A robust neuromuscular system protects rat and human skeletal muscle from sarcopenia

Rat forelimb muscles are resistant to sarcopenia

In order to characterize how aging differentially affects various locomotor muscles, we performed a cross-sectional analysis of muscle mass in male Wistar rats, where skeletally mature adult controls of 8 months of age were compared to aged animals ranging from 18 months to 24 months. As previously reported [8], hindlimb muscle mass progressively declined with age, reaching approximately 50% atrophy by 24 months of age (Figure 1A). Fast type II myofibers are more susceptible to muscle wasting with age [28, 29]. Consistently, age-dependent muscle atrophy was more pronounced in fast twitch than slow twitch muscles. Nevertheless, all hindlimb muscles analyzed had a similar kinetic of muscle mass decline with age regardless of their fiber type composition and of their contractile involvement in planta- or dorsiflexion, suggesting that a mechanism independent of the contractile phenotype could be an upstream determinant of the susceptibility to age-related muscle atrophy. In contrast, forelimb muscles were protected from age-related muscle atrophy and showed stable muscle mass even at old age (Figure 1B). This striking differential phenotype between hind- and forelimbs suggests that anatomical differences are the primary driver for the susceptibility to sarcopenia. In order to uncover the molecular and cellular mechanisms underlying this differential susceptibility to sarcopenia, we separated the decline in hindlimb muscle mass in 2 phases: an early-sarcopenic phase between 18-20 months of age, where the decline is mild; and a sarcopenic phase between 22-24 months of age where the decline is severe. We then analyzed the fiber size distribution of tibialis anterior (TA) and biceps brachii as prototypical hind- and forelimb muscles, respectively. Fiber cross-sectional area specifically decreased with advancing age in TA while remaining stable in biceps brachii, further demonstrating that forelimb muscles do not undergo aged-related atrophy (Figure 1C-D). Finally, analysis of fiber types revealed that the proportion of slow fibers had slightly increased with age in TA, but not in biceps brachii (Supplementary Figure 1). Cross-sectional area analysis for each fiber type confirmed that fast type IIB fibers were the most prone to age-induced atrophy, and fiber IIB atrophy was only observed in the TA and not the biceps brachii of aged sarcopenic rats (Supplementary Figure 1).

Sarcopenia is diagnosed clinically using both muscle mass and function, with gait speed used as one of the clinical measures for function [30]. In order to assess muscle function in the aging rats, we performed a catwalk analysis to evaluate the natural gait of the animals. Gait speed was significantly decreased with age in early sarcopenic rats, due to both a shorter stride length and increased stand time (i.e. the duration of paw contact on the ground) (Figure 2). Differential analysis of hind- and forelimbs further revealed that aged rats have more difficulties to lift their hindlimbs, but not their forelimbs, as indicated by a reduced swing time with age that is specific for hindlimbs (Figure 2B). Muscle function is therefore impaired in aged rats, and differences in the susceptibility to muscle wasting between forelimbs and hindlimbs directly translate into functional locomotor deficits. Altogether, these data demonstrate that rat sarcopenia is highly regional within the body and that the difference in the susceptibility to sarcopenia can be used to uncouple the mechanisms of aging from those of sarcopenia.

One major difference between hindlimb and forelimb muscles is that they are innervated by fully distinct sets of nerves, which emanate from different regions of the spinal cord. To test whether differential neuromuscular dysfunction could explain the different susceptibility to sarcopenia, we assessed electromyographic responses in a subset of hindlimb and forelimb muscles from adult, early-sarcopenic and sarcopenic rats. The maximal amplitude of the compound muscle action potential (CMAP) was recorded with needle electrodes inserted in muscles from the hindlimb and forelimb, after stimulating the sciatic and radial nerves, respectively. The maximal CMAP amplitude progressively declined with advancing age in the hindlimb gastrocnemius and TA muscles (Figure 3A-B) but was fully preserved in the forelimb biceps brachii and triceps brachii muscles (Figure 3C-D). These results demonstrate that neuromuscular transmission declines specifically with age in muscles susceptible to sarcopenia, and suggests that a causal relationship may exist between impaired neuromuscular function and muscle wasting during aging.

Transcriptomic profiling reveals a neuromuscular signature specifically regulated in muscles susceptible to sarcopenia

Given the regional susceptibility to sarcopenia in rats (Figure 1), differential profiling of hindlimb and forelimb muscle gene expression by RNA sequencing during aging provided a way to dissociate the pathways which are specifically regulated during sarcopenia only in hindlimb muscles, from those commonly regulated with age in both hindlimb and forelimb muscles. Many genes were progressively regulated with age in gastrocnemius with more than 3000 genes regulated during early-sarcopenia and more than 6000 genes regulated in sarcopenic hindlimb muscle (Figure 4A). Gene ontology analyses on aged gastrocnemius muscles confirmed that aging alters important biological processes such as neuromuscular function, extra-cellular matrix remodeling & fibrosis, energy homeostasis & mitochondrial function, and inflammation (supplementary table S1), in agreement with previously reported molecular profiling of sarcopenia in the hindlimb [6, 8]. Strikingly, in the triceps brachii muscle, only 196 genes were significantly regulated with age, out of which almost all were regulated at the oldest age (Figure 4B). In addition, nearly all genes for which expression was affected by aging in triceps brachii were also similarly affected by aging in gastrocnemius (Figure 4B). This highly specific molecular signature of aging in gastrocnemius over triceps brachii therefore confirmed that rat sarcopenia is also regional at the molecular level and demonstrated that molecular perturbations in muscle are linked to sarcopenia rather than age (see supplementary table S2 for the full list of genes differentially regulated by age in gastrocnemius and triceps brachii). Gene ontology analysis of the interaction between age and muscle type, which uncouples sarcopenia from aging, demonstrated that sarcopenia specifically alters neuromuscular function, extra-cellular matrix remodeling and energy homeostasis (supplementary table S3). In contrast, genes commonly regulated in both muscles map to inflammation and immune response categories (supplementary table S4), suggesting that at least this subset of chronic inflammatory signaling is not sufficient to drive sarcopenia.

Aging promotes a global up-regulation of many genes controlling NMJ assembly and function in hindlimb muscles, likely through de-repression of extra-synaptic gene inhibition [8]. Given that both neuromuscular decline and altered gene expression correlated with regional muscle wasting, we further analyzed the molecular perturbations in muscle linked to the NMJ. When we focused on genes regulated with age involved in neuromuscular function (Gene Ontology GO:0031594), we could identify 2 major clusters of genes either up- or down-regulated with age specifically in gastrocnemius (Figure 4C). Importantly, none of these genes were regulated with age in triceps brachii in the genome wide profiling and in the follow-up validation by quantitative PCR (Figure 4D), demonstrating that deregulation of neuromuscular genes during aging occurs specifically in the muscle prone to sarcopenia.

Neuromuscular defects with age are directly associated with the susceptibility to sarcopenia

The molecular and electrophysiological defects of regional neuromuscular decline between hindlimbs and forelimbs were then investigated at the cellular level. Increased fragmentation of the neuromuscular endplates through dispersion of acetylcholine receptors is believed to be a hallmark of impaired NMJ functionality [31]. NMJ fragmentation in forelimb versus hindlimb muscles was assessed by isolating fiber bundles from biceps brachii and Extensor Digitorum Longus (EDL), respectively, and scoring the level of fragmentation from the images where acetylcholine receptors were stained with fluorescently labeled α-bungarotoxin. The fragmentation level of the hindlimb EDL muscle increased with age from ∼20% in adult muscle to ∼70% in sarcopenic muscle (Figure 5A), consistent with pre-vious reports demonstrating hindlimb NMJ fragmentation in aged rodents [31]. In biceps brachii, however, the proportion of fragmented junctions remained stable at all ages (Figure 5B), despite a much higher baseline fragmentation that likely arises from inherent NMJ architecture in different muscles [32, 33]. In order to confirm that differential fragmentation in hindlimb and forelimb muscles were associated with functional differences during neuromuscular aging, we performed retro-labeling studies by injecting the fluorescent retrotracer fluorogold into the gastrocnemius and the triceps brachii of adult or sarcopenic rats and quantifying labeled motoneurons in the spinal cord (Figure 5C). The number of labeled spinal cord motoneurons projecting to the gastrocnemius hindlimb muscle declined dramatically by more than 50% in sarcopenic vs adult rats, while motoneurons innervating the triceps brachii maintained a normal connectivity with age (Figure 5D). These results demonstrate that both the morphology of NMJs and the connectivity of motoneurons from the spinal cord are specifically altered in sarcopenic hindlimb muscles. Taken together, they further highlight a tight correlation between age-related neuromuscular decline and sarcopenia, which suggests that alterations in the neuromuscular system may be causal in the initiation and progression of sarcopenia.

Cholesterol biosynthesis declines differentially in nerves according to muscle susceptibility to sarcopenia

Our results demonstrated that defects in the connectivity from the spinal cord to muscle are correlated with impaired neuromuscular transmission and sarcopenia.

To further investigate whether intrinsic molecular defects in nerves could contribute to the susceptibility to sarcopenia, we performed a gene expression analysis comparing hindlimb sciatic nerve to forelimb radial nerve in adult and sarcopenic animals. As shown in Figure 6A, many genes were regulated with age in both nerves, providing a general signature of nerve aging. Gene ontology analysis of genes differentially regulated with age in the 2 nerve types (i.e. the interaction term of the two by two factorial design) converged on many GO terms associated with cholesterol biosynthesis (Figure 6B). Nerve electric signal conduction and velocity are largely dependent on the lipid-rich myelin sheath produced by Schwann cells around the motoneuron axons [34]. Cholesterol is the major constituent of myelin as it accounts for 25% of total myelin lipids, and is essential for proper myelin synthesis and maintenance [35, 36]. Cholesterol is synthesized via a 37-step process regulated by key rate-limiting enzymes such as HMG-CoA reductase, squalene epoxidase and farnesyl-PP-synthase. When looking specifically at the gene expression level of these enzymes, we found a decrease with age in both sciatic and radial nerves, which decrease was significantly stronger in the sciatic nerve (Figure 6C). These data suggest that cholesterol synthesis is impaired with age in correlation with neuromuscular defects, suggesting that defects in cholesterol synthesis could be an early event in the age-related neuromuscular dysfunction. Interestingly, a substantial number of genes showed stronger regulation with age in the sciatic nerve (Table 1), thereby providing potential mechanisms that could drive the differences in neuromuscular severity between hindlimbs and forelimbs. Of particular interest, some neurotrophic factors and regulators of myelination such as Bdnf and Fgf5 were specifically up-regulated with age in the sciatic but not the radial nerve (Figure 6D), most likely because of compensatory mechanisms that are secondary to the severity of nerve alterations. The stronger up-regulation of the neurotrophin receptor p75NTR in aged sciatic nerve further confirmed that nerves afferent to sarcopenic muscles re-activate neurotrophic gene expression programs as p75NTR controls myelination development and is upregulated in demyelinating motoneuron diseases and nerve injury [37].

Neuromuscular dysfunction correlates with loss of muscle mass in elderly humans

Our results in the rat model of sarcopenia revealed a strong correlation between neuromuscular dysfunction and skeletal muscle atrophy during aging. To test this hypothesis in humans, we evaluated muscle mass and CMAP in the vastus lateralis, a muscle that is particularly susceptible to sarcopenia, and tibialis anterior, which does not normally experience muscle mass decline with advancing age [38]. The decline in muscle size with increasing age in the vastus lateralis (VL) muscle was strongly correlated to decreased CMAP amplitude, demonstrating that both loss of muscle and neuromuscular defects affect this muscle in the elderly (Figure 7A-B). In the tibialis anterior, however, neither muscle size nor CMAP amplitude significantly decreased with increasing age (Figure 7C-D). These results demonstrate that, similar to our observations in the rat, human skeletal muscles with intact neuromuscular transmission are protected from sarcopenia.

Animal experiments

All experimental procedures on animals were in agreement with Swiss and EU ethical guidelines and approved by the local animal experimentation committee of the Canton of Vaud (license number VD2630). Male Wistar rats aged 8 months to 24 months were obtained from Janvier Labs (Le Genest-Saint-Isle, France). Upon arrival all animals were housed by two in standard type 4 cages with ad libitum access to food and water on a 12 hour light/dark cycle at a temperature between 20-24°C and a relative humidity between 50-60%. All rats labeled by age were grouped by date of birth within one month, and further grouping was then based on the muscle phenotype in hindlimbs under the following categories: adult (8-10 months of age), early-sarcopenic (18-20 months of age), and sarcopenic (22-24 months of age). Rats were sacrificed by exsanguination under isoflurane anesthesia and skeletal muscles were dissected free of fat, weighed and snap frozen in liquid nitrogen or processed for histology as described below. A piece of about 1cm of sciatic and radial nerves was dissected out from the mid-thigh or arm region and snap frozen in liquid nitrogen.

Electromyography (EMG) measurements were performed on rats under isoflurane anesthesia. The left limbs were shaved and recording needle electrodes (twisted pairs, wire 150cm, needle 0.4 × 13mm, Neurolite, Switzerland) were sequentially placed into the gastrocnemius, tibialis anterior, triceps brachii and biceps brachii muscles. Supra-maximal electric stimulation was achieved via stimulating needle electrodes sequentially placed around the sciatic nerve and the radial nerve, and the resulting compound muscle action potential (CMAP) was recorded using the Keypoint software (Neurolite, Switzerland).

Gait parameters were measured using the Catwalk XT system (Noldus, Netherlands). Animals were placed to spontaneously walk on a glass lane with tangential illumination and each step was recorded by detecting light diffraction using a camera below the lane. At least 3 runs per animal were acquired with 10 minutes break in between. Locomotion patterns and gait speed were then analyzed from smooth runs, which were defined as runs during which the rat maintained constant speed for at least 3-4 prints per paw, using the automated Catwalk analysis software. All runs were required, manually edited for correct detection of the paws.

Immunofluorescence staining and microscopy

For fiber type analysis, tibialis anterior and biceps brachii were frozen in isopentane cooled in liquid nitrogen, stored at −80°C and cryo-sectioned at 10μm. Sections were incubated 2h in blocking solution made with 4% IgG-free bovine serum albumin (001-000-162, Jackson immunoresearch) + 1% fetal bovine serum in PBS and subsequently incubated with mouse MHC2B (clone BFF3, DSHB) and rabbit laminin (L9393, Sigma aldrich) antibodies overnight. After washing, sections were incubated with donkey anti-mouse-IgM-Cy3 (715-165-020, Jackson immunoresearch) + goat anti-rabbit-A405 (A-31556, Thermo Fisher) secondary antibodies for 1h. A second round of incubations were then performed using MHC2A (clone SC71, DSHB) + MHC1 (clone BAD5, DSHB), and then goat anti-mouse-IgG1-A647 (A-21240, Thermo Fisher) + goat anti-mouse-IgG2b-A488 (A-21141, Thermo Fisher). Images were acquired using a slide scanner (VS-120, Olympus) and analyzed using an in-house MetaXpress journal (Molecular Devices, Sunnyvale, USA). Briefly, fibers were segmented based on laminin staining using a combination of morphological operators and thresholding. Segmentation results were manually controlled and corrected prior to create the final segmentation masks and extract individual fiber cross-sectional area. Intensity thresholds for fiber type staining were manually chosen for each image individually before final quantification. Fibers where the average intensity in the channel of interest was above the threshold were annotated positive for this specific marker. Fibers that were negative for MHC1, 2A and 2B were considered positive for MHC2X. Subsequent data processing was performed by a custom workflow using KNIME Analytic Platform (http://www.knime.org) [56].

For neuromuscular junction analyses, muscles were pinned on cork in PBS and injected with α-bungarotoxin-A488 (1:5000, B-13422 Thermo Fischer) for 30min to stain for acetylcholine receptors. Muscles were then rinsed with PBS and injected with 2% paraformaldehyde solution (P6148, Sigma Aldrich) for 15min, rinsed again and stored at 4°C. Muscles were then separated into bundles of 20-30 fibers, mounted with fluorescence mounting medium (S302380, Dako) and pressed overnight at 4°C. Images were acquired on a Leica DMI 6000B microscope. For each muscle, approximately 50 neuromuscular junctions were imaged and each junction was classified into 3 classes (1-2 fragments, 3-4 fragments or ≥5 fragments) according to the number of fragments counted manually.

Motoneuron retrotracing study

Rats were anesthesized using isoflurane and the right legs were shaved and disinfected with ethanol. Fluorogold (Fluorochrome LLC) was diluted at 2% (w/v) and injected into the gastrocnemius and triceps brachii at 4 to 6 points of injection. The total volume of injection was 60-80 μL per muscle. Ten days later, animals were sacrificed by intracardial perfusion of paraformaldehyde and the spinal cord was dissected out, placed in paraformaldehyde solution overnight at 4°C and subsequently placed in 20% sucrose solution for 3-4 days. The cervical and lumbar regions of each spinal cord were then removed and frozen in OCT embedding medium (361603E, VWR Chemicals). Each sample was cryo-sectioned at 40μm on its entire length and 1 out of 4 sections were examined to count the number of labeled motoneurons.

Gene expression profiling

Reverse transcription and qPCR

Total RNA was extracted using a commercial kit (miRNeasy Mini Kit, Qiagen) according to the manufacturer's instruction, and the concentration was measured using a NanoDrop device. Total RNA extracts were diluted at 20ng/ml and reverse transcribed into cDNA using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) and then diluted 1/2. Quantitative PCR reactions were performed using LightCycler DNA Green Master (Roche) on a LightCycler 384 Real-Time PCR System thermocycler (Roche). The amplification curves were analyzed by the LightCycler 480 SW 1.5 software. HPRT was amplified as qPCR normalization gene.

Human clinical study

The human clinical study was approved by the University of Manchester Research Ethics Committee and was conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent prior to inclusion in the study. Volunteers were included if they were male, and for the vastus lateralis aged between 18-35 years (mean (SD) age 25.2 (4.6) years, height 1.77 (0.06) m, body mass 74.8 (11.1) kg and BMI 23.7 (2.9) kg/m²) or 65-90 years (mean (SD) age 71.6 (6.2) years, height 1.72 (0.07) m, body mass 73.2 (8.7) kg and BMI 24.6 (2.7) kg/m²). For the tibialis anterior (TA) they were aged between 18-35 years (mean (SD) age 27.5 (44.6) years, height 1.77 (0.05) m, body mass 76.7 (10.9) kg and BMI 24.5 (3.7) kg/m²) or 65-90 years (mean (SD) age 74.8 (7.6) years, height 1.71 (0.06) m, body mass 74.7 (7.6) kg and BMI 25.9 (4.02) kg/m²). All participants were physically active and living independently. They were free from neuromuscular, metabolic and cardiovascular diseases (except for controlled hypertension), had not suffered a broken bone within the past two years and reported no mobility problems that affected their ability to complete usual activities of daily living.

Muscle anatomical cross sectional area. Magnetic resonance imaging (MRI) was used to measure peak VL and TA muscle anatomical cross sectional areas using a T1-weighted turbo 3D sequence on a 0.25-T G-Scan (Esaote, Genoe, Italy). The motor point over the VL muscle belly was marked, and contiguous transverse-plane slices of 6mm thickness were collected. Images were analyzed off-line using Osirix imaging software (OsiriX medical imaging, OsiriX, Atlanta, USA) and the cross-sectional at the motor point was recorded.

Surface EMG

The maximal compound muscle action potential (CMAP) was evoked by a manually triggered stimulator (model DS7A; Digitimer, Welwyn Garden City, Hertfordshire, UK) using percutaneous stimulation (Medserve, Daventry, UK). For VL, stimulation was applied to the femoral nerve and a carbon-rubber anode electrode (Dermatrode self-adhering electrode, 5.08cm diameter, Farmadomo, NL) was placed over the skin overlying the gluteus muscle. For TA, the anode was placed on the medial knee joint cleft and stimulation was applied to a superficial part of the common peroneal nerve on the lateral aspect of the shank close to the head of the fibula. For each muscle, the recording surface-EMG electrode was positioned over the motor point of the respective muscle belly to ensure the largest CMAP (negative peak amplitude) with fastest rise-time. For both VL and TA, the stimulator voltage was fixed at 400V and the pulse width at 50μS, and the current was then increased incrementally until the CMAP negative peak amplitude plateaued. At this point the current was increased again by 30mA to ensure supra-maximal stimulation (usually occurring between 100 and 200mA for VL and 80-120mA for TA).

Statistics

All genome wide statistical analyses were performed using R version 3.1.3 and relevant Bioconductor packages as described in the gene expression profiling section. All other statistical analyses were performed using the GraphPad Prism software, and statistical significance was assessed by the Mann-Whitney test for binary comparisons. For comparison of more than 2 groups, one-way ANOVA followed by Bonferroni's multiple comparison test was used. All data are expressed as mean value +/− s.e.m.