The clinical data of a patient with ATP6AP2-CDG was gathered at the Children’s Hospital of Fudan University. This study was approved by the Ethics Committee of Children’s Hospital of Fudan University (No. FELL-2020-402). Any potentially identifiable photographs or data were published with the parents’ written informed consent.

The material was removed from the liver transplant procedure, fixed in 10% buffered formalin, dried in varying degrees of ethyl alcohol, and then embedded in paraffin. Hematoxylin and eosin (HE) staining was followed by periodic acid Schiff (PAS) with and without diastase, Masson, reticulin, copper, and iron special staining.

Additional 4 µm sections were deparaffinized, rehydrated, and pretreated with 3% H₂O₂ to eliminate endogenous peroxidase activity. Besides, the sections were heat-mediated antigen-retrieved with EDTA (pH 9) or citrate buffer (pH 6) before immunohistochemical staining was undertaken. A variety of antibodies were used as primary antibodies, including ATP6AP2 (purchased from www.ptgcn.com, catalog number 10926-1-AP, 1:50 dilution), LC3A/B (purchased from www.affbiotech.com, catalog number AF5402, 1:100 dilution), and mTOR (purchased from www.affbiotech.com, catalog number AF6308, 1:100 dilution). After incubating overnight at 4°C, the sections were incubated at room temperature for an hour with the general secondary antibody. Then, the sections were developed with DAB and counterstained with hematoxylin. To serve as a control, we used a normal liver sample donated by a surgical patient.

We collected the patient’s and his parents’ EDTA-anticoagulated whole blood specimens to perform WES004 (Human All Exon V4) using the Illumina HiSeq X ten platform by MyGenostics (a commercial genetic testing company). Using the TruSeq DNA Library Preparation Kit, DNA libraries were created. The raw data were aligned to the reference human genome (GRCh37/hg19). For variation calling to compile single nucleotide variations (SNVs) and indels, the Genome Analysis Toolkit (GATK) (https://gatk.broadinstitute.org) was applied. The annotation and interpretation processes used ANNOVAR software and the Enliven^® Variants Annotation Interpretation System. Data were filtered in 1,000 Genome (www.1000 genomes.org), NHLBI Exome Sequencing Project (ESP6500) (https://esp.gs.washington.edu), Genome Aggregation Database (gnomAD) (https://gnomad.broadinstitute.org), dbSNP152 (https://www.ncbi.nlm.nih.gov/snp), and Exome Aggregation Consortium (ExAC) (http://exac.broadinstitute.org). In order to score the genetic variants and further compare them, damage prediction was conducted using Combined Annotation Dependent Depletion (CADD) (https://cadd.gs.washington.edu) and Mutation Significance Cutoff (MSC) (https://lab.rockefeller.edu/casanova/MSC). With a 99% confidence level and HGMD and ClinVar as the database sources, the MSC server was deployed to CADD, PolyPhen 2 (http://genetics.bwh.harvard.edu/pph2), and SIFT (https://sift.bii.a-star.edu.sg/www/SIFT_indels2.html). For analysis based on variant-disease and gene-disease relationships, Genomics England PanelApp, a crowdsourcing tool, was used (https://panelapp.genomicsengland.co.uk). The HGMD database, Online Mendelian Inheritance in Man (OMIM), and Human Phenotype Ontology (HPO) were utilized to correlate phenotype descriptions with variant and gene prioritization results. Genetic variants were categorized into five categories by the American College of Medical Genetics and Genomics (ACMG): pathogenic, likely pathogenic, variant of uncertain significance (VUS), likely benign, and benign. Following that, a VUS of ATP6AP2 was discovered in the patient with the transcript NM_005765, the mutation site was confirmed by Sanger sequencing in his parents.

MutationTaster (http://www.mutationtaster.org) predicted the toxicity of the amino acid shift brought on by the gene mutation. In addition, the conservation of the mutation site was analyzed among multiple species. We modeled the protein using SWISS-model (https://www.swissmodel.expasy.org) with the UniProtKB code O75787, and PyMol (http://www.pymol.org) was used to analyze and visualize the altered structure.

HEK293T cells were cultured in high-glucose DMEM (Gibco) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin at 37°C under 5% CO₂. The cDNA of ATP6AP2 was generated by gene synthesis (www.generalbiol.com) and cloned into pcDNA3.1 with an N-terminal FLAG-tagged version. The original plasmid was used as a template to introduce the site mutation ATP6AP2^Gly62Glu by homologous recombination. All constructs were confirmed by DNA sequencing. The plasmids were chemically transformed into competent DH5α E. coli (TIANGEN). Individual colonies were selected, and plasmid DNAs were extracted using OMEGA Endo-Free Plasmid Midi Kit D6915. Then, Beyotime Lipo8000™ Transfection Reagent was used for plasmid DNA transfection following the manufacturer’s instructions.

Total RNA was extracted using AG RNAex Pro Reagent. Reverse transcription of the total RNA into cDNA was carried out using Evo M-MLV RT Premix (Accurate Biotechnology). The relative expression of endogenous, wild-type, and mutant ATP6AP2 was determined by quantitative real-time polymerase chain reaction (qRT-PCR) using SYBR Green Premix Pro Taq HS Kit (Accurate Biotechnology) on a Bio-Rad system. The forward primer sequence is 5′CTG​CAT​TGT​CCA​TGG​GCT​TC3′, and the reverse primer sequence is 5′AAC​AGG​TTA​CCC​ACT​GCG​AG3′.

Cells were harvested in 100 µL RIPA buffer supplemented with 1 µL phenylmethanesulfonyl fluoride lysed on ice for 30 min. The lysates were then centrifuged at 12,000 rpm for 20 min at 4°C. The supernatant was measured for the protein concentration using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) and was denatured with 1× SDS-PAGE Sample Loading Buffer (Beyotime) at 100°C for 10 min. We measured 10 µg total protein of each aliquot, and it was analyzed by SDS-PAGE and immunoblotted to a hydrophobic polyvinylidene fluoride membrane. After 1 h of milk blocking, the membranes were incubated with primary antibodies at 4°C overnight followed by second antibodies. Bands signal was detected by Tanon 5,200 automatic chemiluminescence image analysis system, and the relative expression was calculated by ImageJ software.

The following antibodies were used: ATP6AP2 (purchased from www.ptgcn.com, catalog number 10926-1-AP, 1:1,000 dilution), LC3A/B (purchased from www.affbiotech.com, catalog number AF5402, 1:1,000 dilution), mTOR (purchased from www.affbiotech.com, catalog number AF6308, 1:1,000 dilution), FLAG tag (purchased from www.huabio.cn, catalog number HA601080, 1:5,000 dilution) and GAPDH (purchased from Zsbio, catalog number TA-08, 1:2000 dilution). HRP-labeled Goat Anti-Rabbit (purchased from www.beyotime.com, catalog number A0208, 1:20000 dilution) and HRP-labeled Goat Anti-Mouse IgG (purchased from www.beyotime.com, catalog number A0216, 1:20000 dilution) were used as second antibodies.

HEK293T cells harboring ATP6AP2^WT and ATP6AP2^Gly62Glu were cultured in 10 cm dishes. Each group of six samples was extracted with extractant containing internal standard (methanol:acetonitrile:water = 2:2:1, V/V/V) for metabolite extraction. Metabolomic data analyses were conducted using non-targeted metabolomic profiling by Biomarker Co., Ltd. The Waters Acquity I-Class PLUS ultra-high performance liquid tandem mass spectrometer and the Waters Xevo G2-XS QTof high-resolution mass spectrometer made up the LC/MS equipment used for the metabolomics investigation. The column used was purchased from Waters Acquity UPLC HSS T3 column (1.8 um 2.1*100 mm). Based on the Progenesis QI software’s online METLIN database and Biomark’s own self-built library for identification, the raw data collected using MassLynx V4.2 were processed by Progenesis QI software for peak extraction, peak alignment, and other data processing operations. At the same time, theoretical fragment identification and mass deviation were both within 100 ppm. The follow-up study was carried out after the first peak area data was normalized with the total peak area. In order to assess the repeatability of the samples within the group and the quality control samples, principal component analysis and Spearman correlation analysis were utilized. The KEGG, HMDB, and lipid mapping databases were examined for classification and route information pertaining to the discovered chemicals. The OPLS-DA modeling was carried out using the R language package, and the model’s dependability was examined using 200 permutations.

Data were analyzed using SPSS 20.0 and statistical plotting was performed using GraphPad Prism 8.0 software. Metric data were expressed in (mean ± standard deviation, ± s). A parametric t-test was used for comparisons of metric data that followed a normal distribution, and a nonparametric t-test was used when the normal distribution was not followed. Statistics were considered significant for p-values under 0.05.

The boy was born to healthy non-consanguineous parents and was delivered by a full-term cesarean section with a birth weight of 3.5 kg. From 3 days to 2 months age, he was hospitalized three times for jaundice. During routine blood tests, remarkably elevated total bilirubin, direct bilirubin, alanine aminotransferase, aspartate aminotransferase, γ-glutamyl transferase, total cholesterol, low-density lipoprotein (LDL) cholesterol, international normalized ratio, prolonged prothrombin time, and decreased serum ceruloplasmin were noticed. There was no hepatosplenomegaly, no significant anemia, no elevated reticulocytes, and no hemolytic manifestations. The patient was discharged after the condition improved, followed by treatment with ursodeoxycholic acid (UDCA), compound glycyrrhizin tablets, and traditional Chinese medicine (details are not available). However, the hyperbilirubinemia, hypertransaminasemia, coagulopathy, and decreased serum ceruloplasmin persisted, and massive ascites emerged during the approximate 9-month follow-up. Besides, alpha-fetoprotein (AFP) (1,395 ng/mL, reference range 0–8.1 ng/mL) was elevated at 10 months of age, which might result from his young age and liver impairment.

Then the 11-month-old boy was admitted to our hospital for further therapy. Physical examination showed that his height was 70 cm (<3rd percentile by the WHO standard) and weight was 7.4 kg (<3rd percentile). There was mild yellowing of the sclera and skin throughout the body without rash or bleeding points. Cutis laxa was noticed in the patient. There was an II/VI murmur of the heart at the left sternal border. Abdominal distention was present. The liver was located 3.5–4 cm below the xiphoid process, with a tough texture. The spleen was located reaching the iliac fossa below the left costal, and right to the midline, with a diameter of 9–10 cm. No anomalies were discovered during the nervous system evaluation. Echocardiography and X-ray revealed cardiomegaly (Figure 1A-a). Normal cardiac function was assessed since a normal left ventricular ejection fraction without manifestations of left or right ventricular insufficiency. The patient did not present with hypertrophic cardiomyopathy as well. Abdominal ultrasonography indicated multiple hepatic focal lesions and splenomegaly; the former presented as an enhancement on computed tomography (CT) (Figure 1A-b). CT showed portal hypertension and lateral branch circulation formation as well (Figure 1A-c). The development screening test (DST) showed growth retardation of the patient. The majority of biochemical liver test results were abnormal, including low albumin, elevated aspartate aminotransferase, γ-glutamyl transferase, alkaline phosphatase, creatine kinase isoenzyme, total bilirubin, direct bilirubin, and total bile acid. Prolonged activated partial thromboplastin time and prothrombin time, increased international normalized ratio and D-dimer, and decreased fibrinogen were present. Complete blood count demonstrated anemia and thrombocytopenia without immunodeficiency. The patient’s primary laboratory investigations at various stages are summarized in Table 1. Symptomatic treatment was then carried out by albumin and fibrinogen infusion, oral administration of D-galactose, UDCA, and vitamin D/AD/E/K1 supplements. Among them, D-galactose was administered to correct abnormal glycosylation metabolic pathway in purpose. Due to chronic liver failure with life-threatening decompensated cirrhosis and an increased risk of transplantation as his physical condition deteriorated, the patient underwent living-related liver (donated by his mother) transplantation at that time. At subsequent follow-up, ultrasound and CT showed portal anastomotic stenosis (Figure 1A-d). Liver function parameters gradually normalized, and cutis laxa also improved post transplantation. Tacrolimus was given to control the graft versus host reaction. The patient has been followed through 1 year postoperatively and recovered well.

The liver specimen resected from transplantation surgery showed distorted architecture of liver lobules and morphology of pseudo-lobule (Figure 1B-a). High power displayed hepatic cells with finely granular cytoplasm and clear cell membrane, as well as obvious vesicular steatosis in the portal area rather than around the central vein (Figure 1B-b). The portal area was enlarged with infiltration of a lot of lymphocytes (Figure 1B-c). PAS staining identified various sizes of vacuoles in the hepatocytes with steatosis (Figure 1B-d); while staining for PAS with diastase, copper and iron were negative. The liver lobules were divided by hyperplastic fibrous tissue in the portal area, which created pseudo-lobules and bridging-like fibrosis that were identified by Masson staining. Reticulin staining revealed preserved reticular scaffold structure. According to the Ishak staging system (Ishak. et al., 1995), it was diagnosed as stage 6 cirrhosis.

Immunohistochemical staining of the patient’s liver showed a cytoplasmic positive pattern in hepatocytes for ATP6AP2 (Figure 1C-a), as well as LC3A/B (Figure 1C-b) and mTOR (Figure 1C-c). Compared with the healthy control, there were no obvious alterations observed.

WES identified a hemizygous variation c.185G>A in ATP6AP2 exon 3. The c.185G>A was inherited from the healthy mother (Figure 2A), leading to a change of glycine to glutamic acid at the amino acid position 62 (p.Gly62Glu), which was not reported in the HGMD. The amino acid residue of the site mutation was fairly well conserved (Figure 2B), which might be important for the protein’s function. The CADD score of c.185G>A was 27.9, predicted as deleterious. According to the ACMG guidelines, the variant was initially determined as clinically insignificant (uncertain) PM2: the mutation frequency was 0 in the general population database; there were no reports in the literature; comprehensive bioinformatic predictions indicated that it was harmful; and it was verified through family analysis. The Human Genome Variation Society (HGVS; http://www.hgvs.org/varnomen) guidelines served as the foundation for the nomenclature of variants. Wild-type and mutated ATP6AP2 were modeled by PyMol (http://www.pymol.org), which showed that the non-polar aliphatic amino acid glycine was substituted by acidic amino acid glutamic acid with altered polar contact with Arg31 and Trp60 (Figure 2C). The absence of a side chain in the wild-type glycine increases the flexibility of the protein at this site. The variant residue glutamic acid has a negatively charged side chain, making it hydrophilic and resulting in it preferring the surface of the protein to its interior. The glycine to glutamic acid residue change had a high “disease propensity” value of 1.67 (Figure 2D). Figure 2E illustrates the ATP6AP2 protein domains and location of amino acid changes of the reported variants in CDG so far.

In HEK293T cells, qRT-PCR showed mRNA of ATP6AP2 was slightly increased in mutant rather than in wild type exogenous constructs (Figure 3A), but the difference was not statistically significant (p = 0.1000). Western blotting (Figure 3B) revealed that ATP6AP2 protein (Figure 3C) was also increased without significant difference (p = 0.0563). Whereas the expression of FLAG (Figure 3D) tagged in the N-terminal of ATP6AP2 confirmed the elevated level in mutant cells (p < 0.001). The autophagosome marker, LC3A/B (Figure 3E), was upregulated without significant difference (p = 0.5053). And mTOR signaling (Figure 3F) was significantly upregulated (p < 0.05).

In HEK293T cells, a total of 5,225 peaks were detected, of which 943 metabolites were annotated. Compared with the wild-type group, there were 66 differential metabolites, containing 29 ups and 37 downs. As shown by volcano plot, metabolite downregulation was more pronounced (Figure 4A). The difference in metabolism between wild types and mutants is depicted in a heat map as well (Figure 4B). It was observed that the top 10 metabolites with the smallest p-value were significantly lower in ATP6AP2^Gly62Glu than in ATP6AP2^WT: tyr-pro-phe-pro-gly-pro-ile, 7-[(6-Hydroxy-3,7-dimethyl-2,7-octadienyl)oxy]-2H-1-benzopyran-2-one, testosterone, corticosterone acetate, narbomycin, 9(S)-HPOT, oleandomycin, praziquantel, phenylalanyl-leucyl-leucyl-arginyl-asparagine, and antibiotic JI-20B (Figure 4C). In organisms, different metabolites interact to create various pathways. The KEGG database was used to annotate the differential metabolites, and the top 20 entries with the most differential metabolites in the pathway were selected. A summary bar chart and annotation category are illustrated in Figure 4D. The most involved metabolic pathway was steroid hormone biosynthesis, steroid biosynthesis, and arachidonic acid metabolism in lipid metabolism. Metabolism of terpenoids and polyketides as well as the endocrine system were involved. The enrichment netplot shows similar results (Figure 4E).