Biochemical characterization of two novel mutations in the human high-affinity choline transporter 1 identified in a patient with congenital myasthenic syndrome

Abstract

Congenital myasthenic syndrome (CMS) is a heterogeneous condition associated with 34 different genes, including SLC5A7, which encodes the high-affinity choline transporter 1 (CHT1). CHT1 is expressed in presynaptic neurons of the neuromuscular junction where it uses the inward sodium gradient to reuptake choline. Biallelic CHT1 mutations often lead to neonatal lethality, and less commonly to non-lethal motor weakness and developmental delays. Here, we report detailed biochemical characterization of two novel mutations in CHT1, p.I294T and p.D349N, which we identified in an 11-year-old patient with a history of neonatal respiratory distress, and subsequent hypotonia and global developmental delay. Heterologous expression of each CHT1 mutant in human embryonic kidney cells showed two different mechanisms of reduced protein function. The p.I294T CHT1 mutant transporter function was detectable, but its abundance and half-life were significantly reduced. In contrast, the p.D349N CHT1 mutant was abundantly expressed at the cell membrane, but transporter function was absent. The residual function of the p.I294T CHT1 mutant may explain the non-lethal form of CMS in this patient, and the divergent mechanisms of reduced CHT1 function that we identified may guide future functional studies of the CHT1 myasthenic syndrome. Based on these in vitro studies that provided a diagnosis, treatment with cholinesterase inhibitor together with physical and occupational therapy significantly improved the patient’s strength and quality of life.

Introduction

The neuromuscular junction (NMJ), situated at the interface between the motor neuron end plate and the muscle, mediates acetylcholine (ACh) signaling from the neuron that triggers muscle contraction. Pathogenic variants of NMJ proteins lead to congenital myasthenic syndromes (CMS) ranging in severity from fetal akinesia (reduced prenatal movements) and neonatal lethality to non-lethal fatigable weakness, hypotonia, and respiratory and feeding difficulties (1). In some patients, electromyography may reveal decreased muscle response to repetitive nerve stimulation (2). Currently, mutations in 34 different genes are associated with CMS (3). These include genes coding for proteins involved in every step of ACh signaling: its synthesis, its receptor subunits, presynaptic reuptake of its metabolite choline and uptake of ACh into presynaptic vesicles (1,4). Although most cases of CMS are caused by mutations in genes responsible for post-synaptic ACh signaling (4), rare forms of CMS are caused by mutations affecting presynaptic processes (5).

In the synapse, ACh is rapidly broken down by acetylcholinesterase into choline, which is then transported back into the presynaptic motor neuron by CHT1, a Na⁺-dependent and hemicholinium-3 (HC-3) sensitive (K_i 1–3 μM) choline transporter (choline K_m = 0.5–3 μM) encoded by the SLC5A7 gene. Additional neuronal choline transport is mediated by Na⁺-independent, less HC-3 sensitive (K_i 20–200 μM) lower affinity (choline K_m = 20–200 μM) choline transporters including SLC44A1 [originally termed the choline transporter-like protein 1, CTL1; (6–8)]. CHT1 forms homo-oligomers at the plasma membrane and is proposed to contain 13 transmembrane domains, with an extracellular amino terminus along with an intracellular carboxyl terminus (9). The 63-kDa molecular weight protein is also N-glycosylated (9,10). In addition, evidence suggests that CHT1 function is regulated via phosphorylation (11–14). Importantly, biallelic mutations in SLC5A7 that code for CHT1 cause a presynaptic CMS syndrome with a spectrum of severity, though often the course is severe with a high rate of neonatal lethality (2,5,15,16). Consistent with the severe form of the human SLC5A7 CMS phenotype, homozygous Slc5a7 knockout mice die from apnea within an hour of birth.

Given the role of CHT1 in choline reuptake that is required for resynthesis of ACh, patients with CMS because of SLC5A7 mutations may be amenable to treatment with cholinesterase inhibitors (ChEIs). ChEIs were first described as a treatment for myasthenia gravis in 1934 (17), and by preventing degradation of ACh, they increase its availability in the NMJ. ChEIs, including pyridostigmine, are given as first line therapy for presynaptic CMS (4), and they have been shown to improve muscle strength, including resolution of episodic apnea in some cases, in patients with SLC5A7 mutations. Therefore, identification and characterization of genetic mutations in patients with CMS can have invaluable clinical impact. Early initiation of therapy is especially critical for those with impaired respiratory function, which is a common presentation of SLC5A7-related CMS.

Here, we report a patient with moderate (not neonatally-lethal) CMS symptoms (hypotonia, developmental delay and a history of significant neonatal respiratory difficulties) in whom we identified two novel, compound heterozygous CHT1 mutations: p.I294T and p.D349N. Through extensive biochemical studies, we show that each mutation leads to loss of CHT1 function in a different manner: premature protein degradation with some residual transporter function (p.I294T), and transporter inactivation (p.D349N). Subsequent treatment of the patient with ChEI led to a significant improvement of symptoms and quality of life.

Results

Clinical description of the proband

The proband is an 11-year-old male with hypotonia and global developmental delay. His prenatal and birth history were unremarkable, and he was discharged home on the second day of life. He was admitted to the pediatric intensive care unit (PICU) at 1 week of age for lethargy, poor feeding, decreased weight and hypothermia, and was discharged home after treatment for presumed sepsis, dehydration, hypernatremia and hyperbilirubinemia. He was admitted again at 3 weeks of age for respiratory distress, bradycardia and desaturations requiring chest compressions and positive pressure ventilation. He was intubated and transferred to another hospital where he received feeds through a nasogastric tube. Barium esophagram showed severe oral stage dysphagia and delayed swallow. During this hospital course, he failed three extubation attempts. He was transferred to a third hospital at 7-week old where he received a tracheostomy, gastrostomy tube for supplemental feeding and Nissen fundoplication. Physical examination was notable for hypotonia, and muscle biopsy showed non-specific findings of some peri-fascicular atrophy with mild variation in fiber size and hypercontracted fibers. Electromyography was normal. At age 4 months, he was discharged to a pediatric rehabilitation center and then home at age 7 months on continuous positive airway pressure (CPAP) support with tracheostomy collar during sleep.

The proband had several subsequent admissions to the PICU between ages 9 and 24 months becuase of respiratory deterioration from infections and hypopharyngeal hypotonia. Global developmental delay was noted during these hospitalizations, and he was initiated on early intervention services. He first rolled over at 12-month old. At 13-month old, he was noted to have ophthalmoplegia and decreased muscle bulk and tone. He first socially smiled at 18-month old. At 24-month old, he sat unsupported and spoke his first word. He was weaned off ventilatory support by age 2.5 years, followed by closure of his tracheostomy. At 3-year old, he was able to follow simple commands and use two-word phrases. He continued to receive supplemental feeds via gastrostomy tube until age 3.5 years, when the gastrostomy was closed becuase of leakage, and he subsequently underwent percutaneous endoscopic gastrostomy (PEG) placement. At 5-year old, he was able to walk and received lower extremity bracing to prevent equinus contracture. He also underwent orchiopexy for bilateral undescended testes. At 6-year old, he discontinued bracing after improvement. Electromyography repeated at 10-year old showed occasional variably decreased amplitude motor unit action potentials and normal repetitive nerve stimulation.

Prior to enrollment into our research program, clinical genetic testing of the proband was non-diagnostic, including karyotype, microarray, PHOX2B gene sequence analysis, mitochondrial sequencing, trio whole-exome sequencing and clinical reanalysis of trio whole-exome sequencing. At 10 years of age, the proband was enrolled into our Pediatric Undiagnosed Diseases Program at the New York University Grossman School of Medicine. At this time, he was able to ambulate independently with bilateral foot pronation and external rotation with tight heel cords. He was able to walk ~80 m, but tired easily, and he required a wheelchair for longer ambulation with worsening weakness throughout the day. Cognitively, per the mother, he was approximately like a 4-year old and had some regression of prior skills such as recognizing numbers, letters and colors. He was also impulsive and had difficulty with attention. He was not toilet-trained, and he wore diapers. He was in special education with a para-assistant and an individualized education and support program including physical, speech and occupational therapy. He was diagnosed with mild to moderate intellectual disability after a neuropsychological evaluation. His parents are non-consanguineous and family history is non-contributory.

Research genomics analysis identifies p.I294T and p.D349N variants in SLC5A7

A complete medical history review of the patient suggested that his presentation was consistent with a congenital neuromuscular disease that warranted further genomic analyses. Reanalysis of whole-exome sequencing has a well-documented increase in diagnostic yield (18). Therefore, as our initial strategy for identifying a genetic diagnosis, prior clinical whole-exome sequencing data were obtained and reanalyzed.

Research reanalysis of whole-exome sequencing identified compound heterozygous variants in the proband in SLC5A7: Variant (1) chr2:108008614G > A (hg38), c.1045G > A (NM_021815.5), p.D349N (NP_068587.1); and Variant (2) chr2:108006188 T > C (hg38), c.881 T > C (NM_021815.5), p.I294T (NP_068587.1; Fig. 1A). Figure 1B shows a topological model of CHT1 with previously reported mutations and the two mutations reported in this manuscript. Although the father was originally available for the clinical whole-exome sequencing, he was not available for genotyping of these variants as part of our research program. Sanger sequencing confirmed variant 1 was absent in the mother but present in a heterozygous state in a healthy brother, whereas variant 2 was present in a heterozygous state in the mother but absent in the healthy brother, thereby confirming that the proband was compound heterozygous for the two variants. Variant 1 has a CADD score of 29.6—an InMeRF score of 0.947 (19)—and is absolutely conserved across vertebrates (20). Variant 2 has a CADD score of 22.5—an InMeRF score of 0.7—and is conserved across vertebrates with the exception of a few species. High CADD and InMeRF scores predict the variant is likely to result in a deleterious effect. Both variants are absent in gnomAD (21). Given that the variants did not meet criteria of the American College of Medical Genetics to be clinically reportable as ‘Likely Pathogenic’ or ‘Pathogenic’ (22), we decided to conduct a functional characterization of the mutants to fully assess the impact of each variant to confirm the diagnosis.

Figure 1

Identification of CHT1 mutations, and CHT1 topological model with CMS-causing mutations. (A) Proband whole-exome sequencing (WES) variants were analyzed with three tiers of filters: Tier I (genotype quality and population frequency), Tier II (predicted pathogenicity and phenotype-driven filtering) and Tier III (inheritance mode). Phenotype-driven filtering was performed using Qiagen’s biological context algorithm which ranked correlation of genes to human phenotype ontology (HPO) terms derived from medical chart review. This approach produced a set of candidate variants, of which only the compound heterozygous SLC5A7 variants matched the patient’s phenotype. Additional exploratory analyses with less stringent filters did not identify additional candidate variants. n = number of variants after each filtration step; GQ = genotype quality and AF = allele frequency. (B) Topological model of CHT1 highlighting previously reported CMS-causing mutations. Light blue corresponds to the membrane, dark blue ovals to CHT1 transmembrane domains with residue number at the cytosolic and extracellular end indicated. Newly identified CHT1 mutations I294T and D349N are indicated in red.

Heterologous p.D349N CHT1 is more abundant than WT CHT1 whereas p.I294T CHT1 is significantly less abundant than WT CHT1 protein in HEK 293 cells

Although the proband is a compound heterozygote, each mutant was characterized separately to determine the mechanism by which each mutant may lead to alteration of function. In the absence of an immortalized cell line appropriately reproducing motoneurons, HEK 293 cells were chosen because they are easy to transfect and they do not express CHT1 endogenously, thereby facilitating functional studies. HEK 293 cells stably expressing p.I294T, p.D349N or WT CHT1 proteins were generated, and cell lysates prepared to assess protein abundance in this cell model. As shown in Figure 2A and B, in comparison with WT CHT1, p.I294T mutant abundance was reduced to 36.1 ± 1.5% of WT CHT1 (n = 4, ± SEM), whereas p.D349N abundance was 135.5 ± 1.5% (n = 4, ± SEM) of WT CHT1 protein level. Similar relative abundances between mutant and WT CHT1 were observed in four independent transient transfections and in three clones isolated from CHT1 WT, I294T and D349N parental cell lines (Supplementary Material, Fig. S2). The migration profile of WT CHT1 shows an abundant upper band corresponding to complex glycosylated protein (~60 kDa) and a lower broad band (~48 kDa) corresponding to core glycosylated and non-glycosylated protein, as previously described (23). The reduced abundance of p.I294T CHT1 mutant could originate from increased protein degradation, or reduced protein synthesis so we next assessed p.I294T CHT1 mutant degradation rate.

Figure 2

Assessment of WT and mutant CHT1 abundance and half-life. (A) Representative immunoblot from HEK 293 cells stably expressing WT, p.I294T or p.D349N CHT1 mutants or empty vector (EV). Black circles indicate CHT1 carrying complex oligosaccharides, white circle shows core glycosylated CHT1 and triangles show unglycosylated CHT1. (B) Quantification of CHT1 abundance in EV, WT, p.I294T or p.D349N CHT1 mutants. Error bars correspond to mean ± SEM, n = 4. ^**P < 0.01, ^****P < 0.0001 using 1-way ANOVA with Tukey’s multiple comparison post-hoc test. (C) Representative immunoblot of WT and p.I294T CHT1 expressing cells treated with cycloheximide. Error bars correspond to mean ± SEM, n = 3. ^**P < 0.01 using two-tailed unpaired t-test ANOVA. CHX, cycloheximide. (D) Quantification of WT and p.I294T CHT1 abundance after treatment with cycloheximide. Error bars correspond to mean ± SEM, n = 4. ^**P < 0.01 using 1-way ANOVA with Tukey’s multiple comparison post-hoc test.

p.I294T has a shorter half-life compared to WT

To address whether p.I294T CHT1 mutant has a shorter half-life, we incubated HEK 293 cells stably expressing this mutant with cycloheximide for 0, 2, 4, 6, 9 or 24 h and lysates were subsequently subject to immunoblot using rabbit polyclonal CHT1 antibody. As shown in Figure 2C and D (Supplementary Material, Fig. S1), p.I294T CHT1 mutant had a significantly reduced half-life compared with WT CHT1. Although WT CHT1 had a half-life of 23 h, the p.I294T CHT1 mutant had a half-life of 7 h. Immunofluorescence analysis showed a uniform layer of low fluorescent cells rather than a few high expressing cells, supporting that the viral infection was successful and uniform but resulted in low CHT1-expressing cells (Supplementary Material, Fig. S3). Although we did not assess the rate of protein synthesis, this result indicates that the decreased abundance of p.I294T CHT1 mutant is at least in part becuase of rapid degradation compared to WT CHT1 protein. Given the high abundance of p.D349N CHT1 mutant, its half-life was not examined.

The proportion of p.I294T and p.D349N mutants reaching the plasma membrane is similar to WT CHT1

The reduced half-life of p.I294T CHT1 mutant may translate into a decreased abundance of the protein at the cell surface. To assess this, we conducted cell surface biotinylation assays in cells expressing WT CHT1, p.I294T CHT1 or p.D349N CHT1. For each CHT1 genotype, the ratio of biotinylated or unbiotinylated protein was normalized to the total amount of CHT1 protein. This strategy allowed us to assess the relative ability of CHT1 WT or mutant protein to reach the cell surface independently of its overall abundance. As shown in Figure 3, despite reduced overall abundance, the proportion of cell surface p.I294T CHT1 mutant is similar to WT CHT1, when normalized to total protein abundance (86.3 ± 40.6% for p.I294T CHT1 and 126.1 ± 40.1% for p.D349N CHT1; n = 4, mean ± SEM). This result demonstrates that although less total protein is present in these cells, the remaining protein is translocated to the plasma membrane in the same proportion as WT CHT1, indicating normal processing to its final destination in the cell membrane. Furthermore, the fraction of unbiotinylated CHT1 is comparable between WT, p.I294T and p.D349N. Thus, the p.D349N CHT1 mutant reached the cell surface in a similar proportion to WT CHT1 when normalized to total CHT1 protein abundance. These results suggest mistrafficking of either mutant is not a likely cause of the disease.

Figure 3

Cell surface abundance of WT, p.I294T or p.D349N CHT1 in HEK 293 cell lines. Cell surface biotinylation was performed as described in section ‘Materials and methods’. (A) Representative immunoblot of total (input), unbiotinylated and biotinylated fractions isolated from HEK 293 cells stably expressing WT, p.I294T or p.D349N CHT1. Black circle indicates CHT1 carrying complex oligosaccharides, the white circle shows core glycosylated CHT1, arrowhead indicates unglycosylated CHT1. (B) Quantification of the total (input), unbiotinylated or biotinylated fraction of WT or p.I294T CHT1 normalized to total protein expression of each mutant and to WT abundance. Error bars correspond to mean ± SEM, n = 5. ^*P < 0.05, ^**P < 0.01 using 1-way ANOVA with Dunnett’s multiple comparison post-hoc test.

p.D349N CHT1 transport is inactive whereas p.I294T CHT1 transport is partially active

Given that both p.I294T CHT1 and p.D349N CHT1 mutants reach the cell surface at similar proportions to WT CHT1, we next assessed their transport activity. We measured the transport rate of 0.1, 2.5 and 10 μM tritiated choline over 15 min in HEK 293 cells expressing empty vector, WT or mutant CHT1 proteins. Figure 4A shows that compared to WT CHT1, the transport activity of the highly abundant p.D349N CHT1 mutant is comparable with HEK 293 cells expressing the empty vector, which is significantly lower than WT CHT1. This result indicates that the p.D349N substitution results in a non-functional protein, potentially through interference with binding or release of choline, protein misfolding (without triggering premature protein degradation) or blocking the transporter in an inactive state. Interestingly, the p.I294T CHT1 mutant shows 49.5 ± 1.7% (n = 3, mean ± SEM) residual transport function compared with WT CHT1 when cells are incubated with 0.1 μM tritiated choline, which is significantly higher than empty vector expressing cells (Fig. 4A). However, the significant difference compared with empty vector expressing cells is abolished with 2.5 and 10 μM tritiated choline. Correction for CHT1 abundance makes I294T function similar to WT (Fig. 4B). This suggests that I294T, whereas lower in total abundance, retains some activity. Notably, this residual activity may explain the less severe phenotype observed in the patient and raises the possibility that this mutant may support more choline transport activity if rescued from premature degradation.

Figure 4

Transport of [³H]-choline by HEK 293 cells stably expressing WT, p.I294T or p.D349N CHT1 or empty vector (EV) at a single time point of 5 min at multiple choline concentrations (0.1, 2.5 and 10 μM), either (A) uncorrected or (B) corrected for CHT1 protein level. Values are shown as percentage of WT transport activity. Error bars correspond to mean ± SEM, n = 4. ^**P < 0.01, ^****P < 0.0001 using 1-way ANOVA with Tukey’s multiple comparison post-hoc test.

Staurosporine partially rescues p.I294T CHT1 mutant processing

Given the partial transport activity observed for the p.I294T CHT1 mutant, we wondered whether chemical chaperones that rescue cystic fibrosis transmembrane conductance regulator (CFTR) mutant proteins [VX 809 (24) and C3 (25,26)] or that activate CHT1 [MKC 231 (27) and staurosporine, STS (28)] would rescue p.I294T CHT1 mutant abundance and activity. HEK 293 cells stably expressing WT, p.I294T or p.D349N CHT1 mutants were incubated with VX 809 (2 or 5 μM), 1% DMSO, C3 (5 or 10 μM), MKC231 (1 or 10 μM) or 5 μM STS for 24 h prior to immunoblotting. None of the treatments altered p.D349N CHT1 mutant abundance (data not shown), and none of the treatments improved p.I294T CHT1 mutant abundance except for the STS treatment that significantly increased the glycosylated/total CHT1 ratio of the mutant from 0.51 ± 0.07 to 0.88 ± 0.08 (n = 3, mean ± SEM; Fig. 5). An increased glycosylation ratio may correlate with increased trafficking to the cell surface. Despite an overall low abundance of p.I294T, combined with appropriate trafficking (Fig. 3), a strategy may be to further increase trafficking to the membrane, to rescue transport activity and correct the transport deficit seen in this patient. However, STS was toxic to cells and its effect could not be tested on CHT1 transport function. This result highlights that development of future drugs that increase proper processing of CHT1 mutants with residual transport function might be a viable therapeutic strategy.

Figure 5

Chemical chaperone treatments of WT or p.I294T CHT1 stably expressing cells. (A) HEK 293 cells stably expressing WT (left panel) or p.I294T (right panel) CHT1 were treated with VX809 (vehicle DMSO or 5 μM), DMSO (none or 1%) or C3 (vehicle DMSO or 5 μM) for 24 h prior to lysis and immunoblotting. Black circles indicate CHT1 carrying complex oligosaccharides, the white circles indicate core glycosylated CHT1, arrowhead indicates unglycosylated CHT1. (B) HEK 293 cells stably expressing WT (left panel) or p.I294T (right panel) CHT1 mutant were untreated (‘−’), treated with MKC231 (MKC, 1 or 10 μM), staurosporine (STS, 5 μM) or volume matched DMSO control for 24-h prior to lysis and immunoblotting. ‘−’ indicates volume matched vehicle control (water for DMSO treatment), n = 4. Black circle indicates CHT1 carrying complex oligosaccharides, the white circle shows core glycosylated CHT1, arrowhead indicates unglycosylated CHT1. (C) (top panel) Quantification of total CHT1 abundance for WT and p.I294T under the indicated treatment conditions. (bottom panel) Quantification of the ratio of glycosylated to total CHT1 protein for WT and p.I294T CHT1 expressing cells under the indicated treatment conditions. Error bars correspond to mean ± SEM, n = 3. ^**P < 0.01 using a 2-way ANOVA with Tukey’s multiple comparison post-hoc (top panel) or Sidak’s multiple comparison post-hoc (bottom panel) test.

Treatment with ChEI and physical and occupational therapy significantly improved the proband’s symptoms and quality of life

Based on the functional evidence provided by these functional studies, the diagnosis of SLC5A7-related CMS was made and reported to the family, referring neurologist and geneticist. Given the diagnosis of SLC5A7-related CMS, the proband was prescribed the ChEI Pyridostigmine (30 mg, 3 times daily at patient weight 30 kg) and additional physical and occupational therapies. Treatment resulted in improved locomotion, strength, and endurance as shown by increasingly achieved distances in standardized 2-min walk tests, and total time, speed and incline on a treadmill (Fig. 6). Subjective reports from his mother also stated that he was able to walk longer distances in single excursions (from 80 to 250 m), and to run and eat dry food, which he was unable to previously perform.

Figure 6

Physical therapy endurance measurements after initiation of treatment. After diagnosis, the proband was initiated on pyridostigimine (30 mg, three times per day, patient weight 30 kg) and additional physical therapy sessions (once per week). (A) Two-minute walk test distances. The 2 min walk test is a standardized assessment of endurance that measures total distance after 2 min of walking through a 15.2 m out-and-back course. Distances were recorded over three sessions. (B–D) The proband performed an endurance exercise by walking on a treadmill over the course of 12 physical therapy sessions. (B) Total time on treadmill. (C) Speed on treadmill. (D) Percent incline on treadmill. Percent incline is calculated from the ratio of vertical to horizontal distance on the treadmill. Linear regressions with R² values (square of Pearson correlation coefficient) are shown, except for 2 min walk test distance (A) for which there are too few measurements.

Discussion

Here, we report a proband with novel compound heterozygous mutations in CHT1, p.I294T and p.D349N, which result in a non-lethal CMS. Using biochemical assays, we show that in HEK 293 cells, the p.I294T CHT1 mutant is less abundant than CHT1 WT, but retains some cell surface localization and transport function at low choline concentrations. On the other hand, the p.D349N CHT1 mutant is more abundant than WT CHT1 and retains cell surface localization but is inactive at any concentration of choline tested. Only STS improved p.I294T CHT1 processing, resulting in an increased fraction of CHT1 carrying complex glycosylation. However, this treatment caused cell death and could not be tested on mutant transport function. As a result of this genetic diagnosis and biochemical characterization, the proband was treated with a ChEIs, and his symptoms improved.

Five studies to date have correlated SLC5A7 mutations with NMJ defects. Wang et al. (16) described three homozygous mutations in SLC5A7 in three families, including highly conserved p.S94R and p.V112E, and p.P210L mutations. The phenotypes varied from the classical presentation of CMS (P210L) who responded to ChEIs, to generalized neurodevelopmental delay coupled with brain atrophy (S94R) and infantile lethality (V112E), who did not respond to a ChEI. Wang et al. hypothesized that aberrant protein trafficking and localization lead to these phenotypes by in vitro studies. Pardal-Fernández et al. (5) next described a compound heterozygous infant with acute respiratory failure, generalized weakness and severe hypotonia who carried compound heterozygous mutations in SLC5A7, and they proposed that the phenotype was becuase of low levels of ACh released into the synapse leading to aberrant muscle fiber potentials. In that case, ChEIs improved the patient’s condition but only for a short period of time because the impaired presynaptic CHT1 is not dependent on the choline gradient (5,16). In addition, Bauche et al. (2) not only described 11 recessive SLC5A7 mutations in six unrelated families clinically ranging from CMS with episodic apnea to a lethal prenatal type of arthrogryposis, but also complemented their studies through biochemical characterization of some of the mutations identified. Finally, three novel homozygous mutations (p.R107H, p.A296T and p.Y414N) resulted in presynaptic CMS and in varying degrees of weakness, including fatality, and varying responses to ChEIs ranging from improvements in gross motor development and dysphagia (R107H and A296T) to no response [Y414N; (29)]. Rodriguez Cruz et al. proposed that response to treatment corresponds to presenting phenotypes and residual activity of CHT1. Notably, our patient has had sustained improvement in motor strength at least over a 6-month follow-up period, consistent with residual function of the I294T mutant in our biochemical assay. Therefore, genotype–phenotype correlations and biochemical characterization of CHT1 mutations may help guide treatment management in CHT1-related CMS.

Our group previously identified a recessive p.S263F mutation in SLC5A7 in two siblings of El Salvadoran descent (30). This amino acid substitution manifested with electromyography suggestive of CMS, arthrogryposis and reduced prenatal movements. Both siblings were withdrawn from intensive care a few months after birth. Characterization of the mutation in vitro showed that the substitution resulted in a complete loss of function, explaining the severe phenotype of the homozygous patient.

A structure for human CHT1 has not been published yet. According to the predicted structure of human CHT1 with Alphafold [https://alphafold.ebi.ac.uk/entry/Q9GZV3; (31,32)], D349 is buried within transmembrane helix 9 at the center of the transmembrane helices bundle (Fig. 7A and B), and its carboxyl group interacts with S350 and Y175. Substituting D349 to N would likely impair one of these interactions and may explain the lack of activity in the CHT1 D349N variant, which was not rescued by increasing the choline concentration. Furthermore, D349 is highly conserved across vertebrates, highlighting the likely importance of this specific residue. Bauche et al. (2) reported a compound heterozygous patient carrying the p.V344L on one allele and p.Y175C substitution on the other allele. V344 is only five amino acids away from the p.D349N substitution reported here. This patient experienced some symptoms similar to the proband reported here, including continuing episodic apnea, motor delay, facial weakness, difficulties chewing, limb fatigability and weakness (2). A mild improvement of the symptoms was observed upon treatment with ChEIs. No functional characterization of this substitution was reported.

Figure 7

Alignment and location of D349 or I294 on CHT1 structure predicted by AlphaFold. (A) Alignment of the choline transporter primary sequence from various species, highlighting the location of D349 (red rectangle). (B) Predicted human CHT1 structure using AlphaFold highlighting the position of D349 (pink) (arrow) buried at the center of the transmembrane domain. Left panel shows the sideview and right panel shows the cytosolic view of the protein with D349 shown in pink (arrow). (C) Alignment of the choline transporter primary sequence from various species, highlighting the location of I294 (red rectangle). (D) Predicted human CHT1 structure using AlphaFold highlighting the position of I294 (pink) at the extracellular side of transmembrane helix 8 (arrow). Left panel shows a side view and right panel shows the extracellular view of the protein with I294 shown in pink (arrow).

The I294T mutant results in the introduction of a polar uncharged group. According to the Alphafold predicted structure of human CHT1, this highly conserved residue is located at the extracellular end of transmembrane domain 8 in proximity to A206, L207 and V212 (Fig. 7C and D). Substitution to a polar uncharged residue may disrupt these interactions and cause inappropriate recognition by the protein quality-control machinery, leading to premature degradation of the protein. Surprisingly, in HEK cells stably expressing CHT1 I294T, at 0.1-μM choline, residual transport function was observed, but at higher choline concentrations, no function was detected. Although it is unclear why transport activity was abolished at high choline concentrations, this finding suggests that this substitution does not cause a dramatic misfolding with complete loss of function, in line with the milder phenotype of this patient compared to lethal CMS conditions caused by transport dead mutants. Two substitutions in the vicinity of I294 have been reported to cause CMS. Three residues away, p.I291T substitution was found in the compound heterozygous state with p.R446G in a patient with a mild form of CMS (2). This patient presented with episodic apnea, delayed motor development, limb fatigability and significant response to treatment with ChEIs including resolution of apneic episodes. The second substitution, p.A296T is only two residues away from I294 (29). Two siblings homozygous for this substitution displayed various apneic episodes, feeding difficulties, delayed motor development, fatigue, significant learning and behavioral difficulties. Both patients were treated with pyridostigmine and salbutamol with improvement of symptoms (29).

Dietary choline is essential, and choline synthesis does not occur in cholinergic neurons, which is why high-affinity choline uptake (HACU) via CHT1 is required for most cholinergic neurotransmission. HACU is limited to cholinergic nerve terminals where the choline concentration is low because it is converted to ACh (33,34), whereas the ubiquitous CHT1/Na⁺-independent transport occurs where the choline concentration is high. Approximately 60% of the choline that is taken up by the cell via HACU is converted to ACh (33), whereas only a small amount of choline transported by the Na⁺-independent pathway is used to synthesize ACh (8). Therefore, HACU is necessary for ACh synthesis, and this is evident in mouse CHT1 knockout models that have a lethal phenotype (6,10,35,36). Since the proband has a reduced but residual capacity for choline uptake, increasing the concentration of ACh in the synapse via a ChEI seems to be an appropriate intervention to help improve symptoms, which was subsequently demonstrated clinically.

In summary, we report here a child with CMS carrying newly identified compound heterozygous p.I294T and p.D349N CHT1 mutations. In vitro characterizations suggest two mechanisms by which these mutations lead to loss of CHT1 function. Based on these findings, the child was treated with a ChEI that significantly improved the patient’s condition. This study highlights the benefits of bedside to bench to bedside approaches for rare diseases, and the clinical applicability of genotype–phenotype correlations.

Materials and Methods

Ethical compliance and subject enrollment

The proband, parents and unaffected siblings were enrolled into the Pediatric Undiagnosed Diseases Program at the NYU Grossman School of Medicine. The study was approved by the NYU Grossman School of Medicine Institutional Review Board (Protocol s18-01743) and by the University of Alberta Health Research Ethics Board (Pro00121079). Informed consent was obtained from the parents, and conduct of the study followed all regulatory and ethical guidelines per the approved study protocol and US Federal regulations for the protection of human subjects.

Review of medical records

To obtain a complete view of the proband’s medical history, medical records were gathered from all prior treating physicians. A complete review of the medical records was performed, including prenatal, birth, medical and family histories along with physical examination, imaging and laboratory tests. To help guide the research strategy given the prior non-diagnostic clinical evaluations, the summarized medical history was also reviewed by the Pediatric Undiagnosed Diseases Program study committee, consisting of a clinical geneticist, a pediatric neurogeneticist, two research geneticists and three genetic counselors.

Research sample collection

Genetic research studies utilized saliva collected from the proband and mother using the Oragene OGR-600 collection kit. The father was unavailable for sample collection. Blood was collected from three unaffected siblings for segregation analysis. For CLIA-grade validation of research results, to enable reporting of the diagnosis to the family, new blood samples were collected from the proband and mother.

Genomic data analysis

Raw whole-exome sequencing data from prior non-diagnostic clinical testing (Roch VCRome 2.1 exome panel) were obtained and aligned to the human reference genome (hg38) with Burrows-Wheeler Alignment Tool [BWA; (37)]. Single-nucleotide and indel variants were called using the Genome Analysis Toolkit Best Practices pipeline (38) run via the Terra platform on Google Cloud. Aligned reads produced a mean target exome coverage of 154x, with 96% of the exome at ≥30x. Single-nucleotide and indel variants were filtered and analyzed using Ingenuity Variant Analysis (Qiagen, Germantown, MD, USA). Variants were filtered for genotype quality (>30), population frequency [< 1% in the gnomAD database; (21)], missense or loss-of-function variants, and variants compatible with autosomal recessive, compound heterozygous and X-linked modes of inheritance. Additional analyses were conducted to examine: (i) all rare variants in genes associated with muscle weakness syndromes, (ii) all rare variants with Combined Annotation Dependent Depletion (CADD) score > 25 (20) and, (iii) all rare variants in candidate genes from differential diagnoses suggested by the medical records review. All resulting variants were examined for consistency with the proband phenotype, inheritance pattern, predicted pathogenicity and classification criteria of the American College of Medical Genetics (22).

Sanger sequencing validation of genetic variants

Genomic DNA was extracted from saliva (mother and proband) using prepIT-L2P reagent (DNA Genotek) and from blood (unaffected siblings) using the MagAttract HMW DNA kit (Qiagen, Germantown, MD, USA). The inheritance pattern of the two SLC5A7 variants was initially evaluated on a research basis by amplifying with the GoTaq G2 Flexi PCR kit (Promega, Fitchburg, WI, USA) using primers spanning the two SLC5A7 variants (chr2:108006188 T > C: TCTTCCTCAGCCACCTATGC and CCCACCATTCCCCAGATGAT; chr2:108008614 G > A: GGCTTCCAGATCCCAAGACT and CCACCAAGATTTTCCAGTGTCA; hg38 coordinates). PCR products were purified using AMPure magnetic beads, followed by Sanger sequencing (Genewiz, South Plainfield, NJ, USA). CLIA-grade validation of variants and the inheritance pattern for the mother, proband and an unaffected sibling was performed on blood samples by the Precision Genomics Laboratory at Columbia University.

Reporting of diagnosis and clinical outcomes

The diagnosis was reported to the mother and the treating physician by the Pediatric Undiagnosed Diseases Program. The family was provided a diagnostic report and counseling and was referred to a neurologist and clinical geneticist for clinical follow-up. Clinical outcomes of the proband were obtained from the subsequent medical records and via post-diagnosis interviews and survey of the mother. Quantitative physical performance data were obtained from physical therapy medical records, including 2-min walk test distances, treadmill time, speed and incline. Physical performance was plotted relative to days after initiation of treatment with pyridostigmine.

Site-directed mutagenesis

The human CHT1-WT cDNA was a gift from Randy Blakely [Addgene, plasmid # 15 766; (39)] and was cloned into the pLVX-IRES-Hygromycin B (Hyg) vector (Clontech) as described previously (30). The point mutations 881 T > C (I294T) and 1045G > A (D349N) were introduced using the Q5 site-directed mutagenesis kit [New England Biolabs (NEB, Ipswich, MA, USA)]. The mutagenic primers were designed with the NEB Q5 software and the forward primers are as follows I294T: chr2:108006188 T > C (hg38) c.881 T > C (NM_021815.5) 5′–CATTGGGGCCactGGAGCATCAA– 3′. D349N: chr2:108008614 G > A (hg38) c.1045G > A (NM_021815.5) forward primer: 5′ –GTCATCAGCAaatTCTTCCATCTTG– 3′. The resulting cDNA was sequenced to confirm the presence of the variant (The Applied Genomics Core, University of Alberta, Edmonton, Canada).

Cell culture, generation of stable cell lines and transient transfection

Human embryonic kidney (HEK) 293 cells (Clontech, Mountain View, CA, USA) were used for a lentiviral production procedure (Clontech, Mountain View, CA, USA). Cells were grown in complete medium [Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Bend, OR, USA) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich, Saint Louis, MO, USA), 100 units/ml of penicillin and 100 μg/ml of streptomycin (ThermoFisher Scientific, Bend, OR, USA)] in 5% CO₂ at 37°C in a humidified incubator until 80% confluency. Next, the cells were transfected with the pLVX-IRES-Hyg plasmid bearing human WT or mutant CHT1 cDNA using a Single Shot packaging kit (Clontech, Mountain View, CA, USA). After 48 h, the viral supernatant was collected and passed through a 0.45-μm filter, then supplemented with 8 μg/ml of polybrene (Sigma-Aldrich, Saint Louis, MO, USA) and used to infect fresh HEK 293 cells at 90% confluency. Selection of cells stably expressing WT or mutant CHT1 was performed by replacing the viral supernatant after 3 days with complete growth medium containing 0.2 mg/ml hygromycin B (Gibco, Bend, OR, USA). Unless otherwise indicated, cells were used in experiments directly after selection. For Supplementary Material, Figure S2B, single cell clones were generated using a cell sorter (Sony MA900, University of Alberta Faculty of Medicine & Dentistry Flow Cytometry Facility) to place individual cells in wells of 96-well plates. Transient transfections were also performed using the constructs described previously and Lipofectamine 3000 according to the manufacturer’s instructions (1 μg DNA/1.5 μl lipofectamine ratio). Experiments were performed 48 h after transfection.

Immunoblot and immunofluorescence

Confluent HEK 293 cell lines expressing mutant CHT1, wild-type CHT1 (positive control) or empty expression vector (negative control) were lysed with radioimmunoprecipitation assay (RIPA) lysis buffer (0.3 M NaCl, 20 mm Tris/HCl pH 7.5, 2 mm EDTA, 2% Deoxycholate, 2% Triton X-100, 0.2% SDS, pH 7.4) with protease inhibitors [Complete Mini EDTA-free (Roche, Basel, Switzerland)] and phosphatase inhibitors (PhosSTOP, Roche, Basel, Switzerland). A bicinchoninic acid (BCA, Thermo Scientific, Bend, OR, USA) assay was performed to determine the total protein concentration in the cell lysates. Twenty microgram of lysates were loaded onto 7.5% SDS-PAGE gels and proteins then transferred onto a PVDF membrane (Millipore, Burlington, MA, USA). The membrane was blocked for 1 h with skim milk (3% w/v; BioBasic, Markham, ON, Canada) in Tris-buffered saline with 0.1% Tween 20 (TBST). Next, it was incubated with anti-CHT1 rabbit polyclonal antibody (1: 5000 overnight; Millipore, Burlington, MA, USA) in TBST containing 1% skim milk, followed by anti-rabbit horseradish peroxidase (HRP)-conjugated secondary antibody (1: 10 000 for 1 h; GE Healthcare, Chicago, IL, USA) incubation. Clarity Western chemiluminescent detection reagent (Bio-Rad, Hercules, CA, USA) was used for detection and visualization performed with the ChemiDoc MP Imaging system (Bio-Rad, Hercules, CA, USA). Quantification of relative band intensity was performed using the ImageLab software (Bio-Rad, Hercules, CA, USA). The membrane was then subjected to anti-β-actin HRP antibody (1:10 000; Biolegend, San Diego, CA, USA) in TBST containing 1% skim milk, and the bands were visualized as described previously.

Immunofluorescence experiments were conducted on sub-confluent HEK 293 cell lines expressing I294T, D349N mutant CHT1 or wild-type CHT1. Cells were grown on poly-l-lysine coated glass coverslips, fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 and blocked with 1% bovine serum albumin (BSA). Cells were then incubated with anti-CHT1 rabbit polyclonal antibody followed by goat anti-rabbit secondary antibody coupled to Alexa 488 (Molecular Probes, Carlsbad, CA, USA), 4′,6-diamidino-2-phenylindole (DAPI) was used to stain the nuclei. Samples were examined using a Zeiss laser confocal microscope LSM 510 and a 60 X oil objective.

Cycloheximide chase experiment

HEK 293 cell lines stably expressing mutant CHT1, wild-type CHT1 (positive control line) or empty expression vector (negative control line) were grown to 90% confluency on poly-l-lysine treated 6-well plates and subsequently treated with 0.3 mg/ml of cycloheximide (Fluka, Buchs, Switzerland) for 0, 2, 4, 6, 9 and 24 h. The cells were lysed the following day with RIPA lysis buffer containing protease inhibitors [Complete Mini EDTA-free (Roche, Basel, Switzerland)] and phosphatase inhibitors (PhosSTOP Roche, Basel, Switzerland). A BCA assay was performed, and CHT1 levels were detected via immunoblot as described previously.

Rescue treatments

HEK 293 cell lines stably expressing mutant CHT1, wild-type CHT1 (positive control) or empty expression vector (negative control) were grown to 90% confluency and on poly-l-lysine treated 6-well plates and subsequently treated with Corrector quinazolinone C3 (gift from the Cystic Fibrosis Foundation Therapeutics, Bethesda, MD), VX 908 (SELLECK Chemicals, Houston, TX, USA), volume matched DMSO controls for both or 1% DMSO treatment with a volume matched water control for 24 h. The cells were also treated with MKC231 (MedChemExpress LLC, Monmouth Junction, NJ, USA) or STS (FroggaBio Inc., North York, ON, Canada) and volume-matched DMSO control and a negative control for 23 h. The cells were then lysed in RIPA lysis buffer containing protease inhibitors [Complete Mini EDTA-free (Roche, Basel, Switzerland)] and phosphatase inhibitors (PhosSTOP, Roche, Basel, Switzerland). BCA assay was performed and CHT1 level was detected by immunoblot as described previously.

Choline uptake assay

HEK 293 cells stably expressing WT-CHT1, mutant CHT1 lines or empty vector were grown on poly-l-Lysine coated coverslips in 6 well plates until >90% confluency. The cells were washed twice with warm (37°C) Krebs-Ringer-HEPES (KRH) buffer (130 mm NaCl, 1.3 mm KCl, 2.2 mm CaCl₂.2H₂O, 1.2 mm MgSO₄.7H₂O, 1.2 mm KH₂PO₄, 10 mm HEPES, 10 mm glucose, pH 7.40), and then incubated with warm KRH for 10 min at 37°C. The cells were then incubated with warm KRH containing [³H]-choline (0.1 μM, 100 nCi/ml) at 37°C for 5 min. After 3 washes with cold KRH, they were subsequently lysed using 750 μl of 0.5% Triton. The transport activity was measured using liquid scintillation (Beckman model LS6000IC) and activity was normalized by measuring the total protein concentration in each well. Untreated cells from the same seeding date were subject to immunoblot as described previously to assess CHT1 expression.

Cell surface biotinylation

HEK 293 cells stably expressing WT-, CHT1 mutants or empty vector were seeded on six well plates coated with Poly-l-lysine. The cells were subsequently subjected to cell surface biotinylation, as described previously (23), with minor modifications. Following incubation with membrane impermeant EZ-Link biotinylation reagent (Thermo Scientific, Bend, OR, USA) and lysis as described previously, total protein levels were measured. An aliquot of the lysate was saved (total fraction) and 450 μg of each lysate was subsequently incubated with 140 μl streptavidin slurry beads for 1 h on a rocker at 4°C. Following centrifugation, the supernatant was collected, and an aliquot kept as the unbiotinylated fraction. After washing, the beads were resuspended in 50 μl of 2X Laemmli buffer and incubated at room temperature for 30 min. The eluted biotinylated proteins were subsequently collected by centrifugation (biotinylated fraction). The biotinylated fraction (25 μl) was loaded on SDS-PAGE gel for immunoblot analysis along with 2.5 μg of the total fraction and a matched volume of unbound fraction per well. In addition to CHT1, the blots were probed for cytosolic glyceraldehyde-3-phosphate dehydrogenase (GAPDH) to ensure cell membrane integrity was intact during the biotinylation procedure [mouse monoclonal anti-GAPDH 6C5 (1:10 000); Santa Cruz Biotechnology, Dallas, TX], and for Na⁺/K⁺-ATPase as loading control [mouse monoclonal anti Na⁺/K⁺-ATPase H-3 (1:10 000); Santa Cruz Biotechnology, Dallas, TX]. Since the Na⁺/K⁺-ATPase is present in both the unbound and biotinylated fractions, each fraction was normalized for loading only within that particular fraction.

Statistical analysis

A 1-way analysis of variance (ANOVA) with Tukey’s multiple comparison post-hoc test was used to analyze the results of the immunoblot. The half-life experiment was analyzed using a two-tailed unpaired t-test. A 1-way ANOVA with Dunnett’s multiple comparison post-hoc test was used to analyze surface level CHT1 abundance. A 1-way ANOVA and Tukey’s post-hoc test was used to analyze the transport assay results. A 2-way ANOVA was done using Tukey’s multiple comparisons test to analyze the rescue experiments and comparisons done within group using 2-way ANOVA and Šídák’s multiple comparisons test. GraphPad Prism software was used to represent the data. Statistical significance was attributed to a P value ≤ 0.05.

Acknowledgements

We thank Kris MacNaughton and Xiaohua Song for excellent technical assistance. This work has been funded by the generous support of the Stollery Children’s Hospital Foundation through the Women and Children’s Health Research Institute (WCHRI Innovation grant #3134). EC and EML laboratories are supported in part by grants from the Canadian Institutes of Health Research (PS-168871, PJT-159547) and the Natural Sciences and Engineering Research Council of Canada (NSERC; funding reference numbers RGPIN-2017-06432, RGPIN-2017-06154). M.R. received a Sir Frederick Banting and Dr Charles Best Canada Graduate Scholarship-Master’s (CGS-M) from the Canadian Institutes of Health Research; Walter H. Johns Graduate Fellowship; a University of Alberta Faculty of Medicine and Dentistry/Alberta Health Services Graduate Student Recruitment Studentship (GSRS) and an Alberta Graduate Excellence Scholarship (AGES). J.R.Z. received support from the University of Alberta Faculty of Medicine and Dentistry through the HE Bell award; International Research Training Group-Motyl Studentship in Membrane Biology; Doctoral Recruitment Studentship; AGES; Faculty of Medicine and Dentistry 75th Anniversary Award; and the Bell McLeod Educational Fund Graduate Research Prize. G.D.E, T.K.T. and the Pediatric Undiagnosed Diseases Program at New York University School of Medicine were supported by the Jacob Goldfield Foundation and the RTW Charitable Foundation.

Conflict of Interest statement. The authors declare no conflict of interest.

Author notes

Midhat Rizvi, Tina K. Truong, Janet Zhou, Gilad D. Evrony, Elaine M. Leslie and Emmanuelle Cordat authors contributed equally to the work reported in this manuscript.