INTRODUCTION

Asparagine synthetase deficiency (ASNSD; OMIM615574) is an ultrarare and lethal neurodegenerative autosomal recessive disease caused by homozygous or compound heterozygous variants in the asparagine synthetase (ASNS; MIM108370) gene. It is a neurometabolic disorder with significant clinical homogeneity characterized by severe congenital microcephaly, severe developmental delay, axial hypotonia followed by progressive hypertonia, spastic quadriplegia, intractable seizures and hyperreflexia, dysmorphic features, feeding difficulties, and cortical blindness.1 Brain magnetic resonance image (MRI) findings include progressive cerebral atrophy, simplified gyral pattern, delayed myelination, pontine hypoplasia, thin corpus callosum, and ventriculomegaly.

Human asparagine synthetase is an enzyme of 561 residues. It belongs to the family of glutamine amidotransferases type II (GATase) asparagine synthase B type that catalyze the adenosine triphosphate (ATP)–dependent conversion of aspartate to asparagine.2 This enzyme is a homodimer, with each monomer containing both a glutaminase domain (residues 29–160) and an asparagine synthetase domain (residues 381–547). The enzymatic conversion of aspartate to asparagine occurs via three separate reactions: (1) activation of aspartate by formation of an aspartyl-AMP intermediate; (2) hydrolysis of glutamine to glutamate and ammonia, and (3) breakdown of the aspartyl-AMP intermediate via nucleophilic attack by ammonia.3

ASNS is highly expressed in both the adult and developing embryonic brain, particularly in the cortical plate and ventricular layers of the cerebral cortex.4 Asparagine is a nonessential amino acid, required for cell proliferation and early embryonic brain development.5,6 Studies on cultured fibroblasts from ASNSD patients demonstrated reduced cellular proliferation when cultured in asparagine-restricted growth medium compared with fibroblasts from unaffected relatives. This would suggest that de novo synthesis of asparagine is impaired, rendering asparagine an essential amino acid in those patients.6,7 Thus, the dependence of the brain on the de novo synthesis of asparagine, and subsequent cerebral deficiency of this amino acid, largely explains the neurological sequelae associated with this disorder. As such, demonstration of reduced asparagine levels in body fluids, and in particular, cerebrospinal fluid (CSF), has been traditionally used in the identification of patients with ASNSD. However, there are significant challenges associated with biochemical screening for ASNSD, with poor reliability and sensitivity often cited as limitations. Several patients have been reported with unremarkable plasma and/or CSF asparagine levels, limiting the use of biochemical testing as a frontline screening investigation.4,7,8,9,10 In contrast, molecular testing offers a more definitive diagnostic screen for patients presenting with ASNSD-like phenotype, particularly in cases where biochemical data is equivocal. To date, 36 ASNSD families have been reported globally, 12 of whom were Saudi families with homozygous variants reported by our group.9,11,12,13 Given the high rate of consanguinity in Saudi Arabia, it is expected that there is a significant number of additional unreported cases of ASNSD with known or novel variants and clinical features. In the absence of robust biochemical testing, it is imperative that molecular testing is used to identify these individuals. In turn, this will provide data to enable the delineation of the phenotype and direct genetic testing, variants interpretation, and clinical diagnosis of this rare disorder.

The main objectives of this study were to collect and analyze ASNSD patients’ data from large genetic and metabolic centers across the country and characterize them at the clinical, biochemical, and molecular levels.

MATERIALS AND METHODS

Ethics statement

Informed consent for genetic testing and participation in research was obtained from all study participants. The study was approved by the Institutional Review Board (IRB) at Taibah University (MLT-2019-07) and King Fahad Medical City (19–512).

Human subjects

A retrospective review and collation of clinical, biochemical, and molecular data was performed across four large genetic metabolic centers in Saudi Arabia for 13 unpublished Saudi families with an established diagnosis of ASNSD. Informed consent for genetic testing and participation in research was obtained from all study participants. The study was approved by the Institutional Review Board (IRB) at Taibah University (MLT-2019-07) and King Fahad Medical City (19–512).

MRI studies

A neuroradiologist retrospectively reviewed the MRI studies for 13 patients (with the exception of patient F13) and the findings were recorded. Microcephaly was evaluated on MRI and correlated with clinical head circumference measurement. Brain atrophy was evaluated based on the imaging findings of cortical volume loss, reduced white matter bulk, thinning of the corpus callosum, and ventriculomegaly. Myelination was assessed considering the age at the time of imaging. Additionally, MRI images were evaluated for migration and organization malformations of cortical development.

Molecular testing and variants curation

Exome sequencing (ES) and Sanger confirmation for the affected cases were performed in CAP-accredited, in-house, or commercial molecular laboratories as described previously.11,14 Sequence variants were classified and reported by board-certified geneticists following Human Genome Variation Society (HGVS) nomenclature and American College of Medical Genetics and Genomics (ACMG) guidelines.15

Structural modeling of missense variants

Molecular modeling of asparagine synthetase was performed using the automated homology modeling server SWISS-MODEL (http://swissmodel.expasy.org/SWISS-MODEL.html) based on the structure of the E. coli ortholog (PDB: 1CT9; UniProtKB: P22106) as the template.16,17 The sequences were first aligned with the ClustalX program to optimize insertions/deletions.18 The model was visualized and analyzed with the Molmol program.

Statistical analysis

Statistical analysis of biochemical data was performed using Prism version 7.04 (GraphPad Software, San Diego, CA). Plasma and CSF asparagine and glutamine levels were measuring by conventional amino acid methods. We further interrogated CSF data by adjusting asparagine concentration for glutamine concentration and expressing as a ratio. Median CSF asparagine concentrations were compared between ASNSD patients and pediatric controls (n = 637) by Mann–Whitney U test. Similarly, comparison of median values for the CSF glutamine:asparagine ratio was also performed by Mann–Whitney U test.

RESULTS

Known clinical features

Consanguinity was reported in all families. Our study included 13 ASNSD families (31 patients) with median age of 30 months at last follow-up (age range: 0.26–84 months) (Table 1). There were 15 patients with molecular confirmation, of whom 14 had detailed clinical characterization. There were 17 additional relatives with similar history but without detailed clinical characterization and molecular confirmation due to inability to collect DNA samples as a result of early deaths in 76% (13/17) of those individuals (Fig. 1).

Table 1 The demographic, clinical features and molecular results of the Saudi asparagine synthetase deficiency (ASNSD) families included in this study.
Fig. 1: Pedigrees of 13 asparagine synthetase deficiency (ASNSD) families with remarkable history and mortality of ASNSD.
figure 1figure 1

F13 had molecular autopsy by proxy testing of parents due to lack of available DNA from either deceased child.

Congenital or prenatal-onset microcephaly, severe developmental delay, and facial dysmorphism were present in all ASNSD patients for whom clinical information was available (Fig. 2a). Other features included infantile seizures 93% (13/14), progressive spasticity in 86% (12/14), abnormal EEG (for those who have undergone EEG) in 91% (10/11), hyperekplexia (including the probable hyperekplexia cases19) in 36% (5/14), and cortical blindness in 21% (3/14) of the patients. Brain MRI, which was performed for all patients except F13, showed brain atrophy in 100% (13/13) and delayed myelination in 77% (10/13) (Fig. 2b, c). The main cortical developmental finding was simplified gyration and sulcation, present in 38% of the patients (5/13). The cortical developmental malformation was found in only one patient, characterized by focal cortical thickening associated with simplified gyri and sulci (Fig. 2b). Therefore, the penetrance of the brain abnormalities, after excluding patient F13, was 100% (13/13) of the patients who had brain MRI. Infantile or childhood mortality was reported in 71% (22/31) (median age: 21.5 months; age range: 0.26–72 months) of all patients with or without molecular confirmations. An updated review of the clinical features of the current and published cases of ASNSD was conducted to show the total percentage contribution of the individual features to the ASNSD phenotype (Table 2 and S1). Congenital microcephaly, severe developmental delay, brain atrophy, spastic quadriplegia, seizure, and abnormal EEG were reported in most of the patients.

Fig. 2: Clinical features of the asparagine synthetase deficiency (ASNSD) patients.
figure 2

(a) Percentage representation of the main clinical features of ASNSD in this study. (b) Axial T2 magnetic resonance image (MRI) (patient F1) at the age of 7 months shows microcephaly, frontal cortical thickening, and simplified gyri and sulci. There is T2 hyperintense signal in the white matter related to delayed myelination. (c) Axial T1WI and T2WI (patient F2), respectively, at the age of 14 months show microcephaly, severe brain atrophy, and delayed myelination. (d) Ultrasound image of the umbilical region demonstrated (patient F2). Wide neck hernia with 2.5-cm defect in the mid anterior abdominal wall muscles (solid arrows). Small bowel loop herniated in the hernia sac with no complications (dotted arrow). Skeletal survey (patient F11-2) of the (e) upper extremities showed generalized osteopenia, fixed contracture, deformity of the elbow and wrist joints with dislocation of the left distal radioulnar joint; (f) legs showed generalized osteopenia, long bone thinning with contractures and deformities of the knee and ankle joints; and (g) femurs showed anterior bowing of the right femur. There are a plate and screws transfixing the proximal left femur associated with abnormal deformity and bowing of the distal femur. Periosteal reaction is noted at the posterior cortex of distal femur that could be related to healing fracture.

Table 2 Summary of the clinical features of all reported asparagine synthetase deficiency (ASNSD) patients, including those reported in this study.

Additional clinical features in this study

In addition to the known features associated with ASNSD, we have described some of the following previously unreported phenotypes in patients with ASNSD (Table 1).

Umbilical hernia

Found in two patients from different families (F2 and 8) with different ASNS variants. Patient F2 had small reducible umbilical hernia not associated with any skin changes or signs of intestinal obstruction. Abdominal ultrasound in the second patient revealed an anterior abdominal wall defect with evidence of gas-containing bowel loop herniation (Fig. 2d).

Osteopenia

Described in two siblings (F11), with additional feature of pectus carnatum in the index case. Skeletal survey showed generalized reduction of bone density and contractures of the elbow and ankle joins, bilateral deformity of the foot, abnormal deformities and bowing of the distal left and anterior right femurs (Fig. 2e–g).

Eczema

Found in two patients from different families (F8 and 10) with different ASNS variants. However, the eczema was mild and distributed over the cheeks, dorsum of both hands and feet. It was treated by application of topical steroids.

High arched palate

Observed in two patients from different families (F10 and 11) with different ASNS variants.

Lung hypoplasia

Described in one neonate (F13) who had nonreassuring fetal heart monitoring at the time of labor prompting emergency C-section at 36 + 4 weeks of gestation. Reported abnormalities included congenital microcephaly, abnormal brain MRI, dysmorphic features, lung hypoplasia, hypotension, severe edema, and hypoalbuminemia. Despite increasing mechanical ventilation due to frequent desaturation attacks, he died on day 8 of life with respiratory failure.

Lactic acidosis

Observed in one family (F6), likely attributed to the presence of a homozygous loss of function (LOF) variant in the CLPB gene (NM_030813.5: c.654dupA; p.Gln219fs). A homozygous LOF variant in CLPB gene has been previously reported in four newborn siblings with congenital microcephaly, severe encephalopathy, 3-methylglutaconic aciduria, and lactic acidosis.20 Thus, he had dual molecular diagnosis.

Plasma and CSF amino acids analysis

A summary of biochemical data in patients with ASNSD is provided in Table S2. Plasma asparagine (and other relevant amino acids concentrations), from this and previous studies, were collected for 21 patients with ASNSD. The results are summarized in Fig. S1a. Of 21 patients, 9 had decreased plasma asparagine concentrations compared with local population reference intervals, and a further 3 patients had low/normal values. From the data it may be inferred that the sensitivity of plasma asparagine as an indicator of ASNSD, even in fasting samples, is limited. Unsurprisingly, CSF asparagine levels represent a more reliable marker of ASNS deficiency. In the ASNS patients (n = 8) where CSF data were available, the majority (75%) had low or low/normal concentrations. Comparison of median CSF asparagine concentration (0.5; 95% Confidence Level (CL): 0.0–5.0) with 637 unaffected controls (7.1; 95% CL: 6.8–7.6) demonstrated statistical significance (p < 0.0001) at the 0.01 level. Fig. S1b shows CSF asparagine concentrations in six patients expressed as a function of the glutamine:asparagine ratio. Similarly, comparison of the median CSF glutamine:asparagine ratios in affected patients (544; 95% CL: 128–922) versus controls (median: 71; 95% CL: 67–75) also demonstrated statistical significance (p < 0.0001) at the 0.01 level. These results indicate that the majority of ASNS patient values are clustered in a small region bordered by the 5th and 95th percentiles for asparagine (1.7 mcmol/L) and glutamine:asparagine ratio (477), respectively (Fig. S1b). If we used these criteria as screen positive cut-off values, this translates to a diagnostic sensitivity of 83% (95% CL: 36–99) and specificity of 98% (95% CL: 96–99) for ASNS deficiency.

Molecular testing and genotype–phenotype correlation

Most of the index cases underwent chromosomal analysis (karyotype and chromosomal microarray [CMA]) and ES performed stepwise as genetically indicated. We detected seven homozygous variants in the ASNS gene (Table S3). Three variants were LOF variants (two splicing and one nonsense) and four were missense variants. All of the variants were previously detected in Saudi ASNSD patients, with the exception of the p.Thr475Asn and c.674-1G>A variants, which are novel in this study (Fig. 3a). All variants are clustered in the highly conserved asparagine synthetase domain of ASNS protein, except for the p.Asn75Ile variant, which is located in the glutamine aminotransferase type II domain (Fig. 3c).

Fig. 3: Variants landscape and predicted structural impact.
figure 3

(a) Variants detected to date in Saudi  asparagine synthetase deficiency (ASNSD) families. (b) Mapping the clinically important variants on the surface of asparagine synthetase. The side chains of the ASNS variants p.Asn75Ile, p.Tyr398Cys, p.Arg404His, and p.Phe452Ile are depicted in red; those of the residues that form interactions with these residues are shown in light blue. The side chain of Arg340 previously reported for a variant to an alanine is shown in green. The position of the loop, which should host Thr475, is shown in orange. AMP is shown in green in full atoms representation. (c) Schematic diagram of the functional domains of ASNS protein (the longest isoform, 561 AAs) along with the corresponding position of all reported variants. The red-colored variants represent the Saudi-specific variants.

The p.Tyr398Cys and p.Asn75Ile variants account for ASNSD in 31% (4/13) and 23% (3/13) of the families in this study, respectively. These variants also account for 32% (8/25) and 24% (6/25) of the total Saudi families, respectively, including those reported (Fig. 3a). Haplotype analysis of the overlapping region of homozygosity (ROH) in patients with the p.Tyr398Cys variant supported its founder nature (data not shown). The ROH data were not available for the other variants.

Molecular modeling data suggest that the variants occurred at buried or semiburied residues distant from the dimer interface (Fig. 3b and Supplemental Information). Asp54 is a semiexposed residue close to the surface. The conserved Asn75 is positioned in the interface between the glutaminase domain and the asparagine synthetase domain, and forms an H-bond with the conserved Arg27, which is predicted to have a stabilizing effect on the protein fold. A variant changing Tyr398 to cysteine could potentially introduce a reactive thiol residue, able to cross-link to form intra- or intermolecular disulfide bonds with other cysteine residues. Arg404 is close to the AMP molecule of the active site and is predicted to have a role in the functionality of the active site itself. Trp452 is deeply buried in a narrow pocket. Finally, Thr475 is positioned in the loop where it may have a stabilizing role in the confirmation of AMP.

According to ACMG criteria, these missense variants are classified as pathogenic or likely pathogenic based on their complete segregation with this single-gene disease, conservation across species, low-allele frequency in controls, biochemical effect, and predicted impact by molecular modeling (Table 1 and Table S24).

The pathogenic homozygous LOF variants (p.Arg407*, c.674-1G>A and c.1137+1G>A) were detected in four patients; three died at the age of 3 months or earlier, and the other died at the age of 3.6 years (Table 1). The p.Thr475Asn variant is likely a milder variant, detected in four patients (F10 and 11) who are still alive at ages 2.2, 5, 1.8, and 7 years, respectively. This prediction is supported by molecular modeling studies, which showed that this substitution resulted in relatively milder perturbation of the conformation of the loop.

DISCUSSION

Asparagine synthetase deficiency is reported in 36 families. Of the reported families, we contributed all of the Saudi ASNSD families (12/36). This is the largest study of this rare and neurodegenerative disease. In this study, we reported an additional 13 Saudi ASNSD families to expand the total number of ASNSD families to 49, of whom 25 are Saudi families. The p.Tyr398Cys and p.Asn75Ile variants account for ASNSD in 56% of the Saudi families. More importantly, we have contributed to expansion of the clinical phenotype, established the ASNS variants landscape, and determined the utility of biochemical testing as a frontline investigation in ASNSD.

ASNSD is a clinically homogeneous neurological disease.1 In our study, congenital microcephaly, global developmental delay, facial dysmorphism, and brain abnormalities were present in all ASNSD patients for whom clinical information and imaging data were available. The majority of the patients had infantile seizures, progressive spasticity, and abnormal EEG. Recently, diaphragmatic eventration, a rare additional feature of ASNSD, was reported in siblings with ASNSD.21 We also reported lissencephaly (four families), strabismus (two families), and bilateral microphthalmos (one family) in association with ASNSD.13 In this study, we have observed additional clinical features of umbilical hernia, osteopenia, eczema, high arched palate, lung hypoplasia, and bilateral hearing loss. Some of these features could be incidental and not related to ASNSD. However, most of these features are present in more than one family and were associated with different ASNS variants. Additionally, chromosomal analysis and ES in these patients revealed no additional variants in other known or candidate genes to account for the additional phenotypes. Taken together, the additional clinical features may represent a phenotypic expansion of ASNSD. Future studies are warranted to investigate the association of the additional phenotypes with ASNSD.

A total of 35 ASNSD-causing variants have been reported (Table S4). The majority of these variants are missense variants, clustered in the asparagine synthetase domain (Fig. 3c). Of the ASNSD-causing variants, we previously reported seven homozygous variants in 12 families. In this study, in addition to other reported variants, we described two novel homozygous variants that are classified as pathogenic according to the ACMG guidelines given the available clinical, segregation, frequency, molecular, biochemical, and prediction data. Of the nine Saudi ASNS variants, eight variants have never been reported in families from outside the Arabian Peninsula. Consistently, the p.Tyr398Cys variant, which accounts for 32% of ASNSD in the Saudi population, has been previously reported in an Emirati family.8 Indeed, haplotype analysis supported the founder nature of the p.Tyr398Cys variant. It is noteworthy that point variants, small deletions, and duplications account for 97% of all ASNSD cases (Fig. 3c and Table S4). Therefore, gene sequencing–based assays are recommended to investigate cases with ASNSD-like phenotype.

In inborn errors of metabolism, biallelic LOF variants resulting in absent or truncated enzymes are usually associated with more severe clinical phenotype and poorer outcome compared with the missense variants. There is insufficient genotype–phenotype correlation in ASNSD, especially in nonconsanguineous families where biallelic LOF variants are never reported.1 However, we previously reported the homozygous LOF p.Arg407* variant in an ASNSD patient who died at 6 weeks of age, and had family history of four maternal aunts who died at ages 4, 5, 3, and 6 weeks, respectively.9 Another homozygous LOF variant, p.Lys66fs, was reported in a Yemeni patient who died at age 3 months from respiratory distress (ACMG poster, https://epostersonline.com/acmg2016/node/1304). In our study, we reported three LOF variants, one of which is p.Arg407*, in three patients: two died at ages 3 months, and the patient with the p.Arg407* variant died of respiratory failure at age 8 days. In contrast, the p.Thr475Asn variant is present in four patients who are still alive at ages 2.2, 5, 1.8, and 7 years, respectively. Therefore, the neonatal or prenatal lethality due to biallelic LOF variants in ASNS might be underdiagnosed and might explain the underrepresentation of the LOF variants compared with the missense variants in ASNSD patients (Fig. 3c).

Supplementation of the deficient amino acid has proven effective in amino acid synthesis deficiencies (e.g., serine biosynthesis disorders and glutamine synthetase deficiency).21,22,23 However, our attempts to correct low cerebral asparagine levels through asparagine supplementation in two siblings with the p.Tyr398Cys variant resulted in increasing frequency and duration of seizures.12 In contrast, a recent study described asparagine supplementation in two siblings with ASNSD and low CSF asparagine levels. The results showed that supplementation was well tolerated, and halted disease progression in both patients. One of the patients also demonstrated improvements in cognition and nonverbal communication.10 However, one of the improved patients suffered from congenital microcephaly, which means that microcephaly has started in utero and is irreversible if treatment is initiated postnatally. Therefore, with the small sample size of only two patients and our previous study that does not support a beneficial but rather an adverse outcome of asparagine supplementation, it is very difficult at this stage to conclude that asparagine supplementation is beneficial for this disease. Functional animal studies and clinical trials of sufficiently large sample size are warranted to evaluate the efficacy of asparagine supplementation.

Our study also demonstrated the limitations associated with biochemical screening for ASNSD. The limitations of plasma asparagine analysis are well recognized, particularly the difficulties associated with obtaining fasting samples. Although CSF asparagine analysis is more useful, a significant number of ASNSD patients present with asparagine levels within the reference intervals.4,9,10 There may be more utility in expressing CSF asparagine as a ratio with other amino acids, such as glutamine. Using this approach, we determined the sensitivity of CSF analysis to be approximately 83%, which is not ideal for a screening marker. Additionally, the invasiveness of obtaining spinal fluid is another important factor to be considered. Rather, direct assessment of ASNS enzyme activity as a biochemical indicator of deficiency may have a role in both screening and response to treatment in patients with suspected or reestablished disease. In contrast to biochemical screening, molecular analysis provides a definitive, noninvasive, cost-effective alternative as a frontline investigation.

Further studies are warranted to uncover the mechanism by which the variants in ASNS gene fully explain the phenotype associated with ASNSD. Prenatal-onset microcephaly is well established in ASNSD; therefore asparagine supplementation in families with history of ASNSD warrants further experimental investigations as the case for serine deficiency to rescue or mitigate the congenital central nervous system (CNS) and other extraneurological features of the disease.4,9,24 Gene therapy–based approaches can be considered as alternative means of treatment to asparagine supplementation for those with milder variants, such as the p.Thr475Asn variant, to account for and rescue the known and possibly unknown functions of the ASNS enzyme.

A limitation of this study is that it is a retrospective observational study; therefore, the selection and information biases are high. Another limitation is the lack of detailed clinical features for 16 of the relatives. This is mainly because of the high mortality and early infantile death rate among our ASNSD patients (71% [22/31]). In many cases, the family history of similar disease and early deaths triggered the first visit for the index cases to the clinic. An additional limitation is the lack of biochemical data for most of the patients in this study due to the early deaths before the genetic results or the rejection of the CSF sampling by the families. However, to mitigate this issue, we analyzed the results after we combined our data with the published biochemical data for the ASNSD patients.

In conclusion, this is the largest ASNSD study to date, which expands the number of published ASNSD patients, clinical features, biochemical utility, and variant spectrum associated with ASNSD. The findings presented in this study have several clinical values. First, knowing the full spectrum of the disease, including the rare clinical features, helps to narrow the differential diagnosis and to follow an efficient pathway for the diagnosis of patients with ASNSD presentations. Second, it provides the ASNS variant landscape for particular populations such as the Saudi population, with the supporting clinical, genetic, and functional data for accurate variant classification. This is critical to shorten the diagnostic odyssey and reach accurate and timely diagnosis, and to provide the proper genetic counseling for reproductive options to avoid the recurrence of this lethal genetic disease. For the Saudi population, these findings should be of great value to the comprehensive premarital screening program to be launched in the country in the near future.