- American Society of Plant Biologists
Many factors contribute to genetic and phenotypic variation within an interbreeding population. Phenotypic variation within a species or population is highly complex; it is often polygenic and quantitative and influenced by environmental and genetic factors. In addition to considerations of allelic variation (dominance), it is recognized that both additive and nonadditive (epistatic) interactions between genes and within gene networks may play an important role. Epistasis describes a nonadditive genetic interaction that results from the activity (or mutation) of one gene masking the phenotype or effect caused by the activity of another gene. This can be contrasted with dominance, which describes an interaction between different alleles at a single genetic locus. There is considerable debate in the literature regarding the importance of epistasis in determining genetic and phenotypic variation. It is often considered that the bulk of genetic variation in populations of both plants and animals is due to additive interactions (i.e., the complementary actions of additive alleles), and nonadditive epistatic interactions are of little consequence (Rieseberg et al., 1999; Hill et al., 2008). However, Rieseberg et al. (1999) noted that this might merely reflect the difficulty of detecting epistatic effects.
Mapping of quantitative trait loci (QTL) has been used for many years to identify genetic loci that control or influence phenotypic traits. In recent years, the combination of QTL analysis and genomic methods, such as genome-wide gene expression analysis, has allowed greater insights into genetic mechanisms underlying phenotypic variation (e.g., Keurentjes et al., 2007; West et al., 2007). In the issue of The Plant Cell, Rowe et al. (pages 1199–1216) combine large-scale metabolomic analysis with QTL analysis of metabolic traits to investigate QTLs that influence the Arabidopsis thaliana metabolome. The authors identified a large number of metabolite QTLs with moderate phenotypic effects and found frequent epistatic interactions controlling a majority of the metabolic variation. The results suggest that the Arabidopsis metabolome is organized into epistatically interacting networks or clusters that regulate central metabolism.
Rowe et al. conducted untargeted metabolomic analyses on an Arabidopsis Bayreuth-0 (Bay) × Shahdara (Sha) recombinant inbred line (RIL) population described by Loudet et al. (2002) and previously used for targeted metabolite QTL and global expression QTL analyses (Loudet et al., 2003; Calenge et al., 2006; Kliebenstein et al., 2006; Wentzell et al., 2007; West et al., 2007). The untargeted metabolomic approach aims to measure as many metabolites as possible, rather than determining a priori what class of compounds or specific compounds will be investigated. Rowe et al. conducted replicated experiments using gas chromatography–time of flight–mass spectrometry to measure metabolite accumulation in the Arabidopsis Bay and Sha parental accessions and each of 210 Bay × Sha RILs.
The authors provide evidence of transgressive segregation for metabolite accumulation, in that more variation was detected in the RILs than in the parents: a significant fraction of metabolites was found in one or both parents but not detected in all RILs or found in a number of RILs but not detected in either parent, and for metabolites detected in all lines, the RILs showed extreme values compared with the parents. Transgressive segregation refers to the presence of traits or phenotypes in hybrid populations that are extreme relative to either of the parental lines and is considered an important phenomenon by which novel adaptations can arise in hybrid plant populations (Rieseberg et al., 1999). There are a number of mechanisms that could be responsible for transgressive segregation in hybrids. Rieseberg et al. (1999) describe seven possibilities: (1) an elevated mutation rate, (2) reduced developmental stability, (3) nonadditive epistatic effects between alleles, (4) overdominance, (5) unmasking of rare recessive alleles, (6) chromosome number variation, and (7) complementary action of additive alleles. Here, Rowe et al. show that epistatic interactions contribute significantly to the metabolic variation observed in the Arabidopsis Bay × Sha RILs.
The data from metabolites present in the 210 Bay × Sha RILs were used to map QTLs, and 438 QTLs affecting 243 metabolites were identified and mapped. Eleven regions of the genome were identified as metabolite QTL clusters or hot spots that contained more metabolite QTLs than expected by chance, and five of these clusters were coincident with expression QTL hot spots identified by West et al. (2007), suggesting a link with transcript variation. Metabolites were then clustered based on QTL position and allelic effect, thereby connecting the metabolite QTL clusters based on shared regulation of specific metabolites. The authors found that at least four metabolite QTL loci had global effects on central metabolism.
Genome-wide epistasis was assessed by conducting formal ANOVA pairwise tests of epistasis between the 11 identified metabolite QTL clusters. All 55 putative pairwise epistatic interactions between the eleven metabolite QTL clusters were tested against the average accumulation of 557 metabolites within the RILs. For each metabolite, r2 estimates were determined for all significant main effect terms and all significant epistatic terms. For a majority of metabolites, the significant epistatic interactions explained as much or more of the genetic variation than the main effect QTLs. Another ANVOA test involving three of the four metabolite QLT clusters associated with central metabolism revealed a significant three-way epistatic interaction between these clusters. These results suggest that epistasis played a significant role in determining the outcome of metabolite transgressive segregation observed in the Bay × Sha RILs.
Epistasis was also harnessed as a tool for metabolite classification. This use was based on a known epistatic interaction between the AOP and Elong QTLs, which are involved in glucosinolate metabolism and transcript accumulation for aliphatic glucosinolate biosynthetic genes. The authors identified 31 metabolites whose accumulation was determined by an epistatic interaction between these two loci, the majority of which were unidentified compounds that share common QTLs with known glucosinolates and showed either positive or negative epistasis between AOP and Elong. In this way, these known epistatic loci allowed for the identification of additional candidate metabolites that may be associated with glucosinolate biosynthesis. Finally, Rowe et al. applied a logic-based approach to using metabolite differences between RILs to predict biochemical networks. This approach led to the identification of two putative biochemical networks involving previously unknown or uncharacterized metabolites.
This work illustrates the power of combining metabolomics platforms with mapping populations for investigating the genetic regulation of complex biochemical networks. The results show that metabolic pathways are regulated by a number of genetic loci, which frequently interact in a nonadditive or epistatic fashion. A similar study by Lisec et al. (2008) found more limited evidence of epistatic interactions between metabolic QTL in Arabidopsis recombinant inbred and introgression lines derived from accessions Col-0 and C24. Thus, the debate on the importance of the contribution of epistasis to genetic variation in natural populations continues. It is possible that there is an underlying difference in the level of epistasis between the two populations or that some normalization techniques can amplify additivity and/or diminish epistasis in this RIL structure; testing these possibilities requires further investigation. Another recent study by Schauer et al. (2008) examined the inheritance of metabolic QTL in a population of introgression lines of tomato and showed that most of the metabolic QTL introgressed from a wild ancestor of tomato in the domesticated line were dominantly inherited. In this case, the population structure (introgression lines) was specifically designed to maximize the focus on the dominance versus recessivity of additive QTL. The importance of epistasis in populations also is likely to be affected by breeding strategy; for example, Rieseberg et al. (1999) noted differences in transgressive segregation between outcrossing and inbreeding plants. The work of Rowe et al. suggests that increasing the experimental ability to detect epistatic interactions with advanced genomics methods, the use of larger RIL populations, and appropriate statistical analyses may continue to show that epistasis plays important roles in genetic and phenotypic variation in plant populations.
References
