Expanding the phenotype of CARS1 variants to include congenital hyperinsulinism

Abstract

Background

See More

CARS1 loss of function compound heterozygous or homozygous variants have been reported in five individuals to cause a neurodevelopmental phenotype that includes microcephaly and brittle hair and nails. Additional multisystem involvement in these five people have included neurologic, cardiac, ophthalmologic and endocrine problems.

Case presentation

We report a sixth person with novel compound heterozygous variants in CARS1. In addition to the previously reported features such as intellectual disability, neurologic features, microcephaly and hair abnormalities, this patient had persistent hypoglycemia due to congenital hyperinsulinism.

Conclusions

This report identifies two novel variants in CARS1 and expands the phenotype of this multisystem disorder to include congenital hyperinsulinism.

Background

CARS1 gene (previously known as CARS, CYSRS, OMIM 123859) codes for cysteinyl-tRNA synthetase 1. This aminoacyl tRNA synthetase is responsible for pairing cysteine to its cognate tRNA^Cys as part of protein translation [1]. Cysteinyl-tRNA synthetase 1 has also been demonstrated to act as a cysteine persulfide synthase [2]. Cysteinyl-tRNA synthetase is one of 35 aminoacyl tRNA synthetases that have been implicated in human disease [3]. Several different neurologic phenotypes have been reported in people with variants in CARS1. A report by Turner et al. in 2019 identified a de novo heterozygous CARS1 variant (p.Asn348Ser) in a male with autism in a large screening study looking for de novo variants in people with neurodevelopmental disorders [4]. In 2024 a heterozygous variant in CARS1 (p.Glu795Val) was reported to segregate in an autosomal dominant pattern with Parkinsonism-spinocerebellar ataxia in nine individuals of a large family [5]. However, the largest report to date of variants in CARS1 leading to human disease was by Kuo et al. [6], where they described four patients with compound heterozygous or homozygous loss of function variants in CARS1 causing an apparently autosomal recessive neurodevelopmental syndrome (OMIM 618891, Microcephaly, Developmental Delay, and Brittle Hair Syndrome) [6]. The four patients all had developmental delay, microcephaly and brittle hair and nails. Additional medical problems found in several of the reported patients included brain abnormalities (including prominent lateral ventricles and sulci; mild cerebral atrophy; apparent delayed myelination; decreased white-matter volume; prominence of cerebellar folia; small corpus striatum; hypoplasia of the corpus callosum; mild, globally diffuse cerebral and cerebellar atrophy; hemiatrophy; wide right ventricle), seizures, neurological problems including hypotonia, dystonia and pyramidal signs, cardiac abnormalities, steatosis and hepatomegaly, ophthalmologic and vision problems, palate abnormalities, and endocrine problems including hypospadias, delayed puberty, and type 2 diabetes [6]. Consistent with an autosomal recessive syndrome, Del Greco et al. [7] recently reported an additional patient who was compound heterozygous for CARS1 variants and had a similar phenotype compared to patients reported by Kuo et al. In addition, this patient had hypoglycemia of unknown etiology [7].

In the two prior reports by Kuo et al. [6] and Del Greco et al. [7], functional studies were performed demonstrating the impact of CARS1 variants on cysteinyl-tRNA synthetase. Aminoacylation assays showed reduced enzyme activity and yeast complementation assays confirmed loss of function effects of the CARS1 variants [6, 7]. These studies further implicate CARS1 variants in human disease.

Here we provide a third report describing a sixth patient with novel compound heterozygous variants in CARS1 and expand the phenotype of this autosomal recessive multisystem condition.

Case presentation

Our proband is a 15-year-old male who was born at 34 3/7 weeks gestation as part of a diamniotic dichorionic twin pregnancy conceived via in vitro fertilization and born to non-consanguineous parents in their late thirties of European and Filipino ancestry. Pregnancy course was complicated by echogenic bowel noted at 20 weeks gestation, intrauterine growth restriction (which was also noted in his twin) and decreasing amniotic fluid. After his twin had absent end-diastolic flow, he and his twin were delivered via caesarian section secondary to breech/vertex position. His birth measurements were weight 1.162 kg (Z-score: −2.85), length 36.8 cm (Z-score: −3.38) and head circumference 28 cm (Z-score: −2.33). He required continuous positive airway pressure (CPAP) the first two days of life and was transferred to the neonatal intensive care unit. He received phototherapy for hyperbilirubinemia. A gastrostomy tube was placed due to feeding difficulties. An echocardiogram obtained to evaluate a heart murmur revealed a tiny patent ductus arteriosus (PDA) and muscular ventricular septal defect (VSD), with repeat echocardiogram at two months of life noting small conoventricular VSD, tiny and tortuous PDA and small patent foramen ovale. Renal ultrasound obtained due to hypertension noted small kidneys but was otherwise normal, and he was treated with captopril until three months of age. Head ultrasound obtained due to small size noted a tiny cystic area anterior and lateral to the anterior horn of the left lateral ventricle, which was confirmed on repeat head ultrasound at one month of age as well as brain magnetic resonance imaging (MRI). Radiology noted that the small cystic area could reflect cystic encephalomalacia versus a prominent perivascular space. Brain MRI noted normal cerebral volume and age-appropriate myelination. Neurology exam noted hyperreflexia (lower greater than upper) and ankle clonus (right greater than left).

Patient had hypoglycemia that required intravenous fluids. At two weeks of life a critical sample was obtained, noting plasma glucose less than 50 mg/dL, plasma insulin of 11.9 µU/mL, growth hormone of 9 ng/mL and cortisol of 4.8 mcg/dL. Based on this critical sample demonstrating inappropriately elevated insulin concentration during hypoglycemia, he was diagnosed with hyperinsulinism. Therapy with 12 mg/kg/day diazoxide was initiated at 14 days of life. He failed a safety fast on 12 mg/kg/day but because of concerns of fluid overload and hypertension, his diazoxide dose was not increased. He was discharged on 11.2 mg/kg/day and bolus gastrostomy feeds over 60 min. Diazoxide was discontinued around 28 months of age due to excess fluid retention in combination with obstructive sleep apnea that resulted in pulmonary hypertension, and he was treated with continuous dextrose through gastrostomy. Diazoxide at 5 mg/kg/day was then re-trialed around 39 months of age due to persistent hypoglycemia on dextrose alone, and he was able to be weaned off enteral dextrose. He remained on diazoxide until 12.5 years of age, though a fast completed at that time off diazoxide revealed ongoing evidence of hyperinsulinism based on inappropriately suppressed ketones during hypoglycemia and positive glucagon stimulation test (plasma glucose 49 mg/dL, betahydroxybutyrate 0.9 mmol/L, insulin < 2 µU/mL, c-peptide 0.3 mg/mL; positive glycemic response to glucagon with plasma glucose rising 60 mg/dL in 34 min). Levothyroxine was started at three months of age following abnormal thyroid studies (TSH 27 µIU/mL, T4 1.4 mcg/dL). Patient was evaluated by Genetics at 2.5 months of age and was noted to have dysmorphic features including epicanthal folds, micrognathia, high arched palate, low set ears, and microphallus. He had mild clinodactyly of left and right fifth digits which was also present in his father. Small palpebral fissures and unilateral iris coloboma [left eye] were also noted by ophthalmology. Patient was diagnosed with osteopenia around six months of age following a fracture but had a normal Dual-Energy X-ray Absorptiometry (DEXA) around eight years of age. He was small at birth and continues to have microcephaly (Z-score at 9 years 4 months: −6.69) and short stature (Z-score at 15 years 3 months: −3.56). Growth hormone treatment was initiated at 10.5 years of age and continues as of the writing of this report.

Additional medical history includes brittle nails, scoliosis, testicular hernia and hydrocele repair as well as strabismus repair. He has been followed by cardiology for history of pulmonary hypertension (resolved by 33 months of age and considered multifactorial in origin), tricuspid valve insufficiency (now trivial), history of mild aortic root dilation (resolved by seven and a half years of age) and elevated NT-pro BNP with normal cardiac function. He also has delayed puberty with no evidence of pubertal development and prepubertal levels of gonadotropins and testosterone at age 15 years. Spastic diplegia was observed by neurology around two years of age and an ataxic gait was appreciated around five and a half years of age.

Patient was delayed in meeting his developmental milestones and currently has intellectual disability and cerebral palsy. He ambulates and uses a walker for assistance. He has many words but uses an adaptive language device or sign language to assist with communication. He takes most of his nutrition through a gastrostomy tube with some limited food intake by mouth. His neurologic phenotype is overall consistent with what has been reported in the prior reports by Kuo et al. [6] and Del Greco et al. [7].

Family history is significant for patient’s fraternal twin sister having IUGR prenatally and transient hyperinsulinism that resolved and was presumed to be perinatal stress-induced hyperinsulinism, given her perinatal history of prematurity and IUGR. She additionally has been diagnosed with Graves disease, accelerated bone age, immune thrombocytopenic purpura and ketotic hypoglycemia. Development has been normal. Patient’s mother, maternal aunt and maternal grandmother all report symptoms of hypoglycemia that have been managed with frequent snacking; they have not had confirmed hypoglycemia and have not undergone formal evaluations of their symptoms. Patient’s father has gout and eczema and is otherwise healthy. A paternal male cousin has possible autism spectrum disorder. The family history is otherwise unremarkable with respect to neurodevelopmental disorders.

Results

Given his complicated medical history, the patient had multiple genetic tests completed over the years. Due to the presence of inverted nipples, prominent fat pads and abnormal dimpling over his arms and buttocks, along with mildly abnormal glycosylation studies, he was suspected around two to three years of age to have an unidentified congenital disorder of glycosylation. However, additional extensive metabolic testing in addition to a congenital disorders of glycosylation gene panel and exome sequencing were not consistent with a clear diagnosis of a congenital disorder of glycosylation. Additional unrevealing genetic testing included karyotype, SNP chromosomal microarray, congenital hyperinsulinism panel, Kallmann syndrome panel, and exome reanalysis (See Supplemental Table 1 for genetic tests and gene list). HK1 intron 2 sequencing through research testing was also negative, and based on the presence of several common heterozygous single nucleotide polymorphisms, a large deletion was also excluded. We therefore obtained trio genome sequencing (GeneDx) at 13.5 years of age. Testing revealed two variants in the CARS1 gene: c.2093A > G, p.Tyr698Cys (maternally inherited) and c.1102G > A, p.Asp368Asn (de novo) (NM_001751.5). Paternity was previously confirmed based on the exome data. Patient’s mother does not have autism or symptoms of Parkinsonism-spinocerebellar ataxia. The reference laboratory was unable to determine if these variants were in cis or trans given that one variant was de novo, and no additional haplotype studies were performed. Both variants were classified by the laboratory as variants of uncertain clinical significance. The variants both appear to be rare, as the p.Asp368Asn variant was present in three people in gnomAD (v.4) while the p.Tyr698Cys variant is absent from gnomAD (v.4) [8]. Both variants are predicted to be “probably damaging” by PolyPhen-2 [9] and “deleterious” by PROVEAN [10]. Both CARS1 gene variants are considered variants of uncertain clinical significance based on American College of Medical Genetics and Genomics variant classification criteria, since CARS1 is a candidate gene given the limited case reports to date [11]. These variants were not previously identified by the initial exome analysis since an association with a neurodevelopmental phenotype had not yet been identified at the time of testing, and reanalysis was performed shortly after the Kuo et al. [6] paper was published and the CARS1 gene was not yet integrated into the laboratory’s gene analysis database. Genome sequencing did not identify any pathogenic variants, likely pathogenic variants or variants of uncertain clinical significance in genes associated with congenital hyperinsulinism.

Table 1 ^**Comparison of patient phenotype to previously reported patients

Patient was subsequently referred to dermatology for additional testing due to case reports of specific hair findings on microscopy in the medical literature [6]. While prior cases have been reported with brittle and shortened hairs, this patient had “a full head of hair” and there were no reported parental concerns about hair issues. Nonetheless, hairs were cut near the base of the shaft and examined under light microscopy. Visible light examination demonstrated rare instances of trichorrhexis nodosa with focal fracture of the hair shaft. Polarized light revealed abnormal and focal banding of pigment within the hair shaft in a small subset of hair shafts that was subtle and not equivalent to the more typical and prominent “tiger banding” associated with trichothiodystrophy. These hair findings overlap with the hair findings noted in prior published patients [6]. Overall, our patient’s phenotype has significant overlap with the five patients reported in the medical literature (see Table 1).

Discussion and conclusions

The de novo p.Asp368Asn variant in the CARS1 gene is located within the catalytic domain of the protein, within the region of five of the previously reported variants [6, 7] (Fig. 1). The maternally inherited p.Tyr698Cys variant is located near the frame shift variant reported by Kuo et al. within the C-terminal extension (CTE) [6]. Liu et al. has shown that the CTE plays an important role in anticodon recognition by cysteinyl-tRNA synthetase 1 [12].

Fig. 1

Location of CARS1 variants in gene. Legend: Adapted from Fig. 5B from Kuo et al. [6]. CARS1 gene variants are positioned above. Previously reported variants are in black and novel variants from this report are in red. Numbers below indicate amino acid positions. CARS functional domains shown in purple (catalytic domain) and yellow (anticodon-binding domain). The p.Asp368Asn variant from this report was de novo and the p.Tyr698Cys variant from this report was inherited

Our patient’s clinical features and hair microscopy findings share significant overlap with the five previously reported patients in the medical literature with compound heterozygous or homozygous variants in CARS1. In particular, the five reported patients all had developmental delays including motor and language delays. While our patient’s intellectual disability appears more significant than the mild to moderate delays observed in the patients in Kuo’s report, our patient had several risk factors for delays which may have compounded the severity of delays observed in him. These include prematurity [13, 14] (which was reported in only one of the four patients in Kuo et al.’s report) and hypoglycemia due to congenital hyperinsulinism, another risk factor for cognitive impairment [15,16,17]. Three out of the four reported patients in Kuo’s report had spastic ataxia, along with ataxic gait noted in Del Greco’s patient, and spastic ataxia was also observed in our patient. Various brain abnormalities were noted on MRI in our patient as well as those previously reported. The brain abnormalities in our patient do not overlap with those found in the other reported patients. However, our patient’s brain MRI was performed at one month of age and not repeated, while the other patients had brain imaging performed between one year and 33 years of age. We therefore do not know what our patient’s brain imaging would have revealed at an older age. All patients had short stature and three of the five patients shared microcephaly with our patient. While our patient’s hair was noted to be normal on clinical exam in contrast to the reported patients, as with four of the previously reported patients, hair abnormalities were evident on microscopy, including trichorrhexis nodosa. Trichorrhexis nodosa can be acquired or inherited [18], though overall is not a frequent hair abnormality in children [19]. Our patient’s small phallus and delayed puberty overlapped with one of the reported patients who had hypothalamic hypogonadism and delayed puberty. Interestingly, two of the patients had type 2 diabetes diagnosed at 24 years and 30 years of age, respectively. This is earlier than people diagnosed in the general population, where the typical age of diagnosis of type 2 diabetes in the United States is 49.9 years of age [20]. At 15 years of age, our patient has not been diagnosed with diabetes. However, he has a diagnosis of congenital hyperinsulinism. Congenital hyperinsulinism due to dysregulated insulin secretion by the pancreatic beta cells is the most common cause of persistent hypoglycemia in infants and children [21, 22]. In addition to monogenic causes of congenital hyperinsulinism, hyperinsulinism has been identified in multiple genetic syndromes [21, 23], though the underlying cause of hyperinsulinism in those syndromes is not always well understood [23, 24]. The patient described by Del Greco et al. had hypoglycemia that was diagnosed at three months of age, though the underlying cause of the hypoglycemia was not characterized in that report [7]. The presence of dysglycemia, both hypo- and hyperglycemia in these cases, and specifically hypoglycemia due to hyperinsulinism in our case, suggests that insulin secretion by pancreatic beta cells may be impacted by pathogenic CARS1 variants. Indeed, expression data from rat beta cells as well as data from the Human Protein Atlas both suggest expression of the CARS1 gene in beta cells [25, 26]. Hyperinsulinism was also previously reported in a patient with a homozygous variant in a gene encoding a different aminoacyl tRNA synthetase, specifically the YARS1 gene (OMIM 603623) which encodes tyrosyl-tRNA synthetase [27]. Hypoglycemia has also been reported in five other people with a homozygous YARS1 variant [28].

We propose a possible mechanism for how pathogenic CARS1 variants may lead to the hyperinsulinism observed in our patient. Akaike et al. demonstrated that cysteinyl-tRNA synthetases are involved in the creation of persulfides, including cysteine hydropersulfide (CysSSH) and cysteine polysuflides [2]. While the mitochondrial cysteinyl-tRNA synthetase (CARS2, OMIM 612800) accounted for the majority of the observed CysSHH in Akaike’s paper, their mouse studies noted the cytosolic cysteinyl-tRNA synthetase also produced cysteine polysulfides, dependent on the presence of pyridoxal phosphate [2]. Studies by Kline et al. have shown that calcium/calmodulin-dependent protein kinase II (CaMKII) regulates voltage gated potassium ATP (KATP) channels in beta cells via phosphorylation of T224 of the Kir6.2 subunit [29], where phosphorylation leads to significantly decreased beta cell KATP channels. [29] CaMKII in turn can be inhibited by persulfidation [30], and cysteinyl-tRNA synthetases have also been shown to have cysteine persulfide synthase activity [2]. Reduced/absent cysteinyl tRNA synthetase 1 due to CARS1 pathogenic variants may therefore reduce formation of CysSSH, decreasing the inhibition of CaMKII activity, resulting in increased phosphorylation of beta cell KATP channels and therefore contributing to the hyperinsulinism observed in our patient. Indeed, Dadi et al. demonstrated impaired glucose tolerance when CaMKII was inhibited [31], further supporting a causal role for this pathway in causing the opposite phenotype of hyperinsulinism.

The underlying genetic cause remains elusive in 21% of people with congenital hyperinsulinism [32]. While our patient underwent extensive genetic testing, including karyotype, SNP chromosomal microarray, congenital hyperinsulinism panel, exome, exome reanalysis and most recently genome sequencing, none of which identified alternative causes of his hyperinsulinism, we cannot exclude the possibility that this patient’s hyperinsulinism has some other genetic cause unrelated to the CARS1 variants. The patient’s sister had transient hyperinsulinism with a clinical course consistent with perinatal stress-induced hyperinsulinism and therefore, we believe this is unrelated to the hyperinsulinism of the patient in this report [33]. Furthermore, given that this multisystem syndrome has only relatively recently been described with only five other patients described in the medical literature with overlapping phenotypes, we have limited patient data to draw from. Though there are few patients described to date, it is interesting that four of the six have evidence of dysglycemia. Finally, since one of our patient’s CARS1 variants was de novo, we are unable to confirm if the de novo variant is in trans with the maternally inherited variant. Future studies such as haplotype analysis or reanalyzing the gene via long read sequencing may clarify this. However, given the strong phenotypic overlap to the previously reported patients, we believe the two variants are most likely on opposite alleles.

This case further illustrates the utility of continuing to pursue the genetic etiology of a multisystem neurodevelopmental disorder after initial genetic testing is negative. In addition to the negative panel genetic testing for congenital disorders of glycosylation, congenital hyperinsulinism and Kallmann syndrome, our patient’s initial chromosomal SNP microarray and exome sequencing (including exome reanalysis) were negative. Genome sequencing at 13.5 years of age finally identified the likely causative variants to explain this patient’s complicated medical history since it was performed several years after Kuo et al. described four patients with CARS1 gene variants [6]. While genome sequencing is known to have superior detection of variants compared to exome sequencing [34], in our patient’s case it was not the difference in technology but rather the advancement in scientific knowledge that led to the identification of these CARS1 gene variants. The testing laboratory (Baylor Genetics) that performed the exome sequencing has confirmed the CARS1 gene variants were detectable on their platform but not analyzed at the time exome analysis and reanalysis were performed (personal communication from Baylor Genetics). While newer technology such as genome sequencing and even more recently long read sequencing platforms will invariably lead to additional pathogenic variants being identified in patients in the future [35], equally important are the continued advancements in gene discovery and understanding of the molecular mechanisms of disease to render interpretation of these variants intelligible and usable for patients and clinicians.

In closing, we have described a sixth patient with an autosomal recessive multisystem disorder due to cysteinyl-tRNA synthetase 1 variants causing developmental delay, microcephaly, short stature and hair abnormalities. This case report expands the phenotype to include congenital hyperinsulinism, highlighting the beta cell dysfunction that appears to be present in some patients. The identification of additional patients in the future will further clarify the phenotypic spectrum associated with variants in the CARS1 gene, and future research is needed to understand how variants in this gene affect the many organs involved in this condition.