The many faces of pleiotropy

Abstract

Pleiotropy is the well-established phenomenon of a single gene affecting multiple traits. It has long played a central role in theoretical, experimental, and clinical research in genetics, development, molecular biology, evolution, and medicine. In recent years, genomic techniques have brought data to bear on fundamental questions about the nature and extent of pleiotropy. However, these efforts are plagued by conceptual difficulties derived from disparate meanings and interpretations of pleiotropy. Here, we describe distinct uses of the pleiotropy concept and explain the pitfalls associated with applying empirical data to them. We conclude that, for any question about the nature or extent of pleiotropy, the appropriate answer is always 'What do you mean?'.

Figure 1

Different meanings of pleiotropy denote unrelated biological phenomena. Here, considering a single locus with alternate alleles (A and a) in a haploid organism, we represent genotype:gene function:phenotype:fitness mappings for alternate alleles that illustrate a few of the possible relationships. (a) Pleiotropy by all accounts. A loss-of-function mutation in a multifunctional protein abolishes two molecular activities that affect two different morphological phenotypes that in turn affect two independent aspects of fitness. (b) Molecular-gene pleiotropy occurs when a gene product carries out multiple independent biochemical functions. A gene deletion may abolish both functions, but if they are both redundant there may be no measureable phenotype (in this genetic background). (c) Genes vs. mutations: A mutation in a multifunctional gene may affect only one function, in which case this instance of molecular-gene pleiotropy does not correspond to any developmental or selectional pleiotropy. (d) Developmental pleiotropy occurs when a mutation affects two aspects of phenotype. This phenomenon does not require any multifunctionality at the level of the molecular gene nor any particular pattern of impact on fitness. (e) A mutation that affects only a single phenotype may nevertheless yield selectional pleiotropy if the phenotype has varied effects on fitness in different circumstances –different environments or sexes, for example. (f) Perhaps the least intuitive case is a mutation that affects only a single phenotype and has identical effects on orthogonal fitness components. Because these fitness components are independent traits in the selectional context, a mutation with uniform effects on more than one is necessarily pleiotropic.

Figure 2

Two ways mutations can affect multiple traits. (a) An allelic difference (A/a) independently affects the blue trait and the red trait. (b) An allelic difference affects the blue trait, which in turn influences the red trait. The former case may be more susceptible to genetic modifiers of pleiotropy. In the latter case, independent environmental and genetic perturbations that affect the blue trait also affect the red trait.

Figure 3

Distributions of pleiotropy for single-gene deletions. Previous estimates of pleiotropy have suggested that most genes have few effects, and the distribution of effects per gene is L-shaped. The data responsible for those estimates [15] are replotted here in blue: most deletions affect few traits, and a few deletions affect many. Data excluded from those estimates, plotted in red, shows that the distribution of genic pleiotropy is actually U-shaped, with a second mode at maximal pleiotropy. (a) Yeast deletions affect a fraction of the 253 measured morphological traits, if one excludes those deletions that affect all 253 traits (lethals, red bar at “all”). Deletions with no detectable effects constitute an even larger fraction of genes (red bar at “none”). (b) Yeast deletions affect growth rate in only a small fraction of environments, excluding lethals, which affect growth rate in all 21 studied environments (“all”). In this dataset, the large category of genes with no detectable effect (“none”) actually includes genes that uniformly decrease growth rate in all environments; only deletions exhibiting gene-by-environment interactions were counted as pleiotropic. (c) A live-imaging study of C. elegans embryos in which 661 essential genes had been knocked down by RNAi found that each knockdown resulted in only a subset of 44 possible embryonic phenotypes. (d) If any number of adult traits or life-history traits are included in the analysis of the 661 genes plotted in panel c, each gene is maximally pleiotropic –all traits are affected as –every gene result in embryonic lethality.