An induced pluripotent stem cell-derived NMJ platform for study of the NGLY1-Congenital Disorder of Deglycosylation

Abstract

There are many neurological rare diseases where animal models have proven inadequate or do not currently exist. NGLY1 Deficiency, a congenital disorder of deglycosylation, is a rare disease that predominantly affects motor control, especially control of neuromuscular action. In this study, NGLY1-deficient, patient-derived induced pluripotent stem cells (iPSCs) were differentiated into motoneurons (MNs) to identify disease phenotypes analogous to clinical disease pathology with significant deficits apparent in the NGLY1-deficient lines compared to the control. A neuromuscular junction (NMJ) model was developed using patient and wild type (WT) MNs to study functional differences between healthy and diseased NMJs. Reduced axon length, increased and shortened axon branches, MN action potential (AP) bursting and decreased AP firing rate and amplitude were observed in the NGLY1-deficient MNs in monoculture. When transitioned to the NMJ-coculture system, deficits in NMJ number, stability, failure rate, and synchronicity with indirect skeletal muscle (SkM) stimulation were observed. This project establishes a phenotypic NGLY1 model for investigation of possible therapeutics and investigations into mechanistic deficits in the system.

Graphical Abstract

NGLY1-CDDG is a rare and recently identified congenital deglycosylation disorder that impairs motor function. Motoneurons derived from NGLY1-deficient patients were co-cultured with skeletal muscle in a NMJ microphysiological system. Myotube contractile response from motoneuron stimulation in diseased and wild-type systems were quantified and compared. Findings highlight the contribution of NGLY1-deficient IPSC motoneurons to neuromuscular dysfunction. Stock art obtained from Biorender.com

1. Introduction

Neuromuscular junctions (NMJs) formed between the presynaptic terminals of spinal cord motoneurons (MNs) and fused skeletal muscle (SkM) myotubes are the connections between the CNS and muscular system that allow for voluntary muscle control. MN depolarization and transmission of action potentials (APs) from the soma to the nerve terminal triggers the opening of voltage-gated calcium channels in the active zone that ultimately results in the release of acetylcholine (ACh) from synaptic vesicles into the synaptic cleft. This ACh then binds to nicotinic ACh receptors (nACHR) on the endplate of the SkM, triggering sodium ion influx and transmission of the AP to the SkM through the creation of an endplate potential that propagates across the membrane and drives myotube contraction.^[1]

A number of heritable diseases can negatively impact the complex process of AP generation and transmission from the MN to the SkM. For example, patients with familial amyotrophic lateral sclerosis (fALS) can possess mutations in superoxide dismutase 1 (SOD1), Fused in Sarcoma (FUS), and chromosome 9 open reading frame 72 (C9orf72), among others, that have been characterized by NMJ denervation and axon retraction that manifests as choreoathetosis (twitching), muscle impairment, difficulties breathing or eating, and peripheral neuropathy.^[2] Similarly, spinal muscular atrophy (SMA) can manifest as hypotonia, paralysis, or tongue fasciculation that stems from spinal cord MN degradation as a result of homozygous loss-of-function mutations in survival of motor neuron 1 (SMN1).^[3] Misdiagnoses of neurodegenerative diseases are unfortunately common due to the nonspecific commonalities between them, resulting in treatments that could harm the patient either directly or by neglecting the actual underlying illness.^[4]

Deglycosylation disorders are rare genetic disorders that manifest similarly to neurodegenerative diseases and have only been discovered in recent years with the application of exome sequencing for identification of undiagnosed genetic conditions.^[5] One disorder in particular, NGLY1-congenital disorder of deglycosylation (NGLY1-CDDG, NGLY1 deficiency), arises as a result of a biallelic inactivation or deletion of NGLY1. When present, NGLY1 encodes the cytosolic N-glycanase 1 (NGLY1) which normally functions to cleave intact N-glycans from glycoproteins for eventual degradation through ER-associated degradation (ERAD).^[6] N-glycosylation has been linked in multiple studies to the normal processes of neural transmission including synaptic transmission through the neuromuscular junction, neuronal voltage gated ion channel regulation, and development of the nervous system.^[7] Further, impairments to deglycosylation and ERAD disruption have been linked to neurodegenerative diseases such as ALS and Alzheimer’s disease, the latter of which is characterized by extracellular accumulation and aggregation amyloid beta and tau proteins.^[8] Little is known about the mechanism of NGLY1 or how its inactivation drives its deficiencies’ varied symptoms, however a study conducted by Fujihira et al. has implicated the activity of another de-N-glycosylating enzyme, endo-β-N-acetylglucosaminidase (ENGase), in the progression of this disease.^[9] The glycosidases ENGase and mannosidase alpha class 2C member 1 (Man2C1) process cytoplasmic N-glycans after they are cleaved from misfolded N-glycoproteins by NGLY1. In the absence of NGLY1, ENGase will directly target N-acetylglucosamine (GlcNAc) residues attached to the glycanated protein leaving a single GlcNAc residue which disrupts ERAD and leads to cytoplasmic accumulation and aggregation of misfolded N-GlcNAc proteins.^[9–10]

NGLY1 deficiency is an exceedingly rare disease and as of 2018 there were only 48 reported cases worldwide based on genotypic analysis and clinical phenotype, with only 18 cases described in the literature.^{[5a, 11]} NGLY1 is ubiquitously expressed among the parenchyma of most major organs, and as such a biallelic loss of function mutation manifests as a range of clinical phenotypes ranging from neonatal jaundice, elevation of liver transaminases, and liver fibrosis to neurological symptoms such as hypotonia, movement disorder, alacrima, hypoalacrima, ocular apraxia, and peripheral neuropathy.^{[11a, 12]} The symptoms and their severity vary from patient to patient, but clinical examination of two patients with confirmed NGLY1 deficiencies demonstrate progressive neuronal impairment and rapid onset of axon loss.^[12] These findings allowed researchers to hypothesize that NGLY1 deficiency negatively affects neuronal cells’ ability to properly respond to environmental stress as a result of excessive cytoplasmic protein accumulation, leading to progressive loss of neurological function.^{[11b, 13]} With the relative obscurity of the disorder and the apparent phenotypic commonalities between NGLY1 deficiency and neurodegenerative diseases, it raises the possibility that some early onset diagnoses of ALS or SMA may, in fact, be misdiagnoses. This could be problematic in the determination of appropriate treatment.

A significant barrier to the study of NGLY1 deficiency is the scarcity of patients in which the disorder can be studied and through which a potential treatment could be developed via their participation in standard clinical trials. Murine models of NGLY1 deficiency have been developed that faithfully replicate many aspects of the disorder.^{[9, 13]} However, due to the myriad of physiological differences between diseased humans and knockout animal models there is has been markedly poor translatability from human to animal studies.

Recent studies approximate that 89% of all novel drugs in clinical trials with the proof of concept (POC) performed in animal models, have failed in clinical trials, with a concomitant increase in drug development costs.^[14] Such a high failure rate is unsustainable for the long-term development of new therapeutics, especially for rare diseases where there is limited economic incentives compared to other therapeutic targets. The development of phenotypic human cell-based models in which specific cell functions and patient genetic background can be studied for the discovery of novel treatments without known molecular targets can assist in overcoming some of the difficulties in rare disease therapeutic development. Induced pluripotent stem cell (iPSC) technology is the basis of patient-specific phenotypic disease models for studying the progression of neurological diseases in vitro, as they possess nearly limitless potential for expansion and differentiation into a multitude of cell types. By utilizing one of many non-integrating methods of gene delivery into somatic cells, researchers can reprogram somatic cells from patients with congenital disorders into proliferative, pluripotent stem cells that maintain the patients’ genetic signature.^[15] These iPSCs can then be differentiated down a neuronal lineage for the purpose of building a personalized model of neurological disease that can specifically replicate the pathology associated with a patient’s condition.^[16] This is especially relevant for rare congenital disorders, wherein a personalized in vitro model containing patients’ cells could be the best avenue available to study the disease pathophysiology and to evaluate the drug efficacy. In addition, the CRISPR technology now allows for knockouts or mutation correction to be developed as a control supplementing the patient-cell based studies.^[17]

In this study, we present a human iPSC-based neuromuscular junction (NMJ) coculture model comprised of primary SkM and MNs derived from dermal fibroblasts of patients carrying biallelic mutations in the NGLY1 gene. The NMJ coculture platform features two adjacent, electrically isolated cellular compartments connected by an array of microtunnels.^[18] These microtunnels have been microfabricated with dimensions that allow for axon extension from the neuron to the muscle side, while maintaining electrical and physical isolation between the compartments. This allows for independent broad field electrical stimulation and interrogation of each cell type. We also describe a method to maintain the pluripotency of multiple NGLY1-deficient patient-derived iPSCs in culture as well as a guided differentiation of WT and NGLY1 deficient iPSCs into spinal cord MNs. Through the use of this NMJ coculture system, we demonstrate that these human iPSC-derived motoneurons (iPSC-hMNs) possess electrophysiological properties characteristic of spinal cord MNs and are capable of developing functional NMJs when cocultured with fused SkM myotubes. Indirect and direct electrical interrogation of these NMJ yields clinically relevant data that can be used to investigate the efficacy and toxicity of novel therapeutics not only for NGLY1 deficiency, but other deglycosylating disorders with associated neural deficits.

2. Results

2.1. Differentiation of NGLY1-Deficient iPSCs into iPSC-hMNs

Induced pluripotent stem cells (iPSCs) were generated previously from the dermal fibroblasts of healthy and NGLY1 deficient patients^[19] as well as a line of iPSCs that had been CRISPR modified to possess a homozygous knockout of the NGLY1 gene. In the context of this paper, the lines will be referred to as 268A (WT), X2–9 (Isogenic KO), 592D (Patient 1), and 594A (Patient 2). 592D Patient 1 iPSCs were derived from dermal fibroblasts isolated from a 16 year old female patient with NGLY1 deficiency (Coriell GM26612) and possess the homozygous mutation c.1201A>T (p.Arg401X) in the NGLY1 gene.^[19b] Patient 2 594A iPSCs were derived from dermal fibroblasts sampled from a 2-year-old male with NGLY1 deficiency (Coriell GM26602) and possess compound heterozygous mutations L318P and R390P in the NGLY1 gene.

After multiple sequential passages using the methods described in the Materials section we were able to adapt each line to a similar culture criteria and cryopreserve each iPSC line en masse. Both WT and diseased iPSCs adopted typical stem cell morphology as determined by phase contrast microscopy, wherein the cells proliferated as adherent colonies with rounded edges, multinucleated proliferative individual cells, and tight compaction in the colony centers (Figure 1a). There was one exception to this, as the 594A IPSCs had a greater predilection towards spontaneous differentiation and detachment following cryopreservation recovery and passage. Immunocytochemical analysis of readily proliferating and morphologically-normal stem cell colonies indicated that each line was positive for NANOG, OCT4, and SOX2, three characteristic markers of pluripotency (Figure 1b–c). G-band karyotype analysis indicated that none of the four lines harbored clonal abnormalities and were karyotypically normal at the time of cryopreservation (Figure S1).

Figure 1.

Morphological characterization of WT and diseased induced pluripotent stem cells (iPSCs). (a) Phase images of large iPSC colonies captured at ~80% culture confluency, typically occurring at day 3 post-plating. Scale bar = 100 μm. (b,c) Immunocytochemistry for characteristic markers of stem-ness OCT4, NANOG, and SOX2 in iPSC cultures fixed at ~80% confluency. Scale bar = 20 μm.

Motoneuron differentiation was initiated on morphologically normal cultures of these iPSCs nearing the end of log phase growth. The differentiation protocol employed was based on a previously published protocol with minor alterations^{[18, 20]} and yielded functionally active hMNs after 26 days of differentiation. These terminally-differentiated cells primarily stained positive for multiple proteins characteristic of spinal cord MNs. Positive stains for microtubule-associated protein 2 (MAP2) and the LIM homeodomain transcription factor Islet 1 (Islet1) at days 10 and 14 indicated the specialization into spinal cord MNs that possessed sufficient axon extension capabilities^[21] (Figure 2a). Additionally, all four iPSC-hMN lines stained positive for non-phosphorylated neurofilament (SMI-32) and choline acetyltransferase (ChAT), further indicating these cells’ differentiation into cholinergic α-MNs^[22] (Figure 2b). Quantification of immunocytochemical information can be found in Figure S2.

Figure 2.

Immunocytochemical characterization of terminally differentiated WT and diseased iPSC-derived MNs (a,b) Immunocytochemical analysis for iPSC-hMN expression of characteristic MN markers MAP2, Islet1/2, ChAT, and SMI-32. Scale bars = 10 μm.

2.2. Morphological Characterization of NGLY1 Deficient iPSC-hMNs

Potential occurrences of axon regression was investigated by analyzing high-contrast phase microscopy images taken at post-plating days 1, 6, 10, and 14 to derive the number of axon branches per micron, mean length of axon segments, and the number of varicosities per micron (Figure 3a). 268A hMNs maintained steady rates of axon branching at ~0.014 branches/μm (b/μm) early in their maturation, however progressive pruning events were noted at days 10 and 14 with the final branching value at 0.050 b/μm. 592D hMNs exhibited high branching after one day in culture, but underwent significant pruning events later in culture to drop to match the 268A hMNs at 0.014 b/μm on day 6. Relative increases and decreases in branching at day 10 by 592D and 594A hMNs, respectively, indicated lower and higher pruning capabilities in each line (Figure 3b). Average segment length increased steadily in WT neurons over time, beginning at 30 μm on day 1 and increasing daily to 67 μm by day 14. This indicated both that the axons were extending and that mid-segment branching decreased over time, a characteristic normally observed in neurite maturation in healthy hMNs. Mean segment length was found to be comparatively lower in all three diseased hMNs at various points in their maturation with little overlap with the respective rates of axon branching (Figure 3c). Average number of varicosities in 268A hMNs varied slightly by day in vitro (DIV) but trended downward from 0.016 varicosities/μm (va/μm) to 0.006 va/μm. X2–9 hMNs repeated this downward trend with statistically insignificant variations noted at day 10. 592D hMNs trended higher than both 268A and X2–9 hMNs, remaining relatively steady at a count of 0.014 va/μm to 0.012 va/μm. Most interestingly, 594A hMNs maintained the lowest amount of varicosity formation, dropping from 0.020 va/μm on day 1 to 0.035 from days 6–14 (Figure 3d).

Figure 3.

Morphological characterization of terminally differentiated WT and diseased iPSC-derived MNs by phase microscopy. (a) Representative phase contrast image of 268A WT iPSC-hMN neurite tracing as well as representative selections of neurite varicosity (yellow arrows, dots) and branch (red arrows, dots) selection. (b) Neurite branching, (c) mean length, and (d) varicosity quantification for each iPSC-hMN line as observed from phase contrast images taken at post-plating days 1, 6, 10, and 14. N = 3. Data points are mean ± SEM, * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.

2.3. Electrophysiological Characterization of NGLY1 Deficient iPSC-hMNs

In vitro studies of neurodegeneration have traditionally included patch clamp electrophysiological readouts in their evaluation of a singular neuron’s ability to regulate charged ion flow in and out of the cell in the process of AP generation. Whole-cell patch clamp techniques were used to quantify intracellular flux of sodium and potassium ions, as well as the ability of the neurons to generate APs, the amplitude of the APs, and the rate at which spontaneous APs were generated when clamped at a consistent voltage. Representative traces recorded from 268A WT hMNs are depicted in Figure 4a, and representative traces from diseased hMNs can be found in Figure S3.

Figure 4.

Electrophysiological characterization of WT and diseased iPSC-hMNs via patch clamp. (a) Representative patch clamp traces of WT iPSC-hMNs, including Na⁺/K⁺ currents, induced APs, spontaneous APs, and baseline membrane potentials. (b-e) Quantification of b) inward Na⁺ and outward K⁺ flow, c) spontaneous APs as the number of APs per second, d) resting membrane potentials, and e) action potential amplitudes for each iPSC-hMN line at post-differentiation day 14. N = 5–10. Data points are mean ± SEM, * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.

Evoked APs from 268A hMNs revealed average inward Na⁺ currents at 1500 pA and average outward K⁺ currents at 1100 pA. There were no statistically significant variations in K⁺ flux between hMN lines, however inward Na⁺ was restricted in X2–9 isogenic KO hMNs and in the patient-derived 592D hMNs with currents of −760 pA and −800 pA, respectively (Figure 4b). The most notable differences in electrophysiological activity of these hMNs was with spontaneous AP generation. 268A hMNs were observed to fire at an average rate of 4.5 AP/s, while diseased hMNs all fired at a significantly slower rate, 0.5 – 1.5 AP/s (Figure 4c). The resting membrane potential of WT hMNs was characteristic of normal hMNs at −66 mV. NGLY1-deficient patient-derived hMNs both had elevated membrane potentials, up to −49 mV for 592D hMNs and −55 mV for 594A hMNs (Figure 4d). Action potential amplitude varied between lines, with X2–9 hMNs exhibiting a 20% decrease in amplitude in comparison to WT 268A hMNs (Figure 4e).

Interestingly, aberrant patterns of spontaneous bursting behavior were observed in each line of NGLY1-deficient hMNs and to some degree in WT hMNs (Figure 5a). These bursting patterns are identified as multiple APs firing in a short timeframe (0.5 – 1 s) in a semi-consistent manner and have been associated with disease phenotypes.^[23] Secondary characteristics of these burst patterns could include incomplete depolarization to baseline and decreasing AP amplitude with each successive firing. These abnormal bursting patterns could only be observed in cells that fired action potentials absent exogenous stimulation, therefore the percentage of spontaneously active cells from each hMN line was quantified; 100% of all 268A WT hMNs, 71% of X2–9 hMNs, 50% of 592D hMNs, and 60% 594A hMNs demonstrated an ability to spontaneously fire APs (Figure 5b). Burst firing was independent of the line’s ability to generate spontaneous APs, with lowest incidence rates in the 268A WT hMNs at 13% and highest in the X2–9 KO hMNs at 43%. Incidence rates of burst firing in patient derived 592D and 594A hMNs was 13% and 20%, respectively (Figure 5c).

Figure 5.

Bursting behavior in patched WT and NGLY1-deficient IPSC-motoneurons. (a) Representative spontaneous action potential traces from 268A, X2–9, 592D, and 594A iPSC-MNs depicting differing degrees of burst patterns in WT and NGLY1-deficient hMNs. (b) The percentage of patched motoneurons that adopt spontaneous burst firing patterns. (c) Percentage of patched motoneurons from each line that fire spontaneous action potentials at any detectable frequency. R = 5 – 12, data points mean ± SEM, * p ≤ 0.05.

2.4. Effect of NGLY1 Deficiency on Ion Transporter Expression

The Na-K-Cl cotransporter 1 (NKCC1) is one of multiple transmembrane proteins in the family of cation-chloride cotransporters responsible for the transport of Cl⁻ with K⁺ or Na⁺ across the plasma membrane and are integral to the mediation of intracellular chloride concentrations and maintenance of cell volume. NKCC1 in particular has been demonstrated to play a critical role in the facilitation of high intracellular uptake of Cl⁻ ions in damaged peripheral neurons, a state that drives axon growth and recovery from axonal injury.^[24] Following its synthesis and assembly, NKCC1 undergoes rapid, complex N-glycosylation, a process that has been hypothesized to be necessary for normal co-transport function due to the prevalence of core/high-mannose type glycosylated NKCC1 at the plasma membrane, although the effects of N-glycosylation on the plasma membrane is currently unknown.^[25] NGLY1, or PNGase, is a de-N-glycosylating enzyme that removes N-linked or asparagine-linked glycans (N-glycans) from glycoproteins. A recent report demonstrated that N-glycosylated NKCC1 is a substrate of NGLY1.^[26] Due to NKCC1’s apparent need for N-glycosylation to maintain normal intracellular ion concentrations, as well as the observed electrophysiological deficits in diseased hMNs, NGLY1 deficient hMNs were investigated for their overall expression of NKCC1. Any potential disturbance of the balance between glycosylation and de-glycosylation could alter the subcellular distribution and function of NKCC1 which would drive the observed electrophysiological deficits by preventing adequate intracellular ion concentrations from being established, inhibit plasma membrane targeting, or through other unforeseen issues associated with misfolded protein aggregation.

Initial investigation of the mature iPSC-hMNs indicated expression of NKCC1 by all four lines (Figure 6a). Based on the distributions reported in the literature, 90% of NKCC1 is localized to intracellular compartments while only 10% is located on the plasma membrane.^[25] The uniform staining observed in the cultures appears to align with this distribution. Diminished amounts of detectable NKCC1 were evident in western blots of differentiated iPSC-hMNs by thin, lightly colored blots at 165 kDa in diseased (X2–9, 592D, and 594A) iPSC-hMN lanes in comparison to the much darker and wider blot in the WT 268A iPSC-hMN lane (Figure 6b). Complete images of western blot and total protein stained gels can be found in Figure S5. The western blots were conducted on whole cell lysates of iPSC-hMNs, as any insoluble proteins or membrane-bound proteins would have been removed in the process of centrifugation and lysate preparation. These data suggest that there was less soluble NKCC1 protein in the supernatant of the lysates obtained from diseased iPSC-hMNs, potentially due to incomplete de-glycosylation of NKCC1 in NGLY1-deficient hMNs which affects their subcellular distribution, membrane binding and/or solubility.

Figure 6.

Aberrant glycosylation of Na-K-Cl cotransporter (NKCC1) in iPSC-hMNs. (a) Immunocytochemical analysis of each iPSC-hMN line indicated strong expression of NKCC1 in WT and diseased lines. Scale bar = 10 μm. (b) Western blot analysis of NKCC1 indicates aberrant de-glycosylation of NKCC1 as evidenced by diminished stains in the NGLY1 KO and diseased lanes (X2–9, 592D, 594A) in comparison to the 268A WT isolates. N = 4. Data points are mean ± SEM, **** p ≤ 0.0001.

2.5. Characterization of NGLY1-deficient NMJ Coculture Systems

The NMJ systems utilized in this study have been previously described for the investigation of neurodegenerative diseases and evaluation of the compounds in a dose-dependent manner.^{[18, 20b]} Improvements to the preparation of these systems as outlined in the Methods section have increased the batch to batch consistency from previous iterations. These systems maintain physical and electrical isolation between hMN and SkM compartments while still allowing for axonal penetration through the microfabricated microtunnels and subsequent NMJ formation. Optimal plating timelines, culture time, and recording days were determined experimentally and were impacted by the SkM contractility, hMN maturity, and NMJ formation.

Primary WT skeletal myoblasts were seeded into one chamber, cultured for 48–72 hours depending on myoblast confluency, then switched to differentiation medium to facilitate fusion and myotube formation. Initiation of myoblast fusion was denoted as “NMJ Day 0” and served as the reference point around which hMN seeding and NMJ recordings were planned. NGLY1 deficient iPSC-hMNs were then seeded on NMJ day 2 and the resulting cocultures were monitored daily for axon penetration and innervated myotube morphology until NMJ interrogation on NMJ Day 12 (Figure 7a). The resulting NMJ coculture consisted of fused myotubes innervated by axons extending from iPSC-hMNs in a separate, chemically and electrically isolated compartment that contained electrodes necessary for broad field stimulation of the neurons to drive indirect contraction of the myotubes in the adjacent chamber (Figure 7b). NMJ formation in the NGLY1 deficient coculture systems was investigated using immunocytochemistry at NMJ Day 12 to visualize colocalization of presynaptic hMN terminals and Ach clustering by staining for NF-H, synaptophysin, and α-bungarotoxin (Figure 7c–d). Characteristic NMJ morphologies were observed with each line at NMJ Day 12, signifying each iPSC-hMN line’s capability to form NMJs in this coculture system.

Figure 7.

NMJ On-A-Chip Coculture chamber. (a) Differentiation and plating timeline to achieve iPSC-hMN induction and subsequent NMJ formation in our NMJ coculture chambers. (b) Schematic of an NMJ coculture chamber, depicting the physically and electrically separated cellular compartments, innervation of SkM by iPSC-hMNs, and electrodes used for broad-field electrical stimulation of iPSC-hMNs during functional evaluation. Phase images depict typical morphology of iPSC-hMNs (left) and innervated primary SkM (right). Arrows indicate contact points between axons and SkM, signaling the potential presence of NMJs. Scale bars = 20 μm. (c-d) Immunocytochemical analysis of NMJs formed between iPSC-hMNs and primary SkM, as depicted by characteristic staining patterns for b) α-bungarotoxin with neurofilament-H and c) α-bungarotoxin with synaptophysin. Scale bars = 10 μm.

2.6. Functional Analysis of NGLY1 Deficient NMJs

In order to evaluate the contractility and function of NMJs in this system, we designed and implemented a custom, multi-ROI image acquisition platform that allowed for the simultaneous interrogation of four NMJs simultaneously in one platform. Since each NMJ is a functionally distinct unit, this allowed the potential to quadruple the replicates obtained per experiment without requiring the culture of additional systems (Figure 8a). Of the observed myotubes that contracted with indirect stimulation, up to four NMJs in a single frame were selected at random and their X/Y coordinates were logged as distinct ROIs for continuous monitoring during stimulation. The method of testing utilized in this study employs image subtraction software to capture the planar movement of contractile myotubes following their direct (SkM myotube stimulation) or indirect (MN stimulation of SkM myotubes) stimulation. NMJ number, stability, fidelity, and the percentage of ROIs in tetanus were derived from the data obtained using this recording method.

Figure 8.

Functional evaluation of NMJs using a custom-designed, multi-ROI image acquisition software. (a) Representative image depicting the ROI selection and image subtraction panes used to select and simultaneously record the contractility of multiple NMJs. (b) Number of observable NMJs present in each chamber prior to electrical interrogation. (c) NMJ stability represented as a percentage of NMJs present following direct and indirect stimulation. (d) NMJ chamber failure as a percentage of chambers that failed to form functional NMJs by the optimal testing date (NMJ day 12) in relation to the total number of chambers plated for each line. (e) Fidelity of NMJs quantified as the percentage of successful SkM contractions following direct and indirect iPSC-hMN stimulation at 0.33 Hz, 0.5 Hz, 1.0 Hz, 2.0 Hz and 4.0 Hz. (f) Percentage of ROIs in tetanus with direct and indirect stimulation at 2.0 Hz and 4.0 Hz. N’s of 6–8 for NMJ stability and number, 8–9 for NMJ failure rate, and 4–9 for NMJ fidelity and tetanus. Data points are mean ± SEM, * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001.

Prior to intensive stimulation and testing, the hMNs were pulsed once to drive indirect SkM contractions and the number of observable NMJs in each system was recorded. This was performed with each system tested and at the end of each experiment all recorded values were averaged and charted. It was found that the WT NMJ systems formed, on average, four NMJs while NGLY1 deficient NMJ systems formed two (Figure 8b). NMJ number was again counted after stimulating with the full direct and indirect stimulation sweep and was divided by the number acquired pre-stimulation. This ratio is depicted as NMJ stability. Following indirect and direct stimulation, 60% of the NMJs in 268A systems maintained their function whereas systems seeded with X2–9, 592D, and 594A iPSC-hMNs retained 35%, 26%, and 14% of their functional NMJs, respectively (Figure 8C). NMJ number is entirely dependent on the number of NMJs that are formed from NMJ Day 2, the day of iPSC-hMNs seeding, to NMJ Day 12, the day on which they were evaluated for contractility. In some cases, however, no functional NMJs were observed pre- or post-stimulation even with visible axons in the SkM compartment. Of all systems seeded and assembled, only 29% of 268A systems failed to form NMJs while that number trended upward from 47% to 63% in NMJ systems seeded with NGLY1 deficient hMNs (Figure 8d).

NMJ fidelity is the metric that is most analogous to clinical evaluations of the neuromuscular junction in vivo. To investigate this parameter, recordings of each system’s contractile response to indirect hMN stimulation at 0.33 Hz, 0.5 Hz, 1.0 Hz, 2.0 Hz, and 4.0 Hz was obtained followed by direct SkM stimulation at the same frequencies as a control. Each frequency was pulsed 12X for 50 ms each. The number of pulses for which synchronous SkM contraction was observed was recorded and divided by the total number of stimulation pulses for each ROI in each system. With direct stimulation, 268A NMJs had high fidelity while X2–9 NMJs were significantly lower in fidelity at 0.5 Hz - 4.0 Hz. NMJs from 268A iPSC-hMN systems maintained a near 100% fidelity when stimulated indirectly at each frequency. X2–9 NMJs consistently expressed a lower indirect fidelity, ranging from 69% at lower frequencies to 39% at higher frequencies. 592D NMJs ranged from 74% at lower frequencies to 54% at higher frequencies, and 594A NMJs ranged from 73% at slower frequencies to 53% at faster frequencies (Figure 8e).

At higher stimulation frequencies the SkM was often driven into either fused or unfused tetanus, a direct result of the SkM not being able to relax completely between stimulated contractions. This led to a characteristic peak shape and was marked as 100% fidelity. As a result of this, many of the 2.0 Hz, and 4.0 Hz traces from diseased NMJs are depicted as having high fidelities according to the metrics analyzed in this study. In order to address this, the percentage of ROIs in tetanus with each stimulation frequency were tallied. A progressive increase in the percentage of ROIs in tetanus was observed with direct SkM stimulation, however myotubes innervated with X2–9 iPSC-hMNs entered into tetanus less often than NMJs from WT 268A iPSC-hMNs. Notably, the percentage of ROIs in tetanus with indirect stimulation was decreased significantly in NGLY1-deficient NMJs, further indicative of motor deficits in diseased NMJs. (Figure 8f).

3.0. Discussion

NGLY1 deficiency is caused by mutations in the NGLY1 gene, resulting in the deficiency of deglycosylation of N-linked glycoproteins. However, the precise mechanism of disease pathology is unclear and disease phenotype in the patient cells is still not characterized. We have developed and described a compartmentalized, microfabricated NMJ system that allows for the coculture of primary SkM and NGLY1 deficient hMNs in a single device. Each chamber of the system was designed to be connected by a multitude of parallel microtunnels that are large enough to facilitate the projection of axons through each microtunnel while maintaining chemical and electrical isolation between cellular compartments. This system was used to investigate the neuromuscular deficits associated with the progressive neuronal impairment experienced by patients who suffer from the deglycosylation disorder NGLY1 deficiency. NGLY1 deficient hMNs and an isogenic KO utilized in this study were differentiated from reprogrammed dermal fibroblast iPSCs from patients with confirmed LOF mutations in NGLY1 and were confirmed to possess morphological and electrophysiological deficits in MN monoculture. Additionally, numerous electrophysiological and contractile deficits were observed in the NMJs formed by these hMNs that correspond with the clinical manifestations of this disease.

Analysis of individual case reports of patients with NGLY1 deficiency have pinpointed shared clinical pathologies of disease. In addition to hypotonia and liver disease, neuronal and musculoskeletal impairments were prevalent among this population, as was global development delay, hypotonia, abnormal electroencephalography (EEG) results, movement disorders, and diminished reflexes.^[11a] Analogous neuromuscular pathologies were reported in the patients from whose cellular material the iPSCs were generated from. Patient 1, from whom the 592D iPSCs were derived, possessed a homozygous LOF mutation in the NGLY1 gene and exhibited muscular hypotonia, decreased strength, and axonal neuropathy as well as broadly described developmental delay, motor, and speech impairments. Patient 2, the source for the 594A iPSCs, possessed compound heterozygous mutations at multiple loci in the NGLY1 gene. He also exhibited hypotonia, a movement disorder, and delayed speech development in addition to ataxia, chorea, and a slower than normal rate of myelination. We theorized and ultimately observed that while the two patients may manifest similar clinical pathologies, when isolated into their own respective microphysiological NMJ systems the differences in LOF mutations will drive deficits of differing degrees.

The iPSCs derived from patients with NGLY1 deficiency were confirmed to both retain their pluripotency and remain karyotypically normal after multiple consecutive passages, which was interesting due to the developmental impairments associated with LOF mutations in NGLY1. The multi-stage hMN induction protocol was successful in facilitating the differentiation of the iPSCs into spinal cord MNs, which was confirmed by immunocytochemical staining with multiple MN-specific markers. Morphological analysis of the MN neural processes (neurites) evaluated characteristics of the MN length and patterns of axon arborization. During development, presynaptic neurons extend axons and undergo a series of branching steps and continued axonal lengthening to form connections with postsynaptic cells such as SkM. When sufficient levels of contact have been made neurons will undergo axon pruning that allows for the removal of misguided axon branches without death of the individual cell.^[27] Late stage neurodegenerative disorders such as ALS and Alzheimer’s have been associated with impairments in anterograde mitochondrial transport that are thought to drive “dying back” events and subsequent degradation of neuromuscular junction function.^[28] Impairments in retrograde or anterograde transport of proteins and mitochondria can be physically observed as localized swelling on developing axons. These swelling points are herein referred to as varicosities and have previously been linked to further deficits in axonal transport.^{[20b, 29]}

We theorized that if similar axon retraction and swelling events were occurring with diseased NGLY1 deficient iPSC-hMNs it would point to a shared phenotype between NGLY1 deficiency and other neurodegenerative diseases. The gradual increases observed in mean axonal segment length in conjunction with the decreases in branching indicate some degree of normal axonal growth and pruning in each line. However, NGLY1-deficient patient-derived 592D iPSC-hMNs maintained a consistently shorter mean axon length with increased levels of branching, suggesting a deficit in the neurons’ pruning efficiency. Mean axon length at day 14 for NGLY1-deficient 594A, X2–9, and 592D iPSC-hMNs was also observed to be significantly lower than the WT 268A hMNs at the same day with no corresponding elevation in the number of branches. Taken together these data point to an axonal retraction event akin to that observed in neurodegenerative diseases and in the clinical cases reported by Lam et al. Patient-derived 592D iPSC-hMNs also possessed a significantly higher number of axonal varicosities at day 14, indicating potential deficits in axonal transport. Interestingly, the number of axonal varicosities observed in patient-derived 594A iPSC-hMNs was significantly lower than those in the WT 268A iPSC-hMNs. This would indicate that this line had a lesser degree of aberrant axonal transport, however the functional ramifications of this require further investigation.

Patch clamp electrophysiology was employed to investigate if ion channel dysregulation was present in the differentiated NGLY1-deficient iPSC-hMNs, as it is a method commonly utilized to determine severity and progression of neurodegenerative diseases.^[30] Striking differences were observed between healthy and diseased iPSC-hMNs, primarily in the ability of the diseased hMNs to properly regulate intracellular Na⁺ levels and with diminished AP generation. Many NGLY1-deficient patients experience epileptic events with myoclonic absences, such as staring and tonic events caused by brief cessations in muscular activity.^[31] The increased incidence rates of burst firing observed in two of the three studied NGLY1-deficient iPSC-hMN lines could correlate with hyperkinetic and myoclonic movement observed in vivo. Similar patterns of neural bursting are normally found in neurons derived from the cerebral cortex, hippocampus, or brainstem. Neural bursting can occur in spinal cord motoneurons in primarily network sourced events. However, network-independent motoneuron bursting has been reported in spinal cord motoneurons in experiments utilizing penicillin-induced epileptiform burst analysis.^[32] Generally, intrinsic bursting is caused by a disbalance between inward calcium conductance and outward potassium conductance, and should alter voluntary muscle contraction behavior in the context of a functional NMJ.^[33] Considering that complex movement disorders have been observed in the majority of NGLY1-deficiency cases, the bursting observed in these NGLY1-deficient neurons could be a major phenotype for this disease, as normal MN activity is necessary for consistent and error-free transmission of contractile signals to SkM through the NMJ. Furthermore, reduced and infrequent rates of spontaneous AP generation and decreased AP amplitude in the X2–9 iPSC-hMNs foreshadowed the clinically relevant contractile deficits observed in our NMJ systems. An additional clinical manifestation of NGLY1-deficiency is global developmental delay. A metric that is commonly used to gauge hMN maturity in vitro is resting membrane potential, and both patient-derived iPSC-hMN lines utilized in this study had elevated resting membrane potentials in comparison to WT hMNs.

One of the few confirmed substrates of NGLY1 deglycosylation is the NKCC1 protein, which functions as a primary regulator of intracellular ion concentrations, a factor that will negatively affect action potential generation and propagation in neurons lacking the functional protein.^{[24, 26, 34]} Western blot and immunocytochemical analysis for NKCC1 expression indicated a reduction in the levels of NKCC1 in diseased iPSC-hMNs, which could explain the deficits in diseased electrophysiological profiles as NKCC1 activity has previously been demonstrated to be reliant on proper regulation of N-glycosylation of the intracellular domains of NKCC1.^[35] These findings suggest a potential delay in the maturation of diseased hMNs and that further maturation could correct the phenotype, otherwise it indicates the presence of further neurological deficit, however additional experimentation is required to elucidate the cause and ramifications of this decreased expression.

Utilization of the NMJ coculture system allowed investigation of the iPSC-hMNs beyond the limits of single neuron electrophysiological evaluation and was used to directly investigate the capability of these neurons to drive SkM contraction through hMN stimulation. The dual-chamber NMJ platform developed has previously demonstrated its capacity for drug evaluation by the modeling amyotrophic lateral sclerosis (ALS) utilizing patient-derived iPSCs.^[18,20b] A recent on-chip 3D NMJ model utilized MN spheroids and muscle bundles in the respective chambers, where the MNs were excited through optogenetic stimulation and the contraction of the SkM bundles were captured by recording pillar displacement.^[36] This system demonstrated comparable competency in drug evaluation and disease modeling to other systems. Other systems have been developed to model NMJs using optogentics^[37] and chemical stimulation.^[38] In comparison to these systems, the platform described here activated MNs through electrical stimulation so there was no need for genetic modifications for optogenetic labeling of the iPSCs, while allowing for precise frequency control. Compared to utilizing 3D muscle bundles, it also allows the discerning of individual myofibers for their innervation and contraction behavior, which would be desired when studying muscle subtype-specific characteristics in disease modeling. The dual chamber system also allows separate drug dosing of the SkM and MNs

The readouts obtained with the platform are representative of the varied disorders observed in NGLY1-deficient patients. Hypotonia is an observed pathology of NGLY1 deficiency and the source of this disorder is multifactorial, however the secretion of neurotrophic factors by spinal cord MNs during development contribute to the differentiation of myotubes and maintenance of the NMJ in vivo.^[39] Muscular development is stalled, and in some cases reversed, if secretion or binding of these trophic factors is inhibited early in development. Additionally, the continual presence of these trophic factors is required for maintenance and persistence of functional NMJs. Downward trends in NMJ number, stability, and NMJ formation rate observed in the NGLY1-deficient NMJs point towards the contribution of diseased hMNs to impairment of NMJ formation and persistence in post-stimulation function. Whether this contribution is a result of decreased trophic support or reduced electrical activity is unclear and is a potential topic of future study.

In addition to those discussed in previous sections, common clinical pathologies of NGLY1 include asynchronous myotonic jerking of the upper extremities that can either be caused by sudden muscle contraction with positive myoclonus or brief periods of muscle relaxation in negative myoclonus. Choreoathetotic movements of the hands and fingers have also been reported in patients with NGLY1 deficiency. The NMJ fidelity metric reported in Figure 8e is representative of the synchronicity of myotubes to MN stimulation delivered by way of the innervating axons. Impaired synchronicity observed in the diseased hMNs at increasing frequencies in comparison to the high synchronicity observed with the WT hMNs suggests the entry of the innervated SkM into a negative myoclonic state, wherein the diseased hMNs are unable to potentiate signals with complete fidelity, potentially as a result of the electrophysiological deficits observed in the patch-clamp electrophysiological examination. Interestingly, direct NMJ fidelity in NMJs containing the isogenic NGLY1 knockout was significantly lower than that of the WT or patient-derived lines. This is another indication that this diseased line of iPSC-hMNs may slow or inhibit complete and effective formation of NMJs. Additionally, NGLY1-deficient neurons exhibited deficits in their ability to sustain tetanic contractile profiles in comparison to WT neurons, suggesting a link between NGLY1-deficienct hMNs and increased fatiguability of the SkMs they innervate. Investigation of fatigue index with prolonged stimulation will be performed in future experiments. WT SkM was utilized in these NMJ systems to directly investigate the contribution of NGLY1 deficient neurons to NMJ dysfunction, however in NGLY1-deficient patients NMJs would be comprised of diseased MNs and SkM. Now that a link between NGLY1-deficient MNs and NMJ dysfunction has been established, future experiments with NGLY1-deficient SkM can be conducted to achieve a more physiologically relevant model for mechanistic studies.

Taken together, the data presented in this study demonstrates that NGLY1-deficient iPSCs can successfully be differentiated into functional MNs that recreate certain neurological phenotypes observed in patients with NGLY1 deficiency on a cellular level. Reduced axon length, increased and shortened axon branches, increased MN bursting, decreased action potential firing rate and amplitude were observed in the NGLY1-deficient MNs in monoculture. The NGLY1 iPSC-hMNs can then be cocultured with primary SkM in an NMJ system, where the functional deficits observed correlate with neuronal and neuromuscular disorders seen in vivo. When transitioned to the NMJ coculture system deficits in NMJ number, stability, failure rate, and synchronicity with indirect SkM stimulation were observed. The NMJ platform can be used to better understand the effect of deglycosylation disorders and impaired ERAD on the development and function of the NMJ, as well as the screening of potential therapeutics for the treatment of neuromuscular deficiencies in a phenotypic model of patients with deglycosylation disorders or other congenital rare diseases.

4.0. Conclusion

This work presents an iPSC-derived NGLY1-NMJ platform for the phenotypic investigation of the NGLY1 congenital disorder. Motoneurons differentiated from patient-derived or gene edited iPSC lines harboring NGLY1 gene mutations demonstrated altered electrophysiological properties such as reduced spontaneous activity and more prominent bursting behavior. These changes can be at least partially attributed to the aberrant de-glycosylation of NKCC1 protein in the NGLY1 deficient MNs which was indicated by Western blot analysis. Integration of these MNs into the NMJ platform demonstrated deficient NMJ function, which was manifested as reduced functional NMJ number, reduced NMJ stability and fidelity, but increased NMJ fatigue index. This NGLY1-NMJ platform can be utilized for deeper mechanistic study of this disease as well as for therapeutic screening.

5.0. Experimental Section/Methods

5.1. WT and Diseased Induced Pluripotent Stem Cell Culture

Wild type, isogenic knockout, and two lines of NGLY1 deficient patient-derived induced pluripotent stem cells were provided by the National Center for Advancing Translational Sciences. A generic stem cell culture protocol was used for their culture with some minor variations depending on the optimal growth conditions necessitated by each line. A range of 1 μM to 5 μM Y27632 (Tocris 1254) was utilized when plating, passaging, and cryopreserving these lines, as each individual line had a different tolerance for ROCK inhibition before excessive death or spontaneous differentiation was induced.

Cryopreserved IPSCs were thawed at 37°C, combined with the ROCK inhibitor Y-27632-supplemented mTeSR Plus (StemCell Technologies 05825) and centrifuged at 200 × g for 3 minutes to remove residual cryoprotectant from the medium. Following aspiration of the cryoprotectant-containing medium, intact IPSC colonies were resuspended in mTeSR Plus containing Y-27632 at a pre-determined optimal concentration and were plated onto hESC-qualified matrigel-coated (Corning 354277) wells of a six well plate for feeder-free recovery and expansion. These plates were coated with matrigel according to the manufacturer’s lot specific recommendations. A subsequent full media change was conducted with fresh mTeSR Plus within 16 hours and every day thereafter in order to facilitate rapid recovery and expansion.

IPSCs were passaged when at 70–80% confluency while still in their logarithmic growth phase (4–6 days) using StemCell Technologies’ ReLeSR dissociation reagent (StemCell Technologies 05872) to selectively dissociate undifferentiated stem cells as colonies and passage without centrifugation. These cells were rinsed once with 1X phosphate buffered saline (PBS), then incubated with ReLeSR at 37°C for up to six minutes, or until significant colony detachment was observed. Detached colonies were collected in mTeSR Plus with Y-27632 and dispensed onto matrigel-coated surfaces in a 1:6 to 1:12 ratio by surface area depending on the optimal growth conditions for each line. A full media change was conducted with fresh mTeSR Plus within 16 hours followed by daily full media changes.

Stem cell colonies were cryopreserved at 70–80% confluency using a similar process to that employed during passaging. Following dissociation using ReLeSR, colonies were collected in mTeSR Plus with Y27632 and centrifuged at 200 × g for three minutes. The resulting pellet was gently but quickly resuspended in cold freezing medium containing mTeSR Plus supplemented with 20% Knockout Serum Replacer (KOSR, ThermoFisher 10828028), 10% DMSO, and Y-27632 at 1–5 μM. The resuspended colonies were then transferred to cryovials and frozen at a rate of −1°C per minute before transferring them to liquid nitrogen for long-term storage.

5.2. Differentiation of IPSCs into Mature Motoneurons

Motoneuron induction was performed on each of the four iPSC lines using a slightly modified protocol than the method previously described by Qu et al. [20a] and Guo et al. [20b]. Briefly, iPSCs were plated at a low density of ~20% confluency and cultured to ~90% confluency in mTeSR Plus. At this point the cells were transitioned through each of the four hMN differentiation medias, each for a different culture period (Figure 7A) and with a differing composition (Figure S4): hMN Diff. I for three days, hMN Diff. II for two days, hMN Diff. III for 15 days, and hMN maturation medium for 11 days. On the final day of hMN induction, differentiated hMNs were dissociated using 0.05% Trypsin/EDTA (Sigma-Aldrich 59417C), centrifuged at 270 × g for three minutes, frozen at a rate of −1°C per minute, and stored in liquid nitrogen for long-term storage.

5.3. Patch Clamp Electrophysiology of Motoneurons

Following differentiation and cryopreservation, MNs of each line were plated onto laminin-coated coverslips and cultured to maturity. Whole-cell patch clamp was performed on these MNs at post-plating day 14 according to a method described previously.^{[20b, 40]} Briefly, on the day of recording one of the aforementioned hMN coverslips was transferred to the recording stage of a Zeiss Axioscope 2FS Plus microscope in hMN maturation medium adjusted to an osmolality of 300 mOsm. The surface of the coverslip was inspected for the presence of morphologically typical hMNs. Once a suitable ROI containing one or more healthy hMNs was identified, a borosilicate glass patch clamp pipette (Sutter Instrument Company BF 150-86-10) was pulled to a resistance of 6–10 MΩ using an automated Sutter P97 pipette puller, filled with an intracellular solution, and interrogated according to the method that has been described previously.^[20b]

Data collected during electrophysiological interrogation was analyzed using Axon pCLAMP 10. Action potential amplitude, resting membrane potential, membrane resistance, and spontaneous AP firing rate were obtained from the various stimulation methods employed. Inward and outward currents were obtained from evoked-AP traces, wherein the last five sweeps in the stimulation sequence were selected and the y-axis minimum and maximum points of these traces were identified as the maximum inward sodium and maximum outward potassium currents, respectively. Five to ten biological replicates were utilized in the determination of statistical significance.

5.4. Microfabrication, Assembly, and Coating of the NMJ Coculture system

SU-8 system molds were fabricated according to a previously published protocol with minor alterations.^[18] Briefly, silicon master molds were layered with two coatings of SU-8 2002 (Microchem, MA) utilizing standard photolithography methods. Each NMJ system pattern on the master mold contained two reservoirs (5 mm × 10 mm) separated by an array of 240 microtunnels (200 μM × 10 μm × 3.5 μM) that allow axonal penetration but effectively impede chemical signaling or transmission of electrical impulses from one chamber to the other. A polydimethylsiloxane (PDMS) mixture was prepared using a Sylgard 184^™ Silicone Elastomer Kit (Dow Corning, USA) following the standard 10:1 w/w ratio of monomer to curing agent. A PDMS cast of the NMJ system was produced by pouring the mixture over the SU-8 master mold, degassing at −80 kPa, and heat-curing at 65°C for four hours. The PDMS cast was removed from the SU-8 master mold and cut into individual systems, which were then sterilized in 70% isopropyl alcohol (IPA) for 24 hours, followed by a 48 hour soak in sterile water to leach out unpolymerized monomers and contaminants.

Bonding of PDMS systems to 22 × 22 mm glass coverslips was achieved by exposure to oxygen plasma at 30 W and 600 mTorr for 40 seconds using a plasma cleaner (Model PDC-001, Harrick Plasma). Post-treatment, the systems and coverslips were contacted within a period of 90 seconds and heated at 80°C for 30 minutes to ensure maximum bond strength. Bonded systems were sterilized via germicidal UV-C exposure for 10 minutes. Separate extracellular matrix coatings were applied to the chamber interiors; laminin solution (Thermo Fisher 23017015; 3 μg mL−1) on the MN side and collagen type I solution (Thermo Fisher A1048301; 60 μg mL−1) on the SkM side. The coated systems were incubated at room temperature for 1.5 hours, after which the collagen I coating was removed and replaced with 1X DPBS. The systems were stored at 4°C overnight. After 24 hours, the laminin coating was removed and replaced with 1X DPBS.

5.5. Skeletal Myoblast Culture and NMJ System Coculture

Primary human SkM myoblasts (hSMM) were obtained from Lonza, Allendale, NJ, USA, from a healthy donor (Lonza CC2580, Lot No. 8TL180368), passaged once in serum-free Adult Growth Medium (AGM) [22] (Figure S4c), and cryopreserved at 60 – 70% confluency. Cryopreserved P1 hSMM were thawed and seeded along the microtunnels of the NMJ systems at a density of 300 cells/mm² in AGM. Myoblasts were proliferated to 85% confluency and switched to a serum-free differentiation medium (NBActiv4, Brainbits) supplemented with 1% Antibiotic-Antimycotic (Thermo Fisher 15240–062) to promote fusion into myotubes. Cultures were maintained at 37°C and 5% CO₂ with full medium changes every two days.

Myotube fusion was typically observed after 48 hours of incubation in differentiation medium, at which point differentiated iPSC-hMNs were seeded into the laminin-coated chamber opposite the SkM chamber with hMN maturation medium (Figure S4b). MNs and SkM were allowed to mature in the adjacent chambers for 10 to 12 days, during which the iPSC-hMNs extended axons through microtunnels in the PDMS barrier and innervated the fused myotubes. NMJ systems were maintained at 37°C and in 5% CO₂ with full medium changes using hMN maturation medium and Nba4 every two days, functionally evaluated on NMJ day 12, and fixed for immunocytochemical analysis immediately after testing.

5.6. Functional Interrogation of NMJ Systems

On NMJ Day 12, cultured NMJ systems were assembled into acrylic housings and secured to a testing apparatus fitted with a chlorinated silver wire. Prior to testing, the trans-barrier electrical resistance of each NMJ system was measured using a Voltohmmeter (EVOM2, WPI); systems with resistance values below 5,000 Ω were demonstrated to have sufficient electrical leakage from the MN to the SkM chambers such that the skeletal muscle would be driven to contract due to broad field stimulation. Due to the poor electrical isolation between SkM and MN chambers in these systems, these low-resistance systems were excluded from testing. Recordings were collected using video output of an inverted phase microscope (Eclipse Ti2-A, Nikon, USA) to a Hamamatsu digital camera (model C-13440). A custom-designed, multi-ROI acquisition software built in LabVIEW was used to collect functional data of the NMJ systems. This software captured both a phase image of the systems as well as a subtracted view.

Broad-field electrical stimulation was achieved using a custom-designed dual channel stimulator to deliver bi-phasic square pulses of amplitude, +/− 20 mA and constant pulse width (PW) of 50 ms at pre-specified frequencies with varying times for each stimulation frequency. A sweep consisting of uniform 12 pulses at 0.33, 0.5, 1.0, 2.0, and 4.0 Hz frequencies was delivered first to the MN compartment (indirect stimulation), then to the SkM compartment (direct stimulation). Myotube contractions, incited either via direct electrical stimulation or MN activation, were visualized by a pixel differential generated from cell movement. The pixel differential was co-plotted with the delivered stimulation pulse to identify synchronicity between cell response and stimuli, and was interpreted further to assess NMJ fidelity, a parameter defined as the number of synchronous contractions divided by the number of delivered stimulation pulses. Prior to intensive testing, the number of NMJs per system was quantified. This value was later divided by the number of remaining NMJs post-stimulation to define NMJ stability.

5.7. Analysis of Functional Data

Recorded data were converted to traces and interpreted using a custom-built, semi-automated analysis software. NMJ response peaks were co-plotted against stimulation pulses to assess synchronization and to visualize NMJ transmission defects ranging from “skips” to complete failure, phenomenon that directly affect the fidelity parameter. Only response peaks corresponding to stimulation pulses were selected, as well as regions of tetanus, which were considered to be at 100% fidelity. Regions of interest (ROI) that responded to indirect but not direct stimulation were excluded from analysis, as well as ROIs that contracted with 0% fidelity at all frequencies of indirect stimulation but were active with direct stimulation. Aberrant ROIs that indicated errors in implementation of the recording software or uncharacteristically impaired NMJ maturity were also excluded from analysis.

5.8. Immunocytochemistry and Fluorescent Microscopy

Cells were fixed with 4% paraformaldehyde for ten minutes at 4°C followed by one rinse with 1X PBS and storage at 4°C in 1X PBS for up to two weeks. On the day of primary addition the cells were permeabilized with 0.05% Triton X-100 (Sigma-Aldrich X100) in PBS for 30 minutes, blocked with permeabilization solution containing 1% bovine serum albumin (Sigma-Aldrich A2153) and 5% goat serum (Sigma-Aldrich S26) for 60 minutes, all at room temperature. Primary antibody solutions were prepared in blocking solution using the following antibodies and dilutions, followed by overnight incubation with the requisite culture-specific antibody combination at 4°C: NANOG (Abcam ab62734, 1:200), SOX2 (Abcam ab137385, 1:200), OCT4 (Abcam ab18976, 1:100), MAP2 (Millipore Sigma AB5622, 1:1000), Choline acetyltransferase (Abcam ab181023, 1:100), ISL1/2 (Developmental Studies Hybridoma Bank 39.4D5, 4 μg/mL), SMI-32 (Millipore Sigma 5598440001, 1:1000), NKCC1 (Millipore Sigma MABS1237, 1:1000), Synaptophysin (Abcam ab8049, 1:100), and Neurofilament H (Millipore Sigma AB5539, 1:1000). The cells were then rinsed three times with PBS, after which the following secondary antibodies matching the species of the sample were incubated with the sample at room temperature for two hours: Goat anti-mouse 568 (ThermoFisher A-11004, 1:250), Goat anti-rabbit 568 (ThermoFisher A-11011, 1:250), and Goat anti-mouse 488 (ThermoFisher A-11001, 1:250). Alexa-fluor-conjugated α-Bungarotoxin (ThermoFisher B13422, 1:100) was added to select NMJ systems during this secondary antibody addition step. Residual secondary antibodies were rinsed out with consecutive five-minute rinses with PBS and the samples were incubated with the nucleic acid stain 4′,6-diamidino-2-phenylindole (DAPI) in a 300 nM solution for five minutes, followed by another two consecutive five-minute rinses with PBS.

Immunostained coverslips were mounted to glass slides with ProLong Gold Antifade Mounting solution (ThermoFisher P36931) and imaged up to one week after staining. Stained NMJ systems were not mounted to coverslips, as this would require PDMS removal and subsequently cause NMJ destruction. Instead, a custom-built adapter was utilized to image them via confocal microscopy. Stained coverslips and NMJ systems were imaged using a Nikon Ti2 microscope and C2 confocal laser running off of Nikon NIS Elements AR software. Quantification of immunofluorescence was determined by manual calculation of the percentage of cells in each imaged frame that were dual positive for DAPI and the marker of interest.

5.9. Axon Length and Branching Analysis

High-contrast phase microscopy images representative of the overall culture health were taken of three separate cultures of iPSC-hMNs at post-plating days 1, 6, 10, and 14. Images were analyzed using the trace feature in NeuronJ. Axons extending from the neuronal soma were traced as individual segments, where each point at which the axon split into additional segments was marked as a branch, with each continuing segment being traced as a new path. Axon branches per unit length, mean segment length, and varicosities per unit length were derived from these data.

5.10. Western Blot Analysis

Terminally differentiated iPSC-hMNs were harvested and lysed with cell extraction buffer (Invitrogen FNN0011) according to the manufacturer’s recommended protocol following its supplementation with 1 mM phenylmethylsulfonyl fluoride (PMSF) and protease inhibitor cocktail (Sigma-Aldrich P2714). Cellular debris was removed by centrifugation at 13,000 RPM for ten minutes at 4°C. Protein quantification of supernatants was performed via Bradford assay (Thermo Scientific 23246). 20 μg of total protein was diluted with 4X Laemmli buffer containing 50 mM β-mercaptoethanol, denatured at 40°C for 2 hours, loaded onto a 4–15% polyacrylamide gel, and electrophoresed at 100 V for 10 minutes then at 120 V for 90 minutes. Total protein stains were obtained for quantification purposes in lieu of potentially variable loading controls using the Pearce MemCode Reversible Protein Stain kit (ThermoFisher 24580) and were destained immediately after imaging. Proteins were transferred to nitrocellulose membranes at a constant voltage of 100V, 350 mA for 16 hours and were blocked with 5% dry milk for 1 hour at room temperature (Bio-Rad 1706404) prior to staining. NKCC1 was detected using a T4 antibody obtained from the Developmental Studies Hybridoma Bank at the University of Iowa (DSHB, T4-s) diluted to 0.5 μg/mL and incubated for 16 hours at 4°C. Following four successive rinses with TBS+0.1% Tween-20 (Bio-Rad 1706531), secondary staining was performed at room temperature for 60 minutes with an HRP-conjugated goat antibody against mouse IgG (Sigma-Aldrich 12–349). The membrane was again washed multiple times with TBS+0.1% Tween-20, then with TBS to remove any residual Tween. Immunostained membranes were imaged with a AlphaImager 2200 Gel Doc (Alpha Innotech), band intensity was normalized to the corresponding total protein stain, and expression was plotted relative to WT 268A expression levels.

5.11. G-banded Karyotyping

G-band karyotype analysis was performed by Cell Line Genetics (Madison, WI) using healthy, proliferating cultures of iPSCs. Standard GTL banding techniques were employed and 20 cells were selected for analysis. Any observed clonal or non-clonal mutations were noted and reported to Hesperos by NCATS.

5.12. Statistical Analysis

Data is displayed as the mean of a minimum of three biological replicates ± the standard error of the mean (SEM). Unpaired student’s T-tests assuming unequal variances with a two-tail distribution and one-way ANOVAs were used to determine statistical significance between WT and diseased readouts and is indicated by the presence of asterisks (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001).

Data availability

The datasets generated during and/or analyzed during the current study are not publicly available due to government restrictions on the data but are available from the corresponding author on reasonable request.