Selection, Epistasis, and Genotype-Phenotype Relationships

Investigation of model organisms can sometimes be misleading about the biology of humans, but this is often because expectations about similarities between different species, especially when studied in substantially different contexts, are unrealistic (Casanova and Abel, 2004). Viewed as entities of roughly comparable complexity, model organisms can be of value more as approximate guides to what complexities and intricacies might be encountered than as templates permitting precise one-to-one extrapolations among comparable genes, gene products, or networks of interactions.

It is in the spirit endorsed in the preceding paragraph that a paper (2009) by Ralph Greenspan (no known familial connection) on the genetics of behavior in Drosophila is of particular interest. Greenspan reviews studies of selection (of fruit flies) for several different behavioral phenotypes (e.g., aggression, mating speed, and locomotor activity), in which gene expression changes in fly heads (i.e., primarily brain tissue) are assessed. Notable results from these investigations include the relatively large number of genes implicated, the modest levels of expression differences between phenotypically-distinguishable strains, and the lack of overlap between the genes identified by expression profiling and those associated with the same phenotypes by classical approaches that identify large-effect mutants. Greenspan also notes that in a variety of fruit fly selection studies, the extent of epistasis among genes influencing the phenotype of interest is related to the length of time or number of generations during which selection was applied.

The briefest selection experiments (on the order of 20-30 generations) tended to yield phenotypic changes correlated with changes in multiple genes that acted with individually small effects and with relative independence (limited epistasis). In cases where the investigators applied selection for intervals of intermediate duration (roughly one to several hundred generations), the associated phenotypic changes correlated with changes in multiple genes acting with small effects but with a high degree of interaction (substantial epistasis). The lengthiest selection experiments tended to yield phenotypic changes correlated with changes in multiple genes that acted with significant interdependence (strong epistasis) but with notable examples of individual genes having major phenotypic effects. Greenspan notes that these three scenarios correspond approximately (and respectively) to the views on the genetic basis of selection expressed by the three founders of population genetics: R. A. Fisher, Sewall Wright, and J. B.S. Haldane.

Greenspan also reviews the evidence that the gene networks associated with behavioral phenotypes, in Drosophila at least, are somewhat variable. Which genes were associated, and the nature of gene-gene interactions, appear to change when one participating element is altered through mutation. In other words, the effects of particular genes and of particular genotypes at a limited number of loci are somewhat context-dependent, where the shifting context is genetic. In several independent studies of what appear to be distinct phenotypes (and as alluded to above), the selection-related changes in gene expression patterns typically do not include loci identified by mutational screens as influencing the phenotypes in question. This extremely interesting result also has an insufficiently but crucial implication that is also consistent with the frequent context-dependence of gene function: genes that may not vary between genetically distinct and phenotypically-distinguishable organisms may nevertheless contribute to the phenotypic differences in question. In other words, genes that substantially or even crucially influence physiological or pathophysiological phenotypes may be relatively invisible to standard genetic approaches.

Of particular interest to those interested in human physiology and pathophysiology, there is ample evidence that these findings in fruit flies are roughly representative of the situation for mammalian species such as mice and humans. In a review on the genetics of human behavior Kendler and Greenspan (2006) present numerous examples of multiple genes influencing particular human behavioral traits and individual genes influencing multiple traits. They also review the compelling evidence for the influence of gene-environment interactions on behavior in both model organisms and humans. Their overall conclusion is that the available data on the complex patterns of relationships between genotypes and phenotypes are fully consistent with the evolutionary origins of humans from pre-existing species.

References

Casanova JL, Abel L. The human model: a genetic dissection of immunity to infection in natural
conditions. Nat Rev Immunol. 2004 Jan;4(1):55-66.

Greenspan RJ. Selection, Gene Interaction, and Flexible Gene Networks. Cold
Spring Harb Symp Quant Biol. 2009 Nov 10. [Epub ahead of print] PubMed PMID:
19903749.

Kendler, K.S. and Greenspan, R.J. (2006) The nature of genetic influences on behavior: Lessons from “simpler” organisms. Am. J. Psychiatry 163: 1683-1694.