Review of Congenital Myasthenic Syndrome Caused by Pathogenic Variants in GFPT1

ABSTRACT

Glutamine:fructose‐6‐phosphate transaminase 1 (GFPT1) catalyzes the first and rate‐limiting step of the hexosamine biosynthetic pathway (HBP) to generate UDP‐GlcNAc. GFPT1 exon 9 is specifically spliced in in striated muscles, which makes a long isoform of GFPT1 (GFPT1‐L). In contrast, a short isoform (GFPT1‐S) is generated in the other tissues. GFPT1‐L was likely acquired in evolution to suppress the HBP in striated muscles to flow more glucose into the glycolytic pathway. Loss‐of‐function variants of GFPT1 cause limb‐girdle congenital myasthenic syndrome (CMS). A total of 146 patients in 115 pedigrees with GFPT1‐CMS have been reported with 71 pathogenic variants. The mean age of onset was 8.3 ± 10.1 years (range 0 to 69 years). Limb‐girdle muscle weakness, tubular aggregates in muscle biopsy, and elevated serum CK were observed in 100%, 66.7%, and 42.5%, respectively. Involvements of palpebral, extraocular, facial, bulbar, and respiratory muscles were rare and were observed in less than one in seven patients. Pyridostigmine, amifampridine, and salbutamol were effective in 95.8%, 76.3%, and 86.4%, respectively. Mechanistic studies show that hypoglycosylation of the acetylcholine receptor δ subunit is likely to be a key for defective acetylcholine receptor clustering. In mouse models, complete lack of GFPT1 in skeletal muscle developed CMS in 6 weeks of age, whereas lack of GFPT1‐L in skeletal muscle required 12 months to develop CMS, which was likely to be accounted for by the expression of a low level of GFPT1‐S. In accordance with this notion, null variants were enriched in exon 9 in GFPT1‐CMS.

1. GFPT1 and GFPT2 in the Hexosamine Biosynthetic Pathway (HBP)

Congenital myasthenic syndromes (CMS) are heterogeneous disorders caused by defective neuromuscular signal transmission, and are characterized by fatigable muscle weakness, muscle hypoplasia, and minor anomalies. Pathogenic variants have been identified in 40 genes [1], and GFPT1 encoding the glutamine:fructose‐6‐phosphate transaminase 1 (GFPT1) is one of the causative genes.

Glucose transported into the cells flows into three pathways. First, the glycolytic pathway generates adenosine triphosphate (ATP) and reduced nicotinamide adenine dinucleotide (NADH). Second, the pentose phosphate pathway generates ribose 5‐phosphate for nucleotide biosynthesis and NADPH. Third, the hexosamine biosynthetic pathway (HBP) generates uridine diphosphate N‐acetylglucosamine (UDP‐GlcNAc), which is a key molecule to make N‐ and O‐linked glycosylation of glycoproteins, as well as to make glycosaminoglycans and glycolipids (Figure 1A) [2, 3]. Both the N‐ and O‐linked glycosylation is essential for the maintenance and function of skeletal muscle. The rate of glucose consumption by HBP in primary adipocytes obtained from rat epididymal fat was as low as 2%–3% [4]. UDP‐GlcNAc also modulates signaling pathways by O‐GlcNAcylation of signaling molecules, and is a metabolic regulator for both stress response and nutrient sensing [5, 6].

FIGURE 1.

The hexosamine biosynthetic pathway (HBP) and genomic/domain structures of GFPT1: (A) GFPT1 and GFPT2 constitute the first and rate‐limiting step of the HBP to generate UDP‐GlcNAc that is an essential substrate for O‐GlcNAcylation and N‐glycosylation pathways. (B) Genomic/domain structures of human GFPT1. Alternatively, spliced exon 9 is included in the skeletal and heart muscles. (C) Monomeric presentation of the human GFPT1 dimer that was crystalized with glucose‐6‐phosphate and glutamine (PDB: 6R4E) [11].

The first and rate‐limiting step of the HBP is mediated by glutamine:fructose‐6‐phosphate transaminase (GFPT). Mammals carry paralogous GFPT1 and GFPT2, which are 76% identical at the amino acid level in both humans and mice. According to the GTEx project [7], both GFPT1 and GFPT2 are ubiquitously expressed in all tissues at variable levels in human [8, 9]. In human skeletal muscle, the expression level of GFPT1 is 5.20 times higher than that of GFPT2. The enzymatic activity of GFPT2 is lower than that of GFPT1, but it is less sensitive to the feedback inhibition of UDP‐GlcNAc [10]. Similar to GFPT1, GFPT2 is increased in breast, colon, pancreas, and non‐small‐cell lung cancers [3], but GFPT2 has been less studied compared to GFPT1.

GFPT1 has three domains: the glutamine amidotransferase type‐2 domain (GATase), the sugar isomerase domain 1 (SIS1), and the sugar isomerase domain 2 (SIS2) (Figure 1B). GFPT1 makes a homodimer (Figure 1C) [11]. The GATase domain catalyzes the hydrolysis of glutamine to glutamate and free ammonia (NH₃), whereas the SIS domains catalyze the conversion of fructose‐6‐phosphate (F‐6‐P) to glucosamine‐6‐phosphate (GlcN‐6‐P) using ammonia (NH₃) channeled from the GATase domain (Figure 1A) [2].

2. Striated Muscle‐Specific GFPT1‐L Was Likely to Be Acquired in the Course of Mammalian Evolution to Flow More Glucose into the Glycolytic Energy Production Pathway

Alternative splicing of GFPT1 exon 9 generates a short ubiquitous isoform (GFPT1‐S) and a long striated muscle‐specific isoform (GFPT1‐L) (Figure 1B) [12, 13]. In humans and mice, skeletal muscle almost exclusively expresses GFPT1‐L, whereas the heart muscle expresses both GFPT1‐L and GFPT1‐S. In contrast, the other tissues exclusively express GFPT1‐S [12, 13]. GFPT1 exon 9 encodes 18 amino acids (54 bp) in humans and 16 amino acids (48 bp) in mice. Compared to GFPT1‐S, V_MAX, K_M of fructose‐6‐phosphate, and K_i of UDP‐GlcNAc of GFPT‐L are 53% [13], 215% [12], and 20% [12], respectively. Thus, GFPT1‐L has a lower enzymatic activity and a higher feedback inhibition by UDP‐GlcNAc compared to GFPT1‐S.

RNA‐binding protein, the serine and arginine rich splicing factor 1 (SRSF1) and the RNA binding fox‐1 homology 1/2 (Rbfox1/2), bind to GFPT1 exon 9 and intron 9, respectively, and cooperatively enhance the inclusion of GFPT1 exon 9 by reinforcing the recruitment of U1 snRNP at the 5′ splice site [14]. In contrast, another RNA‐binding protein, hnRNP H/F, binds to GFPT1 exon 9, and suppresses the inclusion of GFPT1 exon 9. In accordance with their roles in splicing of GFPT1 exon 9, Rbfox1/2 increases and hnRNP H/F decreases with myotube differentiation.

Knockout (KO) of GFPT1 exon 9 markedly increased the protein level of GFPT1‐S and UDP‐GlcNAc/UDP‐GalNAc, which subsequently impaired glycolytic energy production and induced abnormal formation and maintenance of the neuromuscular junction (NMJ) [14]. This suggested that GFPT1 exon 9 was likely to be acquired in the course of evolution in mammalian striated muscles to attenuate the HBP for efficient glycolytic energy production.

3. Pathogenic Variants Causing GFPT1 ‐CMS

Pathogenic variants of GFPT1 were first reported in 22 patients with limb‐girdle CMS with tubular aggregates in 2011 [15]. A total of 78 pathogenic variants comprised 68 single nucleotide variants and 10 indels have been reported in CMS (Table S1). Out of the 78 variants, 34 were recurrent variants with multiple reports. Especially, c.44C>T, c.*22C>A, and c.331C>T have been reported 7, 8, and 9 times, respectively. The 78 variants were classified into 55 missense variants, eight splicing variants, seven frameshifting variants, six nonsense variants, and one inframe variant, as well as one variant at the 3′ untranslated region (UTR).

Compared to patients with only missense variants, patients with nonsense, frameshift, splicing, gross deletion, or initiation codon variants showed significantly earlier onset, a higher proportion of females, and more involvement of bulbar muscles [16]. The other reports also support the notion that patients with homozygous missense variants show mild phenotypes [15, 17, 18]. In contrast, biallelic loss‐of‐function variants affecting muscle‐specific GFPT‐L show a severe phenotype including apnea, severe hypotonia, and multiple joint contractures [17, 19].

No clustering of pathogenic variants was observed in any domain. Similarly, no phenotypic difference was noted between patients with and without variants in the GATase domain [16]. Out of the eight variants in muscle‐specific exon 9, seven were frameshifting or nonsense variants and one was a missense variant. The reason for the enrichment of null variants in exon 9 will be discussed below.

4. Clinical and Therapeutic Features of GFPT1 ‐CMS

A total of 146 patients in 115 pedigrees have been reported to date, although a single patient may be counted in duplicate if the patient was presented in multiple reports (Table 1). The ages of onset were 8.3 ± 10.1 years (mean and SD) ranging from 0 to 69 years (Figure 2). Neonatal onset was reported in 10.4% of the cases. In contrast, the ages of onset over 40 years have also been reported [20, 21], but environmental or genetic factors that delayed the disease development remain elusive.

TABLE 1.

Clinical Features of 146 patients in 115 pedigrees with GFPT1‐CMS.

	Pedigrees	Patients	Intellectual disability	Ptosis	EOM	Facial	Bulbar	Limb girdle	Respiration	Decrement in RNS	Elevated CK	Tubular aggregates	Pyridostigmine	Amifampridine	Salbutamol	References
1	14	24	2/24	1/24	0/24	5/24	0/24	24/24	1/24	18/22	8/23	13/18	20/22	9/9	—	[15, 30]
2	1	1	0/1	0/1	0/1	0/1	0/1	1/1	—	1/1	1/1	1/1	1/1	—	—	[33]
3	11	11	0/11	0/11	0/11	0/11	2/11	11/11	2/11	11/11	1/11	7/9	10/11	5/11	1/1	[17]
4	1	3	0/3	0/3	0/3	1/3	0/3	11/11	0/3	0/1	1/1	0/1	3/3	1/1	—	[46]
5	1	1	—	—	—	—	—	—	—	—	1/1	0/1	—	—	—	[24]
6	1	1	—	0/1	0/1	—	—	1/1	—	—	—	—	—	—	—	[50]
7	9	11	0/11	0	4/11	0/11	1/11	11/11	0/11	11/11	7/11	6/6	9/9	9/9	—	[31]
8	5	8	0/8	0/8	0/8	0/8	0/8	8/8	—	—	—	—	8/8	4/4	—	[27]
9	1	1	0/1	0/1	0/1	0/1	0/1	1/1	0/1	—	1/1	0/1	—	—	—	[26]
10	3	3	0/0	0/3	0/3	0/3	0/3	3/3	0/3	3/3	—	—	1/1	—	—	[51]
11	2	4	0/4	0/4	0/4	0/4	0/4	4/4	4/4	2/2	0/2	—	2/2	—	—	[52]
12	1	2	—	0/2	0/2	0/2	0/2	2/2	—	2/2	1/2	1/1	2/2			[53]
13	1	1	0/1	0/1	0/1	0/1	0/1	1/1	1/1	1/1	0/1	1/1	1/1	—	—	[32]
14	2	3	1/3	1/3	1/3	1/3	1/3	3/3	3/3	2/2	0/1	0/3	2/2	1/3	—	[19]
15	1	3	—	—	—	—	—	—	—	—	—	—	—	—	—	[54]
16	3	3	—	—	—	—	—	2/2	—	—	0/2	—	—	—	—	[55]
17	1	1	0/1	1/1	0/1	0/1	0/1	1/1	0/1	1/1	1/1	1/1	1/1	—	1/1	[34]
18	3	3	0/3	0/3	0/3	0/3	0/3	3/3	0/3	3/3	1/3	3/3	3/3	—	—	[56]
19	2	2	—	—	—	—	—	—	—	—	—	2/2	—	—	—	[28]
20	1	1	—	—	—	—	—	1/1	—	1/1	—	1/1	1/1	—	—	[57]
21	2	2	2/2	0/2	0/2	0/2	0/2	2/2	0/2	2/2	0/2	1/2	2/2	—	1/2	[29]
22	1	3	0/3	1/2	0/3	0/3	3/3	3/3	3/3	—	1/2	1/2	—	—	—	[18]
23	1	1	0/1	0/1	0/1	0/1	0/1	1/1	0/1	1/1	0/1	1/1	0/1	—	—	[21]
24	1	1	0/1	0/1	0/1	0/1	0/1	1/1	0/1	1/1	1/1	1/1	1/1	—	—	[20]
25	11	15	—	—	—	—	—	15/15	1/15	—	7/12	7/9	13/14	10/11	6/7	[47]
26	10	10	0/10	4/10	4/10	—	0/10	10/10	0/10	—	—	—	10/10	—	10/10	[58]
27	1	1	0/1	0/1	0/1	0/1	0/1	1/1	0/1	1/1	0/1		1/1	0/1	0/1	[35]
28	2	2	0/2	0/2	0/2	1/2	0/2	2/2	0/2	2/2	2/2	0/1	2/2	—	—	[59]
29	22	24	0/24	5/24	3/24	5/24	3/24	24/24	0/24	21/21	10/17	6/13	22/22	—	16/17	[16]
Total	115	146	5/115	13/109	12/121	13/110	10/120	139/139	15/124	84/89	44/99	53/78	115/120	39/49	35/39
Ratio	—	—	4.3%	11.9%	9.9%	11.8%	8.3%	100.0%	12.1%	94.4%	44.4%	67.9%	95.8%	79.6%	89.7%

Note: Ratios represent the number of patients with indicated symptoms, response to drug, and laboratory findings divided by the total number of patients. Hyphens indicate no description. Articles are sorted in chronological order.

FIGURE 2.

Ages of onset of 106 patients with GFPT1‐CMS. The ages of onset are plotted in 5‐year intervals except for the neonatal period.

Limb‐girdle muscle weakness, tubular aggregates (TAs) in muscle biopsy, and elevated CK were documented in 100%, 67.9%, and 44.4% of cases, respectively (Table 1). Among CMS, serum creatine kinase (CK) is elevated ~1.5‐fold in slow‐channel CMS, ~3‐fold in GFPT1‐CMS, and ~10‐fold in GMPPB‐CMS [22, 23]. Collation of laboratory findings of GFPT1‐CMS patients in this communication also underscored the elevated serum CK levels, which, however, were not observed in about half of the patients. Intellectual disability was rare (4.3%). The involvements of palpebral, extraocular, facial, bulbar, and respiratory muscles were 11.9%, 9.9%, 11.8%, 8.3%, and 12.1%, respectively. Thus, GFPT1‐CMS always exhibits limb‐girdle muscle weakness and rarely involves eye and bulbar muscles. In addition to the clinical features collated in Table 1, patients with GFPT1‐CMS sometimes show scoliosis [19, 24, 25, 26, 27, 28, 29], pes cavus [19, 26], scapular winging [16, 30, 31, 32], joint contractures [17, 25, 33], and high arched palate [16, 19], and rarely show hyperextension of distal joints [34], agenesis of the pectoralis major muscle [20], pectus carinatum [35], and cranial synostosis [17].

Pyridostigmine was effective in 95.8% of the patients, which unnecessitated the use of the other drugs. Indeed, amifampridine and salbutamol were less frequently used, but were effective in 79.6% and 89.7% of the patients, respectively. These three drugs are generally effective in glycosylation‐deficient CMS comprised of GFPT1‐CMS, DPAGT1‐CMS, ALG2‐CMS, ALG14‐CMS, and GMPPB‐CMS [1, 36].

5. Mechanistic Studies of GFPT1 ‐CMS

Gftp1‐deficient zebrafish showed curled and shortened tails, severely impaired swimming and touch‐evoked escape responses, abnormal somites, and delayed NMJ development [15]. These findings underscored the roles of GFPT1 at the NMJ.

Patient‐derived myoblast/myotubes showed significantly reduced cell‐surface expression of acetylcholine receptor (AChR) [37]. Similarly, reduced AChR expression was also observed in GFPT1 inhibitor‐treated or Gfpt1‐silenced TE671 muscle cells. The reduced expression of AChR was attributed to the decreased steady‐state levels of the AChR α, δ, and ε subunits, which were likely due to defective N‐linked glycosylation [37].

Mice with muscle‐specific knockout of Gfpt1 showed fatigable muscle weakness and tubular aggregates in muscle fibers, as well as fragmentation and reduction in size of AChR clusters at the neuromuscular junction (NMJ) [38]. Although Gfpt1 was knocked out only in skeletal muscles, presynaptic abnormalities including disorganized axons, thinner irregular myelin sheaths, and remodeling of motor nerve terminals were observed. The mouse muscle also had accumulated subsarcolemmal vesicular structures suggesting ER‐Golgi stress. Proteomic analysis of affected muscles showed the upregulation of proteins involved in NMJ differentiation and maintenance. Analysis of the same knockout mice showed that AChR δ subunit but not the other AChR subunits was hypoglycosylated [39], which was likely to account for the reduced cell‐surface expression of AChR observed in patient‐derived myoblasts/myotubes [37].

We generated a knock‐in mouse model carrying a frameshifting variant in Gfpt1 exon 9 that was observed in a patient with GFPT1‐CMS [40]. In contrast to the complete lack of GFPT1 in skeletal muscle in the knockout mice stated above [38, 39], the knock‐in mice expressed GFPT1 to 20% of wild‐type mice, because a small amount of GFPT1‐S could be generated. The mice showed markedly decreased UDP‐GlcNAc/GalNAc, CMP‐NeuAc, and protein O‐GlcNAcylations. At age 12 months, the mice exhibited tubular aggregates in muscle fibers, poor exercise performance, fragmented AChR clusters, and simplified NMJ ultrastructures. At age 6 months, Hsp70 was increased in skeletal muscle, but Grp78‐p62 colocalization was not observed indicating the adaptive unfolded protein response (UPR). In contrast, at age 12 months, Grp78‐p62 colocalization was abundantly observed in skeletal muscle and the apoptosis markers of Chop and Bax were induced, both of which indicated maladaptive UPR. Prolonged existence of ER stress due to lack of UDP‐GlcNAc was likely to switch UPR from adaptive to maladaptive with aging. Complete lack of GFPT1‐S/L [38] and 20% expression of GFPT1‐S [40] were likely to account for the difference in the ages of disease onset in the two mouse models. The enrichment of null variants in exon 9 stated above may represent that missense variants but not null variants in exon 9 can be compensated for by the expression of a low level of GFPT1‐S in skeletal muscle.

TAs present in 67.9% of GFPT1‐CMS patients (Table 1) are comprised of densely packed membranous tubules originating from the sarcoplasmic reticulum (SR) [41]. In addition to GFPT1‐CMS, TAs are observed in other glycosylation‐deficient forms of CMS (DPAGT1‐CMS and ALG2‐CMS but not ALG14‐CMS or GMPPB‐CMS) [36], as well as in muscle diseases associated with abnormal calcium handling including tubular aggregates myopathy (TAM) due to pathogenic variants in STIM1 and ORAI1 and periodic paralysis due to pathogenic variants in CACNA1S [31]. The stromal interaction molecule 1 (STIM1) and the ORAI calcium release‐activated calcium modulator 1 (ORAI1) are glycosylated SR proteins in the store‐operated Ca²⁺ entry (SOCE) system that enables Ca²⁺ uptake by the SR when the SR Ca²⁺ level is reduced [42, 43]. Hypoglycosylated and functionally deficient STIM1 and ORAI1 may result in a Ca²⁺ shift from the SR to the cytosol, which subsequently causes TAs [17, 44, 45]. As indicated above in our GFPT1‐L knock‐in mice, a maladaptive response to ER stress due to hypo‐glycosylation may also account for the formation of tubular aggregates [40].

A pathogenic variant at the 3′ UTR (c.*22C>A) has been reported in 8 articles [15, 17, 20, 27, 30, 31, 46, 47]. Dusl and colleagues showed that c.*22C>A created a new binding site for miR‐206*, which reduced the translation of GFPT1 to 66% of wild‐type [48]. In addition, anti‐miR‐26* treatment substantially rescued GFPT1 expression levels in patient‐derived myoblasts. As miR‐206* is abundantly expressed in skeletal muscle, the translation of GFPT1 mRNA is efficiently suppressed in skeletal muscle.

Similar to the HBP, the Leloir pathway catalyzes the conversion galactose to UDP‐GalNAc. The generated UDP‐GalNAc can be converted to UDP‐GlcNAc via the UDP‐GalNAc epimerase (GALE). Administration of galactose in skeletal muscle‐specific Gfpt1‐deficient mice rescued neuromuscular deficits, improved muscle fatigue, restored NMJ morphology, and increased protein O‐GlcNAcylation in skeletal muscle [49]. Thus, galactose is expected to be a promising therapeutic option for patients with GFPT1‐CMS.

6. Conclusions

GFPT1‐CMS primarily presents with limb‐girdle muscle weakness. It typically spares the ocular, bulbar, and facial muscles. The combination of limb‐girdle muscle weakness, mildly elevated serum CK levels, and tubular aggregates on muscle biopsy possibly leads to misdiagnosis as other myopathies [18, 35]. However, decremental responses to repetitive nerve stimulation and favorable responses to pyridostigmine, amifampridine, and salbutamol are crucial indicators for including GFPT1‐CMS in the differential diagnosis.

Author Contributions

Kinji Ohno: conceptualization and writing final draft. Mohammad Nazim: splicing analysis of GFPT1. Ruchen Zhang: analysis of lack of muscle‐specific GFPT1 transcript. Paniz Farshadyeganeh: analysis of knockout of muscle‐specific exon of GFPT1. Mikako Ito: supervision of GFPT1 projects. Bisei Ohkawara: supervision of GFPT1 projects.

Ethics Statement

We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Pathogenic variants in GFPT1 causing CMS.

Ohno K., Nazim M., Zhang R., Farshadyeganeh P., Ito M., and Ohkawara B., “Review of Congenital Myasthenic Syndrome Caused by Pathogenic Variants in GFPT1 ,” Muscle & Nerve 74, no. S1 (2026): S58–S64, 10.1002/mus.70065.

Data Availability Statement

No new data were generated in this review.