Discovery of a CNS penetrant small molecule SMN2 splicing modulator with improved tolerability for spinal muscular atrophy

Spinal muscular atrophy (SMA) is a motor neuron disease, typically resulting from loss-of-function mutations in the survival motor neuron 1 (SMN1) gene. Nusinersen/SPINRAZA, a splice-switching oligonucleotide that modulates SMN2 (a paralog of SMN1) splicing and consequently increases SMN protein levels, has a therapeutic effect for SMA. Previously reported small-molecule SMN2 splicing modulators such as risdiplam/EVRYSDI and its analog SMN-C3 modulate not only the splicing of SMN2 but also that of secondary splice targets, including forkhead box protein M1 (FOXM1). Through screening SMA patient-derived fibroblasts, a novel small molecule, designated TEC-1, was identified that selectively modulates SMN2 splicing over three secondary splice targets. TEC-1 did not strongly affect the splicing of FOXM1, and unlike risdiplam, did not induce micronucleus formation. In addition, TEC-1 showed higher selectively on galactosylceramidase and huntingtin gene expression compared to previously reported compounds (e.g., SMN-C3) due to off-target effects on cryptic exon inclusion and nonsense-mediated mRNA decay. Moreover, TEC-1 significantly ameliorated the disease phenotype in an SMA murine model in vivo. Thus, TEC-1 may have promising therapeutic potential for SMA, and our study demonstrates the feasibility of RNA-targeting small-molecule drug development with an improved tolerability profile.

Spinal muscular atrophy (SMA) is a neurodegenerative disorder in which there is a loss of lower motor neurons (MNs) that project from the spinal cord and the brain stem. This leads to muscle atrophy and difficulties in breathing and walking, which may require tracheotomy and artificial respiration assistance.
SMA is typically inherited as an autosomal recessive trait, with most patients having loss-of-function mutations in the survival motor neuron 1 (SMN1) gene. The SMN2 gene, a paralog of SMN1, differs from SMN1 at only 2 base pairs in the open reading frame, but the amino acid sequences encoded by the two genes are identical. One of the base pair differences causes a translationally silent, single-nucleotide transition in SMN2 at position 6 of exon 7 (c6t), which not only disrupts binding sites for positive splicing regulators 1 but also creates binding sites for negative splicing regulators 2 . In addition, this change (c6t) strengthens an extended inhibitory context 3,4 . As a result, most of the mRNA transcribed from the SMN2 gene is of the Δ7 form, which skips exon 7 via splicing. However, the full-length SMN2 mRNA containing exon 7 (FL-SMN2) is also produced at a rate of 5-10% of the total transcripts. An increase in the copy number of the SMN2 gene ameliorates the severity of SMA, and loss and compensation strongly correlate with the onset and progression of the disease 5 ; hence, in recent years, SMN2 has become an attractive target of drug development.
Nusinersen/ASO10-27/SPINRAZA, an approved antisense oligonucleotide (ASO) drug for the treatment of SMA, directly targets intronic splicing silencer N1 (ISS-N1) in intron 7 of SMN2 6,7 , which modulates SMN2 splicing, and leads to an increase in SMN protein levels 8 . According to a Phase III trial (ENDEAR), nusinersen showed moderate therapeutic effects with delayed disease progression and ventilation timing 9 , but was unable to penetrate the blood-brain barrier (BBB). However, nusinersen treatment requires repeated intrathecal administration, and is associated with several side effects, including post-lumbar puncture syndrome characterized by back pain, headache, and fever, and the production of neutralizing antibodies (NAb), raising concerns associated with its clinical use 10 . Furthermore, an advanced medical procedure to treat SMA patients with scoliosis

TEC-1 increases the expression level of FL-SMN2 mRNA and decreases the expression level of Δ7 mRNA.
To identify splicing modulators with improved selectivity for SMN2, over 300 compounds were designed from our compound collection and evaluated, optimized using SMA patient-derived fibroblasts 20 . Quantitative polymerase chain reaction (qPCR) was then used to evaluate the splicing activity of these candidates against SMN2 and secondary splice targets (e.g., FOXM1). Subsequently, a single oral dose was administered to FVB mice to assess bioavailability and brain penetration. As a result of these screens, we identified a compound, TEC-1, which increased the form of SMN2 that includes exon 7 (FL-SMN2), and decreased the form that skips exon 7 (Δ7) in a concentration-dependent manner (Fig. 1a,b,e,f). These results led us to conclude that TEC-1 is an SMN2 splicing modulator. The SMN-C series (SMN-C3 and risdiplam) also modulated SMN2 splicing, with effects similar to those noted in previous reports 15,16 (Fig. 1c-f).
TEC-1 does not strongly affect FOXM1 splicing. FOXM1 is a protein involved in cell division, and reduced abundance of its major splicing isoform FOXM1b/c is observed in the G1/S phase, whereas an increase in this isoform is detected in the G2/M phase. Moreover, FOXM1 splicing contributes to chromosome missegregation and toxicity at the level of cell cycle/division 21,22 . Previous reports revealed that risdiplam induces micronucleation in vitro and in vivo 16 . We therefore evaluated changes in FOXM1 splicing variants that occurred under the influence of the SMN-C series by reverse transcription (RT)-PCR (Fig. 2a,b). In GM03813 fibroblasts treated with the SMN-C series, the level of the FOXM1b/c variant was decreased, while the levels of the FOXM1 ΔC and 1a variants were increased, suggesting that the SMN-C series modulates the splicing of FOXM1 (Fig. 2be). Interestingly, TEC-1 did not strongly affect the splicing of FOXM1 (Fig. 2b-e). To estimate the selectivity of splicing between SMN2 and FOXM1, we determined the EC 1.5× value of SMN2, representing the concentration at which there is a 50% increase in total FL-SMN2 mRNA. Next, we calculated EC 50 values of FOXM1b/c to determine the concentrations showing a 50% reduction in total FOXM1b/c mRNA levels. We then normalized EC 50 of FOXM1b/c based on the EC 1.5× value of FL-SMN2 (FOXM1b/FL-SMN2) and found that its value in the presence of TEC-1 was over 61 (Table 1 and Fig. 2c-e). By contrast, these values in the presence of SMN-C3 or risdiplam was 8 and 8, respectively, indicating that the SMN-C series strongly affects FOXM1 splicing compared with TEC-1 (Table 1 and Fig. 2c-e). We next examined whether TEC-1 induces chromosomal damage using an in vitro micronucleus assay in human lymphoblasts (TK6 cells) under three treatment conditions (Table 2). In a 3-h treatment without S9 mix, followed by a 21-h recovery period, 2.3 µg/mL (5.9 µM) TEC-1 caused a significant increase in the incidence of micronucleated cells. We confirmed that TEC-1 induces a concentrationdependent increase in the number of micronucleated cells. However, the relative population doubling (RPD) value at 2.3 µg/mL (5.9 µM) was 35.0%, suggesting that this positive response was secondary to cytotoxicity. In a 3-h treatment including S9 mix, followed by a 21-h recovery period, 2.7 µg/mL (6.9 µM) TEC-1 significantly increased the incidence of micronucleated cells. The RPD value at this concentration was 58.7%. However, in this condition, a concentration-dependent increase was not observed. In a 24-h treatment without S9 mix, there was no significant increase in the number of micronucleated cells at any TEC-1 concentration compared to the nega- www.nature.com/scientificreports/ tive control group. We therefore concluded that TEC-1 and its metabolites in the liver do not induce micronucleation, indicating that no chromosomal damage occurs. Conversely, risdiplam induces micronucleation in vitro and in vivo 16 . In summary, these results suggest that avoiding FOXM1 splicing helps to prevent micronucleation. The FOXM1 sequence in cynomolgus monkeys is nearly identical to that in humans, with only a single mismatch in the sequence between exon 9 and the donor site ( Supplementary Fig. 1a, yellow highlight). Notably, FOXM1 mis-splicing was similarly observed in cynomolgus monkey cells (NCMDF) exposed to risdiplam (Supplementary Fig. 1b-d), for which the toxicology is discussed below.
TEC-1 only slightly modulated GALC splicing. SMN depletion was reported to reduce the total amount of GALC mRNA in mouse MNs 23 . The SMN-C series, which increases the amount of SMN protein, unexpectedly decreased the total amount of GALC mRNA in RNA sequencing studies of SMA patient-derived fibroblasts 15,17 . We speculated that this reduced GALC expression is caused by the effects of the SMN-C series on secondary splice targets, rather than by an effect mediated by the SMN protein itself. To test this possibility, we evaluated the ability of SMN-C3 or TEC-1 to change the expression of GALC mRNA. SMN-C3 strongly reduced the GALC mRNA level compared to TEC-1 (Fig. 3a). The EC 50 values of GALC, which reflects a 50% reduction in total GALC mRNA, normalized to the EC 1.5× value of FL-SMN2 (GALC/FL-SMN2) was over 61 for TEC-1, whereas that of SMN-C3 was 4, suggesting that TEC-1 is a more selective SMN2 splicing modulator (e) qPCR analysis of FL-SMN2 transcripts in SMA type II fibroblasts (GM03813) exposed to each compound for 24 h. The amounts of mRNA were normalized to those of GAPDH. (f) qPCR analysis of Δ7 transcripts of GM03813 cells exposed to each compound for 24 h. The amounts of mRNA were normalized to those of GAPDH. Data are from two biologically independent samples in e, f. a-d were derived by by Axcelead Drug Discovery Partners, Inc.

Scientific Reports
| (2020) 10:17472 | https://doi.org/10.1038/s41598-020-74346-9 www.nature.com/scientificreports/ RT-PCR analysis of FOXM1 variants of GM03813 fibroblasts exposed to each compound for 24 h. RT-PCR products of FOXM1 variants were separated by agarose gel electrophoresis and visualized with ethidium bromide. Standard FOXM1 fragments corresponding to three isoforms (Oligo STD) are shown. Molecular weight markers are shown on the right (bp). The full-length gel is presented in Supplementary Fig. 1e. (c-e) qPCR analysis of each FOXM1 variant of GM03813 cells exposed to each compound for 24 h. The amounts of each variant were normalized to those of GAPDH. Data are from two biologically independent samples in (c-e). www.nature.com/scientificreports/ compared with the SMN-C series ( Table 1, Fig. 3a). To investigate the mechanism of action for the reduction in GALC mRNA expression induced by the SMN-C series, we assessed whether the decrease in GALC mRNA by SMN-C3 is affected by cycloheximide (CHX), which suppresses nonsense-mediated mRNA decay (NMD). The GALC mRNA reduction was rescued by CHX exposure (Supplementary Fig. 2a). Direct sequencing of RT-PCR products revealed that a 34 nucleotide (nt) sequence derived from intron 6 was inserted between exons 6 and 7 ( Supplementary Fig. 2b cryptic exon start, Fig. 3b). Thus, we confirmed a previously unreported 34 nt cryptic exon by RT-PCR and qPCR, the inclusion of which was more significantly induced by SMN-C3 than by TEC-1 (Fig. 3c,d). This cryptic exon inclusion in GALC causes NMD since an in-frame TAG codon (premature termination codon) exists 12 nt from the 5ʹ-terminus of the cryptic exon (22 nt from the 3ʹ end of the cryptic exon) ( Supplementary Fig. 3a, highlighted green). Thus, we conclude that the SMN-C series promotes inclusion of this abnormal 34 nt cryptic exon, which contributes to the reduction in total GALC mRNA expression. Notably, reductions in GALC by the SMN-C series were not observed in cells derived from dogs and rats, whose genomes lack the long intronic region that includes the 34 nt sequence ( Supplementary Fig. 3a, d, e). However, the reduction in GALC caused by the SMN-C series in cynomolgus monkey cells was stronger than that in green monkey cells ( Supplementary Fig. 3a-c). This difference is likely related to the number of mismatched sequences near the cryptic exon and donor site ( Supplementary Fig. 3a-c). Furthermore, the sequences of the human GALC cryptic exon and human FOXM1 exon 9/A2 are quite similar (only 3 mismatches in the included exon and donor site, Supplementary Fig. 11, highlighted yellow). These results suggest that sequence similarity contributes to the inclusion of the GALC cryptic exon by the SMN-C series.
GALC encodes a lysosomal enzyme that degrades psychosine, a highly toxic glycolipid 18 . Mutations in GALC underlie Krabbe disease, an autosomal recessive disorder in which psychosine accumulates in the brain. To test whether reduced GALC mRNA impacts GALC enzymatic function, the activity of GALC protein was measured. Expression of UDP glycosyltransferase 8 (UGT8), a key enzyme for the production of psychosine, is low in the skin 24 . Since UGT8 expression is high in oligodendrocytes 25 , its glioma cell line Hs683 was utilized for this assay. SMN-C3 strongly decreased the activity of GALC compared to TEC-1, similar to its effects on GALC mRNA expression (Fig. 3a,e). TEC-1, even at a concentration of 3000 nM, induced a slight decrease in GALC enzymatic activity, and caused only a 1.9-fold increase in psychosine compared to the control (Fig. 3e,f). Notably, a 1.9-fold increase in psychosine is tolerable, as heterozygous carriers of Krabbe disease have an approximately 1.8-fold increase in plasma psychosine compared with unaffected controls 18 . By contrast, 3000 nM SMN-C3 triggered an approximately 22-fold increase in psychosine (Fig. 3f), which is greater than the approximately 15-fold increase in plasma psychosine reported in early-infantile-onset Krabbe disease 18 . Table 2. TEC-1 did not induce micronucleus formation in TK9 cells with or without a rat liver microsome fraction (S9). Experiments were performed under the following three treatment conditions. 1: short-term treatment for 3 h without S9 mix, followed by a 21-h culture (3 h −S9 mix, shown in groups 1-5), 2: short-term treatment for 3 h with S9 mix, followed by 21-h culture (3 h + S9 mix, shown in groups 6-10), 3: continuous treatment for 24 h without S9 mix (24 h −S9 mix, shown in group [11][12][13][14][15]. Mitomycin (MMC) for 3 h −S9 mix, cyclophosphamide (CP) for 3 h + S9 mix, and colcemid (COL) for 24 h −S9 mix were used as positive controls. These articles have been recommended for use in the OECD test guideline (Test No. 487) 49 . DMSO was served as negative control. The numbers of micronucleated cells were counted in 4000 cells and percentages of micronucleated cells were calculated for statistical analyses. ##  www.nature.com/scientificreports/ TEC-1 does not induce abnormal HTT splicing. Knockdown of the HTT gene in young mice was reported to cause acute pancreatitis 19 . Previous RNA sequencing studies have reported that the SMN-C series reduces total HTT mRNA 15,17 , which we confirmed in the present study using qPCR (Fig. 4a). However, SMN depletion does not alter the level of HTT mRNA in mouse MNs 23 , suggesting that HTT reduction by the SMN-C series is caused by secondary splice target effects. To elucidate the mechanism by which HTT is reduced, RT-PCR products from cells cultured with both risdiplam and CHX were sequenced. In the presence of risdiplam with CHX, the HTT mRNA included a previously unreported 115 nt derived from part of intron 49, just after the end of exon 49 (Fig. 4b, Supplementary Fig. 4b), which we also confirmed by RT-PCR and qPCR (Fig. 4c,d). As in the case with the GALC splicing isoforms, an in-frame TAG codon (premature termination codon) was present 58 nt upstream from the 3ʹ end of the cryptic exon (57 nt from the 5ʹ end) ( Supplementary Fig. 5a, highlighted green). These results suggest that the HTT mRNA with the cryptic exon was also degraded via NMD. Importantly, TEC-1 did not substantially impact the inclusion of the HTT cryptic exon or reduction of HTT at the mRNA and full-length protein levels (Fig. 4a,d,e). www.nature.com/scientificreports/

TEC-1 modulates SMN2 splicing and shows disease-modifying effects in induced pluripotent stem cell (iPSC)-derived MNs of an SMA patient. To test whether TEC-1 impacts SMN2 splicing in
MNs, we cultured MNs differentiated from iPSCs derived from a patient with SMA type II 26 . TEC-1 showed clear ability to modulate SMN2 splicing at a concentration of 30 nM, and increased SMA protein at a concentration of 100 nM (Fig. 5a-c). Consistent with previous results, we found that SMN-C3 modulates SMN2 splicing in our culture conditions 15 , with similar effects to those of TEC-1 ( Supplementary Fig. 6). TEC-1 has also been shown to increase the levels of choline acetyltransferase (ChAT), which are decreased in MNs from patients with SMA compared to those of healthy subjects 27 (Fig. 5d). These results suggest that TEC-1 can modulate SMN2 splicing in MNs, and is therefore a potential therapeutic drug.
To ensure its therapeutic benefit to SMA patients in combination with nusinersen, it is important to know whether TEC-1 inhibits the activity of nusinersen. Low concentrations of TEC-1 did not inhibit the effects of nusinersen in SMA patient-derived fibroblasts, whereas at 1 and 3 µM, it significantly enhanced the effects of nusinersen ( Supplementary Fig. 7).
TEC-1 ameliorates the disease phenotype in a murine model of SMA. To examine whether TEC-1 alleviates the SMA phenotype, we evaluated the effects of TEC-1 on a murine model of SMA (SMNΔ7 mice), which show a severe disease phenotype 28 . First, we confirmed that orally administered TEC-1 could penetrate the BBB in adult wild-type FVB mice (Supplementary Fig. 8 and Supplementary Table 4). Next, since oral administration to neonatal SMNΔ7 mice increases the rate of unexpected mortality 29 , we intraperitoneally www.nature.com/scientificreports/ administered TEC-1 before weaning (P2 to P23). Intraperitoneally administered TEC-1 was absorbed, excreted, and crossed the BBB in juvenile wild-type FVB mice (Supplementary Fig. 9 and Supplementary Table 5). SMNΔ7 mice were then treated with 2, 6, and 20 mg/kg TEC-1 once a day by intraperitoneal injection from P2 to P23. After P23, the route of administration was changed to oral administration, since oral drug development is the ultimate objective. SMNΔ7 mice treated with TEC-1 showed an increased body mass compared to that of vehicle-treated SMNΔ7 mice (Fig. 6a,b). To examine the effect of TEC-1 on motor function in SMNΔ7 mice, a righting reflex test was performed at P6, P11, and P16. Latency to righting in vehicle-treated SMNΔ7 mice was prolonged compared to that of control heterozygous mice. However, SMNΔ7 mice treated with 2 mg/kg TEC-1 showed a significantly shortened latency time compared to that of vehicle-treated mice at P16 (Fig. 6c-f). Importantly, TEC-1 enhanced the survival of SMNΔ7 mice compared with that of vehicle-treated SMNΔ7 mice (Fig. 6g). Taken together, TEC-1 has potential to ameliorate SMA phenotypes, improving both survival and motor function.

Discussion
TEC-1 was identified and optimized using a cell-based screening system with fibroblasts from a patient with SMA, in contrast to the previous use of a cell-based assay using an SMN2 minigene reporter, which led to the identification of the SMN-C series 15 and the NVS-SM series 30 . To improve the clinical tolerability, we simultaneously evaluated the splicing of SMN2 and secondary splice targets (e.g., FOXM1) in SMA patient-derived fibroblasts, which maintain the intrinsic structure of cellular mRNAs. Our approach confirmed that TEC-1 has improved selectivity toward SMN2 splicing over three representative secondary splice targets. The characterization of the molecular target or binding site of TEC-1 is beyond the scope of this study. Previous reports revealed that the SMN-C series and NVS-SM series directly interact with the major groove of the RNA duplex generated by the 5ʹ splicing site of exon 7 and U1 snRNA 17 . Furthermore, the SMN-C series also binds to purine-rich regions within exon 7, and this interaction is proposed to be affected by several RNA-binding protein factors 17,31,32 . Thus, multiple interactions of RNA structure and compounds contribute to selective SMN2 splicing by the SMN-C  Supplementary Fig. 10. Data in (a-d) represent means ± SEM of three independent assessments per concentration. *p < 0.05, **p < 0.01, ***p < 0.001 as assessed by one-way ANOVA followed by Dunnett's test using DMSO-treated cells as a control (100%). www.nature.com/scientificreports/  www.nature.com/scientificreports/ series. Further research is required to reveal that TEC-1, which has a similar pharmacophore to the SMN-C series (Fig. 1), interacts in the same manner and strongly and/or selectively binds to SMN2 over the three secondary splice targets compared with the SMN-C series. Secondary splice target effects may lead to a high risk of toxicity in vivo. Indeed, in a 39-week toxicological study of cynomolgus monkeys treated with risdiplam, pathological intestinal and pancreatic changes, and irreversible retinal degeneration were observed 16 . Ratni et al. 16 hypothesized that FOXM1 may be the most important target contributing to these toxicities, although the involvement of other secondary splice targets such as STRN3, APLP2, and MADD remains unclear. Our in vitro analysis of cynomolgus monkey cells also suggested that aberrant splicing of FOXM1 contributes to the toxicity in risdiplam-treated cynomolgus monkeys (Supplementary Fig. 1). To further characterize the selectivity of splicing by TEC-1, in future studies, we plan to investigate not only these secondary splice targets (STRN3, APLP2, and MADD) but also other unexpected targets using RNA sequencing and qPCR. Roche/PTC, when developing risdiplam, had set an exposure cap in a Phase I clinical trial with its concentration showing an area under the curve (AUC) 0-24h value of 1500 ng·h/mL in plasma (18 mg/individual) 14 . Risdiplam administered under this exposure cap showed a comparable clinical effect to AVXS-101 in motor function assessments. Only about half of the patients achieved a clinical readout, which was the ability to sit unsupported for several seconds (5 or 30 s) during interim analysis 12,13 . Therefore, there is still room to improve motor function delays. Based on these reports and our results, we believe that risdiplam was being tested at lower doses in clinical trials to avoid toxic effects such as FOXM1 or GALC splicing and micronucleus induction 16 . Risdiplam aberrantly included FOXM1 exon 9, whose nucleotide sequence is similar to an unreported GALC cryptic exon ( Supplementary  Fig. 11, highlighted yellow). We hypothesize that the secondary splice target effects of the SMN-C series stem from these conserved nucleotide sequences of RNA. By contrast, TEC-1 showed relatively higher selectivity against FOXM1, GALC, or HTT. Our screening strategy with cells which maintain the intrinsic structure of cellular mRNAs further proved the validity of excluding toxic cryptic exons in specific secondary splice targets, enabling the identification of small-molecule SMN2 splicing modulators with a more tolerable clinical profile. The biological activity of 106 representative proteins that are drug discovery targets was investigated with a selectivity-panel from Eurofins Inc. TEC-1 weakly affected the biological activity of only two targets, human acetylcholinesterase and rat N type calcium channel, with IC 50 values of 1.19 μM and 8.80 μM, respectively. TEC-1 is therefore an effective and safe drug since the EC 1.5× value of FL-SMN2 (49.1 nM) is markedly lower than the two aforementioned IC 50 values. Furthermore, TEC-1 showed good profiles with general in vitro toxicity and adsorption, distribution, metabolism, and excretion (ADME) tests (items: solubility, PAMPA/membrane permeability; MDR1/BBB penetration, stability; CYP inhibition, CYP induction; PXR, hERG, cytotoxicity/liver toxicity, umu/mutagenicity, Ames/mutagenicity, in vitro micronucleus test/tumorigenesis). It should be noted that TEC-1 did not induce micronucleus formation in human cells in vitro, in contrast to risdiplam. Collectively, our findings demonstrate that TEC-1 is a key member of a class of compounds with a low risk of acute and chronic side effects for SMA treatment.
Cardiac abnormalities 33,34 , pancreatic defects 35,36 , and liver deficits 37,38 have been reported in patients with SMA and in the murine model of SMA. Small-molecule compounds delivered systemically have the potential to increase the expression of SMN proteins, which are expressed ubiquitously in humans. Pinna and tail necrosis have been reported in the SMA murine model treated with SMN gene therapy 39 or low doses of nusinersen 40 , which may result from insufficient systemic delivery or the impact of large-molecule drugs. Long-term clinical observations of these therapeutics are therefore recommended 41,42 . The small-molecule compound TEC-1 is expected to be a more promising drug for the long-term treatment of patients with SMA compared with the two approved large-molecule drugs whose systemic delivery is limited. From this single pharmacokinetics studies in juvenile (intraperitoneal) and adult (oral) FVB mice, the Kp values, which indicate the ratio of the compound in the brain and plasma, were greater than 1 (Supplementary Figs. 8 and 9). Thus, TEC-1 was efficiently distributed to the brain in both juveniles and adults, suggesting a favorable systemic delivery that includes the brain. Although the drug-metabolizing enzyme of TEC-1 has not yet been identified, TEC-1 was excreted from the plasma and brain in a time-dependent manner in juvenile mice. This indicates that TEC-1 can be developed for use in juvenile patients whose compositions of drug-metabolizing enzymes are different from those of adults. All SMA model mice administered 20 mg/kg TEC-1 intraperitoneally survived until weaning. TEC-1 was absorbed from the peripheral circulation, crossed the BBB, and completely rescued SMA phenotypes in vivo until weaning age. Interestingly, 66% (2 of 3 mice) of the SMA mice administered TEC-1 intraperitoneally survived, whereas 40% (4 of 10 mice) of SMA mice receiving oral administration survived, suggesting that bioavailability of the drug administered intraperitoneally is more favorable than when administered orally after weaning. To improve oral bioavailability, new formulation technologies 43 will be applied to TEC-1 for the development of SMA therapeutics. At this time, the TEC-1 dosage is higher than that of risdiplam to extend the lifespan in the SMA murine model 16 . Finally, we will examine whether formulated TEC-1 increases the oral bioavailability, and enhances the exposure cap and therapeutic effects compared with risdiplam for SMA in preclinical and clinical studies.
In conclusion, these findings indicate that TEC-1 has selectivity toward SMN2 splicing over three secondary splice targets, suggesting that TEC-1 is a disease-modifying drug with a potentially higher therapeutic window compared to the SMN-C series, including risdiplam. Furthermore, TEC-1 did not inhibit the action of nusinersen in a cell culture system, supporting the possibility that TEC-1 could be utilized concomitantly with this existing SMA drug. Identification of TEC-1 contributes not only to the development of a promising SMA therapeutic but also to the feasibility of RNA-targeting small-molecule drug discovery that ensures clinical tolerability.

Oligonucleotides for qRCR, RT-PCR. Oligo DNA standards, primers and probes were synthesized by
Integrated DNA Technologies Inc. (USA). Their sequences are also available in Supplementary Tables 1, 2  . Human lymphoblastoid-derived TK6 cells were purchased from American Type Culture Collection (USA) and cultured in the culture medium consisted of RPMI1640 supplemented with 10% heatinactivated horse serum, 2 mmol/L sodium pyruvate, 100 unit/mL penicillin and 100 µg/mL streptomycin. All above cells were kept in a 37 °C incubator supplied with 5% CO 2 .

Differentiation of iPSCs to motor neuron. GM24468 cells were differentiated to motor neurons (MNs)
via a motor neuron progenitor (MNP) as previously reported 26 Table 1). Cycle conditions were as follows; 95 °C for 1 min; followed by 40 cycles of denaturation at 95 °C for 15 s; annealing and elongation at 60 °C for 1 min in the ViiA7 RT-PCR system (Thermo Fisher Scientific). The copy numbers of each target were normalized to those of GAPDH. www.nature.com/scientificreports/ In the instances of using NRK-49F, MDCK, CV-1 or NCMDF, cell lysis and reverse transcription were carried out using Custom Cells to CT Lysis Components (Ambion) and Cells-to-CT RT Components (Ambion), each in accordance with manufacturer's recommendations. qPCR was conducted using TaqMan Fast Advanced Master Mix. The final concentrations of primers and probes were adjusted to 0.4 µM and 0.15 µM, respectively. Cycle conditions were as follows; 50 °C for 2 min and 95 °C for 2 s; followed by 40 cycles of denaturation at 95 °C for 1 s; annealing and elongation at 60 °C for 20 s in the ABI7900HT RT-PCR system (Applied Biosystems, USA). The relative quantification in gene expression was determined using the 2−ΔΔCt method.
Total RNA was extracted with RNeasy Mini kit and RNase-Free DNase Set (Qiagen, USA) from compoundtreated cells, followed by cDNA synthesis using High capacity cDNA revere transcription kit (Applied Biosystems) according to their protocols. PCR was done with PrimeStar GXL DNA polymerase (Takara). Briefly, cDNA of GALC coding full length sequence was amplified by PCR using the primer set, 5ʹ-GAG TCA TGT GAC CCA CAC AATG-3ʹ and 5ʹ-GAA TGT TAG GGA ACA CAC CAG GTA -3ʹ. Cycle conditions for GALC transcripts were used as follows; 98 °C for 1 min followed by 35 cycles of denaturation at 98 °C for 10 s; annealing at 65 °C for 10 s; elongation at 72 °C for 4 min, and a final incubation at 72 °C for 5 min. The products were analyzed with agarose gel electrophoresis. cDNA synthesis was carried out using SuperPrep Cell Lysis & RT Kit for qPCR (Toyobo, Japan). PCR amplification was carried out using Takara Ex Taq Hot Start (Takara). The final concentrations of primers was adjusted to 1 µM (Supplementary Table 2). Cycle condition was as follows; 98 °C for 1 min followed by 45 cycles of denaturation at 98 °C for 10 s; annealing at 65 °C for 10 s; elongation at 72 °C for 1 min. PCR products were separated with electrophoresis on a 3% agarose gel and visualized after ethidium bromide staining under UV light.

Analysis of cryptic GALC and HTT transcripts. Hs683 cells which were co-treated with 1 µM SMN-C3
and 200 µg/ml CHX for 6 h were used for analysis of GALC transcript, while Hs683 cells which co-treated with 0.3 µM risdiplam and 200 µg/ml CHX for 6 h were used for the analysis of HTT.
Total RNA was extracted with RNeasy Mini kit and RNase-Free DNase Set from compound-treated cells, followed by cDNA synthesis using High capacity cDNA revere transcription kit according to their protocols. PCR was done with PrimeStar GXL DNA polymerase. Briefly, cDNA containing from exon 5 to exon 7 of GALC coding sequence was amplified by PCR using the primer set, 5ʹ-GAG TCA TGT GAC CCA CAC AATG-3ʹ and 5ʹ-GAA TGT TAG GGA ACA CAC CAG GTA -3ʹ. Cycle conditions for GALC transcripts were used as follows; 98 °C for 1 min followed by 35 cycles of denaturation at 98 °C for 10 s; annealing at 65 °C for 10 s; elongation at 72 °C for 4 min, and a final incubation at 72 °C for 5 min. The products were analyzed with 1% agarose gel electrophoresis. The second PCR was done using the first PCR products as templates with the same conditions but for only 25 cycles. One of the amplified products, derived from the cells treated with the compound, was sequenced directly using a primer, 5ʹ-CAC AAT GGC TGA GTG GCT ACTC-3ʹ and Big Dye terminator v3.1 cycle sequencing kit. For analysis of the HTT transcripts, cDNA sequence from exon 42 to exon 55 was amplified by PCR using the primer set, 5ʹ-GAG GAT TCT GAC TTG GCA GCCA-3ʹ and 5ʹ-CAC AGG CAC AGT CAT TGC ACTGA-3ʹ with next condition; 98 °C for 1 min followed by 35 cycles of denaturation at 98 °C for 10 s; annealing at 65 °C for 10 s; elongation at 72 °C for 2 min, with a final incubation at 72 °C for 4 min. The second PCR was performed using the first PCR product as a template under the same conditions but only for 20 cycles. The amplified product was sequenced directly using a primer, 5ʹ-ATG AAT GCC TTC ATG ATG AAC TCG -3ʹ and the above sequencing kit. DNA sequences were determined using an ABI-PRISM 3100 automatic sequencer (Applied Biosystems).

SDS-PAGE and western blotting.
GM03813 fibroblasts and GM24468-derived MNs were seeded at 1.5 × 10 4 and 4 × 10 4 cells/well in 96-well plates and were treated with TEC-1 or SMN-C3 for 72 h at the concentration of 0.01, 0.03, 0.1, 0.3, 1, or 3 µM. After 72 h exposure of compound, GM24468 derived MNs and GM03813 fibroblasts were washed once with D-PBS (-), lysed with RIPA buffer (Wako) containing cOmplete (Sigma-Aldrich) and PhosSTOP (Sigma-Aldrich). The lysates were spun at 15,000 rpm for 15 min at 4 °C, and the supernatants were collected. The protein concentration was quantified with BCA kit (Wako). The lysates, adjusted to have the same protein amount, were mixed with NuPAGE LDS sample buffer containing DTT and heated at 95 °C for 5 min. Except for HTT detection, each sample (2 µg total protein/lane) was loaded Perfect NT Gel 7.5-15% gel (DRC, Japan) and run at 150 V for 60-70 min with Tris/glycine/SDS running buffer (BIO-RAD, USA). The proteins were transferred from the gel to Immobilon-P PVDF Membrane (Millipore, USA) with a transfer solution comprising 10% methanol and 1 × Novex Tris-glycine transfer buffer at 25 V for 180-240 min using Criterion blotter (BIO-RAD). In the case of HTT detection, each sample (2 µg total protein/lane) was loaded to Perfect NT Gel 7.5-15% gel and run at 150 V for 60-70 min with NuPAGE Tris-Acetate SDS Running Buffer. The proteins were transferred with a transfer solution comprising 10% methanol and 1 × Novex Trisglycine transfer buffer at 25 V for 960 min using Criterion blotter. The membranes were blocked with Pierce Scientific Reports | (2020) 10:17472 | https://doi.org/10.1038/s41598-020-74346-9 www.nature.com/scientificreports/ The population doubling (PD) and relative population doubling (RPD) at each concentration was calculated by the following formula.
For the treated groups, the 3 consecutive concentrations were selected as follows. For −S9 mix assay condition, cytotoxicity, as indicated by a RPD of less than 50% was noted at 2.3 μg/mL of TEC-1 and higher; for + S9 mix assay condition, cytotoxicity was noted at 2.9 μg/mL of TEC-1 and higher; for 24-h assay condition, precipitation was noted at 1.5 μg/mL of TEC-1 and higher. Based on these results 1.9, 2.1 and 2.3 μg/mL of TEC-1 were analyzed for −S9 mix assay condition; 2.5, 2.7, and 2.9 μg/mL of TEC-1 were analyzed for + S9 mix assay condition; 1.1, 1.3, 1.5, μg/mL of TEC-1 were analyzed for 24-h assay condition. For all concentrations, 4000 mononuclear cells with cytoplasm were analyzed and the number of cells with micronucleus (micronuclei) was counted using a microscope (× 600). Micronucleus is identified as a round or oval small nucleus in the cytoplasm with the same staining intensity as the main nucleus. The diameter of each micronucleus should be less than 1/2 of the diameter of the main nucleus. Two investigators observed the slides. Each investigator scored 1000 cells for the frequency of micronucleated cells. Data from the two investigators were combined for each culture and the combined data from duplicate cultures were pooled for each concentration for each treatment condition. A cell having one or more micronuclei was recorded as a micronucleated cell. The test article was judged to be positive if the incidence of micronucleated cells between any test article treatment groups satisfied both criteria (1) and (2) shown below.
(1) A significant difference in the incidence of micronucleated cells between the negative control group is detected in the statistical analysis and concentration-dependent increase was also detected in the statistical analysis. (2) The incidence of micronucleated cells is more than the historical control range (mean + 2SD).
Otherwise, the test article was judged to be negative.
Pharmacokinetic study in adult wild type FVB mice. TEC-1 was administered at doses of 2, 6, and 20 mg/kg for oral (suspended in 0.5% methyl cellulose aqueous solution) to male FVB mice (aged 8 weeks, 3 mice per a time point). After 0.5-120 h administration, blood and brain samples were collected. Blood samples with heparin were centrifuged for plasma collection. Brain samples were homogenized in saline 20% (w/v). The samples were deproteinized with acetonitrile and following by centrifugation, and the supernatants were analyzed by LC/MS/MS to obtain TEC-1 concentrations of the plasma and brain.
Pharmacokinetic study in neonatal FVB mice. TEC www.nature.com/scientificreports/ analyzed by Student's t-test (unpaired, two-tailed). Dose-responsibility test was analyzed by one-way analysis of variance followed by William's test or Dunnett's test. In micronucleus test, Fisher's exact test was performed in order to compare the incidence of micronucleated cells in the test article groups or the positive control group with that of the negative control groups for each treatment condition. When Fisher's exact test showed statistical significance, the exact Cochran-Armitage trend test was performed to evaluate any dose-response relationship. Statistical processing was performed using the Microsoft Excel or GraphPad Prism Software (Prism 6 or 8), EXSUS statistical software package (CAC Croit Corporation, Japan).

Data availability
The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.