Experiment
Hydrocephalus is a devastating neurological condition, characterized by the abnormal accumulation of cerebrospinal fluid (CSF) within the cerebral ventricles. Humans with L1 cell adhesion molecule (L1CAM) mutations exhibit X-linked hydrocephalus, as well as other severe neurological disorders. L1-6D mutant mice, which are homozygous for a deletion that removes the sixth immunoglobulin-like domain of L1cam, seldom display hydrocephalus on the 129/Sv background. However, the same L1-6D mutation produces severe hydrocephalus on the C57BL/6J background.
To begin to understand how L1cam deficiencies result in hydrocephalus and to identify modifier loci that contribute to X-linked hydrocephalus by genetically interacting with L1cam, the authors conducted a genome-wide scan on 92 (55 male, 37 female) F2 L1-6D mice, bred from L1-6D 129S2/SvPasCrlf and C57BL/6J mice (129S2.Cg-L1camtm1.1Vlem and B6.Cg-L1camtm1.1Vlem).
The hydrocephalic phenotype of F2 L1-6D mice utilized for linkage analyses was quantified as follows. For each mouse, we chose the imaged section that showed the greatest expansion of the lateral ventricles. The positions of these sections corresponded to reference sections positioned within +1.0 mm from Bregma [61]. Next, the perimeter of the coronal section and its respective lateral ventricles were manually traced with ImageJ software. ImageJ was then further utilized to calculate the areas of the traced regions. This generated three measurements: coronal section area, left ventricle area, and right ventricle area. Left and right ventricle areas were added together to calculate total lateral ventricular area. To account for differences in brain size and standardize measurements, a ratio was calculated for each brain: total lateral ventricular area divided by coronal section area. Ratios were then transformed, using a natural log scale [ln (1,000 x ratio)], to approximate a normal distribution for statistical analyses. The authors refer to these transformed ratios as hydrocephalus severity scores. Student's t tests were performed to determine whether there were significant differences in hydrocephalus severity between the different cohorts, as well as between the two sexes.
L1-6D genotyping was performed using PCR amplification and the following primer set: forward 5- CCAGCCAGGATCCTAACAAAAGAC, reverse wildtype allele 5- AGT GATGCTGGCCTGCAAAG, and reverse knock-in allele 5- AACCACACTGCTCGACCTG. The genome-wide scan was performed at the Affymetrix GeneChip laboratory of the Yale Microarray Center for Research on the Nervous System, through the NIH Neuroscience Microarray Consortium. Illumina's Mouse Low Density Linkage panel and Golden Gate Assay were used to genotype 92 F2 L1-6D homo-/hemizygote pups with hydrocephalus at 377 SNPs; 99% of the genotypes were present. Genotypes were classified as 129S2 or C57BL/6J, and SNP performance was evaluated by the Mouse SNP Genotyping Service at Harvard University. The genotypes of the 129S2 mice matched that of 129S1 mice at all 375 working SNPs except one (rs13481099), which appeared monomorphic between the 129S2 and C57BL/6J mice. Ultimately, 246 informative SNPs were polymorphic between 129S2 and C57BL/6J mice; the average genomic distance between these SNPs was 9.9 Mb.
Linkage analysis was performed using a total of 288 SNPs (272 autosomal and 16 X-linked). Mice with hydrocephalus and those without were analyzed separately. In addition, sex was used as a covariate. Data from male and female mice were analyzed separately, as well as combined and analyzed together. Progeny 6.9.04 software was used to visualize haplotypes. R/qtl 1.07-12 software was used for QTL analyses to test for linkage to severity modifier loci. Single-locus and two-QTL, two-dimensional genome-wide scans, using a multiple imputation method with 2.5-Mb steps over the entire genome and 200 imputations, were performed utilizing 229 autosomal genome-wide markers.
QTL analysis showed no main-effect loci linked to hydrocephalus severity. The authors found a linkage peak on chromosome 10 at 119 Mb (mCV25429984) with a LOD score of 2.9. However, a permutation test of the single-locus scan did not indicate significant evidence of linkage at this locus. A two-loci scan was then performed. Three pairs of loci had overall LOD scores above 6 (full LOD) and interaction LOD scores, which measure the statistical interaction between loci, above 3.5 (all genome coordinates relative to NCBI Build 37 (mm9)):
Chromosome 4 at 153.7 Mb and Chromosome 9 at 112.2 Mb
Chromosome 5 at 109.2 Mb and Chromosome 9 at 102.2 Mb
Chromosome 7 at 101.6 Mb and Chromosome 9 at 87.2 Mb
However, again, the permutation test revealed no significant results for this two-QTL scan. Furthermore, no QTL were detected in males nor females when data were segregated by sex.
Analyses of the X chromosome required the authors to take into consideration that 129/Sv DNA was originally utilized to generate the L1-6D knock-in allele. Consequently, all L1-6D C57BL/6J mice carry a 129 congenic interval, which includes the L1cam locus (positioned at 71.1 Mb). Sequencing revealed that this L1-6D associated 129 region is <25 Mb, and it lies between SNPs rs24884396 and rs3166693 (genome coordinates, 46.6-71.5 Mb).
A genome-wide QTL scan was performed, including X-chromosome markers (14 X-chromosome markers + 229 autosomal markers). The L1-6D associated 129 region was excluded because the 129/B6 genotype of parental females complicated the analysis. Furthermore, only 39 F2 L1-6D hemi-/homozygous mutants derived from F1 L1-6D/L1-6D females were used for the X-chromosome QTL study. F2 L1-6D mutants derived from F1 L1-6D/+ females were not used because their siblings, non-hydrocephalic wild-type males and L1-6D/+ females, were not genotyped with the genome-wide linkage panel. Since these F2 wild-type males and L1-6D/+ females would have inherited their X chromosomes from the B6 parental strain, the missing data would confound analysis. Single- and two-loci scans did not identify any X-chromosome markers linked to hydrocephalus severity.
An extension study was next conducted to determine whether stronger linkage evidence could be obtained for the candidate modifier loci identified through the genome-wide scan. Seven genomic regions were analyzed in total. Two of these regions, located on chromosomes 4 and 5, were originally identified by both chi-square analysis and two-loci QTL scan. For the extension study, 43 informative SNPs, dispersed throughout the candidate modifier regions, were used to genotype two new groups of mice: (1) a control set of 30 F2 L1-6D hemi/homozygous mutant siblings without hydrocephalus (unaffecteds) and (2) 64 additional F2 L1-6D hemi-/homozygous mutant mice with hydrocephalus (affecteds).
Combined analysis of all 156 F2 L1-6D mutants with hydrocephalus (64 from the extension study and 92 from the original genome-wide scan) resulted in the significant linkage of the eight extension study SNPs. All markers passed Bonferroni correction (alpha = 0.005), and the lowest p value was exhibited by rs3694887 (104.2 Mb), p = 4.04 x 10^11. Linkage was confirmed for the single locus:
L1hydro1 (L1cam hydrocephalus modifier 1) maps to Chr 5: 100 - 129.2 Mb with a peak p value of 4.04 x 10^11 at 104.2 Mb (rs3694887). Mice with a B6/B6 genotype at marker rs3654076, located within L1hydro1, tended to have more severe hydrocephalus than mice with the 129/B6 or 129/129 genotype.
No single genotype in the L1hydro1 region was shared among all hydrocephalic mice, and the genotype frequencies of F2 L1-6D mutant mice without hydrocephalus did not deviate from Mendelian segregation at L1hydro1, suggesting that L1hydro1 exhibits incomplete penetrance.
Analysis Tools