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.

Summary A recent convergence of technological innovations has re‐energized the ability to apply genetics to research in human craniofacial development. Next‐generation exome and whole genome sequencing have significantly dropped in price, making it relatively trivial to sequence and analyze patients and families with congenital craniofacial anomalies. A concurrent revolution in genome editing with the use of the CRISPR‐Cas9 system enables the rapid generation of animal models, including mouse, which can precisely recapitulate human variants. Here, we summarize the choices currently available to the research community. We illustrate this approach with the study of a family with a novel craniofacial syndrome with dominant inheritance pattern. The genomic analysis suggested a causal variant in AMOTL1 which we modeled in mice. We also made a novel deletion allele of Amotl1. Our results indicate that Amotl1 is not required in the mouse for survival to weaning. Mice carrying the variant identified in the human sequencing studies, however, do not survive to weaning in normal ratios. The cause of death is not understood for these mice complicating our conclusions about the pathogenicity in the index patient. Thus, we highlight some of the powerful opportunities and confounding factors confronting current craniofacial genetic research.

Forward genetics in the mouse continues to be a useful and unbiased approach to identifying new genes and alleles with previously unappreciated roles in mammalian development and disease. Here, we report a new mouse allele of Cse1l that was recovered from an ENU mutagenesis screen. Embryos homozygous for the anteater allele of Cse1l display a number of variable phenotypes, with craniofacial and ocular malformations being the most obvious. We provide evidence that Cse1l is the causal gene through complementation with a novel null allele of Cse1l generated by CRISPR-Cas9 editing. While the variability in the anteater phenotype was high enough to preclude a detailed molecular analysis, we demonstrate a very penetrant reduction in Pax6 levels in the developing eye along with significant ocular developmental phenotypes. The eye gene discovery tool iSyTE shows Cse1l to be significantly expressed in the lens from early eye development stages in embryos through adulthood. Cse1l has not previously been shown to be required for organogenesis as homozygosity for a null allele results in very early lethality. Future detailed studies of Cse1l function in craniofacial and neural development will be best served with a conditional allele to circumvent the variable phenotypes we report here. We suggest that human next-generation (whole genome or exome) sequencing studies yielding variants of unknown significance in CSE1L could consider these findings as part of variant analysis.

No abstract