A short antisense oligonucleotide ameliorates symptoms of severe mouse models of spinal muscular atrophy.

Recent reports underscore the unparalleled potential of antisense-oligonucleotide (ASO)-based approaches to ameliorate various pathological conditions. However, in vivo studies validating the effectiveness of a short ASO (<10-mer) in the context of a human disease have not been performed. One disease with proven amenability to ASO-based therapy is spinal muscular atrophy (SMA). SMA is a neuromuscular disease caused by loss-of-function mutations in the survival motor neuron 1 (SMN1) gene. Correction of aberrant splicing of the remaining paralog, SMN2, can rescue mouse models of SMA. Here, we report the therapeutic efficacy of an 8-mer ASO (3UP8i) in two severe models of SMA. While 3UP8i modestly improved survival and function in the more severe Taiwanese SMA model, it dramatically increased survival, improved neuromuscular junction pathology, and tempered cardiac deficits in a new, less severe model of SMA. Our results expand the repertoire of ASO-based compounds for SMA therapy, and for the first time, demonstrate the in vivo efficacy of a short ASO in the context of a human disease.

Introduction

The survival motor neuron (SMN) gene is ubiquitously expressed and produces an essential protein.[1] Low SMN levels cause spinal muscular atrophy (SMA) (OMIM 253300; 253550; 253400; 271150), an autosomal recessive neuromuscular disease that is the leading hereditary cause of infant death.[2,3] SMA is characterized by the loss of motor neurons within the spinal cord and resultant muscle weakness and atrophy.[4] Clinically, SMA is heterogeneous, and patients are stratified into clinical types based on the age at disease-onset and maximum motor function achievements.[5,6,7,8,9] Genetically, SMA is homogeneous in that all forms of the disease are caused by the loss of function mutations within the SMN1 gene, with concurrent retention of at least one copy of a highly similar gene, SMN2.[2] SMN2 differs from SMN1 functionally by a single C to T nucleotide transition at position 6 (C6U transition in transcript) in exon 7 (ref. 10). The C6U transition creates a silencer and/or strengthens an inhibitory motif, which leads to a predominant amount of the transcripts skipping SMN2 exon 7 (refs. 11,12), and reviewed in Singh et al.[13] SMNΔ7, the translated product from exon 7-skipped transcripts, is less functional and rapidly degraded.[14]

A recent update on potential therapeutic options for SMA treatment identified antisense-oligonucleotide (ASO)-based therapy as one of the most promising strategies.[15] Blocking of intronic splicing silencer N1 (ISS-N1) by an ASO fully corrects SMN2 exon 7 splicing and restores a high level of SMN from endogenous SMN2.[16] The original study performed with the ISS-N1 target employed a 20-mer ASO encompassing uniform phosphorothioate backbone and 2′-O-methyl modifications (2′OMeP).[16] Subsequent preclinical studies have revealed efficacy differences between ASO chemistries and the route of ASO administrations.[17] ASO-10–27, an 18-mer ASO encompassing phosphorothioate backbone and 2′-O-methoxyethyl modifications (abbreviated as “MOE”), has shown better efficacy when delivered peripherally but not when delivered through an intracerebroventricular (ICV) route.[18] Twenty-mer or longer phosphorodiamidate morpholino oligomers (abbreviated as “PMOs”) have shown unprecedented efficacies when delivered through the ICV route.[19,20] These findings underscore the challenge in selecting an optimal ASO with the right size and chemistry. Short ASOs offer several advantages over longer ASOs, including low to zero tolerance for mismatch base pairing, reduced cost of synthesis, and ease of modifications.[21] Despite these advantages, the lack of a validated short target coupled with the possibility of rapid clearance from the bloodstream, have limited exploring the in vivo efficacy of very short ASOs.

The lessons learnt from ISS-N1 as an antisense target underscore certain basic principles that could be exploited for the validation of additional targets for SMA therapy.[16] For instance, any strong inhibitory element within an intronic sequence, sequestration of which does not interfere with downstream processes, could serve as an ideal antisense target. Incidentally, an eight-nucleotide GC-rich sequence (GCRS) partially overlaps with ISS-N1, and simultaneously participates in the formation of an internal stem formed by long-distance interaction (LDI), termed internal stem through LDI-1 (ISTL1), a unique inhibitory RNA structure.[21,22] Sequestration of GCRS by an 8-mer ASO encompassing 2′OMeP (3UP8) fully corrects aberrant SMN2 exon 7 splicing and restores high levels of SMN in SMA patient cells.[21] Encouraged by these findings, here we examine the therapeutic efficacy of 3UP8i, an improved version of 3UP8 in two severe mouse models of SMA. Our results provide the first evidence of therapeutic efficacy of a very short ASO in the context of a preclinical study of a human disease.

Results

Optimization of an 8-mer ASO for in vivo application

We previously reported that an 8-mer 2′OMeP ASO (3UP8) that targets a GCRS that partially overlaps with ISS-N1 corrects SMN2 exon 7 splicing in SMA patient cells (, lane 3).[21] Unlike large ASOs, 3UP8 showed zero tolerance for single-nucleotide mismatch base pairing.[21] Another 8-mer ASO (F8) that targeted the first eight residues of ISS-N1 had no effect on SMN2 exon 7 splicing, suggesting the effect of 3UP8 is independent of ISS-N1 and specific to the GCRS (, lane 2).[21] To test the in vivo efficacy of 3UP8, we performed preliminary experiments in healthy, adult Smn heterozygous mice containing human SMN2.[23] Intraperitoneal (IP) administration of a moderate dose (20 μg/g body weight) of 3UP8 did not produce any noticeable effect on SMN2 exon 7 splicing in the liver. Among several factors, the lack of the stimulatory effect of 3UP8 in vivo could be attributed to the fast clearance of a short ASO from the blood. To circumvent this problem, we generated 3UP8i, an improved version of 3UP8 by incorporating PEG-282 and propyl modifications at the 5′ and 3′ ends, respectively (). These modifications confer enhanced stability and resistance to enzymatic degradation or modification.[24,25] As expected, 3UP8i retained the GCRS-specific stimulatory effect on SMN2 exon 7 splicing in SMA patient cells (, lane 5). A control 8-mer ASO with similar modifications and a single-nucleotide substitution (CON8i) produced no appreciable effect on SMN2 exon 7 splicing (, lane 4). In vivo, 3UP8i (20 μg/g body weight) showed noticeable stimulation of SMN2 exon 7 splicing in liver (, lane 4). We next examined the effect of high dosages (80 μg/g body weight) of 3UP8i and CON8i delivered through IP and subcutaneous (SC) routes. 3UP8i showed substantial stimulation of SMN2 exon 7 inclusion in the liver, whereas CON8i had no noticeable effect (, lanes 5–10). Upon ICV administration of a moderate dose of 3UP8i, we observed substantial inclusion of SMN2 exon 7 in neuronal tissues, with some concomitant peripheral increases as well (Supplementary Figure S1c). These initial experiments demonstrated that 3UP8i shows a wide range of in vivo function.

In vivo efficacy of 3UP8i in severe 5058-Hemi SMA mice

To examine the in vivo efficacy of 3UP8i, we employed the “Taiwanese” severe SMA mouse model, which we refer to as 5058-Hemi SMA mice.[26] This model harbors two copies of the SMN2 transgene and a targeted mutation of murine Smn exon 7 (ref. 27). Previous studies established that early, peripheral ASO delivery was more effective than an ICV delivery strategy for long-term rescue of the SMA phenotype in this SMA model.[18] Thus, in blinded studies, we administered 3UP8i via IP injection starting at postnatal day 0 (P0), with two subsequent doses at P5 and P10. Both low (20 µg/g body weight) and high (80 µg/g body weight) doses were tested to assess the dose-dependent effect (). The low dose of 3UP8i produced no apparent benefit in terms of survival or weight gain compared to CON8i, although a significant improvement in righting reflex was seen by P9 (–). At the higher dose, 3UP8i resulted in a modest, yet statistically significant increase in weight gain starting at P8 compared to the CON8i-treated 5058-Hemi SMA mice (). Compared to heterozygous littermates, the 3UP8i-treated SMA mice were still noticeably underdeveloped (). The higher 3UP8i dose also resulted in a significant increase in median survival compared to CON8i, from 10 to 13 days (). Although general mobility was diminished, motor function of 3UP8i-treated SMA mice, as assayed by the time-to-right, was indistinguishable from heterozygous controls (). Overall, although 3UP8i showed modest efficacy, it did not produce the level of therapeutic efficacy seen with the longer ASOs in the same model.[18]

Effect of 3UP8i on SMN2 exon 7 splicing is age-dependent

Based on the modest improvements in survival and function of 5058-Hemi SMA mice treated with 3UP8i, we postulated that 3UP8i might possess developmentally variable efficacy. To test this, we examined the ability of 3UP8i to increase full-length SMN2 (FL-SMN2) transcripts over time. We used homozygous 5058-SMA mice, which harbor four copies of SMN2 on a Smn mutant background; other than mild peripheral necrosis, they are phenotypically akin to unaffected wild-type mice.[27,28,29] Quantitative RT-PCR (qRT-PCR) of liver samples from 5058-homozygous mice administered 3UP8i or CON8i revealed that 3UP8i had limited SMN2 exon 7 splice-correcting activity at early postnatal time points (P2: 3UP8i, 1.55 ± 0.180; CON8i, 0.876 ± 0.126) (, left panel). At later time points, 3UP8i displayed a substantial increase in FL-SMN2 transcript relative to CON8i (P10: 3UP8i, 2.12 ± 0.153; CON8i, 0.759 ± 0.437. P15: 3UP8i, 3.88 ± 0.128; CON8i, 2.73 ± 0.115). End-point PCR of the same samples showed a similar pattern (). The qRT-PCR analysis at P20 showed only a modest increase of FL-SMN2 transcripts (P20: 3UP8i, 1.38 ± 0.124; CON8i, 0.974 ± 0.085). However, the end point PCR analysis at P20 showed the highest ratio of FL:Δ7 transcripts of any time-point (, panel 4). The Δ7 transcripts of the CON8i-treated samples were significantly higher for all time-points except P2 (, right panel). These results confirmed that SMN2 exon 7 inclusion by 3UP8i is age-dependent.

Effect of 3UP8i in a less severe mouse model of SMA

5058-Hemi SMA mice are one of the more severe models of SMA, with a limited therapeutic window.[18,26,27] Knowing that the current formulation of 3UP8i is not entirely suitable for such early interventions, we developed an SMA mouse with a slightly larger therapeutic window. This new SMA model, termed 5058-Hemi hybrid, was generated using the same breeding strategy for 5058-Hemi SMA mice, except instead of intercrossing 5058 homozygous FVB/N mice with the outbred Δ7Smn allele from this line (Smn), we used a different, but equivalent, Δ7Smn allele (Smn) in a C57/BL6 background (Smn).[27,30] By generating an F1 SMA hybrid, we more than doubled the median survival (P8 versus P22), and the age at which weight differentiated from controls (P4 versus P8) ().

Using the same dosing strategy as with 5058-Hemi SMA mice, the median survival of 3UP8i-treated Hemi-hybrid SMA mice was undefined at study end point (P50), whereas the median survival of CON8i-treated Hemi-hybrid SMA was 16 days. This represented at least a threefold increase in median survival of the 3UP8i-treated SMA mice (log rank, P < 0.0001) (). Untreated SMA mice had a median survival of 22 days, which was significantly better than CON8i-treated SMA mice (log rank, P = 0.0016), and likely due to a larger sampling size in conjunction with the stress of dosing the CON8i-treated SMA mice. Alternatively, the CON8i ASO may not have been well tolerated, as there was also a small decrease in overall weight compared to the untreated Hemi-hybrid SMA mice (). The study end point at P50 was chosen since the frequency of the 3UP8i-treated deaths had slowed substantially, and no further decline in health or motor function was evident. Moreover, we found that 3UP8i treatment resulted in a significant improvement in weight gain of 5058-Hemi hybrids, starting at P13 (3UP8i, 6.19g ± 0.457; CON8i, 5.03 ± 0.322) (). This trend continued, and the slope of weight gain was similar to control mice.

Since heart defects and bradyarrythmia had been noted in other SMA mouse models,[31,32,33,34] we studied the ECG waveforms of 5058-Hemi hybrid mice. Recording intervals were limited to preweaning ages due to post-weaning variability between the sexes.[35] By P10, heterozygote littermates diverged from all SMA mice for heart rate (HR) and PR interval (Supplementary Figure S2a and c, respectively). The 3UP8i-treated SMA mice displayed significant improvement in HR by P18 (bpm: HET, 776 ± 11.2; 3UP8i, 701 ± 31.9; CON8i, 608 ± 39.3; Supplemenatry Figure S2a). The lower HR and elongated PR interval (ms: HET, 23.31 ± 0.459; 3UP8i, 26.4 ± 0.930; CON8i, 30.43 ± 2.49) confirmed that this model presents with bradyarrythmia, although to a lesser extent than in the more severe, 5025-▵7 model and severe inducible SMA mice.[32,33] The QRS interval of the SMA pups was slightly elevated compared to control (HET, 9.69 ± 0.211; 3UP8i, 10.7 ± 0.594; CON8i, 10.75 ± 0.543) (Supplementary Figure S2d). This elevation was indicative of delayed ventricular depolarization times, and suggested a bundle branch block.[36] Treatment with 3UP8i resulted in a corrective, downward trend toward heterozygote levels, but it did not reach statistical significance compared to CON8i-treated SMA mice by P18. No significant difference was observed in heart rate variability (HRV) between SMA mice and control littermates (Supplementary Figure S2b). Together, the bradyarrythmia and heart block suggested that the 5058-Hemi hybrid SMA mice suffer from deficits in sympathetic innervation, and to some extent, 3UP8i was able to correct this.

One property of this particular model is the appearance of hind limb edema and tail necrosis, starting at ~P8 in untreated mice (, left panel). Treatment with 3UP8i prevented hind limb edema, and delayed tail necrosis (, middle panel). Although the tail eventually necrosed in treated mice, they were generally spared from any additional peripheral necrosis (, right panel). Further, 3UP8i-treated SMA mice did not show any noticeable deficits in motor function over the course of the study.

The primary mode-of-action of 3UP8i is to promote the inclusion of exon 7 in SMN2 transcripts. Thus, we expected the levels of FL-SMN2 transcripts to substantially increase in the surviving 5058-Hemi hybrid SMA population that received 3UP8i. We examined the relative SMN2 expression levels in the liver of 5058-Hemi hybrid mice by TaqMan qRT-PCR, and found that 2 days postfinal dose, 3UP8i-treated liver samples showed a fourfold increase in FL-SMN2 transcripts compared to CON8i-treatment (3UP8i, 5.95 ± 0.220; CON8i, 1.40 ± 0.113) (, left panel). Correspondingly, the exon 7-skipped isoform (Δ7-SMN) significantly decreased threefold (3UP8i, 0.591 ± 0.117; CON8i, 1.78 ± 0.190) (, right panel). By P50, there was no significant difference in amount of FL-SMN2 and Δ7-SMN2 transcripts between 3UP8i-treated SMA mice and untreated heterozygotes (). This suggested that the pharmacodynamic range of 3UP8i is less than 40 days in vivo. End-point PCR of the same samples confirmed the trend seen via qRT-PCR, with a noticeable increase in the FL-versus Δ7-SMN2 transcript ratio following 3UP8i treatment at P12 (). Previous studies have demonstrated that peripheral administration of ASOs produces a splice-correction that is generally most pronounced in the liver, with other peripheral tissues displaying a reduced response.[18] Examination of FL-SMN2 levels in the quadriceps of 3UP8i-treated mice revealed consistent findings. FL-SMN2 levels were elevated compared to CON8i-treated mice, although less than the change seen in the liver (Supplementary Figure S3a, b).

Ultimately, SMA is caused by a deficiency in SMN protein levels. Therefore, we determined whether the increases seen at the transcript level in the liver of 3UP8i-treated 5058-Hemi hybrid SMA mice translated to increased levels of SMN. Compared to CON8i-treated mice, 3UP8i treatment resulted in nearly a threefold increase in SMN levels (HET, 1.821 ± 0.133; CON8i, 0.441 ± 0.096; 3UP8i, 1.317 ± 0.183) (). These increased levels persisted through P50, although they were substantially lower compared to heterozygous littermates (HET, 3.205 ± 0.177; 3UP8i, 1.683 ± 0.311). As expected, the decreased levels of FL-SMN2 transcripts from P12 to P50 translated to lower SMN levels (,). However, the SMN levels from 3UP8i-treated mice were still much higher at P50 than the levels of CON8i-treated mice at P12. Moreover, SMN levels were also elevated in muscle, albeit to a smaller extent than in the liver (Supplementary Figure S3c,d). Taken together, the SMN protein levels and SMN2 exon 7 splicing data verified the molecular basis for the improved survival and function in the 5058-Hemi hybrid SMA mice.

Effect of 3UP8i on neuromuscular junction pathology

SMA is characterized as a disease of the motor unit, and viable therapeutics should improve synaptic and muscular pathologies. To examine possible improvements in motor unit pathology following 3UP8i treatment, we analyzed the neuromuscular junction (NMJ) of intercostal (IC) muscles from P12 5058-Hemi hybrid SMA mice (). IC muscles are critical for maintaining proper respiratory function, and are one of the more vulnerable muscle groups in both SMA mouse models and human SMA patients.[37,38,39] The ICs from CON8i-treated SMA mice showed a significant decrease in fully innervated motor endplates (CON8i, 62.00% ± 1.780 versus HET, 86.25% ± 2.18) with a concomitant increase in partial innervations (CON8i, 36.00% ± 3.34 versus HET, 13.75% ± 2.18) (). Treatment with 3UP8i significantly increased the percentage of fully innervated endplates relative to CON8i-treated SMA mice (3UP8i, 77.25% ± 1.93 versus CON8i, 62.00% ± 1.780); and accordingly, brought partial innervations back toward those observed in control mice (HET, 13.75% ± 2.18; 3UP8i, 21.50% ± 2.10; CON8i, 36.00% ± 3.34) (). There was no change in motor end plate size (). These results indicated an inherent defect in either the formation and/or maintenance of the NMJ within the 5058-Hemi hybrid SMA mice, which was responsive to 3UP8i treatment.

Discussion

Recent preclinical studies employing ASO-based strategies aimed at SMN2 exon 7 splicing correction have achieved important milestones for the treatment of SMA, the leading genetic cause of infant mortality. While 18-mer or larger ASOs have been successfully employed in preclinical and clinical studies for the potential treatment of several diseases including SMA, in vivo studies that demonstrate the therapeutic potential of a splice-switching short ASO have yet to be completed. The only in vivo study reported thus far with a short ASO pertains to an 8-mer locked nucleic acid (LNA) against the seed sequence of a microRNA.[40] Interestingly, an iso-sequential 8-mer ASO with 2′OMeP modifications targeting the same microRNA turned out to be totally ineffective.[40] Although interesting, these findings indicated a limited scope of a short ASO to the LNA-based chemistry. On the other hand, our earlier report that a GCRS-targeting 8-mer ASO (3UP8) could efficiently correct SMN2 exon 7 splicing in SMA patient cells suggested the therapeutic potential of the widely used 2′OMeP chemistry.[21] However, despite stabilizing 2′OMeP modifications, in vivo 3UP8 failed to correct SMN2 exon 7 splicing. This was possibly due to terminal degradation and/or fast clearance from the blood. To combat these potential limitations, we explored chemistries that would confer greater in vivo efficacy. 3UP8 with PEG-282 and C3-propyl terminal spacers (3UP8i) induced robust SMN2 exon 7 inclusion in vivo, and a high level of specificity—the introduction of a single mismatch base in the middle of ASO completely eliminated the antisense response.

Bolstered by these initial results, we first tested the in vivo efficacy of 3UP8i in 5058-hemizygous mice, a severe model of SMA. A high dose (80 μg/g body weight) of 3UP8i, we observed an encouraging trend of body weight gain, improved righting reflex, and a modest 3-day extension in survival of these severe mice. The outcome was more convincing in the milder 5058-Hemi hybrid SMA mice, which provided a wider therapeutic window. In fact, treatment of 5058-Hemi hybrid SMA mice with 3UP8i produced improvements in weight and survival that began to mirror the efficacy of larger ISS-N1 targeting ASOs reported recently. Although it is difficult to pinpoint the exact cause of the 5058-Hemi hybrid's lessened severity, we suspect that this was the result of hybrid vigor. The noticeable caveat of this model is the development of hind limb edema. This pathology limits the types of motor tests that can be performed. Central nervous system (CNS) administration of ISIS-SMNRx, an ISS-N1 targeting MOE ASO, into SMA type III mice (5058-4copy) rescued a similar phenotype,[29] which may explain why peripheral administration routes were unable to correct this necrosis. The improvements in the Hemi-hybrid SMA mice corresponded to increased FL-SMN2 transcript and total SMN levels at P12. The P50 3UP8i-treated SMA mice showed decreased FL-SMN2 transcript and SMN levels compared to levels at P12, yet they were functionally normal aside from mild necrosis and decreased body size. Previous studies have indicated that the need for increased SMN levels is more critical at early as opposed to later time periods.[18]

While previous studies have demonstrated the effectiveness of very small ASOs (7-mer-11-mer) in vitro,[41,42] our findings with 3UP8i validate, for the first time, the therapeutic potential of a very small splice switching oligonucleotide in an animal model of a human disease. In addition to a better uptake, a small ASO imparts greater efficacy, provided it maintains target specificity.[43,44,45] Considering 3UP8i did not tolerate a single mismatch with the target, it fulfilled one of the rigorous requirements of target specificity. Future studies will be aimed at determining global off target effects of 3UP8i, particularly at higher concentrations. This information will be valuable for SMA research and more generally regarding the effects of short ASOs as they are developed for the treatment of other human diseases. The small helix generated by the annealing of 3UP8i to the GCRS is predicted to disrupt ISTL1 that we recently reported to be an inhibitory RNA structure formed by a unique LDI.[22] Hence, the mechanism of 3UP8i function is likely to be distinct from the group of ISS-N1 targeting ASOs that displace inhibitory protein(s) hnRNP A1/A2.[21,46] To further underscore the promising therapeutic potential of 3UP8i, it did not elicit an inflammatory response, as measured by qRT-PCR of common inflammation markers AIF1, CD68, IL-1A, IFNA, and IL-18 (Supplementary Figure S4). Our results also establish the potential of combining new chemistries to further enhance 3UP8i efficacy. In a broader context, our results validate the ability of a short ASO (8-mer) to correct a human disease in an animal model. This has important implications for the development of other therapeutics directed at small targets, including seed sequences of microRNAs, where the use of large ASOs carries the risk of poor specificity.

Materials and methods

Mice. Animals were cared for and monitored in accordance with the approved Lurie Children's Hospital Institutional Animal Care and Use Committee protocols. Animals were kept in a controlled environment at 25 °C and 50% humidity, with a 12 hours light/dark photoperiod, and were monitored daily for health. Three SMA mouse models were utilized in this study (see Supplementary Table S1): 5058-homozygous mice (SMN2;SMN2;Smn) were generated as described;[28,29] 5058-Hemi FVB/N SMA mice (SMN2;Smn) were generated as described;[18,27] 5058-Hemi hybrid SMA mice (SMN2; Smn) were generated by crossing 5058-homozygous FVB/N mice (SMN2;SMN2;Smn) with the heterozygous mice harboring the ▵7Smn Melki allele (Smn; C57-congenic), which produces litters in which 50% of the pups are SMA. The resulting SMA mice are FVB/N:C57/BL6 F1-hybrids, hemizygous for the SMN2 allele and compound heterozygotes for the Δ7Smn allele (SMN2;Smn). All offspring were genotyped on P0 using multiplex PCR of tail biopsies as previously described.[28] Mice were culled or combined to maintain litter sizes between six and nine pups, with at least two heterozygous siblings per cage. When possible, each cage of SMA mice under study (5058-Hemi FVB/N and 5058-Hemi hybrid) was evenly split between mismatch control (CON8i) and 3UP8i, which were delivered as blinded drugs known as oligo A and oligo B. One caveat of note was that by P6 or later, 3UP8i-treated 5058-hemizygous mice could often be unblinded based on weight and righting reflex.

Drug treatments and phenotyping. The 3UP8i and mismatch control (CON8i) 8-mer ASOs incorporated 2′OMeP, with a 5′ PEG-282 modification (Spacer 18) and a 3′ C3-propyl spacer, were synthesized (TriLink Technologies, San Diego, CA) as previously described.[16,21] ASOs were resuspended in 0.9% saline and passed through a 0.22 µm filter. Samples were coded as oligo A or oligo B to blind the studies. Mice were dosed on P0, P5, and P10 at either 20 or 80 µg/g of body weight via IP injection. Mice were monitored every morning for survival, weight, righting reflex (5058-Hemi only), and ECG (5058-Hemi hybrid only). For the righting reflex test, mice were placed in a supine position and assayed for the time-to-right on all four paws. Failure to right was set at 60 seconds. Functional death was defined as two consecutive days of greater than 20% weight loss or righting reflex failure. Noninvasive ECG measurements of conscious 5058-Hemi hybrid mice were recorded between P8 and P18, using the ECGenie system (Mouse Specifics, Quincy, MA) as previously described.[35,47]

RNA and protein analysis. Total RNA was isolated from 50 mg of tissue using the TRIzol reagent (Life Technologies, Carlsbad, CA) as per the manufacturer's instructions. DNase treatment and first-strand cDNA synthesis were performed as previously described.[48] FL- and Δ7-SMN2 end point PCR was performed as previously described.[29] Cycle conditions: 94 °C/45 seconds, 60 °C/45 seconds, 72 °C/45 seconds, 30 cycles. TaqMan assays of FL- and Δ7-SMN2 were performed on an Applied Biosystems (Carlsbad, CA) 7500 Fast Real-Time PCR System utilizing the following conditions: 1 cycle: 2 minutes 50 °C, 1 cycle: 10 minutes 95 °C, 40 cycles: 15 seconds 95 °C, 1 minutes 60 °C. The data were analyzed using the ΔΔCt method. Primer and probe sets used were previously described.[29] Mouse Gapdh (#4352339E, Applied Biosystems) was used to normalize samples. The data were analyzed using the ΔΔCt method.[49] Radioactive RT-PCR was performed as previously described.[21] 5058-Hemi hybrid mouse tissue was collected and homogenized as previously described.[50] Lysates were analyzed via immunoblot using the Li-COR Odyssey (Li-COR Technologies, Lincoln, NE) system as previously described.[48]

Immunostaining and microscopy. NMJ staining and imaging of whole-mount intercostal (IC) muscles from P12 5058-Hemi hybrid mice were performed as previously described.[32] Images were blinded and randomized prior to the analysis of NMJ motor endplate area and innervation status. Innervation status was determined as previously described.[28] Motor endplate area was defined by measuring the area of the background-subtracted of the α-BTX-594 signal using NIH Image J software (National Institutes of Health, Bethesda, MD). For each pup, n = 50 or more NMJs were analyzed.

Statistical analysis. Statistical analyses were performed using Graph Pad Prism software, V6.0 (GraphPad Software, La Jolla, CA). Kaplan–Meier survival curves were compared using log-rank tests. Statistical significance was determined using either a one- or two-way analysis of variance with a Bonferroni posthoc comparison, or Student's t-test when appropriate.

Figure S1. Effect of 3UP8i treatment on splicing of SMN2 in vivo. Figure S2. 3UP8 improves cardiac deficits in Hemi-hybrid SMA mice. Figure S3. 3UP8i increases FL-SMN2 levels in skeletal muscle of Hemi-hybrid SMA mice. Figure S4. 3UP8i ASO does not increase expression of several inflammation-related genes. Table S1. SMA models utilized in this study.