Figure 11.2 Genetic diversity and variation in gene expression in yeast. (A) Data are shown for the gene expression of YLL007C for all 40 segregants (seg) of a cross between two strains BM and MY, for the parental strains, and for segregants inheriting a marker found to be in linkage with YLL007C expression. (B) When the number of linkages was analysed by chromosomal location (based on 20 kb bins) many more linkages were observed at certain bins than expected by chance (no bin would, for example, be expected to contain more than five linkages by chance). Eight of the groups showing more than five linkages are indicated with dashed arrows with the common function of linked expression transcripts indicated. From Brem et al. (2002), reprinted with permission from AAAS.
linkages fell into only eight bins (groups) with evidence of common function for linked gene transcripts at a particular group (Fig. 11.2B).
In a subsequent study from the same group, potential trans-acting effects of genetic variation were explored in more detail using a cross of the same two yeast strains, BY and RM, analysing 86 segregants (Yvert et al. 2003). The strategy here involved hierarchical clustering of genome-wide gene expression data based on similarities between genes in their expression profiles followed by linkage analysis using the mean expression of a cluster as a quantitative phenotype. Many more genes than expected by chance showed significant clustering, with three-quarters involving genes on different chromosomes. Linkage analysis defined significant linkage for 304 clusters containing 1011 genes (ten clusters expected by chance) with the majority of genes and clusters (75% and 80%, respectively) not showing self linkage. This work emphasized the importance of trans-acting variation and allowed positional cloning and molecular analysis to be applied with the definition of specific genetic variants responsible for the observed linkage. It also showed that in yeast, at least, genetic variation involving transcription factors did not account for a significant proportion of trans-acting loci but rather such variation was broadly dispersed across different genes in the yeast genome (Yvert et al. 2003).
Following on from the studies in yeast, Schadt and colleagues demonstrated in mice the utility and application of the genetical genomics approach (Schadt et al. 2003). Here two inbred mouse strains (C57BL/6J and DBA/2J) were crossed, producing an F1 generation who were in turn crossed to produce an F2 generation, among whom genetic variation could be used to map particular phenotypes to a specific genomic location. Gene expression profiles were established using a mouse expression microarray. Schadt and colleagues analysed a total of 111 F2 generation mice using 100 microsatellite markers for 7861 gene expression phenotypes that were significantly different between the two parental strains, or more than 10% of the F2 individuals. eQTLs were defined for 2123 genes with a lod score (Box 2.4) greater than 4.3 (P <0.00005). As found in yeast, local and distant regulatory variation was resolved, probably cis- and trans-acting eQTLs, respectively. Here those eQTLs with the highest lod scores were likely cis-acting.
An example is provided by the eQTL identified at the C5 gene. DBA/2J mice, in contrast to C57BL/6J mice, are known to be deficient in C5 at both the transcript and protein level as a result of a 2 bp deletion in the C5 gene, which leads to a frameshift and premature stop codon (Wetsel et al. 1990; Karp et al. 2000). In this case the eQTL is found at the gene itself (with a very high lod score of 27.4) and has a clearly defined functional mechanism with specific genetic variation (the 2 bp deletion) responsible for the difference in gene expression (Schadt et al. 2003).
Rodent recombinant inbred lines are a very powerful resource for mapping eQTLs, being a mosaic of the two crossed inbred parental strains after repeated sib matings (Fig. 11.3) (Broman 2005). Such studies in mice and rats allowed the definition of multiple cis-acting eQTLs and integration with data available for many other pheno-types on such strains, for example behavioural pheno-types with the eQTLs mapped in mouse brain tissue by Chesler and colleagues (2005). Evidence of tissue-specific effects was also possible, for example between haematopoietic stem cells and brain tissue in the same mice (Bystrykh et al. 2005). These datasets and others are available as part of WebQTL at GeneNetwork (www. genenetwork.org) which seeks to integrate networks of genes, transcripts, and traits for different organisms with biological traits and gene expression data (Wang et al. 2003b; Chesler et al. 2004).
The work of Hubner and colleagues analysing gene expression in recombinant inbred lines from a cross of the spontaneously hypertensive rat and the normo-tensive Brown Norway rat strains demonstrated more than 1000 eQTLs for gene expression in fat and kidney tissues, with more significant eQTLs predominantly cis-acting. Within previously mapped physiological QTLs, specifically hypertension-related loci, 73 significant cis-regulated eQTLs with human homologues were defined that were proposed as candidates for gene loci where genetic variation may be important determinants of hypertension (Hubner et al. 2005). The genetical genomics approach has also been applied to the fly Drosophila melanogaster (Wayne and McIntyre 2002), the worm Caenorhabditis elegans (Li et al. 2006b), and plants such as maize Zea mays (Schadt et al. 2003), thale cress Arabidopsis thaliana (Keurentjes et al. 2007), and Eucalyptus (Kirst et al. 2004).
Was this article helpful?