Identification of rare variants causing urea cycle disorders: A clinical, genetic, and biophysical study

Abstract Urea cycle disorders (UCDs) are a group of rare metabolic conditions characterized by hyperammonemia and a broad spectrum of phenotypic severity. They are caused by the congenital deficiency in the eight biomolecules involved in urea cycle. In the present study, five cases of UCD were recruited and submitted to a series of clinical, biochemical, and genetic analysis with a combination of high throughput techniques. Moreover, in silico analysis was conducted on the identified missense genetic variants. Various clinical and biochemical indications (including profiles of amino acids and urinary orotic acids) of UCD were manifested by the five probands. Sequence analysis revealed nine diagnostic variants, including three novel ones, which caused Argininosuccinic aciduria (ASA) in one case, Carbamoyl phosphate synthetase 1deficiency (CPS1D) in two cases, Ornithine transcarbamylase deficiency (OTCD) in one case, and Citrin deficiency in 1case. Results of in silico biophysical analysis strongly suggested the pathogenicity of each the five missense variants and provided insight into their intramolecular impacts. In conclusion, this study expanded the genetic variation spectrum of UCD, gave solid evidence for counselling to the affected families, and should facilitate the functional study on the proteins in urea cycle.

, argininosuccinate lyase (ASL, EC 4.3.2.1) and arginase (ARG, EC 3.5.3.1), and two amino acid transporters, namely ornithine transporter (ORNT1; ornithine/citrulline carrier) and citrin (aspartate/ glutamate carrier). 5,6 Generally speaking, hyperammonemia leads to anorexia, cerebral oedema, lethargy, vomiting, hypothermia, hyperventilation (or hypoventilation), neurologic posturing, and coma. 6 The ability to accurately identify the pathogenesis in early stage of the disease is beneficial for taking appropriate and necessary management to prevent irreversible neurological damage. 7,8 Plasma ammonia level is the most direct indicator of UCD patients at any age, yet it does not meet the requirement for screening for early treatment. Currently, tandem MS of dried blood spots is the widely used method to screen for neonatal metabolic diseases, including UCDs. 9 However, the sensitivity and specificity of this method for detection of reducing citrulline are not high, resulting in a decrease in the detection rate of some types of UCDs, such as NAGS, CPS1,and OTC deficiencies. 10 The characteristics of clinical, genetic heterogeneity and variable severity of UCDs promoted the development of more comprehensive and individualized diagnostic and therapeutic strategies. 11 In addition, identification of the proband's genotype is of great importance in the selection of reproductive options, such as prenatal diagnosis (PD) or pre-implantation genetic diagnosis (PGD), for affected families. Enzyme activity analysis, although being a time/labour consuming assay, can help confirming the diagnosis and the residual activity if the genetic test does not give a clear result, 6 so it should be retained.
In the present study, 5 families with early onset metabolite disorders were recruited and underwent a series of clinical, biochemical and genetic tests. The results provided clinicians with sufficient evidence to develop specific treatment measures. Furthermore, structural prediction and molecular dynamic analysis were conducted to elucidate the intramolecular impact of missense variants.

| Clinical and biochemical analysis
Probands of these families received comprehensive clinical examination and laboratory tests. The diagnostic criteria for UCD referred to that summarized by Matsumoto et al. 6 General biomedical tests were performed using Cobas C501 Automatic System (Roche, Switzerland). Tests of coagulation function were carried out using ACL TOP 700LAS System (Instrumentation Laboratory, Werfen, Spain). The plasma amino acid profiles were tested using API 3200MD LC-MS/MS tandem mass spectrometry System (AB SCIEX, Canada), and the urinary orotic acid levels were tested using GC-2010 Pro Capillary Gas Chromatograph System (Shimadzu, Japan).
Specific management strategy was adopted to treat each proband according to the combined results of clinical, biomedical and genetic analysis.

| Genetic analysis
Conventional G-bangding karyotyping was performed on peripheral blood specimens from the 5 probands according to the AGT F I G U R E 1 The overall experimental workflow in this study. ACMG, American College of Medical Genetics and Genomics; Aqua words represent patented software of Berry Genomics; CNV, copy number variation; HGMD, Human Gene Mutation Database; InDel, insertion/deletion; Mt, mitochondrial variation; QC, quality control; ROH, runs of homozygosity; SNP, single nucleotide polymorphism; WES, whole-exome sequencing (American Genetic Technician) cytogenetics manual, 12 to detect overall chromosomal anomalies.
Whole-exome sequencing (WES) was conducted on samples of the 5 probands as described in previous studies. 13,14   (a combined method of pathogenicity prediction) and pLI score (representing the tolerance for truncating variants) were also introduced.
To validate variants, Sanger sequencing was introduced as a confirmatory method using 3500DX Genetic Analyzer (Applied Biosystems, Thermo Fisher Scientific, USA). The sequencing PCR primers, reaction conditions and reagents were included in Table S1.

| In silico protein structural and molecular dynamic analysis
The evolutionary conservatism of all amino acids affected by missense variants was analysed using MEGA7 (http://www.megas oftwa re.net/previ ousVe rsions.php) with default parameters.
Protein modelling was conducted using Modeller 9V17 software 18 with default parameters based on the structure models of homozygous sequences indexed in PDB database (http://www. rcsb.org/). Molecular dynamic analysis was conducted on 1 novel missense variant. Briefly, both wild-type and mutant models were generated by Modeller 9V17; the programme CHARMM22 was used to add hydrogen atoms, N-and C-terminal patches to the models. 19 The models were solvated and neutralized in a box with TIP3P water at a minimum of 13 Å between the model and the wall of the box. All simulations were run using NAMD 2.9 with periodic boundary conditions (PBC) applied. 20 The temperature was held at 300 K while the pressure was controlled at 1 atm. The time step was set to 2 fs and the particle mesh Ewald method was applied to model the electrostatics and the van der Waals interactions cutoff was set at 12 Å. Both models followed a three-step pre-equilibration totalling 600 ps, the last snapshots of which were chosen as the starting structures for 20 ns productive simulations without constraints.

| Clinical findings
The pedigree diagrams of the 5 families were demonstrated in  Table 1, and the brief medical histories are as follows.
Detailed treatment course and highlighted remarks were included in Table S1.
Patient I-1, male, was hospitalized for poor response, sucking rejection and hypotonia at 12 days of age. Laboratory tests suggested hyperammonemia, hepatic dysfunction, dysfunction of blood coagulation and abnormal profiles of amino acids (AA) and urinary orotic acids (UOA) ( Table 1). The peak area of the arginosuccinate was 1.8275 (677 times of the control level). The patient was treated with ornithine, aspartic acid and arginine injection and the ammonia level decreased to 286 μmol/L after 12 hours, and to 64 μmol/L 3 days later. Compound glycyrrhizin and reduced glutathione were used for protecting liver function.
After treatment for 28 days, the ammonia level decreased to 54 μmol/L. Then the patient received a liver transplantation (LT) at 6 months. The surgery was successful and the infant is currently in good condition.
Patient II-1, male, was hospitalized for poor response, sucking rejection at 4 days of age. Laboratory tests suggested hyperammonemia, yet normal hepatic function and blood glucose with no metabolic acidosis. The abnormal AA and UOA profiles were demonstrated in Table 1. The infant was treated with arginine hydrochloride injection and the ammonia level had been reduced to 84 μmol/L.
But with a tiny amount of protein intake, his blood ammonia was elevated (like taking 30ml milk, once every 3 hours, then the blood ammonia would increase to ~ 300 μmol/L, with no effectiveness by taking oral anti-hyperammonia drugs such as sodium benzoate). The parents rejected LT. At the 53th day, his condition deteriorated (ammonia level > 500 μmol/L) and developed into death.
Patient III-1, male, was hospitalized for poor response and hypotonia at 2 days after birth, and developed into convulsions and coma during hospitalization. Laboratory tests indicated severe hyperammonemia and abnormal AA/UOA profiles (Table 1). After 3 weeks' dietary and medicinal treatment, the infant's ammonia level dropped to normal, but a cranial magnetic resonance imaging scan revealed abnormity. He was diagnosed with cerebral palsy at 1 years old.
Patient IV-1, male, having a suspected family history (details in Table 1), was hospitalized for hyperammonemia right after birth.
Shortly after admission, the patient developed pulmonary bleeding, upper gastrointestinal bleeding, convulsions and coma. Despite of aggressive emergency treatment, the child died 24 hours after birth.
Patient V-1, female, was hospitalized for jaundice at 35 days after birth. She was once with in utero foetal hypoxia, and her birth weight was 2100g with 45cm of birth length (<3rd percentile, both).
Laboratory results indicated her with liver dysfunction (probably intrahepatic cholestasis) and irregular AA/UOA profile (Table 1). After a thorough genetic test, she was given treatment with lactose-free milk powder and ursodeoxycholic acid. Her hepatic function returned to normal at 3 months old; and the AFP level decreased gradually, returning to normal by 1 year old.

| Genetic variations
The karyotyping results were normal for each of the probands.

| Structural and molecular dynamic analysis
In this study, 5 diagnostic missense variants were detected, namely prefers the first round of helices, but Q810P placed in the middle of a helix, which always causes an apparent bent (shown in Figure 5E).
Moreover, the number of hydrogen bonds formed between Q 810 and the rest residues were more than that formed between Q810P and the rest residues (1.155:0.20) ( Figure 5D).

| D ISCUSS I ON
Ammonia is produced by intestinal urease-positive bacteria as well as during amino acid metabolism, and the urea cycle (UC) performs the critical function of converting ammonia into urea. 6 In the current study, the 5 cases involved 4 various UCD types, respectively. ASA, resulted from ASL deficiency, is consistent with autosomal recessive (AR) pattern and may present a broad spectrum of severity depending on the residual activity of ASL. 24,25 Recently, a model was developed to predict the phenotypic severity of ASA patients by cohort based multi-platform data, which is worth learning from, but also needs to be optimized by expanding the sample size. 26 Another study proved that cerebral condition in ASA involves neuronal oxidative/nitrosative stress independent of hyperammonaemia, which should be highly valued in clinical practice. 27   study, patient I-1 manifested significant noenatal elevation of plasma ammonia and citrulline, and the Arginosuccinate level was more than 600 times higher than the control; furthermore, the identification of compound variants in ASL gene confirmed the ASA diagnosis.
The use of ornithine, aspartic acid and arginine injection combina- (with in-vitro fertilization, IVF) are recommended, which also applies to case IV. 40,41 As a matter of fact, families I, III and V have made it clear that IVF approach will be considered. It was also suggested that the future spouse of subject I.4 should check his ASL genotype before they consider a pregnancy, and they should be made aware that de novo mutations are more common in males. As for case IV, the risk of male siblings of patient IV-1 will be as high as 50%; while the female siblings, if being heterozygous, will not be entirely safe either and should avoid strong stressors as much as possible. 36 As BabySeq projects emerged, the field of neonatal screening for mendelian diseases may enter a new era and fundamentally improve the efficiency of screening. [42][43][44] However, there are many details, such as the design of gene panels, the curation of variation interpretation criteria, the choice of screening methods, the scope of application of subjects, etc, which need to be discussed in a long process. 45 With regard to disorders like UCDs with acute neonatal presentation, it may be more appropriate to perform pre-pregnancy carrier screening using a meticulously -designed gene panel.
A limitation of this study is that the first-line genetic detection method we use is WES, covering only the gene coding exonic region and the ad- In summary, we identified the causative variants in five families with UCDs, including three novel variants, which provided solid evidence for genetic counselling. In addition, thorough in silico analysis would benefit the interpretation of variations.

ACK N OWLED G M ENTS
We thank the participating patients and investigators for their cooperation.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Based on reasonable requirements, all raw data can be obtained from the corresponding author.