Inhibition of lncRNA MAAT Controls Multiple Types of Muscle Atrophy by cis- and trans-Regulatory Actions

Abstract

Muscle atrophy is associated with negative outcomes in a variety of diseases. Identification of a common therapeutic target would address a significant unmet clinical need. Here, we identify a long non-coding RNA (lncRNA) (muscle-atrophy-associated transcript, lncMAAT) as a common regulator of skeletal muscle atrophy. lncMAAT is downregulated in multiple types of muscle-atrophy models both in vivo (denervation, Angiotensin II [AngII], fasting, immobilization, and aging-induced muscle atrophy) and in vitro (AngII, H₂O₂, and tumor necrosis factor alpha [TNF-α]-induced muscle atrophy). Gain- and loss-of-function analysis both in vitro and in vivo reveals that downregulation of lncMAAT is sufficient to induce muscle atrophy, while overexpression of lncMAAT can ameliorate multiple types of muscle atrophy. Mechanistically, lncMAAT negatively regulates the transcription of miR-29b through SOX6 by a trans-regulatory module and increases the expression of the neighboring gene Mbnl1 by a cis-regulatory module. Therefore, overexpression of lncMAAT may represent a promising therapy for muscle atrophy induced by different stimuli.

Graphical Abstract

Xiao and colleagues report that lncMAAT is a common regulator of muscle atrophy. Inhibition of lncMAAT is sufficient to induce muscle atrophy by cis- and trans-regulatory actions. Overexpression of lncMAAT can ameliorate multiple types of muscle atrophy. Their work provides a promising therapy for muscle atrophy induced by different stimuli.

Introduction

Muscle atrophy commonly occurs in aging, disuse, starvation, and many chronic diseases, including heart failure and cancer.¹ Muscle atrophy leads to muscular weakness and reduced quality of life, which also significantly increases morbidity and mortality.^2,3 While progress has been made in understanding the molecular underpinnings of muscle atrophy, currently there are no effective approved drugs to combat muscle atrophy.² Thus, effective therapies for muscle atrophy would address an important unmet clinical need.³

As muscle atrophy is commonly associated with many diseases, there may exist shared signaling pathways important for its pathogenesis.⁴ Identification of key signaling entities in these common pathways may lead to the discovery of a common therapeutic target for muscle atrophy.⁴ Two muscle-specific ubiquitin ligases, MuRF1 (for muscle RING finger 1) and Atrogin-1/MAFbx (muscle atrophy F-box), have been found to be consistently elevated and required for multiple types of muscle atrophy.⁴ The decreased phosphatidylinositol 3-kinase (PI3K)/serine/threonine kinase Akt signaling pathway also contributes to many types of muscle atrophy.^5,6 A common program of transcriptional adaptations in muscle atrophy has also been reported.^7,8 Recently, we reported that a microRNA (miRNA or miR), miR-29b, controls multiple types of muscle atrophy.^5,6 We hypothesized that common molecular regulatory mechanisms in these pathways governing muscle atrophy may yield effective therapeutic targets for muscle atrophy.^5,6

Long non-coding RNAs (lncRNAs) are transcribed RNAs longer than 200 nucleotides with little or no protein-coding capacity.⁹ lncRNAs are emerging as important regulators of many essential biological processes including cell survival, cell growth, differentiation, development, and metabolism.^9,10 Aberrant expression of lncRNAs has been linked to diverse pathologies, including cancer and metabolic and cardiovascular diseases.^11,12 lncRNAs have also been related to muscle physiology and disease.^11,13 Several lncRNAs have been reported in myogenesis, including Linc-YY1, Linc-RAM, lncRNA-Six1, H19, SYISL, lncMyoD, lncmg, and linc-MD1.^11,13, 14, 15, 16, 17, 18, 19, 20 However, few lncRNAs have been mechanistically linked to skeletal muscle diseases (apart from lncRNAs linked to Duchenne muscular dystrophy [DMD] and facioscapulohumeral muscular dystrophy [FSHD]).^14,21,22 A limited number of lncRNAs, including lnclRS1, lncMUMA, and lncRNA Atrolnc-1, have been found to participate in specific types of muscle atrophy.23, 24, 25 However, if a common regulatory lncRNA can mediate muscle atrophy in different disease models remains unclear.

In the present study, we identify an lncRNA (muscle-atrophy-associated transcript, MAAT) as a common regulator of muscle atrophy. lncMAAT is downregulated in multiple types of muscle atrophy both in vivo (denervation, Angiotensin II [AngII], fasting, immobilization (Imo), and aging-induced muscle atrophy) and in vitro (AngII, H₂O₂, and tumor necrosis factor alpha [TNF-α]-induced muscle atrophy). Downregulation of lncMAAT is sufficient to induce muscle atrophy, while overexpression of lncMAAT can protect multiple types of muscle atrophy. Mechanistically, lncMAAT negatively regulates the transcription of miR-29b through sex-determining region Y-box 6 (SOX6), and lncMAAT also increases the expression of the neighboring gene muscleblind-like 1 (MBNL1) by a cis-regulatory module. These data suggest that lncMAAT overexpression is a potential effective therapy for multiple types of muscle atrophy.

Results

lncMAAT Is Decreased in Multiple Types of Muscle Atrophy and Controls Muscle Atrophy In Vitro

To identify lncRNAs that play a role in muscle atrophy, lncRNA microarray were performed on the gastrocnemius muscles from mice that have undergone denervation of the right sciatic nerve. A total of 3,030 dysregulated lncRNAs were identified (p < 0.05, fold change ≥ 2.0), among which 1,913 lncRNAs were upregulated and 1,117 lncRNAs were downregulated (Figure 1A; Table S1). Among all the lncRNAs identified above, we are more interested in identifying a lncRNA that might be a comediator of multiple types of muscle atrophy. Our previous studies have demonstrated that miR-29b contributed to muscle atrophy in a variety of disease models.^5,6 Thus, among these dysregulated lncRNAs identified by lncRNA microarray, we sought to identify lncRNAs that could potentially function as the sponge of miR-29b. Using miRanda (v3.3a) to computationally predict lncRNAs that can function as a sponge for miR-29b with more than 3 binding sites, a list of 18 lncRNAs were identified (Figure 1B). We next determined changes in the expression of these lncRNAs in mouse denervated gastrocnemius muscles and found that four of these lncRNAs were decreased, four were increased, six were unchanged, and four were not detected (Figure 1C). As we were interested in lncRNAs that could sponge miR-29b, we focused only on those lncRNAs that were decreased with muscle atrophy (which might be able to lead to increased miR-29b). In fully differentiated C2C12 myotubes, two independent small interfering RNAs (siRNAs) to each lncRNA were used to silence the 4 lncRNAs that were decreased with atrophy. lncMAAT knockdown significantly reduced myotube diameter and elevated Atrogin-1 and MuRF-1 (Figure S1A), while knockdown of the other three lncRNAs did not consistently affect the myotube diameter and expressions of Atrogin-1 and MuRF-1 (Figures S1B–S1D). Together, these data suggested that downregulation of lncMAAT may contribute to muscle atrophy.

Figure 1.

lncMAAT Expression Is Reduced in Multiple Types of Muscle Atrophy

(A) lncRNA microarray showed dysregulated lncRNAs in gastrocnemius muscle from denervation (Den) mouse model (n = 4 per group). (B) The list of lncRNAs that might be sponges of miR-29b as predicted among the dysregulated lncRNAs identified by lncRNA microarray. (C) qRT-PCR analysis of the candidate lncRNAs in gastrocnemius muscle from Den mice compared to sham (n = 3 for sham and 4 for Den). (D) qRT-PCR analysis of lncMAAT in TA, EDL, and soleus from Den mice compared to sham (n = 6 per group). (E) qRT-PCR analysis of lncMAAT in different animal models and cellular models of muscle atrophy. Animal model, AngII (n = 6 per group); fasting (n = 6 per group); immobilization (Imo) (n = 4 per group); aging (n = 4 per group). Cellular model, AngII (n = 6 per group); H₂O₂ (n = 4 per group), TNF-α (n = 4 per group). ∗p < 0.05; ∗∗p < 0.01; data represent mean ± SEM. See also Figures S1 and S2.

To further confirm that, we also investigated if the downregulation of lncMAAT in denervation was a generalized process or specific to gastrocnemius muscles. We checked the level of lncMAAT in other muscles after denervation including tibialis anterior (TA), extensor digitorum longus (EDL), and soleus and found that it was decreased in all these tested muscles (Figure 1D). To further prove that downregulation of lncMAAT is associated with multiple types of muscle atrophy, we determined the expression level of lncMAAT in four other muscle atrophy in vivo models, including AngII infusion, fasting, immobilization, and aging-induced muscle atrophy. We found that lncMAAT was consistently downregulated in all these models (Figure 1E). Finally, lncMAAT was also decreased in in vitro models of induced-muscle-atrophy differentiated C2C12 myotubes treated with AngII, H₂O₂, and TNF-α (Figure 1E). These data suggest that downregulation of lncMAAT may be implicated as a common mediator of muscle atrophy.

To determine if the change of lncMAAT is specific to skeletal muscle or a general response in muscle during the factors (denervation, immobilization, and AngII) that induce skeletal muscle atrophy, we determined the expression of lncMAAT in the heart and found that lncMAAT was unchanged in all these models (Figures S2A and S2B). Furthermore, we also assessed the expression of lncMAAT in doxorubicin (Dox)-induced cardiac atrophy tissues and cancer cachexia-induced cardiac atrophy tissues26, 27, 28, 29 and found that lncMAAT was unchanged in these cardiac atrophy models (Figure S2C). These data suggest that the inhibition of lncMAAT is specific to skeletal muscle atrophy.

Knockdown of lncMAAT with two independent shRNAs reduced myotube diameter, elevated Atrogin-1 and MuRF-1, increased autophagy, protein ubiquitination, apoptosis, and decreased mtDNA copy numbers, and decreased major histocompatibility complex (MHC) levels (Figure 2; Figures S3A–S3C). Importantly, the AKT/FOXO3A/mTOR pathway was also inhibited by lncMAAT knockdown as evidenced by decreased phosphorylation of AKT (Ser-473), FOXO3A (Ser-253), mTOR, P70S6K, and 4EBPI (Figure S3D). Additionally, PI3K was positively regulated by lncMAAT (Figure S3E), suggesting that lncMAAT mediated AKT/FOXO3A/mTOR pathway via PI3K. Similar results were observed with siRNAs to lncMAAT (Figures S4 and S5). Collectively, these data suggest that a decrease in lncMAAT is related to muscle atrophy, and lncMAAT knockdown is sufficient to cause muscle atrophy.

Figure 2.

Inhibition of lncMAAT Is Sufficient to Induce Muscle Atrophy In Vitro

(A) qRT-PCR analysis of lncMAAT for C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 4 per group). (B) Immunofluorescent staining and quantification for C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 4 per group; scale bar, 100 μm). (C) qRT-PCR analysis of Atrogin-1 and MuRF-1 for C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 6 per group). (D) Western blot analysis for ubiquitin protein expression in C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 3 per group). (E) Western blot analysis for P62 and LC3II expression in C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 3 per group). (F) Western blot analysis for Bax, Bcl2, and cleaved-caspase3 expression in C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 3 per group). (G) qRT-PCR analysis of the ratio of mtCol1 and genomic GAPDH in C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 6 per group). (H) Western blot analysis for MHC expression in C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 3 per group). ∗p < 0.05; ∗∗p < 0.01; data represent mean ± SEM. See also Figures S3, S4, and S5.

To further investigate if the decreased expression of lncMAAT is required for muscle atrophy in fully differentiated C2C12 myotubes, the effects of lncMAAT overexpression in AngII-, H₂O₂-, and TNF-α-induced muscle atrophy in vitro models were determined. It was found that lncMAAT overexpression could attenuate muscle atrophy in all these in vitro models, as evidenced by an attenuated decrease in myotube diameter and suppressed increase in the expressions of Atrogin-1 and MuRF-1 (Figure 3). Taken together, lncMAAT is decreased in multiple types of muscle atrophy and controls muscle atrophy in our models.

Figure 3.

lncMAAT Overexpression Alleviates Muscle Atrophy In Vitro

(A) qRT-PCR analysis of lncMAAT for C2C12 myotubes after transfection with lncMAAT overexpression lentivirus (n = 6 per group). (B) Immunofluorescent staining and quantification (n = 3 per group; scale bars, 100 μm); qRT-PCR analysis of Atrogin-1 and MuRF-1 (n = 4 per group) for C2C12 myotubes after transfection with lncMAAT overexpression lentivirus in AngII-induced muscle-atrophy model. (C) Immunofluorescent staining and quantification (n = 3 per group; scale bar, 100 μm); qRT-PCR analysis of Atrogin-1 and MuRF-1 (n = 4 per group) for C2C12 myotubes after transfection with lncMAAT overexpression lentivirus in H₂O₂-induced muscle-atrophy model. (D) Immunofluorescent staining and quantification (n = 3 per group; scale bars, 100 μm); qRT-PCR analysis of Atrogin-1 and MuRF-1 (n = 6 per group) for C2C12 myotubes after transfection with lncMAAT overexpression lentivirus in TNF-α-induced muscle-atrophy model. ∗p < 0.05; ∗∗p < 0.01; data represent mean ± SEM.

lncMAAT Negatively Regulates the Transcription of miR-29b through SOX6

As we hypothesized that the capacity of lncMAAT to regulate muscle atrophy is probably mediated via miR-29b, we examined whether the expression of miR-29b was regulated by lncMAAT in C2C12 myotube. We found that overexpression of lncMAAT suppressed miR-29b, while lncMAAT knockdown increased miR-29b expression level (Figure 4A). To our surprise, lncMAAT was mainly located in the nucleus (Figures 4B and 4C), and lncMAAT could not interact with miR-29b directly (Figure S6A). Thus, we hypothesized that lncMAAT could not regulate the expression of miR-29b via competing endogenous RNA or molecular sponge mechanisms but by intranuclear transcription regulation. To confirm this hypothesis, we further tested whether lncMAAT could regulate the expression of pri-miR-29b, the primary transcript of miR-29b. As expected, we found that lncMAAT could negatively regulate the expression of pri-miR-29b (Figure 4D). These findings suggested that lncMAAT could negatively regulate the expression of miR-29b via intranuclear transcription regulation.

Figure 4.

lncMAAT Negatively Regulates the Transcription of miR-29b through SOX6

(A) qRT-PCR analysis of miR-29b for C2C12 myotubes after transfection with lncMAAT overexpression lentivirus or siRNAs (n = 5 per group for lncMAAT overexpression lentivirus; n = 6 per group for siRNAs). (B) qRT-PCR analysis of lncMAAT, GAPDH, Xist, and Malat1 in nuclear or cytosolic fraction of C2C12 cell (n = 6). (C) RNA FISH of lncMAAT in C2C12 cells using an antisense probe. Scale bar, 10 μm. (D) qRT-PCR analysis of pri-miR-29b for C2C12 myotubes after transfection with lncMAAT overexpression lentivirus or siRNAs (n = 6 per group for lncMAAT overexpression lentivirus; n = 4 per group for siRNAs). (E) Transcription factors (TFs) bind to the promotor of pri-miR-29b and interact with lncMAAT. (F) RIP assay for the binding ability of SOX6 to lncMAAT (n = 5 per group). (G) ChIP-PCR assay for the relationship between Sox6 and the promotor of miR-29b (myh7 works as a positive control; n = 5 per group). (H) ChIP-PCR assay for the binding ability of SOX6 to miR-29b with lncMAAT knockdown (n = 6 per group). (I) Immunofluorescent staining of myotube and diameter quantification of myotubes cotreated with lncMAAT shRNA and miR-29b sponge (n = 3 per group; scale bar, 100 μm); qRT-PCR analysis of the expression of Atrogin-1 and MuRF-1 for C2C12 myotubes cotreated with lncMAAT shRNA and miR-29b sponge (n = 6 per group). See also Figure S6.

In order to address how lncMAAT regulates the transcription of miR-29b, we first searched for transcription factors that could bind to the promotor of pri-miR-29b-1 and -2 by the JASPAR database (http://jaspardev.genereg.net/). After that, by assessment of the interaction between these transcription factors and lncMAAT using RegRNA 2.0 (http://regrna2.mbc.nctu.edu.tw/detection.html), we found 16 transcription factors that not only bound the promotor of pri-miR-29b but also interacted with lncMAAT (Figure 4E). Interestingly, SOX6 appeared to directly interact with lncMAAT as demonstrated by RNA binding protein immunoprecipitation (RIP) assay and electrophoretic mobility shift assays (EMSA) (Figure 4F; Figure S6B). To assess whether SOX6 could bind to the promotor of miR-29b, we performed a chromatin immunoprecipitation (ChIP) assay, which confirmed the binding of SOX6 to the promoter of miR-29b (Figure 4G), and inhibition of lncMAAT promoted the binding ability of SOX6 to miR-29b (Figure 4H). Taken together, these results indicate that lncMAAT regulates the transcription of miR-29b by direct interaction with the transcription factor SOX6. The binding of lncMAAT to SOX6 may sequester the transcription factor away from the promoter region of miR-29b, thereby explaining the inverse correlation between lncMAAT and miR-29b expression.

To further investigate if lncMAAT inhibition mediated muscle atrophy by miR-29b, we silenced both lncMAAT and miR-29b in C2C12 myotube cells. We found that inhibition of miR-29b could alleviate lncMAAT-knockdown-induced muscle atrophy (Figure 4I), which suggested that miR-29b is indeed downstream of lncMAAT in the pathogenesis of muscle atrophy. In addition, we found that Sox6 could be negatively regulated by lncMAAT (Figure S6C). Besides, Sox6 was upregulated in cellular models of muscle atrophy (Figure S6D). Furthermore, the expression of Sox6 related with the expressing lncMAAT in muscle cells through regulating miR-29b (Figures S6E and S6F).

lncMAAT Positively Regulates the Expression of Neighbor Gene Mbnl1 by a cis-Regulatory Module

The gene location of lncMAAT was chr3: 60604384-60607757(+), which is located in the third intron of Mbnl1 (Figure S7A). As cis-regulation of neighboring genes is a known mode of action of lncRNAs, we also investigated the possibility of lncMAAT directly regulating Mbnl1 transcription. lncMAAT overexpression increased the expression of MBNL1 both at mRNA and protein levels, while lncMAAT inhibition decreased it (Figures 5A–5D; Figure S7B). Mbnl1 was a splicing regulator and could result in specific dysregulation of alternative splicing of other pre-mRNAs in muscle weakness.³⁰ CnAbeta1 enhances myoblast proliferation and muscle regeneration, which was an important target of Mbnl1.^31,32 Interestingly, overexpression of lncMAAT promoted generation of CnAbeta1, while lncMAAT inhibition decreased it (Figure S7C), which was shown to be the same function of Mbnl1. In addition, Mbnl1 inhibition could rescue the increase of CnAbeta1 in lncMAAT overexpression. (Figures S7D and S7E). These data suggested that lncMAAT could have an effect on the regulation of alternative splicing via Mbnl1.

Figure 5.

lncMAAT Positively Regulates the Expression of Neighbor Gene Mbnl1 by a cis-Regulatory Module

(A) Western blot analysis for MBNL1 expression in C2C12 myotubes after transfection with lncMAAT overexpression lentivirus (n = 3 per group). (B) qRT-PCR analysis of Mbnl1 expression in C2C12 myotubes after transfection with lncMAAT overexpression lentivirus (n = 6 per group). (C) Western blot analysis for MBNL1 expression in C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 3 per group). (D) qRT-PCR analysis of Mbnl1 expression in C2C12 myotubes after transfection with lncMAAT shRNA lentivirus (n = 6 per group). (E) Immunofluorescent staining of myotube and diameter quantification of myotubes treated with Mbnl siRNA (n = 4 per group; scale bar, 100 μm). (F) qRT-PCR analysis of the expression of Atrogin-1 and MuRF-1 for C2C12 myotubes treated with Mbnl1 siRNA (n = 6 per group). (G) Immunofluorescent staining of myotube and diameter quantification of myotubes cotreated with lncMAAT shRNA and Mbnl1 overexpression lentivirus (n = 4 per group; scale bar, 100 μm); qRT-PCR analysis of the expression of Atrogin-1 and MuRF-1 for C2C12 myotubes cotreated with lncMAAT shRNA and Mbnl1 overexpression lentivirus (n = 5–6 per group). See also Figure S7.

Inhibition of Mbnl1 induced muscle atrophy in C2C12 myotube cells, as indicated by reduced myotube diameter and elevated expressions of Atrogin-1 and MuRF-1 (Figures 5E and 5F; Figure S7D), while Mbnl1 overexpression could reverse the pro-atrophy effects of lncMAAT knockdown in C2C12 myotube cells (Figure 5G; Figure S7F). Collectively, these data prove that lncMAAT could positively regulate the expression of the neighboring gene Mbnl1 by a cis-regulatory module and could therefore regulate muscle atrophy in a miR-29b-independent manner as well.

To further determine the dominant effect of lncMAAT in the target genes, we performed the rescue experiment by intervening the miR-29b and Mbnl1 alone or together. It was found that overexpressing Mbnl1 or inhibiting miR-29b could rescue muscle atrophy induced by lncMAAT inhibition. Mbnl1 overexpression and miR-29b inhibition together showed a better rescue effect in myotubes (Figures S7F and S7G). Furthermore, miR-29b and Mbnl1 could not influence each other in myotubes (Figures S7H and S7I). These data suggest that miR-29b and Mbnl1 regulate muscle atrophy through independent signaling pathways.

lncMAAT Controls Muscle Atrophy In Vivo

To reveal the in vivo relevance of lncMAAT knockdown, we used lentivirus-mediated knockdown of lncMAAT to decrease the expression level of lncMAAT in mouse gastrocnemius muscles. Using two independent lentivirus-mediated shRNAs for lncMAAT, muscle atrophy was induced, as confirmed by decreased gastrocnemius weight, the ratio of gastrocnemius weight to body weight, grip strength, myotube diameter, and elevated Atrogin-1 and MuRF-1, apoptosis, autophagy, and protein ubiquitination (Figure 6). In addition, the MHC level was also decreased, while autophagy-related genes, including Atg12, Atg4b, and Vps34, and some atrogenes, including Traf6, Cblb, and Nedd4, were increased by lncMAAT knockdown (Figures S8A–S8C). Finally, knockdown of lncMAAT negatively regulated the AKT/FOXO3A/mTOR growth pathway, as noted by a consistent decrease in the phosphorylation of AKT (Ser-473), FOXO3A (Ser-253), mTOR, P70S6K, and 4EBPI (Figure S8D). In addition, immunofluorescent staining with MHC antibodies showed that the proportion of type I fiber in fiber-type composition was decreased, while that of type IIB fiber was increased by lncMAAT inhibition (Figure S8E). Besides, all types of fiber underwent atrophy by lncMAAT inhibition (Figure S8E). We next determined whether the targets identified in vitro were regulated by lentivirus-mediated shRNAs for lncMAAT in vivo. It was found that miR-29b and Sox6 were increased while Mnbl1 was decreased in gastrocnemius muscles when mice were injected with lentivirus-mediated shRNAs for lncMAAT (Figure S9). Thus, these results prove that lncMAAT knockdown is sufficient to induce muscle atrophy in vivo.

Figure 6.

lncMAAT Controls Muscle Atrophy In Vivo

(A) qRT-PCR analysis for lncMAAT in gastrocnemius muscle from mice injected with lncMAAT knockdown lentivirus (n = 7 per group). (B) Gastrocnemius muscle morphology of mice injected with lncMAAT knockdown lentivirus (scale bar, 0.5 cm). (C) Gastrocnemius weight (GW) and GW/body weight (GW/BW) ratio of mice injected with lncMAAT knockdown lentivirus (n = 7 per group). (D) The grip strength of right hind limbs of mice injected with lncMAAT knockdown lentivirus (n = 7 per group). (E) WGA staining for myofiber of mice injected with lncMAAT knockdown lentivirus (n = 7 per group; scale bar, 80 μm). (F) qRT-PCR analysis for Atrogin-1 and MuRF-1 gene expression in gastrocnemius muscle from mice injected with lncMAAT knockdown lentivirus (n = 7 per group). (G) TUNEL staining for gastrocnemius muscle from mice injected with lncMAAT knockdown lentivirus (n = 6 per group; scale bar, 100 μm). (H) Western blot analysis for Bax, Bcl2, and cleaved caspase3 expression in gastrocnemius muscle from mice injected with lncMAAT knockdown lentivirus (n = 3 per group). (I) Western blot analysis for LC3II and P62 expression in gastrocnemius muscle from mice injected with lncMAAT knockdown lentivirus (n = 3 per group). (J) Western blot analysis for ubiquitin protein expression in gastrocnemius muscle from mice injected with lncMAAT knockdown lentivirus (n = 3 per group). ∗p < 0.05; ∗∗p < 0.01; data represent mean ± SEM. See also Figures S8 and S9.

To determine whether lncMAAT overexpression can protect muscle atrophy, we treated mice with intramuscular injection of lncMAAT overexpression lentivirus, followed by AngII infusion. lncMAAT overexpression lentivirus could significantly increase the expression level of lncMAAT whether in the presence of AngII or not. The AngII-induced muscle-atrophy mouse model is notable for significantly decreased gastrocnemius weight, the ratio of gastrocnemius weight to body weight, grip strength, myotube diameter, and elevated Atrogin-1, MuRF-1, and apoptosis (Figure 7). lncMAAT overexpression significantly attenuated those changes observed in the AngII-induced muscle-atrophy mice (Figure 7). Additionally, the decreased phosphorylations of AKT (Ser-473), FOXO3A (Ser-253), mTOR, P70S6K, and 4EBPI in AngII-induced muscle atrophy were also attenuated by lncMAAT overexpression (Figure S10A). Furthermore, the changes of target genes could be rescued by lncMAAT overexpression in AngII-induced muscle atrophy. (Figures S10B and S10C). Collectively, these data confirmed that lncMAAT overexpression is able to protect AngII-induced muscle atrophy.

Figure 7.

lncMAAT Overexpression Protects AngII-Induced Muscle Atrophy

(A) qRT-PCR analysis for lncMAAT in gastrocnemius muscle from mice injected with lncMAAT overexpression lentivirus in AngII-induced muscle atrophy (n = 6, 10, 6, 10). (B) Gastrocnemius muscle morphology of mice injected with lncMAAT overexpression lentivirus in AngII-induced muscle atrophy (scale bar, 1 cm). (C) GW and GW/BW ratio of mice injected with lncMAAT overexpression lentivirus in AngII-induced muscle atrophy (n = 6, 10, 6, 10). (D) The grip strength of right hind limb of mice injected with lncMAAT overexpression lentivirus in AngII-induced muscle atrophy (n = 6, 10, 6, 10). (E) qRT-PCR analysis for Atrogin-1 and MuRF-1 gene expression in gastrocnemius muscle from mice injected with lncMAAT overexpression lentivirus in AngII-induced muscle atrophy (n = 6, 10, 6, 10). (F) TUNEL and WGA stainings for myofiber of mice injected with lncMAAT overexpression lentivirus in AngII-induced muscle atrophy (n = 6 per group; scale bar, 100 μm). (G) Western blot analysis for Bax, Bcl2, and cleaved caspase3 expression in gastrocnemius muscle from mice injected with lncMAAT overexpression lentivirus in AngII-induced muscle atrophy (n = 3 per group). ∗p < 0.05; ∗∗p < 0.01; data represent mean ± SEM. See also Figures S10–S13.

As we hypothesized that lncMAAT downregulation is a common mediator of muscle atrophy, we also determined the effects of lncMAAT overexpression in two additional muscle-atrophy mouse models, namely denervation-induced muscle atrophy and immobilization-induced muscle atrophy. We found that in the denervation-induced muscle-atrophy model, gastrocnemius weight, the ratio of gastrocnemius weight to body weight, and myotube diameter were significantly decreased, while the expression of Atrogin-1, MuRF-1, and apoptosis were elevated (Figures S11A–S11F). In addition, the phosphorylations of AKT (Ser-473), FOXO3A (Ser-253), mTOR, P70S6K, and 4EBPI were decreased (Figure S11G). lncMAAT overexpression significantly attenuated those changes (Figure S11). Additionally, similar protective effects were achieved by lncMAAT overexpression in the immobilization-induced muscle-atrophy model (Figure S12). Moreover, similar effects of alleviating the target genes was confirmed in different muscle-atrophy models by lncMAAT overexpression (Figure S13). Therefore, these data confirm that lncMAAT downregulation contributes to multiple types of muscle atrophy and forced expression of lncMAAT can ameliorate muscle atrophy in several different in vivo models.

Discussion

Muscle atrophy is a debilitating response to a variety of conditions, including disuse, fasting, cancer, and other systemic diseases including heart failure.^25,33,34 Identification of a common target for muscle atrophy would have a significant impact.4, 5, 6 MuRF1 and MAFbx have been found to be common drivers of muscle atrophy,⁴ and our previous work demonstrated that miR-29b controls multiple types of muscle atrophy.^5,6 Here, we found that lncMAAT was downregulated in multiple models of muscle atrophy. Based on our results, we propose a model for the mechanisms of action of lncMAAT during muscle atrophy (Figure 8). lncMAAT is decreased when muscle atrophy occurs. Inhibition of lncMAAT represses the expression of MBNL1 by a cis-regulatory module. On the other hand, through interaction with transcription factor SOX6, lncMAAT inhibition promotes the binding with the promotor of miR-29b, thus promoting the transcription of miR-29b. Thus, downregulation of lncMAAT contributes to muscle atrophy in a miR-29b-dependent as well as a miR-29b-independent manner. As lncRNAs are emerging as promising therapeutic targets for the drug development,^11,35 our study provides an exciting avenue to pursue in a setting where there are no treatments at present.

Figure 8.

Proposed Model of lncMAAT Function during Muscle Atrophy

lncMAAT is decreased when muscle atrophy occurs. Inhibition of lncMAAT represses the expression of MBNL1 by a cis-regulatory module. Additionally, inhibition of lncMAAT could increase the expression of SOX6 and promote the transcription of miR-29b via binding to its promotor site. Thus, downregulation of lncMAAT contributes to muscle atrophy.

lncRNAs have been identified as an emerging class of regulators of skeletal muscle physiology.^11,14,23 Aberrant expression of lncRNAs has been found in several muscular disorders, including muscle atrophy.^13,14 There are only a few studies previously implicating lncRNAs in the pathogenesis of muscle atrophy.^23,24,36 lnclRS1 has been reported to control muscle atrophy by sponging miR-15 family to activate insulin growth factor (IGF)-PI3K/AKT pathway.²⁴ lnclRS1 was identified from RNA sequencing based on hypertrophic (WRR) and leaner broilers (XH), and it was found to be upregulated in hypertrophic broilers.²⁴ lnclRS1 could rescue dexamethasone-induced muscle atrophy in chicken primary myotubes.²⁴ However, whether lnclRS1 can protect against muscle atrophy in vivo is not determined.²⁴ lncRNA MAR1 was highly expressed in mouse skeletal muscle and positively correlated with muscle differentiation and growth both in vitro and in vivo.³⁶ MAR1 overexpression in mice attenuated muscle atrophy induced by either aging or unloading through miR-487b/Wnt5a targeting.³⁶ Finally, lncRNA Pvt1 was upregulated during muscle atrophy and was shown to damage mitochondrial respiration and morphology resulting in changes in mito/autophagy, apoptosis, and myofiber size in vivo.³⁷ Pvt1 downregulation could protect from denervation atrophy.³⁷ Although lncRNA has been linked to a single type of muscle atrophy, whether a common lncRNA regulator for muscle atrophy exists is not clear.^13,36,37 Here, we identified lncMAAT downregulation as a common driver of muscle atrophy as evidenced by the following: (1) lncMAAT is downregulated in several in vivo (denervation, AngII, fasting, immobilization, and aging) and in vitro (AngII, H₂O₂, and TNF-α) muscle-atrophy models. (2) Downregulation of lncMAAT is sufficient to induce muscle atrophy both in vitro and in vivo. (3) Overexpression of lncMAAT can protect against atrophy in multiple models of muscle atrophy in vitro (AngII, H₂O₂, and TNF-α) and in vivo (AngII, denervation, and immobilization). Thus, we provide strong evidence that downregulation of lncMAAT contributes to multiple types of muscle atrophy.

As miR-29b was reported as a common regulator of muscle atrophy,^5,6 we were especially interested in knowing if lncMAAT might regulate miR-29b. We found that lncMAAT could negatively regulate miR-29b expression, and miR-29b inhibition could attenuate lncMAAT-knockdown-induced muscle atrophy. However, lncMAAT was mainly located in the nucleus. RNA localization is fundamental to its function. Some pentamer motif and nuclear retention elements (NREs) were reported to be necessary for the nuclear localization, such as motif AGCCC in lncRNA BORG and U1 snRNP components for MEG3.^38,39 Additionally, U1 motif CAGGUGAGU was reported vital for chromatin retention of non-coding RNAs (ncRNAs) and several U1 small nuclear ribonucleoprotein particle (snRNP) components (SNRNP70, SNRPA, and SNRPC) facilitated the retention.⁴⁰ As expected, bioinformatic analysis by FIMO (http://meme-suite.org/tools/fimo)⁴¹ showed that U1 motif CAGGUGAGU existed in lncMAAT (Figure S14A). Additionally, RPISeq (http://pridb.gdcb.iastate.edu/RPISeq/)⁴² revealed that lncMAAT could bind to U1 snRNP components (Figure S14B). These may imply that the U1 motif facilitates the retention of lncMAAT in the nucleus.

Because of its nuclear localization, lncMAAT would be more likely to regulate miR-29b by intranuclear transcription regulation instead of competing endogenous RNA or molecular sponge. By bioinformatic analysis, ChIP, and RIP assay, we found that SOX6 could bind to the promotor of pri-miR-29b and also interact with lncMAAT. SOX6 belongs to the Sry-related HMG-box family of transcription factors, which is a key regulator of proliferation, apoptosis, and terminal differentiation of many different cell types and organ development.⁴³ SOX6 serves tumor-suppressive functions in many types of cancer, including esophageal squamous cell carcinoma, pancreatic cancer, colorectal cancer, hepatocellular carcinoma, and chronic myeloid leukemia.44, 45, 46, 47, 48 SOX6 is also involved in insulin resistance and the developmental origins of obesity by promoting adipogenesis.⁴⁹ SOX6 is enriched in skeletal muscle, and it plays an important role in fetal skeletal muscle to terminally differentiate into a fast or slow fiber and skeletal muscle formation.50, 51, 52, 53 Here, we demonstrated that SOX6 was upregulated in muscle atrophy and lncMAAT regulated the transcription of miR-29b by directly binding to SOX6. With lncMAAT inhibition, we found the switch of slow-twitch fiber to fast-twitch fiber, which is consistent with the function of SOX6 in maintaining of fast-twitch fiber.^52,54 Collectively, lncMAAT negatively regulates the transcription of miR-29b through SOX6.

MBNL1 is a member of the MBNL family of RNA-binding proteins that can regulate alternative splicing, alternative polyadenylation, mRNA stability and localization, and miRNA biogenesis.⁵⁵ MBNL1 could regulate the pluripotency of embryonic stem cells negatively, and MBNL1 knockdown led to the increase of key pluripotency genes required for induced pluripotent stem cell generation.⁵⁶ Besides, MBNL1 is a metastasis suppressor in breast cancer.⁵⁷ In addition, knockdown of MBNL1 impairs erythroid terminal proliferation and differentiation.⁵⁸ MBNL1 is involved in the development of heart, heart valve, and central neuron system.59, 60, 61 Loss of MBNL1 could abrogate the transformation of fibroblasts into myofibroblasts⁶² and attenuate the fibrotic phase of wound healing in myocardial infarction and dermal injury mouse models.⁶³ In the muscle system, MBNL1 is also involved in the development of muscle, and MBNL1 promotes muscle cell differentiation in mouse myoblasts.⁶⁴ Perturbation of MBNL1 is associated with myotonic dystrophy type 1 (DM1), leading to cataract formation, abnormal muscle relaxation, heart and nerve dysfunction, and other pathologies.^61,65 However, the role of MBNL1 in muscle atrophy is unknown. Based on the cis-regulation model, we found that lncMAAT positively regulated the expression of neighbor gene Mbnl1. Knockdown of Mbnl1 could lead to muscle atrophy, while Mbnl1 overexpression could reverse the pro-atrophy effects of lncMAAT knockdown. Interestingly, Mbnl1 overexpression is generally well tolerated in skeletal muscle, and it has been reported to be able to rescue the muscle mass loss in DM1.⁶⁶ Thus, decreased lncMAAT lead to the decrease of its neighbor gene Mbnl1 in a cis-regulatory module, leading to muscle atrophy in a miR-29b-independent manner. Mbnl1 overexpression may be a valuable strategy for treating muscle atrophy.

In conclusion, inhibition of lncMAAT contributes to multiple types of muscle atrophy. Mechanistically, lncMAAT negatively regulates the transcription of miR-29b through SOX6. Meanwhile, lncMAAT also increases the expression of neighbor gene Mbnl1 by a cis-regulatory module. lncMAAT overexpression represents a promising therapy for multiple types of muscle atrophy.

Materials and Methods

Mouse Models

All experimental mice were 8- to 10-week-old C57BL/6 wild-type mice. C57BL/6 mice were purchased from Cavens Lab Animal (Changzhou, China). All mice maintained in specific pathogen-free (SPF) laboratory animal facility of Shanghai University (Shanghai, China). All procedures with animals followed guidelines on the use and care of laboratory animals for biomedical research published by the National Institutes of Health (no. 85-23, revised 1996),and approved by the committee on the Ethics of Animal Experiments of Shanghai University (2017-007).

The model of denervation-induced muscle atrophy was achieved through cutting a small section of the sciatic nerve by microclipping and ensuring no docking of the severed nerve. Mice that had not severed the sciatic nerve were used as controls.

The immobilization-induced muscle-atrophy model was achieved through putting the screw (0.4 × 8 mm) from the calcaneus and talus into the shaft of the tibia. The control mice were untreated.

The AngII-induced muscle-atrophy model was achieved through implanting a pump (Alzet 2001) containing 2 μg/μL AngII (Sigma, #A9525) into the back; the control mice were implanted with a pump containing PBS and received the same amount of food as the experimented mice.

All mice were sacrificed 1 week later, and muscles were dissected in rapidly frozen in liquid nitrogen for subsequent analyses.

lncRNA Microarray

Total RNAs extracted from the gastrocnemius muscles from the denervation-induced muscle-atrophy mouse model were used for lncRNA microarray. lncRNA profiling was performed with OE Biotech’s (Shanghai, China) lncRNA microarray service, based on Agilent mouse lncRNA (4∗180K, design ID 049801). The MIAME-compliant data have been submitted to Gene Expression Omnibus (GEO: GSE151300). The dysregulated lncRNAs are shown in Table S1.

Gene Expression Analyses

Total RNA was extracted from muscle tissue or cells using RNAiso (Takara). 400 ng RNA was used to reverse transcribe into cDNA by a RevertAid first-strand cDNA synthesis kit (Thermo, #K1622). Then, cDNA samples were amplified on the real-time PCR system (Bio-Rad) using the TB green premix Ex Taq (Takara, #RR420A). All data were normalized to 18 s expression. The primer sequences used are shown in Table S2.

Immunoblotting

Total protein was extracted from muscle tissues or cells using lysis buffer (KeyGEN) containing protease inhibitor (KeyGEN) and phosphatase inhibitor (KeyGEN). The protein concentration was measured with a BCA protein assay kit (Takara). Equal amounts of protein samples were separated by SDS-polyacrylamide gel electrophoresis. After transferring protein to polyvinylidene fluoride (PVDF) membrane (Millipore), the primary antibodies and appropriate secondary antibodies were incubated to detect the protein. All proteins were visualized by ECL chemiluminescent kit (Tanon), and chemical luminescence of membranes was detected with a Tanon luminescent imaging system. Primary antibodies included Phospho-mTOR (Ser-2448) antibody (Cell Signaling Technology, #2971S); mTOR antibody (Cell Signaling Technology, #2972S); phospho-FoxO3a (Ser-253) antibody (Cell Signaling Technology, #9466S); phospho-FoxO3a (Thr-32) antibody (Cell Signaling Technology, #9464S); FoxO3a polyclonal antibody (Abclonal, #A0102); phospho-p70S6 kinase (Thr-389) antibody (Cell Signaling Technology, #9205S); p70S6 kinase antibody (Cell Signaling Technology, #9202S); phospho-Akt (Ser-473) (587F11) mouse monoclonal antibody (mAb) (Cell Signaling Technology, #4051); phospho-Akt (Thr-308) (C31E5E) rabbit mAb (Cell Signaling Technology, #2965); AKT antibody (Proteintech, #10176-2-AP); phospho-EIF4EBP1-S65 polyclonal antibody (Abclonal, #AP0032); EIF4EBP1 polyclonal antibody (Abclonal, #A1248); BAX polyclonal antibody (Abclonal, #A12009); BCL2 polyclonal antibody (Abclonal, #A11025); caspase3 polyclonal antibody (Abclonal, #A2156); LC3A/LC3B polyclonal antibody (Abclonal, #A5618); P62/SQSTM1 antibody (Proteintech, #18420-1-AP); and MBNL1 polyclonal antibody (Abclonal #A8054). Proteins were stripped by stripping buffer when phosphorylated protein expression was detected.

Imaging

After harvesting the C2C12 myotube, cells were fixed by 4% paraformaldehyde (PFA) for 30 min at room temperature, permeabilized with 0.5% Triton X-100 in PBS for 15 min, and then blocked with 5% BSA in phosphate-buffered saline with Tween (PBST) for 1 h at room temperature. Then, myotubes were incubated with primary antibody MF-20 (1:100, DSHB) for more than 12 h and then incubated with secondary antibody Cy3-AffiniPure rabbit anti-mouse immunoglobulin G (IgG) (heavy chain + light chain [H+L]) or Alexa Fluor 488-conjugated Affinipure goat anti-mouse IgG (H+L) (1:200, Jackson ImmunoResearch) for more than 2 h, and finally stained the nucleus with Hoechst (1:2000, KeyGEN) for 20 min. More than 10 randomly selected regions were photographed using a fluorescence microscope (20× magnification) (Leica) in each well and at least 3 wells were used in each treatment group. ImageJ was used to measure the diameter of the myotube, and at least 100 myotubes/group were used for analysis.

Cell Culture and Transfection

C2C12 myoblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum and 1% penicillin-streptomycin mixture at 37°C with 5% CO₂. For differentiation, DMEM containing 2% horse serum and 1% penicillin-streptomycin mixture was used as differentiation medium to induce C2C12 myoblasts to differentiate into myotubes. After differentiation for 4 days, multinuclear myotubes were formed.

The differentiated myotubes were used to induce a muscle-atrophy model in vitro. The cellular models of muscle atrophy were introduced by incubation with TNF-α (PeproTech, #315-01A) at 100 ng/mL for 24 h, H₂O₂ (Sigma, #323381) at 400 μM for 24 h, and AngII (Sigma, #A9525) at 500 nM for 48 h in differentiation medium. After incubation, cells were harvested or used for morphological analysis.

Myotube transfection was performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. The transfection concentration of siRNA was 100 nM. The transfection was performed after myotubes formed, and 24 h later, the muscle-atrophy models were performed followed.

The sequences for siRNA, short hairpin RNA (shRNA), and plasmid were listed in Table S3.

Wheat Germ Agglutinin (WGA) Staining

Muscle samples were obtained, flash frozen in OCT compound (optimal cutting temperature compound; Sakura) and liquid nitrogen, and cut at 10 μm per section. The frozen sections were rewarmed at room temperature for 30 min, and then muscle tissues were fixed with 4% PFA for 15 min. After being washed with PBS three times, muscle tissues were incubated with WGA (1:100; Sigma, #L4895) for 2 h to determine the outline of myofiber, and incubated with Hoechst (1:2,000, KeyGEN) for 20 min. Finally, the images were captured by fluorescent microscope (20× magnification) (Zeiss) and were analyzed by ImageJ.

TUNEL Staining

Muscle tissues with 10-μm-thick frozen sections were rewarmed at room temperature for 30 min. Then TUNEL staining was performed by DeadEnd fluorometric TUNEL system (Promega, #G3250) following a protocol suggested by the manufacturer. In brief, the tissue was fixed in 4% PFA for 15 min and permeabilized with proteinase K for 20 min, then incubated with equilibration buffer and incubation buffer. Finally, the Hoechst (1:2,000, KeyGEN) was used for nuclear staining. Images were captured by fluorescence microscope (Leica) (20× magnification) (Zeiss).

Isolation of Nuclear and Cytoplasmic RNA

The protocol for isolation of nuclear and cytoplasmic RNA was as previously described.⁶⁷ In brief, cells were harvested and washed with PBS. Then, cell pellets were incubated in 200 μL lysis buffer A (10 mM Tris [pH 8.0], 140 mM NaCl, 1.5 mM MgCl₂, 0.5% N P-40) for 5 min on ice. After centrifugation at 1,000 × g for 3 min at 4°C, the supernatant was harvested, and 1 mL Trizol was added for cytoplasmic RNA. The pellets were lysed with buffer A with containing 1% Tween 40 and 0.5% deoxycholic acid after being washed with buffer A twice. The purified nuclear pellets were then resuspended in 1 mL Trizol for nuclear RNA.

Interaction Analysis

To predict miRNA binding sites in lncRNAs, the interaction between lncRNA and miR-29b was analyzed by miRanda v3.3a algorithm. To find the transcript factor binding to the specific promotor, the JASPAR database (http://jaspardev.genereg.net/) was used.⁶⁸ To find the identifying functional RNA motifs and sites, the RegRNA2.0 (http://regrna2.mbc.nctu.edu.tw/detection.html) was used.⁶⁹

Fluorescence In Situ Hybridization (FISH)

Biotin-labeled sense and antisense RNA probes were synthesized with PCR products by 10× biotin-labeling mix (Roche, #11685597910) and T7 polymerase (Roche, #10881767001). Then the probes were purified by purification with NucAway spin columns (Ambion, #AM10070).

The primers used were as follows:

• FRNA-F, 5′-TGTCAGACCAGCTTCATGTCG-3′;

• FRNA-T7-R, 5′-TTGTAAAACGACGGCCAGTGAATTGTAATACGACTCACTATAGGGCGGAATTCTAGGAGCGTCTCT-3′;

• negative FRNA-T7-F,

• 5′-TTGTAAAACGACGGCCAGTGAATTGTAATACGACTCACTATAGGGTGTCAGACCAGCTTCATGTCG-3′;

• negative FRNA-R, 5′-CGGAATTCTAGGAGCGTCTCT-3′.

The cells were fixed with 2%–4% freshly prepared PFA at room temperature and permeabilized with 0.2%–0.5% Triton X-100, 2 mM ribonucleoside vanadyl complex (VRC) (NEB) on ice. After being washed with 2× saline sodium citrate (SSC) buffer, the cells were hybridized at 37°C overnight in a humidified chamber with probe (10 ng/mL). Cy3-labeled streptavidin was added to cells for 1 h at room temperature after being washed with SSC buffer. After being washed three times with PBS, the cells were then mounted with prolong Gold antifade reagent with DAPI (Invitrogen, Carlsbad, CA, USA), and images were captured with a Zeiss microscope with 63× oil lens.

ChIP

A total of 10⁷ C2C12 cells were collected. Then, cells were cross-linked with 37% formaldehyde and 2 M glycine formaldehyde. Cells were lysed with cell lysis buffer with protease inhibitor (100×) (Roche) and incubated on ice for 15 min. Nuclei were obtained after centrifugation and supernatant removal. Nuclear components were extracted with nuclei lysis buffer with protease inhibitor (100×) (Roche) and incubated on ice for 30 min. Ultrasonic crushing was carried out with an ultrasonic breaker (Scientz) with 60% ultrasonic power to keep the DNA fragment at 250∼500 bp as far as possible. The supernatant was collected, and total volume was recorded.

The supernatant was incubated with IgG (Sigma, #SAB3700848) and Sox6 (Abcam, #ab30455) at 4°C for more than 12 h. 1% supernatant was taken as input and stored at 4°C. Balanced cleaning G-sepharose beads were added into the mixture and incubated at 4°C for 90 min. Then, the supernatant was discarded after centrifugation, and mixture debris was washed with elution buffer. After collection with centrifugation, mixture debris was resuspended with 10 M NaCl. Then, unlocked crosslinking was performed by heating at 65°C overnight. Finally, DNA were extracted for qRT-PCR analysis. The primer sequences were as follows: Myh7 (promotor), forward primer, 5′-ACACCGCCCACTCAATACAC-3′; reverse primer, 5′-GCCCTCTCCAAACACTCTTG-3′. miR29b-1 (promotor), forward primer, 5′-GGCTCTGGTAGCCTGTTTTAGA-3′; reverse primer, 5′-GGAGAACTAC TTCGCTGCCG-3′; miR29b-2 (promotor), forward primer, 5′-AGGCTGGTTCTTCTGA CTGC-3′; reverse primer, 5′-GTCTACCCTCTGCTGTGCTG-3′.

RIP

A total of 10⁷ C2C12 cells were collected. C2C12 cells were resuspended in 1mL pre-cooled RIP buffer. Then, cells were repeatedly frozen and thawed to make the cells fully lysed. Clearing cell lysate was done by centrifugation at 4°C. Meanwhile, Dynabeads protein G were washed with RIP buffer and blocked with BSA and yeast tRNA. The collected cell supernatant was mixed with Dynabeads protein G and incubated with a roller at 4°C for 2 h to preclear the lysate. Then, 5% of the supernatant was absorbed as an input group, and the remaining supernatant was divided into two groups, the Sox6 antibody (Abcam, #ab30455) and the IgG antibody (Sigma, #SAB3700848), and incubated overnight. The next day, the supernatant was mixed with Dynabeads protein G on roller for 2h at 4°C, and then the supernatant was discarded and Qiazol was added to extract RNA (miRNeasy mini kit; QIAGEN, #217004); the same operation was performed in the input group.

Measurement of mtDNA Copy Number

The ratio of mtDNA to genomic DNA was calculated by dividing copies of Co1 with copies of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) in each experiment. A TIANamp genomic DNA kit (TIANGEN) was used to extract total DNA from cells or gastrocnemius muscles. Then, DNA was amplified on the real-time PCR system (Bio-Rad) using the TB green premix Ex Taq (Takara). The primers used are as follows:

• mt-Co1 forward primer, 5′-CAGTCTAATGCTTACTCAGC-3′; reverse primer, 5′-GGGCAGTTACGATAACATTG-3′;

• GAPDH forward primer, 5′-GGGAAGCCCATCACCATCTTC-3′; reverse primer, 5′-AGAGGGGCCATCCACAGTCT-3′.

Force Measurements

The grip force of mice was measured with the mouse grip-force-measuring instrument (YLS-13A, Yiyan Technology, China). Simply, the adaptive training on the mice was conducted 10 min before grip-strength test. Mice were allowed to grab the metal pull bar. Then the mouse was dragged, and the force was recorded at the time when the mouse released the metal bar. Each mouse was tested 3 times with a 30-s break between tests. The experiments were blindly performed by the investigator, who did not know the group allocation.

Lentivirus Injections in Mice

For knockdown lncMAAT in mouse muscle, a single intramuscular injection of lentiviral particles was performed at the dose of 10⁸ TU per mouse.

For overexpression of lncMAAT in mouse muscle, each mouse was injected with lentivirus FUGW or overexpressed lncMAAT at the dose of 10⁸ TU, and the muscle-atrophy operation was performed 1 week after the injection.

Statistical Analysis

Results were presented as mean ± SEM using GraphPad Prism 8.0. An unpaired, two-tailed Student’s t test was used for comparisons between the two groups. A one-way ANOVA test was performed to compare multiple groups, followed by Bonferroni’ or Dunnett T3’s post hoc test, based on homogeneity of variance test. All analyses were performed using SPSS Statistics 20.0. Differences were considered significant with p <0.05.