CNS glycosylphosphatidylinositol deficiency results in delayed white matter development, ataxia and premature death in a novel mouse model

Abstract

The glycosylphosphatidylinositol (GPI) anchor is a post-translational modification added to approximately 150 different proteins to facilitate proper membrane anchoring and trafficking to lipid rafts. Biosynthesis and remodeling of the GPI anchor requires the activity of over 20 distinct genes. Defects in the biosynthesis of GPI anchors in humans lead to inherited glycosylphosphatidylinositol deficiency (IGD). IGD patients display a wide range of phenotypes though the central nervous system (CNS) appears to be the most commonly affected tissue. A full understanding of the etiology of these phenotypes has been hampered by the lack of animal models due to embryonic lethality of GPI biosynthesis gene null mutants. Here we model IGD by genetically ablating GPI production in the CNS with a conditional mouse allele of phosphatidylinositol glycan anchor biosynthesis, class A (Piga) and Nestin-Cre. We find that the mutants do not have structural brain defects but do not survive past weaning. The mutants show progressive decline with severe ataxia consistent with defects in cerebellar development. We show that the mutants have reduced myelination and defective Purkinje cell development. Surprisingly, we found that Piga was expressed in a fairly restricted pattern in the early postnatal brain consistent with the defects we observed in our model. Thus, we have generated a novel mouse model of the neurological defects of IGD which demonstrates a critical role for GPI biosynthesis in cerebellar and white matter development.

Introduction

Inherited glycosylphosphatidylinositol deficiency (IGD) is defined by a deficiency of the cell surface glycosylphosphatidylinositol (GPI) anchor (1,2). The GPI anchor is a glycolipid added post-translationally to nearly 150 proteins and is required for membrane anchoring and lipid raft trafficking of these proteins (1,3). Over 20 genes are required for the biosynthesis and remodeling of the GPI anchor in the endoplasmic reticulum and Golgi apparatus. IGD patients display a wide array of phenotypes including neural, craniofacial, cardiac, renal, hepatic, ophthalmologic, skeletal, dental, dermatologic and sensorineural defects (1,2,4). This wide array of phenotypes highlights the broad requirement for GPI biosynthesis and remodeling in the development of many organ systems.

A recent review of IGD patients found that the most commonly affected organ system, regardless of the mutated gene in the GPI biosynthesis pathway, is the central nervous system (CNS) (2). These CNS phenotypes include structural CNS deficits including microcephaly, hypoplastic corpus callosum, hypoplastic cerebellum, white matter immaturity and cortical atrophy. Clinical findings related to these structural deficits include epilepsy, developmental delay, intellectual disability, hypotonia, hyperreflexia, chorea, ataxia and behavioral abnormalities with variable penetrance (2,5–7). Electroencephalogram imaging reveals abnormal electrical activity including hypsarrhythmia indicative of abnormal brain development in nearly all patients (2,8,9). Often, the epilepsy is intractable and severely impairs quality of life. Some reports have shown that vitamin B₆ is beneficial in treating IGD-related epilepsy. It has been hypothesized that the benefit of vitamin B₆ is related to the defects in GPI-anchored tissue nonspecific alkaline phosphatase, which dephosphorylates B₆, thereby allowing it to pass the membrane and participate in γ-aminobutyric acid (GABA) synthesis (10,11). However, it is unclear if all patients respond to vitamin B₆ supplementation and no rigorous study has tested this hypothesis, at least in part due to a relative lack of mouse models of IGD. Little is known about the requirement for GPI biosynthesis in the CNS or the mechanisms that lead to these clinical phenotypes. Therapy is currently limited to symptomatic treatment and most patients will die before age five by cardiac arrest, aspiration pneumonia or central respiratory failure (12). Thus, there is a clear need to understand the pathophysiology that results from GPI deficiency in the CNS.

The initiation of GPI biosynthesis requires the phosphatidylinositol glycan anchor biosynthesis class A (PIGA) protein, a component of the phosphatidylinositol N-acetylglucosaminyltransferase complex. This complex generates N-acetylglucosamine-phosphatidylinositol (GlcNac-PI) by transferring N-acetylglucosamine (GlcNac) from UDP-GlcNac to phosphatidylinositol (PI). GlcNac-PI is the first precursor in the biosynthesis of the GPI anchor and deletion of Piga completely abolishes GPI biosynthesis (1,3). Patients with germline mutations in PIGA develop a form of IGD termed multiple congenital anomalies-hypotonia-seizures syndrome 2 (MCAHS2: OMIM 300868) (13). A recent review of 10 MCAHS2 patients found that all had seizures, abnormal EEG findings, developmental delay and intellectual disability. A total of 8/10 patients had hypotonia and 5/10 had hyperreflexia, cerebellar hypoplasia and cortical atrophy (12). A total of 7/10 patients developed white matter defects including thin corpus callosum and white matter immaturity. Thus, PIGA deficiency serves as a representative model to study the broad requirement for GPI biosynthesis in CNS development.

Other groups have taken a genetic approach to ablate GPI biosynthesis in specific mouse tissues by combining a conditional allele of Piga with flanking loxP sites (Piga^flox) with tissue-specific Cre recombinase transgenics. Recently, we used this conditional allele to delete Piga in the developing mouse neural crest cells with Wnt1-Cre. This deletion resulted in cleft lip and cleft palate as well as hypoplasia of the craniofacial skeleton, highlighting a critical role for GPI biosynthesis in neural crest cell survival (14). Others have used this approach to reveal an important role for GPI biosynthesis in skin, limb, blood and craniofacial development (15–17). However, germline knockout of Piga using CMV-Cre resulted in embryonic lethality with neural tube defect, edema and cleft lip/palate precluding the analysis of any postnatal neurological phenotype (18). To overcome this, we sought to conditionally ablate GPI biosynthesis after the closure of the neural tube, at embryonic day E9.5 in mouse. To completely abolish CNS GPI biosynthesis at this stage, we chose to use Nestin-Cre to drive deletion of Piga broadly in the central and peripheral nervous system starting at E11.5 (19). As Piga resides on the X-chromosome, we generated Piga^flox/Y, Nes-Cre + hemizygous conditional knockout males (hemizygous cKO) with complete deletion of Piga in the Nestin lineage and Piga^flox/WT, Nes-Cre mosaic cKO females (mosaic cKO) containing one conditional allele of Piga and one wild-type Piga allele. Due to random X inactivation, these females develop a mosaic cKO of Piga in the Nestin lineage. Both hemizygous cKO and mosaic cKO animals survived to postnatal stages and developed some of the CNS phenotypes observed in IGD patients including white matter immaturity, gait imbalance, motor incoordination and early death. Our conditional knockout approach allowed us to investigate the requirement for GPI biosynthesis for the first time in vivo in the mammalian CNS. This model may also be used to test novel IGD therapeutic interventions in the future.

Results

Piga is expressed broadly in the CNS with enrichment in the corpus callosum and Purkinje cells

We initially sought to determine the expression of Piga in the embryonic and postnatal mouse brain. Published and publicly available single cell RNA sequencing data sets are now a robust tool to query gene expression. Zhang and colleagues performed one such experiment in the P7 mouse cortex and found that Piga is broadly expressed in a variety of neural cell types, but with a clear enrichment in astrocytes and newly formed and myelinating oligodendrocytes (20) (Fig. 1A). We further explored Piga expression with immunofluorescence in the postnatal CNS with a commercially available polyclonal αPIGA antibody. We found that PIGA protein is enriched in the corpus callosum and cortical periventricular areas at P0 (Fig. 1B and C). At P10, higher expression was maintained in the periventricular areas, and PIGA expression was evident in the Purkinje cells of the cerebellum at this stage (Fig. 1D and E). Further examination revealed clear immunoreactivity in both the cytoplasm and dendrites of the Purkinje cell (Fig. 1E). We concluded from these data that PIGA shows increased expression at P0 in developing white matter areas such as the corpus callosum, consistent with the P7 RNA-Seq data with highest expression in myelinating oligodendrocytes. Once the mature morphology of the cerebellum is established at P10, PIGA is strongly expressed in the Purkinje cells, the sole output of all motor coordination in the cerebellar cortex.

Figure 1.

Piga expression in the CNS. Data in A is replotted directly from (20) (brainrnaseq.org). P0 PIGA (Green) expression in the corpus callosum (CC) and periventricular areas (B, C). P10 PIGA expression in the cerebellar Purkinje cells (D, E). PIGA is shown as green NeuN as a neuronal marker is in red, DAPI counterstain is blue. Fb: forebrain, Mb: Midbrain, BA1: Branchial Arch 1, MNP: Medial Nasal Process, Lb: Limb bud, Oligo.: Oligodendrocyte, OPC: Oligodendrocyte Progenitor Cell. (Scale bars = 500 μm in B, D, 100 μm C, E).

Deletion of Piga in the Nestin-Cre lineage results in CNS GPI deficiency

As very few studies have focused on the role of glycosylation in neural development, we aimed to first take a broad approach and delete GPI biosynthesis in the entire CNS/PNS. We hypothesized this would illuminate the critical requirement for GPI biosynthesis in neural development beyond neurulation. We are unable to study the effect of GPI biosynthesis at later neurogenesis stages in the germline Piga null mice because these mutants develop neural tube defects or die very early in development (18).

We obtained the conditional Piga allele (B6.129-Piga^<tm1>) which has been used in several studies to delete Piga with Cre/lox technology (14,16,18). Upon Cre recombination, the final exon of the conditional Piga is deleted and GPI biosynthesis is severely disrupted indicating the allele efficiently abolishes Piga function (14,16,18). We crossed these mice to Nestin-Cre mice (B6.Cg-Tg(Nes-cre)1^Kln/J). This Nestin-Cre allele has been shown to mediate loxP excision at E11.5 in the forebrain, midbrain, hindbrain and spinal cord and maintain expression in neural stem cells throughout development (19,21,22). Therefore, targeting by this Nestin-Cre transgene is broad and includes neurons, astrocytes and oligodendrocytes.

Piga is located on the X chromosome in mouse and human resulting in different degrees of Piga deletion in male versus female progeny from Piga^flox crosses with Nestin-Cre. Males with one copy of Piga^flox/Y; Nestin-Cre^+/− are termed hemizygous conditional knockout (hemizygous cKO) males and should lack all Piga expression in the Nestin lineage. Females used in this study are mosaic cKOs with the genotype Piga^flox/X; Nestin Cre^+/− carrying one floxed allele and one wild-type allele of Piga. We chose to continue our behavioral studies with mosaic cKO females because they survive significantly longer than hemizygous cKO males and develop phenotypes similar to those observed in IGD patients including ataxia, decreased lifespan and neurological decline.

Mice were analyzed from E16.5 to postnatal stages. To confirm the Nestin-Cre transgenic was ablating Piga in the CNS, we performed PCR with three primers spanning the targeted region to differentiate between wild-type, floxed (Piga flox) and recombined (Piga del) alleles (Fig. 2A and B). By analyzing the Piga deleted PCR product in hemizygous cKO males, we found Nestin-Cre targeted Piga in multiple CNS tissues including the cortex, cerebellum, as well as subcortical structures including the thalamus, hypothalamus, midbrain, pons, and medulla (Fig. 2B, lane 2–4). However, Piga excision was not noted in genomic DNA from the tail of the same animal, confirming that the targeting was specific and consistent with previous reports using the Nestin-Cre (Fig. 2B, lane 1). We did observe some remaining Piga flox PCR products in the CNS tissues indicating that the Cre did not target absolutely every cell in these tissues. We suspect this Piga DNA may be derived from non-neural tissues such as blood vessels, blood and microglia which are not targeted by Nestin-Cre. We conclude that our conditional approach targets Piga for deletion in the CNS.

Figure 2.

Deletion of Piga in the Nestin-Cre lineage results in CNS GPI deficiency. Genotyping strategy for identifying Piga flox (primers 1 and 2 generate a 350 bp product), Piga deleted (del) (primers 1 and 3 generate a 600 bp product) (A). PCR of tail and various CNS structures of Piga hemizygous cKO male mouse (B). Western blot of PIGA (Green) in subcortical tissues from WT and Piga hemizygous or mosaic cKO female littermates with tubulin loading control (Red) (C). Quantification of PIGA/tubulin signal from western blot of N = 6 WT and N = 3 mosaic cKO subcortical lysates and two littermatched sets of hemizygous knockouts (D). CB = cerebellum, Subcort = subcortical tissues.

It is known that mothers with one mutant allele of Piga display skewing of X inactivation to silence the affected X chromosome carrying the mutant allele (12). Thus, we were concerned that skewing toward the wild-type X chromosome would result in residual PIGA expression higher than the expected 50% of wild-type in mosaic cKOs. To confirm that the PIGA protein was reduced in the brain of Piga mosaic cKO females, we performed western blotting for PIGA in subcortical lysate including the thalamus, basal ganglia, pons, midbrain and medulla. Western blotting showed a reduction of PIGA protein compared to wild-type littermate controls of approximately 50–75% when normalized to tubulin loading control (Fig. 2C–D). Similar results were shown from Piga hemizygous cKO animals (Fig. 2C–D). These data confirm PIGA protein reduction in the brain of the Piga cKO mice. Consistent with the PCR results (Fig. 2B) and known expression of Nestin-Cre (23), the reduction of PIGA is not complete. We conclude that this is from tissues not expressing Nestin-Cre.

CNS GPI deficiency results in decreased survival and weight, but does not grossly affect the structure of the brain

We hypothesized that GPI deficiency in the CNS lineage would result in structural defects of the brain including cerebellar hypoplasia, dilated lateral ventricles and microcephaly as has been observed in a subset of IGD patients (13). To our surprise, we found no significant structural defects in the cortex of Piga mosaic cKO mice at birth (Fig. 3A and B) or up to weaning stages (Fig. 3C and D). We observed mild hypoplasia of the cerebellum in Piga mosaic cKO mice as evidenced by H&E staining (Fig. 3E and F). We observed that both the Piga mosaic cKO females and the Piga hemizygous cKO males gained less weight than their WT littermates and were easily identifiable by this phenotype (Fig. 3G). By P7, all mosaic cKO mutants were approximately half the weight of their WT littermates. We also noticed that mutants died postnatally with the oldest mutant only surviving to P23 (Fig. 3H). Piga hemizygous cKO males that lack all GPI biosynthesis in the CNS die on average twice as quickly as the mosaic cKO females suggesting a gene dosage effect (Fig. 3H). We hypothesize that the genetic mosaicism in females allows for residual expression of Piga allowing longer survival than hemizygous cKO males.

Figure 3.

CNS GPI deficiency results in cerebellar hypoplasia, decreased weight gain and decreased survival. Coronal H&E section of WT and mosaic cKO cortex at E18.5 (A, B). Gross whole brain of WT and Mosaic cKO at P22 (C, D). H&E section of the cerebellar hemisphere in WT and mosaic cKO (E, F). Average weight of WT and mosaic cKO mice from P0 to P15 (G). Survival curve of WT (black), mosaic cKO (Red), and hemizygous cKO (Blue) mice (H). All paired images shown at same magnification.

CNS GPI deficiency results in neurological decline and ataxia

While the cause of death in mutants remains unclear, all mosaic cKO mice developed progressive ataxia, tremor and a hindlimb clasping phenotype. Moribund mutants were unable to walk but moved their forelimbs and hindlimbs spontaneously while laying on their side in the host cage. Moribund mutants were euthanized at this stage.

Between P10 and P19, we observed a progressive hindlimb clasping phenotype that worsened over time in the mosaic cKO mutants. Hindlimb clasping is common to many models of neurological disease with a wide variety of underlying pathophysiologies in the brain and spinal cord (24). In the tail suspension test, we scored the hindlimb clasping phenotype according to standard protocols and noted a statistically significantly lower score in the mutants indicating a neurological phenotype (P < 0.0001, Fig. 4A–C).

Figure 4.

CNS GPI deficiency results in hindlimb clasping, ataxia, and tremor. Hindlimb clasping score of wild-type and mosaic cKO littermates at P16 (A). Images of wild-type and mosaic cKO littermates with mosaic cKO displaying hindlimb clasping (B, C). Surface righting reflex test in wild-type and mosaic cKO littermates (D). Forelimb Wire Hang test in wild-type and mosaic cKO littermates (E). Rotarod testing of wild-type and mosaic cKO littermates at P12 and P15, distance travelled (F) and latency to fall (G).

Given this preliminary indication of neurological disease in postnatal mosaic cKO animals, we sought to test other aspects of neurodevelopment, including developmental landmarks such as reflexes. First, we performed the righting reflex test, which tests the ability of the mouse to right itself into a prone position after being placed in a supine position. Mosaic cKOs were severely impaired in their ability to right themselves as measured by the latency to right themselves (P < 0.0001; Fig. 4D). These data indicate that the mutants were defective at either the labyrinthine reflex, limb coordination and/or strength as they actively, but unsuccessfully, attempted to right themselves. To test their strength, we performed the forelimb wire hang test in which the mice were suspended on a wire by their forelimbs over a padded surface and measured their latency to fall. Interestingly, we found that the mosaic cKOs were able to hang onto the wire statistically longer than their WT littermates, P = 0.0019, Fig. 4E). These data indicate that the function of the skeletal muscle system and strength was not impaired in the mutant forelimbs. As the mosaic cKOs show deficits in surface righting, but not the forelimb wire hang, we suspected that mutants developed defective limb coordination, ataxia.

Over time, the mutants developed ataxia and a persistent tremor (Supplementary Material, Fig. S1). The signs of this were first noticeable at approximately P12 and progressively worsened until approximately P20 when the mutants became moribund. To quantify and further study the ataxia phenotype, we performed rotarod testing. The rotarod assays motor coordination and balance by challenging the mice to maintain their balance on a rotating rod that increases in speed of rotation gradually. We found that the mosaic cKOs performed poorly, with significantly decreased travel and latency to fall compared to WT littermates at P12 and P15 (ANOVA P = 0.0001 for distance traveled and P < 0.0001 for latency to fall; Fig. 4F and G). The early lethality of the hemizygous cKO males precluded rotarod analysis. Taken together, all these data from behavioral testing indicate the mosaic cKOs displayed defective motor coordination and balance.

CNS GPI deficiency delays white matter development

Recent reviews of neural GPI-anchored proteins (GPI-APs) highlighted the diverse importance of GPI-APs in many neural processes including axon outgrowth, regeneration, synapse formation, neuron cell adhesion and oligodendrocyte development (25,26). We reasoned that failure to properly present one or more specific GPI-APs on the cell membrane may be a cause of the phenotypes observed in the mosaic cKOs. Therefore, we reviewed the phenotypes of mouse knockouts for neural GPI-AP genes in the Mouse Genome Informatics resource. Of the GPI-AP knockout mouse models, our Piga mosaic cKO mutants most resemble the phenotype observed in the contactin 1^−/− mouse (27,28). Contactin 1^−/− mice weigh significantly less than their WT littermates, develop progressive ataxia and die at approximately P19. Colakoglu et al. showed that contactin 1 is critical for oligodendrocyte development, and mutants lack proper CNS myelination and display defects in cerebellum micro-organization (27). Indeed, multiple mutants in myelin development display an ataxic tremor and early death phenotype. Given the phenotypic similarity between contactin 1^−/− mice and Piga mosaic cKOs, we sought to determine the degree of myelination in Piga mosaic cKO mutants.

We performed immunohistochemistry for myelin basic protein (MBP) in WT and mosaic cKO mutants at P19 (n = 4 for each). We found that three out of four mutants examined showed reduced MBP staining compared to controls indicating that myelination is defective in the mosaic cKO mutants (Fig. 5A–D). We concluded that developmental myelination is impaired in Piga mosaic cKO mutants. This phenotype has been observed in multiple IGD patients and may be partially responsible for the tremor and early death we observe in the Piga mosaic cKO mutants.

Figure 5.

CNS GPI deficiency delays white matter development. Immunohistochemistry of αMBP in WT (A, C) and mosaic cKO mice (N = 3/4 mutants delayed; B, D). CB: cerebellum, CC: corpus callosum, MB: midbrain, Md: medulla, Ps: pons, Th: thalamus. CNS GPI deficiency impairs Purkinje cell arborization. Purkinje cell (αCalbindin Green) and granule cell (αNeuN Red) immunofluorescence in WT and mosaic cKO mice (A, B). Representative images from N = 4/4 mosaic cKO. (Scale bars = 100 μm. All paired images shown at same magnification.).

CNS GPI deficiency impairs Purkinje cell dendritic arborization

The ataxia we observed in the Piga mosaic cKOs suggested a defect in proper cerebellar function as this is a major center for coordination of motor movement. Defects in either granule cell development or Purkinje cell development can lead to ataxic phenotypes (29–31). Granule cells are the major excitatory input to Purkinje cells, and Purkinje cells are the major inhibitory output to the deep cerebellar nuclei. Given the expression pattern of Piga in the postnatal brain, we hypothesized that defects in the Purkinje cell layer were responsible for the ataxic phenotype in Piga mosaic cKO mice. To assess defects in cerebellar development, we performed immunofluorescence for NeuN to highlight the inner granular layer and Calbindin staining to examine the morphology of the Purkinje cells. The granule cell layer of the cerebellum did not appear to be affected, but the Purkinje cell layer showed marked defects in Purkinje cell dendritic arborization in the molecular layer (n = 4/4 mutants examined). We found the dendritic tree to be consistently less elaborated and branched in the Piga mosaic cKO mice compared to wild-type littermates (Fig. 5E and F).

To obtain a broader and unbiased understanding of the differences between the wild-type and Piga mosaic cKO cerebella, we performed RNA-Seq on bulk right cerebellar hemispheres in 3 wild-type and 3 Piga mosaic cKO littermates at P20. Differential gene expression analysis identified 176 upregulated genes (Log₂Fold Change ≥1, P_adjusted < 0.05), 67 down regulated (Log₂Fold Change ≤ 1, P_adjusted < 0.05) and 17 370 genes that showed no statistical expression difference in the Piga mosaic cKO cerebella compared to wildtype controls (Fig. 6A). Of note, the three most highly expressed genes included myelin protein zero, tyrosine hydroxylase, and perlipin 4 (Fig. 6B). Gene Ontology (GO) analysis identified an enrichment for pathways including neuropeptide signaling, iron transport response to hypoxia and circadian rhythm for the upregulated genes in the Piga mosaic cKO mutant (Fig. 6C).

Figure 6.

RNA-Seq analysis. (A) Volcano plot of differentially expressed genes from RNA sequencing of WT and Piga mosaic cKO right cerebellar hemisphere at P20, blue = downregulated genes and red = upregulated genes. (B) Top 20 upregulated and downregulated genes (C). GO analysis of upregulated genes in the Piga mosaic cKO cerebella.

Discussion

In this work, we investigated the role of GPI biosynthesis in the developing CNS as it is the most affected organ system in patients with pathogenic variants in GPI biosynthesis genes (32). We found that the initiating enzyme required for GPI biosynthesis, Piga, is highly expressed in the developing corpus callosum, periventricular areas of the CNS and Purkinje cells. We then deleted Piga in the CNS/PNS using a conditional approach with Piga^flox mice and Nestin-Cre to determine the requirement for GPI biosynthesis in this lineage. We found that these mutants do not develop structural defects of the brain as we initially hypothesized. Instead, they developed severe ataxia and tremor, failed to thrive and died prematurely. As this phenotype resembled other mouse mutants with defects in myelination and IGD patients develop defects in developmental myelination, we sought to determine the degree of myelination in the mutants. We found by αMBP staining that mosaic cKO mutants have severe defects in developmental myelination compared to WT littermates confirming a requirement for GPI biosynthesis in myelination. We also found severe defects in motor coordination and striking defects in Purkinje cell arborization in the mosaic cKO cerebellum. Similar ataxic phenotypes and reduced lifespan have been observed in other models of CDG including the genetic reductions of Pmm2, Cog7 and Atp6ap2 in Drosophila models, demonstrating that motor coordination critically requires normal glycosylation through phylogeny (33). We found that gene expression was significantly different in mosaic cKO mice compared to controls with an overexpression of tyrosine hydroxylase, a marker of premature cerebellar development. These data illuminate a novel role for the GPI anchor posttranslational modification in the mouse CNS and provides mechanistic insight into the pathophysiology of IGD.

We were interested to find that the expression pattern of Piga in the postnatal CNS suggested a role for GPI biosynthesis in white matter development. Piga was strongly expressed in the corpus callosum, a highly myelinated structure. Myelination is critical to protect axons, increase conduction velocity along the axon and allow communication between neurons in the CNS especially for motor coordination and learning. Classic models of hypomyelination such as the rumpshaker and quaker mouse mutants display similar phenotypes as those observed in the mosaic Piga cKO mice including truncal ataxia, tremor and early death (34). The germline mutants of GPI-anchored contactin family members display a very similar phenotype to that observed in the Piga mosaic cKO mice including ataxia, tremor, small body size and premature death around P20 (27,28). Recently, it was shown that contactin 1 is expressed in oligodendrocytes and is critical for normal myelination (27). We hypothesize that deficiency in the GPI-anchoring of CONTACTIN 1 contributes to the hypomyelination phenotype we observe in the Piga mosaic cKO mice. While CONTACTIN1 is a promising candidate, there are several other GPI-APs in the CNS that regulate many related functions including synaptic plasticity, axon outgrowth and regeneration (25). We hypothesize that the phenotype observed in the Piga mosaic cKO mice highlights the critical early requirements for GPI biosynthesis in the postnatal brain. Indeed, the mosaic cKO mice are moribund by P21 consistent with the lethality seen in a variety of hypomyelination mutants. In contrast, the hemizygous cKO males die even before myelination has really begun, around P10, arguing that GPI biosynthesis is critical for other CNS functions earlier in development. However, we were unable to identify the cause of death in the hemizygous cKO mice and this function of GPI biosynthesis remains unclear.

The most common phenotypes observed in all IGD patients are intellectual disability/developmental delay and epilepsy. We noticed several episodes of mosaic mutants flailing their hindlimbs and forelimbs while on their backs after they fell due to their ataxic gait. They then experienced a sustained period of rigid paralysis with limbs outstretched followed by a short period of inactivity (data not shown). These events may be consistent with a seizure, but the early lethality in our mutants precludes a rigorous analysis of the ‘seizure phenotype’ in the mutants by EEG monitoring. These phenotypes are consistent with the ‘severe’ presentation of Piga deficiency in which patients die earlier and display severely delayed myelination with thin corpus callosum compared to more ‘mild’ forms of Piga deficiency (12,35). Further research with a less severe model of GPI biosynthesis deficiency may prove to be a better model to study the seizure phenotype as observed in IGD patients. Knock-in of patient variants with the ‘moderate phenotype’ by CRISPR-Cas9 technology or Piga deletion with more restricted Cre transgenes may achieve a more moderate defect in GPI biosynthesis and provide a longer-lived model of IGD with the seizure phenotype.

It has been known for decades that somatic mutations in Piga in the hematopoietic lineage result in a hemolytic anemia called paroxysmal nocturnal hemoglobinuria (PNH). PNH red blood cells (RBCs) are missing two critical negative regulators of the complement cascade, the GPI-APs CD55 and CD59. Without GPI biosynthesis, RBCs become the target of aberrant complement activation leading to life threatening hemolytic anemia. Recent research has shown that a clinically available complement inhibitor can almost completely halt the PNH disease process (36,37). Others have hypothesized that the GPI-APs CD59/CD55 may also be critical to protect neurons from complement in the CNS. Indeed, one group derived neurons from Piga-deficient patient-induced pluripotent stem cells and showed that these neurons are more susceptible to complement-mediated lysis in culture (38). We identified a mild enrichment for immune activation genes in the Piga mosaic cKO cerebellum, but they were not among the top hits from GO. The most upregulated genes in mosaic cKO mutants include myelin protein zero, tyrosine hydroxylase and perilipin 4. MYELIN PROTEIN ZERO is the major constituent of peripheral myelin in the peripheral nervous system, and it is unclear why this gene was overexpressed in our samples. PERILIPIN 4 coats intracellular lipid droplets. Perilipin 4 overexpression in the mutants may suggest that there is storage of lipid droplets in the cells of the mutant. Perhaps, defects in GPI biosynthesis result in an accumulation of lipid precursors leading to lipid droplet inclusions in cells.

The strongest signal from our RNA-Seq experiment by adjusted P-value was tyrosine hydroxylase (Th). Interestingly, though Purkinje cells are not dopaminergic, it has been shown that WT Purkinje cells go through a short phase of Th expression during development, which decreases by P19. In a variety of ataxic mutants including the pogo, Lrp5/6, β-catenin cKO and dilute mutants, Purkinje cells abnormally retain Th expression to later postnatal stages (39–42). This retained Th expression is thought to mark a delay in Purkinje cell maturation, though the exact role of Th in this process is unclear.

Our RNA sequencing results also suggest a role for hypoxia in the pathology of the cerebellar defect observed in the Piga mosaic cKO mice. Hypoxia has been shown to delay the maturation of the cerebellum. How GPI deficiency could result in hypoxia remains unclear. Alterations in lipid metabolism due to the blockade in GPI biosynthesis could allow accumulation of lipid precursors including phosphatidylinositol (PI). If PI were to accumulate, then subsequent peroxidation may lead to the generation of reactive oxygen species and a resulting hypoxic response. This remains to be tested. Alternatively, vasculature development may be impaired in the Piga mosaic cKO cerebellum leading to regional hypoxia. Further evaluation of vascular development of the Piga mosaic cKO would help define this defect.

We propose that our conditional knockout model may serve as an excellent model for preclinical trials of drugs for IGD as the phenotype is robust and quantifiable, and the mosaic cKO survives long enough postnatally to be treated with experimental compounds. The degree of myelination as measured by immunohistochemistry and ataxia as tested with the rotarod serve as convenient end points to examine in experimental settings of preclinical drug trials. A synthetic intermediate of GlcNac-PI would be one promising candidate to test in this model as this intermediate would provide the necessary precursor for GPI biosynthesis missing in Piga mutants and possibly rescue GPI expression and the phenotype.

Materials and Methods

Animal husbandry

All animals were maintained through a protocol approved by the Cincinnati Children’s Hospital Medical Center IACUC committee (IACUC2016-0098). Mice were housed in a vivarium with a 12 h light cycle with food and water ad libitum. Piga^flox (B6.129-Piga^tm1, #RBRC06211) mice were previously generated by Taroh Kinoshita and Junji Takeda and obtained from RIKEN (18). Mice were genotyped for the Piga^flox allele using Riken’s three primer protocol. B6.Cg-Tg(Nes-cre)1Kln/J Nestin-Cre mice (Jackson labs #003771) were genotyped using Jackson Lab’s recommended general Cre genotyping. Progeny of the Piga^flox x Nestin-Cre mice were sacrificed when they became moribund (immobile) at approximately P20. Primers used for genotyping are available in Supplementary Material, Table S1.

In situ hybridization

Whole E11.5 embryos were fixed overnight in 4% PFA at 4°C and dehydrated through a methanol series. Samples were treated with 4.5 μg/ml Proteinase K for 7–13 min at room temperature, post-fixed in 4% PFA/0.2% glutaraldehyde and blocked with hybridization buffer prior to hybridization overnight with DIG-labeled in situ probes at 65°C with constant agitation. The samples were washed and incubated with an anti-DIG antibody (Roche #11093274910) o/n at 4°C. Embryos were washed and incubated with BM Purple (Roche #11442074001) from 4 h at room temperature to o/n at 4°C.

Piga plasmid was obtained from Origene (Rockville, MD, #MR222212). Antisense probes were generated from PCR products containing T3 overhangs. A Piga antisense probe was generated from 910 base pair PCR product. Primers are listed in Supplementary Material, Table S1. The PCR products were purified, in vitro transcription was performed with digoxigenin-labeled dUTP (Roche #11277073910) and the probe was purified with the MEGAclear Transcription Clean-Up Kit (Thermo #AM1908) per the manufacturer’s instructions. For sense probes, the plasmids were cut with XhoI restriction enzyme after the coding sequence and T7 RNA polymerase was used for in vitro transcription.

PCR analysis of specific brain regions

Mice were euthanized, and tissues were immediately dissected from a variety of brain regions including the cortex, cerebellum and subcortical areas. Tail biopsies were taken as controls. Tissue was lysed in 50 mM sodium hydroxide and boiled on a hot plate for 15 min. The samples were then neutralized with 1 M Tris and centrifuged. PCR analysis was performed with 60°C annealing temperature and 34 cycles of amplification for Piga flox allele genotyping and a program with 64°C annealing temperature with 34 cycles of amplification for Cre genotyping. Primers are available in Supplementary Material, Table S1.

Histology

Brains were dissected and fixed in 10% formalin for 24–48 h, washed in 70% ethanol and paraffin embedded by the CCHMC Pathology Core. Brains were sectioned by microtome at 10–20 μm and stained with hematoxylin & eosin using standard methods.

Weight

Mice used for weight test were weighed individually every day using a standard metric balance from P1 to P21.

Tail suspension test

P15-P18 mice were held by the tail for 10 s in the air, and their hindlimb posture was observed. The mice were given a score from 0 to 4. A score of 4 indicates the normal hindlimb separation in which they are widely spread when suspended by the tail. Score of 3 means the hindlimbs are more vertical though they barely touch. Score of 2 means the hindlimbs are close and often touching, and 1 indicates profound weakness in which both hindlimbs are almost always clasped together. Score of 0 is given only if the hindlimbs are clasped for the entire 10 s the mouse is suspended (43).

Righting reflex

P7–8 mice were placed supine, and the time to flip over onto their abdomens was measured in seconds and recorded as a latency to right themselves to a prone position (43). Three trials were performed consecutively, and each trial time was plotted.

Forelimb wire hang

Mice were placed on a thin wire suspended by their forelimbs over a padded drop zone. The pups were placed on the wire such that the experimenter could observe their grip on the wire. The pups were then released and the time before they fell off the wire was measured in seconds and recorded as a latency to fall (43). Three trials were performed in sequence and each trial latency to fall was plotted.

Rotarod

Rotarod was performed by the CCHMC Animal Behavior Core (44). Researchers were blinded to the genotypes of the animals tested. Rotarod apparatus (San Diego Instruments) was used with SDI software with the 4 to 40 RPM (revolutions per minute) protocol for mice. The apparatus has four chambers, and four mice were tested at the same time. The photobeam sensor detects and records the time and distance each mouse travels before falling from the rod. The test was allowed to run for 360 s. The test starts at 4 RPM for 30 s and increases to 16 RPM for 110 s, then 28 RPM for 110 s, and finally 40 RPM for 110 s. The latency to fall in seconds and distance traveled was recorded automatically and the mice were allowed to rest for 15 min before each successive trial. Four trials were performed sequentially on each mouse.

Immunohistochemistry

Immunohistochemistry was performed on formalin-fixed, paraffin embedded brain tissue harvested from P19 animals. Tissue was sectioned at 10–20 μm; sections were blocked for 1 h at room temperature in 4% normal goat serum in PBST and incubated in primary antibody (1:500 chicken anti-mouse Myelin Basic Protein antibody, Aves Inc, #MBP) overnight at 4°C. The next day, slides were washed in PBS and incubated in 1:500 biotinylated goat anti-chicken antibody (Aves Inc, #B-1005) for 1 h at room temperature. The slides were washed and incubated in ABC mix (Vectastain ABC HRP Kit, #PK-4000) for 1 h at room temperature. The slides were washed and developed in 0.5 mg/ml DAB (Sigma) activated with 30% hydrogen peroxide. Slides were incubated in DAB for approximately 5 min, washed in PBS, sealed with Cytoseal and imaged by light microscopy.

Immunofluorescence

P0-P23 mice were euthanized with isoflurane and cervical dislocation and their brains were microdissected in PBS. They were fixed in 4% PFA o/n, equilibrated in 30% sucrose o/n, cryo-embedded in OCT and sectioned from 10 to 20 μM with cryostat. Antigen retrieval was performed with citrate retrieval buffer, blocked in 4% normal goat serum, incubated in primary antibodies o/n at 4°C: rabbit anti-PIGA (Proteintech #13679-1-AP, 1:200), mouse anti-NeuN (Millipore #MAB377, 1:1000), Rat anti-CD68 (Biorad #MCA1957T, 1:1000), rabbit anti-Calbindin (abcam #ab25085, 1:4000) and Rat anti-Mylein Basic Protein (Aves #MBP, 1:500). Sections were incubated with Alexafluor 488-congugated goat anti-rabbit (Thermo #A11008, 1:1000) and Alexafluor 594 conjugated goat anti-mouse (Thermo #A11008, 1:1000) secondary antibodies and counterstained with DAPI. Sections were imaged on Nikon C2 confocal 703 microscope.

Western immunoblotting

Brains were microdissected, and the cortex and cerebellum were isolated. Subcortical tissue was lysed in 800 μl RIPA buffer+ Protease inhibitor. Lysate protein concentration was determined by BCA assay, and electrophoresis was performed on a 10% Tris-glycine gel. Protein was transferred to a PVDF membrane, blocked in Odyssey blocking buffer and incubated o/n at 4°C with Rabbit anti-PIGA (Proteintech #13679-1-AP, 1:1000) and Mouse anti-Tubulin (Sigma #T6199, 1:1000) antibodies. Membranes were washed and incubated for 1 hour in goat anti-rabbit IRDye 800CW (LICOR # 926–32 211, 1:15000) and goat anti-mouse IRDye 680Rd (LICOR, #926-68 070, 1:15000) antibodies and visualized on a LICOR Odyssey imaging system. Relative protein concentration was determined by normalizing PIGA signal to Tubulin signal in Image Studio Lite Ver 5.2.

RNA sequencing

3 WT and 3 mosaic cKO cerebella were bisected along the midline of the cerebellar vermis and the right half was snap frozen on dry ice. RNA was isolated and pooled samples of each genotype (i.e one pooled ‘wild type’ and one pooled ‘mutant’ sample were each used for paired end bulk RNA sequencing (BGI Americas, Cambridge, MA). mRNA molecules were purified from total RNA using oligo(dT) attached magnetic beads. mRNA molecules were fragmented into small pieces using fragmentation reagent and first strand cDNA was generated using random hexamer primed reverse transcription, followed by a second strand cDNA synthesis. The synthesized cDNA was subjected to end repair and 3′ adenylated. Adapters were ligated to the ends of these 3′ adenylated cDNA fragments. PCR was used to amplify the cDNA fragments with adapters from previous step. PCR products were purified with Ampure XP Beads (AGENCOURT) and dissolved in EB solution. Library was validated on the Agilent Technologies 2100 bioanalyzer. The double stranded PCR products were heat denatured and circularized by the splint oligo sequence. The single strand circle DNA (ssCir-DNA) were formatted as the final library. The library was amplified with phi29 to make the DNA nanoball (DNB) which had more than 300 copies of one molecule. The DNBs were load into the patterned nanoarray and single end 50(pair end 100/150) bases reads were generated in the way of combinatorial Probe Anchor Synthesis (cPAS). Analysis was performed by BGI RNA Sequencing services which includes a proprietary analysis pipeline.

Statistical analysis

Statistical analysis was performed using Graphpad Prism (GraphPad Software, San Diego, CA). Survival analysis was performed with a log-rank (Mantel-Cox) test. Rotarod analysis was performed with a one-way ANOVA followed by unpaired t-tests. All other tests were unpaired t-tests with the exception of Westerns from hemizygous cKO animals in Fig. 2C and D which were analyzed with paired t-tests.

Conflict of Interest Statement

The authors declare no conflict of interest.