Potential ASO-based personalized treatment for Charcot-Marie-Tooth disease type 2S

Abstract

Immunoglobulin mu-binding protein 2 (IGHMBP2) pathogenic variants lead to a spectrum of disorders characterized by alpha-motor neuron degeneration. We describe a compound heterozygous patient diagnosed with Charcot-Marie-Tooth disease type 2S with variants in IGHMBP2: a pathogenic missense variant acting in trans with a confirmed intronic cryptic splice site variant. This variant was shown to result in the creation of a new splice acceptor site, loss of reading frame and nonsense-mediated decay. We designed a 19-mer antisense oligonucleotide targeting this cryptic intronic variant to restore IGHMBP2 levels. ASO treatment of patient fibroblasts significantly increased the ratio of restored wild-type transcript to cryptic exon-containing transcript and resulted in over a 50% increase in IGHMBP2 protein levels. Neuromuscular junction analyses revealed high fatigue and chaotic tetanus formulation in untreated patient cells. We demonstrate rescue of NMJ function following ASO treatment, captured by a reduction in fatigue and chaotic tetanus responses. Furthermore, toxicity testing revealed that intrathecal administration of the ASO to wild-type Sprague-Dawley rats over 3 months was well tolerated. Our preclinical data support this ASO as a potential CMT2S treatment by rescuing IGHMBP2. N-of-1 ASO-based therapeutics may prove instrumental in the design of treatments for this diverse genetic disorder.

Graphical abstract

Smieszek and colleagues designed an antisense oligonucleotide (ASO) therapeutic targeting a cryptic splice site variant found in the immunoglobulin mu-binding protein 2 (IGHMBP2) gene of an identified Charcot-Marie-Tooth disease type 2S (CMT2S) patient. The authors provide preclinical data showing successful efficacy and safety of the designed ASO treatment for CMT2S.

Introduction

Charcot-Marie-Tooth (CMT) disease (OMIM: 616155) encompasses a heterogeneous group of genetic disorders of the peripheral nervous system, in which motor and/or sensory peripheral nerves are affected.^1,2 These disorders exhibit both high genetic and allelic heterogeneity, posing challenges for both diagnosis and treatment. The severity of these disorders is associated with the nature of the variants, ranging from severe motor neuropathies caused by strong loss of function variants, to axonal disorders such as Charcot-Marie-Tooth disease type 2S (CMT2S) associated with less disruptive variants. Clinical manifestations across the various types of CMT include common features of atrophy and weakness, primarily impacting the distal muscles.^2,3,4 These symptoms are often accompanied by areflexia and variable degrees of sensory involvement.^2,3,4 Over time, muscle weakness and atrophy spread proximally. The prevalence of CMT is estimated to be 1 in 2,500 individuals, with clinical presentations influenced by the diverse genetic variants associated with CMT.^2,3,5

Immunoglobulin mu-binding protein 2 (IGHMBP2) variants can cause a rare autosomal recessive form of axonal CMT, CMT2S, as well as a rare form of spinal muscular atrophy (SMA) called spinal muscular atrophy with respiratory distress type I (SMARD1).^2,3,4,6 IGHMBP2 (OMIM: 600502) is thought to bind directly to DNA and RNA in order to regulate pre-mRNA processing and transcription.² It has been recently shown that IGHMBP2 aids in the maintenance of cellular transcriptome homeostasis and translation of transcripts, including transcription-export (TREX) components THOC1 and THOC2.⁷ The loss of IGHMBP2 results in reduced translation efficiency, downregulating mRNA metabolism.⁷ IGHMBP2 is expressed at high levels in neurons, yet despite its nearly ubiquitous expression, the IGHMBP2 gene product predominately affects alpha-motor neurons.⁸ Consequently, abnormal mRNA metabolism and RNA processing may lead to alpha-motor neuron degeneration and disease manifestation. IGHMBP2 pathogenic variants often cause SMARD1 when they lead to complete protein function loss, while CMT2S is typically associated with variants that lead to partial protein loss. Through examination of levels of IGHMBP2 in fibroblast and lymphoblastoid cell lines, single heterozygous carriers of IGHMBP2-associated CMT2 variants were found to have intermediate IGHMBP2 protein levels between affected and healthy control individuals.² While SMARD1 typically results in a loss of motor neurons (MNs) in the spinal cord accompanied by rapid respiratory distress type 1, CMT2S often manifests as a severe neuropathy without critical breathing problems.^9,10

CMT2S is an axonal sensorimotor neuropathy distinguished by the early onset of progressing distal sensory and motor impairment.³ The severity of the neuropathy is variable, with some experiencing severe scoliosis in addition to limb weakness.¹¹ Prevalence rate estimates from epidemiological studies are variable as the symptoms and disease forms within CMT vary widely, but it is estimated that the prevalence of CMT2S is less than 1 in 1,000,000 worldwide.¹² Management of CMT2S disease primarily revolves around minimizing or delaying symptoms to improve the patient’s quality of life due to the absence of a definitive cure.

We describe an individual CMT2S patient with compound heterozygous variants in IGHMBP2, first discussed by Cassini et al., 2019.⁴ The patient’s development was normal until around 3 months of age when she displayed inversion of her feet when held upright. The patient underwent genetic testing at age 5, where her doctors considered the diagnosis of SMA type 3 or a variant of CMT, but SMA testing was negative. As she grew older, she developed increasing proximal weakness and was unable to stand from a sitting position or climb stairs. At age 9, the patient underwent whole exome sequencing (WES), which revealed a single maternally inherited IGHMBP2 (NM_0021180.2) variant of unknown significance (c.1730T>C; p.Leu577Pro) in coding exon 12. The patient was referred to the Undiagnosed Diseases Network (UDN) at age 11 for further evaluation. Cassini et al. evaluated her genome and identified a paternally inherited cryptic splice site within IGHMBP2. This variant activates a cryptic splice site perturbing IGHMBP2 splicing and was shown to induce nonsense-mediated decay (NMD).⁴ Whole genome sequencing (WGS) through the UDN revealed this paternally inherited noncoding variant (c.1235 + 894C>A) deep in intron 8, predicted to activate a cryptic acceptor site.

Antisense oligonucleotides (ASOs) are small-sized, single-stranded oligonucleotides that can modulate gene expression by interacting with target RNA sequences through Watson-Crick hydrogen bonding.¹³ Delivery of ASOs to the central nervous system (CNS) has been previously successful in several ASO programs. Nusinersen is the first US Food and Drug Administration (FDA)-approved drug to treat SMA, a rare neuromuscular disorder that results in progressive muscular weakness and atrophy. Nusinersen is an ASO designed to increase survival motor neuron (SMN) protein expression by binding to an intron-splicing silencer region of the SMN2 pre-messenger RNA to promote exon inclusion and the production of full-length SMN protein.¹⁴ Several other ASOs in development have successfully penetrated the CNS following intrathecal administration, allowing for the treatment of neurodegenerative and neuromuscular disorders.¹³

In this study, we investigated the efficacy of a novel ASO, VCA-894A, designed for splice modification and cryptic exon skipping to restore the reading frame and ultimately protein function of IGHMBP2, by targeting this patient-specific cryptic splice site variant. This patient exhibits a partial loss of protein expression, attributed to mutation-induced NMD stemming from the paternal variant, coupled with partially functional protein expression from the maternal allele. Given that the variant on the paternal allele triggers the incorporation of a new cryptic exon leading to transcript decay, we hypothesize that ASO-mediated exon skipping can restore IGHMBP2 expression. Consequently, this intervention should yield normal protein expression from the affected allele, increasing IGHMBP2 protein levels. This innovative approach holds promise for restoring function, particularly in light of the asymptomatic presentation in the parent with the other allele. To develop a model of clinically relevant phenotypes of this CMT2S patient, patient cells were reprogrammed into induced pluripotent stem cells (iPSCs) and differentiated into MNs for analysis of the morphology and physiology of the diseased MNs compared with healthy MNs.¹⁵ Their subsequent neuromuscular junction (NMJ) differences were analyzed using a patient-derived “human-on-a-chip” NMJ system created by Hesperos Inc.¹⁶ The derived MNs were then incubated with VCA-894A, and treatment effects were quantified.

Results

Patient description

At 11 years old, the patient had difficulty with ambulation and recurrent falls but was able to walk over short distances.⁴ She has never been able to run or jump. She has difficulty with fine motor tasks due to weakness but displayed no bulbar or sensory symptoms.⁴ Neurologic examination showed normal mental status and cranial nerve function but revealed symmetric diffuse weakness, which was more prominent distally compared to proximally, with atrophy of her distal extremities.⁴

Currently, at the age of 19, the patient can stand with support and walk a few steps with assistance. The patient and family attribute some of her motor function loss to her growth without compensatory increases in strength. She has scoliosis, but because the curves in her spine are symmetric, it has not had an impact on her posture, seating, or breathing. Now that her bones are fully developed, she will not require surgery. While the patient has significant weakness in her arms and hands, she can care for herself, drive, and has graduated high school. Her clinical presentation is consistent with other reported CMT2S patients who have a similarly broad range of weaknesses.

IGHMBP2 cryptic splice site variant present in the patient

To confirm the presence of the patient-specific IGHMBP2 variant, we obtained the patient’s fibroblast cell line and completed WGS analysis. The patient’s IGHMBP2 variants are depicted in Figure 1. We confirmed the previously identified IGHMBP2 cryptic splice site variant in the patient’s DNA, noncoding variant c.1235 + 894C>A, deep in intron 8. We did not detect any other pathogenic variants in this gene. Paternal allele A carries the cryptic splice site variant responsible for the decrease of IGHMBP2 protein level mediated via transcriptional NMD. Additionally, maternal allele A carries a coding single nucleotide variant (SNV) (IGHMBP2; p.Leu577Pro) rs1483165002, as seen in Figures 1 and S1 depicts a pedigree showing the origin of the cryptic splice site variant and its effects on the expression of IGHMBP2. The consequence of this variant is incorporation of a pseudoexon and loss of open reading frame, leading to NMD. This is confirmed by cDNA amplification of cycloheximide-treated patient cells as shown in Figures S2A–S2D. The intronic retention is 182 base pairs (bp). We additionally tested ratios of in-frame vs. out-of-frame PCR products upon treatment with an ASO vs. untreated control. An increased ratio of in-frame (wild-type [WT]) products was observed (Figures S2E and S2F).

Figure 1.

Origin of IGHMBP2 variants

The top panel depicts the origin of the IGHMBP2 variants, including the cryptic splice site variant.⁴ Paternal allele A carries the cryptic splice site variant responsible for the decrease of IGHMBP2 protein level mediated via transcriptional NMD. Maternal allele A carries a coding SNV (IGHMBP2; p.Leu577Pro) rs1483165002. The lower panel shows the section of the predicted 3D protein structure of IGHMBP2, comparing WT with the carrier of the missense variant rs1483165002, where the position of the amino acid substitution is colored in gray. Specifically observed is a loss of two hydrogen bonds and a loss of connection to the chains.

To confirm reduced levels of IGHMBP2 expression, fibroblasts from the CMT2S patient were compared with fibroblasts from a control non-CMT2S patient. CMT2S patient fibroblasts revealed significantly reduced IGHMBP2 expression and protein levels compared with the control patient fibroblasts. The cryptic exon causes immediate NMD and reduction of IGHMBP2 transcript by exactly 50% (Figure S3).

Efficient cellular entry of ASO and cell viability

The proposed ASO intervention with VCA-894A is cryptic exon skipping to restore the reading frame and subsequently IGHMBP2 protein function, as depicted in Figure 2. Proof-of-concept experiments focused on protein rescue and restoration of the decreased IGHMBP2 levels, which are diminished by the paternally inherited intronic splicing variant. Multiple ASOs were designed to target the region harboring the variant located in intron 8 (c.1235 + 894C>A). ASOs were chemically modified with a phosphorothioate and 2′-methoxyethyl (MOE) backbone. The patient’s fibroblast cells were obtained, as fibroblasts are an ideal cell population for ASO therapeutics.¹⁷ Currently, fibroblast cells constitute one of the more frequently used models for ASO therapeutics. They express moderate levels of the gene of interest, IGHMBP2.¹⁸ The patient’s fibroblast cells with the confirmed cryptic splice site variant were used for in vitro cell assays. Flow cytometry studies utilizing gymnotic uptake with dye-labeled ASO (GFP+99.8%) were completed to determine the cellular entry and cell viability of our candidate ASO. Successful cellular entry and viability of the ASO is confirmed in Figure S4.

Figure 2.

Mechanism of ASO Intervention of IGHMBP2 Variant

The ASO targets IGHMBP2, the paternal splicing variant resulting in the cryptic splice site. The objective of the treatment is to increase protein levels by avoiding NMD.

ASO treatment increases mRNA levels of IGHMBP2

The top candidate ASO, VCA-894A, is a 19-mer ASO with a 2′-MOE modification and phosphorothioate backbone, targeting the sequence CACTTCCAC(A)GGGGGAAGA around the variant of interest. VCA-894A treatment of patient fibroblast cells (48-h incubation) boosted IGHMBP2 expression by a factor of approximately 1.88-fold relative to untreated control samples (Figure 3A). This restored nearly all transcriptional levels diminished by the paternal cryptic splice variant. As the maternal missense allele has normal mRNA levels, but the paternal allele is subject to NMD, the patient cell line has ∼50% detectable mRNA levels. A 1.88-fold relative increase therefore restores to nearly normal mRNA levels—although this would be ∼50% nonfunctional maternal mRNAs, and ∼50% functional paternal mRNAs. qPCR results further confirmed an increased ratio of restored WT transcript to cryptic exon-containing transcript (Figure S2F). These results revealed VCA-894A as the top ASO candidate for restoring transcriptional levels. This was further confirmed with RNA sequencing, as an increase in IGHMBP2 expression of 2.5-fold (adjusted p value < 0.002) following treatment of VCA-894A in the patient’s fibroblast cells was observed (Table S1). Dose-response analysis revealed that VCA-894A’s half-maximal potency was in the nanomolar range. Dose-response analysis was completed with VCA-894A to determine IGHMBP2 relative expression at concentrations of 0, 250 nM, 500 nM, and 1.0 μM. Our selected ASO showed an EC50 of ∼300 nM. As seen in Figure 3B, concentrations of 250 nM, 500 nM, and 1.0 μM VCA-894A all increased IGHMBP2 relative expression in the patient’s fibroblasts, as compared with untreated control fibroblasts. Observed is a dose-responsive increase in RNA following treatment, achieving the greatest increase of IGHMBP2 relative expression.

Figure 3.

VCA-894A restores IGHMBP2

(A) Relative mRNA expression of IGHMBP2 in fibroblast-grown cells following VCA-894A treatment (48-h incubation, 1 μM). The mean expression of untreated samples is 0.9968, and the mean of treated samples is 1.744. The difference between means (B − A) ± SEM is 0.7473 ± 0.05081. 95% confidence interval of 0.6063–0.8884. R squared (eta squared) is 0.9818. The expression levels are significantly higher in the VCA-894A-treated samples (unpaired t test, ∗∗∗p value < 0.001). GAPDH expression was used to normalize expression across samples. (B) Dose-response analysis was completed with VCA-894A to determine IGHMBP2 relative expression at concentrations of 0, 250 nM, 500 nM, and 1 μM. A dose-responsive increase in RNA is observed. We utilized three biological replicates/each condition and each experiment was repeated twice. The patient’s fibroblasts underwent 48-h incubation via gymnotic uptake of ASO. A one-way ANOVA test was used to calculate statistical significance (∗∗∗∗p value < 0.0001). (C) IGHMBP2 western blot results show significant protein restoration with 0.5 μM and 1.0 μM VCA-894A treatment (72-h incubation) (∗∗∗p value < 0.001).

VCA-894A ASO restores protein levels

Next, upon treatment of fibroblasts with VCA-894A (72-h incubation), we observed a significant increase of approximately 50%–70% of IGHMBP2 protein levels in ASO-treated samples compared with untreated control samples (p value < 0.0001). Western blot analysis using IGHMBP2 antibody confirmed protein restoration, showing increased levels in the protein lysate obtained from the treated fibroblasts compared with untreated controls. Different time points for analyzing mRNA and protein levels were assessed, as we expected transcriptional changes to be observed faster than protein changes (48-h vs. 72-h). This analysis is depicted in Figure 3C.

VCA-894A has limited off-target binding

Since off-target binding of ASOs can lead to side effects and toxicity, we examined potential off-target binding sites of VCA-894A in the human and rat genomes by in silico analysis. Using the Basic Local Alignment Search Tool (BLAST), we examined the sequence lengths of 19–15 nucleotides. Results for the human genome are provided in Figure S5. VCA-894A had no off-target matches to the human transcriptome until it was allowed ≥4 mismatches. This is suggestive of a limited off-target binding footprint on the human transcriptome. It is currently understood that single base-pair mismatches can significantly reduce binding and antisense activity. In the case of other therapeutically approved products, such as Inotersen, >3 mismatches resulted in no reduction of the mismatched mRNAs in either of the cell lines tested.¹⁹ We confirmed that only one of the three (15 mers) predicted off-targets (MPL) of VCA-894A is antisense and in proximity to critical functional regions (exonic) of annotated human mRNAs. The other 15-mers (4 mismatches) are not in antisense orientation to its target. Additionally, the region surrounding the target sequence is not highly polymorphic.

Generation, expansion, differentiation, and characterization of CMT2S-MNs from patient-derived iPSCs

To develop the CMT2S patient-specific NMJ model, iPSCs and iPSC-MNs were generated through reprogramming from patient-derived fibroblasts and healthy WT controls followed by expansion and differentiation. Bright field imaging and immunostaining of human iPSC (hiPSC) colonies confirmed the expression of pluripotent specific markers octamer-binding transcription factor 3/4 (OCT4), stage-specific embryonic antigen-2 (SSEA-2), and TRA-1-81.

The CMT2S iPSCs reprogrammed from the patient’s fibroblasts were expanded, and the iPSC stock was established. Phase images of these iPSCs after expansion demonstrated typical iPSC colony morphology. MNs were differentiated from CMT2S iPSC utilizing an established protocol.²⁰ Immunocytochemistry (ICC) staining of CMT2S-MNs indicated their positivity to neuronal marker microtubule-associated protein 2 (MAP2) and MN marker choline acetyltransferase (ChAT), confirming cell type and viability (Figure 4A). ICC staining with MAP2 was used to identify axons with axonal hillock morphological features, confirming the presence of axons (Figure S6).

Figure 4.

Characterization of CMT2S iPSC-MN

(A) Immunocytochemistry staining of CMT2S iPSC-MNs indicated their positivity to neuronal marker MAP2 and MN marker ChAT, confirming cell type and viability. (B) Characterization of hiPSC-MN axonal varicosity at 3 weeks. Confocal microscopy of WT-MNs at 20X emphasizing staining morphology for analysis indicating expression of MAP2 and NF. Confocal microscopy of CMT2S-MNs at 20X emphasizing staining morphology for analysis indicating expression of MAP2 and NF. Scale bar, 100 μm. Varicosities are identified by arrows. Quantification of varicosity analysis measured in varicosities per micron at 3 weeks in vitro. n ≥ 40 neurons for each genotype from at least three batches of culture were analyzed. Data represent mean ± SEM. Asterisks indicate that the condition is significantly different from the WT at the same time point: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Morphology and physiology of CMT2S iPSC-MNs

CMT2S iPSC-MNs were compared with WT iPSC-MNs to characterize CMT2S-MN defects caused by IGHMBP2 variants. MN axonal morphology was analyzed by ICC staining utilizing the marker neurofilament (NF) co-stained with MAP2 (Figure 4B). Analysis of CMT2S-MNs on week 3 revealed thinner processes with a higher number of varicosities compared with WT-MNs (Figure 4B). Axonal varicosities can affect axonal conduction and transmission, and represent a key pathological feature believed to develop via slow accumulation of axonal damage that occurs during neuronal degeneration.^15,20 CMT2S-MNs demonstrated an expedited maturation compared to WT-MNs, as determined by axon progression from the MN cell bodies, penetration through the NMJ tunnels, and consequent skeletal muscle (SkM) innervation.

Electrophysiological characterization of CMT2S-MNs demonstrates slight hyperexcitability at early stages, but overall functional decline at later stages

Electrophysiological comparison between control WT-MNs and CMT2S-MNs was performed through patch-clamp recording (Figures S7 and S10). Quantitative analysis of MNs for 4 weeks in the culture indicated a slight increase of action potential (AP) amplitude in weeks 1 and 2 of CMT2S-MNs, and a slight decrease of resting membrane potential (RP) in week 3 (Figure S7). Analysis of spontaneous firing of CMT2S-MNs displayed a significant reduction of firing frequency and a reduction in the percentage of cells that are spontaneously active in weeks 3 and 4 (Figure S8). However, in week 1, there is an approximate double percentage of CMT2S-MNs that are spontaneously active compared with WT (Figure S8C). Likewise, analysis of Na⁺ and K⁺ currents under stepped voltage-clamp conditions indicates a significant reduction of these currents in weeks 3 and 4 of CMT2S-MNs (Figures S9B and S9C), and a slight increase of Na⁺ currents at week 1 (after normalization to cell capacitance, Figure S9D). In the same trend, analysis of induced repetitive firing under stepped current-clamp conditions indicates a significant reduction of firing frequency in weeks 3 and 4 (Figure S10). Altogether, these analyses suggest a significant functional decline in the later stages of the CMT2S-MNs, with a slight hyperexcitability at the early stages.

While the functional decline at later stages of CMT2S-MNs aligns well with clinical symptoms, this hyperexcitability of MNs shares a resemblance to an amyotrophic lateral sclerosis (ALS) phenotype. However, compared with ALS-MNs, hyperexcitability is observed much earlier in CMT2S-MNs, as soon as day 5, while it is not observed until day 20 in ALS-MNs. This may correlate with CMT2S being an early-onset degenerative disorder, while ALS develops later in life. Further, patient-derived spinal MNs of a CMT2A patient and a CMT2E patient showed similar results of hyperexcitable neurons and altered ion channel kinetics.²¹

Functional NMJ evaluation of CMT2S caused by IGHMBP2 variants

To determine the NMJ defects in MNs expressing the CMT2S IGHMBP2 variants, we utilized the NMJ platform established by Hesperos Inc.²² MNs differentiated from the CMT2S iPSC line, as well as those from healthy control iPSCs (WT-MNs), were integrated into a dual-chamber NMJ platform in the co-culture with WT iPSC-derived SkM myofibers (iPSC-SkM) in a defined serum-free medium after CMT2S characterization. The establishment of the NMJ platform and timeline is depicted in Figure S11. The cell culture development of NMJ systems for WT and CMT2S-MNs can be seen in Figure S12. Innervation of both WT and CMT2S-MNs was confirmed with ICC for acetylcholine receptor (bungarotoxin-647), NF heavy chain, and synaptophysin (Figure S13 and Videos S1, S2, S3, S4, S5, and S6). NMJ defects were investigated by analyzing established clinically relevant functional parameters of NMJ number per chamber, NMJ fidelity, and NMJ fatigue index (FI). The NMJ function was interrogated longitudinally at three time points (days 12, 14, and 16 of SkM maturation) to monitor the formation, quality, and maintenance of NMJ function. To determine any potential functional deficits in developing MN cultures, MNs were also plated 1 week later, as fully developed SkM is necessary to interrogate immature MNs (corresponding to days 5, 7, and 9).

To assess NMJ number, chambers were electrically stimulated on the MN side and the number of contracting myotubes was counted using an in-house microscopy software. The percentage of myotube responses was calculated by measuring the number of contracting myotubes exhibiting each distinct phenotype at each time point. The ratio of the number belonging to any one phenotype relative to the total amount is the percentage. A decrease in NMJ number at an early stage of testing may imply a difficulty in NMJ formation, while a decrease at later testing days may suggest a defect in NMJ maintenance. CMT2S and WT systems were significantly different on individual days 7, 12, and 14, as seen in Figure 5. There is a level of stabilization in WT systems with no variable response across all time points, while CMT2S-MNs have a different response depending on the progression of SkM maturation. NMJ fidelity was examined by testing the synchronous firing of a specific NMJ at five different frequencies (0.3, 0.5, 1, 2, and 4 Hz). The percentage of successful muscle contractions driven by MN stimulation was quantified to identify the ability of a CMT2S-NMJ to reliably transmit electrical activity from an MN to muscle to induce muscle contraction at increasing stimulation frequencies. No significant differences were seen in fidelity percentage between the CMT2S-NMJs and WT-NMJs (Figure 5), with both groups expressing similar trends of maturation in functionality.

Figure 5.

NMJ number and fidelity

NMJ number is shown for both WT and CMT2S systems plated on two different timelines (Late plating – days 5, 7, and 9; Early plating – days 12, 14, and 16). CMT2S and WT systems were significantly different on individual days 7, 12, and 14 for NMJ number (two-way ANOVA with Fisher’s LSD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001). Fidelity scores for NMJs formed at these timelines are shown as bar graphs with increasing frequency from left to right (0.33, 0.5, 1.0, 2.0, and 4.0 Hz). No significance was determined between WT and CMT2S fidelity at any time point. Means are shown as data points or bars, with error bars reporting SEM.

Next, we performed an NMJ FI functional readout. FI is generated when muscle contraction demonstrates complete or partial tetanus at 2 Hz for a set duration. Tetanus categories are depicted in Figure 6A. Muscle fatigue is a common clinical symptom seen in CMT2S patients and is observed in our individual patient. Quantification of the FI showed an increased fatigue in the CMT2S-NMJs on all test days, as compared with WT-NMJs (Figure 6B). Compared with the WT FI representative trace, the CMT2S FI representative trace demonstrated quick and chaotic fatigue and presented with a tetanus followed by decay response. Across all time points, WT and CMT2S systems were significantly different. This response may be due to the inability of CMT2S-MNs to sustain axonal transport or vesicular secretion of neurotransmitters, consistent with the axonopathy phenotype of CMT2S.

Figure 6.

NMJ functional parameters

(A) Tetanus categories: Classical tetanus, tetanus followed by decay, chaotic responses, and non-classical tetanus. (B) The FI is defined as in Equation 1. Means are shown as data points or bars, with error bars reporting standard error of the mean (SEM). Quantification of the FI showed increased fatigue in the CMT2S-NMJs on all test days, as compared with WT-NMJs (∗∗∗∗p < 0.0001). Shaded areas show a duration of 2 Hz stimulation (2 min). (C) CMT2S-NMJ systems show a high level of decay and chaotic tetanus and a low level of classical tetanus, as compared to WT-NMJs. CMT2S-NMJ systems show a higher level of non-classical tetanus, as compared with WT-NMJs.

Analysis of the types of tetanus (Figure 6A) present in the CMT2S-NMJ system revealed a dominant phenotype of tetanus followed by decay. Tetanus types in WT vs. CMT2S NMJ systems can be seen in Figure 6C. WT systems overwhelmingly show a classical phenotype in tetanus, whereas CMT2S systems show a predominantly chaotic decay phenotype. As WT systems get more fatigued with repeat testing (days 12, 14, and 16), the proportion of classical response to decay responses decreases. In addition, the more the CMT2S systems are tested, the amount of myotubes that can generate a classical tetanus response decreases. Chaotic responses in both WT and CMT2S systems increase with repeated testing. Non-classical responses appear high in developing WT systems, but once the NMJ has reached functional maturity the amount of non-classical responses becomes minimal. In contrast, CMT2S systems show non-classical responses at varying points along development, with a potential decline noted by day 16.

VCA-894A dosing in NMJ systems

VCA-894A treatment was examined in the NMJ system to evaluate the therapeutic effect on CMT2S-MNs. The following groups were analyzed in two cohorts (Dosing Cohort 1 and Dosing Cohort 2): non-dosed WT-MN control, vehicle dosed CMT2S-MN, CMT2S-MN Dose 1 (10 nM VCA-894A), CMT2S-MN Dose 2 (100 nM VCA-894A), and CMT2S-MN Dose 3 (1 μM VCA-894A). A schematic of the dosing timeline can be seen in Figure 7A.

Figure 7.

ASO dosing NMJ functional assessment – fatigue index

(A) ASO dosing experimental timeline. Two different dosing strategies were utilized for determining recovery from MNs plated on the standard timeline. Dosing Cohort 1 (day 3 dosing) and Dosing Cohort 2 (day 12 dosing) were dosed with VCA-894A according to the depicted timeline. (B) ASO dosing assessment on the Dosing Cohort 1 (left panel) and Dosing Cohort 2 (right panel). There is a significant restoration of functionality in Dosing Cohort 1 with 1 μM on day 12, 100 nM on day 14, and 10 nM and 100 nM on day 16 and day 20. There is a significant improvement in Dosing Cohort 2 systems dosed with 1 μM VCA-894A on day 14, with near significance by day 20 (p = 0.0708). Means are shown as data points or bars, with error bars reporting the standard error of the mean (SEM). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

The two cohorts utilized for the ASO dosing experiments represent “early” and “late” treatment with VCA-894A. Dosing Cohort 1 (dosing on day 3 or “early” treatment) is described as immature MNs, with little to no NMJs present. The purpose of this cohort was to examine whether VCA-894A affects the ability to form stable NMJs and restore functional parameters, modeling patient stem cell populations replacing dysfunctional MNs after treatment, or early childhood intervention. Dosing Cohort 2 (dosing on day 12 or “late” treatment) is described as mature MNs that have stable but dysfunctional NMJs present. The purpose of this cohort was to examine whether VCA-894A affects already-formed NMJs to restore functional parameters, modeling a later therapeutic intervention and a more mature patient population.

To determine the effects of VCA-894A on NMJ stability, NMJ counts were performed before (pre-test) and after (post-test) fidelity measurements, as changes in post-test counts compared with controls may indicate a loss of stability due to testing. An increase in NMJ numbers was previously observed in the CMT2S-MN line compared with WT (Figure 5). For pre-test NMJ number, no significant difference is noted between groups at any time point, except for Dosing Cohort 2 immediately after dosing, where 10 nM VCA-894A significantly decreases the amount of NMJs in the system (Figure S14A). Over time, all systems level out and show no significant difference past the initial dosing day. This demonstrates that VCA-894A is not negatively affecting the amount of NMJs in each system. Post-test NMJ number shows no significant difference in the amount of NMJs that survive after testing between any of the groups (Figure S14B). This demonstrates that VCA-894A is not negatively affecting the formation or stability of NMJs.

Initial comparisons of CMT2S and WT-NMJs revealed no significant differences between fidelity (Figure 5). The results of VCA-894A treatment on fidelity are depicted in Figure S15. Initially, at day 12, there was a significant decrease in the Dosing Cohort 2 vehicle group compared with WT, with additional significant decreases in fidelity for both Dosing Cohort 1 at 100 nM and 1 μM and for Dosing Cohort 2 at 10 nM and 1 μM compared to the vehicle group. This initial deficit in functionality is likely due to a disruption of the NMJ from the initial addition of VCA-894A. By day 20, there is a significant difference between WT and vehicle controls in both dosing cohorts. This was not previously observed, as initial experiments did not go to day 20. At day 20, VCA-894A shows a significant restoration in functionality for the existing NMJs (Dosing Cohort 2) at 100 nM, with non-significant increases in mean fidelity for 10 nM and 1 μM. There is no significant effect for Dosing Cohort 1 on day 20. These data suggest that VCA-894A can enhance the functionality of existing NMJs over time.

The ability of VCA-894A to improve muscle fatigue is an important clinical benefit. As demonstrated in Figure 6B, the CMT2S-NMJ systems display a higher FI than WT. Results of VCA-894A treatment effects on FI in the CMT2S and WT NMJ systems are depicted in Figure 7B. There is a significant restoration of functionality in Dosing Cohort 1 with 1 μM on day 12, 100 nM on day 14, 10 nM and 100 nM on day 16 and day 20, indicating the ability of VCA-894A to ameliorate deficits in newly developing NMJs. There is a significant improvement in Dosing Cohort 2 systems dosed with 1 μM VCA-894A on day 14, with near significance by day 20 (p value = 0.0708). This indicates that VCA-894A may have the ability to restore functionality in currently stable NMJs.

As previously described, CMT2S-MNs elicit a distinct tetanus response, a tetanus and decay phenotype, that is consistent with the axonopathy phenotype of CMT2S. Following treatment with VCA-894A, there is a significant change in the proportion of phenotypes in Dosing Cohort 1, with 10 nM showing a significant restoration at all days except day 14, 100 nM showing significant restoration across all time points, and 1 μM showing significant differences only on day 12 and day 14 (Figure 8). Dosing Cohort 2 systems treated with 100 nM show significance at early time points (day 12 and day 14), with 1 μM dosing showing significant effects at day 12, day 14, and day 20. This indicates the ability of VCA-894A to reverse the type of tetanus and sustain longer SkM contractions in both early and late dosing cohorts, improving NMJ tetanus responses.

Figure 8.

ASO dosing NMJ functional assessment—tetanus type

There is a significant change in the proportion of phenotypes in Dosing Cohort 1 systems, with 10 nM showing a significant restoration at all days except day 14, 100 nM showing significant restoration across all time points, and 1 μM showing significant differences only at day 12 and day 14. Dosing Cohort 2 systems treated with 100 nM show significance at early time points (day 12 and day 14), with 1-μM dosing showing significant effects at day 12, day 14, and day 20. One-way multiple chi-square; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p <0.0001.

Three-month rat toxicology study

The potential toxicity of VCA-894A was examined when given by intrathecal administration over 3 months to 80 WT Sprague-Dawley rats. As nusinersen has been well studied for toxicity and efficacy, and VCA-894A is a similarly sized ASO with identical chemical modifications, we have determined that the optimal dosage of VCA-894A is 12.0 mg. In the 3-month potential toxicity study, we evaluated doses of 0.03 mg, 0.11 mg, and 0.35 mg, which are comparable to 15 mg, 50 mg, and 150 mg of a human dosage. The following parameters and endpoints were evaluated in this study: mortality, clinical observations, body weights, ophthalmology, neurobehavioral evaluations, clinical pathology parameters (hematology, coagulation, and clinical chemistry), cytokine analysis, bioanalysis in plasma and cerebrospinal fluid, toxicokinetic parameters, organ weights, and macroscopic and microscopic examinations.VCA-894A administration on days 1, 29, 57, and 85 was well tolerated at all dose levels and there were no unscheduled deaths during the course of the study. There were no VCA-894A-related effects on body weights, ophthalmology, neurobehavioral evaluations, clinical pathology parameters, cytokine analysis, organ weights, and macroscopic examinations. No glial activation was noted. Effects were limited to transient clinical signs following administration of 0.35 mg VCA-894A (abnormal gait, low carriage, incoordination, limited usage of the hindlimbs, decreased muscle tone, partly closed eyes, erected fur, and decreased activity) and microscopic findings in the brain, spinal cord (included injection site), dorsal root ganglion, nerve roots, and kidneys, which were noted unremarkable. Clinical findings were mainly noted on day 1 and sporadically on days 29, 57, and 85, after recovery from anesthesia and resolved within 24 h. Similar findings of transient (∼24 h) clinical signs have been observed in in vivo studies completed with similar ASOs, milasen and nusinersen, at doses comparable to ≥0.1 mg in rats, suggesting a similarity in safety profiles.²³ Minimal nerve fiber degeneration was observed only in the spinal nerve roots; it was not present in the white matter of the spinal cord or the brain. There was no evidence of degeneration of the neurons in the dorsal root ganglia or in the spinal cord/brain. There was only one animal in the low-dose group that presented a mild nerve fiber degeneration in the cervical nerve root; since it was a single incident and deemed to be non-dose responsive, we consider this finding sporadic and nonadverse. At a minimal grade of severity, approximately 1 to 7–8 degenerated fibers were found out of 50–100 nerve fibers on the slides. Since no degeneration of the neurons was seen in the spinal cord or the dorsal root ganglia, we conclude that there was no direct degeneration of the neuron caused by VCA-894A. The degenerated nerve fibers were mainly seen at the level of the injection site (lumbar), which is more suggestive of a procedure-related effect. This change was not considered adverse based on location (only nerve roots), absence of associated neuronal degeneration, and absence of correlating neuronal deficit clinical signs. Based on the transient nature of the clinical signs and the nature and severity of the pathology findings, the aforementioned changes were considered nonadverse; the no observed-adverse-effect level (NOAEL) was considered to be the highest dose tested at 0.35 mg/dose with an AUC_0–24h accumulation ratio of 0.868 h∗ng/mL. Our findings induced by high doses of VCA-894A provide translatable and readily monitorable indicators that were used to guide safety monitoring in our clinical protocol design, as we plan to move into clinical application.

Discussion

ASOs are uniquely capable of directly targeting cryptic splice variants through steric block, effectively restoring protein levels. A patient was reported with pathogenic variants within IGHMBP2. WGS revealed a paternally inherited cryptic splice site variant (non-coding variant [c.1235 + 894C>A] deep in intron 8). The resulting transcript undergoes NMD. This variant is suspected to cause the disease manifestation of CMT2S in this patient. In this study, we illustrate in vitro proof-of-concept functional experiments utilizing a splice-modulating ASO tailored to the individual genome of this specific CMT2S patient. Our objective was to target this specific cryptic splice site variant, c.1235 + 894C>A, rescuing IGHMBP2 protein levels with a novel ASO, VCA-894A. Similar to nusinersen, VCA-894A is a 19-mer ASO with the same backbone and sugar chemistry modifications, phosphorothioate and 2′-O-MOE. Nusinersen is an FDA-approved ASO drug for SMA that changes the splicing pattern of SMN2 RNA.¹⁴ Following a similar dosing regimen, VCA-894A is intended for intrathecal administration to treat this individual patient’s neuromuscular disease, CMT2S.

The patient’s distinctive CMT2S phenotype is characterized through exploratory studies on MNs derived from the patient’s fibroblast cells. Patient-derived CMT2S iPSC-MNs, compared to WT iPSC-MNs, revealed an early stage hyperexcitable phenotype with a higher percentage of neurons displaying spontaneous firing, comparable to ALS. However, hyperexcitability in CMT2S iPSC-MNs had a much earlier onset than ALS iPSC-MNs. This observation may correlate with the observed early onset of CMT2S clinical manifestation, compared with the later onset of ALS which is an age-related neurodegenerative disease. The hyperexcitability observed in CMT2S-MNs may be caused by the observed reduced membrane capacitance and thus induced higher Na⁺ current density. At a later stage, CMT2S-MNs displayed overall functional decay manifested as a decline in the percentage of MNs exhibiting spontaneous firing, the frequency of spontaneous firing, the peak Na⁺ and K⁺ currents, Na⁺ current density, and the induced repetitive firing frequency. This phenotype aligns well with the motor function decay observed in the patient. Furthermore, NMJ FI analysis shows quick fatigue, as well as a chaotic tetanus phenotype, which may correlate with the observed CMT2S clinical phenotype. Following treatment with VCA-894A, we demonstrated rescue of NMJ functioning, captured by a decrease in FI and decay and chaotic responses. Although these functional assays are novel and show promising advancements in terms of translating effects of rare genetic variants where no benchmarked methods exist, it is important to note their limitations. While advantageous for translating the effects of a phenotype and demonstrating ASO rescue on functional parameters, these assays are suboptimal for dose selection. In this case, standard qPCR-based dose selection was utilized. Further studies on CMT2S-MNs are needed to confirm the effect variants within IGHMBP2 can have on MN function and clinical phenotype. Patient-derived iPSC models have been shown to be successful models of CMT2 for both phenotype characterization and evidence of treatment restoration of disease phenotypes.²⁴

Previous research has revealed the altered neuronal morphology of CMT2S-MNs. In CMT2S cell lines presenting IGHMBP2 variants, diseased MNs displayed shorter neurite lengths, lower numbers of neurites per soma, and higher percentages of neurons without neurites compared to healthy control MNs.²⁵ Treatment with AAV9.IGHMBP2 successfully improved all phenotypes in CMT2S lines; neurite length and percentage of neurons with neurites all significantly increased, while the percentage of neurons without neurites decreased.²⁵ This provides evidence of the role that IGHMBP2 variants possess in MN morphology and the ability to rescue these defects by IGHMBP2 protein rescue. While there is currently no available treatment for CMT2S, there is an ongoing trial initiated by Alcyone Therapeutics for an IGHMBP2 gene replacement therapy to treat individuals with IGHMBP2-related diseases, such as SMARD1/CMT2S (NCT05152823). This group demonstrated the high efficacy of IGHMBP2 gene therapy in both the SMARD1 mouse model and in an in vitro model of the human disease, where AAV9-IGHMBP2 gene transfer resulted in improvements in motor function and survival.²⁶ As IGHMBP2 expression was increased, consistent preservation of the number of MNs present in treated mice was shown, and neuromuscular phenotype was rescued.²⁶ Additionally, the researchers were able to ameliorate defects seen in NMJs and improve the size of myofibers in skeletal muscle and heart.²⁶ Systemic AAV9-IGHMBP2 gene therapy was determined to be efficacious even when delayed treatment was provided, proving the concept that IGHMBP2-related diseases may be treatable in older individuals who have already established the disease phenotype.^27,28

Several other in vivo studies have demonstrated that introducing IGHMBP2 can rescue disease phenotype in SMARD1 mouse models.^26,27,28,29 The ability to restore IGHMBP2 expression can consequently lead to an increase in IGHMBP2 protein, resulting in a decreased manifestation of CMT2S. VCA-894A has shown success in vitro in rescuing IGHMBP2, which, as previous studies have shown, should result in restored motor function. Limitations of using organ-on-a-chip models include the inability to fully recapitulate the complex pathophysiology observed in patients, especially those with disease presentations involving various tissues.³⁰ However, evaluations of the predictive value of using animal studies is limited due to substantial interspecies differences.³⁰ In the case of rare diseases, for which no model currently exists and specific genetic variants are targeted, organ-on-a-chip models prevail as the most specific model. Our patient-specific iPSC-derived MN and NMJ model will allow us to continue testing the efficacy of the selected VCA-894A ASO treatment to repair observed MN and NMJ functional defects by achieving IGHMBP2 protein rescue.

Overall, the data reveal the ability of VCA-894A to enhance the functionality of existing NMJs over time, ameliorate deficits in newly developing NMJs, restore functionality in currently stable NMJs, and improve tetanus response, sustaining longer SkM contractions in both immature and mature MNs. VCA-894A does not negatively affect the amount, formation, or stability of NMJs after dosing. These results point toward the expected clinical benefits of sustaining longer and more normal muscle contractions, as well as reducing muscle fatigue, without harming developing or developed MNs.

This study assessed the efficacy of a novel ASO in restoring IGHMBP2 protein levels through in vitro experiments performed on an individual CMT2S patient’s fibroblast cell line. The therapeutic potential of ASOs to modulate gene expression and rescue protein has been confirmed in clinical trials for a variety of indications, with success at targeting the CNS.¹³ While improved clinical formulation of the ASO may be necessary, the current preclinical data support this ASO as a potential treatment for this individual CMT2S patient. In a 3-month rat toxicity study, we observed that VCA-894A administration to WT Sprague-Dawley rats by intrathecal lumbar injection was well tolerated with an NOAEL of the highest dose tested.

An increased number of cases of autosomal recessive CMT2S caused by IGHMBP2 pathogenic variants are being reported, which harbor noncoding variants in trans with coding variants.³¹ With the increasing use of WGS in the diagnosis of CMT, more cryptic intronic variants are being delineated, specifically in the intron 8 hotspot of IGHMBP2.³¹ While this IGHMBP2 c.1235 + 894C>A variant is only present in this individual patient, per existing databases, more variants of uncertain significance are being reported in the intron 8 region that are in compound heterozygosity with pathogenic variants in IGHMBP2.³² This hotspot of aberrant splice induction may prove to be a therapeutic intervention in treating a number of CMT2S cases.³² The individualized precision medicine approach may prove instrumental in designing treatments for this highly diverse genetic disorder. Milasen is an investigational drug designed for an individual patient that paved the way for personalized ASO treatment for rare genetic diseases.³³ Our study exemplifies the shifting boundary between rapid WGS-based clinical diagnoses and research capabilities, allowing for ASO-based individualized genomic medicine design.

Materials and methods

Whole genome sequencing

Incoming nucleic acid samples were quantified using fluorescent-based assays (PicoGreen) to accurately determine whether sufficient material was available for library preparation and sequencing. DNA sample size distributions were profiled by a Fragment Analyzer (Advanced Analytics) or BioAnalyzer (Agilent Technologies) to assess sample quality and integrity. HumanCoreExome 24v1.3 array was performed on all human DNA samples sequenced. WGS libraries were prepared using the Truseq DNA PCR-free Library Preparation Kit. Whole Genome data were processed on the New York Genome Center (NYGC) automated pipeline. Paired-end 150-bp reads were aligned to the GRCh37 human reference (BWA-MEM v0.7.8) and processed with GATK best-practices workflow (GATK v3.4.0).

ASO design

Multiple ASOs were designed to specifically target the region harboring the variant located in intron 8 (c.1235 + 894C>A). ASOs were chemically modified with a phosphorothioate and 2′-MOE backbone, previously determined to be an efficacious backbone to promote ASO delivery.¹³ The designed ASO candidates were prioritized based on in silico binding affinity and were independently modeled in silico.

Cell culturing

Fibroblasts were cultured in T-75 vented flasks (Thermo Fisher) in DMEM-high glucose (Cat. No. 11965092, Gibco) supplemented with 10% FBS at 37°C and 5% CO₂. Cells were sub-cultured or split 1:3 at around 80%–90% confluency.

RNA isolation and cDNA synthesis

All RNA isolations were performed with the RNeasy Mini Kit (Cat. No. 74104, Qiagen) according to the manufacturer’s protocol. RNA was quantified by Qubit 4.0 fluorometer and Qubit BR RNA kit (Cat. No. Q10211, Invitrogen). One microgram of total RNA was used for all cDNA synthesis reactions. For qPCR analysis, cDNA was synthesized using a High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Cat. No. 4374966 Applied Biosystems), following the manufacturer’s instructions.

Confirmation of NMD

Abnormal transcript caused by the variant was furthermore confirmed to undergo NMD during treatment of patient cell lines with cycloheximide. Total RNA was extracted post-treatment with cycloheximide at 48 h, and cDNA was synthesized using High-Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Cat. No. 4374966 Applied Biosystems). Primers were designed to amplify the IGHMBP2 region, including the exon8-cryptic exon-exon9 region, to confirm the stable expression of the out-of-frame sequence. The sequences of the primers can be found in the supplemental material (Figure S16).

Cellular entry and cell viability experiments

The in vitro cell assay treatment duration with the selected ASO was 48 h. Initial experiments involved the use of ASOs with FAM end-labeled dye.

Gene expression analysis

Initial screenings for gene expression experiments were conducted at 1 μM concentrations. These ASOs were tested in an in vitro assay using patient-derived fibroblasts maintained in DMEM, high glucose with 10% FBS. qPCR analysis was performed using TaqMan primers (Hs01045556_m1) located in exon 8/9 of IGHMBP2. TaqMan Fast Advanced Mix (7500 and 7500 Fast systems with fast cycling mode) was utilized. PCR conditions were first one cycle of 50°C for 2 min, one cycle of 95°C for 20 s, followed by 40 cycles of 95°C for 3 s and 60°C for 30 s.

GAPDH primers were used as a control. All experiments were performed using gymnotic uptake without a transfection reagent and were incubated for 48 h. Screening included a single concentration of 1 μM for 48 h, gymnotic uptake. Dose response was tested at concentrations of 0, 250 nM, 500 nM, and 1μ M ASO (48-h, gymnotic uptake). Forty-eight-hour incubation was selected as we expect transcription changes to be observed faster than protein changes. We utilized three biological replicates for each condition.

Western blot analysis

Western blot analyses were performed using 1:2,000 mouse anti–IGHMBP2 monoclonal antibody (Cat. No. clone11-24 MABE162, EMD-Millipore); 1:5,000 mouse anti-GAPDH (Cat. No. MA5-15738, Invitrogen); 1:5,000 Goat anti-mouse IgG Secondary horseradish peroxidase antibody (Cat. No. 62–6520, Invitrogen). Each sample was normalized to a total of 10 μg per lane for protein loading (Qubit Protein and Protein Broad Range Assay Kit, Cat. No. Q33212, Invitrogen). Proteins were detected using Western Blotting Luminol Reagent (Cat. No. sc-2048, Santa Cruz). ImageJ was used for quantification of signal intensity for each probe tested. Western blot analysis was conducted with 72-h incubation utilizing gynmnotic uptake. We utilized three biological replicates/each condition and each experiment was repeated twice.

CMT2S iPSC line establishment, expansion, differentiation, and characterization

CMT2S iPSC line was generated by a third-party contractor in collaboration with the Hickman lab at the University of Central Florida (UCF). The established CMT2S iPSC line was then expanded, cryopreserved, and differentiated into MNs and SkM cells. The differentiated cells were then phenotypically characterized by comparison with the cells differentiated from WT iPSC.

Generation of CMT2S iPSC line and establishment of CMT2S iPSC stocks

Footprint-free human iPSC line (Cat# iPS11) was derived from human foreskin fibroblasts (HFFs) by ectopic expression of OCT4, SOX2, KLF4, and L-MYC genes using ALSTEM episomal plasmids (Cat #RF202), as previously described.³⁴ The cells were derived using morphological selection criteria and without the use of fluorescent markers or drug selection. When cultured under standard human embryonic stem (ES) cell culture conditions, the morphology of human iPSCs is identical to that of human ES cells. The cells express the pluripotency markers OCT4, SSEA-3, Nanog, and endogenous alkaline phosphatase as demonstrated by ICC. High-viability, low-passage iPSCs have been pre-adapted to serum-free, feeder-free culture conditions. The established iPSC line was then grown up and expanded in mTesR1 medium (Cat# 85850, Stemcell Technology) and series passaging and cryopreservation. The chromosome integrity of the iPSCs was inspected through karyotype analysis (Cell Line Genetics).

iPSC passage number is counted based on the number of iPSC expansion/harvest cycles. Normally for each cycle, one vial of iPSC is thawed and plated into a six-well plate and grown in expansion medium until they become ∼70%–80% confluent, about 5–7 days. iPSCs are then harvested and dispensed into six cryovials, and cryopreserved into liquid N₂. The iPSC stock provided by the third-party vendor is identified as passage 0 (P0). The iPSCs for CMT2S-MN and SkM differentiation are at passage 4 (P4).

Differentiation of CMT2S-MNs from iPSCs and phenotypic analysis of CMT2S-MNs

MNs were differentiated from CMT2S-iPSCs or WT-iPSCs (iPSCs derived from healthy subjects) following the established protocol as described by Qu et al.³⁵ with modifications for replacing Component C with LDN 193189 (0.1 × 10-6 μM) and SB431542 (4 × 10-6 μM). The differentiated MNs were cryopreserved. Phenotypes of CMT2S-MNs were analyzed by comparing them with WT-MNs through the characterization of phase imaging, ICC, and patch-clamp electrophysiology.

Differentiation of SkM from iPSCs and NMJ culture

Human iPSCs line (ND41865) derived from a healthy subject was obtained from the Coriell Institute for Medical Research. iPSCs were passaged a maximum of 12 times per recommended NIH passaging protocols. Cells from passages 6 to 12 were used in myoblast differentiation experiments. Myoblast differentiation was performed strictly following a protocol published by Chal et al.³⁶ Generated myoblasts were harvested and cryopreserved for downstream applications. For application in the NMJ platform, NMJ chambers were prepared.¹⁵ IPSC-myoblasts were plated in a SkM chamber in Myocult medium (Stemcell Technologies Inc) for growth until confluence. Next, SkM cultures were submitted to the induction of myotube fusion by switching to DK-HI medium.³⁷ Two days later, the medium was switched to NBActiv4 (Brainbits) and was maintained in this medium by half-medium change every two days until the end of the experiments.

Formation of functional NMJs in chambers

SkM myoblasts were plated into NMJ chambers on day −3 (relative to induction of myoblast fusion, when cultures are switched to NBActiv4). CMT2S or WT iPSC-MNs were plated into the MN chamber in MN medium 1 day after plating SkM (day −2 of NMJ culture) and maintained in MN medium by half-medium change every 2 days until the end of experiments. NMJ function was then tested on scheduled days of the co-culture. In the case of investigating maturation of the NMJ, MNs were plated both on the standard timeline and 7 days after the standard timeline, or on day 5 of SkM maturation, to ensure that functionally mature SkM would be present for interrogation of functionally immature MNs. In this case, data from days 5, 7, and 9 in Figures 5 and 6 utilized a staggered plating schedule, while days 12, 14, and 16 were with the standard plating timeline. All ASO dosing experiments utilized the standard plating timeline. A schematic of the establishment of the NMJ testing platform can be found in Figure S11.

CMT2S cell line functional testing in NMJ

MNs differentiated from the CMT2S iPSC line (CMT2S-MNs), as well as those from WT-MNs, were integrated into the dual-chamber NMJ platform in the co-culture with WT iPSC-SkM in a defined serum-free medium after CMT2S characterization. The presence of NMJs within the NMJ system was confirmed via ICC. Terminal fiber staining using synaptophysin to represent the presynaptic component of the NMJ in combination with nicotinic acetylcholine receptors (nAChR) localized on the SkM myofibers to represent the postsynaptic component.¹⁶ Synaptic vesicles containing synaptophysin can be observed at AChRs localized to the SkM myofibers with innervation highlighted by the presence of NF heavy chain staining in ICC images acquired at 20× and 60× magnification, supporting the formation of NMJs from iPSC-differentiated MNs (Figure S13).

The NMJ functional defects were investigated by analyzing established parameters (NMJ number per chamber, NMJ fidelity, and NMJ FI). NMJ function was interrogated longitudinally at three time points to monitor the formation, quality, and maintenance of the NMJ. The custom data collection and corresponding functional analysis produced from the NMJ platform utilizes a real-time video recording of the SkM culture as it responds to the direct stimulation of the MN culture in the neighboring, electrically isolated chamber.^16,38 Thus, contraction is driven by axonal innervation to the SkM and as a result, the site of NMJs that are present in the SkM chamber of the NMJ bipartite system. As a result, the post hoc analysis of the motion capture video recording of the SkM response can be assessed for clinically relevant readouts of NMJ number, fidelity, and FI to identify conditional NMJ dysfunction. This experimental workflow enables the tracking of NMJ number over time and matching between timepoints for each system, allowing for a more accurate temporal analysis. These data can then be stratified to determine the rate of degradation, utilize normalization techniques, and run matched statistical analyses solely by utilizing system functional data.

Effects of VCA-894A on CMT2S-MNs in bipartite NMJ systems

VCA-894A was examined in the NMJ system to evaluate the therapeutic effect on CMT2S-MNs. A bolus dose was administered at one of three concentrations (10 nM, 100 nM, or 1 μM) of VCA-894A to generate a therapeutic dose response. NMJ systems were re-dosed every 48 h to maintain the target concentration of VCA-894A. Day 19 was the last day of the ASO dosing strategy. Measurements were performed on days 12, 14, 16, and 20. Functional testing was performed 3 h post-dosing of VCA-894A. Establishment of the NMJ testing platform can be found in Figure S11. VCA-894A was prepared in a 20% TE buffer. The final vehicle concentration in all dosed systems was 5% vehicle in standard cell culture media (corresponding to 1% TE in solution).

Study timelines: NMJ testing

Due to the expedited maturation of CMT2S-MNs and a high degree of SkM innervation, CMT2S-MN plating was delayed by 7 days and NMJ chambers were tested on day 10, which is 2 days earlier than the established Hesperos testing protocol. Dosing Cohort 1 (“early” treatment) and Dosing Cohort 2 (“late” treatment) were dosed with VCA-894A on day 3 and day 12, respectively. Dosing Cohort 1 is representative of utilizing the ASO to enhance the formation and function of developing NMJs, while Dosing Cohort 2 represents the ability of the ASO to improve the function of existing NMJs. A schematic of the dosing timeline can be found in Figure 7A.

Immunocytochemistry for CMT2S cell line characterization

Cultures on coverslips were fixed in freshly prepared 4% paraformaldehyde in phosphate-buffered saline (PBS) (pH 7.2), without Mg²⁺ or Ca²⁺ for 15 min. Cells were washed twice with PBS for 10 min at room temperature and then permeabilized with 0.1% Triton X-100/PBS for 15 min. Non-specific binding sites were blocked using Blocking Buffer (5% donkey serum plus 0.5% bovine serum albumin [BSA] in PBS) for 1 h at room temperature. Cells were incubated with primary antibodies overnight at 4°C. Following primary antibody incubation, cultures were washed with PBS 3X at 5-, 10-, and 15-min intervals with PBS. Cultures were incubated with secondary antibodies for 2 h at room temperature and then washed 3X with PBS for 30 min total. After staining, the coverslips were mounted using ProLong(TM) Gold Antifade Mountant with 40-6-diamidino-2-phenylindole (DAPI) (Thermo Fisher Scientific, P36931). Primary antibodies for MN staining were those against MAP2 (Millipore) and ChAT and for SkM staining were those against MyoD (Abcam) and Pax7 (Abcam). Fluorescence Imaging was performed using an UltraView spinning disk confocal microscope (PerkinElmer). Volocity software was used to process z stack projections of scanned images.

Measurement of axonal branching

CMT2S-MNs and WT-MNs were thawed from cryopreservation and plated on the surface coated with DETA/Laminin in MN medium. The cells were fixed for ICC on days 1, 2, and 3, respectively. The cells were immunostained with MAP2 and DAPI. The images were taken under confocal microscopy. The axons were identified with the morphological feature of axonal hillock. The order of the axonal branching was recorded and the length of each segment was measured using Neuron J as an ImageJ plugin.

Electrophysiology

The electrophysiological properties of iPSC-derived MNs were investigated using whole-cell patch-clamp recording techniques as previously detailed.³⁹ Briefly, cultured neurons maintained on glass coverslips were placed in the recording chamber of a Zeiss Axioscope 2FS Plus upright microscope. In culture, the MNs were visually distinguished from nonneuronal cells using an IR differential interference contrast (DIC) video microscope. Borosilicate glass patch pipettes (BF 150-86-10; Sutter Instrument Company), with a resistance of 6–10 MΩ, were made using a Sutter P97 pipette puller (Sutter Instrument Company). Both current-clamp and voltage-clamp recordings were taken using a Multiclamp 700A amplifier (Axon instruments). The pipette (intracellular) solution contained 1 M ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid, 140 M K-gluconate, 2 M MgCl2, 2 Mm Na2ATP, and 10 M 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES; pH7.2). The MN medium was supplemented with 10 M HEPES (pH 7.2) and was used as the extracellular solution for all patch experiments.

NMJ number

NMJ number is defined as the total number of NMJs formed in a system, quantified as a result of indirect stimulation of the SkM. Prior to dosing, NMJ number was taken for each system. Systems were electrically stimulated on the MN side for two pulses with video recording taken of the SkM side. A custom subtraction algorithm was used to determine movement of contracting myotubes, indicating a functionally mature NMJ. NMJs were counted for each system prior to dosing and prior to testing at each time point. Dosing for each chamber was assigned at random to minimize any potential bias toward chamber functionality. For experiments involving VCA-894A, NMJ counts were performed prior to testing for fidelity and after testing fidelity, to determine any impact of the ASO on the stability of the NMJ.

NMJ fatigue index

For NMJ FI analyses, MNs were stimulated at 2-Hz frequency for 2 min and the skeletal muscle side was recorded. Responding myotubes were selected using in-house analysis software and the FI was calculated. FI is defined as 1 minus the ratio between an ideal tetanic contraction (Y_Max∗Time) and the real contraction observed (AUC). In other words, this is the contractile loss the system exhibits over 2 min. The FI is defined as:

$$Fatigueindex=1-\left[\frac{\left(Areaundercurve\right)}{\left(Peakforce\ast Time\right)}\right]$$ (Equation 1)

Three-month rat toxicology study

The 3-month rat toxicology study was conducted at Charles River Laboratories following GLP requirements. VCA-894A was administered by intrathecal lumbar injection using a 30-G hypodermic needle under anesthesia to female WT Sprague-Dawley rats. Endpoints evaluated in this study included mortality, clinical observations, body weights, ophthalmology, neurobehavioral evaluations, clinical pathology parameters (hematology, coagulation, and clinical chemistry), cytokine analysis, bioanalysis in plasma and CSF, toxicokinetic parameters, organ weights, and macroscopic and microscopic examinations.

Statistical methods

Experiments were performed in triplicate with five independent experiments performed for NMJ characterization of the phenotypic differences between WT and CMT2S systems, and three independent experiments performed for ASO dosing experiments. Data were visualized and analyzed utilizing GraphPad Prism 10.0.3. For NMJ number and FI, a one-way ANOVA with Fisher’s least significant difference (LSD) test was utilized to determine significance at each time point. For NMJ fidelity, a two-way ANOVA without matched frequencies (pooled variance across frequency) with Fisher’s LSD test was utilized to determine significance. Comparisons between proportions of tetanus types were performed using a matched chi-square analysis, with independent chi-square tests run for each comparison, with categories defined as the total number of myotubes exhibiting classical tetanus as opposed to tetanus followed by decay. Means are shown as data points or bars, with error bars reporting the standard error of the mean (SEM).

Data availability

Data are available upon request.

Acknowledgments

Vanda would like to acknowledge the patient who contributed to this study and the patient’s family. We thank Dr. Timothy Yu and Dr. Jennifer Kwon for valuable discussions throughout this project. Vanda would also like to acknowledge Hesperos Inc., with whom we collaborated on phenotypic analyses of CMT2S human-on-a-chip neuromuscular junction systems.

Author contributions

S.S.: conceptualization, methodology, validation, formal analysis, investigation, writing – original draft, review & editing, visualization, supervision, project administration. B.P.: conceptualization, methodology, validation, formal analysis, investigation, writing – review & editing, visualization, supervision, project administration. C.T.: writing – original draft, review & editing, project administration. C.J.: writing – original draft, review & editing, project administration. H.B.: writing – review & editing. J.M.K.: writing – review & editing. D.W.H.: writing – review & editing, experimental design, phenotypic model, NMJ data analysis and compilation, graphs. C.N.: experimentation, NMJ data analysis, and compilation. R.B.: experimental design. K.H.: motoneuron differentiation and phenotypic analysis. R.A.: electrophysiology. A.N.: SkM differentiation and characterization. X.G.: MN experimental design and supervision, MN data analysis and presentation. J.H.: writing – review & editing. C.M.P.: conceptualization, methodology, writing – review & editing, supervision. G.B.: conceptualization, methodology, writing – review & editing, supervision. M.H.P.: conceptualization, methodology, writing – review & editing, supervision.

Declaration of interests

This study was fully funded by Vanda Pharmaceuticals Inc. S.S., B.P., C.T., C.J., H.B., C.M.P., G.B., and M.H.P. are employees of Vanda Pharmaceuticals Inc. M.H.P. is CEO of Vanda Pharmaceuticals Inc. J.H. is a Professor at the University of Central Florida’s NanoScience Technology Center and Co-Founder and Chief Scientist of Hesperos, Inc.