Hydrocephalus is categorized as congenital when present at birth and is characterized by abnormalities in cerebrospinal fluid flow and reabsorption. Congenital hydrocephalus (CH) in human populations is a common congenital malformation, with an incidence of 0.5–1.8 per 1,000 births. To date, nine genes have been associated within model organisms such as mouse, rat, and zebra fish. Through linkage analysis of human families with CH, variants in the gene L1CAM were discovered to be responsible for X-linked hydrocephalus. Linkage analysis was also used in an autosomal recessive CH, which revealed causative variants in the gene MPDZ. These mutations only account for 7–15% of reported CH in humans. To better understand this disease pathway, a recently acquired severe CH in Oriental Shorthairs is being examined. An Oriental Shorthair male with atypical small ears was adopted into a breeding program to create cats with low set and rounded ears. Subsequent breeding showed the rounded heads and small ears to be autosomal recessive; however, other subtle traits were noticed, such as abnormal gait and clumsiness (Figure 1). Upon further examination of an affected kitten, these cats were determined to have CH. Quick artificial selection, adopted in the development for desirable trait, small ears, has also kept an undesirable deleterious CH trait in the population.

Figure 1. Affected Oriental Shorthair siblings Affected cats with desired small rounded ears and larger rounded skull due to hydrocephalic status.

A pedigree of the cats segregating for CH was created and confirmed using parentage analysis of 20 STRs and 38 SNPs with the program COLONY. Hydrocephalus status was determined by T2 weighted MRI images and coronal sections of fixed brain (Figure 2). Fifty-one cats from the pedigree were genotyped on the Illumina Infinium Feline 63K iSelect DNA array and a GWAS was performed using the program PLINK. After genotype pruning of the 62,897 SNPs on the array, nine individuals were removed for low genotyping rate (MIND > 0.1) leaving 22 cases and 20 controls in the analysis. Four hundred seventy-eight SNPs failed the missingness test (GENO > 0.2) and 23,599 SNPs failed MAF < 0.05 scan. An extremely significant association was identified on chromosome A3. Cat chromosome A3 is homologous to human chromosome 20 and portions of human chromosome 2. Haplotype analysis was conducted from SNP A3.121152112 to SNP A3.147637938 (~ 20 Mb), which revealed a unique haplotype across all cases spanning ~ 9 Mb. Visual screening of the region identified 64 genes; none have been reported to be associated with hydrocephalus or brain development and none of the known candidate genes from other species are within the haplotype. Lacking a clear candidate gene, a trio of cats was selected for whole genome sequencing.

Figure 2. An unaffected, a moderately affected, and a severely affected cat coronal sections and MRI images a. A unaffected cat with normal ventricles. b. A moderately affected cat with widening of the lateral and third ventricles. c. A severely affected cat with widening of the lateral and third ventricles and thinning of the cerebral cortices.

An affected sire, a carrier dam, and an affected offspring were whole-genome sequenced using two PCR-free libraries (350 bp and 550 bp) per cat. Approximately 30x coverage of Illumina HiSeq 100 bp paired-end reads have been produced per cat. A company, Maverix Biomics, provided variant calling using the algorithm FreeBayes, and alignment to the cat Ver6.2 genome assembly (9) on a UCSC-type browser. A variant report was produced for the 9-Mb haplotype region, which included variant positions, frequency, base change, and impact of the change. Three hundred and twenty-five variants (synonymous, nonsynonymous, deletions, frameshifts, UTR, and splice site variants) were detected in the region, 141 variants segregated concordantly with the phenotype in the sequenced trio. After the inclusion of the sequence from the additional twelve control cats, only nine exonic variants remained concordant with the disease phenotype (1 non-synonymous, 1 start gain, and 7 UTR variants). Two of the variants (start-gain in HADHA and non-synonymous in GPR113) have been genotyped by direct Sanger sequencing in 20 cats; the results were concordant in all affected and carrier individuals and could not be eliminated as possible causative variants.

Thirty-four SNPs were selected from within the 9-Mb region of association (position range on chromosome A3: 121152112 to 130285890) and were designed to run on the Sequenom Mass Array 4.0 using the Sequenom software assay design suite. The nine SNPs detected in coding regions were included in the assay. The remaining 25 SNPs were selected using the following criteria: inter-snp average distance of 260,000 bp within the 9-Mb region to ensure even coverage of the haplotype, high quality score and call rate. The Sequenom assay was run on 96 samples, 79 related cats from the pedigree and 17 unrelated oriental shorthair cats. Of 34 SNPs, four were concordant in the 96 tested cats, 2 SNPs failed genotyping. Mass Array preliminary results reduced the haplotype to a 154-kb critical region. A de novo assembly of the affected sire from the WGS trio is in progress to evaluate the presence of structural variants.