the mutation where identified as pre-symptomatic. This study was approved by the local Ethical Committee (CMO nr. 2005/189) and written informed consent was obtained from all patients. Biopsies from healthy controls were collected at Canisius-Wilhelmina Hospital, Nijmegen, The Netherlands.

Microarray and Statistical Analyses

Cell culture and Lentivirus transduction

The human 7304 immortalized myoblasts [44] were propagated in a medium containing DMEM+20% Fetal Calf Serum supplemented with an equal volume Skeletal Muscle Cell Media (PromoCell, Heidelberg, Germany) at 37 °C under 5% CO₂. Cell fusion was carried out in a medium containing DMEM+5% Horse Serum.

The shRNA in lentivirus expression vectors 121 (TRCN0000000121), 122 (TRCN0000000122) and (TRCN0000000123) 123 were obtained from Sigma-Aldrich. An empty vector, H1, was used as a negative control. Lentivirus production and transduction was previously as described [45]. Cells were cultured with viruses (MOI ~25) overnight, followed by medium refreshing. Transduced cells were maintained in the presence of 5μg/ml puromycin, which was omitted for fusion experiments. PABPN1 down-regulation was determined using RT-qPCR. Six independent transduction experiments were performed. Cell fusion was carried out in triplicates and cell fusion index was determined by dividing the number of nuclei in myotubes to the total number of myotubes.

The lentiviruses, CFP-PABPN1 and H4-CFP, were previously described in ([13, 45], respectively). Transduction of each virus was performed in control (7304), sh121 or sh122 cultures. Only cultures with transduction efficiency >70% were included in the studies presented here.

RNA Analysis

All human biopsies were stored at −80 before RNA extraction. RNA extraction and RT-qPCR were performed as described in [14]. RNA extraction from culture cells was performed with NucleoSpin^® RNA II, MACHEREY-NAGEL, according to the manufacturer protocol and RNA was directly subjected for RT-qPCR using Cyber Green, Invitrogen. Expression levels were calculated according to the DDCT method, and were first normalized to GUSB housekeeping gene for human samples and GAPDH for cell culture. Fold change was calculated from controls. The statistical significance was determined with the Student's t-test. The list of primers used in this study is provided in Supplementary Table 4.

Western blot, immunofluorescence, JC-1 and Nile-red staining and image quantification