Introduction

• why is genetic architecture essential in understanding evolutionary theory? because it describes or determines the variational properties of characters, and thus their evolutionary potential, i.e. the evolution of evolvability, see Kirschner & Gerhart in 1997
• aim: incorporating development and more realistic representations of the genotype-phenotype map into population genetics
• classical model of quantitative genetics: polygenic model proposed by Fisher in 1918 , see also Elston & Stewart in 1971
• genetic canalization: evolution of reduced genetic variability (Delta geno => small Delta pheno), introduced by Waddington in 1942, formalized by Wagner et al in 1997
• genetic assimilation: evolutionary responses based on environmentally induced variation, introduced by Waddington in 1953

Genetic architecture: definition and distinctions

• need to distinguish between statistical/population notions (e.g. in QTL analysis) and biological/physiological notions (what really happens in a given organism), see Cheverud & Routman in 1995 and Hansen & Wagner in 2001
• epistatis: the effect of a gene, or genotype, is different in two different genetic backgrounds, see Rice in 2000 and 2002
• pleiotropy: said of a gene affecting more than one character

Genetic architecture and evolvability

What is evolvability?

• "ability to evolve", but many biological definitions
• ability of the genetic system to produce and maintain potentially adaptive genetic variants

• property of genetic architecture rather than population (even though may be estimated at population level)
• concern the production of variation as well as its maintenance
• focus on adaptation (neglect capacity for producing unconditionally deleterious mutations)

Autonomy

• every trait should not depend on every other trait (unbounded pleiotropy) -> modularity
• evolutionary autonomy of a trait = conditional evolvability (i.e. everything else being fixed)

Mutability

• ultimate source of variation, thus fundamental limit on evolvability
• mutational target size (of a trait): number, size and mutability of genes and regulatory elements affecting this trait

Coordination

• coordinated variability: simple, nonobstructive changes in developmental processes can lead to large changes in phenotype, which may have a higher likelihood of being advantageous because they are structured by the developmental process
• examples: heterochronic or allometric changes, co-option of gene networks

Gene interaction

• directional epistasis (>0 or <0) can have dramatic effect on response to selection
• not found by Fisher's model of variance components as it implicitly assume nondirectional epistasis (in fact the estimates are a mixture of different kind of epistasis)
• see Carter et al in 2005

Hidden and cryptic variation

• a genetic system can hide variation via negative LD or canalization (epistasis with background) and revealed cryptic variation can fuel rapid adaptation
• see evolutionary capacitors as Hsp90 in Drosophila and Arabidopsis
• genetic systems with cryptic variation are also robust, arguably due to overdetermination of each phenotype by many genotypes, see Wagner in 2005 and 2008

Genome dynamics

• recombination, duplication, conversion, etc

The major transitions

• changes in how the organism transmits genetic information, according to Maynard-Smith & Szathmary in 1995 (e.g. sex and recombination, multicellularity)

Evolution of genetic architecture: general principles

Modes of changes

• epistasis with an evolving genetic background
• heritable allelic effects (when correlation between the effects of alleles and the mutations they can generate)
• genotype-by-environment interactions with a changing environment

Modes of selection

• 3 classes of hypotheses for the evolution of variational properties:

• such properties are influenced by selection on variation or variability
• such properties are in fact intrinsic features of the organism, not under direct selection
• congruence: mix of the two, e.g. selection for robustness is a by-product of selection for dominance

• "genetic architecture is a function of general organismal development and structure"
• "organismal architecture is not adapted to structure genetic variation, but it is adapted to structure environmental variation"

=> what does it mean??!

Evolution of the genotype-phenotype map

Canalization and Genetic assimilation

• wild types supposed to be canalized because they show less variation than mutants
• perturbances (mutations or from environment) can push system away from this and thus make variation appear that can be selected upon

Does stabilizing selection favor canalization?

• stabilizing selection is generally expected to be the most common mode of selection

=> really?

Environmental canalization and the congruence hypothesis

• environmental canalization: genetic mechanisms that reduce environmental noise

Does directional selection favor decanalization?

• depends on directionality of epistasis
• lacks theoretical studies of fluctuating selection

Evolution of pleiotropy

• will occur if epistatic interactions on a gene affects several traits differently, but also depends on directionality of epistasis
• can selection "decouple" two traits that are functionally unrelated?

=> what "decouple" means here if the two traits are already functionally unrelated??

Evolution of gene interactions

• evolution of m-order interactions influenced by the directionality of m+1-order interactions

Evolution of dominance

• Fisher: dominance is an adaptation to minimize the deleterious effects of mutations, but such selection is very weak, and dominance also occurs in species spending most of their time in a haploid state
• Haldane & Wright: dominance is a safety factor against disturbances (congruence)
• Kacser & Burns: dominance is an intrinsic consequence of enzymatic biochemistry, but only applies to flux in linear pathways under Michaelis-Menten kinetics

Genetic architecture: intrinsic or adapted?

• general but very abstract model, as Kauffman in 1993
• metabolic networks, as Szathmary in 1993
• possible generic properties, as robustness, see Wagner in 2005

Evolution of genome architecture

• number and types of genes
• impact of gene duplication, see Wagner in 1994
• retention of gene duplications and TEs/introns, see Lynch & Connery in 2003

Empirical challenges

Theory and experiment

• statistical models of classical quantitative genetics were set up in a way that did not facilitate either the empirical or the theoretical study of gene interactions
• most of the dynamic theory is based on models that lack parameters that are experimentally measurable
• fundamental problem with many models: they lack an adequate representation of the phenotype
• study gene interaction networks
• perform artificial selection on phenotypes with well-defined underlying gene networks

Epistatis

• epistatic variance components cannot be used to distinguish between different types of epistasis
• need better statistical models for QTL detection

Pleiotropy

• need to study sign and degree of variation of pleiotropic effect, not only presence/absence
• estimating the underlying dimensionality of characters

Conclusions

• need a better connection to experiment through the identification of measurable parameters and general analytical results
• connecting the masses of genetic data to interesting phenotypes