Spinal muscular atrophy (SMA) is the leading genetic cause of infant mortality and is caused by the loss of a functional SMN1 gene. In humans, there exists a nearly-identical copy gene known as SMN2 that encodes an identical protein as SMN1, but differs by a silent C to T transition within exon 7. This single nucleotide difference produces an alternatively spliced isoform, SMNΔ7, which encodes a rapidly degraded protein. The absence of the short peptide encoded by SMN exon 7 is critical in the disease development process; however, heterologous sequences can partially compensate for the SMN exon 7 peptide in several cellular assays. Consistent with this, aminoglycosides, compounds that can suppress efficient recognition of stop codons, resulted in significantly increased levels of SMN protein in SMA patient fibroblasts. We now examine the potential therapeutic capabilities of a novel aminoglycoside, TC007. In an intermediate SMA model (Smn−/−; SMN2+/+; SMNΔ7), when delivered directly to the central nervous system (CNS), TC007 induces SMN in both the brain and spinal cord, significantly increases lifespan (∼30%) and increases ventral horn cell number, consistent with its ability to increase SMN levels in induced pluripotent stem cell-derived human SMA motor neuron cultures. Collectively, these experiments are the first in vivo examination of therapeutics for SMA designed to induce read-through of the SMNΔ7 stop codon to show increased benefit by direct administration to the CNS.
Spinal muscular atrophy (SMA) is the leading genetic cause of infantile death, affecting every one in 6000 live births. There are four types of SMA based upon disease severity and age of onset (1,2). Even though there is a broad clinical spectrum of SMA, almost all cases of SMA are caused by a homozygous loss of a functional copy of Survival Motor Neuron 1 (SMN1) (3). A nearly-identical copy gene, called SMN2, exists only in humans (4). These two genes are 99% identical at a nucleotide level and encode identical proteins. However, due to a C to T conversion within the 5′ end of exon 7, SMN2-derived transcripts are alternatively spliced such that the primary product is lacking exon 7 (SMNΔ7) (5). The SMNΔ7 protein has been shown to have severely reduced functionality when compared with full-length SMN. SMNΔ7 protein is rapidly degraded (6), has a reduced ability to self-associate (7), is unable to promote UsnRNP assembly (8,9), is unable to promote neurite outgrowth in SMN-deficient neurons (9,10) and is unable to rescue the neuronal defects of a zebrafish model of SMA (11).
SMN is a ubiquitously expressed protein with functions in every cell type. The most well-defined SMN activity is in UsnRNP assembly (12,13). SMN, while in its complex known as the SMN/Gemin complex, loads the Smith Core proteins onto nascent UsnRNAs before importing the assembled UsnRNP complexes into the nucleus (14). Although SMN has been shown to have many roles in all cell types, it is still unknown why motor neurons are left particularly susceptible to the loss of SMN. It is hypothesized that motor neurons could be more sensitive to changes in UsnRNP concentration, especially introns recognized by the minor spliceosome (15). It has also been argued that there may be issues with axonal transport (16–18), axonal pathfinding (19–21), neuromuscular junction (NMJ) maturation (22,23) or even SMN-related perturbations within the muscle (24). For all essential activities, the absence of SMN exon 7 significantly reduces SMN ability to function in these activities.
A cytoplasmic localization signal has been identified within SMN exon 7 and can readily transport SMN and heterologous proteins to the cytoplasm (25). While this peptide is important for SMN protein function, heterologous sequences can seemingly compensate for the SMN exon 7 peptide in vitro in regards to SMN protein localization, stability, UsnRNP assembly and neurite outgrowth (9,26). Consequently, a class of antibiotics known as aminoglycosides have been examined in SMA contexts based upon their ability to suppress translational stop codon recognition. If the first stop codon of SMNΔ7 exon 8 was not recognized, this would allow the translational machinery to elongate the truncated protein by an additional five amino acids. This length has been shown in vitro to confer more functionality to the protein (9,26). As a result, aminoglycosides that suppress efficient recognition of stop codons have been shown to result in significantly increased levels of SMN protein in SMA patient fibroblasts (26–28). Additionally, it has recently been shown that interperitoneal injection of the FDA-approved aminoglycoside genetecin (G418) can increase SMN protein and gross motor function as measured by a tube-test in intermediate SMA mice (26). However, a positive effect on lifespan was not seen, most likely due to toxicity of the compound since even unaffected heterozygous littermates died prematurely when treated with G418 (26). Here we examine the efficacy of a novel aminoglycoside-derivative, TC007, and demonstrate not only improved phenotypic measures, but also prolonged lifespan in SMA mice.
TC007 treatment can elevate SMN ‘gems’ in neuronal tissues
TC007 can elevate SMN in type 1 patient fibroblasts (3813 cells), as shown by western blot and gem counts (27). Gems are punctate nuclear structures and a part of SMN's normal cellular distribution and are an excellent biomarker for SMN levels and are therefore used as a tool to measure the efficacy of potential therapeutics. Generally, gem numbers have been examined in primary patient fibroblasts and lymphoblasts. Since SMA is a neuronal disease, analysis of gems in a more disease-specific context is advantageous. To this end, a mixed population of SMA neurons (∼25–35% of total population) and astrocytes differentiated from induced pluripotent stem (iPS) cells were treated with 150 µg/ml TC007 for 48 h (29). The SMA iPS cells were created from a type I primary patient fibroblasts (3813 cells) and show striking similarities to embryonic stem cells (29). When compared with iPS cells created from phenotypically wild-type carrier fibroblasts (iPS-WT), they exhibit specific defects after differentiation into motor neurons (29). Compounds known to elevate SMN gems in patient fibroblasts such as valproic acid and tobramycin elevate gem numbers in the differentiated SMA iPS cells (29). As expected, gem numbers in unaffected carrier cells (iPS-WT) were relatively high, ∼70 per 100 nuclei, whereas untreated SMA cells contained only 15 gems per 100 nuclei (Fig. 1). When iPS-SMA cells were treated with TC007, gem numbers were significantly elevated to ∼44 gems per 100 nuclei (Fig. 1), demonstrating that TC007 can elevate SMN levels in a disease-specific cellular context.
Direct TC007 administration to the central nervous system of SMA mice significantly elevates SMN protein, increases VHCs and promotes survival
Since it has not been determined whether TC007 can cross the blood-brain-barrier, an administration procedure was examined that allowed TC007 to be delivered directly to the central nervous system (CNS). As a proof of principle that TC007 can induce SMN protein in vivo, a single 50 mg/kg dose was administered directly to the CNS using intra-cerebral ventricle (ICV) injection (30–34) into a p5 Δ7 SMA mouse. Injecting into the ventricles should allow the compound to quickly circulate throughout the CNS. Following a single injection of TC007, the brain and spinal cord were harvested 24 h later. In extracts derived from treated brain tissue, SMN levels were significantly elevated in the brain of the TC007-treated animals (Fig. 2A and B). Analysis of treated spinal cord samples demonstrated that ICV delivery also resulted in significant increases in sites distal to the initial site of injection (Fig. 2C). Real-time qRT-PCR was performed on the same brain tissues to determine whether this elevation could be due to either an increase in SMN promoter activity or an alteration of the splicing. As expected, and in agreement with results obtained from tissue culture models in which SMN read-through was analyzed (27), neither the exon 7 splicing ratio (Fig. 3A and B) nor the amount of total SMN2 RNA (Fig. 3C) was altered following TC007 administration. These results suggest that the increase in SMN protein was likely through a transcriptional mechanism, consistent with the purposed read-through activity for aminoglycosides compounds.
Since TC007 was capable of elevating SMN levels in the CNS, the following experiments were performed to determine whether delivery of TC007 would lessen the severity of the SMA phenotype. TC007 was administered at a modestly lower concentration than in the previous experiments to avoid toxicity issues from repeated administrations: 30 mg/kg via ICV injection on post-natal days 3, 5 and 7. To examine the effect of this compound directly on the CNS (and specifically the motor neurons), the lumbar section of the spinal cord was examined to determine whether there was an increased survival of ventral horn cells (VHCs) at end-stage of disease (p14). TC007-treated mice had a significant increase in p14 VHC numbers (Fig. 4). This increase in VHCs correlates directly with a trend towards improved gross motor function of TC007-treated animals at mid- to end-stage mice, as measured by time to right (Fig. 5A and B). It has previously been demonstrated that SMA mice have defects at the NMJ, particularly within the intercostal muscles (23,35,36). Although VHC numbers were increased, there was no significant difference between treated versus untreated in regards to the NMJ at the same p14 time point, as determined by endplate-diameter, endplate-maturation [as determined by the increased number of ‘holes’ or perforations within the endplate, previously used as a measure of endplate-maturity (23,35,36)] or percent occupancy of the intercostal muscles (Supplementary Material, Fig. S1). It is unknown why the increased VHC numbers did not correlate with a significant improvement within the NMJ, although there may exist an NMJ enhancement that is too slight to measure but is just enough to have an impact on the gross motor function.
The most profound effect seen of the TC007-treated mice (n = 12) is that they lived significantly (P = 0.0386) longer (from an average of 12.64 ± 1.274 to 16.00 ± 1.008 days, or 27%) than the vehicle-treated (n = 11) counterparts (Fig. 6). Additionally, since other SMN-inducing aminoglycosides have high toxicity even in unaffected littermates (26), we administered 30 mg/kg TC007 or vehicle on post-natal days 3, 5 and 7 to heterozygous littermates. Neither lifespan (Fig. 6) nor weight (Supplementary Material, Fig. S2) was adversely affected by TC007 when compared with vehicle or non-injected pups. This indicates that multiple ICV injections of TC007 are non-toxic at the dose examined and in this animal model.
A variety of therapeutic interventions and therapeutic targets are currently being examined for SMA. The obvious and primary goal for treating SMA is to increase SMN, and therefore SMN2 has been shown to be an outstanding target for therapeutic intervention. This has been proposed to be accomplished through gene therapy, which includes gene replacement, changing the alternative splicing pattern by recruiting either splice enhancers/silencers, blocking splice sites, blocking endogenous splice enhancer/silencer sequences, or splicing in an exogenous exon 7 to the native SMN sequence [for review see (37,38)]. Additionally, drug therapy targeting SMN has been proposed for SMA. Several compounds propose to affect the promoter of SMN2 to increase transcription such as phenylbutyrate (39). Others, like HDAC inhibitors, both change the splicing of SMN to promote exon 7 inclusion as well as increase total SMN transcripts (40–42). Some drugs, such as Indoprofen (43), aminoglycosides (27,28) and proteasome inhibitors (44), propose to stabilize SMN through post-transcriptional modifications.
Aminoglycosides were first proposed to be used in a capacity aside from their antibiotic properties when it was determined that they bound not only to the small subunit of the prokaryotic ribosome (45) but also to the small subunit of the eukaryotic ribosome. Importantly, this interaction does not efficiently inhibit eukaryotic protein synthesis since binding is less efficient. However, aminoglycosides can inhibit the stop codon recognition, thereby inducing a translational read-through event at relatively low frequency [(46–48); for review see (49)]. The first disease context for which read-through was a proposed therapy was cystic fibrosis (CF) (50,51). CF is the most common autosomal recessive disorder and is often caused by a nonsense mutation to cause a premature termination codon (PTC) in the gene CF-transmembrane-conductance regulator (CFTR). Following treatment with the aminoglycoside G418, a small, but increased amount of full-length CFTR protein could be produced from a mutant CFTR gene (50,51). Since then, aminoglycosides have been proposed for other diseases caused primarily by PTCs, such as Duchene's muscular dystrophy (DMD), Becker muscular dystrophy (BMD) and Hurler syndrome. Recently, PTC therapeutics has developed a novel non-antibiotic compound, called PTC124, that can induce a read-through event without the aminoglycoside-associated toxicities (52–54). In a promising clinical trial of PTC124 on CF patients, an in vivo increase of full-length CF protein, lung function and body weight was seen (55). This new class of compounds specific for PTCs is an exciting new tool for PTC-mediated diseases that may also have application in SMA.
Using constructs of SMN with varying C-terminal lengths, it was shown that any C-terminal tail of SMN of a certain length was able to regain the proper cellular localization (in both nuclear gems and diffusely cytoplasmic) lost by the primarily nuclear SMN exons 1–6 (28,56). Additionally, in vitro evidence points to increased functionality of the read-through SMNΔ7 (9,26). If additional five amino acids were added to the SMNΔ7 protein (in the context of read-through), it is more stable, can promote neurite outgrowths and can increase UsnRNP assembly (9,26). This points to the probability that the C-terminus of SMN is quite flexible in terms of amino acid sequence, so long as it is of a certain length to denote stability and thereby functionality to the protein.
When Type I SMA primary patient fibroblasts were treated with FDA-approved aminoglycosides, with the goal of inducing an SMNΔ7 translational read-through to stabilize the protein, an increase SMN level was seen (28,57). This was determined both by an increase in gem numbers and by western blotting (28,57). They do this without changing the splicing of SMN2 (27), and are able to induce a read-through event in the context of SMNΔ7's C-terminus (26,27). Additionally, several aminoglycoside-derivatives are able to elevate gem numbers above those of the first generation of compounds (27). Therefore, we wanted to do an in vivo examination of a read-through-inducing compound for an SMA therapy. Specifically, we chose one aminoglycoside-derivative, TC007, previously shown to be a potent SMN inducer in SMA fibroblasts. Here we show that TC007 is not toxic under these experimental conditions, but the animals benefit from a direct administration to the CNS, as indicated by a significant elongation of lifespan and increase in VHC numbers in a model of SMA. It is possible that the ICV delivery adversely impacts the animals as the mean life span was slightly lower than previously reported (12.6 versus 13.4); however, this difference was not statistically significant. Additionally, these experiments are designed to demonstrate that a read-through-inducing compound can—when administered directly to the CNS—elevate SMN levels in the CNS and that the SMN increase correlates with a lessening of the severity of the SMA phenotype.
MATERIALS AND METHODS
Animals and drug treatment
All animal experiments were carried out in accordance with protocols approved by the Animal Care and Use Committee of the University of Missouri. Mice were genotyped and litters were excluded as described previously. Additionally, the mice receiving ICV were excluded if found dead the day after injection or determined to be outliers by Q-test. TC007 was initially resuspended in dH20, further diluted in PBS and administered to the ICV (1 µl/g of body weight) on post-natal days 3, 5 and 7. PBS (vehicle) was injected as a negative control. ICV injections were performed as described previously (30–33). Briefly described, neonates on p3, 5 and 7 were immobilized via cryo-anesthesia and injected using microliter-calibrated sterilized glass micropipette 0.25 mm lateral to the sagittal suture and 0.50–0.75 mm rostral to the neonatal coronary suture. The needles were inserted perpendicular to the skull and were removed after 15 s of discontinuation of plunger movement to prevent backflow. Mice recovered in 5–10 min in a warmed container until movement and general response was restored. To assess gross motor function, righting reflex [as described in (58)] was measured starting at post-natal day 5.
Western blot analysis
Tissues were harvested at indicated times, and analysis was performed as described previously (59). Mouse anti-SMN (BD) was used for western at 1:1000 and rabbit anti-actin (Sigma) was used at 1:250.
Histology and morphometry
Spinal cords were harvested at indicated times, and analysis of VHCs was performed as described previously (59). Intercostal muscle was dissected in Tyrodes solution and post-fixed for 30 min in 3.7% formaldehyde. The muscles were then frozen in TBS tissue freezing compound and cryosectioned with 30 µm thickness. Each section was then permeabilized with methanol before blocking (5% BSA in TBST) for 1 h, incubated with anti-NF 160 (1:500) and SV2 (1:50) at 4°C overnight, washed and then incubated with FITC-conjugated anti-mouse (1:200) and Alexafluor 594 conjugated α-bungarotoxin (1:100) for 1 h. Neuromuscular junctions were visualized using a Leica scope. A total of 200 endplates per mouse, marked with α-bungarotoxin, were measured and ‘holes’ were counted. The same endplates were evaluated for axonal input.
Analysis was performed as described previously (59). Briefly, error bars on graphs represent standard error of the mean (SEM). Significance of lifespan between TC007- and vehicle-treated mice was determined by Mantel–Cox test (P = 0.0386). All other significance indicated was calculated by Student's t-test, P < 0.05 or greater.
RNA was isolated from ICV-injected brain tissue (50 mg/kg) after 24 h (on p6) using TriReagent protocol and treated with RQ1 DNase to eliminate genomic and transgenic DNA. Equal amounts of total RNA were used to make the cDNA. Omitting reverse transcriptase was used as a negative control. Quantitative real-time PCR was performed using TaqMan and primers described previously (40).
Immunofluorescence of SMA iPS cells
iPS cells were generated and differentiated to a neuronal lineage as described previously (60). Described, iPS neurospheres were plated on laminin/poly-ornithine slides in media [70:30 DMEM:F12 plus B27 supplement (Sigma)] not containing epidermal growth factor or fibroblast growth factor to differentiate for 1 week. They were then treated with 150 µg/ml of TC007 for 48 h before being fixed with 1:1 acetone:methanol. They were then prepared for immunofluorescence as described previously (29) using mouse anti-SMN 4B7 (Lorson lab) 1:10 and anti-mouse FITC (Jackson Research) 1:200, with DAPI to mark the nuclei.
This work was supported by grants from FightSMA [C.L.L.]; and the National Institutes of Health [C.L.L, R01 NS41584; R01 HD054413] and [C.W.C., AI053138].
We would like to thank Dr. Polo-Parada for his expertise and guidance on NMJ staining and John Marston for assistance with the mouse colony.
Conflict of Interest statement. None declared.