Experiment
The objective of the current study was to identify genetic modifiers of collagen deposition in the mouse heart.
Transgenic 129P2-Scn5atm1Care/+ mice were crossed with FVB/NJ-Scn5atm1Care/+ transgenic mice to generate F1 progeny that were subsequently intercrossed to generate disease-sensitized F2 progeny. All F2 mice were genotyped for the Scn5atm1Care mutation; 65 mice heterozygous for the mutation were used.
For the genome wide scan the F2 mice and 3 mice from each parental strain were genotyped across the 19 autosomes and the X chromosome using an Illumina Golden Gate mouse medium density (768 SNP) panel. 587 SNPs of the panel were informative in the F2 mice. Mice were euthanized at 12 to 14 weeks. Protein was isolated from the left ventricular (LV) tissue using standard procedures. Genome wide (multiple) QTL mapping for the amount of collagen relative to protein in LV tissue was performed using the R/qtl package in the statistical programming language R. Significance levels for the LOD scores were determined based on permutation FDR<0.05, resulting in a LOD score threshold of 2.9 for significant collagen QTL.
Genome wide (multiple) eQTL mapping was performed using the R/qtl package in the statistical programming language R. The empirical LOD threshold of 6.6 was determined using 100,000 permutations. The expression levels of transcripts for identified eQTL colocalizing with the collagen QTL were tested for nominal correlation (p<0.05) with collagen levels.
No difference in the levels of collagen was observed between the two founder strains. However, the (129P2-Scn5atm1Care/+ x FVB/NJ-Scn5atm1Care/+)F2 mice showed a large spread in relative collagen, suggesting the existence of segregating alleles affecting the trait in the founder strains. The relative amount of collagen was used together with previously generated genotypic data from the same F2 mice as input for QTL mapping.
Table 1:
A significant, dominant QTL, Carcdq1 (cardiac collagen deposition QTL 1) was identified mapping to Chromosome 8 with peak marker rs3672639 (50.2 cM) with a LOD score of 4.1. Carcdq1 mapped between flanking markers rs13479716 (42.2 cM) and rs4137596 (63.8 cM). The FVB/NJ allele was associated with a lower level of cardiac collagen [Fig 3A].
Using multiple QTL mapping a significant covariate QTL Carcdq2 (cardiac collagen deposition QTL 2) was identified mapping to Chromosome 18 with peak marker rs3720827 (32.4 cM) with a LOD score of 2.5. Carcdq2 mapped between flanking markers rs13483183 (0.0 cM) and mCV24836796 (52.1 cM). Homozygotes for the FVB/NJ allele had lower level of cardiac collagen [Fig 3A].
Adding Carcdq2 as a covariate in QTL mapping uncovered an additional QTL Carcdq3 (cardiac collagen deposition QTL 3) mapping to Chromosome 2 with peak marker rs3713997 (0.0) with a LOD score of 3.0. Carcdq3 mapped between flanking markers rs3713997 (0.0 cM) and rs13476399 (22.6 cM). The FVB/NJ allele was associated with a lower level of cardiac collagen [Fig 3A].
As genetic variation underlying a QTL may act through effects on gene expression, the 1.5 LOD drop region around the identified QTL was assessed for overlapping eQTL. The eQTL were identified based on determination of gene transcript abundance in LV tissue from 109 of the F2 mice. Twenty four significant eQTL were identified within the Carcdq1 1.5 LOD interval (Table 2). Seven significant eQTL were identified within the Carcdq3 1.5 LOD interval (Table 3) and 12 were identified within the Carcdq2 interval (Table 4).
As causative transcripts are expected to correlate with the amount of collagen, a Spearman correlation test was performed on the identified transcripts. Of the 31 transcripts with significant eQTL mapping to the Carcdq1 and Carcdq3 loci, five (Pdlim3, Tmem66, Rad23A, Lsm6m and Tnks) and one (Gpr158) transcripts, respectively, correlated with the relative amount of collagen. Of the 12 transcripts with eQTL underlying the Carcdq2 locus the levels of Fgf1 showed a suggestive correlation with collagen.
Analysis Tools