Santa Fe Institute [date: before Jan 2006]
Evolutionary Dynamics: Genotype-Phenotype Map
Lauren Ancel, SFI Postdoctoral Fellow
Walter Fontana, SFI Research Professor

Waddington offers two modes for the evolution of organismal development (1957). The first is environmental canalization: the honing of developmental pathways to buffer environmental perturbations. The other is genetic canalization: the evolution of epistatic gene interactions to buffer the deleterious effects of mutation. The consequence of these two processes is the production of well-defined phenotypes that are preserved across developmental and genetic contexts. Lauren Ancel and Walter Fontana have recently discovered a remarkable link between environmental canalization and genetic canalization. Premised on the common biophysical underpinnings of environmentally induced phenotypic change and genetically induced phenotypic change, we find that genetic canalization and hence a loss of evolvability ensues as a by-product of environmental canalization.

Environmental canalization is the reduction of phenotypic noise induced by micro-environmental variation. In contrast, phenotypic plasticity is the ability of organisms with the same genotype to vary in developmental pattern, phenotype, or behavior in response to varying environmental conditions. We speculate that phenotypic plasticity incurs fitness costs. If plasticity increases phenotypic options with no associated costs, then natural selection should produce maximally plastic organisms. Since this is not the case, scientists have sought to identify the forces that curb plasticity. Schmalhausen identified "erroneous" phenotypic changes as one cost of plasticity (1949). Other potential costs include the energetic costs of regulation, sensing the environment and producing biological structures; and genetic costs like deleterious pleiotropy of genes involved in plasticity (DeWitt et al., 1998).

If plasticity is costly, why does it occur? One explanation rests on the demands of an everchanging environment. Examples of plasticity in nature—leaf morphology, bacteriophage host range, human behavior—seem to correlate with environmental heterogeneity. Several mathematical models, motivated by such observations, have demonstrated the stability of phenotypic plasticity despite associated fitness costs in a temporally or spatially fluctuating environment (Slatkin & Lande, 1976; Via & Lande, 1985; Ancel, 1999). Upon encountering a new environment, a plastic organism with a range of phenotypes is more likely to exhibit an advantageous one than a non-plastic organism adapted to the preceding environment. A model developed by Ancel permits a quantitative analysis of the population dynamics surrounding environmental transitions, and predicts an intuitive relationship between the rate of environmental stochasticity and the extent of phenotypic plasticity of a population (Ancel, 1999). See Figure 1 below.

When there are metabolic or other costs associated with plasticity, natural selection reduces plasticity in a static environment. Using a more complex and realistic phenotype, RNA secondary structure, we have studied the ramifications of such a loss in plasticity (Ancel & Fontana, 1999). This model of plasticity is based on the idea that an RNA molecule in a slightly heterogeneous environment fluctuates between its lowest free energy structures. Examples of environmental variability include thermodynamic fluctuations in a heat bath and the presence of other potentially interacting or colliding molecules.

Every sequence in the model has a plastic repertoire consisting of the most energetically favorable secondary structures it can attain, that is, all secondary configurations of the sequence that lie within a small energetic neighborhood of its minimum free energy structure. The time a sequence spends in each structure is proportional to how energetically favorable the structure is relative to other structures in the repertoire.

They predetermine an optimal structure that is constant throughout the evolutionary simulations. The fitness of a sequence is then a weighted average of the distance between each of the structures in its repertoire and the optimal structure, where the weight is proportional to the time the sequence spends in the structure. This approximates the fluctuations among RNA structures that a sequence experiences in natural environments due to thermodynamic perturbations.

Because highly plastic RNA sequences spend time in structures that cover a range of fitnesses, natural selection produces low-plasticity sequences that spend almost all of their time in the best structure in their repertoire. This is the RNA version of environmental canalization. In the face of micro-environmental (thermodynamic) fluctuations, a low plasticity sequence folds reliably into a single structure. A typical sequence at the beginning of an evolutionary simulation splits its time relatively evenly among over 500 very diverse structures in its plastic repertoire, whereas a typical sequence at the end of evolution has 5 very similar structures in its plastic repertoire and spends most of its time in the most energetically favorable one.

Fontana and Ancel also observe a surprising and dramatic reduction in evolvability. The populations evolve toward the target phenotype for a short time and then suddenly stop, unable to continue their phenotypic approach to the target. This turns out to be a side effect of natural selection for lowered plasticity. It is mediated by plastogenetic congruence which is defined as the correlation between structures within the plastic repertoire of a sequence and the structures in the mutational vicinity of that sequence. Via plastogenetic congruence, then, low plasticity sequences must also lack phenotypic variation in their genetic neighborhood, and therefore lack mutational access to improved phenotypes. In this way, the reduction in plasticity leads to phenotypes that are robust to genetic change, and are therefore unlikely to evolve. In other words, under plastogenetic congruence, environmental canalization entails genetic canalization. This gives a mechanistic explanation for a hypothesis that was put forth originally by Wagner et al. (1997).

Environmental canalization of RNA secondary structure has another dramatic by-product: modularity. Not only do the sequences produced by the evolutionary simulations have highly reduced variation in their plastic repertoires and their mutational neighborhoods, they are also thermodynamically, genetically and kinetically modular. This means that the minimum free energy structures of these sequences—that is, the structures in which they predominantly spend their time—consist of independent substructures that maintain their integrity across a wide range of temperatures, withstand embedding in random genetic contexts, and follow a remarkably fast and directed folding path into the minimum free energy structure.

Plastogenetic congruence and the biophysical connection from low plasticity to modularity are not particular to RNA. Phenocopies—environmentally induced mimics of mutant phenotypes provide evidence for plastogenetic congruence in, for example, flies, moths, and birds (Gibson Hogness, 1996; Goldschmidt, 1940; Miller 1990). If we accept the folding of RNA secondary structures as a metaphor for developmental pathways, then our results have fundamental importance to our understanding of evolution. This work suggests that we look to the environmental context in which populations evolve to explain why some organisms evolve faster than others; how modular organization of regulatory networks and biopolymers originates; and what gives rise to a shift in the syntactical units of phenotypic variation.

Future work is focusing on the following research objectives:

1. The evolvability of mammalian viruses. The remarkable relationship between plasticity and evolvability found in RNA suggests that we look to the evolutionary context of a lineage to explain its evolutionary potential. Studies of clonal interference in vesicular stomatitis virus raise the important question of why, despite a higher overall mutation rate, this virus evolves more slowly than, for example, E. coli (Miralles et al., 1999; Drake, 1993). This discrepancy may result from differences in the evolutionary histories. Viruses evolving under constant conditions for many generations may be canalized toward the environment. By virtue of plastogenetic congruence, these viruses will also be buffered towards the effects of mutation, and thus isolated from novel phenotypes. Recall that plastogenetic congruence is the correlation between environmentally and genetically induced phenotypic modifications.

On the other hand, viruses within immunocompetent hosts confront dynamic cellular environments. Variants that evade the immunological defenses will be the most successful. The ability to generate such superior variants through mutation over the course of infection therefore underlies the long-term dynamics of infection. (Wodarz & Nowak, 1998, 1999; Haraguchi & Sasaki, 1997)

We speculate that temporal and/or spatial heterogeneity in the host-cell environment increases viral evolutionary potential. The logic behind this hypothesis stems from the correlation of plasticity with evolvability apparent in RNA. A heterogeneous immune response or therapeutic protocol will select for increasingly phenotypically plastic viruses that can adapt to a changing host response (Domingo, 1997). By plastogenetic congruence, these plastic strains will be highly sensitive not only to environmental perturbation (as the term plasticity implies) but also to mutations. Ironically then, a multi-stepped immune response may increase the evolvability of infecting pathogens, and thereby be self-defeating for the host.

Based on empirical work, Ancel would like to extend models of immune responses to pathogens (Regoes et al., 1998; DeBoer & Perelson, 1998; Perelson & Nelson, 1999; Bonhoeffer et al., 1997) to include a more explicit viral genotype-phenotype relation, one that captures the environment- gene interactions and rugged phenotypic neutrality (networks of diverse genotypes that produce the same fitness yet have different evolutionary potential).

2. The building blocks and evolvability of developmental networks. Modularity in developmental networks is like a toolbox of self-standing phenotypic pathways that are well buffered to changes in genotypic and environmental context and can be assembled combinatorially to create novel phenotypes. For example, the leaves of lycopsids and the interserninal scales of Bennettitales are thought to have originated through diverted development of reproductive organs (Crane & Kenrick, 1997). Bird locomotion and the structure of reef-building colonies may be examples of evolutionary transitions to modularity from more amorphous organization (Gatesy & Middleton, 1997; Zhuravlev, 1999).

The rise of modularity in genetic regulatory mechanisms that control development shifts the unit of evolution from the individual gene to sets of highly interacting genes that contribute to a module. Modularity is thought, therefore, to facilitate the evolution of complex and better-adapted life forms. Mutations in homeotic genes, for example, cause one body part to transform into the structure of another body part. Although mutations that move a leg to the position of an antenna (in the Antennapedia complex) or transform the balancing organs into a pair of wings (in the Bithorax complex) are likely to be deleterious to the organism, they provide evidence for rapid combinatoric modification of a general body plan (Lewis, 1994). This possible advantage to modularity, however, is not necessarily the reason for the emergence of modularity in the first place (Wagner, 1996; Ancel Fontana, 1999).

Ancel and her collaborators would like to develop theoretical models to study the evolutionary origins of genetic networks. They conjecture that modularity in development, like modularity in RNA structure, may arise as a side effect of environmental canalization, independent of the ultimate contribution of modularity to evolvability.

3. Continued studies of RNA evolution. There are two directions in which Walter Fontana and Ancel hope to continue their work on plasticity in RNA molecules. First they will seek experimental verification of the relationship between selection for thermodynamic stability and the loss of evolvability. The SELEX experimental technique can artificially select for RNA molecules, called aptamers, that bind optimally to preselected target molecules (Ellington & Szostak, 1990; Tuerk & Gold, 1990; Ellington, 1994). In collaboration with Laura Landweber and her laboratory at Princeton University, tjeu will use this experimental set-up to test directly the relationship between the loss of plasticity and the loss of evolvability in RNA sequences.

They will evolve a population of variants based on aptamers to bind with a preselected target and then assess their further evolvability to a different, chemically similar target (experiments like this have been performed for citrulline/arginine targets (Yang et al., 1996) and arginine/guanine targets (Connell and Yarus, 1994)) and their ability to withstand extreme temperatures. The RNA secondary structure folding algorithm provides another analytical assay for the plasticity and modularity of the sequences produced by SELEX. If our theory is correct, we will find that the longer the sequences have evolved toward a fixed target (in number of generations), the harder it is for these sequences to evolve to another target, the more impervious they are to temperature deviations, and the more modular they appear in computations of their kinetic, thermodynamic and genetic integrity.

They will also study the evolution of plasticity in RNA within a heterogeneous environment. According to models for phenotypic plasticity mentioned above, a variable macro-environment can favor developmental flexibility. A heterogeneous environment may thereby enhance evolvability as plastogenetic congruence implies that phenotypic flexibility correlates to genetic sensitivity. They will model environmental fluctuations by varying either the target phenotype, the temperature under which the molecules fold and/or a stream of potentially cofolding molecules.

References

  1. Ancel, L.W. (1999) A quantitative model of the Simpson-Baldwin effect. Journal of Theoretical Biology 196, 197-209.
  2. Ancel, L. and W. Fontana (1999) Plasticity, Evolvability and Modularity in RNA. Journal of Experimental Zoology: Molecular and Developmental Evolution, Submitted.
  3. Bonhoeffer S., R.M. May, G.M. Shaw and M.A. Nowak (1997) Virus dynamics and drug therapy. Proceedings of the National Academy of Sciences 94, 6971-6976.
  4. Connell, G.J. and M. Yarus (1994) RNAs with dual specificity and dual RNAs with similar specificity. Science 264, 1137-1141.
  5. Crane, P.R. and P. Kendrick (1997) Diverted development of reproductive organs: A source of morphological innovation in land plants. Plant Systematics and Evolution 206, 161-74.
  6. Deboer, R.J. and A.S. Perelson (1998) Target-cell limited and immune control models of HIV infection: A comparison. Journal of Theoretical Biology 190, 201-214.
  7. Dewitt, T.J., A. Sih and D.S. Wilson (1998) Costs and limits of phenotypic plasticity. Trends in Ecology and Evolution 13, 71-81.
  8. Domingo, E. (1997) RNA virus evolution, population dynamics, and nutritional status. Biological Trace Element Research 56, 23-30.
  9. Drake, J.S. (1993) Rates of spontaneous mutation among RNA viruses. Proceeding of the National Academy of Sciences 90, 4171-4175.
  10. Ellington, A.D. (1994) RNA selection: Aptamers achieve the desired recognition. Current Biology 4, 427-429.
  11. Gatesy, S.M. and K.M. Middleton (1997) Bipedalism, flight, and the evolution of theropod locomotor diversity. Journal of Vertebrate Paleontology 17, 308-329.
  12. Gibson, G. and D.S. Hogness (1996) Effect of polymorphism in the Drosophila regulatory gene ultrabithorax on homeotic stability. Science 271, 200-203.
  13. Goldschmidt, R.B. (1940) The material basis of evolution. New Haven, Connecticut: Yale University Press.
  14. Haraguchi, Y. and A. Sasaki (1997) Cross-reactivity in immune-response. Transactions of the Royal Society of London Series B-Biological Sciences 352, 11-20.
  15. Kimata, IT., L. Kuller, D.B. Anderson, P. Dailey and J. Overbaugh (1999) Emerging cytopathic and antigenic simian immunodeficiency virus variants influence AIDS progression. Nature Medicine 5, 535-541.
  16. Lewis, E.B. (1994) Homeosis: The first 100 years. Trends in Genetics 10, 341-343.
  17. Miralles, R., P.J. Gerrish, A. Moya and S.F. Elena (1999a) Clonal interference and the evolution of RNA viruses. Science 285, 1745-1747.
  18. Müller, G.B. (1990) Developmental mechanisms at the origin of morphological novelty: A side-effect hypothesis. In Evolutionary Innovations (M.H. Nitecki, ed.) Chicago: University of Chicago Press 99-130.
  19. Perelson, A.S. and P.W. Nelson (1999) Mathematical analysis of HIV-1 dynamics in vivo. SIAM Review 41, 3-44.
  20. Regoes, R.R., D. Wodarz and M.A. Nowak (1998) Virus dynamics: The effect of target cell limitation and immune responses on virus evolution. Journal of Theoretical Biology 191, 451-462.
  21. Rohn, J.L., M.S. Moser, S.R. Gwynn, D.N. Baldwin and J. Overbaugh (1998) In vivo evolution of a novel syncytium-inducing and cytopathic feline leukemia virus variant. Journal of Virology 72, 2686-2696.
  22. Schmalhausen, I.I. (1949) Factors of evolution (Dordick, I., trans.; Dobzhansky, T., ed.) Philadelphia: Blakiston.
  23. Slatkin, M. and R. Lande (1976) Niche width in a fluctuating environment—Density independent model. American Naturalist 110, 31-55.
  24. Tuerk, C. and L. Gold (1990) Systematic evolution of ligands by exponential enrichment (SELEX): RNA ligands to bacteriophage T4 DNA polymerase. Science 249, 505-510.
  25. Via S. and R. Lande (1985) Genotype-environment interaction and the evolution of phenotypic plasticity. Evolution 39, 505-522.
  26. Waddington, C.H. (1957) The strategy of the genes. New York: MacMillan Co.
  27. Wagner, G.P. (1996) Homologs, natural kinds and the evolution of modularity. American Zoologist 36, 36-43.
  28. Wagner, G.P., G. Booth and H. Bagheri-Chaichian (1997) A population genetic theory of canalization. Evolution 51, 329-347.
  29. Wodarz, D. and M.A. Nowak (1998) The effect of different immune responses on the evolution of virulent CSCR4-trophic HIV. Proceedings of the Royal Society of London Series B-Biological Sciences 265, 2149-2158.
  30. Yang, Y.S., M. Kochoyan, P. Burgstaller, E. Westhof and M. Famulok (1996) Structural basis of ligand discrimination by 2 related RNA aptamers resolved by NMR-spectroscopy. Science 2727 1343-1347.
  31. Zhuravlev, A.Y. (1999) The modularity and development of Cambrian reef ecosystem. Zhurnal Obshchei Biologil 60, 29-40.

Baldwin Effect / Genotype-Phenotype Map