- Top of page
- MATERIALS AND METHODS
- Author contributions
- For more information
- Supporting Information
SMA is characterized by skeletal-muscle weakness and wasting, due to α-motor neuron degeneration, and no effective therapy is currently available. Loss-of-function mutations or deletion of SMN1 and the resulting deficiency in the encoded SMN protein, which mediates snRNP assembly, cause SMA, although how this specifically affects α-motor neurons remains unclear (Burghes & Beattie, 2009). A closely related gene, SMN2, is present only in humans. Although SMN2 exon 7 is predominantly skipped by alternative splicing, which results in a truncated defective protein, called SMNΔ7, SMN2 acts as a disease modifier and reduces SMA severity as its copy number increases (McAndrew et al, 1997). Based on the age of onset and clinical severity, SMA is subdivided into types I, II, III and IV, with type I being the most severe form. Types I–III affect infants and children usually under the age of 3, whereas type IV shows adult onset (Lunn & Wang, 2008).
Several SMA models have been generated to reproduce SMA with various severities. Knockout of the murine Smn gene results in embryonic lethality (Schrank et al, 1997). Introduction of a human SMN2 transgene rescues this phenotype, such that Smn−/− SMN2 mice have SMA-like phenotypes whose severity inversely correlates with the SMN2 copy number (Hsieh-Li et al, 2000; Monani et al, 2000). Severe-SMA mice harbouring two SMN2 copies, or with an extra SMNΔ7 cDNA transgene (SMA Δ7 mouse model), develop early and rapidly progressive pathology, dying within 1–2 weeks postnatally (Hsieh-Li et al, 2000; Le et al, 2005; Monani et al, 2000; Riessland et al, 2010). In contrast, SMA mice harbouring four SMN2 copies survive normally and do not develop paralysis, but have an abnormal, short and thick tail and develop tail and ear necrosis, beginning around 3 weeks and 3 months postnatally, respectively (Hsieh-Li et al, 2000). These models provide distinct advantages, including the testing of therapeutic strategies based on targeting the human SMN2 transgene by means of splicing correction or upregulation (Park et al, 2010a).
RNA splicing requires pre-mRNA cis-acting elements recognized by trans-acting factors, such as spliceosome components and auxiliary RNA-binding proteins (Cartegni et al, 2002). Antisense oligonucleotides (ASOs) can be designed to target cis-element(s) on a given pre-mRNA, so as to preclude binding of trans-acting factors and thereby modulate splicing patterns. These properties enable the development of RNA-targeted therapeutics to correct disease-associated splicing defects or restore the translational reading frame (Bennett & Swayze, 2010). We previously reported that a 2′-O-(2-methoxyethyl) (MOE) therapeutic ASO (ASO-10-27 or ISIS-SMNRx) that promotes exon 7 inclusion in SMN2, rescues the phenotypes of several SMA mouse strains (Hua et al, 2010, 2011; Passini et al, 2011; Sahashi et al, 2012). Recently, a phase Ib/IIa clinical study of ASO-10-27 has been initiated in children with intermediate SMA.
We have also employed ASO technology to phenocopy SMA in transgenic mice, utilizing ASOs that exacerbate the SMN2 splicing defect and persistently promote pathogenesis. Intracerebroventricular (ICV) administration of an exon-7-complementary MOE ASO (ASO-20-37) that promotes SMN2 exon 7 skipping in neonatal four-copy SMN2-transgenic mice successfully phenocopies intermediate SMA, including ∼1-month lifespan and progressive motor dysfunction, with α-motor neuron loss and abnormal neuromuscular junctions (NMJs) (Sahashi et al, 2012). These phenotypes are shared by other intermediate SMA models with point mutations (Smn2B/− mice) or exon deletions (SmnF7/Δ7, NSE-Cre mice) in murine Smn, although these strains lack an SMN2 transgene, which is being actively pursued as a therapeutic target in human SMA (Park et al, 2010a).
Available SMA mouse strains, including those with inducible expression of SMN, are extremely useful for studying the temporal and spatial requirements for SMN (Gavrilina et al, 2008; Le et al, 2011; Lutz et al, 2011; Park et al, 2010b), although the physiological roles of SMN and pathological roles of SMN deficiency after the developmental stages, remain unclear. A recent report showed that removal of ectopic SMN induction after postnatal Day 28 in an SMA Δ7 mouse background resulted in some of the mice surviving for >8 months (Le et al, 2011). However, the tissue-specific effects of adult-onset SMN deficiency have not been addressed. Many SMA patients reach adulthood, and there is an adult-onset form of the disease, namely type IV SMA, characterized by progressive paralysis and decline in daily-living activities. Therefore, addressing the effect of SMN levels and the phenotypic effects of SMN deficiency/restoration in adult mice should contribute to the understanding of SMA pathogenesis and to the development of targeted therapies. Animal models of adult-onset SMA would be extremely valuable for such studies.
Here we extended our antisense exon-skipping approach to adult mice with four copies of an SMN2 transgene. We found that ICV-administered ASO phenocopies adult-onset SMA. The extent of SMN2 mis-splicing in the central nervous system (CNS) determined the severity of the SMA-like motor symptoms. SMN2 mis-splicing was exacerbated during late-stage disease, which should accelerate the decline. In addition, systemically administered exon-skipping ASO also affected survival, resulting in striking liver and heart lesions, and the combination of central and peripheral administration exacerbated the pathology. We demonstrated effective rescue with therapeutic ASO-10-27, suggesting that there is a broad temporal therapeutic window for treatment of adult-onset SMA. The ability to persistently modulate splicing of a target gene using ASOs provides a powerful method to model and characterize diseases in animals.
- Top of page
- MATERIALS AND METHODS
- Author contributions
- For more information
- Supporting Information
Severe type I-II SMA patients show symptoms during the newborn and early-infant periods, suggesting a potential developmental component of SMA pathogenesis. On the other hand, many type II–III SMA patients reach adulthood, and there are also adult-onset type IV SMA patients. The pathological impact of post-developmental SMN deficiency remains unclear, which prompted us to address the spatiotemporal requirements for SMN in adult mice.
We have shown that exon-skipping ASOs can be used to phenocopy certain diseases by gene-targeted modulation of alternative splicing. By inhibiting SMN2 exon 7 inclusion with an appropriate ASO in neonatal (Sahashi et al, 2012) or adult (this study) transgenic mice, we elicited phenotypes resembling canonical SMA. We first used ICV delivery of exon-skipping ASO-20-37 in adult SMA mice with four copies of an SMN2 transgene, a method whose effectiveness we previously showed in neonates of the same strain (Sahashi et al, 2012). A single injection had dose-dependent effects on SMN2 splicing in the CNS, correlating with lifespan, growth and motor-function phenotypes (Figs 1 and 3). Notably, compared with our previous finding that neonatal ICV injection causes progressive paralysis, beginning after 2 weeks (Sahashi et al, 2012), in the present study, abnormal gait or rearing became apparent only 1 month after ASO injection. One hundred micrograms of ASO-20-37 robustly inhibited SMN2 exon 7 splicing as early as PS7 (Fig 1), revealing a time lag between SMN depletion and the motor-phenotype onset. Importantly, our results suggest that the late-onset paralysis reflects a tolerance of mature motor units against SMN deficiency (discussed below), although it is accompanied by motor-unit pathologies. IP injection failed to elicit overt paralysis (Supporting Information Fig S3), even during end-stage disease, revealing that motor dysfunction is not merely secondary to end-stage disease conditions. However, adult-onset paralysis in type IV SMA also reflects compensatory sprouting of nerve terminals, which is not seen in our mice, likely because of the rapid disease course (Dubowitz & Sewry, 2007) as well as the small reduction in muscle SMN.
ICV injection of 10 µg ASO, which reduced the proportion of exon 7 inclusion in the spinal cord of mice with four SMN2 copies by at most ∼60%, failed to elicit overt phenotypes (Figs 2 and 3). This implies that a threshold for the adult-onset phenotypes would be less than two transgene copies, in this strain background. Another explanation is that ICV injection of 10 µg ASO might elicit only minimal phenotypes that are hardly detectable, or might not sustain SMN2 mis-splicing to result in significant phenotypic defects, although mis-splicing was evident until at least PS120 (Fig 2). Considering the severity of SMA in neonatal mice with two SMN2 copies, our results indicate that the SMN levels required for healthy adult mice, especially in the CNS, are less than for neonatal mice. This notion is partly supported by our previous study in severe-SMA mice rescued by therapeutic ASO-10-27: high levels of correction of SMN2 splicing are not persistently required, at least in the liver, for long-term survival (Hua et al, 2011). Also, for SMN2, a minimum of two copies are thought to be necessary after 1 month postnatally in SMA Δ7 mice (Le et al, 2011), which in turn indicates a lower post-developmental demand for SMN.
ICV injection of ASO-20-37 resulted in a marked decrease in the number of SMN-containing nuclear gems—an indicator of SMN protein abundance (Coovert et al, 1997) in α-motor neurons (Fig 4), which is expected to cause neuronal dysfunction. A dose of 25 µg ASO caused α-motor neuron loss and shrinkage, by 2 months post-injection (Fig 4). A dose of 100 µg ASO did not cause such changes by 1 month post-injection, although we observed a loss of glutamatergic or cholinergic excitatory synapses onto motor neurons, which should affect neuronal circuits and in turn motor function (Fig 4). Glutamatergic synaptic pathology occurs early in the course of disease (Ling et al, 2010; Mentis et al, 2011), and cholinergic synaptic pathology may also play a role in the motor deficits (Zagoraiou et al, 2009). Considering that neonatal ICV injection of the ASO leads to significant motor-neuron loss at P30 (Sahashi et al, 2012), morphological defects in α-motor neurons and also NMJs (Fig 4, see below), once their maturation is complete, might only occur to a limited extent and in a delayed manner in the context of SMN deficiency, which may account for the lack of muscle denervation (Fig 4; Supporting Information Fig S4).
ICV injection of ASO-20-37 elicited motor-neuron pathology, and especially when combined with IP injection, it led to NMJ pathology, as well as overt motor-function deficits. These findings reveal that SMN function is required for α-motor neuron and NMJ maintenance even after embryonic development, although the threshold levels of SMN appear to be lower than for neonate mice, meaning that mature motor units can tolerate relatively low SMN levels.
Whether SMN deficiency in adulthood affects NMJ function has not been determined. Removal of SMN at 1 month postnatally in the SMA Δ7 mouse results in no impairment in synaptic transmission at NMJs, 1 month later, at a time when the mice are dying (Le et al, 2011). Thus, it is unlikely that neuromuscular transmission was impaired in the mice in which we ICV-injected 100 µg of ASO-20-37, which also die in 1 month. Instead, pathology in the motor-neuron circuitry, as reported previously (Ling et al, 2010; Mentis et al, 2011; Park et al, 2010b), could explain the motor phenotypes in our model.
A higher proportion of abnormal NMJ structures were observed with a longer disease course (Fig 4), indicating a temporal requirement for the defects to arise in adults with SMN deficiency. The delayed-onset motor neuron dysfunction may also affect efferent nerve terminals and endplates (Gogliotti et al, 2012; Park et al, 2010b). In contrast, our finding that combined ICV and IP injection of ASO exacerbated NMJ defects suggests additional NMJ pathology with peripheral SMN deficiency (Fig 4). We demonstrate that relatively mild NMJ pathology could be seen in adult-onset SMA mice, in contrast to early-onset SMA mice (Kariya et al, 2008; Kong et al, 2009; Lee et al, 2011; Ling et al, 2010, 2012; Murray et al, 2008; Park et al, 2010a; Sahashi et al, 2012) whose endplate maturity or NMJ innervation is affected. These motor-unit pathologies in mouse models imply that SMN is especially important during development, which in turn suggests that the pathogenesis of SMA has important developmental components.
Because of the inability of MOE ASOs to cross the mature BBB, systemic IP injection of ASO-20-37 inhibited SMN2 splicing only in peripheral tissues, especially liver. However, we observed only limited splicing effects of ASO-20-37 in muscle and heart (Fig 2), presumably reflecting the limited ASO biodistribution in these tissues and/or different levels of trans-acting factors that regulate SMN2 splicing. As IP administration affected survival and unveiled some heart and liver pathology (Figs 3 and 5), analysing the effect of enhanced, persistent SMN2 mis-splicing in muscle or heart may further help to address post-developmental pathogenesis, especially using different administration routes and/or different ASOs.
Echocardiography before end-stage disease showed no significant defects in mice injected ICV with ASO-20-37 to substantially inhibit SMN2 splicing in the CNS (Table 1; Fig 2). However, despite the absence of a significant decline in heart rate, our finding that some mice suddenly died suggests a possible end-stage involvement of severe central autonomic dysfunction, which has been implicated in cardiac dysfunction in severe SMA (Bevan et al, 2010; Heier et al, 2010). In contrast to ICV injection of ASO-20-37, IP injection induced cardiac hypertrophy and a reduction in cardiac mass (Fig 5; Table 1; Supporting Information Table S2). Because there was no cellular pathology, impairments in cardiac growth should be considered in this case. SMN deficiency in other tissues could also contribute to cardiac complications. In particular, liver dysfunction might induce cardiomyopathy, in part by hyperdynamic circulation (Ma & Lee, 1996; Moller & Henriksen, 2010), and/or an accompanying decrease in circulating IGF1 levels (see below) might affect cardiac function (Juul et al, 2002; Vasan et al, 2003).
We identified pronounced liver pathology in mice that received IP injection of ASO-20-37, but not in mice that received ICV injection. Accordingly, in the IP-injected mice, the mRNA levels of hepatic Igf1 and Igfals decreased (Fig 5). Most of the circulating IGF1 is in the form of stable complexes with IGFALS and IGFBP3 (Baxter & Dai, 1994). Therefore, the reduction in IGF1 and IGFALS would account for the observed low levels of circulating IGF1 (Fig 5; Sjogren et al, 1999). These liver pathologies must be a consequence of hepatic SMN2 mis-splicing, because they were prevented by subsequent IP injection of splicing-correcting ASO-10-27 (Supporting Information Fig S5). They may contribute to SMA progression (Hua et al, 2011), underscoring the importance of SMA pathology in peripheral tissues. Knockout of SMN in mouse liver results in defective liver development, with lack of regeneration (Vitte et al, 2004). In contrast, our approach did not completely suppress SMN2 splicing at the adult disease stages, and resulted in distinctive liver pathology, including oval cell-mediated regeneration (Figs 2 and 5). The combination of ICV and IP injections of ASO-20-37 further shortened lifespan, compared to either injection alone, emphasizing that hepatic and cardiac complications should be also considered as features of adult SMA, especially at the advanced stages, when SMN2 mis-splicing is exacerbated (see below; Fig 2; Sahashi et al, 2012).
SMN2 was further mis-spliced in both the CNS and peripheral tissues in the late-stage mice that were ICV-injected with ASO-20-37 (Fig 2), in a manner that did not correlate with the extent of ASO uptake in various tissues. This finding is indicative of progressive SMN2 mis-splicing in the context of end-stage SMA (Sahashi et al, 2012). ASO-induced SMN deficiency may trigger a further decrease in SMN2 splicing through a feedback loop (Jodelka et al, 2010; Ruggiu et al, 2012), on top of which, end-stage-disease conditions, such as nutritional deficiency and hypoxia may cause widespread splicing alterations, including in SMN2 (Bebee et al, 2012; Sahashi et al, 2012), each of which may potentially accelerate SMA progression. Thus, controlling stress conditions caused by SMA progression and/or other complications may be very important for its prognosis.
ASO-20-37, which specifically targets the human SMN2 transgene, did not affect normal mice with an intact Smn gene. We previously showed that neonatal ICV injection of ASO-20-37 phenocopies SMA in SMN2-transgenic mice (Sahashi et al, 2012), and here, we similarly recapitulated this phenotype in adult mice of the same strain. The phenotypic amelioration of these ASO-20-37-treated mice by a therapeutic ASO that restores SMN2 splicing (Fig 6) demonstrates that the SMA-like phenotypes were elicited through SMN2 mis-splicing, excluding potential off-target effects as contributors to disease onset.
Compared with RNAi-based or antisense-knockdown approaches, our method, which we previously dubbed TSUNAMI (for targeting splicing using negative ASOs to model illness), retains the primary transcript and thus enables testing of therapeutics that correct splicing of the target pre-mRNA (Sahashi et al, 2012). By targeting the human SMN2 transgene, we were able to use therapeutic ASO-10-27 for splicing-rescue experiments in the SMA-phenocopy mouse model. Its ability to correct splicing in the CNS or peripheral tissues correlated with phenotypic and histological amelioration in ASO-20-37-treated mice (Fig 6; Supporting Information Fig S5).
We reported that neonatal systemic injection of ASO-10-27 efficiently rescues severe-SMA mice (Hua et al, 2011). However, this systemic treatment had no therapeutic benefit in adult mice that were first administered ASO-20-37 by ICV injection (Fig 6). This is similar to what we observed with neonatal mice sequentially treated with these ASOs (Sahashi et al, 2012). As we discussed in that study, the inconsistency in the effect of systemic ASO treatment among these studies may reflect spatial and temporal differences in SMN2 splicing patterns; here SMN2 splicing (from four copies of SMN2) was predominantly inhibited in the CNS at the adult stages, whereas in severe-SMA mice SMN2 mis-splicing (from two copies of SMN2) is ubiquitous and begins embryonically. Another important difference between neonatal and adult mice is the extent of BBB closure (Stewart & Hayakawa, 1987), which determines whether a systemically administered MOE ASO reaches the CNS. However, the present results do suggest that SMN levels in CNS tissues determine SMA-like motor phenotype at the adult stage, whereas those in both CNS and peripheral tissues correlate with overall prognosis. Our results also underscore a potentially wider therapeutic time window for adult SMA than previously believed based on studies in severe-SMA mice (Foust et al, 2010; Hua et al, 2011; Le et al, 2011; Lutz et al, 2011).
Except for the tail and ear necrosis, our model has an asymptomatic phase before disease onset at the adult stage, as seen in type IV SMA. Although basal SMN levels are below normal, due to the mouse genotype, SMN is further reduced in a tissue-specific manner, by ASO treatment, which is distinct from the situation in SMA patients. The disease progression after onset in our model is more acute, and without chronic compensation. However, here and also in our previous study (Sahashi et al, 2012), TSUNAMI can help to elucidate relevant phenotypes, and our spatial and temporal analyses of SMN's roles provide new insights into SMA pathogenesis and targeted-therapeutic strategies.
Here, we have shown that a single injection of an exon-skipping ASO into the cerebral ventricles phenocopies adult-onset SMA. Compared with neonatal administration, this procedure elicited late-onset motor phenotypes. In adult mice with Smn-null background, the SMN levels required for normal CNS function correspond to an SMN2 copy number less than two, suggesting that only moderate SMN levels are necessary for therapy in adult-onset SMA. The effects of peripheral SMN2 mis-splicing induced by TSUNAMI highlight the potential importance of specific pathogenesis in the liver and heart. We also showed a ubiquitous exacerbation of SMN2 mis-splicing during late-stage disease, which likely accelerates disease progression. Finally, our results suggest a broad therapeutic time window for ASO-10-27.
Compared with early-onset SMA, adult-onset SMA may have distinct pathology, as well as therapeutic responses and requirements, perhaps reflecting key differences between development and post-development.