Ciliary neurotrophic factor-induced sprouting preserves motor function in a mouse model of mild spinal muscular atrophy

Ciliary neurotrophic factor-induced sprouting preserves motor function in a mouse model of mild spinal muscular atrophy Christian M. Simon1, Sibylle Jablonka1, Rocio Ruiz2, Lucia Tabares2and Michael Sendtner1, 1 Institute for Clinical Neurobiology, Josef-Schneider-Str. 11, 97080 Wuerzburg, Germany and 2 Department of Medical Physiology and Biophysics, School of Medicine, University of Seville, 41009 Seville, Spain Received August 28, 2009; Revised November 7, 2009; Accepted December 15, 2009 Proximal spinal muscular atrophy (SMA) is caused by homozygous loss or mutation of the SMN1 gene on human chromosome 5. Depending on the levels of SMN protein produced from a second SMN gene (SMN2), different forms of the disease are distinguished. In patients with milder forms of the disease, type III or type IV SMA that normally reach adulthood, enlargement of motor units is regularly observed. However, the underlying mechanisms are not understood. Smn 1/2 mice, a mouse model of type III/IV SMA, reveal progressive loss of motor neurons and denervation of motor endplates starting at 4 weeks of age. Loss of spinal motor neurons between 1 month and 12 months reaches 40%, whereas muscle strength is not reduced. In these animals, amplitude of single motor unit action potentials in the gastrocnemic muscle is increased more than 2-fold. Confocal analysis reveals pronounced sprouting of innervating motor axons. As ciliary neurotrophic factor (CNTF) is highly expressed in Schwann cells, we investigated its role for a compensatory sprouting response and maintenance of muscle strength in this mouse model. Genetic ablation of CNTF results in reduced sprouting and decline of muscle strength in Smn 1/2 mice. These findings indicate that CNTF is necessary for a sprouting response and thus enhances the size of motor units in skeletal muscles of Smn 1/2 mice.Thiscompensatory mechanism could guide the way to new therapies for this motor neuron disease. INTRODUCTION Spinal muscular atrophy (SMA) is an autosomal recessive disorder with various phenotypes, ranging from severe forms that lead to rapid paralysis and early death in type I patients to milder forms (type III and type IV) in which patients reach normal age despite impairment by muscle paralysis (1,2). A common feature of these various forms is the loss of a -motor neurons in the spinal cord, which causes muscle weakness and muscle atrophy, in particular, in the proximal muscle groups of the body axis (3). All these disease forms are caused by homozygous deficiency and/or mutation of the survival motor neuron gene (SMN1) on human chromosome 5, which encodes for the ubiquitously expressed SMN protein (4). It plays a role in the assembly of small nuclear ribonucleoprotein particles (U snRNPs) (5–8), the central components of the spliceosomes. Furthermore, the SMN protein has an additional function, particularly in motor neurons, in promoting presynaptic differentiation by contributing to the translocation of the b-actin mRNA to the presynaptic compartment (9). Thus, Smn-deficiency leads to highly reduced b-actin protein levels and defective clustering of voltage-gated Ca 2þ channels in axonal growth cones in motor neurons isolated from a mouse model of SMA type I (10). Defects in presynaptic components of neuromuscular junctions and corresponding defects in neuromuscular transmission are also observed in early postnatal mice with this disease (11). This report demonstrated that the loss of synaptic vesicles occurred from active zones in a mouse model of type II SMA, and that the depletion of synaptic vesicles correlated with increased phosphorylated neurofilament in this mouse model. Dysorganized cytoskeletal elements such as increased levels of neurofilament in axon terminals are prominent in mouse models of SMA (12). Recent reports confirm these findings and provide evidence that the accumulation of neurofilaments occurs as part of a dying back process in mouse models of SMA (13).  To whom correspondence should be addressed. Tel: þ49 93120144000; Fax: þ49 93120144009; Email: [email protected] #The Author 2009. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected] Human Molecular Genetics, 2010, Vol. 19, No. 6 973–986 doi:10.1093/hmg/ddp562 Advance Access published on December 18, 2009 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 In humans, SMA type I becomes clinically apparent within 6 months after birth, SMA type III in the second or third year after birth (14). Disease progression differs between type I and type III SMA. Type I patients die within few years after birth, type III patients normally reach adult age. In some of these patients, disease progression slows down with age, and compensatory mechanisms are thought to underlie this phenomenon (3,14). Mice with inactivation of one allele of Smn that show a 50% reduction of SMN protein levels exhibit progressive postnatal loss of spinal motor neurons of up to 50% at an age of 12 months (15), although these mice remain phenotypically unaffected. This indicates that motor function does not correlate with the loss of spinal motor neurons, and that compensatory mechanisms are active in milder forms of SMA but not in the severe forms that lead to early postnatal death. In order to identify the underlying compensatory mechanisms, we analysed Smn þ/2 mice electrophysiologically and histopathologically to characterize single motor units. Along with a reduction of motor neurons by over 40%, the size of the average single motor unit action potential (SMUAP) in the gastrocnemic muscle was increased by a factor of at least 2. To investigate how this electrophysiological observation corresponds to changes at the structural level, we analysed axon terminals in the gastrocnemic muscles. We found increased sprouting, a compensatory way to restore nerve– muscle connectivity. CNTF is highly expressed in Schwann cells close to neuromuscular endplates. In order to characterize the role of this neurotrophic factor in this context, Smn þ/2 mice were crossbred with Cntf 2/2 mice, and the resulting double mutants investigated by electrophysiology, immunohistochemistry and analysis of muscle strength. In agreement with previous data showing that CNTF can induce terminal sprouting and sprouting from nodes of Ranvier within denervated muscle (16,17), our data suggest that CNTF is necessary for the intramuscular sprouting observed in Smn þ/2 mice, and thus prevents a drop of muscle strength. This explains why disease progression is slow in type III SMA mouse models in comparison to type I SMA in which sprouting and enlargement of motor units does not normally occur. This finding could guide the way to new therapies in SMA. RESULTS Increased mean motor unit size in the gastrocnemic muscle of Smn 1/2 mice Smn þ/2 mice do not develop signs of muscle weakness despite the loss of at least 40% of spinal motor neurons (L1–L7) at an age of 12 months (Fig. 1A and B) (15). To get an insight into the underlying compensatory mechanisms, we analysed motor function of the gastrocnemic muscle by electromyographic (EMG) measurements (18) in 12-month-old Smn þ/2 and Smn þ/þ littermate controls. We have chosen the gastrocnemic muscle because it contains large motor units and is more affected than i.e. the small finger muscles in SMA type III patients. Moreover, the innervation pattern of this muscle has been studied in detail (19), and this muscle is easily accessible for electrophysiological analysis and analyses of alterations of motor units such as sprouting (16). We first stimulated the sciatic nerve by a single current pulse of supramaximal amplitude and recorded the maximum compound motor action potential (CMAP), which directly relates to the total number and size of muscle fibres. Despite the loss of lumbar motor neurons in Smn þ/2 mice at this age (15), the mean amplitude of the maximum CMAP in the Smn þ/2 mice was not different (data not shown). This finding corresponds to the lack of weakness and normal gross morphology of the gastrocnemic muscle (Fig. 1B and Supplementary Material, Fig. S1) in 12-month-old Smn þ/2 mice. We then investigated the size of the motor units by applying successive incremental stimuli (Fig. 1C–E), starting from a subthreshold level, until recruiting 12 individual responses. As shown in Fig. 1C and D, the size of the amplitude increments was larger in Smn þ/2 than in Smn þ/þ control mice. The average amplitude of the responses, measured as the mean single motor unit action potential (SMUAP), was more than three times higher in Smn þ/2 mice than in controls (Fig. 1E), indicating that this massive enlargement of the mean motor unit size is responsible for maintenance of muscle strength in Smn þ/2 mice. This indicates that compensatory mechanisms allow muscle fibres to remain active in this mouse model of SMA type III despite the loss of motor neurons in the spinal cord (15). Enhanced arborization of nerve endings in gastrocnemic muscle in 12-month-old Smn 1/2 mice Twelve-month-old Smn þ/2 mice exhibit a reduction in lumbar spinal motor neurons and simultaneously increased amplitudes of SMUAPs in the gastrocnemic muscle. In order to study the underlying structural alterations in motor axons innervating the skeletal muscle and in neuromuscular junctions, we crossed Smn þ/2 mice with mice expressing the YFP protein under the Thy1 promoter (line YFP-H) in individual motor neurons (20). In these mice, less than 10% of motor neurons express the fluorescent tracer. This allows tracking all terminals of individual motor neurons within the gastrocnemic muscle. We isolated muscles from Smn þ/2 thy1-YFP-H tg and Smn þ/þ thy1-YFP-H tg mice. After counterstaining with Alexa594-conjugated a -bungarotoxin, nerve fibres were visualized in whole mount gastrocnemic muscles by confocal microscopy. The gastrocnemic muscle is innervated by at least three major branches from the tibial nerve (19). We quantified the number of branching points of individual axons in the upper medial branch of the tibial nerve (white arrow in Fig. 2A), because it could be reproducibly identifiedand analysed intheisolated gastrocnemic muscles. This branch comes from the main trunk and enters the medial gastrocnemic muscle (Fig. 2B). We traced individual axons back from neuromuscular endplates tothe trunk of the nerve and counted the number of branching points. Quantification is shown in Fig. 2C. In Smn þ/2 gastrocnemic muscle, enhanced branching and sprouting of nerve terminals became detectable (Fig. 2C and D). The number of axons with 4, 5 and6 branches is increased in Smn heterozygous deficient mice, and the number of axons with one to three branches is decreased, indicating that enhanced axonal sprouting of terminal motor fibres could be responsible for re-innervation of denervated muscle fibres, due to the loss of spinal motor neurons in the corresponding region of the lumbar spinal cord. 974 Human Molecular Genetics, 2010, Vol. 19, No. 6 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 Development of motor neuron loss, denervation of neuromuscular endplates and compensation by sprouting in Smn 1/2 mice In order to find out when denervation and re-innervation of individual neuromuscular endplates first occurs in Smn þ/2 mice, we counted motor neuron numbers in the lumbar spinal cord (L1–L7) in 4-week and 6-month-old Smn þ/2 and control mice (Fig. 3). In 4-week-old mice, motor neuron numbers were not reduced (Fig. 3A), whereas 23% loss of motor neurons was observed in 6 month and 42% in 12-month-old Smn þ/2 mice, thus confirming previously published data (15). This corresponded to enhanced fibre grouping in 12-month-old Smn þ/2 mice (Supplementary Material, Fig. S2), as visualized by ATPase staining of frozen gastrocnemic muscle sections at pH 4.3 that marks slow twitch fibres. However, despite the loss of motor neurons, no loss of muscle strength was observed in 4-week-, 6-month- and 12-month-old mice (Fig. 3B). First signs of denervation of neuromuscular endplates were already observed in 4-week-old Smn þ/2 mice (Fig. 3C and D). In 4-week-old gastrocnemic muscles, individual neuromuscular endplates became detectable in which the postsynaptic staining with fluorescence-coupled BTX was not covered by immunofluorescence against neurofilament light chain (NFL), a presynaptic marker of innervating motor axons (Fig. 3D). NFL antibodies were applied in this experiment in order to avoid potential bias caused by the selective expression of YFP. In contrast to NFL antibodies that label all neuromuscular endplates, Figure 1. Characterization of motor neuron numbers in lumbar spinal cord and motor unit size in the gastrocnemic muscle of Smn þ/2 mice. (A) Counts of motor neuron cell bodies in the lumbar spinal cord (L1–L7) of 12-month-old Smn þ/þ (n¼4) and Smn þ/2 (n¼5) mice. (B) Grip strength in 12-month-old Smn þ/þ (n¼5) and Smn þ/2 (n¼8) mice. (Cand D) Representative examples of SMUAP amplitude increments in gastrocnemic muscle of 1-year-old control (C) and in Smn þ/2 heterozygous mice (D) in response to 12 nerve stimuli of increasing amplitude after stimulation and recording with needle electrodes. (E) Normalized Mean +SEM SMUAP in WT and Smn þ/2 mice (n¼7). P,0.05, P,0.01, P,0.001. Human Molecular Genetics, 2010, Vol. 19, No. 6 975 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 Figure 2. Enhanced arborization and terminal sprouting of motor fibres innervating the gastrocnemic muscle in Smn þ/2 mice. (A) Overview of a mechanically squeezed gastrocnemic muscle and the arborization pattern of the innervating tibial nerve in the muscle whole mount preparation. The arrow points to the investigated area. (B) The upper medial branch of the tibial nerve in Smn þ/þ thy1-YFP-H tg and Smn þ/2 thy1-YFP-H tg mice. Scale bar ¼250 mm, arrowheads point to sprouting events. (C) Quantification of the arborization level (white bars represent data from Smn þ/þ thy1-YFP-H tg mice and grey bars from Smn þ/2 thy1-YFP-H tg mice). Individual axon terminals (at least n¼50) were traced back to the main medial branch innervating the MGC muscle as outlined in (A), and the number of branching points were counted. Levels of axon arborization reflect formation of additional branches distal of the main trunk. In Smn þ/2 gastrocnemic muscle, axons with additional branching points, in particular in the distal segments, are significantly more abundant than in Smn þ/þ muscle, indicative of enhanced axonal sprouting. In addition, in Smn þ/2 gastrocnemic muscle, more axons are present that show additional sprouts close to neuromuscular endplates (arrowheads) (n¼3 independent animals for each group, muscles from both sides were investigated from each animal). P,0.05, P,0.01, P,0.001. (D) High power micrographs of sprouting axons. Arrowheads point to sprouting. The panel in the middle shows terminal sprouting. Scale bar ¼20 mm. 976 Human Molecular Genetics, 2010, Vol. 19, No. 6 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 Figure 3. Development of motor neuron loss, denervation and re-innervation of neuromuscular endplates in Smnþ/2mice. (A) Motor neurons were counted in the lumbar spinal cord (L1–L7) of 4-week-, 6-month- and 12-month-old Smn þ/2 and Smn þ/þ mice (n¼at least 3 for each group). Whereas motor neuron numbers appeared unaffected in 4-week-old Smn þ/2 mice, a significant loss was observed in 6- and 12-month-old Smn þ/2 mice. (B) Grip strength was unaffected in Smn þ/2 mice at any stage between 4 weeks and 12 months, indicating that the loss of motor neurons was functionally compensated in Smn þ/2 mice. (C) A significant increase in the frequency of axon terminals innervating two neuromuscular endplates was observed in 6- and 12-month-old Smn þ/2 mice. A first tendency of enhanced sprouting, although not significant, was also observed in 4-week-old Smn þ/2 mice. At least 150 axon terminals were investigated per animal in the gastrocnemic muscles from both sides (n¼at least three animals per group). (D) The figure shows examples of denervated neuromuscular endplates in 4-week-old (upper two panels) and 6-month-old (lower four panels) Smn þ/2 mice. Morphology of neuromuscular endplates in corresponding control animals (Smn þ/þ mice from the same litters) is also shown. Denervation is shown by lack of covering of the postsynaptic side of neuromuscular endplates, stained with BTX by innervating axonal branches that are stained with antibodies against neurofilament light chain. The corresponding 3D reconstructions generated from the original confocal stacks with Imaris software are shown on the right side. Individual denervated neuromuscular endplates are marked by arrows. Scale bar ¼10 mm. P,0.05, P,0.01, P,0.001. Human Molecular Genetics, 2010, Vol. 19, No. 6 977 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 YFP is expressed in less than 10% of the motor neurons in Thy1-YFP-H mice, a subgroup that is potentially not representative of all lumbar motor neurons. We counted denervated neuromuscular junctions and found 8 from 334 (2.4%, in comparison to 1.7% (6/344) in Smn þ/þ mice) in the gastrocnemic muscle of 4-week-old Smn þ/2 animals. At least three independent animals were investigated, with at least 80 neuromuscular junctions screened per animal. Denervation of neuromuscular endplates was more prominent (15/293 corresponding to 5.1%) in 6-month-old Smn þ/2 mice (Fig. 3D), in comparison to 7/280 (2.5%) in age-matched Smn þ/þ mice. For analysis of sprouting, more than 150 endplates in different areas of the gastrocnemic muscle were investigated in order to quantify synapses innervated by axon terminals that had developed terminal branches by sprouting, thus innervating two or more nearby endplates (see also Fig. 5D and E). Compensatory mechanisms by enhanced terminal arborization and sprouting already occurred in 4-week-old Smn þ/2 mice. In 4-week-old Smn þ/2 mice, a tendency towards enhanced sprouting was observed (Fig. 3C), and the difference became significant in 6-month-old Smn þ/2 mice. This confirms previously published data that the disease process in mouse models for SMA reflects a dying back process (13). Our data indicate that denervation of neuromuscular endplates occurs early starting in 4-week-old Smn þ/2 mice before the loss of motor neuron cell bodies becomes apparent in the lumbar spinal cord. No loss of muscle strength was observed in 4-week-, 6-month- or 12-month-old mice, indicating that the mechanism of compensation for denervation of individual neuromuscular endplates and motor neuron loss are robust in these mice. Ciliary neurotrophic factor is located in Schwann cells of innervating motor axons in Smn 1/2 gastrocnemic muscle In adult mammals, myelinating Schwann cells express high levels of the neurotrophic factor ciliary neurotrophic factor (CNTF). CNTF has also been shown to protect axonal destruction in pmn mice, a mouse model for amyotrophic lateral sclerosis (ALS) (21–23). Furthermore, CNTF application protects maintenance of innervation of motor endplates in SOD G93A mice, a mouse model of familial ALS (19). Therefore, we tested expression of CNTF in Schwann cells surrounding motor axons in the gastrocnemic muscleof Smn þ/2 mice. CNTF immunoreactivity is prominent both in myelinating Schwann cells and in Schwann cells close to neuromuscular junctions (Fig. 4A and B). The specificity of this staining was tested by the analysis of neuromuscular endplates from Smn þ/2 Cntf 2/2 and Smn þ/þ Cntf 2/2 mice from the same litters (Fig. 4A, third and fourth panel from top). These nerve fibres did not show any CNTF immunoreactivity in surrounding Schwann cells. CNTF expression was also detected in the cytoplasm of myelinating Schwann cells in the sciatic nerve of Smn þ/2 mice (Fig. 4C), as previously observed in other models of motor neuron diseases such as pmn (24) and SOD1 G93A mutant mice (25). Reduced axonal sprouting in Smn 1/2 Cntf 2/2 skeletal muscle To investigate whether the loss of CNTF reduces compensatory sprouting that is detectable in Smn þ/2 muscle by morphological and electrophysiological techniques (Figs 1E and 2C), we crossbred Smn þ/2 with Cntf 2/2 mice in order to obtain Smn þ/2 Cntf 2/2 and corresponding Smn þ/2 and Cntf þ/þ control mice. First, we counted numbers of motor neuron cell bodies in the spinal cord (L1–L7). Previous studies (26–28) have shown that Cntf 2/2 mice exhibit loss of motor neurons between 4 weeks and 6 months. Similar losses of motor neurons were observed in the Cntf 2/2 mice used in this study (Fig. 5A). Four-week-old mice of all genotypes do not show motor neuron loss (Fig. 5A, Supplementary Material, Fig. S3), and the grip strength was normal in all four genotypes (Fig. 5B). A tendency of enhanced sprouting in Smn þ/2 was detectable (Fig. 5C). Smn þ/2 ,Smn þ/2 Cntf þ/þ and Smn þ/2 Cntf 2/2 show a significant loss of motor neurons at 6 and 12 months in comparison to controls (Fig. 5A and Supplementary Material, Fig. S4). However, there were no additive effects of CNTF and Smn deficiency. The difference between Smn þ/2 Cntf þ/þ and Smn þ/2 Cntf 2/2 was not significant at any age studied. Functional analysis by grip strength measurements (28) revealed a reduction in muscle strength of 15% in Cntf 2/2 and Smn þ/2 Cntf 2/2 mice. However, although Smn þ/2 Cntf þ/þ mice exhibit even higher losses of spinal motor neurons than Cntf 2/2 mice, there was no reduction in muscle strength at 6 and 12 months. In contrast, 6- and 12-month-old Smn þ/2 mice that were CNTF-deficient showed reduced muscle strength (Fig. 5B). To address the question whether CNTF-induced sprouting provides a functional compensation for motor neuron losses in Smn þ/2 Cntf þ/þ mice, we investigated the innervation pattern of endplates in the gastrocnemic muscle. Therefore, terminal axons were stained with antibodies against neurofilament. More than 150 endplates in different areas of the gastrocnemic muscle were investigated in order to quantify synapses innervated by axon terminals that had developed terminal branches by sprouting, thus innervating two or more nearby endplates. Figure 5D shows a typical innervation pattern found in wild-type control muscles. Some synaptic areas are magnified. In the upper panel (Fig. 5D, 1–4), examples of axons are shown that innervate only one endplate (no sprouting). Figure 5D (5–7) and E shows examples of axons innervating at least two endplates (sprouting). Applying these criteria, we investigated endplates with the nearby axon innervating more than one synapse in Smn þ/þ ,Smn þ/2 Cntf þ/þ ,Cntf 2/2 and Smn þ/2 Cntf 2/2 . In 6- and 12-month-old Smn þ/2 Cntf þ/þ gastrocnemic muscles, the number of axon terminals innervating two endplates (Fig. 5C) is significantly enhanced by a factor of about 2 (Fig. 5C) compared with control Smn þ/þ mice. In CNTF-deficient mice, only few terminal sprouts were observed, both in Cntf 2/2 and Smn þ/2 Cntf 2/2 mice. These data indicate that CNTF induces sprouting and compensates for the loss of muscle strength in Smn þ/2 Cntf þ/þ mice. Reduction of the mean single motor unit action potential size in Smn 1/2 Cntf 2/2 muscle to wild-type levels correlates with reduced muscle strength To investigate the functional consequences of reduced sprouting after deletion of the Cntf gene in Smn þ/2 mice, we measured muscle strength (Fig. 5B) and the soma size of lumbal motor 978 Human Molecular Genetics, 2010, Vol. 19, No. 6 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 Figure 4. Localization of CNTF immunoreactivity in control and Smn þ/2 muscle and nerves. (A) Localization of CNTF in 100 mm thick longitudinal frozen sections of the gastrocnemic muscle from Smn þ/þ ,Smn þ/2 ,Smn þ/2 Cntf 2/2 double mutant and Cntf 2/2 mice. Triple staining against a -bungarotoxin (BTX, blue), neurofilament (NF, green) and CNTF (CNTF, red) was performed. Individual channels for CNTF, NF and BTX staining are shown, as well as an overlay (OV) on the right side. In Smn þ/þ and Smn þ/2 muscle, CNTF is located in Schwann cells surrounding the axons. As a control, the CNTF signal was undetectable in Smn þ/2 Cntf 2/2 double mutant and Cntf 2/2 single mutant muscle tissues. Scale bar ¼10 mm. (B) High magnification of a Smn þ/2 neuromuscular endplate to demonstrate CNTF expression in Schwann cells. Scale bar ¼10 mm. A 3D reconstruction of the triple staining, generated by Imaris software from the complete confocal stack, is shown in the lower panel. CNTF immunoreactivity is detectable at high levels around the axons before they enter and branch into the pretzel like structure of the neuromuscular endplate. (C) Double staining for CNTF and S100 in frozen sections of the sciatic nerve from Smn þ/2 and Cntf 2/2 mice. A S100 antibody was used as a Schwann cell marker in combination with the CNTF antibodies to confirm the localization of CNTF in the Schwann cells, as seen in Fig. 3A. CNTF colocalizes with S100 in the cytoplasm of Schwann cells and in Smn þ/2 nerves. No signal for CNTF was detected in Cntf 2/2 nerve. Scale bar ¼5mm. Human Molecular Genetics, 2010, Vol. 19, No. 6 979 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 Figure 5. Lack of CNTF abolishes the sprouting of motor nerves in the gastrocnemic muscle of Smn þ/2 mice. (A) Counts of motor neuron cell bodies in the lumbar spinal cord (L1–L7) of 4-week-, 6-month- and 12-month-old Smn þ/þ (n¼3), Smn þ/2 Cntf þ/þ (n¼5), Cntf 2/2 (n¼3) and Smn þ/2 Cntf 2/2 double mutant mice (n¼4). (B) Grip strength in 4-week- (n¼at least 3 in each group), 6-month- (n¼at least 3 in each group) and 12-month-old Smn þ/þ (n¼5), Smn þ/2 Cntf þ/þ (n¼8), Cntf 2/2 (n¼5) and Smn þ/2 Cntf 2/2 (n¼5) mice. (C) Quantification of axonal sprouting, using the same technique as in Fig. 2C. In Smn þ/2 Cntf þ/þ muscles, a significantly increased number of synapses innervated by collateral or terminal sprouts was observed in comparison to Smn þ/þ muscle at 6 and 12 months. Increased sprouting in Smn þ/2 Cntf þ/þ mice was abolished when CNTF was depleted in Smn þ/2 Cntf 2/2 double deficient mice. At least 150 neuromuscular endplates were investigated per animal (n¼at least 3 in each group). (D) Overview of the innervation pattern in the gastrocnemic muscle of a 12-month-old wild-type mouse stained with a -bungarotoxin and antibodies against neurofilament in order to trace all motor axons. Seven areas with neuromuscular synapses were magnified to demonstrate terminal sprouting. The upper panel shows examples of neuromuscular endplates with axon terminals innervating only one endplate (no sprouting, indicated by arrows). The right column shows rare examples of axon terminals innervating more than one endplate (sprouting, indicated by arrows). Scale bar overview ¼100 mm, scale bar magnified area ¼30 mm. In total, more than 150 endplates were investigated for each genotype to quantify the frequency of terminal sprouting. (E) Examples of neuromuscular endplates in gastrocnemic muscle from mice with the four different genotypes. Arrows point to the position of terminal sprouts. Scale bar ¼10 mm. P,0.05, P,0.01, P,0.001. 980 Human Molecular Genetics, 2010, Vol. 19, No. 6 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023 neurons of the spinal cord (L4–L5) of 4-week and 12-month-old mice (Supplementary Material, Fig. S5). We found a slightly increased number of motor neurons with a size over 1000 mm 2 in 12-month-old Smn þ/2 Cntf þ/þ mice. We also performed electrophysiological analyses to determine the size of individual motor units. Using a ring electrode placed at mid-thigh, the mean single motor unit action potential (SMUAPs) sizes in Smn þ/þ ,Smn þ/2 Cntf þ/þ ,Cntf 2/2 and Smn þ/2 Cntf 2/2 were estimated by applying successive incremental stimuli starting from a subthreshold level until recruiting 10 individual responses (Fig. 6A). As shown before by the results obtained with the needle electrode placed into the gastrocnemic muscle (Fig. 1C-E), Smn þ/2 Cntf þ/þ mice show about two times larger SMUAPs than Smn þ/þ mice. Cntf 2/2 mice exhibit lower SMUAPs than Smn þ/2 mice. Smn þ/2 Cntf 2/2 exhibit almost 4-fold lower SMUAPs than Smn þ/2 mice (Fig. 6B). This result indicates that motor units are smaller in the double mutant mouse. These data confirm the functional relevance of the morphological alterations shown in Figure 5. In Figure 6C, a summary of the progressive loss of motor neurons in Smn þ/2 Cntf þ/þ mice from 4 weeks to 12 months is shown. In 6-month-old mice, the loss is 20%, and reaches over 40% in 12-month-old-mice. To maintain the same grip strength, axonal sprouting and size of motor units increase correspondingly. Thus, CNTF appears necessary for the compensatory sprouting response that maintains muscle strength in Smn þ/2 Cntf þ/þ mice. DISCUSSION Smn þ/2 mice, a mouse model of mild forms of SMA, can maintain muscle strength by compensatory increase of the amplitude of single motor action potentials in skeletal muscle. The loss of motor neurons is thus compensated by sprouting from remaining motor axon terminals so that neuromuscular endplates remain innervated. Lack of CNTF that is highly expressed in Schwann cells surrounding axon terminals reduces this sprouting response, both on a morphological and functional level as determined by electrophysiological analyses and by measurement of muscle strength. The most severe form of spinal muscular atrophy, SMA type I or Werdnig–Hoffmann disease, differs from the milder forms SMA type II, type III and type IV by its fast progression and the high mortality in early childhood (3,14). Similarly, Smn 2/2 SMN2 tg mice that express two copies of the human SMN2 gene on a mouse Smn null background, die within few days after birth, and these mice do not survive beyond postnatal day 6 (29). However, this mouse model of type I SMA shows only relatively moderate motor neuron loss in the spinal cord (17%) (29) at a final stage of the disease when the mice are completely paralytic at 3–5 days after birth. The fast progression of the disease points to an absence of compensatory mechanisms in comparison to mouse models with a more moderate phenotype and a slower disease progression, as it has been found in Smn þ/2 mice, a model for SMA type III (15). This corresponds to observations that survival of isolated Smn 2/2 SMN2 tg motor neurons in cell culture is not impaired, but that axon growth (9) and presynaptic differentiation are disturbed. Defective axon growth and dysfunction of neuromuscular transmission have also been observed in Smn-deficient zebrafish (30), fly larvae (31) and mice (11,32,33). In contrast, Smn þ/2 mice that lack only 50% of Smn protein levels survive and do not show any overt sign of motor neuron disease despite a progressive and significant loss of motor neuron cell bodies. These motor neurons are apparently able to sprout and to re-innervate motor endplates that would become denervated when motor neurons are lost. This correlates with an increase in the amplitude of motor units. Similar electrophysiological findings, the lack of abnormal spontaneous activity and large motor units have also been made in patients with milder forms of SMA (3). Our data indicate that CNTF is responsible for the sprouting response that leads to this enlargement of motor units and thus compensates for loss of motor neurons in the milder forms of the disease. Why does CNTF not compensate in the severe forms of SMA in Smn 2/2 SMN2 tg mice? CNTF expression is low during embryonic development (34,35), and in the peripheral nervous system CNTF expression only becomes upregulated in Schwann cells starting at the end of the first postnatal week (34,36). In rodents, CNTF expression reaches the high levels found in the adult nervous system in the third postnatal week. Therefore, CNTF is not present during the first days after birth when these mice become severely paralyzed (29) and show pathology such as depletion of synaptic vesicles at active zones and reduced synaptic transmission (11). Addition of CNTF by application via CNTF-secreting stem cells (21) or by local injection into skeletal muscle (23) leads to improved maintenance of motor axons in peripheral nerves of pmn mutant mice or neuromuscular endplates in SOD G93A mice (19). Interestingly, this effect appears relatively specific for CNTF, as other neurotrophic factors for motor neurons such as the glial-derived neurotrophic factor (GDNF) were without any effect in the same disease models (19,23). Schwann cells close to neuromuscular endplates play a major role in triggering terminal sprouting (37,38). As shown here, these cells express CNTF, and lack of CNTF expression strongly reduces terminal sprouting and augmentation of motor unit size. Terminal Schwann cells have been found to express Semaphorin-3, and it has been suggested that the upregulation of Semaphorin-3 in terminal Schwann cells could suppress sprouting and contribute to loss of neuromuscular synapses in motor neuron disease (39). Other studies have shown that the depletion of synaptic vesicles precedes the loss of synapses in a mouse model of ALS, and that CNTF could prevent the depletion of synaptic vesicles and thus maintain function of these synapses. This effect of CNTF correlates with reduced accumulation of neurofilaments (19), a characteristic hallmark of motor neuron disease in Smn deficient mice (11–13) and in SOD G93A mice, and reduced expression of stress-related genes such as Bcl-2a1-a that are normally upregulated in SOD G93A mice at a stage when synapse loss occurs. How could CNTF act on maintenance of motor endplates and induction of sprouting in Smn þ/2 mice? After binding to the CNTF receptor complex, classical signalling pathways are activated, including B-Raf (40) and the phosphatidylinositol 3 kinase (PI-3K) pathways, that do not only maintain survival but also mediate effects on the cytoskeleton (41). In addition, activation of the CNTF receptor Human Molecular Genetics, 2010, Vol. 19, No. 6 981 Downloaded from https://academic.oup.com/hmg/article/19/6/973/2527012 by Universidad de Sevilla user on 11 May 2023