PAICS deficiency, a new defect of de novo purine synthesis resulting in multiple congenital anomalies and fatal outcome

Abstract

We report for the first time an autosomal recessive inborn error of de novo purine synthesis (DNPS)—PAICS deficiency. We investigated two siblings from the Faroe Islands born with multiple malformations resulting in early neonatal death. Genetic analysis of affected individuals revealed a homozygous missense mutation in PAICS (c.158A>G; p.Lys53Arg) that affects the structure of the catalytic site of the bifunctional enzyme phosphoribosylaminoimidazole carboxylase (AIRC, EC 4.1.1.21)/phosphoribosylaminoimidazole succinocarboxamide synthetase (SAICARS, EC 6.3.2.6) (PAICS). The mutation reduced the catalytic activity of PAICS in heterozygous carrier and patient skin fibroblasts to approximately 50 and 10% of control levels, respectively. The catalytic activity of the corresponding recombinant enzyme protein carrying the mutation p.Lys53Arg expressed and purified from E. coli was reduced to approximately 25% of the wild-type enzyme. Similar to other two known DNPS defects—adenylosuccinate lyase deficiency and AICA-ribosiduria—the PAICS mutation prevented purinosome formation in the patient’s skin fibroblasts, and this phenotype was corrected by transfection with the wild-type but not the mutated PAICS. Although aminoimidazole ribotide (AIR) and aminoimidazole riboside (AIr), the enzyme substrates that are predicted to accumulate in PAICS deficiency, were not detected in patient’s fibroblasts, the cytotoxic effect of AIr on various cell lines was demonstrated. PAICS deficiency is a newly described disease that enhances our understanding of the DNPS pathway and should be considered in the diagnosis of families with recurrent spontaneous abortion or early neonatal death.

Introduction

Purines are critical to energy production and are the fundamental units that transmit the genetic code. De novo purine synthesis (DNPS) includes a series of 10 enzymatic reactions that are critical to purine formation (Fig. 1). To date, two genetically determined defects of DNPS have been identified—adenylosuccinate lyase (ADSL) deficiency (ADLSD [OMIM 103050]) (1, 2) and 5-aminoimidazole-4-carboxamide ribosiduria (AICA-ribosiduria) (OMIM 608688) (3), which are caused by mutations in ADSL or ATIC genes, respectively. Both disorders are characterized by neurological involvement of variable severity (4).

Inherited defects in the four other enzymes involved in the early steps of DNPS—phosphoribosyl pyrophosphate amidotransferase (PPAT, EC 2.4.2.14), trifunctional GART—glycinamide ribonucleotide synthetase (EC 6.3.4.13)/phosphoribosylglycinamide formyltransferase (EC 2.1.2.2)/phosphoribosylaminoimidazole synthetase (EC 6.3.3.1), phosphoribosylformylglycinamidine synthetase (PFAS, EC 6.3.5.3) and bifunctional enzyme phosphoribosylaminoimidazole carboxylase (AIRC, EC 4.1.1.21)/ phosphoribosylaminoimidazole succinocarboxamide synthetase (SAICARS, EC 6.3.2.6) (PAICS)—have not yet been identified in humans and have been found to be lethal in respective knockout models in mouse (5), zebrafish (6) and Drosophila melanogaster (7). However, the range of genetic variation provided in the gnomAD database (https://gnomad.broadinstitute.org/) indicates that, except for PPAT, there are no evolutionary constraints against loss of function or missense mutations in these genes in the population (8).

In this investigation, we report for the first time an autosomal recessive condition resulting from mutations in PAICS. We demonstrate similar laboratory findings as in ADSL deficiency and AICA-ribosiduria and a similar but very severe neonatal phenotype with fatal outcome. PAICS deficiency is a newly described disease entity that increases our understanding of the DNPS pathway and the critical nature of the participating enzymes.

Results

A family with recurrent polyhydramnios associated with fetal malformations and resulting in early neonatal death

Through an international collaboration and GeneMatcher (9), we investigated a family from the Faroe Islands that had total of seven pregnancies (Fig. 2A). Four pregnancies were uneventful and resulted in three healthy unaffected males and one healthy unaffected female. One pregnancy ended in a spontaneous abortion at 13 weeks of gestation based on polyhydramnios, and two pregnancies led to the live births of two children with multiple malformations who died on Days 2 and 3 based on progressive hypotension and hypoxia.

Figure 1

De novo purine synthesis (DNPS) and purine nucleotide cycle (PNC). The inosine monophosphate (IMP), the final product of DNPS, is generated from phosphoribosyl pyrophosphate (PRPP) in 10 enzymatic reactions catalyzed by six enzymes: phosphoribosyl pyrophosphate amidotransferase (PPAT), trifunctional GART glycinamide ribonucleotide synthetase/phosphoribosylglycinamide formyltransferase/phosphoribosylaminoimidazole synthetase, phosphoribosylformylglycinamidine synthetase (PFAS), bifunctional PAICS phosphoribosylaminoimidazole carboxylase/phosphoribosylaminoimidazole succinocarboxamide synthetase, adenylosuccinate lyase (ADSL) and bifunctional ATIC 5-aminoimidazole-4-carboxamide ribonucleotide transformylase/inosine monophosphate cyclohydrolase. In response to purine need, these enzymes transiently assemble into a multi-enzyme structure, the purinosome. The IMP is transformed in PNC to adenosine monophosphate (AMP) in reactions catalyzed by adenylosuccinate synthetase (ADSS) followed by ADSL. AMP could by converted back to IMP by adenosine monophosphate deaminase (AMPD). PAICS deficiency should lead to accumulation of aminoimidazole ribotide (AIR) and its dephosphorylated derivative aminoimidazole riboside (AIr).

Case 1 (II.4)

The index case was a male noted to have polyhydramnios during a routine ultrasound at 20 weeks of gestation. Ultrasound images also indicated an unspecified malformation of one leg. Spontaneous birth occurred at 36 weeks accompanied by discharge of 7 l of clear amniotic fluid. Moderate asphyxia was noted at birth, with APGAR scores of 3/1, 5/8 and 8/8. The birth weight was 1850 g (P3—200 g), length 41 cm (P3—2.5 cm) and head circumference 33.5 cm (P50). Multiple malformations were identified, including a small body, short neck, short stature, brachycephaly, craniofacial dysmorphism especially of midface with flat face, severe nasal hypoplasia, low nasal bridge, anteverted nostrils, bilateral choanal atresia, hypertelorism, low-set and poorly modulated ears, mild clinodactyly of the fifth fingers bilaterally, shortened distal part of the left leg with club foot, duplication of the left first toe, broadened first toe on right foot, small penis with subcoronal hypospadias, bilateral cryptorchidism, esophageal atresia without tracheoesophageal fistula, several costal and vertebral malformations and hypoplasia of left lung (Fig. 2B and C). Normal muscle tone and reflexes, good spontaneous movements and reactions were noted. Ultrasound of the heart, kidneys and brain was without abnormalities. The patient was intubated 2.5 h after delivery based on respiratory failure. Progressive hypotension and hypoxia developed despite standard ventilation. The baby succumbed from cardiorespiratory failure on Day 3 before further diagnostics could be performed.

Figure 2

Pedigree and clinical picture of the proband II.4 with PAICS deficiency. (A) Pedigree of the family with several common ancestors. Black symbols denote affected individuals, open symbols denote unaffected parents and siblings. (+/−) denotes presence (+) or absence (−) of the chr4:57307970 A/G variant. (B) Multiple malformations including small body, short neck, flat face with hypertelorism, nasal hypoplasia, low nasal bridge, low set ears, small penis, clubfoot left in Proband 1 (patient is breathing through an oral Guedel tube for bilateral choanal atresia). (C) Whole body X-rays of proband showing multiple aplasias of costae (only eight rips present), hypoplasia of left thorax, multiple deformations of cervical and thoracic vertebrae (multiple butterfly and hemi-vertebrae) and hypoplasia of left tibia leading to clubfoot.

Case 2 (II.6)

An affected sister of the index case was born spontaneously at 34 weeks of pregnancy. Polyhydramnios was observed. The APGAR scores were 7/1 and 8/5. The umbilical cord pH was 7.28. The weight was 2020 g (P10–P25). Short neck, brachycephaly, flat face, low-set ears and small nose with a low nasal bridge were noted. There was right choanal atresia and left choanal stenosis. No fluid was returned with nasogastric suction. An X-ray confirmed esophageal atresia with a tracheoesophageal fistula. Malformation of several ribs and lumbar hemivertebrae were observed. Cystic malformation of the left lung was suspected. The infant had a similar downhill course with progressive hypotension and hypoxia and died at 28 h of age.

Figure 3

Structural effect of the p.Lys53Arg mutation and its impact on PAICS in cultured skin fibroblasts. (A) Analysis of PAICS structure (PDB 4JA0) suggesting that presence of the arginine residue at the position 53 (R53) may lead to a more favorable interaction of this residue with aspartic acid residue at the position 212 (D212). This may move the beta sheet 13, destabilize the catalytic pocket of the SAICARS and alter enzymatic activity of PAICS. (B) PAICS cDNA analysis; agarose gel electrophoresis profiles of RT-PCR products amplified from total RNA isolated from fibroblasts of control and proband. (C) Western blot analysis of PAICS in cell homogenates; detection with mouse monoclonal anti-PAICS antibody shows a specific immune-reactive protein of a molecular weight ~ 46 kDa corresponding to the predicted molecular weight of the processed PAICS. The amounts of PAICS were comparable in the proband and control when normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH) detected with mouse monoclonal anti-GAPDH. (D) Enzyme activity of PAICS in lysates of fibroblasts was reduced to 10% in the proband and to 50% in heterozygous carriers in contrast to controls. Each data point represents the mean of three experiments. Whiskers represent 10 to 90 percentile values, and the mean is shown as plus and median as line (n = 3). (E) Enzyme activities of the recombinant maltose binding PAICS fusion proteins expressed in E. coli and purified on amylose resin. PAICS activity of mutated protein MBP-PAICS_K53R was reduced to 25% in contrast to the wild-type protein MBP-PAICS_wt. Each data point represents the mean of three expression experiments. Whiskers represent 10 to 90 percentile values, and the mean is shown as plus and median as line (n = 3).

Homozygosity mapping and exome sequencing identified a candidate homozygous missense variant in the PAICS

Genealogy and pedigree structure revealed that both parents have several common ancestors (Fig. 2A), and genomic analysis was pursued. We first genotyped genomic DNA using the Affymetrix 250K NspI SNP array. Using data derived from SNP probes, we did not identify any copy-number alterations that were compatible with an expected autosomal recessive inheritance model. In agreement with the expected founder effect, we identified three homozygous regions, on Chromosome 6 (chr6: 26393539–28 555 894), Chromosome 8 (chr8: 70174745–73 185 849) and Chromosome 4 (chr4: 54484130–62 069 735), which were shared by both probands (data not shown).

To identify potential disease-causing mutations, we sequenced and analyzed the exomes of the father, mother and two affected children. We searched for variants that had allele frequencies less than 0.1% in the gnomAD database (10) and whose genotypes were compatible with an expected autosomal recessive model of the disease. This analysis revealed in the homozygous region on Chromosome 4 a single homozygous missense variant (chr4: 57307970 A/G) in the PAICS. The identified variant, rs192831239, located in Exon 2 of PAICS (c.158A>G; NM_001079525.1), encodes the substitution of a lysine 53 to an arginine (p.Lys53Arg) of the PAICS (NP_001072993.1). Using Sanger sequencing, we confirmed homozygosity for the chr4: 57307970 A/G variant in both affected children and the heterozygous status of the healthy parents and siblings (Fig. 2A). The ‘G’ allele has been identified in a heterozygous state at a maximum frequency of 0.001 in non-Finnish Europeans. At lower frequencies, it has been identified in Finnish (0.0002), Africans (0.0002), and Latinos (0.0007) (https://gnomad.broadinstitute.org/gene/ENSG00000128050). All 143 Faroe individuals tested for this SNP showed the homozygous wild-type genotype A/A.

p.Lys53Arg mutation is predicted to affect the catalytic site of PAICS

The human PAICS assembles into octameric complexes that form well-defined tunnels connecting the AIRC and SAICARS catalytic sites (11). Using the available crystal structure of PAICS (PDB 4JA0, http://www.rcsb.org/pdb/home/home.do) and structural modeling, we found that the presence of the arginine residue at position 53 (R53) may lead to more favorable interaction with the aspartic acid residue at Position 212 (D212), which would cause movement of beta sheet 13 and thus destabilize the catalytic pocket of the SAICARS. To further examine these structural effects, we performed bioinformatic analysis using mCSM software, which revealed significant destabilization of the mutant protein (∆∆G = −0.165 kcal/mol). These findings suggest that Lys53Arg mutation affects PAICS catalytic activity for succinyl aminoimidazole carboxamide ribotide (SAICAR) production based on impaired structural stability (Fig. 3A).

Figure 4

Purinosome formation in skin fibroblasts cultured in purine depleted media. (A–D) In control fibroblasts, (A) PPAT and (B) GART show coarsely granular staining and (C, D) significant colocalization in the cytoplasm that is characteristic for purinosome formation. (E–H) In affected fibroblasts, (E) PPAT and (F) GART show diffuse intracellular staining with (G, H) no significant signal overlap suggesting that the formation of purinosome is affected by the presence of structurally mutated PAICS. (I–L) Transfection of affected fibroblasts with wild-type PAICS restores purinosome formation as shown by granular staining of (I) PPAT and (J) GART and their (K, L) colocalization in the cytoplasm. (M–P) Transfection of affected fibroblasts with mutated PAICS does not restore purinosome formation. (D, H, L and P) The degree of PPAT and GART colocalization is demonstrated by the fluorescent signal overlap (Manders) coefficient values that range from 0 to 1. The resulting overlap coefficient values are presented as the pseudocolor whose scale is shown in the corresponding lookup table.

Figure 5

Cytotoxicity of AIr and SAICAr. Cells were cultured in a growth media containing either AIr (circles) or SAICAr (squares) at 14 concentrations ranging from 1.7 μmol/l to 1 mmol/l. Cell viability was assessed 72 h later with luminescent reagent ATPlite™ (PerkinElmer), which takes the level of intracellular ATP as a measure of cell viability. Data were normalized, and IC50 values were calculated using GraphPad Prism software. (A) AIr is cytotoxic to control fibroblasts with IC50 = 120 μmol/l in contrast to SAICAr which does not show any effect in the tested range of concentrations. (B) AIr and SAICAr are only mildly toxic to CAD5 cells with IC50 values 332 and 618 μmol/l, respectively. Each data point represents the mean of three experiments. Vertical bars represent S.D. (n = 3).

p.Lys53Arg mutation compromises PAICS activity in the patient’s skin fibroblasts and of the corresponding purified recombinant protein

To characterize the molecular consequences of the identified mutation, we investigated cultured skin fibroblasts obtained from proband II.4 (Case 1), the heterozygous parents (I.1 and I.2), one heterozygous sibling and three healthy controls.

We isolated total RNA and performed reverse transcription polymerase chain reaction (RT-PCR) analysis and found a single PCR product of an expected size of 1509 bp in an amount that was comparable to control specimens suggesting that the mutation has no negative effects on PAICS mRNA expression, splicing and stability (Fig. 3B).

Using western blot analysis, we found that the amounts of the PAICS enzyme were similar in the proband and controls (Fig. 3C). Using AIR as a substrate, we found in cell lysates, in contrast to controls, the enzyme activities of PAICS were reduced to approximately 10% in the proband and to 50% in heterozygous carriers (Fig. 3D).

Using established methods (12, 13), we cloned the c.158A>G mutation into the pMAL-c2 vector and expressed and purified the wild type (MBP-PAICS_wt) and mutated MBP-PAICS_K53R proteins from E. coli. Using AIR as a substrate, we found that the mutant protein had reduced activity to 25% in contrast to the wild type (Fig. 3E).

Identified mutation prevents purinosome formation in the patient’s skin fibroblasts, and this phenotype can be corrected by transfection with the wild-type but not the mutated PAICS

We showed earlier that mutations of DNPS enzymes affect purinosome assembly in fibroblasts from individuals with ADSL deficiency and AICA-ribosiduria (14) and in CRISPR-Cas9 genome-edited HeLa cells deficient for the individual steps of DNPS (12). To assess the effect of the K53R PAICS mutation on purinosome formation, we cultured skin fibroblasts of the proband in purine-depleted (PD) medium and immunodetected PPAT and GART using confocal fluorescent microscopy as previously described (14). When cultured in the PD medium, control fibroblasts demonstrated purinosome formation with characteristic granular staining and significant colocalization of PPAT and GART in the cytoplasm (Fig. 4A–D). In contrast, affected fibroblasts cultured under identical conditions showed diffuse intracellular staining of PPAT and GART with no significant signal overlap (Fig. 4E–H). Purinosome formation was restored in affected fibroblasts cultured in the PD medium after transient transfection with the eukaryotic expression vector encoding the wild-type PAICS (pTagBFP-PAICS_wt) (Fig. 4I–L). Transfection with the vector encoding mutant PAICS (pTagBFP-PAICS_K53R) did not restore purinosome formation (Fig. 4M–P).

AIR and AIr that accumulate in the cellular models of complete PAICS deficiency were not detected in the patient’s skin fibroblasts

The other two known DNPS defects—ADSL deficiency and AICA-ribosiduria—are characterized by the accumulation of two dephosphorylated substrates of ADSL—succinyladenosine (SAdo) and succinylaminoimidazole carboxamide riboside (SAICAr). These substrates can be measured in the urine, plasma, cerebrospinal fluid (CSF) and dried blood spots of affected patients (15, 16). Recently, we developed a quantitative liquid chromatography–tandem mass spectrometry (LC-MS/MS) method for detection of AIR and AIr, which are predicted to accumulate in the body fluids of patients with PAICS deficiency (17, 18) similar to the accumulation of these metabolites in the growth media of Chinese hamster ovary (CHO) cells and HeLa cells that are completely deficient in PAICS (8, 19). As there were no body fluids available from the affected individuals, we measured but could not detect AIR or AIr in the cell lysate and growth medium of the patient fibroblasts cultured in purine-depleted media and in purine-rich media. The limits of detection (LODs) and quantification (LOQs) of AIr (defined using signal-to-noise ratios of 3:1 and 10:1, respectively) were 12 and 62 nmol/l, respectively, in growth medium and 3.6 and 18.6 nmol/l, respectively, in water.

The metabolite AIr that is predicted to accumulate in the body fluids of patients with PAICS deficiency demonstrates higher cytotoxicity then SAICAr

The main pathogenic mechanism of ADSL and ATIC deficiencies has been attributed to the cytotoxicity of accumulated dephosphorylated succinylpurines, especially of SAICAr (20). To assess the cytotoxic and neurotoxic effects of AIr, which is expected to accumulate in PAICS deficiency, and compare it to SAICAr, which accumulates in ADSL deficiency, we cultured control fibroblasts and Cath a-differentiated catecholaminergic cells (CAD 5) (21) in a growth medium containing AIr or SAICAr in a concentration gradient from 1.7 μmol/l to 1 mmol/l.

Cell viability was assessed after 72 h of cultivation by determining the level of intracellular ATP with ATPlite™ (PerkinElmer) luminescence system. We observed cytotoxic activity of AIr on both fibroblasts and CAD 5 cells with an estimated half-maximal inhibitory concentration (IC50) of 120 and 332 μmol/l, respectively. In contrast to AIr, SAICAr demonstrated weaker cytotoxicity on both cell types. It was negligible on fibroblasts (IC50 > 1 mmol/l) and showed an IC50 of 618 μmol/l on CAD 5 cells (Fig. 5).

Stimulatory effects of AIr and SAICAr in concentrations <14.3 μM in CAD 5 cells and a stimulatory effect of AIr in concentrations <41.4 μM in fibroblasts were identified as well (Fig. 5).

Discussion

We describe two siblings from The Faroe Islands with multiple malformations and early neonatal death who were born to consanguineous parents. Both affected individuals were homozygous for a rare missense mutation in PAICS encoding PAICS, a bifunctional enzyme involved in DNPS. We demonstrate similar laboratory findings as in two known DNPS disorders, ADSL deficiency and AICA-ribosiduria, and a similar but very severe neonatal phenotype with fatal outcome (Table 1). The mutation affects the catalytic site of the enzyme and causes enzyme deficiency in affected skin fibroblasts and in corresponding recombinant mutant proteins. Moreover, it disturbs purinosome formation in the patient’s skin fibroblasts, which can be corrected by transfection with the wild-type but not the mutated PAICS.

Table 1

Comparison of malformations in three DNPS disorders processed by the OMIM database

	AICA-ribosiduria	ADSL deficiency	PAICS deficiency
Growth retardation		+	+ (Intrauterine)
Polyhydramnios			+
Brachycephaly	+	+	+
Microcephaly		(+)
Short neck			+
Prominent forehead	+
Prominent metopic suture		+
Flat face			+
Hypertelorism			+/−
Low-set ears	+	+	+
Poorly modulated ears			+
Congenital blindness	+		n.a.
Optic atrophy	+		n.a.
Strabismus/nystagmus		+
Small nose		+	+
High nasal bridge	+
Low nasal bridge			+
Anteverted nostrils	+	+	+
Choanal atresia			+
Wide mouth	+	+
Thin upper lip	+	+
Long smooth philtrum		+
Atrial septal defect	+
Pulmonary hypoplasia/malformation			+
Esophagus atresia			+
Genitourinary abnormalities	Prominent clitoris, fused labia minora		Micropenis, subcoronal hypospadias
Multiple skeletal malformations			+++
Cognitive difficulties/encephalopathy	+	+	−

ADSL deficiency and AICA-ribosiduria lead to a diagnostic accumulation of the corresponding dephosphorylated enzyme substrates—SAICAr, SAdo and aminoimidazole carboxamide riboside (AICAr)—in the bodily fluids of patients (2, 3). Similarly, PAICS deficiency should lead to intracellular accumulation of the enzyme substrate AIR and the presence of its dephosphorylated analog AIr in body fluids. Using liquid chromatography coupled either with electrochemical detection (LC-EMD) or with tandem mass spectrometry (LC-MS/MS), we also showed earlier that CHO cells and CRISPR-Cas9 genome-edited HeLa cells deficient in PAICS accumulate AIr in the growth medium (8, 19). Based on the lack of body fluids from affected individuals, we measured but detected neither AIR nor AIr in patients’ fibroblasts. This is similar to fibroblasts from individuals with ADSL deficiency, in which we did not detect SAICAr (data not shown). Our inability to detect AIr could be attributed to the approximatively 4-times higher LOD and LOQ of AIr in the growth medium in contrast to water.

The main pathogenic mechanism of ADSL deficiency has been attributed to the cytotoxicity of SAICAr (20), which has previously been reported to have levels of 0.12 mm in cerebrospinal fluid and 1.3 mm in the urine samples of affected individuals (22). Therefore, higher levels of AIr metabolite in bodily fluids may cause similar clinical impairment (Table 1). The neurotoxic and cytotoxic effects of presumably accumulating AIr were demonstrated on CAD5 and fibroblasts cells. Moreover, we found stimulatory effects of both AIr and SAICAr in these cells. The biphasic dose-response effect has been shown recently in a study of the anti-cancer effect of l-ascorbic acid in a colorectal cancer cell line (23, 24). This phenomenon called also hormetic dose-response model is a biological feature commonly found in toxicological studies (25). The mechanism of stimulatory/inhibitory action can be explained by different affinity of molecule to particular receptor or part of the signaling pathway within the metabolism of the cell (26). This hypothesis considers further examination.

In conclusion, we have identified a new inherited metabolic disorder of de novo purine synthesis—PAICS deficiency. The previous lack of detection and rarity of this condition may be based on recurrent spontaneous abortion or early neonatal death. We are interested to study other similar cases and can provide the genetic testing and biochemical screening described here. Please contact [email protected].

Material and Methods

Study subjects and clinical examination

The study was approved by the appropriate institutional review boards and the investigations were performed according to the Declaration of Helsinki principles. Adults provided informed consent, and the siblings of the affected individuals provided assent with parental consent. Participants provided venous blood samples, and genomic DNA was isolated using standard technology. Punch skin biopsies were obtained under local anesthesia, and fibroblasts were cultured according to standard protocol.

Chemicals

DNPS intermediates—SAICAR, AIR and their dephosphorylated forms SAICAr and AIr—were synthesized as previously described (8, 27). Dulbecco’s minimum essential medium (DMEM), Opti-MEM/Reduced Serum Medium, F12 nutrition mix, Glutamax and fetal bovine serum (FBS) were obtained from Life Technologies. All other chemicals were purchased from Sigma-Aldrich unless otherwise stated in the text.

Homozygosity mapping and sequencing analysis

DNA was isolated by standard methods from venous blood and fibroblasts obtained after skin biopsies. Genotyping was performed on the Affymetrix 250K NspI array (262 000 SNPs) according to the manufacturer’s instructions. SNP calling was calculated using the BRLMM algorithm in the Genotyping Console 3.0.2 software (Affymetrix). Homozygous regions were determined using the MERLIN algorithm. Whole exome sequencing (WES) was performed at the genomic platform of the Imagine Institute, Paris, France. Agilent SureSelect libraries were prepared using the exome capture kit 51 Mb SureSelect Human All Exon kit V5 (Agilent Technologies) from 3 μg of sheared genomic DNA with a Covaris S2 Ultrasonicator. NGS was carried out using HiSeq 2500 (Illumina) generating paired-end reads.

After demultiplexing, paired-end sequences were aligned to the reference human genome (NCBI build37/hg19 version) using the Burrows–Wheeler Aligner. Downstream processing was carried out with the Genome Analysis Toolkit (GATK), SAMtools and Picard (http://www.broadinstitute.org/gatk/guide/topic?name5best-practices). Variant calls were made with the GATK Unified Genotyper. For each sample, the mean depth of coverage obtained was 103 with at least 94% of the exome covered at least 30×. An in-house software (Poly-Web based on Ensembl release 71, http://www.ensembl.org/index.html) was used to annotate and filter variants according to a homozygous genetic model. We excluded variants from public data bases (dbSnp, 1000 genomes, Evs, Exac and gnomAD) with a frequency >1% as well as variants previously identified in ‘in house exomes’.

In order to verify the familial segregation of the variant a bidirectional Sanger sequencing was performed using the ABI BigDye Terminator v.3.1 Cycle Sequencing Kit (Life Technologies).

RNA analysis and cDNA sequencing

Total RNA and cDNA were isolated from cultured skin fibroblasts of proband II.4 (Case 1) and the control by ProtoScript II First Strand cDNA Synthesis Kit (NEB). PCR analysis was performed according to standard procedures. The PCR was conducted in 25 μl reaction mixtures containing Red PCR Master Mix (Rovalab), 1.5 mm MgCl₂, 8% DMSO and 0.4 μM specific primers covering cDNA from start to stop codon. The amplicons were sequenced using the specific 813L primer by Sanger sequencing service (Eurofins Genomics).

Structural impact of the identified mutation

For structural mapping, available crystal structure of PAICS (PDB ID 2H31) (11) was analyzed. Structural models were visualized using the PyMOL Viewer (DeLano Scientific, Palo Alto, CA, USA). Protein stability of the mutant was further assessed using mCSM software (28).

Cell cultures

Primary skin fibroblasts were maintained in Dulbecco’s minimum essential medium/F12 nutrient mixture (DMEM/F12, Gibco, Life Technologies) supplemented with 10% FBS (Gibco, Life Technologies), 1% penicillin/streptomycin (Sigma-Aldrich) and 0.03 mm adenine. The purine-depleted media was prepared from DMEM (Gibco, Life Technologies) supplemented with dialyzed 10% FBS and 1% penicillin/streptomycin. FBS was dialyzed against 0.9% NaCl at 4°C for 48 h using a 10 kDa MWCO dialysis membrane to remove purines.

CAD-2A2D5 (CAD5) cells derived from Catha-differentiated (CAD) cells were provided by Sukhvir Mahal, The Scripps Research Institute, FL, USA. CAD5 cells were maintained in modified Eagle’s Minimum Essential Media (Opti-MEM/Reduced Serum Media, Gibco, Life Technologies) supplemented with 10% FBS, 2 mm Glutamax, 1 mm pyruvate, 1% penicillin/streptomycin without phenol red.

Preparation of fibroblasts’ lysates

Cells (1.5 × 10⁶) were washed with PBS and centrifuged at 400g for 5 min at 4°C. The pellet was suspended in 30 μl of lysis buffer containing 30 mm KH₂PO₄, pH 6.0, 0.5% polyethylene glycol ether W-1 (Sigma-Aldrich) and Protease Inhibitor Cocktail Tablets (Roche) and incubated on ice for 45 min. The solution was sonicated four times for 5 s and centrifuged at 17000g for 20 min at 4°C.

Catalytic activity of PAICS enzyme in patient’s skin fibroblasts

Proband II.4 1.5 × 10⁶ (Case 1), the heterozygous parents (I.1 and I.2), a heterozygous sibling and three control skin fibroblasts were lysed, and protein concentrations were determined by Bradford reagent (Sigma-Aldrich). PAICS activities were assayed at 37°C in 50 mm Tris, pH 7.4, 5 mm aspartate, 0.3 mm ATP, 180 mm KHCO₃, 24.5 mm MgCl₂, 350 mg/l SF lysate and 0.25 mm AIR for 2 h. The reaction product SAICAR was quantified by HPLC analysis as previously described (8, 13). All experiments were repeated three times.

Western blot

The cell lysates were separated by 10% SDS-PAGE, and proteins were blotted onto a PVDF membrane. The membrane was blocked with 5% BSA in PBS-T (PBS with 0.05% Tween 20) and probed with primary antibodies: mouse monoclonal anti-PAICS (OriGene) and mouse monoclonal anti-GAPDH (Sigma-Aldrich) diluted in 5% BSA in PBS-T. The target proteins were detected using peroxidase conjugated secondary antibodies: goat anti-mouse IgG (Sigma-Aldrich) and goat anti-mouse IgM (Pierce).

Chemiluminescent detection was performed using the Clarity Western ECL Substrate (Bio-Rad) and visualized by x-ray imaging.

LC-MS/MS analysis

Before analysis, the cell media were concentrated from 1 mL to 60 μL under a nitrogen stream at 37°C. Five percent perchloric acid (16.6 μl) was added to 30 μl of concentrated medium or cell lysate, and samples were incubated 5 min at 4°C. After centrifugation, the supernatants were neutralized with 3.2 μl of 2.5 M KHCO₃, incubated 10 min at 4°C and centrifuged again. Five microliters of the sample was injected into the LC-MS/MS system consisting of an Agilent 1290 Infinity LC System (Agilent Technologies) equipped with a ProntoSIL 120-3-C18 AQ column (200*4 mm, 3 μm, Bischoff) coupled with an API 4000 triple quadrupole mass spectrometer with an electron spray ionization operated by Analyst software (Applied Biosystems), as previously described (8). The limits of detection and quantification were determined using signal-to-noise ratios of 3:1 and 10:1, resp. from 3 nM to 50 μM concentrations of AIr standard diluted in water or in the purine-depleted (PD) media. The AIr samples were treated in the same manner as the growth media of patient fibroblasts before LC-MS/MS analysis. A retention time of AIr in PD media was 5.2 min, in water 4.7 min.

Cloning of pMAL-PAICS_wt, pMAL-PAICS_K53R, pTagBFP-PAICS_wt and pTagBFP-PAICS_K53R

The cDNA coding for wild-type PAICS was cloned into the expression vectors pMAL-c2 (NEB) and pTagBFP-C (Evrogene) as previously described (8, 12). The mutation K53R was cloned into both vectors by the GeneArt™ Site-Directed Mutagenesis System (Thermo Fisher Scientific) using standard procedures.

Catalytic activities of recombinant PAICS proteins

The proteins MBP-PAICS_wt and MBP-PAICS_K53R were expressed in E. coli and affinity purified as previously described (29). PAICS activities were assayed as described previously, using 25 mg/l of the purified fused protein.

Transfection and immunofluorescence

For transfection, 1 × 10⁵ of patient or control skin fibroblasts were transiently transfected with 1.5 μg of pTagBFP-PAICS_wt or pTagBFP-PAICS_K53R. The transfection and immunofluorescence labeling was performed as previously described (12) with the following primary antibodies: rabbit polyclonal IgG anti-PPAT (Sigma-Aldrich) and mouse polyclonal IgG anti-GART (Abnova). For fluorescence detection, species appropriate Alexa Fluor® 488 and 555 secondary antibodies (Thermo Fisher Scientific) were used. All immunofluorescence experiments were repeated at least twice.

Image acquisition and analysis

Slides were mounted with ProLong® Gold Antifade Mountant (Thermo Fisher Scientific) as the fluorescence mounting medium and analyzed by confocal microscopy as previously described (12, 14). The resulting overlap coefficient values are presented in pseudocolor, and the scale is shown in the corresponding lookup tables (LUT). Image analysis was performed on at least 10 cells from each cell type. The conditions for image acquisition were identical for all cells evaluated in the experiment.

Substrate toxicity testing

Cells (7 × 10²) cells in 24 μl volume were seeded per well in a 384-well plate format. Cells were maintained in Dulbecco’s minimum essential medium (DMEM) or a modified Eagle’s Minimum Essential Media (Opti-MEM/Reduced Serum Media) supplemented with 10% FBS, 2 mm Glutamax, 1 mm pyruvate and 1% penicillin/streptomycin. For feeding experiments, cells were transferred to DMEM without phenol red supplementation as mentioned earlier or Opti-MEM for CAD5 cells, respectively. Cells were grown in 12 μl of media overnight at 37°C using a 5% CO₂ incubator. The following day, individual substrates, SAICAr and AIr, were prediluted in a media in a series of 14 concentrations between 1.7 μM and 1 mm. Twelve microliters of prediluted substrates was added to cells to achieve final concentration in a total volume of 24 μl per well.

The ATPlite™ luminescence assay system was purchased from Perkin Elmer. A luciferase-catalyzed reaction of ATP and luciferin results in production of light. The ATP concentration (representing cell viability) is proportional to emitted light. A total volume of 24 μl of media with cells was treated with 11 μl of ATPlite solution followed by 15 min incubation at room temperature and shaking on an orbital shaker. After the incubation step, luminescence was recorded on an EnVision plate reader (PerkinElmer) with proper optical aperture for a 384-well plate. Subsequent analysis was performed on standard software Microsoft Office Excel and GraphPad Prism software.

Funding

The part of work conducted at the First Faculty of Medicine was supported by Charles University [program numbers PRIMUS/17/MED/6, PROGRES Q26/LF1, UNCE/MED/007, SVV 260367/2017], by the Ministry of Health of Czech Republic [grant number NV19-07-00136] and by the Ministry of Education, Youth and Sports of CR: The National Center for Medical Genomics [grant number LM2015091], National Sustainability Programme II [grant number LQ1604] and National infrastructure for chemical biology [grant number LM2015063]. The part of work conducted at Imagine Institute was supported by State funding from the Agence Nationale de la Recherche, ‘Investissements d’avenir’ [program number ANR-10-IAHU-01].

Acknowledgements

We acknowledge the dedicated care of the index case by member of the Neonatal Intensive Care Unit at University Hospital, Copenhagen, and the pediatricians at the National Hospital of the Faroe Islands, Tórshavn, where both patients were treated; the Genetic Biobank of the Faroe Islands in Tórshavn for providing DNA of the patients’ family and information used for creating the family pedigree; and Ernst Christensen and Marianne Schwartz at the Metabolic Laboratory as well as Susanne Kjærgaard at the Chromosome Laboratory at University Hospital, Copenhagen, for providing DNA of the deceased infants and fibroblasts of the patient 1.

Conflict of Interest statement. None declared.

Author notes

These authors contributed equally.