Heterosis: one boat at a time, or a rising tide?

Authors


The proverb, ‘a rising tide lifts all boats’, is often used in a political context, but is also useful in considering mechanisms of heterosis. Debates about the underlying basis of heterosis usually focus on mechanisms defined by standard quantitative genetic models (Falconer & Mackay, 1996), but often include discussions of whether there are other novel mechanisms that are independent of specific genes in biochemical and developmental pathways that more broadly affect many traits by increasing overall vigor. In this issue of New Phytologist, Stephen Goff proposes in his Tansley review (pp. 923–937) that hybrid vigor results from a reduced metabolic cost of protein recycling in hybrids owing to the opportunity for the cell to select alleles that produce stable proteins and thereby reduce the metabolic expense of processing nonfunctional proteins.

‘…research on heterosis is still limited by the number of genotypes that can be measured for performance in relevant contexts …’

The terminology utilized by Goff distinguishes single-trait heterosis from multigenic heterosis. As most traits for which we might measure heterosis are likely to be controlled by multiple genes, I believe that it is more accurate to distinguish single-trait heterosis (the single ‘boat’) as that clearly attributable to defined underlying pathways within the context of standard models of quantitative inheritance vs pleiotropic heterosis (the ‘rising tide’), which is pathway independent and results, in some way, from general vigor stimulated by heterozygosity.

One argument for mechanisms resulting in pleiotropic heterosis is that a large number of genetic and genomic studies have not defined specific biochemical or developmental pathways underlying heterosis. Traits such as grain yield, for which heterosis has most often been measured, are highly complex traits that result from the combination of a suite of developmental, biochemical and plant-protection component traits. A recent study in tomato corroborates that even a very simple single-gene example of heterosis for yield is manifested through the multiplicative interaction of component traits (Krieger et al., 2010). Results from quantitative trait locus mapping studies (Buckler et al., 2009) that have the scale to detect and resolve small effects are consistent with the hypothesis that genetic variation for most quantitative traits is controlled by many genes with small effects. In addition, recent genomic studies (Gore et al., 2009) provide experimental support for previous observations (Moll et al., 1963) that recombination (or lack thereof) influences our interpretations of quantitative genetic parameters, such as the average degree of dominance, and provides a basis for heterotic patterns in natural and breeding populations. Given a model of many genes with small effects in the context of current evidence that there are a number of regions in the genome with high linkage-disequilibrium, it is reasonable to make the counter-argument that our inability to identify specific pathways is more an issue with the genetic complexity of the target trait and the scale of our experiments and less that there is a need to identify novel mechanisms of pleiotropic heterosis.

Various models have been proposed that generally fall into the category of mechanisms resulting in pleiotropic heterosis. These include genome-wide changes in DNA methylation (Tsaftaris & Polidoros, 2000), organellar complementation (Srivastava, 1981), small RNA expression (Ha et al., 2009) and the current proposal of Goff. One attractive feature of the Goff proposal is that it provides a cell-based mechanism to account for differences in metabolic efficiency that have been observed in hybrids vs inbreds. Manifestation of very small differences in efficiency among cells and across development can have a substantial cumulative effect over the lifetime of an organism. Current advances in sequencing, proteomics, nanotechnology, imaging and metabolomics provide an unprecedented depth of information with which to study biological phenomena such as heterosis. This wealth of information, coupled with novel computational algorithms and informatics approaches, will provide new insights. However, despite the exponential advances in technology, increasing outputs and decreasing costs, research on heterosis is still limited by the number of genotypes that can be measured for performance in relevant contexts (e.g. maize yield in the field). This limitation reduces the power to detect small genetic effects and complex interactions. Furthermore, limited amounts of genetic recombination confound linkage with the estimation of various genetic parameters.

Is the mystique of heterosis warranted?

Hundreds of papers have been written and multiple conferences have been held since the origin of the concept of heterosis debating the underlying genetic and molecular basis. In many of these forums, there appears to be a fascination with the idea that standard genetic models are not sufficient to explain heterosis, and that pleiotropic heterotic mechanisms must contribute to the overall vigor observed in outcrossed organisms. In plants, conceptually simple models of additivity, dominance and epistatic interactions are sufficient to explain much of the genetic variation observed. General combining ability based on these models is highly predictive of hybrid performance, although specific combining ability often separates the most productive hybrids. Genetic responses in recurrent selection studies are consistent with expectations based on standard quantitative genetic models. There are few systematic studies of heterotic response across multiple traits in any species that allow interpretation of single trait vs pleiotropic heterosis. A recent study in maize that evaluated heterosis for 17 traits (Flint-Garcia et al., 2009) indicates that the degree of heterosis is only weakly correlated among traits. This observation is consistent with the notion that single-trait heterosis is predominant, but that some level of pleiotropic heterosis might exist.

The wealth of genomic information that is available has expanded and refined our concept of what an allele might be to include presence/absence variation and copy number variation (Springer et al., 2009) and stable epialleles (Makarevitch et al., 2007), in addition to more canonical types of variation such as single nucleotide polymorphisms. Regardless of the molecular basis of an allele, all of these diverse classes of alleles are variants of specific genes that affect specific pathways and therefore fit into standard quantitative theory which is consistent with single-trait heterosis. Some pathways implicated in heterosis, such as the circadian clock (Chen, 2010) and hormone synthesis (Rood et al., 1988), have a higher probability of having broader influence across traits than pathways expressed late in development and in terminal tissues. Nonetheless, the complexity of the genetics in terms of average magnitude of effect or the degree of interactions does not diminish the appropriateness of standard theory to define parameters of inheritance.

Ample evidence now exists across species for the role of epistasis, dominance and overdominance in heterosis. This evidence comes from variance component analysis, examples from specific genes, as well as genome-wide transcriptional and proteomic analysis. While the proportional contribution of any type of gene action in any specific manifestation of heterosis remains poorly defined, it is likely that all types of gene action will contribute. Genetically defining complex pathways whose variation is controlled by many genes with small effects will require very large population sizes, accurate phenotypic analysis and detailed genomic information coupled with maximizing genetic resolution by recombination. The balance has now shifted in genome-wide genetic studies such that the cost to phenotype agronomic performance traits, or fitness and fecundity in natural settings, now often exceeds the cost of high-density genotyping or genome resequencing. Therefore, one area of advance in the study of heterosis will be the development of creative approaches to enable large-scale phenotypic analysis. In addition, the development of genetic materials and approaches that clearly isolate variation which is not consistent with simple genetic models are necessary to define the proportional role of single-trait heterosis from pleiotropic heterosis. A shortcoming in this regard from previous research is that studies are often focused on one primary trait. Clear characterization of the pleiotropic mechanisms of heterosis will allow us to challenge and refine long-standing quantitative models of inheritance to define the basis of quantitative trait inheritance more effectively and, in applied settings, to predict the performance of the best cultivars.

Ancillary