Phosphoribosylformylglycinamidine Synthase (PFAS) Deficiency: Clinical, Genetic and Metabolic Characterisation of a Novel Defect in Purine de Novo Synthesis

Abstract

Purine de novo purine synthesis involves 10 reactions catalysed by six enzymes, including phosphoribosylformyglycinamidine synthase (PFAS). To date, genetic defects of three of these enzymes, namely ATIC, ADSL and PAICS, have been characterised in humans. Here, we report for the first time two individuals with PFAS deficiency. Probands were identified through metabolic and genetic screening of neurologically impaired individuals. The pathogenicity of the variants was established by structural and functional studies. Probands C1 and C2 presented with prematurity, short stature, recurrent seizures and mild neurological impairment. C1 had elevated urinary levels of formylglycineamide riboside (FGAr) and bi-allelic PFAS variants encoding the NP_036525.1:p.Arg811Trp substitution and the NP_036525.1:p.Glu228_Ser230 in-frame deletion. C2 is a 20-year-old female with a homozygous NP_036525.1:p.Asn264Lys substitution. These amino acid changes are predicted to affect the structural stability of PFAS. Accordingly, C1 skin fibroblasts showed decreased PFAS content and activity, with impaired purinosome formation that was restored by transfection with pTagBFP_PFAS_wt. The enzymatic activities of the corresponding recombinant mutant PFAS proteins were also reduced, and none of them, after transfection, corrected the elevated FGAR/r levels in PFAS-deficient HeLa cells. While genetic defects in purine de novo synthesis are typically considered in patients with severe neurological impairment, these disorders, especially PFAS deficiency, should also be considered in milder phenotypes.

Keywords: FGAR; PFAS deficiency; formylglycinamide riboside; metabolic disorder; purine de novo synthesis; purinosome.

FIGURE 1

Purine de novo synthesis and its defects. Purine de novo synthesis consists of a series of 10 enzymatic reactions catalysed by six key enzymes: Phosphoribosylamidotransferase (PPAT), trifunctional phosphoribosylglycinamide formyltransferase/phosphoribosylglycinamide synthetase/phosphoribosylaminoimidazole synthetase (GART), phosphoribosylformylglycinamidine synthase (PFAS), bifunctional aminoimidazole ribonucleotide carboxylase/phosphoribosylaminoimidazolesuccinocarboxamide synthase (PAICS), adenylosuccinate lyase (ADSL) and bifunctional 5‐aminoimidazole‐4‐carboxamide ribonucleotide transformylase/IMP cyclohydrolase (ATIC). Enzyme deficiencies in this pathway are characterised biochemically by the accumulation of specific intermediates in patients' body fluids: PFAS deficiency leads to the accumulation of formylglycinamide riboside (FGAr); PAICS deficiency to aminoimidazole riboside (AIr); ADSL deficiency to succinylaminoimidazole carboxamide riboside (SAICAr) and succinyladenosine (SAdo) – the latter being an intermediate of the purine nucleotide cycle (not shown); and AICAribosiduria (ATIC deficiency) to the accumulation of 5‐aminoimidazole‐4‐carboxamide ribonucleoside (AICAr), SAICAr and SAdo.

FIGURE 2

Pedigree, cDNA sequencing and LC–MS/MS profiling of purine metabolites in urine and serum samples from the case 1 family. (A) Pedigree of the family: Black symbols represent the affected individual, (+/−) indicates the presence (+) or absence (−) of the noted variant. (B) Accumulation of formylglycinamide riboside (FGAr) in the urine and (C) serum of the patient and his siblings or parents. FGAr concentrations were significantly increased in both the urine and serum of the patient compared to control samples. Data are presented in standardised boxplot graphs, showing the box from the first to third quartiles, whiskers representing the minimum and maximum, and all data points displayed. Statistical significance was assessed using one‐way ANOVA in GraphPad software, with p‐values indicated as **** for p ≤ 0.0001 and ns (non significant) for p > 0.05. (D) cDNA sequencing of PFAS transcripts from patient blood shows an in‐frame deletion (NM_012393.3:c.681_689del) caused by the genomic PFAS variant NC_000017.10:g.8159584G>A. This aberrant splicing is absent in the paternal cDNA, which shows only the wild‐type sequence. A second variant, NM_012393.3:c.2431C>T (missense), is present in the patient and the maternal cDNA displays only the wild‐type sequence at this site.

FIGURE 3

Characterization of PFAS variants: Structural and functional analyses in fibroblasts and recombinant proteins. (A) Structural mapping of the variants using a theoretical model generated by Alpha‐Fold (AF‐O15067‐F1). The positions of the auxiliary ATP and active sites were localised based on the crystal structures of PFAS from Salmonella typhimurium (PDB ID: 1T3T) and from Thermotoga maritima (PDB ID: 2HS4). The positions of the variants are highlighted as orange spheres. The in‐frame deletion Glu228_Ser230del affects a flexible part of the helical region connecting the linker between the N‐domain and the FGAM synthetase domain. The Asn264Lys and Arg811Trp substitutions are located at the FGAM synthetase domain, distant from the active site. They can induce local destabilization due to steric clashes and gain of polar interaction with Asp761 in the case of Asn264Lys or due to the presence of a more hydrophobic residue at the protein surface in the case of Arg811Trp. (B) Western blot analysis of skin fibroblasts using antibodies against GART, PFAS, PAICS, ADSL and ATIC revealed reduced levels of PFAS and PAICS, while the levels of other DNPS enzymes were comparable to those in control fibroblasts. Protein levels were normalised to GAPDH, Actin and MYH9. (C) PFAS enzyme activity in fibroblasts from case 1 was reduced to 16% of control levels. (D) The catalytic activity of the recombinant Flag_PFAS protein Glu228‐Ser230del was decreased to 14%, Arg811Trp to 57% and Asn264Lys to 11% activity compared to Flag_PFAS_wt. Data are presented as standardised boxplot graphs, with the box spanning the first to third quartiles, whiskers showing the minimum and maximum values, individual data points displayed, and the median represented by a line. Experiments were performed at least three times (n ≥ 3). Statistical significance was assessed using one‐way ANOVA in GraphPad software, with p‐values indicated as ** for p ≤ 0.01, *** for p ≤ 0.001 and **** for p ≤ 0.0001.

FIGURE 4

Functional complementation of PFAS deficiency. (A) Purinosome formation in skin fibroblasts of case 1. Patient and control skin fibroblasts were cultured in purine‐depleted medium and confocal fluorescence microscopy was used to detect fluorescently labeled endogenous proteins PPAT and GART. In control fibroblasts, purinosome formation was observed, characterised by granular staining and significant colocalization of PPAT and GART in the cytoplasm. In contrast, fibroblasts from case 1 displayed diffuse cytosolic staining of PPAT and GART with no significant overlapping signals. (B) Purinosome formation was restored following transient transfection with an eukaryotic expression vector encoding wild‐type PFAS (pTagBFP_PFAS_wt). However, transfection with vectors encoding pTagBFP_PFAS variants, Glu228‐Ser230del and Arg811Trp, failed to restore purinosome formation. (C) Transfection of PFAS‐deficient HeLa cells with the pcDNA4_Flag_PFAS_Glu228‐Ser230del, pcDNA4_Flag_PFAS_Arg811Trp and pcDNA4_Flag_PFAS_Asn264Lys constructs corresponding to patients' variants did not correct FGAR/r levels in cell lysate, whereas transfection with the wild‐type construct decreased both.