High Concentration or Combined Treatment of Antisense Oligonucleotides for Spinal Muscular Atrophy Perturbed SMN2 Splicing in Patient Fibroblasts

Spinal muscular atrophy (SMA) is caused by survival motor neuron 1 SMN1 deletion. The survival motor neuron 2 (SMN2) encodes the same protein as SMN1 does, but it has a splicing defect of exon 7. Some antisense oligonucleotides (ASOs) have been proven to correct this defect. One of these, nusinersen, is effective in SMA-affected infants, but not as much so in advanced-stage patients. Furthermore, the current regimen may exhibit a ceiling effect. To overcome these problems, high-dose ASOs or combined ASOs have been explored. Here, using SMA fibroblasts, we examined the effects of high-concentration ASOs and of combining two ASOs. Three ASOs were examined: one targeting intronic splicing suppressor site N1 (ISS-N1) in intron 7, and two others targeting the 3′ splice site and 5′ region of exon 8. In our experiments on all ASO types, a low or intermediate concentration (50 or 100 nM) showed better splicing efficiency than a high concentration (200 nM). In addition, a high concentration of each ASO created a cryptic exon in exon 6. When a mixture of two different ASOs (100 nM each) was added to the cells, the cryptic exon was included in the mRNA. In conclusion, ASOs at a high concentration or used in combination may show less splicing correction and cryptic exon creation.


Introduction
Spinal muscular atrophy (SMA) is a common motor neuron disease that is inherited in an autosomal recessive manner [1]. It has a high incidence, affecting 1 in 11,000 newborns [1].
SMA is clinically divided into five groups [1]: type 0 (the most severe form with onset in the prenatal period; severe respiratory problems after birth), type 1 (a severe form with onset before 6 months of age; unable to sit unsupported), type 2 (an intermediate form with onset before 18 months of age; able to sit unaided, but unable to stand or walk), type 3 (a mild form with onset after 18 months of age; able to stand and walk unaided), and type 4 Genes 2022, 13, 685 2 of 13 (the mildest form with age of onset from adolescence to adulthood; able to stand and walk unaided). Almost all patients with SMA type 0 die within a few weeks after birth, and most patients with SMA type 1 die or become dependent on a ventilator before reaching 2 years of age [1].
SMA is caused by defects of the survival motor neuron 1 (SMN1) gene located on chromosome 5q13 [2]. More than 90% of SMA patients are homozygous for SMN1 deletion [2]. Defects of SMN1 reduce the level of survival motor neuron (SMN) protein, leading to motor neuron dysfunction [3].
A gene homologous to SMN1 is present at the same locus of chromosome 5q13 [2], which is named the survival motor neuron 2 (SMN2) gene. SMN2 encodes the same SMN protein as SMN1 produces. A previous report [2] described that complete loss of SMN2 was not observed in any SMA patients with homozygous SMN1 deletion, suggesting that its complete loss results in embryonic lethality [4,5]. Thus, SMA patients with homozygous deletion of SMN1 have at least one copy of SMN2. However, SMN2 cannot fully compensate for homozygous SMN1 deletion because SMN2 produces only a small amount of full-length SMN protein due to the splicing defect of exon 7 [6].
These drugs are completely different in terms of their dosage form and route of administration. Nusinersen is an ASO drug that is intrathecally administered every few months [19]. It targets an intronic splicing suppressor site (ISS-N1) in SMN2 intron 7, corrects the splicing defect of SMN2 exon 7, and leads to the production of full-length SMN protein [15,16]. Onasemnogene abeparvovec is an adeno-associated virus vector drug carrying SMN complementary DNA encoding the missing SMN protein, which is intravenously administered only once [20]. Meanwhile, risdiplam is a small-molecule drug that corrects the splicing defect of SMN2 exon 7 and leads to production of the full-length SMN protein. It is orally administered every day [21][22][23].
Currently, more than 10,000 patients worldwide are being treated with nusinersen [24]. Nusinersen is effective for improving the motor function of newborns and infants with SMA [25,26], but its efficacy is limited in advanced-stage patients [27]. In addition, a ceiling effect of the current dosage regimen regarding the improvement of motor function has also been reported [28]. To overcome these problems, the effects of nusinersen at a high dose or in combination with different drugs should be explored.
Concerning treatment with a high dose of nusinersen, a clinical trial of such treatment, the DEVOTE study, is now ongoing (ClinicalTrials.gov Identifier: NCT04089566). However, based on studies using SMA cells, Ottesen et al. suggested that a high concentration of nusinersen might have off-target effects on the expression of certain genes [29].
Concerning treatments involving nusinersen in combination with other drugs, Pao et al., reported that dual targeting of both the ISS-N1 site in SMN2 intron 7 and the 3 splice site/5 region of SMN2 exon 8 by two ASOs increased the level of full-length SMN transcription more effectively than a single ASO targeting the ISS-N1 site [30]. However, such dual targeting therapy with two ASOs has not been pursued further. Recently, Harada et al., reported five infants who received nusinersen and onasemnogene abeparvovec [31]. In their study, no adverse effects were noted in the patients, and all patients exhibited improvements in their motor function. Nonetheless, the prolonged use of corticosteroid and careful monitoring of liver toxicity are necessary when onasemnogene abeparvovec therapy is combined with nusinersen therapy [31].
In this study, we examined the effect of a high dose of ASOs and the effect of combining two ASOs. For this purpose, SMA cells were transfected with three ASOs: one targeting the ISS-N1 site in SMN2 intron 7, and two others targeting the 3 splice site or the 5 region of SMN2 exon 8. Our study clarified that ASOs at a high concentration or in combination may result in adverse effects including reduced correction of the splicing of SMN2 exon 7 and the activation of cryptic splice sites in SMN2 intron 6.

Cell Culture and Transfection Protocol
GM00232 SMA type 1 patient fibroblasts were obtained from Coriell Cell Repositories (Camden, NJ, USA). According to the provider's information, donor subject had 2 copies of the SMN2 gene and is homozygous for deletion of exons 7 and 8 of the SMN1 gene. The cells were cultured in Eagle's minimum essential medium containing 1% antibiotic and antimycotic solution (consisting of 10,000 U/ml penicillin, 10,000 µg/mL streptomycin, 25 µg/mL amphotericin B) (Nacalai Tesque, Inc., Kyoto, Japan) and 15% fetal bovine serum (Sigma Aldrich, St. Louis, MO, USA), and incubated at 37 • C in a 5% CO 2 humidified atmosphere prior to transfection. We transfected 1 × 10 6 SMA type 1 fibroblasts in a transfection mixture with a final volume of 10 µL containing ASO and transfection buffer using the Neon transfection system (Invitrogen, Carlsbad, CA, USA). The cells were pulsed twice with a voltage of 1400 V and a width of 20 ms (with each pulse lasting 20 ms). Then, the cells were quickly transferred into 1 mL of medium and incubated for 48 h before harvesting. All plates were pre-coated with StemSure®0.1% w/v Gelatin Solution (WAKO, Osaka, Japan), in accordance with the manufacturer's protocol.
Our study using the above-mentioned ASOs consisted of two parts. First, to explore the dose effect of ASOs, each ASO was transfected into SMA type 1 fibroblasts at three concentrations (50, 100, and 200 nM), followed by analysis of SMN2 transcript expression after 48 h of incubation at 37 • C. Second, to explore the effect of combining ASOs, combinations of two ASOs, each at a concentration of 100 nM (total 200 nM), were also transfected into SMA type 1 fibroblasts, after which SMN2 transcript expression was analyzed following 48 h of incubation at 37 • C. The combinations were ASO-NUS/ASO-EX8, ASO-NUS/ASO-SSJ, and ASO-EX8/ASO-SSJ.

RNA Extraction and cDNA Synthesis
Forty-eight hours after ASO transfection, total RNA was isolated using Sepasol-RNA I reagent (Nacalai Tesque), in accordance with the manufacturer's protocol. Total RNA was then treated with DNaseI Amplification Grade (Invitrogen, Carlsbad, CA, USA), in accordance with the manufacturer's protocol. Freshly prepared RNA was immediately used for cDNA synthesis to avoid degradation because the intron-retaining transcript was unstable.
An aliquot of cDNA, equivalent to 100 ng of total RNA, was subjected to RT-PCR. The RT-PCR was performed in a reaction mixture with a total volume of 30 µL containing 1× PCR buffer, 2 mM MgCl 2 , 0.2 mM each dNTP, 0.3 µM each primer, and 1.0 U Fast Start Taq DNA Polymerase (Roche Applied Science, Mannheim, Germany).
To determine the transcript levels, we performed semi-quantitative RT-PCR analysis followed by gel electrophoresis. The RT-PCR for the SMN transcript was performed as follows: initial denaturation at 94 • C for 7 min; 35 cycles of denaturation at 94 • C for 1 min, annealing at 56 • C for 1 min, and elongation at 72 • C for 1 min; and then final elongation at 72 • C for 7 min. Meanwhile, the RT-PCR for the GAPDH transcript was performed as follows: initial denaturation at 94 • C for 7 min; 25 cycles of denaturation at 94 • C for 1 min, annealing at 60 • C for 1 min, and elongation at 72 • C for 1 min; and final elongation at 72 • C for 7 min.
All amplicons were electrophoresed on 4% agarose gel and visualized by Midori Green staining (Nippon Genetics, Tokyo, Japan). The sizes of the RT-PCR products are shown in Figure 1B.

Subcloning and Sequence Analysis
The PCR product of SMN transcript was purified and cloned into a TA cloning vector, pGEM-T vector (Promega, Madison, WI, USA), before being transformed into competent Escherichia coli (Invitrogen, Carlsbad, CA) and propagated in LB solution. The transformed E. coli was grown on an LB plate containing IPTG and X-gal for blue/white colony selection.
The white colonies were selected and subjected to PCR using T7-SP6 primer set and were run on an agarose gel to select the target colonies based on the amplified product size. The PCR product corresponding to cryptic-exon (exon 7a)-containing transcripts was then purified using Nucleospin kit, in accordance with the manufacturer's protocol (Takara Bio, Tokyo, Japan), and subjected to direct sequencing analysis. This sequencing analysis was outsourced to Fasmac Co., Ltd.

Statistics
To compare the splicing efficiencies among the ASOs, the software ImageJ (Version 2.1.0; National Institutes of Health, Bethesda, MD, USA) was used. All assays were carried out in triplicate and statistical analyses were performed using Microsoft Excel with the add-in software Statcel 3 (The Publisher OMS Ltd., Tokyo, Japan). Results reported as mean ± SD were analyzed by ANOVA with Tukey-Kramer post hoc test for comparisons between groups. * p < 0.01 was considered statistically significant.
To analyze the effect of the dose of ASOs on the level of transcripts containing exon 7, each ASO was transfected into the SMA type 1 fibroblasts at three concentrations (50, 100, and 200 nM), followed by analysis of SMN2 transcript expression ( Figure 2A). Here, distilled water without ASOs was used as a reference. The intensity of each band was determined to evaluate the levels of transcript isoforms (Supplementary Tables S1-S3).

ASO-NUS
Exon 7 inclusion: All concentrations of ASO-NUS used in this study increased the level of exon-7-containing transcripts (Ex6/Ex7/Ex8) while decreasing the level of exon-7lacking transcripts (Ex6/Ex8) (Figure 2A). This means that all ASO-NUS concentrations corrected the SMN2 exon 7 splicing in SMA fibroblasts. Among the three concentrations, 100 nM ASO-NUS corrected the splicing most effectively (Ex6/Ex7/Ex8 in Figure 2A Figure 2A). Nucleotide sequencing analysis demonstrated that this transcript included a sequence of exon 7a between exons 6 and 7, which we previously reported as a cryptic exon in intron 6 [32] (Figure 3). The cryptic-exon-containing transcript was observed only in the cells treated with 200 nM ASO-NUS, but not in the cells treated with 50 and 100 nM ASO-NUS ( Figure 2C).   Figure 2A).
Intron 7 retention: ASO-EX8 at 200 nM produced a significant amount of intron-7retaining transcript (a clear band of Ex6/Ex7/In7/Ex8 in Figure 2A), but at concentrations of 50 and 100 nM it produced only a trace amount of this transcript (only faint bands of Ex6/Ex7/In7/Ex8 in Figure 2A).

Effects of ASOs in Combination
In this experiment, combinations of two ASOs, each at a concentration of 100 nM (total 200 nM), were transfected into SMA type 1 fibroblasts, which were then analyzed for SMN2 transcript expression after 48 h of incubation at 37 • C. The ASO combinations applied were ASO-NUS/ASO-EX8, ASO-NUS/ASO-SSJ, and ASO-EX8/ASO-SSJ. Single ASOs (100 nM) and distilled water without ASOs were used as references. The intensity of each band was determined to evaluate the levels of transcript isoforms (Supplementary Tables S4-S6).
Cells transfected with 100 nM ASO-NUS had the highest RFD among all groups (p < 0.01), which was consistent with the results of the dose effect analysis. The mean RFD values from ASOs applied in combination arranged in ascending order were as follows: 0.25 with ASO-Ex8/ASO-SSJ (100 nM each), 1.29 with ASO-NUS/ASO-SSJ (100 nM each), and 2.07 with ASO-NS/ASO-EX8 (100 nM each) (p < 0.01) ( Figure 4B and Supplementary Table S6).

ASO-NUS/ASO-EX8
Exon 7 inclusion: The combination of ASO-NUS/ASO-EX8 (100 nM each) corrected the SMN2 exon 7 splicing to some degree ( Figure 4B) but showed significantly lower RFD values than the use of ASO-NUS alone (Supplementary Table S6). The presence of ASO-EX8 somewhat canceled out the effect of ASO-NUS ( Figure 4B, Supplementary Table S6).
Cryptic exon creation: It should be noted that the above combination of ASO-NUS/ASO-EX8 (100 nM each) was associated with high production of exon-7a-containing transcript (Ex6/Ex7a/Ex7/Ex8), as shown in Figure 4A,C. However, in the additional experiments with ASO-NUS/ASO-EX8, the cryptic exon was not activated when the total of combined dose was 100 nM, regardless of combination (or mixing) ratios (Supplementary Figure S1). Thus, the excess of total amount of ASOs might be a factor to produce the cryptic-exoncontaining transcript.
Cryptic exon creation: It should be noted that the above combination of ASO-NUS/ASO-EX8 (100 nM each) was associated with high production of exon-7a-containing transcript (Ex6/Ex7a/Ex7/Ex8), as shown in Figure 4A,C. However, in the additional experiments with ASO-NUS/ASO-EX8, the cryptic exon was not activated when the total of combined dose was 100 nM, regardless of combination (or mixing) ratios (Supplementary Figure S1). Thus, the excess of total amount of ASOs might be a factor to produce the cryptic-exon-containing transcript. Intron 7 retention: ASO-NUS/ASO-EX8 (100 nM each) produced only a trace amount of intron-7-retaining transcript (a faint band of Ex6/Ex7/In7/Ex8 in Figure 4A).

ASO-NUS/ASO-SSJ
Exon 7 inclusion: The combinations of ASO-NUS/ASO-SSJ (100 nM each) corrected the SMN2 exon 7 splicing to some degree ( Figure 4B) but were associated with significantly lower RFD values than the use of ASO-NUS alone (Supplementary Table S6). The presence of ASO-SSJ, as well as ASO-EX8, canceled out the effect of ASO-NUS.  Table S6). The presence of ASO-SSJ, as well as ASO-EX8, canceled out the effect of ASO-NUS.
Cryptic exon creation: Notably, the combination of ASO-NUS/ASO-SSJ (100 nM each) produced a high level of the exon-7a-containing transcript (Ex6/Ex7a/Ex7/Ex8), as shown in Figure 4A,C. The cryptic exon activation was not clearly observed when the total dose of combined ASO-NUS/ASO-SSJ was 100 nM, regardless of mixing ratio (Supplementary Figure S1). This observation was similar with the ASO-NUS/ASO-EX8 combination when the total dose was 100 nM. These observations suggested that the total amount of ASOs might be more critical than combination ratios. Intron 7 retention: The combination of ASO-NUS/ASO-SSJ (100 nM each) produced only a trace amount of intron-7-retaining transcript (a faint band of Ex6/Ex7/In7/Ex8 in Figure 4A).
Cryptic exon creation: Moreover, the combination of ASO-EX8/ASO-SSJ (100 nM each) led to almost no production of the exon-7a-containing transcript ( Figure 4A,C). Taken together with the results of other combinations, the activation of cryptic splice sites in intron 6 may be closely related to the presence of ASO-NUS and/or the total amount of ASOs. Intron 7 retention: However, the combination of ASO-EX8/ASO-SSJ (100 nM each) produced a significant amount of the intron-7-retaining transcript (Ex6/Ex7/In7/Ex8 in Figure 4A).

ASO Targeting ISS-N1 Site
Many ASOs for treating SMA have been studied, such as ASOs inhibiting alternative 3 splice site pairing of SMN2 exon 8 [9], ASOs targeting an intronic splicing suppressor site in SMN2 intron 6 [10], peptide nucleic acid (PNA) with an arginine-serine (RS) domain that is a site for binding to exon 7 (known as the ESSENCE method) [11], ASOs containing a sequence complementary to exon 7 and a sequence non-complementary to some ESE motifs (known as the TOES method) [12,13], trans-splicing RNA carrying the exon 7 sequence [14], ASOs targeting the ISS-N1 site [15,16], and ASOs targeting the 5 region of exon 8 [17].
As mentioned in the Introduction section, an ASO targeting ISS-N1, nusinersen, was approved as the first drug for SMA by the FDA. The ISS-N1 site was discovered by Singh et al. [36]. They were also the first to describe that deletion of this site restored inclusion of SMN2 exon 7 [15]. In addition, in 2008 Hua et al., demonstrated that heterogeneous nuclear ribonucleoprotein (hnRNP) A1 and hnRNP A2 bind to the ISS-N1 site and that an ASO masking the motifs in the ISS-N1 site fully restored inclusion of SMN2 exon 7 [16]. Subsequently, Ionis Pharmaceuticals (formerly ISIS Pharmaceuticals) began the clinical development of nusinersen, an antisense drug targeting the ISS-N1 site [37]. Nusinersen showed very promising results at all stages of clinical development and was approved by the FDA in 2016 [37].
Our study confirmed the findings in the previous studies reported by Singh et al., and Hua et al. [15,16]. According to our results, 100 nM ASO-NUS almost fully restored the inclusion of SMN2 exon 7. However, 200 nM ASO-NUS restored this inclusion with lower efficiency than 100 nM ASO-NUS.
In addition, 200 nM ASO-NUS produced a cryptic exon (exon 7a) transcript in our experiment. In line with this, Ottesen et al. reported that a high concentration of an ASO targeting ISS-N1 caused massive perturbation of the transcriptome, including activation of a cryptic splice site in intron 6 [29]. Remarkably, combinations of ASOs, ASO-NUS/ASO-EX8 and ASO-NUS/ASO-SSJ, as well as a high dose of ASO-NUS, also activated the cryptic splice sites in intron 6 ( Figure 4A). We discuss below the activation of the cryptic exon in SMN2 intron 6.

ASO Targeting SMN2 Exon 8
In 2001, Lim et al. reported that an ASO directed toward the 3 splice site of SMN2 exon 8 altered SMN2 splicing in favor of exon 7 inclusion [9]. This was surprising, as it would be expected that disruption of 3 splice site function of exon 8 may cause intron 7 to be retained. Actually, they indeed showed that the level of intron 7 retention was slightly increased. However, the inclusion of exon 7 into the mRNA was clearly improved.
In 2018, Flynn et al. tested the possibility of developing a strategy to retain SMN2 intron 7 [17]. Because the authentic stop codon is located within exon 7, the intron 7retaining transcript sequence should encode the full-length SMN protein. Flynn et al., used ASOs targeting not only the 3 splice site but also the ESE motifs (e.g., hTra2beta-1 binding motif, SF2/ASF binding motif, SRp 40 binding motif) in exon 8. According to them, the ASOs increased the level of intron 7-retaining SMN2 transcript, but this was not linked to an increase in the level of SMN protein because of the function of the longer 3'UTR in the intron 7-retaining transcript [17].
The results of Lim et al. and Flynn et al. were re-tested in our study [9,17]. According to our results, ASO-SSJ and ASO-EX8 slightly increased the level of transcripts with intron 7 retention (Ex6/Ex7/In7/Ex8 in Figures 2A and 4A) and further enhanced the level of transcripts with exon 7 inclusion (Ex6/Ex7/Ex8 in Figures 2A and 4A). ASOs targeting the 3 splice site/5 region of exon 8 may improve exon 7 inclusion, suggesting the existence of novel mechanisms potentially associated with unexpected exon 7 splicing.
Pao et al. previously reported a synergistic effect of two ASOs [30]. According to them, the efficiency of exon 7 inclusion by two ASOs masking the ISS-N1 site and the 3' splice site of exon 8 was much higher than that by a single ASO masking the ISS-N1 site [30]. However, our data did not show such a synergistic effect of a combination of ASOs. In our study, the combination of ASO-EX8/ASO-SSJ (100 nM each) was not associated with exon 7 inclusion, while the combination of ASO-NUS/ASO-EX8 (100 nM each) or ASO-NUS/ASO-SSJ (100 nM each) restored exon 7 inclusion to some degree, albeit with limited efficiency. The efficiency of exon 7 inclusion by a single ASO masking the ISS-N1 site (ASO-NUS) was much higher than that of two ASOs masking the ISS-N1 site and the 3 splice site/5 region of exon 8 (ASO-SSJ or ASO-EX8). Our results suggested that ASO-EX8 and ASO-SSJ may inhibit the ability of ASO-NUS to include exon 7.

Activation of Cryptic Splice Sites in SMN2 Intron 6 by ASOs
Our study demonstrated that a high dose (200 nM) of ASO-NUS, ASO-EX8, and ASO-SSJ activated cryptic splice sites in SMN2 intron 6, leading to a cryptic exon, exon 7a. Ottesen et al., previously reported that a high concentration of ASOs targeting ISS-N1 activated cryptic splice sites in SMN2 intron 6, leading to a cryptic exon, and considered it an off-target effect of ASOs targeting ISS-N1 [29].
SMN exon 7a is a cryptic exon that has been described independently by our group and Singh's group (according to their nomenclature, it is exon 6B) [32,38]. The cryptic splice sites in intron 6 can be activated in both SMN genes, SMN1 and SMN2. Exon 7a harbors two premature stop codons and gene products containing it may be subject to nonsense-mediated decay (NMD) [32]. If an exon-7a-containing transcript were translated into a truncated protein by a particular regulatory system, the product could have some functional activity and intermediate stability [38]. Even so, the inclusion of exon 7a in SMN2 mRNA should be considered undesirable for the treatment of SMA with ASOs.
In this study, we also observed that the combination of ASOs targeting ISS-N1 and the 3 splice site of SMN2 exon 8 activated the cryptic splice sites in intron 6, even though the concentration of each ASO was only 100 nM. A high concentration of single ASOs or the simultaneous use of two ASOs may alter the dynamic secondary structure of SMN2 pre-mRNA, activating cryptic splice sites in intron 6. In other words, exon 7 splicing mechanisms are so delicate that some alteration of the dynamic secondary structure of pre-mRNA can easily affect the splicing pattern [39].

Research Limitations
We did not perform any protein and functional analyses of SMN in this study because we focused on the alternative splicing patterns caused by the administration of high dose of ASOs and combination of ASOs. However, we expected that the increase of full-length SMN2 transcript led to the increase of functional full-length SMN protein in the cells and nuclear gems based on the report of Hua et al. [40], and the cryptic-exon-containing transcript could produce a truncated protein with some degree of functional activity and intermediate stability based on the report of Seo et al. [38].

Conclusions
Nusinersen, an ASO targeting the ISS-N1 site, is effective for improving the motor function of infants with SMA, but not so effective in advanced-stage patients. In addition, a ceiling effect of the current regimen has also been reported. To overcome these problems, studies have explored the effects of a high dose of nusinersen or the synergistic effects of combinations of different drugs. Based on our study, the RFD value with 100 nM ASO-NUS (an ASO targeting ISS-N1) was significantly higher than that with 50 nM ASO-NUS, suggesting a beneficial dose-dependent effect on exon 7 inclusion. However, 200 nM ASO-NUS showed significantly lower RFD values than 100 nM ASO-NUS and activated cryptic splicing sites. Above a certain threshold, the higher the concentration of ASO-NUS, the less effective and more harmful it may be. It should be noted that cryptic exon activation is linked to the decrease of full-length SMN2 transcript, leading to the reduction of functional full-length SMN protein. In addition, combinations of ASO-NUS and other ASOs produced similar results to a high concentration of ASO-NUS.
In conclusion, although it is difficult to directly apply the results of cultured cell studies to the clinical setting, our data suggested that a high concentration of ASOs or their use in combination may show unanticipated effects, including a lower rate of splicing correction and the creation of a cryptic exon.