Intraperitoneal delivery of a novel drug‐like compound improves disease severity in severe and intermediate mouse models of Spinal Muscular Atrophy

Spinal muscular atrophy (SMA) is an autosomal recessive neurodegenerative disorder that causes progressive muscle weakness and is the leading genetic cause of infant mortality worldwide. SMA is caused by the loss of survival motor neuron 1 (SMN1). In humans, a nearly identical copy gene is present, called SMN2. Although SMN2 maintains the same coding sequence, this gene cannot compensate for the loss of SMN1 because of a single silent nucleotide difference in SMN2 exon 7. SMN2 primarily produces an alternatively spliced isoform lacking exon 7, which is critical for protein function. SMN2 is an important disease modifier that makes for an excellent target for therapeutic intervention because all SMA patients retain SMN2. Therefore, compounds and small molecules that can increase SMN2 exon 7 inclusion, transcription and SMN protein stability have great potential for SMA therapeutics. Previously, we performed a high throughput screen and established a class of compounds that increase SMN protein in various cellular contexts. In this study, a novel compound was identified that increased SMN protein levels in vivo and ameliorated the disease phenotype in severe and intermediate mouse models of SMA.

RG-7916 is currently enrolled in two clinical trials and belongs to the SMN-C class of Roche/PTC splicing modulators. SMN-C5 was reported to cause chemical shift perturbations of 7 nucleotides in the 5′-ss of exon 7, and it binds to U1-pre-mRNA complex at Exonic Splice Enhancer 2 (ESE2) in exon 7, which is upstream of the NVS binding site 14 . SMN-C2 and SMN-C3 recently have been shown to bind to AGGAAG motif, which is almost identical to the ESE2, and leads to enhanced binding of FUBP1 and KHSRP 15 . Previously, we described the optimization of a small molecule series that increased SMN protein levels through a distinct mode-of-action. These compounds elevated SMN and increased SMN protein half-life [16][17][18][19][20] . We identified a novel isoxazole compound "4 m" 16,20 , which was active in vitro but demonstrated poor brain and plasma exposure following intraperitoneal (IP) administration to mice. During bioavailability structure-activity relationship (SAR) driven optimization of the isoxazole series, the amide linkage was shown to be associated with poor plasma stability, but was improved by changing the heterocycle and reversing the amide linkage 16 . The identified lead molecule compound (Compound 27), referred herein as "LDN-2014", showed improved brain exposure after IP injection. This study represents a preliminary assessment of the effect of a novel compound, where we report the biological efficacy of LDN-2014 in two SMA animal models: the severe SMN∆7 and the intermediate Smn 2B/− mice.

Analysis of SMN protein levels in the SMNΔ7 mouse model of SMA following LDN-2014 delivery.
To examine the activity of LDN-2014 in vivo, compound was dissolved in DMSO and delivered via IP injections into the well-established severe mouse model of SMA, "SMNΔ7" (Smn −/− ; SMN2 +/+ ; SMNΔ7 +/+ ) 21 . DMSO was selected as the delivery vehicle due to limited aqueous solubility of this molecule (31 µM). LDN-2014 has an EC 50 of 0.3 µM in vitro; results in a brain exposure of C max = 41 μM and plasma levels of C max = 23 μM with a plasma half-life (T 1/2 ) of 2.2 hours following a single IP dose at 20 mg/kg ( 16 , Supplementary Table 1A,B), rendering 20 mg/kg suitable for preliminary assessment of in vivo activity. Starting on P2, animals were injected daily with 20 mg/kg LDN-2014 and sacrificed on P7. Brain, spinal cord and skeletal muscle (gastrocnemius) were collected, solubilized and SMN protein expression measured by semi-quantitative immunoblot in comparison to the previously published transcriptional SMN inducer LDN-76 20 (Fig. 1A,B). Administration of LDN-2014 significantly increased SMN protein expression in brain and spinal cord tissues compared to untreated and vehicle (DMSO) treated SMNΔ7 mice; however, there were no significant SMN increase in muscle (Fig. 1B).

IP injections with LDN-2014 extend survival and improve the phenotype of severe SMA mice.
The increases in SMN protein led us to investigate whether LDN-2014 extended survival of SMNΔ7 mice. Treated animals received a single IP injection daily from P2 onward, corresponding to a daily dose of 20 mg/kg LDN-2014 and DMSO. The vehicle-only cohort ( Fig. 2A,B) exhibited a median survival of 7.7 days compared to the 12.7 days for the untreated SMNΔ7 mice cohorts, an effect that we reported previously 20 . While the dose (2 ml/kg) of 100% DMSO is below the LD 50 of 6.2 ml/kg 22 observed in adult mice, this effect is enhanced in neonatal mice with progressive SMA phenotype. It is possible that the administration of DMSO exacerbates reported cardiovascular defects detected as early as P2 in SMNΔ7 mice 23,24 , as lower chronic doses of 100% DMSO also have been associated with toxicity in adult mice 25,26 . Administration of 100% DMSO resulted in cardiac failure, a pathology that has been observed in SMNΔ7 mice 23,27 . In contrast, LDN-2014 increased survival, with the longest lived animals living beyond 20 days ( Fig. 2A,B). Compared to vehicle-treated, the LDN-2014-treated cohort lived significantly longer (~200%) and showed an increased lifespan of ~30% beyond untreated SMA mice. The increase in survival was accompanied by a significant weight gain in LDN-2014 treated mice compared to vehicle and untreated SMA animals ( Fig. 2C), especially more evident past P9. To normalize for birth weight, body weight was represented as "birth-to-peak" weight, demonstrating that LDN-2014 treatment improves total body weight gain in SMA model mice (Fig. 2D).
To determine the impact of drug treatment upon gross motor function and strength, time-to-right assays were performed throughout the animals' life span starting at P7 (Fig. 3). This time frame was selected as a starting point since this is the age unaffected mice are consistently able to right themselves from a prone position. Drug-treated SMA mice were able to right themselves starting at P9 and showed further improvement by P12, with additional improvements until P14. In contrast, vehicle-alone (DMSO) and untreated SMA mice rarely were able to right themselves within the 30 seconds limit. These improvements in motor function are consistent with the weight and survival improvements in LDN-2014 treated animals (Fig. 3).

Delivery of LDN-2014 improves neuromuscular junction size and maturation in SMNΔ7 mice
after treatment. To begin to examine the structural integrity of the motor units, neuromuscular junctions (NMJs) in the transversus abdominis (TVA) of vehicle and LDN-2014 treated SMN∆7 mice were analyzed by indirect immunofluorescence. The TVA was selected because it has previously been shown that neurons innervating this muscle are consistently vulnerable to disease and exhibit a significant degree of pathology in SMN∆7 mice and other mouse models of SMA. Pathological hallmarks include endplate morphology and denervation as quantified by a reduction in endplate occupancy [28][29][30][31][32] . Similarly to above, four mice per group were injected daily with LDN-2014 (20 mg/kg) or vehicle by IP delivery until P7. For control NMJ immunohistochemistry, unaffected mice (n = 4) were sacrificed at P7 for comparison. Pre-synaptic axons and nerve terminals were labeled with antibodies against total neurofilament and synaptic vesicle protein 2 (SV2) and post-synaptic acetylcholine receptors were labeled with rhodamine-conjugtaed α-Bungarotoxin (Fig. 4A). Administration of LDN-2014 significantly improved NMJ occupancy, with nearly 90% of endplates remaining fully occupied after drug-treatment compared to 50-60% in DMSO and untreated samples (Fig. 4B,C). Collectively, these results demonstrate that LDN-2014 increased SMN expression in disease-relevant tissues, and that this correlated with phenotypic improvements the motor units of SMA model mice, ultimately resulting in an extension in survival. Previous , and untreated (n = 10) SMN∆7 mice. (D) Body weight expressed as percentage weight gain from "birth-to-peak", from all cohorts. Statistical significance is represented by "*" where p ≤ 0.05; "**"p ≤ 0.01; "***"p ≤ 0.001; "****"p ≤ 0.001; "ns" p > 0. 05 studies have shown that targeted SMN expression in motor neurons rescued the neuronal SMA phenotype in SMN∆7 mice 33 ; however, this was not associated with a similar extension in survival. In contrast, an antisense oligonucleotide delivered by intracerebroventricular (ICV) injection significantly rescued the SMA phenotype in the SMN∆7 mice 34,35 . Therefore, we speculate that the increases in the periphery and the central nervous system result in the improvements in survival as well as gross motor function.   (Fig. 5A,B). Treatment with the LDN-76 analog resulted in a similar average lifespan compared to vehicle and untreated Smn 2B/− mice. In contrast, a significant extension in survival was observed in LDN-2014 treated Smn 2B/− animals, as the median survival for these animals was 73 days; however, more than 40% of the cohorts were alive when the experiment was halted at P110 (Fig. 5A,B). This was accompanied by improvements in weight gain, although these mice did not reach the maximum weights of unaffected control animals (Fig. 5C,D). The LDN-2014 treated mice maintained their weights throughout until P110, when the experiment was halted. The general appearance of LDN-2014 treated mice, normal posture and smooth hair coats, was improved compared to LDN-76 or untreated mice (Supplementary Fig. 1).

Discussion
Clearly, increasing SMN protein is advantageous for SMA therapy, and exon 7 splicing and SMN2 promoter activation are two well-examined modes-of-action. However, additional molecular pathways exist to elevate SMN levels; including approaches to prolong the half-life of SMN, as described for the proteasome inhibitor Bortezomib 37 , and the autophagy inhibitor 3-methyladenine 38 . Administration of Bortezomib to the severe SMN∆7 mice improved motor function and neuromuscular junction size, but did not extend survival. Due to low brain exposure, a follow-up study explored the combined treatment Bortezomib with a p-glycoprotein inhibitor to increase the CNS uptake. The combined treatment starting at P4 resulted in increased SMN protein levels in the brain and enhanced survival by 1.5 days in the severe SMN∆7 mice 39 . Administration of an autophagy inhibitor increased SMN protein levels and extended the lifespan by 1.5 days, while Rapamycin, an autophagy inducer, worsened the condition of the severe SMN∆7 mice 38 . Recently, administration of ML372, another novel small molecule that stabilizes the SMN protein through blockade of SMN ubiquitination 40 , resulted in increased SMN protein levels in brain, spinal cord and muscle and a survival extension of four more days. Although we do not know the mechanism by which LDN-2014 alters SMN protein levels, we previously found that LDN-2014 does not alter general proteasome or autophagy activity 17 . Current research to determine the pathway is ongoing, however, the mechanism(s) appears to be able to stabilize SMN protein derived from the human SMN2 gene as well as the endogenous Smn (albeit mutated) gene. Examining the commonalities of these genes and/or the downstream regulators could provide insight into the mechanism as well as identifying a target for further drug development. Based on these studies and our results, we propose that post-translational stabilization of the SMN protein may show a greater benefit in less severe SMA mice models, which could render them suitable to be of clinical benefit in combination with splicing modifiers and antisense oligonucleotides to prolong the time between treatments. In summary, we conclude that LDN-2014 is biologically active in two important animal models of SMA, and although limited by its solubility, it expands this mechanistically distinct avenue for the treatment of SMA.

Methods
Animal procedures and delivery of therapeutics. All animal experiments were carried out in accordance with protocols approved by the Animal Care and Use Committee of the University of Missouri. The animals had free access to water and food. Original breeder pairs of Smn +/− ; SMN2 +/+ ; SMN∆7 +/+ mice were purchased from The Jackson Laboratory (JAX ® Mice and Services, 610 Main Street Bar Harbor, ME 04609 USA). Litters were genotyped on the day of birth (P1) as previously described 41 , utilizing multiplex PCR on tail biopsy material with the following primer sets for the Smn gene: mSmn-WT FWD (5′-tctgtgttcgtgcgtggtgacttt-3′) and mSmn-WT REV (5′-cccaccacctaagaaagcctcaat-3′) and for the Smn knockout: SMN1-KO FWD (5′-ccaacttaatcgccttgcagcaca-3′) and SMN1-KO REV (5′-agcgagtggcaacatggaaatcg-3′). To control for litter size, treated and untreated control SMA mice were raised with two unaffected heterozygous siblings. The Smn 2B/− mice were bred from two colonies: mSmn +/− heterozygotes (stock no. 006214; Jackson Laboratory) and Smn 2B/2B homozygotes. Intraperitoneal injections (IP) started the day after birth and continued daily or every other day as indicated. Drug was solubilized in 100% DMSO. Drug solutions and vehicle (DMSO) were injected in 10 µl per gram body weight. The Time-toright (TTR) assessment started on P7 since healthy animals start to turnover at this time and the measurements were recorded daily. The test was performed on a flat surface and terminated after 30 seconds. A 'Failure' was recorded if an animal did not turn in this time frame. All TTR trials concluded on P20.
Immunohistochemistry of NMJs. Animals from LDN-2014 treatment and vehicle groups were anesthetized at age P7 by inhalant Isoflurane, (1-chloro-2, 2,2-trifluoroethyl difluoromethyl ether; Fluriso TM , VetOne, Boise, ID 83705) followed by transcardiac perfusion with PBS solution and fixed with 4% paraformaldehyde (Sigma-Aldrich). Transversus abdominis muscle was rapidly dissected to yield whole-mount muscle preparations. Pre-synaptic axons and nerve terminals were stained with anti-neurofilament (1:2,000; catalog AB5539, Chemicon, EMD Millipore) and anti-SV2 (1:200 dilution, Chemicon ® , MilliporeSigma, St Louis, MO, USA) for synaptic vesicles (NF/SV2). Acetylcholine receptors (AchR) were labeled with Alexa Fluor 594-conjugated α-Bungarotoxin (Life Technologies). Endplate occupancy was quantified to the following criteria: fully occupied is defined as complete coverage of the post-synaptic endplate by the branches of the pre-synaptic nerve terminal; partially occupied is defined as partial withdrawal of the pre-synaptic nerve terminal from the post-synaptic endplate; vacant is defined as complete withdrawal of the pre-synaptic nerve terminal from the post-synaptic endplate. Labeled pre-synaptic nerve terminals were visualized using 488 nm excitation and 520 nm emission optics, while endplates were visualized using 543 nm excitation and 590 nm emission optics. Quantification was performed blind, and incorporated a minimum of 50 NMJs from at least three distinct and randomly-selected fields of view per muscle per mouse (×40 objective; Leica DM5500 B, Leica Microsystems Inc.). Immunohistochemistry and quantification. SMN∆7 mice were treated with vehicle DMSO (vehicle) or LDN-2014 (20 mg/kg) for 7 days. Mice were sacrificed by brief exposure to general anesthetic (Isoflurane; Fluriso TM , VetOne, Boise, ID 83705) followed by cervical dislocation. Tissues (brain, spinal cord and gastrocnemius muscle) were extracted and immediately frozen in liquid nitrogen. The frozen tissues (~100 mg) were homogenized in JLB buffer (JLB: Jurkat lysis buffer; 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 20 mM NaH2(PO4), 25 mM NaF, 2 mM EDTA, 10% glycerol, 1% Triton X-100 and protease inhibitors (Roche, Indianapolis, IN, USA)). Equal amounts of protein were separated on 12% SDS-PAGE gels and transferred onto PVDF membranes. Membranes were blotted using anti-SMN monoclonal antibody (SMN specific monoclonal antibody (BD Biosciences, San Jose, CA, USA) diluted 1: 300 in TBST (Tris-buffered Saline Tween20 (10 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.2% Tween20)) in 1.5% dry milk. Blots were visualized by chemiluminescence on a Fujifilm imager LAS-3000 (FujiFilmUSA, Hanover Park, IL, USA) and the corresponding software. To verify equal loading the Westerns blots were then stripped using H 2 O 2 for 30 minutes at room temperature and re-probed with anti-Actin antibody produced in rabbit