Compound heterozygous mutations in the gene PIGP are associated with early infantile epileptic encephalopathy

Abstract

There are over 150 known human proteins which are tethered to the cell surface via glycosylphosphatidylinositol (GPI) anchors. These proteins play a variety of important roles in development, and particularly in neurogenesis. Not surprisingly, mutations in the GPI anchor biosynthesis and remodeling pathway cause a number of developmental disorders. This group of conditions has been termed inherited GPI deficiencies (IGDs), a subgroup of congenital disorders of glycosylation; they present with variable phenotypes, often including seizures, hypotonia and intellectual disability. Here, we report two siblings with compound heterozygous variants in the gene phosphatidylinositol glycan anchor biosynthesis, class P (PIGP) (NM_153681.2: c.74T > C;p.Met25Thr and c.456delA;p.Glu153AsnFs*34). PIGP encodes a subunit of the enzyme that catalyzes the first step of GPI anchor biosynthesis. Both children presented with early-onset refractory seizures, hypotonia, and profound global developmental delay, reminiscent of other IGD phenotypes. Functional studies with patient cells showed reduced PIGP mRNA levels, and an associated reduction of GPI-anchored cell surface proteins, which was rescued by exogenous expression of wild-type PIGP. This work associates mutations in the PIGP gene with a novel autosomal recessive IGD, and expands our knowledge of the role of PIG genes in human development.

Introduction

Glycosylphosphatidylinositol (GPI) is a glycolipid utilized by cells to anchor proteins to the cell surface. The addition of this posttranslational modification requires GPI biosynthesis via phosphatidylinositol glycan anchor biosynthesis (PIG) proteins, and remodelling via post GPI-attachment to proteins (PGAP) proteins (1). To date, more than 150 GPI-anchored proteins have been identified, and at least 27 known proteins are involved in the GPI anchor biosynthesis and modification process. GPI-anchored proteins (GPI-APs) have a variety of functions in the cell including, but not limited to, hydrolytic enzymes, adhesion molecules, receptors, protease inhibitors, and complement regulatory proteins (1,2). Given these essential functions, it is not surprising that approximately half of the genes required for GPI anchor biosynthesis and remodelling have already been linked to disease, specifically PIGA (MIM 300868) (3–5), PIGQ (MIM 3008350) (6), PIGC (MIM 601730) (7), PIGY (MIM 616809) (8), DPM2 (MIM 603564) (9), PIGL (MIM 280000) (10,11), PIGW (MIM 616025) (12), PIGM (MIM 610293) (13), PIGV (MIM 239300) (14–17), PIGN (MIM 614080) (18–23), PIGO (MIM 614749) (17,24–26), PIGG (MIM 616917) (27), PIGT (MIM 615398) (28–30), PGAP1 (MIM 611655) (31–33), PGAP3 (MIM 615716) (34,35), and PGAP2 (MIM 614207) (36–38) (Supplementary Material, Table S1). This group of disorders has been termed ‘inherited GPI deficiencies’ (IGDs) (1), and falls under the broader group of congenital disorders of glycosylation. Interestingly, all of these disorders are autosomal recessive or X-linked recessive (PIGA). IGD associated phenotypes often include seizures, intellectual disability, coarse facial features and hypotonia, but can also include microcephaly, hearing impairment, joint contractures, skin anomalies, congenital heart defects, urinary tract defects, skeletal anomalies and others (3–38).

Here we report a non-consanguineous family with a novel autosomal recessive disorder in two siblings characterized by early-onset refractory seizures, hypotonia, and profound global developmental delay. Whole exome sequencing identified compound heterozygous mutations in the gene phosphatidylinositol glycan anchor biosynthesis, class P (PIGP). Experiments in patient cells showed that these mutations result in defective PIGP function. Our findings add PIGP to the expanding list of GPI biosynthesis genes that when mutated cause rare genetic diseases.

Results

Patient descriptions

Two children with refractory epilepsy (Fig. 1A–C) were born to non-consanguineous parents of French-Irish ancestry; they have one healthy sister.

Figure 1

Clinical features of two children with compound heterozygous mutations in PIGP. (A) Photographs of affected male child at 6 months and 9 years of age and (B) affected female child at 15 months of age showing sibling resemblance and hypotonic long facies. (C) MRI of male child at 5 years of age showing thin corpus callosum. (D) Mutations in PIGP localized on isoforms 1 and 2, which differ by 24 amino acids at the beginning of the first exon. The first mutation [Met25Thr (Isoform 1) or Met1Thr (Isoform 2)] leads to a missense or a loss of the start codon in each isoform, respectively. The second mutation, a deletion, causes a frameshift leading to a 34aa extension of the C-terminus. Inset shows the locations of the transmembrane domains (TM1 and TM2). Included are the location of the forward (F primer) and reverse (R primer) primers used for real-time PCR on cDNA.

The male proband (Fig. 1A and C) was born to a healthy G1P0 mother after an uncomplicated pregnancy. He was born after 41 weeks and 4 days, and weighed 4360 g (90–97^th percentile). APGAR scores were 9 at both 1 and 5 minutes. He was admitted to the NICU after birth due to cyanosis and dyspnea associated with feeding. At 14 days of life, he started having seizures that comprised right-sided facial twitching. There were no structural lesions identified on MRI at the time. EEG showed abnormal frequent sharp waves from the bitemporal area with electrographic seizures from the right central region spreading to right temporal region areas.

Physical examination at 6 months of age showed his weight was 8.48 kg (50^th percentile), height was 71.2 cm (90^th percentile), and head circumference was 44.5 cm (75^th percentile). By 3 years, 10 months of age, he weighed 12.15 kg (3^rd percentile), was 98.2 cm in length (10–25^th percentile), and his head circumference was 47.3 cm (less than 2^nd percentile). He was noted to have cortical visual impairment. He had significant right-sided plagiocephaly (Fig. 1A) and held his lower limbs in abduction with contractures at the knees.

Repeat MRIs performed at 1 and 5 years of age showed a thin corpus callosum (Fig. 1C), with progressively increased T2 signal in the periventricular and subcortical white matter. His seizures persisted and were difficult to control. He never experienced infantile spasms. For example, at age 7 years, he was experiencing 12–25 seizures per day, each lasting 2–3 minutes. He has been trialed on phenobarbital, clonazepam, topiramate, vigabatrin, and levetiracetam in various doses and combinations. Since the introduction of levetiracetam at 8 years of age, in combination with clonazepam and phenobarbital, he has not had any reported seizures. An EEG at 9 years of age showed frequent multifocal epileptiform discharges and slow background activity for his age.

At his current age of 10 years, he has a profound intellectual disability with little purposeful movements, no head control, and no vocalizations. He has profound central hypotonia with peripheral hypertonia, and hyperlaxity of the small and larger joints. He receives his nutrition by G-tube since age 8 years to manage recurrent aspirations but is experiencing progressive feeding intolerance.

The proband’s younger sister (Fig. 1B) was born at 37 weeks gestation and weighed 3997 g (90–97^th percentile). She started having complex partial seizures at 7 weeks of age, each seizure lasting several minutes and occurring 1–2 times daily. Seizures comprised eye twitching and apnea for 30–40 seconds. EEG at that time showed a poorly organized background with multifocal epileptiform discharges, and she was trialed on phenobarbital and topiramate. A brain MRI performed at 1.5 months of age found no structural abnormality. At 3 months of age, she began having infantile spasms and EEG showed modified hypsarrhythmia. She was started on vigabatrin which resulted in cessation of clinical spasms, but persistent modified hypsarrhythmia on EEG. Since then she has been trialed on gabapentin, valproic acid, levetiracetam, pyridoxine, pyridoxal-5’-phosphate, folinic acid, and prednisolone, with no change in her modified hypsarrhythmia. Her seizures remained poorly controlled and at 17 months she was having 1–2 seizures a day. At approximately 2 years of age, she began having episodes of sustained eye deviation lasting hours which were non-epileptic and assumed to be episodes of oculogyric crises.

At 6 months, she weighed 6.76 kg (27^th percentile), was 66.1 cm in length (57^th percentile), and had a head circumference of 42 cm (45^th percentile); at 14 months her head circumference was essentially unchanged at 45.3 cm (46^th percentile). A G-tube was placed at 19 months to manage recurrent aspiration. She had no eye contract or tracking, and was diagnosed with a cortical visual impairment. She had a low central muscle tone, with fisted hands and absent postural reflexes. She had positive clonus bilaterally, and an exam at 23 months noted 5–8 beats of clonus, as well as hyperreflexia (4+). She experienced progressive feeding intolerance and passed away at 26 months of age secondary to complications of her seizure disorder.

Investigations to identify the cause of this severe seizure disorder in this family included transferrin isoelectric focusing and enzyme testing (β-glucocerebrosidase, hexosaminidase A&B, β-galactosidase and arylsulfatase A), which were normal in the brother. Genetic tests for mutations in ARX, STXBP1, MECP2 in the brother were normal, as was a microarray for chromosomal copy number variations. Plasma amino acids and urine organic acids were normal in both children. CSF neurotransmitter analysis in the sister revealed low homovanilic acid at 205 nmol/L (normal: 337–1299 nmol/L) and low 5-hydroxyindoleacetic acid at 195 nmol/L (normal: 208–1159 nmol/L). CSF neopterin and biopterin were within normal limits. Liver function tests and alkaline phosphatase (ALP) were normal in both children.

Due to a lack of a molecular diagnosis for these siblings, the family was offered participation in the Care4Rare Canada research study to try to identify the cause of their rare disease using whole-exome sequencing.

Whole-exome sequencing identified compound heterozygous mutations in PIGP

To investigate the genetic cause of this clinical presentation, we performed whole-exome sequencing on genomic DNA from both affected children and their mother. Average coverage for the exomes was 144.68X for the male child, 121.59X for the female child, and 129.47X for their mother, with at least 96.7% of CCDS exons in each exome covered at 10X. Common variants [≥1% minor allele frequency in the 1000 Genomes project phase 1 data set (April 2012 release) (39), the Exome Variant Server (http://evs.gs.washington.edu/EVS/; date last accessed December 2016), the ExAC server (40) or ∼2000 internal exomes] were excluded. An autosomal recessive mode of inheritance was suspected given the concordant phenotype in the siblings and lack of related family history. There were no homozygous rare variants shared by both affected children. After filtering for shared compound heterozygous mutations which were inherited and in trans, we identified a single gene, PIGP; both affected children carried maternally inherited NM_153681.2:c.74T > C (p.Met25Thr) and a second variant in trans, NM_153681.2c.456delA (p.Glu153Asnfs*34). Sanger sequencing confirmed the variants were present in both children, one inherited from each parent. The variants were submitted to ClinVar.

The c.74T > C (p.Met25Thr) PIGP variant occurs at a highly conserved residue (GERP 5.4, (41)) and was predicted to be deleterious by multiple in silico programs [CADD (42), Polyphen2 (43), SIFT (44), Mutation Taster (45), and Align GVGD (46,47)]. This variant has been reported in the ExAC database (40) with very low frequency (2.47 × 10⁻⁵). In the longest PIGP isoform 1 (NM_153681.2), this mutation causes a methionine to threonine missense change, though in isoform 2 (NM_153682.2), this variant leads to a loss of the start codon (p.Met1Thr) (Fig. 1D). The second variant [NM_153681.2:c.456DelA (p.Glu153Asnfs*34)] occurs in the final exon of all transcripts (Fig. 1D). Notably, only a handful of loss-of-function variants in PIGP have been observed in ExAC, all of which are considered rare and never seen in a homozygous state. Interestingly, mutations in other genes whose proteins are in complex with PIGP (PIGA, PIGC PIGQ, PIGY and DPM2) have been identified as causative for similar phenotypes (3–9). Taken together, these findings indicate that the compound heterozygous variants in PIGP are likely the cause of this novel autosomal recessive condition and prompted us to further investigate the functional impact of these variants.

PIGP transcript levels are decreased in patient cells

We began by evaluating whether the mutations impacted PIGP mRNA levels by performing real-time PCR analysis on fibroblasts from the affected male child and age-matched controls (fibroblasts from the affected female child were not available). We found that fibroblasts from the affected child had reduced levels of PIGP mRNA (Fig. 2A). Western blot analysis of PIGP was attempted with four different antibodies; however, no antibodies specifically detected PIGP (data not shown). We conclude that patient cells have decreased levels of PIGP mRNA, and potentially protein.

Figure 2

Impact of the PIGP mutations on patient cells. (A) Real-time PCR on fibroblast cell extracts, showing that the affected male has reduced transcript levels of PIGP. Error bars represent standard error of the mean. GAPDH was used for normalization. (B) Flow cytometry analysis of granulocytes from fixed fresh blood of the affected male and control. Results show cell surface expression of fluorescently labeled proaerolysin ‘FLAER’, which binds directly to the GPI anchor, as well at expression of the GPI-APs CD16, and CD55 from triplicate experiments on fixed cells. Numbers above each histogram represent the mean fluorescence intensity (MFI).

Patient and PIGP deficient HAP1 cells showed reduced cell surface expression of GPI-anchored proteins, but artificially overexpressing wild type PIGP cDNA rescued this phenotype

PIGP is an integral component of the first reaction which commences GPI anchor biosynthesis, therefore, we next assessed whether the GPI-anchoring process was deficient in patient cells. To determine if patient cells had reduced cell surface expression of GPI-anchored proteins (GPI-APs), we stained fixed whole blood of patients and controls with fluorescent antibodies for GPI-APs (CD16, CD55, CD59), as well as with fluorescein-labeled proaerolysin ‘FLAER’ which binds to the GPI anchor itself, and performed flow cytometry to assess relative fluorescence (26). Analysis on granulocytes indicated that patient cells had reduced signal of CD16 (25% vs. controls) and FLAER (51% vs. controls) (Fig. 2B). We also analysed lymphocytes and repeated the experiments in live granulocytes from unfixed whole blood (Supplementary Material, Fig. S1). We conclude that insufficient levels of PIGP and/or defective PIGP function in patient cells leads to reduced levels of GPI-anchored proteins at the cell surface.

To investigate the effects of each mutation on isoforms 1 and 2 independently, we performed the functional analysis of the mutant cDNAs using PIGP deficient cells made by exon trapping mutagenesis (48). We transiently transfected PIGP deficient HAP1 cells with either mutant or wild-type PIGP cDNA of two isoforms and analyzed the surface expression of GPI anchored proteins by flow cytometry. We found that the cells transfected with PIGP-Met25Thr isoform 1 (NM_153681.2) cDNA had reduced expression of CD55, CD59, and CD87 compared to cells transfected with wild-type cDNA (Fig. 3A), and that this difference was even more pronounced with PIGP-Met1Thr isoform 2 (NM_153682.2) cDNA (Fig. 3B), but there seemed to be residual activity despite the loss of the start codon. Cells transfected with PIGP-Glu129AsnFs*34 cDNA from isoform 2 had slightly reduced levels of GPI-anchored proteins (Fig. 3C) as assessed by strong promoter driven constructs, and the reduction was more clear when driven by weaker promoter driven constructs, indicating that the frameshift mutation in the very C-terminal location also affected the PIGP activity (Fig. 3D). Western blot analysis using HA antibody on lysates from HEK293 cells transfected with mutant or wild-type HA-tagged PIGP cDNA showed that wild-type isoform 2 (Fig. 3E, lane 3) was found to be more highly expressed than wild-type isoform 1 (Fig. 3E, lane 1). However, the faint band detected in the wild type of isoform 1 was approximately the same band size as isoform 2, so it is possible that it was translated from the same methionine. Activity of wild type C-terminal tagged isoform 1 was lower than C-terminal tagged isoform 2 (Supplementary Material, Fig. S2), but neither the expression of PIGP-Met25Thr (isoform 1) nor PIGP-Met1Thr (isoform 2) produced a detectable amount of protein (Fig. 3E, lanes 2 and 4). The expression of PIGP- p.Glu129Asnfs*34 (NM_153682.2) in isoform 2 led to the expected increase in molecular weight of the protein (Fig. 3E, lane 6; wild-type in lane 5). These findings conclusively demonstrate that both patient mutations compromise PIGP function.

Figure 3

Flow cytometry analysis of PIGP deficient HAP1 cells transfected with either wild type PIGP cDNA or mutant PIGP cDNA, analysed as the expression of the GPI-APs CD55, CD59, and CD87. (A) The effect of c.T74C on the longest isoform (NM_153681.2), causing the variant p.Meth25Thr; (B) the effect of c.T2C on the dominant isoform (NM_153682.2) causing a loss of the start codon. The effect of the variant c.384delA on the dominant isoform (NM_153682.2), causing a frameshift and stop loss was analyzed using the strong promoter pME (C) as well as the weak promoter pTK (D). The effect of the first variant caused a stronger reduction of GPI-AP expression in isoform 2, however there was some residual activity despite the loss of the start codon. The effect of the second variant was stronger under the weak promoter (D) than the strong promoter (C). (E) Western blot on lysates from HEK293 cells transfected with various PIGP cDNAs or empty vector. Lanes 1 and 2 are isoform 1 (NM_153681.2) transfected with wild-type and Met25Thr PIGP cDNA, respectively. Lanes 3 and 4 are isoform 2 (NM_153682.2) transfected with wild-type and Met1Thr PIGP cDNA, respectively. Lanes 5 and 6 are isoform 2 (NM_153682.2) transfected with wild type and cDNA causing the frameshift mutation Glu129AsnFs*34, respectively. Lane 7 is empty vector. GAPDH (lower panel) was used as loading control. Lanes 1–4 had a C-terminus HA tag, whereas lanes 5 and 6 had an N-terminus HA tag, and HA antibody was used to identify PIGP, which is indicated by the arrows. There was non-specific background in all lanes at ∼27kDa, the two smaller bands in lanes 5 and 6 are unknown. Isoform 2 (lane 3) was more strongly expressed than isoform 1 (lane 1). Both the Met25Thr and Met1Thr variants (lanes 2 and 4) decreased the expression of their respective isoforms. The Glu129AsnFs*34 mutation in isoform 2 also decreased PIGP expression but lead to an increase in protein size (lane 6), as expected. N-terminal HA tagged PIGP (lane 5) showed higher expression than C-terminal HA tagged PIGP (lane 3) of the same isoform.

Finally, we wanted to see if we could rescue the deficiency in GPI-APs in patient cells. We transduced patient fibroblasts with a lentivirus overexpressing wild type PIGP isoform 2 cDNA (NM_153682.2) and repeated the flow cytometry assays for quantification of GPI-APs at the cell surface, comparing results with and without lentiviral rescue, and expressed relative to control fibroblasts. This resulted in increased detection of CD87, from 47 to 63% relative to control fibroblasts, and FLAER increased from 64% to 98% (Fig. 4).

Figure 4

Flow cytometry analysis of fluorescently labeled proaerolysin ‘FLAER’ (A), CD73 (B), and CD87 (C). Fibroblasts from the affected male proband were compared with unaffected controls. Furthermore, patient cells were transduced with lentivirus overexpressing wild type PIGP cDNA (NM_153682.2), and were analyzed in parallel. The lentivirus restored FLAER on patient cell surface from 64% to 98% (A), and CD87 from 47% to 63% (C). Numbers above each histogram represent the mean fluorescence intensity (MFI).

Taken together, our results show that patient cells have decreased levels of PIGP mRNA, and that expression of mutant PIGP cDNA in HEK293 cells produced a decreased amount of protein. Furthermore, both mutations have a deleterious effect on PIGP function, leading to decreased levels of GPI-APs at the cell surface. Taken together, we conclude that biallelic mutations in PIGP are the cause of this novel epilepsy syndrome.

Discussion

We have identified a novel autosomal recessive disorder characterized by early infantile epileptic encephalopathy, hypotonia, and profound global developmental delay. Both affected siblings carried compound heterozygous mutations in the gene PIGP. Molecular profiling of patient cells and an in vitro recapitulation system showed that these PIGP mutations adversely affect protein function, providing substantial support for causality of this syndrome. To our knowledge, this is the first report implicating PIGP in human disease.

The IGDs are an increasingly recognised group of disorders. To date, all of these conditions are associated with seizures and developmental delay, which is consistent with the affected individuals reported here. Other findings, such as hypotonia, microcephaly and CNS abnormalities are also common, demonstrating the importance of this pathway for proper neurodevelopment. From IGDs characterized thus far, it appears that mutations in earlier steps of the pathway lead to degradation of precursor proteins which fail to become properly GPI-anchored, and this occurs through ER-associated degradation (31), while mutations in later steps can leave residual secretion of improperly GPI-anchored proteins such as ALP, which is released into the blood. The siblings reported here did not have elevated ALP levels. As PIGP is a member of the early pathway, this provides further support for this dichotomy. It also is important to note that the stability of different GPI-APs is variable (49), and different cell types will show variable expression patterns of GPI-APs (50). For these reasons, it is important to use several different GPI-APs as biomarkers, and to test multiple cell types, with emphasis on granulocytes (50). Overall, this novel condition bears remarkable resemblance to other IGDs.

Seizure control was difficult to achieve in these patients, though levetiracetam was effective in the male child. The levetiracetam responsiveness is interesting, as excessive levels of PIGP transcripts in brain tissue have been associated with levetiracetam resistance (51). This suggests that appropriate transcript or protein levels may play a role in the response to anticonvulsants. Levetiracetam binds to the synaptic vesicle protein SV2A, yet its exact mechanism(s) of action remain unknown, and there are no identified links to the GPI anchor. Further studies are needed to further investigate a link between PIGP and response to this medication.

Little is known about how dysfunction of PIG genes leads to the common IGD phenotypes. Given the vast number of GPI-APs, this question is incredibly complicated but some insights exist. For example, homozygous or compound heterozygous mutations in FOLR1, a GPI-AP, have been associated with seizures refractory to antiepileptic therapy, as well as intellectual deficiencies, and severe physical handicap (52). It is therefore logical to predict that compromising the FOLRI GPI anchor would also have similar effects. Another example is CNTN2, a GPI-AP involved in the organization of axonal subdomains at the nodes of Ranvier (53). Mutations in CNTN2 have been associated with cortical myoclonic tremor and epilepsy (54). These are merely 2 of the 150+ GPI-APs which are likely disrupted in our patients and contribute to their clinical presentation. Broadening our knowledge of this pathway and the downstream effects of PIG and PIGP mutations is necessary to begin to elucidate the molecular pathogenesis of these disorders and investigate targeted treatment options.

In summary, we report a novel autosomal recessive disorder in a sibling pair affected by early-onset refractory seizures, hypotonia, and profound global developmental delay. Whole exome sequencing identified novel compound heterozygous mutations in the gene PIGP. Molecular findings showed that patient-derived cells have reduced levels of PIGP mRNA transcript, and expressing cDNA recapitulating the patients’ mutations in vitro resulted in a reduced detection of PIGP protein compared with wild type cDNA. These mutations caused reduced expression of GPI-APs, which was rescued by providing wild type PIGP cDNA. These findings broaden our understanding of the disease pathogenesis of mutations occurring in the GPI anchor biosynthesis and remodeling pathway.

Materials and Methods

Patients

A family with two children presenting with neonatal onset seizures and hypotonia were evaluated by the genetics service at the Children’s Hospital of Eastern Ontario. The parents gave informed consent to participate in the Care4Rare Canada research study, which was approved by the Children’s Hospital of Eastern Ontario Research Ethics Board. In addition to clinical evaluation, blood and tissue samples were collected for DNA extraction and the establishment of cell lines.

Whole-exome sequencing, and validation of identified variants

DNA from both affected siblings and the mother was extracted from whole blood. The Agilent SureSelect Clinical Research Exome kit was used to select for exonic DNA, and sequencing was performed on an Illumina HiSeq 2000. Alignment, variant calling, scoring and annotation were performed as described in previous FORGE and Care4Rare projects (55). We compared variants to the 1000 genome phase 1 data set (April 2012 release) (39), the Exome Variant Server (http://evs.gs.washington.edu/EVS/), the ExAC server (40), and our in-house database (∼2000 exomes previously sequenced at the McGill University and Génome Québec Innovation Centre). Variants were excluded if they were found at greater than 1% frequency in any database. Variants identified by whole-exome sequencing were validated by PCR amplification and sequenced using bi-directional Sanger sequencing. DNA from both parents was also Sanger sequenced. Primers: Met25Thr Fwd 5’ CACCCCTTCTGTTGCGG 3’ and Rev 5’ CTTCCCTTGTCACTGAACGC 3’; Glu153AsnFs*34s Fwd 5’TGG GT CT CC ATCAGACACAG 3’ and Rev 5’ AAATG CCTC TCTGG AGG AA C 3’. Due to allelic dropout of the Met25Thr variant in the mother, the primers Fwd 5’ CCAGCCTGGGTGTCTGTATG 3’ and Rev 5’ AATCGCTCTTTCTGGCAATG 3’ were used for Sanger sequencing her DNA.

Cell line establishment

A skin biopsy was taken from the affected male child to establish a fibroblast cell line. The cell line was established at the Centre for Applied Genomics (Toronto, Canada). Fibroblasts were maintained in HyClone DMEM media (GE Healthcare Life Sciences) supplemented with 10% fetal bovine serum, Penicillin-Streptomycin (SV30010, GE Healthcare Life Sciences), and 2mM L-glutamine (SH3003401, Thermo Scientific). Age-matched control cell lines were cultured in parallel.

Real-time PCR

Total RNA was extracted from patient and control cell lines using the RNeasy Mini Kit (QIAGEN), and cDNA was synthesized using the iScript kit (BioRad Laboratories) per the manufacturer protocols. Control reactions were run in parallel without reverse transcriptase. We amplified cDNA using gene-specific primers and iQ SYBR Green Supermix using the following conditions: an initial denaturing steps of 95 °C for 3 minutes, followed by 39 cycles of 95 °C for 10s, 55 °C for 20s, 72 °C for 30S, and a final melting curve generated in increments of 0.5 °C per plate read, using a CFX96 Touch Real-time PCR Detection System (BioRad Laboratories). We quantified gene expression using the Ct method with the CFX Manager software (BioRad Laboratories) and all data were corrected against GAPDH as an internal control. Primers: GAPDH Fwd 5’ TGCACCACCAACTGCTTAGC 3’ and Rev 5’ GGCATGGACTGTGGTCATGAG 3’; PIGP Fwd 5’ TTACCT CGTGTGGGCCTTTA 3’ and Rev 5’ ATGGATGGAGTCGAGTGGAG 3’. Experiments were repeated with 2 technical replicates.

Analysis by flow cytometry

Fresh blood samples from the affected male child and healthy controls were fixed with 10% formaldehyde, red blood cells were lysed in 0.1% Triton X-100, then the samples were stained with the GPI-AP markers: PE-anti human CD16 (BioLegend), FITC-mouse anti human CD55 and CD59 (BD Pharmingen), or FLAER-Alexa 448 Proaerolysin (Cedarlane) for 1 hour at room temperature. Non-specific binding was washed off before analyzing by a BD FACS Canto system (BD Biosciences). Granulocytes and lymphocytes were selected for by size, and flow cytometry results were analyzed using FlowJo software. For analysis of unfixed granulocytes, the samples were stained for 20 minutes on ice and red blood cells were lysed in FACS Lysing Solution (BD Bioscience).

Functional analysis using PIGP deficient HAP1 cells

PIGP deficient HAP1 cells were generated by exon trapping mutagenesis (a gift from Dr. Morihisa Fujita, Jiangnan University, China) (48). We then cloned two hPIGP isoforms containing the 5’UTR using a Hep3B cDNA library, and ligated to generate C-terminal tagged wild-type and mutant pME hPIGP HA, generated by site-directed mutagenesis in isoforms 1 (NM_153681.2) and 2 (NM_153682.2). We also cloned hPIGP isoform 2 containing the 3’UTR from a Hep3B cDNA library and ligated to generate N-terminal tagged wild-type and mutant pME HA-hPIGP, strong promoter driven constructs, and pTK HA-hPIGP, weaker promoter driven constructs. The HAP1 cells were then transiently transfected with the various PIGP cDNAs (mutants or wild type) containing vectors, or an empty vector, by electroporation. Transfection efficiency was monitored by luciferase assay, and flow cytometry analysis was performed 2 days after transfection. For the analysis of protein expression, HEK293 cells were transiently transfected with various PIGP cDNAs and PIGP protein expression was analyzed by Western Blotting using an anti-HA antibody (Cell Signaling). The intensities of the bands were normalized with the loading control (GAPDH), and luciferase activities used for evaluating transfection efficiencies.

Phenotype rescue in patient fibroblasts

Patient fibroblasts were transduced with lentivirus expressing wild-type human PIGP cDNA NM_153682.2 (pReceiver-Lv105, Genecopoeia) and cultured in medium supplemented with puromycin as selection marker. The lentivirus-infected cells as well as untreated patient and control fibroblasts were then subjected to flow cytometry analysis using the markers FLAER-Alexa 448 (Cedarlane), FITC-anti human CD73, and PE-anti human CD87 (BioLegend).

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

First and foremost, the authors would like to thank the study participants and their family. We would also like to recognise the contributions of the high throughput sequencing platform of the McGill University and Génome Québec Innovation Centre, Montréal, Canada, those of Care4Rare project coordinator Chandree Beaulieu, as well as the help of Meredith Gillespie. We also thank Arran McBride, Kana Miyanagi, Saori Umeshita, Shunsuke Mori, and Naoki Morimoto for their technical assistance. This work was selected for study by the Care4Rare Consortium Gene Discovery Steering Committee: Kym Boycott (lead; University of Ottawa), Alex MacKenzie (co-lead; University of Ottawa), Jacek Majewski (McGill University), Michael Brudno (University of Toronto), Dennis Bulman (University of Ottawa), and David Dyment (University of Ottawa). D.L.J. is supported by a Vanier Canada Graduate Scholarship. M.T. is supported by a Canadian Institutes of Health Research post-doctoral fellowship. Y.M. was supported by the Ministry of Health Labour and Welfare.

Conflict of Interest statement. None declared.

Funding

This work was funded by Care4Rare Canada Consortium (Enhanced Care for Rare Genetic Diseases in Canada), which is funded by Genome Canada, the Canadian Institutes of Health Research, the Ontario Genomics Institute, Ontario Research fund, Génome Québec and the Children’s Hospital of Eastern Ontario Foundation. This work was also funded by the Japanese Agency for Medical Research and Development, AMED, and the Ministry of Health Labour and Welfare. P.M.C. is supported by Clinician Scientist funding from the Canadian Institutes of Health Research and the Fonds de Recherche du Québec – Santé.

Author notes

†

T.T.M.N and Y.M contributed equally as second authors.

‡

K.M.B and P.M.C contributed equally as senior authors.