Functional and histopathological identification of the respiratory failure in a DMSXL transgenic mouse model of myotonic dystrophy

Myotonic dystrophy or dystrophia myotonica type 1 (DM1) is an autosomal dominant genetic neuromuscular disorder, that can be accompanied by abnormalities in other organs and systems with a highly variable phenotype. DM1 is caused by an expansion of a CTG repeat located in the 3′ untranslated region of the dystrophia myotonica protein kinase gene (DMPK) on chromosome 19q (Fu et al., 1992; Mahadevan et al., 1992). The number of repeats tends to increase from generation to generation, accounting for the genetic anticipation characteristic of this disease (Ashizawa et al., 1992; Mahadevan et al., 1992; Salehi et al., 2007). Healthy individuals have alleles with 5–35 repeats that are stably transmitted, but when this number exceeds 50, instability occurs and leads to disease. The number of CTG repeats is 50–150 in late-onset, mild DM1; 100–1000 in classic DM1; and more than 2000 in the congenital form. A significant inverse correlation is noted between onset age and repeat number and there is a general correlation between the degree of expansion and the severity of clinical manifestations (Gennarelli et al., 1996; Hunter et al., 1992).

The neuromuscular clinical manifestations of the disease include myotonia and progressive muscle weakness, with dystrophic changes in skeletal muscles including the diaphragm, cranial and limb muscles (Harper, 2001). The pathology of DM1, however, extends to other organs and systems and, as such, can be associated with several significant complications.

Respiratory and cardiac problems have long been recognized as the main complications of DM1 patients and are the major factors contributing to mortality. In fact, a study based on a cohort of 115 patients with congenital myotonic dystrophy reported that only about 50% of patients survived to their mid-30s and that the death of about 66% of surviving patients is due to respiratory causes (Reardon et al., 1993). Moreover, large cohort studies have also reported that respiratory problems are the leading cause of death (around 40%) in adult-onset DM1 (de Die-Smulders et al., 1998; Groh et al., 2008; Mathieu et al., 1999). The progressive respiratory muscle weakness leads to impairment of airway clearance, inadequate ventilation and respiratory failure (Berry et al., 1996), whereas no reduction in central airway caliber was found in the patients with the adult form of DM1 (Fodil et al., 2004). DM1 patients frequently develop pneumonia and respiratory failure as a result of alveolar hypoventilation, particularly in the later stages of the disease (Bogaard et al., 1992; Jammes et al., 1986). Despite the many clinical examinations showing respiratory failure in DM1 patients, the pathological mechanisms underlying this failure are still not completely known. Several studies suggest that the causes of respiratory failure are varied and could be due either to the involvement of respiratory muscles in the dystrophic and myotonic process (Bégin et al., 1980; Bogaard et al., 1992; Jammes et al., 1985; Serisier et al., 1982; Zifko et al., 1996) or to abnormalities in the nervous system, which generates and controls breathing drive (Kilburn et al., 1959; Ono et al., 1996; Takasugi et al., 1995).

In neurodegenerative diseases in humans, access to affected tissue is generally limited to the end-stage of the disease and to biopsies, making animal models extremely useful (Gomes-Pereira et al., 2011; Wansink and Wieringa, 2003). In the present study, we have used DMSXL transgenic mice. These mice express the DMPK gene in a variety of tissues with a pattern similar to the patterns of the murine Dmpk gene in mice and the DMPK gene in human tissues (Huguet et al., 2012). To address whether the DMSXL transgenic mice animal model, which carry a long CTG repeat and display DM1 features (Gomes-Pereira et al., 2011), have respiratory problems, we tested and compared the breathing function in DMSXL and control mice. To identify the pathological changes underlying any respiratory complications, we analyzed the diaphragm muscle and the neural network that generates, controls and transmits the respiratory rhythm.

To investigate whether the DMSXL transgenic mice have respiratory alterations, we first measured, under the same conditions, the most relevant respiratory parameters in awake young or adult DMSXL and control (wild-type and DM20) mice. The analysis of breathing patterns (Fig. 1) and parameters recorded in young awake mice (2 months old) revealed a small but significant decrease in tidal volume (TV) and volume per minute (minute volume, MV) in DMSXL mice.

We noticed that awake mice, especially the DMSXL mice, appeared highly stressed and agitated during recording, and so we assessed the respiratory function in anesthetized mice. The results of these experiments showed that in 2-month-old anesthetized mice, the respiratory rate (RR) and the TV were significantly decreased in DMSXL mice compared with control mice taken from the same litter and with DM20 mice that do not express the DM1 phenotype (wild type and DM20). Consequently, the mean value for minute volume/body weight (tidal volume × breathing frequency/body weight) was also reduced in DMSXL mice (Table 1). In addition, a significant reduction was detected in arterial blood oxygen saturation (SpO₂) in DMSXL mice. Overall, we found that alterations in respiratory function observed in young DMSXL mice did not really change as a function of age. In fact, measurement of respiratory parameters in 6-month-old mice confirmed the decrease in RR, TV, MV and SpO₂ in DMSXL mice (Table 2). To be sure that the alterations in respiratory pattern observed in DMSXL mice are not related to their small size, we measured the respiratory function in wild-type mice having similar size and weight (24±0.6 g). The results of these experiments showed no changes in RR and in SpO₂ in the small control mice compared with normal control mice (RR, 135.7±8.2 in small control versus 126±15.1 in normal control; SpO₂, 98.23±0.21 versus 98.62±0.38). This result ruled out the possibility that the respiratory alterations observed in DMSXL mice are caused by their small size. Moreover, the comparison and statistical analysis of all our results revealed two points: first, the breathing function in DM20 transgenic mice, which do not express the phenotype of DM1 disease, was similar to that in wild-type mice; and second, anesthesia caused a reduction in breathing parameters in all mice compared with awake mice.

To assess whether modifications in the diaphragm are responsible for the impairment in respiratory function in DMSXL mice, we first examined diaphragm sections stained for either ATPase activity or with hematoxylin and eosin (H&E). In both control (wild type and DM20) and DMSXL mice, we observed that the majority of fibers did not show phosphorylase activity and were only counterstained with Luxol Fast Blue (>85% type II); a small number of fibers were strongly stained (type I) (Fig. 2). However, quantification of muscle fibers showed an increase (14.7%) in the mean percentage of type I fibers in DMSXL mice compared with control mice. Other dystrophic alterations were observed in DMSXL diaphragm sections, such as centrally located nuclei, an increase in interfasicular connective tissue and the presence of infiltrating inflammatory cells (Fig. 2). The abnormality of DMSXL diaphragm muscle was accompanied by an accumulation of DMPK RNA with its expanded CUG repeat, found as foci in the nuclei of muscle fibers (Fig. 3). About 58% of DMSXL diaphragm muscle fiber nuclei contained between one and seven foci, with an average of 2.2 foci per affected nucleus. No nuclear RNA foci were detected in the diaphragm muscle fibers of control mice.

The diaphragm neuromuscular junctions (NMJs) also underwent pathological changes in DMSXL mice. Despite the presence of bundles of axons in the serial muscle sections, about 20% of the endplates were denervated and had no contact with nerve terminals. Moreover, numerous EPs exhibited lengthened shapes and faint labeling compared with controls (Fig. 4). Statistical analysis of the morphometric results confirmed the abnormality of EPs in DMSXL mice. A significant decrease in mean EP area (11%) and shape complexity (16.3%) was found in DMSXL mice, as well as in the density of acetylcholine receptors (AChRs) on postsynaptic membranes (reduced by 19.9%; Fig. 4). Combined staining with α-BTX and FISH (see Materials and Methods) revealed that several foci were located in sub-synaptic nuclei.

Light microscopy revealed no obvious morphological differences between control (wild type and DM20) phrenic nerve transverse sections of DMSXL mice, except for the smaller size of DMSXL nerve sections. In all animals, the phrenic nerve consisted of a single fascicle containing myelinated and unmyelinated axons. In electron microscopy analysis, apart from some pathological signs (aberrant Schwann cell proliferation, increased number of macrophages) observed on DMSXL thin nerve sections, a severe and significant loss (41%; P<0.01) in the number of unmyelinated axons was found (Fig. 5). By contrast, the number of myelinated fibers showed a slight but not significant decrease (5%; P=0.30), although the thickness of myelin sheath was reduced significantly (P<0.01) in DMSXL mice compared with controls (628±9 nm versus 770±54 nm). FISH analysis showed that several Schwann cells that intermingled with axons contained one or two ribonuclear foci in their nuclei (Fig. 5).

Histological examination of cervical phrenic motor neurons showed rare apoptotic features (nuclear eccentricity, loss of Nissl bodies or cytoplasm granulation). The estimation of the number of neurons by the physical disector method revealed a slight but not significant decrease (5%; P=0.13) in the mean number of cervical phrenic motor neurons in DMSXL mice. In addition, only a few mRNA foci were observed in phrenic motor neurons in DMSXL mice.

The histological and morphometric analysis of five brainstem disector pairs containing respiratory neurons located in a region that extends from the compact portion of the nucleus ambiguus to just behind the facial motor nucleus (Fig. 6) ruled out the presence of neuronopathy in DMSXL mice respiratory neurons. In fact, the histological examination of respiratory neurons located in the area of the ventrolateral nucleus of the solitary tract and neurons scattered in the vicinity of the nucleus ambiguus and the surrounding reticular formation did not reveal apoptotic features. Moreover, the physical disector estimation of neuron density showed only a non-significant reduction (3%) in the density of DMSXL respiratory neurons as compared with controls.

The major result provided by this study is the demonstration that DMSXL mice, a transgenic mouse model of DM1, have respiratory impairments characterized by a significant decrease in RR, TV, MV and SpO₂ compared with control mice. The second important result is the detection of a partial denervation of diaphragmatic EPs and significant pathomorphological changes in diaphragm EPs and muscle fibers in DMSXL mice, which indicate a possible breakdown in communication between the diaphragmatic muscles fibers and the nerve terminals. Also, a severe and significant loss in the number of phrenic unmyelinated afferents was detected in DMSXL mice, whereas no significant alterations were observed in phrenic motor neurons or brain stem respiratory neurons.

Taking into consideration that the diaphragm muscle is the principal respiratory muscle and that afferent phrenic unmyelinated fibers are involved in control of the basic respiratory rhythm, we believe that the pathological changes observed in the diaphragm and phrenic nerve afferents in DMSXL mice are responsible for the respiratory failure assessed in these mice. Because central neurons did not show significant neuronopathy, the generation of the breathing rhythm is probably not affected in DMSXL mice.

For the study of mechanisms of myotonic dystrophy diseases, there is a growing use of various animal models (Gomes-Pereira et al., 2011; Wansink and Wieringa, 2003). Consequently, some of the same techniques used in the clinic to examine patients should be adapted for analysis in animal models. Several studies reported that pressure plethysmography can serve as a valuable respiratory function test in conscious or anesthetized small animals. This non-invasive method is particularly appropriate for quick and repeatable screening of respiratory function over short and extended periods of time for large numbers of mice and rats. Moreover, the intra-individual variability in respiratory function measured by pressure plethysmography is acceptable (Huang et al., 2011; Stunden et al., 2001; Yilmaz et al., 2005). Therefore, in the current study we used pressure plethysmography to assess the respiratory function in DMSXL and control mice. The comparison and the statistical analysis of several breathing measurements demonstrated a significant decrease in most relevant respiratory parameters in anesthetized DMSXL mice, indicating an impairment in breathing function in young and adult mice. Because alterations in the breathing parameters observed in young mice do not become more severe in adult animals, we inferred that respiratory impairment in DMSXL mice does not depend on the age of the animals. We also demonstrated that respiratory changes including RR and SpO₂ were not related to the small weight of the DMSXL transgenic mice. In addition, the hypothesis that the respiratory abnormalities observed in these mice are due to overexpression of the genes carried by the transgene is excluded because the DM20 transgenic mice carrying the same transgene but with a normal number of CTG repeats did not show alterations in respiratory function compared with wild-type mice.

Statistical analysis of the results of several recordings showed that the mean values of respiratory parameters in anesthetized mice were reproducible and comparable with values previously reported (Daubenspeck et al., 2008; Hamelmann et al., 1997; Hoymann, 2007; Huang et al., 2011). Several studies reported that the breathing frequency decreases by about 40–60% in anesthetized mice compared with conscious mice (Hamelmann et al., 1997; Takezawa et al., 1980). In our study, the RR and MV were decreased under anesthesia to about 47% in controls and 34% in DMSXL mice compared with conscious animals, which indicates that DMSXL mice are more sensitive to anesthesia. In agreement, clinical reports have mentioned that, in general, patients with DM1 disease present potential respiratory depression during or after anesthesia, although the mechanisms underlying the anesthesia effect on respiratory depression in DM1 patients is not well investigated (Ogawa et al., 1993; Yomosa et al., 1991). It is conceivable that in DMSXL mice anesthesia acts directly on the neural network that generates, controls and transmits the respiratory rhythm and also acts on the diaphragm, which is the main respiratory muscle. Another possibility is that the anesthesia has an effect on several areas of the central nervous system, which in turn exert an influence on the structures involved in respiratory function.

Because DMSXL mice displayed impairment in respiratory function, these mice present a useful tool for investigating the pathological changes underlying the respiratory failure associated with DM1 disease. Importantly, the evaluation of respiratory parameters by non-invasive pressure plethysmography could be used to evaluate the effect of potential therapies in DMSXL mice.

Anatomical and electrophysiological data suggest that the basic breathing rhythm of higher mammals is generated and regulated in brainstem respiratory neurons. Then, the stimuli are transmitted via cervical phrenic motor neurons and phrenic nerves to the diaphragmatic muscle, which is the principal respiratory muscle (Ramirez et al., 2002). Therefore, the impairment in respiratory function observed in DMSXL mice might have a central origin in decreased respiratory drive, or a peripheral origin such as weakness of the diaphragmatic muscles, or both.

Because diaphragmatic muscle is the only effective respiratory muscle pump and its action contributes to more than 66% of the TV (Mead et al., 1995; Singh et al., 2003), our efforts were focused on analyzing diaphragm muscle fibers and NMJs. Several of the alterations observed in DMSXL diaphragm muscle lead us to believe that they might cause, at least partially, the respiratory failure in these mice. In fact, the increase in histochemically defined type I muscle fibers, the presence of central nuclei, the gain in interfasicular connective tissue and inflammatory cells, as well as the accumulation of expanded CUG mRNA in muscle fiber nuclei all indicate that the diaphragm muscle of DMSXL mice is involved in the dystrophic processes. This assumption can be supported, on the one hand, by the presence of similar histopathological changes in biopsies of DM1 patients (Harper, 1989; Romeo, 2012) and, on the other hand, by the changes in fiber type composition of the diaphragm muscles in an animal model of Duchenne muscular dystrophy, which is characterized by severe muscle degeneration (Guido et al., 2010; Petrof et al., 1993). These authors hypothesize that the progressive changes in fiber type probably permit better adaptation of the muscle to its function during progression of the disease. It is possible that the increased number of type I diaphragm fibers (slow-twitch) in DMSXL mice, due to a selective degeneration of type II fibers (fast-twitch), probably causes alterations in diaphragmatic muscle function in these mice.

The detection of an accumulation of CUG expanded RNA as foci in diaphragm muscle fiber nuclei is another indication of the abnormality of the DMSXL diaphragm. The presence of foci and the sequestration of muscleblind protein in the nuclei diaphragm NMJs has also been reported in DM1 muscle biopsies and in another DM1 transgenic mouse model (Wheeler et al., 2007). It is well known that the main pathogenic process at the base of DM1 is a toxic RNA gain-of-function effect of mutant DMPK transcripts, which are retained in distinct ribonuclear foci within the cell nuclei. Taking into consideration this hypothesis, inferring that these nuclear foci sequester essential proteins and disrupt their normal function in the cell (Kuyumcu-Martinez and Cooper, 2006; Nykamp and Swanson, 2008; Querido et al., 2011), we can hypothesize that nuclear retention of mutant DMPK mRNA in diaphragmatic muscle fibers alters the function of the diaphragm, which can lead to respiratory failure.

In DMSXL mice, the diaphragmatic NMJs also undergo pathological changes. About 20% of diaphragm EPs were denervated and the area and the shape complexity of EPs were significantly reduced. Moreover, the AChR density on postsynaptic membranes was significantly lower. It should be noted that in previous experiments in DM1 transgenic mice carrying only 500 CTG repeats we detected similar abnormalities in the structure of diaphragmatic EPs (Panaite et al., 2008). The fact that the statistical analysis revealed that the pathological changes in diaphragmatic EPs found in DMSXL mice, carrying about 1300 CTG repeats, do not vary significantly from those observed in DM1 mice indicates that there is no relationship between the severity of EP alterations and length of the CTG repeat. This is consistent with clinical studies reporting that some DM1 features such as cataract, sleep disorders, gastrointestinal problems, respiratory insufficiency and cardiac abnormalities do not correlate with the size of the unstable base CTG triplet (Jaspert et al., 1995; Marchini et al., 2000).

It is quite possible that the denervation and abnormalities seen in EP structure are induced by the loss of 5% of phrenic myelinated fibers and also by the alterations affecting muscle fibers. A recent study reported that one of the mechanisms underlying the impairment of neuromuscular connections observed in DM1 tissue cultures is the misexpression of two members of the SLITRK family genes, which are involved in neurite outgrowth, neuritogenesis and synaptogenesis (Marteyn et al., 2011).

Because the normal function of the diaphragm in both inspiratory and expiratory phases of breathing requires neural stimulation via normal NMJs, the damage and alterations to the phrenic nerve affect the diaphragmatic DMSXL NMJs and probably results in failure of action potential transmission (Andreose et al., 1995; Kumai et al., 2005). Numerous clinical data report that patients with signs of muscular weakness and myotonia also show respiratory impairment (Bégin et al., 1997; Jammes et al., 1985; Serisier et al., 1982; Zifko et al., 1996), therefore, the pathological changes observed in DMSXL diaphragm muscle are probably sufficient to affect the normal function of the diaphragm during breathing in these mice.

In addition to the motor function, the phrenic nerve is also rich (43%) in afferent axons composed mainly of unmyelinated axons. The motor component of the phrenic nerve is mainly myelinated whereas the sensory component is mainly unmyelinated (Langford and Schmidt, 1983). The thin afferent fibers are believed to be connected to polymodal receptors on the surface of the muscle and within the muscle. Free nerve endings located in muscle are presumed to represent these receptors (Brown and Fyffe, 1978; Duron and Marlot, 1980; Goshgarian and Roubal, 1986; Kumazawa and Mizumura, 1977). Groups of these fibers are activated principally by chemical stimuli rather than by mechanical stimuli (Kumazawa and Mizumura, 1977).

The tracer combination of the anterogradely transported Fast Blue and the retrogradely transported Nuclear Yellow allowed the demonstration that some phrenic nerve afferents terminate at the C4 and C5 spinal cord in rats (Revelette et al., 1988). Other afferents project to the brain stem nuclei, including dorsal respiratory group neurons (DRG) and ventral respiratory group neurons (VRG) (Larnicol et al., 1985; Macron et al., 1985; Marlot and Duron, 1981). The neurons in the DRG and VRG in the brain stem are primarily responsible for the generation of the rhythmic pattern of respiration. However, there is growing evidence that diaphragmatic afferent stimuli exert a modulating influence on the basic respiratory rhythm, although they do not play a crucial primary role in generating or maintaining the rhythm.

Electrical stimulation of phrenic nerve afferents during inspiration produce a premature transition from the inspiratory to expiratory phase, and it was concluded that phrenic afferents exert an inhibitory influence on inspiratory motor drive (Jammes et al., 1986; Marlon et al., 1988). However, other studies report that the activation of small-fiber afferents in the diaphragm of anesthetized dogs induces a marked excitatory effect on phrenic motor neurons and brain stem respiratory neurons (Revelette et al., 1988; Speck and Revelette, 1987). These studies concluded that the phrenic nerve afferent populations are capable of producing at least two distinct effects: a net excitation of inspiratory activity at the pre-motor level and a strong inhibitory effect on motor output at the spinal level.

In our current study, the statistical analysis of morphometric results showed a severe loss of unmyelinated phrenic afferents (>40%), whereas phrenic myelinated fibers undergo a slight and non-significant reduction (5%). Furthermore, the phrenic motor neurons and the brainstem respiratory neurons do not suffer from significant neuronopathy because only a small non-significant loss in the number of neurons was found. With reference to the data reported in the studies mentioned above, we can deduce that in DMSXL mice the afferent feedback is reduced and the regulation of respiratory drive is altered, whereas the generation of breathing rhythm is probably not affected.

In summary, despite the large cohort studies showing that respiratory problems are the leading cause of death, especially in surviving patients with congenital DM1 (Reardon et al., 1993), and to a lesser degree in adult-onset DM1 (de Die-Smulders et al., 1998; Groh et al., 2008; Mathieu et al., 1999), the pathological mechanisms underlying respiratory failure are still not completely known.

Our study in a transgenic mouse model of DM1 sheds light on possible mechanisms of respiratory failure, because the respiratory impairments seen in these mice are caused by pathological changes affecting mainly the diaphragm respiratory pump and phrenic unmyelinated afferents. Because cervical phrenic motor neurons and brainstem respiratory neurons do not show significant neuronopathy, the generation of the breathing rhythm is probably not affected in DMSXL mice.

All experiments were carried out in accordance with the local veterinary guidelines for care and use of experimental animals and all analyses and counts were performed blinded to the animal genotype.

The generation of transgenic mice carrying the human genomic DM1 region with expanded repeats of either ∼500 CTG (DM1 mice; displaying mild DM1 phenotype) or 20 CTG (normal; do not develop DM1 phenotype) has been described previously (Gantelet et al., 2007; Seznec et al., 2001). DMSXL mice carrying more than 1300 CTG were obtained from DM1 mice, after large expansions of the CTG repeat over successive generations (Gomes-Pereira et al., 2007; Panaite et al., 2011). Expression of expanded CUG leads to the formation of numerous foci that co-localize with muscleblind-like protein 1 and 2 (MBNL1 and MBNL2). Mild missplicing of target RNA is observed in muscles and heart tissues. These molecular features of DM1-associated RNA toxicity were associated with high mortality, growth retardation and muscle defects (abnormal histopathology, reduced muscle strength and lower motor performances) (Huguet et al., 2012).

Heterozygous mice, homozygous and wild-type mice are obtained by breeding from the same litter. Only homozygous transgenic mice were used in our study because the heterozygous mice expressing a low level of DMPK transcripts do not have an obvious phenotype. Thirty mice were used: group 1 consisted of five 2-month-old DMSXL mice and five control mice taken from the same litter (body weight 20.2±1.1 g versus 27.6±0.8 g); group 2 consisted of five 2-month-old DM20 mice and five control mice (body weight 28.0±0.6 g versus 28.5±0.8 g); and group 3 consisted of five 6-month-old DMSXL mice and five control mice (body weight 22.0±2.0 g versus 33.8±3.6 g).

To investigate whether the DMSXL mice have respiratory failure, we measured the breathing function in DMSXL and control mice using pressure plethysmography (Respiromax system, Columbus Instruments, Columbus, OH), a precise method adapted for both awake and anesthetized small animals (Stunden et al., 2001; Yilmaz et al., 2005). Individual mice were first weighed then placed in a respiromax cylindrical chamber. A tail rod was moved forward until it secured the animal, preventing it from backing up and maintaining a prone position. The head of the animal emerged through an inflatable latex cuff into a head chamber and the neck cuff was pressurized to seal the body chamber. Each mouse was allowed to acclimatize to the plethysmography chamber for ∼5 minutes before tests began. A sensitive transducer measured the changes in pressure in the body chamber caused by the animal’s respiration. Signals from the pressure transducer were amplified and digitized by a respiratory function software (Columbus Instruments), displayed and stored in a computer for graphical and statistical analysis. In all experiments, the temperature of the body chamber was continuously monitored using a thermal probe. The following parameters were evaluated: tidal volume (ml), respiratory rate (breaths/minute), minute volume (tidal volume multiplied by respiratory rate, ml/minute). The respiratory function of each mouse was tested several times over a period of 2 weeks.

In another series of experiments, the respiratory function was measured in mice anesthetized with isoflurane using a Matrx Quantiflex low flow V.M.C. anesthesia system (Midmark, Versailles, OH). The animals were placed individually in an induction chamber and anesthesia was induced with 5% isoflurane in a gas mix of O₂/N₂O₂ (30/70%). The animals were then quickly placed inside the Respiromax system, with a 1.5% isoflurane flow in the head chamber in the same O₂/N₂O₂ gas mix. A MouseOx pulse oximeter (Starr Life Sciences, Oakmont, PA) was placed on the upper, right hindleg of each mouse to measure the oxygen saturation of arterial hemoglobin (SpO₂) and heart rate; the body chamber was closed and the neck cuff swollen. Animals were allowed to settle for 5 minutes before the first set of measurements.

After testing the respiratory function, mice were deeply anesthetized and the brainstem, cervical spinal cord, right and left phrenic nerves and diaphragm muscles carefully removed from each animal as previously described (Panaite et al., 2008; Panaite et al., 2011).

To examine diaphragm muscle fibers, 10-μm thick transverse cryostat sections were prepared and myofibrillar adenosine triphosphatase (ATPase) activity assessed after pre-incubation at pH 4.2 (Sheehan and Hrapchak, 1980) followed by counterstaining with Luxol Fast Blue. Other sections were stained with hematoxylin and eosin (H&E). The number of each fiber type was assessed using an image-processing program (ImageJ 1.40, National Institutes of Health, Bethesda, MD). To analyze the NMJs, serial longitudinal cryostat sections (20-μm thick) of diaphragm muscle were double labeled with tetramethylrhodamine-conjugated α-bungarotoxin (α-BTX; Invitrogen, Life Technologies, Carlsbad, CA,) and with a primary polyclonal antibody directed against the 200-KDa neurofilament protein (AB1982, Millipore, Billerica, MA).

Muscle sections were observed under a Zeiss AxioPlan 2 microscope (Zeiss; Oberkochen, Germany), photographed using an image-acquisition program (Zeiss AxioVision with Axiocam HRc) and images systematically analyzed in the ImageJ 1.40 program. The percentage of denervated endplates (EPs) was estimated and the morphometric parameters (area, shape complexity and fluorescence intensity of rhodamine-α-BTX labeling) of each EP were measured and calculated as described in our previous studies (Panaite et al., 2008; Panaite et al., 2011). More than 1500 EPs were measured from each mouse line (n=7 DMSXL and control mice).

Fluorescence in situ hybridization (FISH) was performed on 7-μm thick frozen sections as described by Guiraud-Dogan et al. (Guiraud-Dogan et al., 2007). In short, sections of diaphragm muscle, phrenic nerve, cervical spinal cord and brain stem were dried, fixed, then incubated in 30% (v/v) formamide and 2× standard sodium citrate (SSC) for 10 minutes, hybridized with the PNA probe Cy3-OO-CAGCAGCAGCAGCAG (1 ng/μl) (Eurogentec, Liège, Belgium) for 2 hours at 37°C in buffer [30% (v/v) formamide, 2× SSC, 0.02% (w/v) BSA, 66 μg/ml yeast tRNA (Invitrogen) and 2 mM vanadyl complex (Sigma-Aldrich, St Louis, MO)]. Sections were then washed for 30 minutes in 30% (v/v) formamide and 2× SSC at 50°C and then for 30 minutes in 1× SSC at 20°C. Nuclei were stained by incubating with DAPI for 15 minutes at room temperature. The cervical spinal cord, phrenic nerves and brain stem sections were additionally labeled with polyclonal anti choline-acetyltransferase antibody (ChAT, 1:50; AB144P, Millipore) as previously described (Panaite et al., 2011).

For light microscopy analysis, semithin transverse sections (1 μm) were cut at different levels from phrenic nerve trunks and stained with Toluidine Blue. For electron microscopy examination, ultrathin sections (80 nm) were cut and contrasted with uranyl acetate and lead citrate. Sections were viewed in the Zeiss EM 10C electron microscope and morphometric analysis was performed as previously described (Gantelet et al., 2007; Panaite et al., 2008). The number of myelinated and unmyelinated axons was counted and the thickness of the myelin sheath calculated (n=7 for each DMSXL and controls).

Serial transverse cryostat sections (20 μm) were prepared from cervical spinal cords (C3–C5) and from brainstems (caudal and rostral to the obex). Sections were either immunolabeled with polyclonal anti-ChAT antibody, which labels motor neurons, or stained with Toluidine Blue. The physical disector method was used to estimate the number of motor neurons in 2-mm segments of either cervical spinal cord or the medullary reticular formation. Identification of respiratory neurons on brainstem sections was achieved according to their localization, as described in several studies (Alheid et al., 2002; Kuwana et al., 2006; Stornetta, 2008). The physical disector method is based on sampling sections, called disector pairs. Four or five disector pairs were used from each animal to estimate the density of neurons, as previously described (Panaite et al., 2011).

Each respiratory parameter for each mouse (TV, RR, MV, MV/g, heart rate, oxygen saturation) and the data obtained from the diaphragm, phrenic nerves, cervical spinal cord and brainstem examinations were analyzed and compared. ANOVA was used to determine significant difference between the groups, followed by Bonferonni-Holm post-hoc test. Pair-groups were compared by Student’s t-test. In all cases, plethysmograph data were normalized to the weight of the animal. Values are reported as mean ± s.d.; P<0.05 was considered a significant difference. All statistical analyses and histograms were performed using GraphPad Prism software (GraphPad Software, La Jolla, CA).

We wish to thank Dr M. Price for critical reading of the manuscript, and Profs J-P. Hornung and E. Welker for their help in the analysis of the brainstem sections.