Deacetylation Inhibition Reverses PABPN1-Dependent Muscle Wasting.

Reduced poly(A)-binding protein nuclear 1 (PABPN1) levels cause aging-associated muscle wasting. PABPN1 is a multifunctional regulator of mRNA processing. To elucidate the molecular mechanisms causing PABPN1-mediated muscle wasting, we compared the transcriptome with the proteome in mouse muscles expressing short hairpin RNA to PABPN1 (shPab). We found greater variations in the proteome than in mRNA expression profiles. Protein accumulation in the shPab proteome was concomitant with reduced proteasomal activity. Notably, protein acetylation appeared to be decreased in shPab versus control proteomes (63%). Acetylome profiling in shPab muscles revealed prominent peptide deacetylation associated with elevated sirtuin-1 (SIRT1) deacetylase. We show that SIRT1 mRNA levels are controlled by PABPN1 via alternative polyadenylation site utilization. Most importantly, SIRT1 deacetylase inhibition by sirtinol increased PABPN1 levels and reversed muscle wasting. We suggest that perturbation of a multifactorial regulatory loop involving PABPN1 and SIRT1 plays an imperative role in aging-associated muscle wasting. VIDEO ABSTRACT.

Introduction

Muscle wasting is prominent in many pathologies including metabolic disorders and neuromuscular diseases. Muscle wasting is also associated with aging and is highly prevalent in the elderly (Bowen et al., 2015). The molecular and cellular causes for aging-associated muscle wasting are diverse. Recently, we reported that muscle wasting is induced by reduced levels of poly(A)-binding protein nuclear 1 (PABPN1) (Riaz et al., 2016). PABPN1 levels are normally reduced in skeletal muscles during aging and are also associated with muscle weakness in oculopharyngeal muscular dystrophy (OPMD), a late-onset myopathy (Anvar et al., 2013). An expansion mutation in PABPN1 is the genetic cause for OPMD (Brais et al., 1998). In OPMD, PABPN1 aggregates and hence depletes the levels of the functional protein (Raz et al., 2011). PABPN1 regulates poly(A) tail length, alternative polyadenylation (APA) in the 3′ UTR or in internal regions, and nuclear export of transcripts, which together will affect mRNA expression levels, leading to an altered transcriptome (Banerjee et al., 2013). The mechanisms by which PABPN1 depletion leads to muscle wasting are not fully understood.

Muscle wasting in OPMD mouse models is associated with the ubiquitin proteasome system (UPS) (Anvar et al., 2011a, Anvar et al., 2011b, Trollet et al., 2010). Also, in the mouse muscles with PABPN1 knockdown (shPab) muscle atrophy was found to be associated with altered mRNA levels of UPS genes that regulate muscle atrophy (Riaz et al., 2016). Reduced proteasomal activity in shPab muscles suggests that protein homeostasis is altered in shPab muscles. Protein homeostasis plays a central role in aging-associated muscle wasting (Cohen et al., 2014). Muscle contraction demands mitochondrial activity, and unsurprisingly altered mitochondrial activity was found in OPMD models, which was concomitant with shortened poly(A) tail length and APA in the 3′ UTR of mitochondrial genes (Chartier et al., 2015, Vest et al., 2017). Yet, how altered mRNA expression profiles lead to muscle wasting remains obscure.

Here we applied RNA sequencing (RNA-seq) and mass spectrometry in shPab muscles to identify the molecular and cellular causes for muscle wasting. We show that protein acetylation is reduced in shPab muscles and that deacetylation inhibition restores PABPN1-mediated muscle wasting.

Results

PABPN1 Down-regulation Alters Muscle Histology

We investigated the molecular mechanisms that are associated with PABPN1-driven muscle wasting in mouse tibialis anterior muscles expressing short hairpin RNA (shRNA) to Pabpn1 (shPab). PABPN1 levels in muscles from four mice were compared between shPab and scrambled shRNA (scram) after contralateral injection, as previously described (Riaz et al., 2016). In this experimental setup analysis was paired, overcoming natural variations between mice. Muscles were harvested for ex vivo procedures including RNA-seq and mass spectrometry, and validations were carried out using qRT-PCR, western blot, and muscle histology (Figure 1A). Transduction efficiency was assessed by GFP fluorescence, which was included in the expression cassette. Overall, similar fluorescence was found in shPab and scram muscles (Figure S1), indicating that any alterations in PABPN1 levels are not due to variation in transduction efficiency. Analysis of PABPN1 demonstrated reduced levels in shPab muscles (Figures 1B and 1C). Muscle histology was found to be altered between scram and shPab (Figure 1D). We confirmed thickening of the extracellular matrix (ECM) in shPab muscles (Figure 1D; Riaz et al., 2016). We also measured more myofibers per image frame in shPab compared with scram muscles (Figure 1E). Smaller myofibers could result from muscle atrophy, which is consistent with our previous study (Riaz et al., 2016); furthermore, it can concur with muscle regeneration. Central myonuclei and split myofibers were found in shPab muscles (Figure 1D). The fraction of central myonuclei in shPab was higher in three of the four mice (Figure 1F). PAX7 and eMyh3 expression are molecular signatures of muscle regeneration (Lepper et al., 2011, Sambasivan et al., 2011, Schiaffino et al., 2015). qRT-PCR of eMyh3 mRNA revealed higher levels in shPab muscles (Figure 1G). PAX7 staining also showed the same trend, wherein the fraction of PAX7-positive myonuclei was higher in shPab muscles (Figures 1H and 1I). Noticeably, the mouse with the highest PABPN1 fold change showed the most severe histological changes, whereas the mouse with the lowest fold change showed resilient changes.

Figure 1

Reduced PABPN1 Levels Induce Muscle Regeneration

(A) Schematic workflow of the ex vivo analyses in scram and shPab muscles. RNA expression profiles (RNA-seq) are compared with the shPab proteome of the same muscles. The shPab acetylome was then analyzed. Procedures were validated using qRT-PCR, western blot (WB), or muscle histology. Ex vivo experiments were performed on paired muscles (N = 4 mice).

(B) qRT-PCR of Pabpn1 mRNA levels after normalization to Hprt housekeeping control. Paired dot plot is from N = 4 mice.

(C) PABPN1 protein and levels in paired muscles. Representative western blot of PABPN1 and GAPDH loading control are shown. Paired dot plot shows PABPN1 levels after normalization to loading control, N = 4 mice.

(D) Gomori trichrome tissue histology in cross sections. Images are of the mouse with highest PABPN1 fold change. White arrowheads point to ECM thickening, central myonuclei are depicted with red arrowheads, and split myofibers with black arrowheads. Scale bar, 50 μm.

(E) Paired dot plot shows the mean number of myofibers per image frame, calculated from 5 frames per muscle (N = 8 muscles).

(F) Paired dot plot shows the mean fraction of central nuclei in myofibers, calculated from 5 frames per muscle (N = 8 muscles).

(G) Paired dot plot shows eMyh3 mRNA levels in scram and shPab muscles (N = 4 mice). Expression values were calculated after normalization to Hprt and to the average eMyh3 expression of all scram muscles.

(H and I) (H) Representative fluorescent images for scram and shPab muscles stained with PAX7 antibody (green). Nuclei are counterstained with DAPI (blue). White arrowheads indicate nuclear PAX7. Scale bar, 7.5μm. (I) Paired dot plot shows the fraction of PAX7 positive nuclei in paired muscles. The percentage was calculated from over 1,000 nuclei per muscle (N = 6 muscles). In all dot plots paired muscles are connected with lines; blue or green lines mark the mouse with the highest or lowest PABPN1 fold change, respectively. Quantification of the muscle histology was performed over the entire muscle cross section. Means and standard deviations are depicted. Statistical difference was assessed with a paired test.

Proteomic Changes Surpass the Transcriptomic Changes in shPab Muscles

The molecular mechanisms that are associated with PABPN1-mediated muscle pathology were investigated in the shPab transcriptome and proteome from the same muscles, using paired ratio t test. The shPab RNA expression profiles included 363 genes (5% false discovery rate [FDR]; Figure 2A, related to Table S1), and those were predominantly enriched with genes of the ECM, mitochondria, and muscle-related genes (related to Table S2). These expression profiles are consistent with OPMD expression profiles (Anvar et al., 2011a, Raz et al., 2013). Moreover, higher expression of ECM genes is consistent with ECM thickening in shPab muscles (Riaz et al., 2016). qRT-PCR validation was carried out for Pabpn1 and eMyh3 (Figures 1 and 2A). For the proteome, a 5% FDR cutoff was too strict (Related to Table S3). Considering a p value cutoff of p < 0.05, we identified differential expression of additional genes, such as Rad32a and Map1lc3a, which were previously reported as differentially expressed in shPab (Raz et al., 2017, Riaz et al., 2016). The shPab proteome contained 248 differentially expressed proteins (p < 0.05) (Figure 2B, related to Table S3). Western blot for seven proteins validated the proteomics results (Figure S2). A significant correlation (p = 2.2 × 10−16) was found between mRNA and protein fold changes. The regression line is similar to the diagonal, although the correlation was small (R = 0.31) (Figure 2C). Overall, protein fold change was greater compared with mRNA fold change (Figure S3). Focusing on the group of genes or proteins with higher fold changes (average fold change > |1.5|) we found that the majority of the proteins had a positive fold change in shPab muscles (68%, Figure 2D). In contrast, a fold change direction preference was not found for the mRNAs (Figure 2D). The same fold change direction between mRNA and proteins was found for nearly 70% of the proteins, of which most proteins (74%) have higher levels in shPab (Figure 2E). In the group with opposite fold change direction between mRNA and protein (30%), the majority of proteins (80%) had higher levels in shPab (Figure 2E). Together, this suggests that in shPab muscles protein accumulation surpasses mRNA alterations. Moreover, protein accumulation in shPab suggests higher protein steady state, which is in agreement with reduced proteasome activity in shPab (Riaz et al., 2016).

Figure 2

Reduced PABPN1 Levels Result in a Greater Effect on Proteome Compared with Transcriptome

(A) A volcano plot of the average mRNA fold change (FC) against p values in the shPab transcriptome. mRNAs that pass the cutoff p value < 0.05 or p values corrected for false discovery rate are depicted in blue or red, respectively. Examples of validated genes are highlighted.

(B) Volcano plot of the average protein fold change against p values in the shPab proteome. Proteins that pass the cutoff (p < 0.05) are depicted in blue.

(C) Scatterplot shows the fold changes between shPab proteome and transcriptome. Correlation (indicated as gray dashed line) was assessed with a Spearman rank test and compared with the diagonal (solid gray line). Average protein fold changes of 1.5 and −1.5 are indicated with red lines. Proteins with an average fold changes higher than 1.5 or lower than −1.5 are depicted in pink or black, respectively.

(D) Bar chart shows fold change direction of shPab-affected genes (average fold change cutoff > |1.5|) sorted for RNA or proteins.

(E) Bar chart shows the percentage of proteins (average fold change cutoff > |1.5|) with similar fold change direction or opposite fold change direction compared with RNA direction. Fold change higher than 1.5 is denoted in black and fold change lower than −1.5 in pink. Both transcriptome and proteome were determined from the same muscles (N = 4 mice).

A heatmap of the differential expressed proteins (p < 0.05, N = 248 proteins) across the four individual mice revealed that the largest differences between scram and shPab muscles were found in the mouse with the highest PABPN1 fold change, whereas in the mouse with the lowest PABPN1 fold change the differences were the smallest (Figure 3A). This trend was then confirmed using k-means clustering. For 83 proteins, fold changes highly correlated with PABPN1 levels (Figure S4). Validation using western blot was demonstrated for two proteins (Figure S2). Together, these results suggest that the levels of a subset of proteins strongly correlate with the PABPN1 fold change.

Figure 3

PABPN1 Levels Affect Muscle Proteome of Which a Large Part Is Linked to Acetylation

(A) Heatmap of protein abundance for the differentially expressed proteins (p < 0.05). Muscles are paired; the mice with the highest and lowest PABPN1 fold change are depicted in blue and green, respectively. Western blot shows PABPN1 expression for each mouse. GAPDH was used as loading control. Mitochondrial proteins are highlighted with green lines, and ribosomal proteins are highlighted with red lines. Color key indicates the Z scores (based on rows).

(B) Pie chart shows the significantly enriched protein groups (UP-keyword, David) of the differentially expressed proteins (p < 0.05, n = 248). The transparent pies (outer circle) indicate the number of proteins within the protein groups that are linked to protein acetylation group. Numbers indicate the percentage per group.

(C) The protein-protein interaction map shows protein networks for 248 proteins that pass the cutoff (p < 0.05). Mitochondrial (green), translation (red), cytoskeletal (purple), and glycolysis (pink) clusters are highlighted. Within the mitochondrial cluster, the oxidative phosphorylation group is highlighted in dark green. Unconnected proteins are not shown, and networks with n < 4 are removed. Proteins that are found in the acetylation group are highlighted in yellow, and those that were not found in the acetylation group are gray.

The cellular machineries that are affected by PABPN1 were then identified in the shPab proteome using enrichment analysis. The mitochondria, ribosome, cytoskeleton, nucleotide binding, glycolysis, and muscle proteins were found to be the most prominently affected (Figure 3B). Notably, mitochondrial proteins had lower levels, whereas most of the ribosomal proteins had higher levels in shPab muscles (Figure 3A). Next, protein-protein interaction networks of the differentially expressed proteins revealed that networks of the mitochondria, translation, cytoskeleton, and glycolysis were the most connected (Figure 3C). This suggests that the function of these cellular machineries is impaired in shPab muscles. Most striking, 63% of the proteins in the shPab proteome were enriched in the acetylation group (related to Table S4). Most mitochondrial proteins in the shPab proteome were also found in the acetylation group (Figures 3B and 3C). Moreover, acetylation was also found to be abundant in the other enriched cellular machineries (Figures 3B and 3C), suggesting that acetylation might have a prominent role in shPab muscles.

shPab Muscles Are Characterized by Reduced Protein Acetylation

To elucidate the acetylated protein landscape in shPab muscles, a peptide pull-down with an antibody to acetylate lysine was carried out in scram and shPab muscles. For this purpose, protein extracts from seven mice, which were injected contralaterally with scram or shPab adeno-associated virus (AAV) 9 particles, were pooled and the pull-down peptides were identified using mass spectrometry (Figure 4A). In total, 2,229 peptides passed the 1% FDR and overlapped between two independent runs (related to Table S5). A total of 1,830 peptides (82%) passed the coefficient of variation cutoff and were considered for subsequent analysis (Figure 4B). A confidence in the identified acetylated peptides was further assessed comparing our study in tibialis anterior muscles with the following published acetylome studies: in human and in rat skeletal muscles (Lundby et al., 2012) and in rat soleus (Ryder et al., 2015). Only ∼13% of the acetylated peptides overlapped between the human and rodent acetylomes (related to Table S6A). As the human and rat studies came from the same laboratory and have a similar number of acetylated peptides (Lundby et al., 2012), it can be assumed that the limited overlap is not due to experimental settings or technical variations. The overlap between the two studies in rat was only 35% and comparable to the overlap between our study and the study in rat by Lundby et al. The specific muscle type was not specified in the study by Lundby et al. However, peptide overlap between fast-twitch tibialis anterior and slow-twitch soleus was only 13% (200 peptides; related to Table S6B). In the pool of the overlapping peptides between soleus and tibialis anterior the mitochondria was highly enriched (Table S6C), which is consistent with mitochondria enrichment in the acetylome studies in rat and humans (Lundby et al., 2012). Using western blot, we confirmed that the pattern of bulk protein acetylation highly differs between soleus and fast-twitch quadriceps in mouse (Figure S5). Differences in histone acetylation were reported between fast and slow muscles (Kawano et al., 2015). Moreover, 82% of those overlapping acetylated proteins are also annotated in DAVID as acetylated proteins (related to Table S6C). Together, this suggests that in skeletal muscles the acetylated mitochondrial proteins are highly abundant. Furthermore, it indicates that our acetylome study is suitable for further analysis.

Figure 4

The shPab-Dependent Acetylome

(A) A workflow of the shPab acetylome procedure. Tibialis anterior muscles (N = 7 mice, contralateral injection of scram or shPab AAVs) were used. Protein extracts were pooled per genotype. Tryptic digestion, followed by immunoprecipitation with an anti-acetylated lysine (AcK) antibody, and mass spectrometry (MS) were carried out separately for the scram and shPab. Peptide analysis pipeline included an MS quality control (1% FDR) and coefficient of variation (CV) < 0.3. The differentially acetylated peptides were considered with a 1.5-fold change (fold change) cutoff.

(B) Scatterplot shows average peptide fold change (log2) versus average CV (-log2). Peptides with CV < 0.3 and a fold change of >1.5 and < -1.5 are depicted with blue and red, respectively. Peptides that are only found in either shPab or scram muscle are lined as INF (infinite). Unchanged peptides are depicted in gray. The dashed line marks CV < 0.3.

(C and D) (C) Western blot of acetylated proteins in scram and shPab muscle protein extracts. GAPDH is loading control. (D) Immunofluorescence with AcK antibody in scram and shPab muscle cryosections. Left panels show an overlay between bright-field and DAPI staining of the nuclei (blue). The middle panels show AcK signal (yellow). Right panels show an enlarged AcK and DAPI overlay around the nuclei that are marked with an arrow. Intensity distribution plots for DAPI and AcK signal are made from the nucleus, which is marked with a line. Examples of AcK-positive and AcK-negative nuclei are in the upper and lower plots, respectively. Scale bar, 5 μm.

(E) Bar chart shows the percentage of AcK-positive nuclei in scram and shPab muscles from four mice (unpaired). The number of measured nuclei is depicted within each bar. Trend for statistical difference was assessed with the Mann-Whitney test.

(F) Fold change direction of nuclear proteins in the shPab acetylome: positive and negative fold changes are depicted in blue and red, respectively.

(G) Pie chart shows the significantly affected gene ontology terms in the shPab acetylome.

(H) Schematic representation of the acetylated and the shPab differentially acetylated oxidative phosphorylation subunits. The acetylated subunits in tibialis anterior muscle are highlighted in light green, and the shPab differentially acetylated are in dark green. The number of affected proteins in each complex is depicted.

We then investigated differential peptide acetylation between scram and shPab muscles and considered a fold change of 1.5 as a cutoff. In the shPab acetylome, 270 peptides were hypoacetylated and 137 peptides were hyperacetylated (Figure 4B). We found that 87 and 61 acetylated peptides were exclusively found in scram and shPab muscles, respectively (Figure 4B). This suggests an overall reduced acetylation in shPab muscles. Differences in protein acetylation between scram and shPab muscles were confirmed using western blot (Figure 4C). In addition, immunofluorescence also showed a decrease in the acetylation signal in shPab nuclei compared with scram muscles (Figures 4D and 4E). To confirm this observation, acetylation direction was assessed for nuclear proteins in the shPab acetylome. Hypoacetylation was also prominent in the nuclear proteins, including all histones (Figure 4F). Moreover, enrichment analysis of the differentially acetylated proteins (N = 171) showed that the mitochondria was highly enriched (Figure 4G). In addition, 42% of the differentially acetylated proteins were enriched in metabolic pathways (Figure 4G). The majority of the differentially acetylated oxidative phosphorylation proteins were found to be in complexes 1 and 5 (Figure 4H). These complexes catalyze, respectively, the production of NAD+ and ATP (Lenaz et al., 2006). This could suggest aberrant mitochondrial activity in shPab muscles. In agreement, previously we reported reduced mitochondrial membrane potential in shPab muscle cell culture (Anvar et al., 2013). Acetylation of mitochondrial proteins is achieved by the NAD+-dependent deacetylases (sirtuins), among which SIRT1 is suggested to be central in aging-associated metabolic changes in muscles (Chang and Guarente, 2014).

PABPN1 Regulates SIRT1 mRNA Processing

We then examined SIRT1 levels in shPab muscles and found an elevated protein expression in shPab compared with scram muscles (Figures 5A and 5B). Interestingly, SIRT1 fold changes related to PABPN1 fold changes (Figure 5B). Therefore next we investigated whether SIRT1 is regulated by PABPN1 in an established PABPN1 knockdown muscle cell culture (de Klerk et al., 2012). SIRT1 protein levels were also elevated in shPab cell culture (Figure 5C). Thus we examined whether Sirt1 transcript level is directly regulated by PABPN1. APA was determined with the ratio between two primer sets to distal and proximal region in Sirt1 3′ UTR (Figure 5D). We found a lower distal to proximal ratio in shPab cell culture (Figure 5E), indicating APA utilization. APA often resulted in shorter transcripts that are more stable (de Klerk et al., 2012, Jenal et al., 2012). Using the proximal primer set we found higher Sirt1 mRNA levels in shPab cell culture (Figure 5F). Consistently, also in the OPMD mouse model (A17.1) we found APA and higher levels in Sirt1 mRNA (Figure S6). Interestingly, APA and higher Sirt1 expression were only found in fast-twitch quadriceps, but not in slow-twitch soleus (Figure S6). In this mouse model, soleus is also less affected (Trollet et al., 2010).

Figure 5

Sirt1 mRNA Levels Are PABPN1-Regulated via Alternative PAS in the 3′ UTR

(A and B) SIRT1 levels in muscles of four mice. (A) A representative western blot in scram and shPab muscles. GAPDH is used as a loading control and for normalization. (B) Paired analysis of PABPN1 and SIRT1 fold change. Line connects values with a mouse. Mice with the smallest and highest PABPN1 fold changes are depicted with green and blue lines, respectively.

(C) A representative western blot in muscle cell culture. GAPDH was used as loading control. GAPDH-normalized values are plotted in the bar chart.

(D) A schematic presentation of alternative PAS utilization in the 3′ UTR in conditions with reduced PABPN1 levels. The positions of distal (D) and proximal (P) primer sets in Sirt1 mRNA are depicted.

(E and F) Distal to proximal ratio in Sirt1 3′ UTR (E) and Sirt1 fold change (F) in scram or shPab muscle cell cultures. Fold change was calculated after normalization to Hprt housekeeping gene and scram control. Statistical difference was assessed with the Student's t test. p < 0.05.

(G) A schematic presentation showing the position of Sirt1 AONs. AONs are designed to mask proximal PAS in Sirt1 mRNA.

(H and I) Bar charts show the distal to proximal ratio in Sirt1 3′ UTR (H) and the Sirt1 fold change (I). Scrambled AON is depicted in the black bar, and with Sirt1 AONs are depicted in dashed bars. Fold change (I) was calculated after normalization to Hprt housekeeping gene and scram control. Averages and standard deviations of all the experiments here are from three biological replicates. Statistical difference was assessed with the Student's t test. p < 0.05.

Next, we investigated whether Sirt1 mRNA is directly regulated by APA utilization. We designed antisense oligonucleotides (AONs) specific to the proximal polyadenylation signal (PAS) in Sirt1 3′ UTR or scrambled AON (Figure 5G; Related to Table S7). The AONs were transfected into cell cultures, and masking of proximal PAS was assessed using the distal to proximal ratio. Compared with scrambled AON, the distal to proximal ratio was higher in Sirt1-AONs in shPab cell culture (Figure 5H), and consistently, Sirt1 long transcript levels were also elevated (Figure 5I). Together this suggests that in conditions with reduced PABPN1 levels Sirt1 levels are modulated via APA utilization. In control cells, Sirt1 mRNA was also elevated after Sirt1-AON transfection, which was not due to APA utilization (Figures 5H and 5I), suggesting that additional factors affect the stability of long Sirt1 transcript.

SIRT1 Inhibition Elevates PABPN1 Expression in shPab Cell Culture

Next, we investigated whether SIRT1 inhibition could reverse mitochondrial activity in shPab cell culture. Treatment with sirtinol, a SIRT1 inhibitor (Villalba and Alcaín, 2012), caused an increase in mitochondrial membrane potential in shPab cell culture (Figure 6A). Also, in shPab cell cultures that were treated with Sirt1-AON we found an increase in mitochondrial membrane potential (Figure S7). In addition, sirtinol treatment restored muscle cell fusion in shPab cell culture (Figure 6B). Consistently, increased cell fusion was also found in stable SIRT1 knockdown cell culture (Figure S8). This suggests that inhibition of SIRT1 activity restores PABPN1-mediated myogenic defects.

Figure 6

Sirtinol Treatment Reverses Myogenic Defects in shPab Muscle Cell Culture

(A) Mitochondrial membrane potential in shPab vehicle and sirtinol-treated muscle cell cultures. Images show an overlay between monomers (green) and J-aggregates (red) and the nuclei (blue). Scale bar, 10 μm. The ratio is measured from >30,000 cells.

(B) Muscle cell fusion in shPab vehicle and sirtinol-treated cell cultures. Images show an overlay between MyHC staining (green) and the nuclei (blue). Scale bar, 20 μm. The fraction of nuclei within MyHC regions is from >50,000 nuclei.

(C) Western blot analysis of PABPN1 levels in sirtinol-treated scram and shPab cell cultures. Tubulin is used as loading control and for normalization.

(D) Accumulation of nuclear PABPN1. Images show an overlay between PABPN1 (red) and nuclei (blue) in shPab vehicle and sirtinol-treated cell cultures. The segmented nuclei are depicted with a blue line. Scale bar, 10 μm. Bar chart shows the average MFI of nuclear PABPN1 from >50,000 nuclei.

(E and F) Distal to proximal ratio in Sirt1 3′ UTR (E) or mRNA fold change (F) in mock and sirtinol-treated shPab cell cultures. Fold change was calculated after normalization to Hprt and scram mock. Averages and standard deviations are from three biological replicates (A, B, D, E, F, and C, respectively). Statistical difference was assessed with the Student's t test. *p < 0.05.

We then explored if SIRT1 inhibition affects PABPN1 function. We found that PABPN1 protein levels were elevated in a dose-dependent manner with 10 μM as the optimal sirtinol concentration (Figure S9). Consistently, PABPN1 increased in scram and shPab cells treated with sirtinol (Figure 6C). Consistently, PABPN1 levels were also increased in SIRT1 knockdown muscle cell culture (Figure S8). Sirtinol treatment also elevated the levels of nuclear PABPN1 in shPab cell culture (Figure 6D). As nuclear PAPBN1 regulates APA utilization, we determined the distal to proximal ratio of Sirt1 mRNA in sirtinol-treated cells, and found a reversion compared with shPab cell culture (Figure 6E). Consistently, Sirt1 fold change was also reduced (Figure 6F). In a recent study we showed that PABPN1 also regulates levels of several autophagy-related genes (Raz et al., 2017). Sirtinol treatment also restored levels of four of five autophagy-related genes (Figure 6F). Together, the results suggest that sirtinol treatment restores PABPN1 levels and its activity. Hence, this treatment could be beneficial in conditions with reduced PABPN1 levels.

Sirtinol Treatment Restores PABPN1-Mediated Muscle Wasting

We then investigated the effect of sirtinol treatment in shPab muscles. Sirtinol or empty vehicle was contralaterally injected into shPab muscles of three mice (Figure 7A). Treatment was repeated twice, and muscles were harvested for ex vivo analyses (Figure 7A). Same as in cell culture, PABPN1 protein levels were increased in sirtinol-treated muscles (Figures 7B and 7C). In the sirtinol-treated shPab muscles ECM thickening was reduced and split myofibers were absent (Figure 7D). In addition, we observed a strong trend wherein the percentage of central nuclei decreased in sirtinol-treated muscles and myofiber cross-sectional area increased (Figures 7E and 7F). Previously we reported a shift in myofiber typing in shPab muscles: myosin-heavy chain (MyHC)-2b reduced, but MyHC-2a slightly increased (Riaz et al., 2016). Sirtinol treatment restored myofiber typing: MyHC-2b was elevated and MyHC-2a was reduced (Figure 7G). Last, we also noticed that MyHC-2b foci were formed in shPab muscles, which were completely absent in scram muscles (Figure S10). This was consistent with our previous study (Riaz et al., 2016). MyHC-2b foci are formed by the loss of ACTN3 (MacArthur et al., 2007). In agreement, reduced ACTN3 protein levels were found in the shPab proteome (related to Table S3). The number and size of MyHC-2b foci were decreased in sirtinol-treated shPab muscles (Figure 7H). Taken together, this suggests that sirtinol treatment could be beneficial in PABPN1-mediated muscle wasting.

Figure 7

Sirtinol Treatment in shPab Muscles Reverses PABPN1-Induced Muscle Pathology

(A) An overview of sirtinol treatment in shPab muscles. Muscles were injected with AAV9 shPab in both right and left tibialis anterior muscles in three mice. After 3 weeks muscles were injected twice, with a 10-day interval, with empty vehicle or sirtinol.

(B) A representative western blot shows PABPN1 protein levels in shPab vehicle and sirtinol-injected muscles. Equal loading is assessed with Coomassie blue (CB). The mouse with the lowest PABPN1 fold change is depicted with a light gray line.

(C) Dot plot shows normalized PABPN1 expression levels. Lines connect paired muscles. Means and standard deviations are depicted with a black line.

(D) Gomori trichrome staining in muscle cryosections. Extracellular thickening is depicted with a white arrow, examples of central myonuclei and split myofibers are indicated with red and black arrows, respectively. Scale bar, 20 μm.

(E) Dot plot shows the fraction of central nuclei in mock and sirtinol-treated shPab muscles; >1,500 myofibers were counted per condition per mouse. Lines connect between paired muscles. Control AAV9 scrambled shRNA was injected into tibialis anterior TA (N = 4 muscles). Means and standard deviations are depicted with a black line.

(F) Cumulative distribution plot of myofiber cross-sectional area (CSA) in vehicle and sirtinol-treated muscles.

(G) Myofiber typing: Images show an overlay of MyHC-2b (green), MyHC-2a (red), and DAPI (blue) staining in vehicle or sirtinol-treated shPab muscles. Scale bar, 20 μm. Cumulative distribution plot of MyHC-2a and MyHC-2b MFI in single myofibers from mock or sirtinol-treated muscles. Statistical difference was determined with the Kolmogorov-Smirnov test; *p < 0.0005.

(H) Representative images of MyHC-2b foci in vehicle and sirtinol-treated shPab muscles. Scale bar, 50 μm. The upper right box shows segmented MyHC-2b foci. Chart bars show mean MyHC-2b foci per myofiber (left panel) and mean foci area (right panel). Statistical difference was assessed with the Student's t test. (F–H) Pooled myofibers from all muscles, number of single myofibers > 1,000.

Discussion

Muscle wasting is characterized by multiple histological alterations such as, atrophy, regeneration, thickening of the ECM, and switches in myofiber typing. Dysregulation of pathways regulating protein homeostasis, like the UPS and autophagy, play a role in muscle atrophy (Schiaffino et al., 2013). Reduced PABPN1 levels regulate expression levels of UPS, autophagy, and mitochondrial genes via APA utilization in the 3′ UTR, which consequently lead to reduced activity of all the cellular machineries (Chartier et al., 2015, Raz et al., 2017, Riaz et al., 2016). Reduced PABPN1 levels cause myogenic defects in muscle cell culture (Anvar et al., 2013, Apponi et al., 2010). Moreover, muscle wasting is found in mouse models for OPMD (Trollet et al., 2010, Vest et al., 2017). ECM thickening and central nuclei were also reported in affected OPMD muscles (Raz et al., 2013). Also, in shPab muscles we found ECM thickening and central nuclei, suggesting that reduced PABPN1 levels cause multiple pathological features of muscle wasting and myogenic defects.

PABPN1 regulates mRNA processing, yet it is not fully understood how an altered transcriptome leads to muscle wasting pathology. Recently we showed that transcripts from APA in the 3′ UTR are under-represented in the translation machinery (Raz et al., 2017). Substantial discrepancies between mRNA fold changes and translation efficiency were also reported in muscle cell cultures (de Klerk et al., 2015). Here we found only limited correlation between transcript and protein fold changes in the same muscles. In agreement, limited overlap between transcriptome and proteome was also reported in muscles from aged adults (Robinson et al., 2017). Limited correlation between outcomes from RNA-seq-based transcriptome and mass-spectrometry-based proteome could be affected by technical issues and in-depth differences. Noteworthy is that for the more affected genes (fold change |>1.5|), variation of their corresponding proteins is generally of a greater magnitude. This suggests that despite a genome-wide effect of PABPN1 on mRNA processing, only a subset of transcripts affect the muscle pathology. Higher protein fold changes in shPab muscles are consistent with lower proteasome activity and reduced autophagosome formation (Raz et al., 2017, Riaz et al., 2016). Posttranscriptional drivers of muscle wasting are associated with protein breakdown regulated by the ubiquitin proteasome and autophagy lysosome pathways (Schiaffino et al., 2013). Also, in aging muscles and in OPMD models, dysregulation of protein homeostasis seems to play a central role (Anvar et al., 2011b, Murgia et al., 2017, Raz et al., 2013, Vest et al., 2017). A broader effect on protein catabolism is implicated by higher levels of the ribosomal proteins in aging muscles, shPab muscles, and OPMD models (Murgia et al., 2017, Robinson et al., 2017, Vest et al., 2017). In addition, reduced expression of mitochondrial proteins is also commonly found in shPab, OPMD, and aged muscles (Murgia et al., 2017, Robinson et al., 2017, Vest et al., 2017). In OPMD muscles, reduced mitochondrial activity is correlated with APA utilization in a subset of mitochondrial genes (Chartier et al., 2015). Mitochondrial activity is more robustly regulated by reversible acetylation (Baeza et al., 2016). Acetylation of mitochondrial proteins is common in skeletal muscles (Lundby et al., 2012). Differential acetylation of mitochondrial proteins was found to be prominent in atrophic rat muscles (Ryder et al., 2015), and consistently also in shPab muscles, with complexes I and V being the most affected. NAD+ production is carried out in complex I, and sirtuins activity is NAD+ dependent. We show that deacetylation inhibition by sirtinol treatment in shPab restored mitochondrial membrane potential and myogenic defects.

Acetylation affects protein levels and function, among which the effect on metabolic processes is widely studied (Drazic et al., 2016). Acetylation is regulated by the sirtuins NAD+-dependent deacetylase gene family (Imai and Guarente, 2016). SIRT1 facilitates metabolic benefits in various cells across different tissues, and a central role in aging muscles was elucidated in many studies (Chang and Guarente, 2014). The mechanisms regulating SIRT1 levels in aging tissues are inadequately understood. Here we show that SIRT1 levels are directly tuned by PABPN1 via APA in the 3′ UTR. Masking the proximal polyadenylation site using specific AONs restored APA in SIRT1 and mitochondrial membrane potential in shPab muscle cell culture. Masking APA using oligonucleotides to pathogenic molecules was suggested as a specific silencing approach (Chen et al., 2016, Marsollier et al., 2016, Raz et al., 2014). Moreover, SIRT1 inhibition restored PABPN1 levels, PABPN1 function, and myogenesis defects, suggesting that SIRT1, directly or indirectly, regulates PABPN1 levels. Acetylated PABPN1 was identified in the SIRT1-dependent acetylome (Scholz et al., 2015), suggesting that PABPN1 could be modulated by SIRT1-dependent deacetylation. Yet, how acetylation of PABPN1 affects its level and function should be explored in future studies. Nevertheless, we show that acetylation inhibition restored muscle wasting in shPab. Our results are consistent with studies suggesting that restoring NAD+ levels or targeting SIRT1 could ameliorate the age-associated disorders (Giblin et al., 2014, Villalba and Alcaín, 2012) (Tarragó et al., 2018). Moreover, sirtinol has been shown to reduce aggregation of the expanded PABPN1 in a C. elegans model for OPMD (Catoire et al., 2008, Pasco et al., 2010). In prokaryotes, increased acetylation leads to enhanced aggregation and formation of inclusion bodies (Kuczynska-Wisnik et al., 2016). Whether deacetylation directly affects protein aggregation in muscles needs to be investigated in future studies.

Taken together, we show that muscle wasting, atrophy, and regeneration are induced by reduced PABPN1 levels. The shPab proteome shows a modulated protein acetylation profile, and the shPab acetylome is predominantly hypoacetylated. Reduced protein acetylation is consistent with higher SIRT1 levels. We show that Sirt1 mRNA is regulated by PABPN1, and in turn, SIRT1 inhibition or knockdown elevates PABPN1 protein levels and restores PABPN1 molecular function. Consequently, cellular defects and muscle wasting are repaired in conditions with reduced PABPN1 levels. Our study suggests that maintaining physiological levels of both PABPN1 and SIRT1 is beneficial for muscles, and disturbance causes muscle wasting. This is the first study showing that reduced PABPN1 levels broadly affect protein acetylation. Future studies in other models and possibly in human aging muscles will further validate the role of PABPN1 in the protein acetylation landscape and its effect on muscle wasting.

Limitations of the Study

This study presents the role of protein deacetylation in a mouse model, which was generated by AAV9-expressing shRNA to PABPN1. Our study design allowed paired analysis, excluding inter-mice variations. Yet, parts of this study should be replicated in other models. Recently, two other mouse models for PABPN1 were published (Vest et al., 2017), and these should be considered in future studies. The role of protein deacetylation in aging muscles should be further studied in human muscles.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.