Final Report Summary - OMUBAS (Overdominance: from spontaneous mutations to balancing selection in natural populations)
Individuals heterozygous for a given locus often display a different phenotype than the average of the two homozygous genotypes they comprise. The most common forms of such non-additive genetic effects are recessive versus dominant mutations, for which heterozygotes show a phenotype similar to homozygotes for the dominant allele. More rarely, heterozygotes may also express a more extreme phenotype than either homozygotes. Such overdominance is particularly relevant for evolution if the trait considered is fitness, the unique trait on which natural selection operates. Overdominance for fitness can maintain genetic diversity in populations even if it generates substantial fitness costs, as exemplified by sickle-cell anemia in humans. It can also contribute to heterosis (e.g. the high performance of hybrids) and inbreeding depression.
OMUBAS aimed to assess the prevalence of overdominance among spontaneous mutations as well as mutations segregating in wild populations. We used the yeast Saccharomyces cerevisiae for both aspects of the project because it is amenable to genetic manipulations and controlled crosses necessary to study the dominance of individual mutations. In yeast, the four products of meiosis can be isolated easily to provide stable haploid clones which may then be auto-diploidized to give four homozygous descendant genotypes. These genotypes can be compared to their heterozygous parent in order to estimate the number of segregating mutations, their effect, and dominance. Support for overdominance entails ruling out that the advantage of heterozygotes is due to genetically linked deleterious recessive mutations; S. cerevisiae is also ideal for this task because of its naturally high recombination rate (which breaks down genetic linkage) and our ability to identify sister chromatids in meiotic products using controlled mating of spores.
We developed a strategy using experimental evolution to isolate overdominant mutations segregating in wild yeast strains. The strategy consisted of alternating rounds of laboratory natural selection to increase the frequency of putative overdominant mutations, with rounds of strong inbreeding that reduce variation at loci with partially dominant mutations or mutations not under selection. The desired outcome of this experiment are populations lacking variation except at overdominant loci. We expect that overdominance will emerge from trade-offs among fitness traits; for example, if allele A is fitter than B in one environment (or for one fitness trait), but B is fitter than A for another, then we expect that selection fluctuating between these two environments will yield overdominance for alleles A and B. We thus carried out the selection experiment in three environments: two homogeneous environments, and one in which selection alternated between these two environments. The experiment consisted of seven evolutionary replicates per treatment and lines were propagated for 20 cycles and ~800 asexual generations (app. 9 months). At the end of the experiment we isolated 8 diploid parents and measured their fitness using head-to-head competitions against two ancestral genotypes. Surprisingly, fitness showed strong negative frequency-dependence, meaning that the fitness of evolved clones declined as their frequency increased. This result is a strong demonstration that genetic variation for fitness can be maintained even in relatively homogeneous environments. Consequently, we performed all fitness assays at low and high frequencies of the focal clones to estimate both their invasion fitness and the slope of their fitness-frequency function. These measurements uncovered large genetic variation for fitness both within and among populations, possibly maintained by trade-offs among fitness traits during asexual selection, or by a trade-off between sexual and asexual fitness (only asexual fitness was measured in our assays). For each experimental population, we chose a single genotype having both high invasive fitness and a shallow fitness-frequency function and tested for overdominance in those genotypes. This was achieved by obtaining sixteen homozygous descendants for each parent (the complete products of four meioses) and measuring their competitive fitness against the same two ancestors at multiple starting frequencies. These assays have uncovered at least one parental clone showing heterozygote advantage and we are now crossing descendants from that parent to test whether this is due to true overdominance or linked deleterious recessive mutations.
Our experiments on spontaneous mutations asked specifically if the proportion of overdominant mutations increases with the initial fitness of a genotype. This relationship was expected based on previous theoretical work from our laboratory. We tested this idea by selecting in a simple laboratory environment asexual populations of autodiploidized parents (that are homozygous at all loci except for the mating-type) that varied in fitness in that environment. Experimental selection was limited to 180 generations to ensure that only one beneficial mutation had spread in the most abundant clonal lineages in each population. We then isolated single-clones from each experimental population and constructed homozygous descendant genotypes to estimate the fitness effect and dominance of the accumulated mutations. This work is still under progress, as many tested genotypes carried none or too many mutations. Completion of this experiment is expected before the summer of 2016.
The experiments of OMUBAS are particularly relevant for scientists using experimental evolution to ameliorate the fitness of microbial strains. To do this, researchers have mostly relied on selection of populations with low standing genetic variation, whereas our experiment on overdominance in wild yeast strains exploited a comparatively large amount of standing variation. The pervasiveness of negative frequency-dependence in these populations is bad news for microbial processes relevant for society (e.g. industrial fermentation) because it means that fitness of single strains will often depend on the particular other strains it is growing with. On the other hand, our results so far do not support the view that overdominance is common, even in very permissive tests. This would suggest that desirable phenotypes due to heterosis – for example in hybrid crops – can be achieved in inbred lines using artificial selection experiments because they are caused by many recessive deleterious mutations that can be purged rather than true overdominant loci.
OMUBAS aimed to assess the prevalence of overdominance among spontaneous mutations as well as mutations segregating in wild populations. We used the yeast Saccharomyces cerevisiae for both aspects of the project because it is amenable to genetic manipulations and controlled crosses necessary to study the dominance of individual mutations. In yeast, the four products of meiosis can be isolated easily to provide stable haploid clones which may then be auto-diploidized to give four homozygous descendant genotypes. These genotypes can be compared to their heterozygous parent in order to estimate the number of segregating mutations, their effect, and dominance. Support for overdominance entails ruling out that the advantage of heterozygotes is due to genetically linked deleterious recessive mutations; S. cerevisiae is also ideal for this task because of its naturally high recombination rate (which breaks down genetic linkage) and our ability to identify sister chromatids in meiotic products using controlled mating of spores.
We developed a strategy using experimental evolution to isolate overdominant mutations segregating in wild yeast strains. The strategy consisted of alternating rounds of laboratory natural selection to increase the frequency of putative overdominant mutations, with rounds of strong inbreeding that reduce variation at loci with partially dominant mutations or mutations not under selection. The desired outcome of this experiment are populations lacking variation except at overdominant loci. We expect that overdominance will emerge from trade-offs among fitness traits; for example, if allele A is fitter than B in one environment (or for one fitness trait), but B is fitter than A for another, then we expect that selection fluctuating between these two environments will yield overdominance for alleles A and B. We thus carried out the selection experiment in three environments: two homogeneous environments, and one in which selection alternated between these two environments. The experiment consisted of seven evolutionary replicates per treatment and lines were propagated for 20 cycles and ~800 asexual generations (app. 9 months). At the end of the experiment we isolated 8 diploid parents and measured their fitness using head-to-head competitions against two ancestral genotypes. Surprisingly, fitness showed strong negative frequency-dependence, meaning that the fitness of evolved clones declined as their frequency increased. This result is a strong demonstration that genetic variation for fitness can be maintained even in relatively homogeneous environments. Consequently, we performed all fitness assays at low and high frequencies of the focal clones to estimate both their invasion fitness and the slope of their fitness-frequency function. These measurements uncovered large genetic variation for fitness both within and among populations, possibly maintained by trade-offs among fitness traits during asexual selection, or by a trade-off between sexual and asexual fitness (only asexual fitness was measured in our assays). For each experimental population, we chose a single genotype having both high invasive fitness and a shallow fitness-frequency function and tested for overdominance in those genotypes. This was achieved by obtaining sixteen homozygous descendants for each parent (the complete products of four meioses) and measuring their competitive fitness against the same two ancestors at multiple starting frequencies. These assays have uncovered at least one parental clone showing heterozygote advantage and we are now crossing descendants from that parent to test whether this is due to true overdominance or linked deleterious recessive mutations.
Our experiments on spontaneous mutations asked specifically if the proportion of overdominant mutations increases with the initial fitness of a genotype. This relationship was expected based on previous theoretical work from our laboratory. We tested this idea by selecting in a simple laboratory environment asexual populations of autodiploidized parents (that are homozygous at all loci except for the mating-type) that varied in fitness in that environment. Experimental selection was limited to 180 generations to ensure that only one beneficial mutation had spread in the most abundant clonal lineages in each population. We then isolated single-clones from each experimental population and constructed homozygous descendant genotypes to estimate the fitness effect and dominance of the accumulated mutations. This work is still under progress, as many tested genotypes carried none or too many mutations. Completion of this experiment is expected before the summer of 2016.
The experiments of OMUBAS are particularly relevant for scientists using experimental evolution to ameliorate the fitness of microbial strains. To do this, researchers have mostly relied on selection of populations with low standing genetic variation, whereas our experiment on overdominance in wild yeast strains exploited a comparatively large amount of standing variation. The pervasiveness of negative frequency-dependence in these populations is bad news for microbial processes relevant for society (e.g. industrial fermentation) because it means that fitness of single strains will often depend on the particular other strains it is growing with. On the other hand, our results so far do not support the view that overdominance is common, even in very permissive tests. This would suggest that desirable phenotypes due to heterosis – for example in hybrid crops – can be achieved in inbred lines using artificial selection experiments because they are caused by many recessive deleterious mutations that can be purged rather than true overdominant loci.