Correction of Clcn1 alternative splicing reverses muscle fiber type transition in mice with myotonic dystrophy

Abstract

In myotonic dystrophy type 1 (DM1), deregulated alternative splicing of the muscle chloride channel Clcn1 causes myotonia, a delayed relaxation of muscles due to repetitive action potentials. The degree of weakness in adult DM1 is associated with increased frequency of oxidative muscle fibers. However, the mechanism for glycolytic-to-oxidative fiber type transition in DM1 and its relationship to myotonia are uncertain. Here we cross two mouse models of DM1 to create a double homozygous model that features progressive functional impairment, severe myotonia, and near absence of type 2B glycolytic fibers. Intramuscular injection of an antisense oligonucleotide for targeted skipping of Clcn1 exon 7a corrects Clcn1 alternative splicing, increases glycolytic 2B levels to ≥ 40% frequency, reduces muscle injury, and improves fiber hypertrophy relative to treatment with a control oligo. Our results demonstrate that fiber type transitions in DM1 result from myotonia and are reversible, and support the development of Clcn1-targeting therapies for DM1.

Introduction

Myotonic dystrophy type 1 (dystrophia myotonica; DM1) is the most common muscular dystrophy in adults. Characteristics of this multisystem disorder include myotonia, progressive weakness, cardiac conduction delays, fatigue, sleep disturbance, and gastrointestinal dysfunction¹. DM1 is caused by an expanded CTG repeat in the 3′ UTR of the DMPK gene². Clinical features of DM1 arise from expression of DMPK transcripts containing an expanded CUG (CUG^exp) repeat that accumulate in nuclear inclusions of affected tissues^3,4,5,6. This pathogenic RNA readily binds proteins in the muscleblind-like (MBNL) family that are required for normal regulation of alternative splicing, gene expression, transcript stability, and alternative polyadenylation, resulting in partial loss of MBNL protein function^7,8,9,10,11.

Myotonia is a hallmark of DM1 and refers to a stiffness of muscles that results from delayed relaxation after voluntary contraction. In DM1, deregulated alternative splicing of the muscle chloride channel Clcn1 pre-mRNA causes myotonia^12,13. Aberrant inclusion of Clcn1 exon 7a shifts the reading frame, creating a premature termination codon in exon 7 and a truncated, poorly functional protein. The corresponding deficit of chloride ion conductance results in hyperexcitability and involuntary persistence of muscle activity^12,14.

Skeletal muscles are composed of individual myofibers. Expression of myosin heavy chain (MyHC) isoforms serves as a convenient indicator of the structural, functional, and metabolic phenotype of muscle fibers¹⁵. MyHC fiber types 1 and 2 A are rich in oxidative enzymes, fatigue resistant, and specialized for continuous activity. MyHC fiber types 2X and 2B are characterized by glycolytic metabolism, fatigue quickly, and are specialized for phasic activity. In rodents, type 2X fibers also contain moderate-to-high levels of the oxidative enzyme succinic dehydrogenase and demonstrate fatigue resistance that is intermediate between 2 A and 2B fibers^16,17. Peak mechanical power of muscle fibers increases in order from MyHC 1 < 2 A < 2X < 2B¹⁵.

The muscle fiber type profile is determined primarily by nerve activity and can change in response to either neural or hormonal influences¹⁵. Intrinsic contraction of muscle fibers that occurs independently of nerve input also is associated with changes in fiber type. For example, the Arrested development of righting (Adr) mouse model of myotonia congenita develops myotonia due to homozygous loss-of-function mutation in the Clcn1 gene and features a predominance of oxidative fibers in most muscles¹⁸, while a chemically-induced experimental myotonia shifted fiber type from glycolytic to oxidative in rats¹⁹. In cross sectional studies of human DM, the degree of muscle weakness is associated with Type 1 oxidative fiber atrophy and a greater proportion of mechanically less powerful oxidative fibers, which may reflect fiber type transition or a preferential loss of glycolytic fibers^20,21. In this study, we used a double homozygous model of DM1 and antisense oligonucleotide exon skipping to test the hypothesis that deregulated alternative splicing of Clcn1 induces a glycolytic-to-oxidative fiber type transition in DM1.

Results

LR41;Mbnl1 ^-/- double homozygous mouse model of DM1

The human skeletal actin-long repeat (HSA^LR) mouse model of DM1 contains an expanded CTG repeat in the 3′ UTR of a human skeletal actin (ACTA1) transgene. HSA^LR line LR41 features a lower transgene copy number, a lower ACTA1-CUG^exp RNA abundance, deregulated alternative splicing to a lesser degree, less frequent myotonia (40% of mice examined by electromyography/EMG), and milder myopathy than HSA^LR line LR20b^12,22. The Muscleblind-like 1 knockout (Mbnl1^−/−) mouse model of DM1 features a homozygous deletion of Mbnl1 exon 3, leading to deregulated alternative splicing and myotonia to a greater degree than in LR41 or LR20b^8,23,24.

A prior comparison of LR20b and Mbnl1^−/− models found that MBNL1 protein loss-of-function can explain >80% of deregulated alternative splicing but only about 50% of the transcriptional changes initiated by CUG^exp RNA⁹. To test the hypothesis that CUG^exp RNA induces pathogenic effects in muscle tissue that are independent of MBNL1 loss-of-function, we crossed the LR41 transgenic model with the Mbnl1^−/− model to create the LR41;Mbnl1^−/− double homozygous model of DM1 (Supplementary Fig. 1; see Methods). LR41;Mbnl1^−/− double homozygous mice experience prominent, severe myotonia that can be exacerbated in the hindlimbs by briefly grasping the base of the tail (Supplementary Movie 1). By contrast, this maneuver fails to induce visible myotonia in homozygous LR41 littermates, while visible myotonia in LR20b is relatively mild (Supplementary Movie 2). To quantify functional impairment, we monitored spontaneous activity in the x-, y-, and z-planes for 30 min using an acrylic cage and infrared lasers²⁵ in three age groups: 1.5–2.5 months, 3.5–4.5 months, and 6–7 months. In LR;Mbnl1^−/− mice, spontaneous activity was significantly lower than in age-matched LR41 littermates, and progressively worsened with age (Fig. 1a; Supplementary Fig. 1). Mean vertical breaks, the number of single rearing events with one second between each event, was similar at the youngest age, but reduced in LR;Mbnl1^−/− by 31% at the middle age, and 89% lower at the oldest age, while mean vertical counts, the total counts from the z-plane sensor, was reduced in LR;Mbnl1^−/− by 32%, 55%, and 95% in the youngest-to-oldest ages examined, respectively. In LR;Mbnl1^−/−, total distance traveled was lower by 31%, 47%, and 59% from youngest-to-oldest, while rest time was 33% longer in 1.5–2.5 month olds, more than double in the 3.5–4.5 month olds, and more than triple in the 6–7 month olds.