Fig 1.
Affected cats have marked microcephaly with polymicrogyria.
Images of whole brain from (A) adult unaffected (+/+), and (B) adult and (C) juvenile affected (-/-) cats. Affected cats have dramatically decreased cerebral cortex size with normal formation of the cerebellum. (D) Brain weights of affected cats are significantly decreased with or without normalization to body weight, which is similar to unaffected cats. (E) Representative sections (left image, normal; right image, affected) from the region of the parietal cortex have gyrification defects characterized by shallow sulci and fusion of small gyri consistent with polymicrogyria, as well as abnormal white matter of the corona radiata and internal capsule.
Fig 2.
Changes in MRI are consistent with microcephaly and attenuation of gyral formation.
(A) Selected images from the frontoparietal region demonstrate marked attenuation and loss of gyral formation and white matter. Note the blurring of gray and white matter boundaries, especially apparent in the corona radiata. (B) Magnified region within the white box highlights the severe attenuation of the (C) anterior cingulate gyrus, outlined in white.
Fig 3.
Zygosity mapping, linkage, and haplotype analysis identifies a frameshift in PEA15 as the cause of cerebral dysgenesis.
(A) Zygosity mapping, identifying all variants that are homozygous in 8 affected animals and heterozygous in 6 obligate carriers. Variants cluster in a region on the distal end of chromosome F1. (B) Diplotypes of 49 cats according to disease status as indicated. 13 diplotypes were imputed from progeny: the top 4 diplotypes are founders (note uncertain haplotype, denoted by X’s), the next 5 diplotypes are for the next generation after the founders, and 4 other cats throughout the pedigree were imputed because a sample was not available. In the unaffected genotyped block, the cat indicated with a › is a cat that has 2 normal diplotypes but is present in the analysis because it was bred with a cat homozygous for the disease diplotype. All affected animals are homozygous for a 1.3 Mb region (dashed black lines) (C) Linkage analysis confirms that the 1.3 Mb region on chromosome F1 identified by zygosity and haplotype analysis cosegregates with cerebral dysgenesis (coordinates given according to Felcat8). (D) CADD scores for all variants in the 1.3 Mb critical region that are homozygous in affected cats and at less than 5% allele frequency in the 99 Lives dataset. Only nine coding variants are present (see detail in Table 1). Six of the coding variants are synonymous. Two variants are missense in LY9 and CD48 (neither of which is expressed in brain). The synonymous variant in SLAMF1 is listed in Table 1 but not plotted here because its CADD score is inflated as it is missense in human, but synonymous in cat. Two other coding variants are also not plotted because they did not lift over to human, but are presented in Table 1. Finally, the two coding variants in the lower left corner are nearby variants in LY9 that cannot be distinguished on this plot because they both have a CADD score near 0. The final coding variant in the region is a frameshift in PEA15, which is highly expressed in brain. The variant is predicted to be highly damaging by CADD.
Table 1.
Key sequence variants (coding or CADD>15) in the 1.3 Mb critical interval.
The table contains all coding variants in the interval with an allele frequency (AF) < 0.05 in the 99Lives dataset, and also contains one non-coding variant that had the next highest CADD score after the PEA15 variant.
Fig 4.
The PEA15 pathogenic variant introduces a premature termination codon, and PEA15 protein is absent in affected cats.
(A) Map of PEA15 demonstrating the pathogenic variant location near the beginning of Exon 2. (B) Overall levels of PEA15 transcripts measured by RNA-Seq are decreased in cats homozygous for the PEA15 pathogenic variant (One-way ANOVA *p<0.0001, *p<0.01 by Tukey’s post hoc). (C) Reads from the mutant PEA15 allele in heterozygous cats are significantly reduced compared to non-mutant reads, while heterozygous variants in nearby genes do not exhibit allele bias (One-way ANOVA *p<0.0001, *p<0.001 by Tukey’s post hoc vs. all of 3 nearby genes comparing the % variant reads per cat as the unit of comparison with 24 to >3,000 reads contributing to each % measurement for each gene). (D) PEA15 is absent from affected animals by western blot.
Fig 5.
Affected cats have a significant loss of white matter.
(A) Subgross sections of MAP2 stained neurons highlight the variable decrease in cortical thickness, and the reduced area of the corona radiata (arrows) and internal capsule. (B) Subgross sections of Luxol fast blue (LFB) stained for myelin indicates decreased white matter, (C) which is confirmed through quantification of LFB stained sections of the frontopareital region.
Fig 6.
Affected cats have loss of normal cerebral cortical layering, increased grey matter astrocytosis, and abnormal neuronal and axonal orientation.
(A) Photomicrographs of Luxol fast blue–Cresyl Echt Violet (LFB-CEV) stained sections depicting vertical columns in the parietal region of unaffected (left) and age-matched affected (right) cats. In unaffected cats, 6 cortical layers extend from below the meninges (beginning with layer I, molecular layer) to the white matter (WM). In affected cats, grey matter thickness and column morphology are altered with disorganized layering. (B) Photomicrographs of GFAP stained sections of vertical columns reveals that affected cats exhibit a relative astrocytosis. Dotted lines indicate separation of white matter and grey matter. (C) Photomicrographs from MAP2 stained sections taken at approximately layers IV and V. Unaffected cats have linear axonal projections oriented perpendicular to the cortical meningeal surface while affected cats lack axonal directionality (bar = 20uM). In all images, the meningeal edge is located at the top.
Fig 7.
Affected cats have significantly increased astrocyte density within the grey matter.
(A-B) GFAP immunohistochemistry (IHC) of grey matter indicates increased density of astrocytes. (C-E) Digital image-analysis algorithms measured a significant increase in GFAP stain density primarily in grey matter stain. (F-J) Olig-2 staining indicates no significant change in the density of oligodendrocytes in grey matter, though an insignificant decrease of ~30% was noted in white matter. (K-L) No significant change in the density or morphology of microglia was detected in affected cats. (M-O) Microglial density findings are confirmed on quantification of IBA-1 stain. (bar = 200um).
Fig 8.
Differential expression analysis.
(A) Log2-fold change vs. magnitude of gene expression for homozygous mutant (M–mutant) (n = 4) vs all unaffected (N&C–normal and carrier) (n = 6) (heterozygous mutant (C–carrier) (n = 3) and homozygous non-mutant (N–normal) (n = 3)). Genes with a significant difference for the strict criteria of significance in both homozygous mutant (n = 4) vs all unaffected (n = 6) and homozygous mutant (n = 4) vs homozygous non-mutant (n = 3) are labeled (triangles). Data was collected for an additional 6 animals, but excluded because of age, cause of death, or principal component analysis results (Methods; S5 Table; S5 Fig). (B) Quantitative changes in collagen gene expression for homozygous mutant, heterozygous, and homozygous non-mutant animals. No differences (p = 0.61, two-way ANOVA) were detected between normal and carrier cats.
Fig 9.
Differences in levels of cell type–specific transcripts suggest altered cellular composition in the brains of affected cats.
(A) Oligodendrocyte precursor cell-specific transcripts are increased in affected cats (Two-way ANOVA *p<0.0001 genotype effect). (B) Oligodendrocyte-specific transcripts are decreased in affected cats (Two-way ANOVA *p<0.0001 genotype effect). (C) Astrocyte-specific transcripts are increased in affected cats (Two-way ANOVA *p<0.0001 genotype effect). (D) Endothelial cell-specific transcripts are increased in affected cats (Two-way ANOVA *p<0.0001 genotype effect). (E) Neuron–specific transcripts did not change significantly (Two-way ANOVA p = 0.60 genotype effect). (F) Microglia–specific transcripts did not change significantly (Two-way ANOVA p = 0.52 genotype effect).
Fig 10.
Differences in cytokine-mediated apoptosis and proliferation in affected fibroblasts.
When treated with TNFα, there is (A) a significant decrease in cell viability and (B) a significant increase in caspase-8 activity of primary dermal fibroblasts from affected cats compared to unaffected cats. (C) When treated with 20ng/mL FGFb, proliferation is significantly increased in primary dermal fibroblasts from affected cats compared to unaffected cats. There was no significant difference of untreated cells for either genotype.
Fig 11.
Hypothesized mechanisms of PEA15 mediated cerebral dysgenesis in domestic cats.
Beginning in late gestation through the early post-natal period of normal animals, there is increased neuronal apoptosis during synaptic pruning. PEA15, which is normally expressed at this time in the brain, protects from excessive apoptosis of neurons and inhibits proliferation of stimulated astrocytes. Therefore, loss of PEA15 is expected to cause increased neuronal apoptosis and increased proliferation of astrocytes. Grey matter astrocytosis may be a direct response to the increased apoptosis or neurons (reactive astrocytosis), and/or and excessive proliferation due to loss of PEA15 function. Abundant astrocytes produce excessive extracellular matrix which can form perineuronal nets and cause a premature end of the critical period for synaptic plasticity. These changes in development result in disorganized axonal development and underdeveloped white matter tracts which manifest as cerebral dysgenesis.