A hypomorphic PIGA gene mutation causes severe defects in neuron development and susceptibility to complement-mediated toxicity in a human iPSC model

Mutations in genes involved in glycosylphosphatidylinositol (GPI) anchor biosynthesis underlie a group of congenital syndromes characterized by severe neurodevelopmental defects. GPI anchored proteins have diverse roles in cell adhesion, signaling, metabolism and complement regulation. Over 30 enzymes are required for GPI anchor biosynthesis and PIGA is involved in the first step of this process. A hypomorphic mutation in the X-linked PIGA gene (c.1234C>T) causes multiple congenital anomalies hypotonia seizure syndrome 2 (MCAHS2), indicating that even partial reduction of GPI anchored proteins dramatically impairs central nervous system development, but the mechanism is unclear. Here, we established a human induced pluripotent stem cell (hiPSC) model containing the PIGAc.1234C>T mutation to study the effects of a hypomorphic allele of PIGA on neuronal development. Neuronal differentiation from neural progenitor cells generated by EB formation in PIGAc.1234C>T is significantly impaired with decreased proliferation, aberrant synapse formation and abnormal membrane depolarization. The results provide direct evidence for a critical role of GPI anchor proteins in early neurodevelopment. Furthermore, neural progenitors derived from PIGAc.1234C>T hiPSCs demonstrate increased susceptibility to complement-mediated cytotoxicity, suggesting that defective complement regulation may contribute to neurodevelopmental disorders.


Introduction
In 2006, the first germline mutation in a gene (PIGM) involved in GPI anchor biosynthesis was described in a child with portal and hepatic vein thrombosis and absence seizures [1]. Since then, dozens of pedigrees describing mutations in genes involved in GPI anchor PLOS  biosynthesis, including the X-linked PIGA gene, have been described [2][3][4][5][6][7][8][9]. While the phenotypes vary, these affected individuals share a number of overlapping features. All mutations are hypomorphs, consistent with the model that complete absence of GPI anchor proteins is embryonic lethal. Overlapping clinical manifestations include early onset infantile spasms, profound developmental delay and intellectual disability, dysmorphic facial features, and multiple central nervous system abnormalities, such as thin corpus callosum, delayed myelination, and hypotonia. Severely affected patients die within the first year of life; others may survive into adulthood, but have significant intellectual disability and seizures. Phosphatidylinositol glycan class A protein (PIGA) is one of over 30 enzymes involved in the biosynthesis of glycosylphosphatidylinositol (GPI), a glycolipid moiety that anchors more than 100 different proteins to the cell surface [10,11]. PIGA is one of seven enzymes essential for the first step in GPI anchor biosynthesis [12]. GPI anchored proteins serve critical functions as adhesion molecules, receptors, complement regulators, enzymes and co-receptors in signal transduction pathways. The PIGA gene is located on chromosome Xp22.2, spans 162 kb and encodes for a widely expressed 484 amino acid protein. The remaining genes involved in GPI anchor biosynthesis are located on autosomes. Until the last decade, only somatic PIGA mutations had been reported in patients with paroxysmal nocturnal hemoglobinuria (PNH) [13,14]; germline mutations had not been reported in PIGA or any other of the genes involved in GPI anchor biosynthesis and were suspected to result in embryonic lethality [15,16].
PNH is a rare hematologic condition that leads to a severe complement-mediated hemolytic anemia [14,17]. The disease develops when a hematopoietic stem cell acquires a PIGA mutation that leads to severe GPI anchor protein deficiency. Following clonal expansion of the PIGA mutant stem cell, PNH patients develop signs and symptoms that correlate with the percentage of GPI anchor deficient blood cells [18]. Hemolysis in PNH is caused by a severe deficiency of two GPI anchored complement regulatory proteins, CD55 and CD59, and the hemolytic anemia can be abrogated by a humanized monoclonal antibody to C5 that blocks terminal complement [19,20]. Thrombosis is the leading cause of death in PNH and also correlates with the size of the PNH clone.
Germline PIGA null mutations are embryonic lethal due to an early block in embryogenesis, before the development of mesoderm and endoderm, that is due to loss of GPI anchored co-receptors involved in BMP4 signaling [16,21]. In 2012, we described the first pedigree of a family with multiple congenital anomalies hypotonia seizure syndrome 2 (MIM316818, MCAHS2) due to a hypomorphic germline PIGA mutation (c.1234C>T) [4]. Neither patient had hemolytic anemia or clinical hemoglobinuria. The findings indicated that even subtle GPI anchor protein deficiency results in severe defects in neuronal development. Since there are limited numbers of GPI anchored proteins involved in neuron development, these rare germline mutations may offer insight into the role that specific GPI anchored proteins play in inherited and acquired neurodevelopmental and neurodegenerative diseases. Since our original report, a number other patients with inherited GPI anchor deficiency and heterogeneous neurodevelopmental congenital anomaly disorders due to hypomorphic PIGA mutations have been described [5,6,[22][23][24][25][26].
Recently, we established a human induced pluripotent stem cell (hiPSC) model of PIGA loss of function using genomic editing to abolish function of the PIGA gene [16]. Differentiation of these PIGAnull hiPSCs resulted in small embryoid bodies that did not produce bloodlike cells; however, using a doxycycline inducible promoter to regulate PIGA expression we were able to establish GPI anchor deficient blood cells by expressing the PIGA gene product early in the differentiation protocol. These data, in conjunction with clinical phenotype of inherited GPI anchor deficiency syndromes, suggest that mutations that lead to reduced GPI anchor protein expression have little to no impact on hematopoiesis. However, they can produce severe defects in neuronal development and predispose to intellectual disability and seizures. In order to study the effects of partial deficiency of PIGA during neuron development, we established hiPSCs containing the hypomorphic PIGAc.1234C>T mutation previously described [4]. These cells have reduced expression of cell surface GPI anchored proteins and normal hematopoietic differentiation; however, they have impaired neuron differentiation, characterized by abnormal proliferation, synapse deficits, and increased susceptibility to the alternative pathway of complement.

Generation of the PIGAc.1234C>T mutation in hiPSC lines
An hiPSC line derived from a healthy male was used to generate hiPSC lines with mutations in the PIGA gene. A complete knock out of PIGA was generated in hiPSCs using zinc finger nuclease (ZFN) technology as described [16]. PIGA gene deficiency was confirmed by lack of CD59 expression. The nonsense point mutation PIGAc.1234C>T was introduced into the PIGAnull hiPSC line using the piggyBac (PB) Transposon System [4,5]. This system was also used to introduce a wildtype PIGA cDNA (PIGAwt) and a truncated PIGA cDNA (PIGAtr411) into the PB backbone vectors for generation of hiPSC lines. PIGAtr411 was used as a negative vector control. All vectors were verified by DNA sequencing. PIGAnull hiPSCs were transfected at 80% confluency with PB-PIGAwt, PB-PIGAc.1234C>T or PB-PIGAtr411 along with PB-Transposase using the 4D Nucleofector system (Lonza). Seventy-two hours post-transfection, CD59 expression was confirmed by flow cytometry (A15705, Molecular Probes) compared to PIGAtr411 as background. The remainder of transfected cells were transferred to a new coated plate and cultured in hESC medium containing 0.5-1.0 μg puromycin for selection. After 2 weeks of antibiotic selection, transfected hiPSCs were harvested and stained with anti-CD59 antibody. CD59 positive cells from PIGAwt and PIGAc.1234C>T hiPSCs were isolated using fluorescence-activated cell sorting and expanded in culture for further study. Using the same method as above, we also introduced PB-PIGAwt, PIGAc.1234C>T, and PIGATr411 into TF1PIGAnull cells.

Mesoderm induction and hematopoietic differentiation
Embryoid bodies (EBs) were generated from the three hiPSC lines via forced aggregation as previously described [16]. The hiPSC lines were enzymatically treated and plated in 96-well plates at a density of 3,000-5,000 cells per well in serum free medium (SFM) supplemented with 10 ng/mL FGF-basic, 10 ng/mL BMP-4, 10 ng/mL VEGF, and 50 ng/mL SCF (R&D System). The cells were aggregated by centrifugation at 3,000rpm for 5 minutes and placed in a humidified incubator at 37˚C with 5% CO 2 . The medium was replaced with fresh SFM containing cytokines every 3 days. Ten days after mesoderm induction, blood-like cells (BLCs) surrounding the EBs were observed. Fourteen days after mesoderm induction, single cells from the three hiPSC lines were collected by filtration through a 70 μm filter. Cell phenotype was analyzed by flow cytometry using CD34 and CD45 as markers of hematopoietic stem cells and progenitor cells. Hematopoietic cells derived from the hiPSC lines were further cultured in suspension in SFM supplemented with 10 ng/mL Flt3, 10 ng/mL IL-3, 10 ng/mL TPO, 50 mg/mL, GM-CSF, and 50 ng/mL SCF plus 3 units/mL EPO (R & D System). After 5 days of hematopoietic differentiation, expression of CD34 (555824, BD Biosciences), CD45 (555483, BD Biosciences), CD15 (11-0159-42, eBioscience), CD33 (555450, BD Biosciences), and CD235a (559943, BD Biosciences) was analyzed by flow cytometry.
Hematopoietic colony-forming unit (CFU) assay Fourteen days after mesoderm induction, single cells differentiated from PIGAwt and PIGAc. 1234C>T hiPSC lines were collected, counted and plated in 3 mL MethoCult TM H4434 classic medium (Stem Cell Technologies) at a density of 50,000 cells per 36 mm ultra-low dish in duplicate, supplemented with 3 units/mL EPO. The dishes were kept in a humidified incubator at 37˚C with 5% CO 2 for 10 to 14 days. The CFUs were counted and characterized under an inverted microscope.

NSC induction and neuronal differentiation
Prior to EB formation, the three hiPSC lines were cultured in a 6-well plate coated with Matrigel (BD Biosciences) in NutriStem TM XF/FF medium (Stemgent). Day 0-4: When 80% confluent, hiPSCs were disassociated with Accutase (Sigma-Aldrich), counted and seeded in Ushaped 96-well plates at a density of 4,000 cells per well in 50μl of hESC medium containing 20 μM ROCK inhibitor (Y-27632, Stemgent). The cells were aggregated by centrifuging at 1500 rpm for 5 minutes, then incubated at 37 0 C overnight. The following day, we added 50% (50μl) of hNIM into each well (100μl final volume per well). The cells were cultured for 4 days for ectoderm induction, with 50% fresh hNIM changed daily. hNIM consists of DMEM/F12 with N-2 supplement (Life Technologies), NEAA 2 mg/mL (Sigma-Aldrich), heparin, 10 μM SB431542 and 100 ng human Noggin (R&D System). Day 4-7: 4 days after ectoderm induction, homogenous EBs derived from the three hiPSC lines (hiPSC-EBs) were pooled into a 100 mm non-treated dish and cultured in suspension in human neural progenitor medium (hNPM) with 50% fresh hNPM changed daily [27]. hNPM consists of neuronal basal medium with N-2 supplement (Life Technologies) and 100ng/mL human noggin (R&D System). Day 7-11: On day 7, floating hiPSC-EBs were re-plated in 6-well plates coated with Poly-L-Lysine (PLL) and Laminin (LM) at high density and cultured in hNPM with fresh medium replaced every other day. By day 11, neural tube-like structures and rosettes appeared in the center of hiPSC-EBs. The EB-derived rosettes were cultured in hNPM changed every other day for 4 additional days. The number of EBs containing neural rosettes (EB-derived rosettes) were counted and the percentage of neural induction was calculated using the formula: % Neural induction = # of EBS with ! 50% neural rosettes / total of EBs x 100 (Formula from Stemcell technologies technical manual). Day 12-19: On day 12, EB-derived rosettes were incubated with STEMdiff TM neural rosette selection reagent (Stemcell Technologies) at 37˚C for 1 hour and manually dislodged to separate them from non-neural ectoderm-like neural crest cells. The EB-derived rosettes were collected, centrifuged at 800 rpm for 3 minutes and resuspended in 2 mL hNPM. Human neural precursor cells (hNPCs) from the EB-derived rosettes were replated in a 6-well plate coated with PLL/LM and cultured in hNPM for 5-7 days (with medium changed daily) until 80-90% confluent, then expanded and frozen, or subjected to further neuronal differentiation. The proliferation of hNPCs derived from PIGAwt, PIGAc.1234C>T, and PIGAnull hiPSC lines, was measured using the Click-iT1 EdU Alexa Fluor1 488 Imaging Kit (Thermo Fisher Scientific). Neuronal stem cells and progenitor cells were characterized by immunofluorescence staining with antibodies against SOX1 (AF3369, R&D System), Nestin (MAB 1259, R&D System) and PAX6 (Ab5790, Abcam). Day 19: 80-90% confluent hiPSCderived hNPCs were dissociated into single cells with Accutase (Sigma) at 37 0 C for 5 minutes, counted, split 1:4 and transferred to a new 6-well plate coated with PLL/LM for neuronal differentiation. For neuronal differentiation, German glass coverslips (NeuVitro) were coated with PLL/LM (Sigma) in 24-well plates and incubated at 4 0 C overnight, washed three times with sterile water and air dried. Mouse astrocytes were seeded as a feeder layer. 100μL of inactive mouse astrocyte suspension P2 (M1800, ScienCell) was applied to the PLL/LM-coated coverslips and incubated at 37 0 C for 2 hours to promote attachment before addition of mouse astrocyte medium (ScienCell). Human NPCs derived from the three hiPSC lines were plated onto PLL/LM-coated coverslips or mouse astrocyte feeder coverslips, at low density in hNPM in a 37 0 C incubator. The next day, hNPM was replaced with Neuronal differentiation medium (NDM) supplemented with 10 ng/mL BNDF and GNDF (R&D System) and 1 mM cyclic AMP (Sigma). NDM consists of Neurobasal medium (Life Technologies) with 1x B27 supplement (Life Technologies) and 1:100 GlutaMax (Life Technologies). The cells were cultured in NDM for 4-6 weeks and fresh NDM was added twice a week. hNPCs grown on mouse astrocyte feeders were cultured in NDM with fresh NDM added twice a week. After 2 to 4 weeks of neuronal differentiation, the neurons from PIGAwt, PIGAc.1234C>T and PIGAnull NPCs were used for electrophysiology studies.

Electrophysiological recording
Whole-cell patch-clamp recordings were performed on 2-week-old hiPSC neurons. Parallel cultures were used for recordings of all three hiPSC NPC lines. The recording chamber was perfused with fresh HEPES-buffered saline composed of 140 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 10 mM HEPES, and 10 mM glucose, pH adjusted to 7.4 with NaOH, osmolarity adjusted to 300 mOsm with sucrose [28]. The recording micropipettes (tip resistance 3-6 MO) were pulled (Model pp-830, Narishige) from borosilicate glass (WPI Inc.). Internal solution containing 135 mM KCl, 1.1 mM CaCl 2 , 10 mM HEPES, 2 mM EGTA, 3 mM Mg-ATP, 0.5 mM Na 4 GTP, pH adjusted to 7.3 with KOH, osmolarity adjusted to 290 mOsm with sucrose. Recordings were made using an Axopatch 200B amplifier (Axon Instruments). Signals were sampled at 10 kHz and filtered at 2 kHz. With voltage-clamp whole-cell configuration (holding at -60 mV), currents were measured with an Axon 700B amplifier and the pCLAMP 9.2 software package (Molecular Devices) in order to characterize Na + /K + current. With current-clamp whole cell configuration, step currents (100 to 900 pA, 100 pA apart) were injected to elicit action potentials. Each injected current lasted 200 ms (or 500 ms) and separated by an interval of 3 s. All recordings were performed at room temperature.

Immunocytochemistry
Cells were fixed with 4% paraformaldehyde for 15 minutes and permeabilized and blocked with 10% donkey serum with 0.3% Triton X-100 (Sigma) in PBS for 30 minutes. Samples were incubated with primary antibodies at 4 0 C overnight, then incubated with appropriate secondary antibodies for 1 hour and stained with Hoechst 33342. Images were taken using a Zeiss confocal microscope (Zeiss LSM510 META) and analyzed with Zen imaging software (Zen 2012 blue edition) and ImageJ (NIH). Cell proliferation of hNPCs during neural induction was assayed using the EdU Alexa Imaging Kit (Thermo Fisher). Fluorescent cells were counted and the percentage of positive cells was calculated relative to the total number cells in the field. Following Edu staining, the hNSCs and hNPCs derived from hiPSCs were characterized by immunofluorescence for SOX1 (AF3369, R&D System) and PAX6 (Ab5790, Abcam). Cells positive for SOX1 or PAX6 were counted manually and the percentage of positive cells was calculated and normalized to the total number of cells per given image field. At least 20 image fields from three slides per cell line were counted. Four-week-old neurons were characterized using mouse anti-MAP2 (1:10,000; M2320, Sigma), rabbit anti-GABA (1:500; A2052, Sigma), mouse anti-synapsin (1:500; 106001, Synaptic System), VGLuT1 (1:500; 135303, Synaptic System), and VGAT (1:500; 131003, Synaptic System). Images were acquired with the same settings for all three cell lines. For analysis of synaptic density, total VGluT1+ in a given image was counted using ImageJ Analyze Particles and VGluT1+ density was determined by using identical settings across multiple images. The numbers of VGAT+ and VGluT1+ were also manually counted. The density of VGAT+ was determined per 100 μM total dendritic length.

Western blot
The three hiPSC lines were incubated with or without BMP4 (50 ng/mL) for 4 hours. Protein was extracted as previously described [16] and concentration was determined by BCA assay (Pierce). Fifteen μg of protein was separated by SDS-PAGE (Invitrogen) and transferred to Hybond-P PVDF membrane (Amersham). The membrane was blocked with 5% non-fat dried milk in TBST for 30 minutes, incubated with diluted PhosphoSmad1/Smad5/Smad8 antibody (9511S, Cell Signaling Technology) in 5%(w/v) BSA, 1xTBST at 4˚C with gentle shaking overnight, then incubated with HRP-conjugated donkey anti-rabbit IgG (1:2,000 in TBST, Cell Signaling Technology) at ambient temperature for 1 hour. Reactive protein bands were visualized using ECL plus Western blotting detection reagents (GE Healthcare) and band intensities were calculated using the Bio Rad Chemi Doc XRS imaging system. Beta-actin (13E5) Rabbit mAB (4970, Cell Signaling Technology) was used for a loading control.

Complement-mediated cytotoxicity assay
The complement-mediated cytotoxicity assay was performed based on a cell viability assay that has been described [27]. PIGAwt, PIGAc.1234C>T and PIGAnull hNPCs were incubated with normal human serum (Sigma-Aldrich), normal human serum containing cobra venom (0.5 μg/mL, Complement Technology, Tyler, TX) to activate the alternative pathway of complement, and human serum from a patient with atypical hemolytic uremic syndrome (aHUS) where there is constitutive overactivity of the alternative pathway of complement. Sera were diluted (20%) in gelatin veronal buffer (GVB) (Sigma-Aldrich) and incubated with TF1PIGAnull cells for 30 minutes at 37˚C. The cells were then washed with PBS and incubated with WST-1 cell proliferation reagent (Roche) for 2 hours at 37˚C. Absorbance was measured in an iMark Microplate Absorbance Reader (Bio-Rad, Hercules, CA) at 490 nm with a reference wavelength at 595 nm. Heat-inactivated serum was used as a negative control. The percentage of non-viable cells was calculated using the formula: 100 -(sample absorbance x 100 / heat inactivated sample's absorbance).

Statistics
Statistics were performed using GraphPad Prism 6 (GraphPad Software). A one-tailed, unpaired Student's t test, Mann-Whitney U test, or one way ANOVA and multiple comparisons was used as appropriate. An F test was performed in Prism to determine whether variances were similar among groups. A P value of less than 0.05 was considered statistically significant.

PIGAc.1234C>T is a read-through mutation
Germline PIGA null mutations are embryonic lethal and lead to a block in mesodermal and endodermal differentiation due to decreased BMP4 signaling. The PIGAc.1234C>T mutation associated with inherited GPI anchor protein deficiency occurs in the last exon of PIGA and is predicted to result in a truncated protein missing the final C-terminal 73 amino acids; thus, we were initially surprised to find this was a hypomorphic mutation in a male patient [4]. To investigate further, we used an expression vector and stably transfected TF1 PIGAnull cells with either wildtype PIGA full-length cDNA (PIGAwt), a full-length PIGA cDNA containing the c.1234C>T mutation (PIGAc.1234C>T), or a truncated PIGA cDNA (PIGAtr411) encoding only the first 411 amino acids as predicted based on the nonsense c.1234C>T mutation (Fig 1A). The reconstitution assay shows that the PIGAc.1234C>T cDNA rendered a reduced, but detectable, level of GPI anchor protein expression (Fig 1B). The hypomorphic state is likely due to "read-through" of the premature stop codon. In contrast, the truncated PIGAtr411 cDNA (lacking the final 73 amino acids) was unable to restore GPI anchor protein expression.

The PIGAc.1234C>T mutation does not impair hematopoiesis
To assess the effect of the PIGAc.1234C>T mutation on hematopoiesis, we used a piggyBac (PB) transposon-based gene transfer system to introduce the mutation into a previously established PIGAnull hiPSC line and differentiated the three hiPSC lines (PIGAwt, PIGAc.1234C>T, and PIGAnull) toward mesoderm. The PIGAc.1234C>T hiPSCs were hypomorphic for the GPI anchored proteins CD59 (Fig 2A) and alkaline phosphatase (Fig 2B). We previously showed that PIGAnull hiPSCs do not generate hematopoietic cells [16]. The level of phosphorylated Smads (forms 1/5/8) upon BMP4 induction was not significantly different between the PIGAc.1234C>T hiPSCs and the PIGAwt hiPSCs (Fig 2C). The hypomorphic PIGAc.1234C>T hiPSCs generated significantly fewer EB-derived blood-like cells (Fig 3A), but were able to generate a similar percentage of CD34+CD45+ hematopoietic cells (Fig 3B  and 3C) and similar numbers of hematopoietic colony forming cells as the PIGAwt hiPSCs (Fig 3D). Taken together, the hypomorphic PIGAc.1234C>T mutation associated with germline GPI anchor deficiency is permissive for hematopoiesis, and this appears to explain why children born with germline mutations that lead to partial GPI anchor protein expression have no demonstrable hematopoietic defects. GPI anchor proteins are important for neuron development. Common features of inherited GPI anchor deficiency are severe intellectual disability, seizures, and other central nervous system abnormalities; thus, we next studied neuronal differentiation of the hiPSCs (S1 Fig). PIGAwt, PIGAc.1234C>T, and PIGAnull hiPSCs were differentiated into neural progenitor cells. By day 11 of neuronal differentiation, neural tube-like structures (rosettes) appeared in the center of hiPSC-EBs, but the PIGAnull hiPSCs formed fewer neuron rosettes ( Fig 4A) and were less capable of differentiating into cells of neuronal fate (Fig 4B). We next measured cell expression of SOX1 (a marker of neural stem cells) (Fig 4C and 4F), PAX6 (a marker of neural progenitor cells) (Fig 4D and 4F), and proliferation using EdU (Fig 4E) in the neural progenitor cells derived from the three cell lines. A PIGA gene dosage effect was observed with the PIGAc.1234C>T derived neural progenitors demonstrating intermediate proliferation and the PIGAnull derived neural progenitors showing virtually no proliferation (Fig 4E). Next, we plated the detached PIGAwt and PIGAc.1234C>T derived rosettes in NDM medium and maintained them in culture for two to four weeks; PIGAnull rosettes were not capable of neuron differentiation (data not shown). After four weeks of neuron differentiation, a phenotypic distinction in synapse formation and proliferation was observed between the PIGAwt and PIGAc.1234C>T derived neurons (Fig 5A). Consistent with the proliferation data, the GABA density of neurons derived from the PIGAc.1234C>T hiPSCs was much less due to reduced maturation of the PIGAc.1234C>T derived MAP2 and GABA expressing neurons (S2 Fig). To examine synapse morphology, we co-immunostained neurons with presynaptic markers, Synapsin and VGLuT (Fig 5B), and Synapsin and VGAT (Fig 5C). Quantification of VGLuT ( Fig 5D) and VGAT (Fig 5E) revealed reduced synapse formation in the PIGAc.1234C>T derived neurons compared to the PIGAwt derived neurons. Taken together, these studies suggest that GPI anchor deficient neurons have significantly reduced proliferation, reduced maturation and presynaptic defects (S3 Fig).

GPI anchor protein deficient neurons exhibit impaired membrane depolarization
We next measured induced action potentials with whole-cell patch-clamp recordings performed on two-week-old neurons derived from the hiPSCs (Fig 6A). As shown in Fig 6B, the PIGAwt neurons showed normal transient inward sodium currents and sustained outward potassium currents in response to membrane depolarization. The PIGAc.1234C>T neurons exhibited a decrease in both currents. The PIGAnull neurons showed almost no sodium current and markedly reduced potassium currents. The number and amplitude of action potentials was also reduced in the PIGAc.1234C>T neurons compared to the PIGAwt neurons ( Fig  6B), further highlighting the synaptic defect associated with GPI anchor protein deficiency. No depolarizations were observed in the PIGAnull line, consistent with our data above demonstrating an inability of this cell line to differentiate into mature neurons (Fig 6C). Spontaneous excitatory postsynaptic currents (EPSCs) were observed only in the PIGAwt neurons, indicating an impaired neuronal property of PIGAc.1234C>T neurons and inability of PIGAnull hiPSCs to properly differentiate (Fig 6D).
PIGAc.1234C>T derived neurons are more vulnerable to complementmediated cytotoxicity GPI anchored proteins CD55 and CD59 are important complement regulatory proteins, suggesting that complement activation could play a pathophysiologic role in the neurodegeneration and or seizures associated with inherited GPI anchor protein deficiency. To test the ability of GPI deficient neuronal stem cells to withstand complement mediated attack, we exposed day 17 neuronal progenitor cells derived from PIGAwt, PIGAc.1234C>T and PIGAnull hiPSCs to 20% normal human serum, 20% normal human serum containing cobra venom (0.5 μg/ mL) to activate the alternative pathway of complement, or 20% human serum from a patient with atypical hemolytic uremic syndrome (aHUS), where there is constitutive activation of the alternative pathway of complement [29,30]. A PIGA gene dosage effect was again observed, with PIGAc.1234C>T and PIGAnull derived neural stem cells demonstrating a marked reduction in cell viability following exposure to activated alternative pathway of complement (Fig 7). . Representative example of images of hiPSC-derived EBs and EB-derived rosettes during neural differentiation. Neural induction and rosette formation upon neural induction was assessed in three cell lines using a serum-free EB generation method. On day 2 (left) of hiPSCderived EBs from PIGAwt, PIGAc.1234C>T, and PIGAnull after forced aggregation (20X magnification, scale bar is 50μm). On day 4 (middle), single homogeneous hiPSC-EBs collected were pooled in a 10 cm plate (4X magnification, scale bar is 100μm). On day 11 (right), neuroepithelial cells appeared and neural tube-like rosettes formed (EBderived rosettes) and scale bar is 50μm. (B). Neural induction rates from EB-derived rosettes. The percentage of EB derived rosettes was 88.8% ± 4.6, 75.5% ± 9.8 and 68.4% ± 6.9 for PIGAwt, PIGAc.1234C>T, and PIGAnull, respectively. PIGAwt versus PIGAc.1234C>T (p>0.05, NS) and PIGAwt versus PIGAnull (*p<0.05, one way ANOVA and Multiple comparisons). Neural induction from PIGAnull hiPSCs was less than 70%. All values were mean ±SD.

Discussion
We described here a model of germline GPI anchor protein deficiency using human iPSC lines. Germline mutations that lead to a subtle decrease in cell surface display of GPI anchored proteins are increasingly being recognized to cause inherited syndromes that manifest with dysmorphic faces, intellectual disability, and intractable seizures. Here, we generated a human model of germline GPI anchor protein deficiency by establishing a human iPSC line containing a hypomorphic mutation PIGAc.1234C>T that mimics the human disease. PIGAc.1234C>T hiPSCs had no substantive hematopoietic defects upon hematopoietic differentiation; however, upon neuronal differentiation, they had significant defects in proliferation and synapse formation. Moreover, GPI anchor deficient neuronal stem cells derived from the PIGAc.1234C>T human hiPSCs were more susceptible to complement-mediated cytotoxicity in an in vitro assay exposing them to activation of the alternative pathway of complement, presumably due to deficiency of GPI anchored complement regulatory proteins.
Acquired PIGA mutations arising from a multipotent hematopoietic stem cell cause PNH and lead to chronic complement-mediated hemolysis due to marked deficiency or absence of GPI anchored complement regulatory proteins CD55 and CD59 [13,31]. Compound heterozygous mutations (one inherited and one acquired) in PIGT and germline mutations involving CD59 also lead to a PNH-like phenotype with severe anemia from marked intravascular hemolysis [32]. Moreover, complete knock-out of Piga in mice is embryonically lethal [33] and knockout of PIGA in human iPSCs does not allow for hematopoietic differentiation due to perturbed signaling through BMP4 [16]. Interestingly, none of the reported patients with germline mutations in PIGA or any of the other genes involved in GPI anchor biosynthesis have exhibited any hematopoietic defect or hemolysis. Similarly, the PIGAc.1234C>T hiPSCs demonstrated no defect in hematopoiesis. The lack of hemolysis in these patients is likely due to 1) the very subtle defect (~a quarter of a log decrease) in cell surface GPI anchored proteins, and 2) the fact that red blood cells from most patients with germline GPI mutations have normal expression of CD55 and CD59; the defect is most conspicuous on granulocytes [1,4,34].
Despite the subtle decrease in GPI anchor proteins observed in the PIGAc.1234C>T hiPSCs and primary cells of patients with germline hypomorphic mutations in PIGA and other GPIrelated genes, the neurologic phenotype is severe. Neurons derived from the PIGAc.1234C>T iPSCs displayed a defect in cell proliferation. In addition, neurons derived from PIGAc.1234C>T iPSCs displayed a decrease in presynaptic markers Synapsin, VGAT, and VGLuT, possibly explaining the CNS imaging abnormalities, seizures, and severe intellectual disability associated with these disorders. Patch-clamp studies showed that the PIGAc.1234C>T derived neurons had a reduction in the ability to fire action potentials, a reduction in amplitude of sodium and potassium currents and a reduction in spontaneous EPSCs, further demonstrating the neurodevelopmental defects and synaptic dysfunction associated with reduced GPI anchor protein expression.
There are more than a dozen known GPI anchored proteins that are expressed during neurogenesis that could be responsible for the neurodevelopmental defects associated with GPI anchor protein deficiency. Members of the nerve growth factor (NGF) and glial cell linederived neurotrophic factor (GDNF) families are crucial for the development and maintenance of the central and peripheral nervous system [35]. NGF and GDNF are added to our neuronal stem cells in culture on day 19 to promote proliferation and differentiation. Susceptibility of hNPCs to complement-mediated cytotoxicity. The complement-mediated cytotoxicity assay was performed in PIGAwt, PIGAc.1234C>T, and PIGAnull hNPCs. The cells were incubated with normal serum, cobra venom factor or atypical HUS serum for 30 minutes. The percentage of non-viable cells was measured using WST-1 cell proliferation reagent. PIGAnull hNPCs were the most susceptible to complement-mediated cell killing, followed by PIGAc.1234C>T cells, which were more susceptible to complement-mediated killing compared to PIGAwt cells (p<0.05). Values were mean ± SD. NS (normal human serum, black), CVF (cobra venom factor, patterned), aHUS (serum from a patient with atypical hemolytic-uremic syndrome, white). https://doi.org/10.1371/journal.pone.0174074.g007 Receptors (GFRα1, GFRα2, GFRα3, and GFRα4) for GDNF-family ligands are GPI anchored and signal through RET (receptor tyrosine kinase). Thus, absence (as in PIGAnull hiPSCs) and deficiency (as in PIGA c.1234C>T hiPSCs) would be expected to result in a decrease in maintenance and proliferation of neurons. Interestingly, Gdnf +/mice have severe cognitive deficits similar to humans with inherited GPI anchor deficiency [36]. Alkaline phosphatase, another GPI anchored protein, is important for neuronal development and plays a major role in vitamin B6 metabolism [37]. Alpl knockout mice develop intractable seizures that are responsive to administration of vitamin B6. Indeed, Kuki et al, described a 9-year-old male with intellectual disability from inherited GPI deficiency due to a germline PIGO mutation with vitamin B6 responsive epilepsy [38]. A variety of other GPI anchored proteins are also critical for the developing nervous system. CD56 (NCAM) is important for neurite growth and development [39,40]. Contactin is an axon-associated adhesion molecule that is important for axon connections in development [41]. Thy-1 [42] and the prion receptor are involved in neurite outgrowth [43]. CD59 is the most important regulator of terminal complement and is widely expressed by the nervous system. Rare patients with congenital, isolated CD59 deficiency have been described [44]. These patients present with a complement-mediated hemolytic anemia and peripheral polyneuropathy. In most cases, the neurologic symptoms are more prominent than the hemolysis, which only becomes clinically relevant during episodes of complement activation, especially during infections. The mechanism behind this observation seems to be the limited neuronal capacity of controlling complement activation because of low neuronal CD59 expression. Administration of eculizumab has been shown to abrogate hemolysis and lead to neuronal regeneration [45]. Neuronal stem cells derived from PIGAwt, PIGAc.1234C>T, and PIGAnull hiPSCs showed a dose-response effect in their ability to protect themselves from complement-mediated attack. Complement has been increasingly implicated in a variety of diseases associated with neurodegeneration and intellectual dysfunction [46][47][48]. Taken together, these data suggest that the neurodegeneration that accompanies inherited GPI anchor deficiency and germline CD59 deficiency may be complement-mediated, and may lend insight into the mechanism of neurodegeneration in macular degeneration [49] and other neurodegenerative diseases [50].

Conclusion
Here we report on a novel model of inherited GPI anchor deficiency using human iPSCs that provides key insights into the phenotypic features of cognitive disability, neurodevelopmental defects, and neurodegeneration. This model will serve as an important platform in combination with precise gene editing approaches to determine which GPI anchor protein(s) are most responsible for the neuronal defects observed in inherited GPI anchor deficiency.