Given the clinical findings with unknown ESRD, we suspected the occurrence of a metabolic disease. After examining the proband affected by PH1 (June 2008), we synthesized the primers for all the exons and exon–intron boundaries of the AGXT gene (Supplementary Table S2). Sanger sequencing was routinely performed on the coding regions and the exon–intron boundaries of the AGXT gene (NM_000030) in subjects 16, 9, and 10. We detected a missense mutation on exon 4 (c.517T > C/p.Cys173Arg; Figure 1B) and validated the mutation by Sanger sequencing using reverse primer. The mutation originated from the patients’ healthy mother (Figure 1A). At this Cys173 amino acid position, the p.Cys173Tyr mutation associated with severely reduced catalytic activity was previously reported on a PH1 patient (Von Schnakenburg and Rumsby, 1998). Human PH1 is clearly caused by AGXT recessive mutation. However, no other pathogenic variant of the AGXT gene was detected in the trios-family despite the successful amplification and sequencing of all the exons and boundaries of AGXT. The PH1 diagnosis in the family was stalled because of another mutated AGXT allele, which was not detected in the family.

On December 2016, subject 17 was admitted in our hospital because of clinical symptoms similar to those in subject 16. The high phenotypic similarity of the two infants prompted us to conduct next-generation sequencing on the family (Figure 1A; whole-exome sequencing on subjects 9, 10, 16, and 17). We envisage that next generation sequencing may generate new data for the diagnosis of the disease in the family (Shen, 2018). We hypothesized that a recessive metabolic disorder affected the family. Considering non-consanguineous mating, we mainly focused on the compound heterozygous mutations in the family. Common variants in public databases (the variant was neither found in ExAC nor 1000 Genomes) and in-house databases were filtered out. Given that both patients died of ESRD, we focused on the genes involved in renal disorders (Supplementary Table S3). We detected the two heterozygous variants (chr2:241810859T > C and chr2:241813466A > C) of the AGXT gene segregated with PH1 in the family (Figure 1 and Supplementary Figure S1). Both AGXT variants were missense variants and caused the disease according to the reports of four prediction software programs (Table 2).

The variant chr2:241810859T > C was identified in exon 4 of the AGXT gene and harbored a change in residue 173 from cysteine to arginine (Figure 1B). The mutation of AGXT-exon 4 (chr2:241810859T > C) was detected as mentioned above. By contrast, variant chr2:241813466A > C was identified in exon 6 of the AGXT gene and harbored a change in residue 223 from serine to arginine. The AGXT-exon 6 mutation (chr2:241813466A > C) was missed in the first round of Sanger sequencing (mentioned above; detailed data are presented in the Discussion section). After re-synthesis, a new pair of primers (2AGXT6F: 5′-CATCTCCCCTGCTATCGTGTAC-3′; 2AGXT6R: 5′-CCTCAGTCCTTTCCTGGTCAC-3′, predicted PCR product size: 498 bp) was used for AGXT-exon 6, and the c.667A > C/p.Ser223Arg mutation (Figure 1C) was detected in the second round of Sanger sequencing.

The wild-type AGXT gene can carry two polymorphic variants: the rs4426527 (p.Ile340Met) and the less frequent minor haplotype (rs34116584, p.Pro11Leu). In this study, we utilized the results of next-generation sequencing (Supplementary Figure S1). However, we discovered that none of the family members carried any of the two variants. Therefore, both mutations are on the major haplotype.

Subject 17 died of ESRD, and a postmortem examination was performed. No visible stone was observed inside both kidneys of the patient (Figure 2A,B). However, global nephrocalcinosis occurred in both kidneys, as observed under a polariscope (Figure 3). We obtained 5 g of the kidney tissue and placed the sample in a hot air oven for 2 h (Figure 4A,B). Crystals were collected for composition analysis (infrared spectroscopy test) and were identified as calcium oxalate monohydrate (Figure 4C).

This study was approved by the ethics committee of the Hunan Children’s Hospital, Changsha City, China. The procedure of the committee conformed with the principles of the 2008 edition of the declaration of Helsinki.

Before this study, written informed consent was obtained from the guardian of the human subject for the publication of this study.

This study includes a Han, Chinese family (resides in the rural area of Hunan Province) with PH1 (Figure 1) and 496 ethnicity- and region-matched controls (the controls were selected from patients who came to our laboratory for GTG banding but who were excluded for having renal disorders).

In the family, subjects 16 and 17 were affected by the infantile form of PH1, whereas subjects 4, 10, 11, and 13 were affected by benign urolithiasis at least once in that a kidney stone composition analysis indicated that all the crystals were calcium oxalate monohydrate, and other family members were asymptomatic.

For each subject, genomic DNA was extracted from peripheral blood (2–5 mL in heparin sodium tubes) through the phenol/trichloromethane method, as described by the standard protocol.

For mutation screening, the coding and intron-exon boundary regions of the AGXT gene (NM_000030, according to GRCh37/hg19) were amplified by PCR, using the primers synthesized by a local biotech company (BGI; Shenzhen, China) and designed with Primer3². The sequencing reaction (BigDye 3.1 Kit, Applied Biosystems, Waltham, MA, United States) of the purified PCR products was performed according to the recommended procedure. The labeled PCR fragments were purified through 70% alcohol precipitation and electrophoresed on an ABI-A3500 genetic analyzer (Applied Biosystems, Waltham, MA, United States). The primer sequences, PCR conditions, and sequencing are shown in Supplementary Table S2.

For the validation of the AGXT-exon6 mutation (c.667A > C: p.Ser223Arg), two pairs of primers were used as presented in Supplementary Figure S3.

The whole exome sequencing in the present study was performed according to the pipeline as reported elsewhere (Jin et al., 2018). In brief, two micrograms of genomic DNA for each sample (Subjects 16 and 17, Figure 1) was sheared into approximately 200-bp fragments. The fragments were enriched with ligation-mediated PCR. Exome capture (NimbleGen 2.1M HD array) was performed according to the manufacturer’s instructions (Roche NimbleGen, Inc., Madison, WI, United States). The captured library was sequenced on a HiSeq2000 sequencing platform (90 bp end reads, Illumina Inc., San Diego, CA, United States), and the Illumina base calling software V1.7 was used for the analysis of raw image files with default parameters. Clean reads were mapped to the reference human genome (hg19) ³, by using the BWA (Burrows–Wheeler Aligner) program with, at most, two mismatches ⁴. The alignment files (.bam) were generated with SAM tools ⁵ and reads of low mapping quality (< Q30) were filtered out. Clonal duplicated reads that may have been derived from PCR artifacts were removed with Picard Tools using default parameters⁶. Short read alignment and annotation visualization were performed by using the Integrative Genomics Viewer⁷. The percentage of alignment of the clean read to the exome regions was obtained by using our custom Perl scripts on the base of the alignment files. Single nucleotide variants (SNVs) and indels were detected with a genome analysis tool kit⁸. All detected SNVs and indels were comprehensively annotated by ANNOVAR⁹, including function implication (gene region, functional effect, mRNA GenBank accession number, amino acid change, cytoband, and so on) and allele frequency, in dbSNP, 1000 Genomes¹⁰, ESP6500 ¹¹, and ExAc ¹². Damaging missense mutations were predicted by SIFT ¹³, PolyPhen-2 ¹⁴, and MutationTaster ¹⁵.