SteveF
03-28-2008, 02:34 PM
Interesting paper in Nature yesterday:
Wagner, G.P. et al. (2008) Pleiotropic scaling of gene effects and the 'cost of complexity'. Nature, 452, 470-472.
As perceived by Darwin, evolutionary adaptation by the processes of mutation and selection is difficult to understand for complex features that are the product of numerous traits acting in concert, for example the eye or the apparatus of flight. Typically, mutations simultaneously affect multiple phenotypic characters. This phenomenon is known as pleiotropy. The impact of pleiotropy on evolution has for decades been the subject of formal analysis1, 2, 3, 4, 5, 6. Some authors have suggested that pleiotropy can impede evolutionary progress (a so-called 'cost of complexity'5). The plausibility of various phenomena attributed to pleiotropy depends on how many traits are affected by each mutation and on our understanding of the correlation between the number of traits affected by each gene substitution and the size of mutational effects on individual traits. Here we show, by studying pleiotropy in mice with the use of quantitative trait loci (QTLs) affecting skeletal characters, that most QTLs affect a relatively small subset of traits and that a substitution at a QTL has an effect on each trait that increases with the total number of traits affected. This suggests that evolution of higher organisms does not suffer a 'cost of complexity' because most mutations affect few traits and the size of the effects does not decrease with pleiotropy.
They conclude:
The observed linear relationship between total effect size and the number of traits implies that the average mutational effect, per trait, is increasing with the square root of the degree of pleiotropy. Mutations with a high degree of pleiotropy have more substantial effects on each trait than mutations with a more limited degree of pleiotropy. This pattern is reminiscent of the decanalizing effects of major mutations14 except that in this case the alleles are not pathological, as other large effect alleles may be, but are part of the natural variation of the species. In addition, this effect is not due to the release of hidden genetic variation, which is a generic feature of genes with epistasis15. We conclude that an increased degree of pleiotropy is accompanied by an increase in the overall phenotypic effects of mutations even among 'minor effect' alleles.
These findings affect predictions about the consequences of complexity on evolvability in two ways. First, the reason that Fisher's geometric model suggests a decrease in evolvability with increasing number of traits (complexity) is that his and all studies following his approach assume that each mutation potentially affects all traits ('universal pleiotropy'). Therefore with increasing complexity it becomes increasingly unlikely that all traits are affected by a mutation in a way that causes fitness to increase. However, the effects we detected in our study are not nearly as widely pleiotropic as assumed by the model of universal pleiotropy. QTL effects are more restricted to parts of the phenotype as suggested by the idea of variational modularity16, 17. Why this is so is unclear, but there is increasing evidence that natural selection can change pleiotropy such that evolvability increases18. If, at any one time, only one or a few characters are maladapted, modularity increases evolvability19, 20. The second factor that was cited as leading to a lower evolvability of complex organisms is the assumption of constant total effect5. This assumption was introduced to accommodate the fact that most mutations have small effects5, 6. In contrast, the euclidean superposition model with universal pleiotropy predicts that the probability of small-effect mutations becomes very small. This is so because if many characters are affected by each mutation, then it would be unlikely that the total effect is small. The constant-total-effect model, however, has the consequence that the average effect per character decreases and thus the rate of response to directional selection also decreases, leading to another cost of complexity prediction. However, our data show that the total effects of mutation actually increase with pleiotropy. It therefore seems that in real organisms the combination of restricted rather than universal pleiotropy, and increasing total effects, could be seen as evolution's answer to the challenges of evolving complex organisms with random variation and selection.
By the way, those of us who frequent the evolution blogosphere, may well have come across an interesting character who goes by the name of "Sanders". He frequents Sandwalk and often Pandasthumb (and is banned at Pharyngula). You might be vaguely interested to learn that he is a member of Gunter Wagners lab (Alexander Vargas):
http://pantheon.yale.edu/%7Egpwagner/people.html
Wagner, G.P. et al. (2008) Pleiotropic scaling of gene effects and the 'cost of complexity'. Nature, 452, 470-472.
As perceived by Darwin, evolutionary adaptation by the processes of mutation and selection is difficult to understand for complex features that are the product of numerous traits acting in concert, for example the eye or the apparatus of flight. Typically, mutations simultaneously affect multiple phenotypic characters. This phenomenon is known as pleiotropy. The impact of pleiotropy on evolution has for decades been the subject of formal analysis1, 2, 3, 4, 5, 6. Some authors have suggested that pleiotropy can impede evolutionary progress (a so-called 'cost of complexity'5). The plausibility of various phenomena attributed to pleiotropy depends on how many traits are affected by each mutation and on our understanding of the correlation between the number of traits affected by each gene substitution and the size of mutational effects on individual traits. Here we show, by studying pleiotropy in mice with the use of quantitative trait loci (QTLs) affecting skeletal characters, that most QTLs affect a relatively small subset of traits and that a substitution at a QTL has an effect on each trait that increases with the total number of traits affected. This suggests that evolution of higher organisms does not suffer a 'cost of complexity' because most mutations affect few traits and the size of the effects does not decrease with pleiotropy.
They conclude:
The observed linear relationship between total effect size and the number of traits implies that the average mutational effect, per trait, is increasing with the square root of the degree of pleiotropy. Mutations with a high degree of pleiotropy have more substantial effects on each trait than mutations with a more limited degree of pleiotropy. This pattern is reminiscent of the decanalizing effects of major mutations14 except that in this case the alleles are not pathological, as other large effect alleles may be, but are part of the natural variation of the species. In addition, this effect is not due to the release of hidden genetic variation, which is a generic feature of genes with epistasis15. We conclude that an increased degree of pleiotropy is accompanied by an increase in the overall phenotypic effects of mutations even among 'minor effect' alleles.
These findings affect predictions about the consequences of complexity on evolvability in two ways. First, the reason that Fisher's geometric model suggests a decrease in evolvability with increasing number of traits (complexity) is that his and all studies following his approach assume that each mutation potentially affects all traits ('universal pleiotropy'). Therefore with increasing complexity it becomes increasingly unlikely that all traits are affected by a mutation in a way that causes fitness to increase. However, the effects we detected in our study are not nearly as widely pleiotropic as assumed by the model of universal pleiotropy. QTL effects are more restricted to parts of the phenotype as suggested by the idea of variational modularity16, 17. Why this is so is unclear, but there is increasing evidence that natural selection can change pleiotropy such that evolvability increases18. If, at any one time, only one or a few characters are maladapted, modularity increases evolvability19, 20. The second factor that was cited as leading to a lower evolvability of complex organisms is the assumption of constant total effect5. This assumption was introduced to accommodate the fact that most mutations have small effects5, 6. In contrast, the euclidean superposition model with universal pleiotropy predicts that the probability of small-effect mutations becomes very small. This is so because if many characters are affected by each mutation, then it would be unlikely that the total effect is small. The constant-total-effect model, however, has the consequence that the average effect per character decreases and thus the rate of response to directional selection also decreases, leading to another cost of complexity prediction. However, our data show that the total effects of mutation actually increase with pleiotropy. It therefore seems that in real organisms the combination of restricted rather than universal pleiotropy, and increasing total effects, could be seen as evolution's answer to the challenges of evolving complex organisms with random variation and selection.
By the way, those of us who frequent the evolution blogosphere, may well have come across an interesting character who goes by the name of "Sanders". He frequents Sandwalk and often Pandasthumb (and is banned at Pharyngula). You might be vaguely interested to learn that he is a member of Gunter Wagners lab (Alexander Vargas):
http://pantheon.yale.edu/%7Egpwagner/people.html