Higher-fitness yeast genotypes are less robust to deleterious mutations

Abstract

Natural selection drives populations toward higher fitness, but second-order selection for adaptability and mutational robustness can also influence evolution. In many microbial systems, diminishing-returns epistasis contributes to a tendency for more-fit genotypes to be less adaptable, but no analogous patterns for robustness are known. To understand how robustness varies across genotypes, we measure the fitness effects of hundreds of individual insertion mutations in a panel of yeast strains. We find that more-fit strains are less robust: They have distributions of fitness effects with lower mean and higher variance. These differences arise because many mutations have more strongly deleterious effects in faster-growing strains. This negative correlation between fitness and robustness implies that second-order selection for robustness will tend to conflict with first-order selection for fitness.

Figure 1.. Schematic of mutagenesis and fitness assay pipeline.

Plasmids with different colors indicate different regions of homology from the yeast genome.

Figure 2.. Distributions of fitness effects of insertion mutations.

(A) Distributions of fitness effects in the large library experiment. Segregants are organized by background fitness. Color represents the fraction of mutations for each segregant in each fitness effect bin (see scale bar at right). (B) Relationship between background fitness and the mean of the DFE for the large library experiment (P = 0.03, two-sided t test). (C) Distributions of fitness effects in the small library experiment. Color represents the fraction of mutations for each segregant in each fitness effect bin (see scale bar at right). (D) Combined distribution of fitness effects of most-fit and least-fit quartile of segregants in the small library experiment. (E-G) Relationship between background fitness and DFE statistics for the small library experiment (P = 4.13 × 10⁻³¹, P = 7. 83 × 10⁻¹⁴, P = 4.08 × 10⁻¹², respectively, two-sided t test).

Figure 3.. Patterns of epistasis for individual mutations.

Fitness effects of 12 representative insertion mutations are plotted against segregant background fitness. The top left mutation is one of five putatively neutral insertions used as controls. Allelic state at the largest-effect quantitative trait locus for the fitness effect of each mutation is shown by yellow (BY) or blue (RM) color; allelic state at the second largest-effect quantitative trait locus is shown by closed (BY) or open (RM) symbol. If no significant QTLs were detected, all data points are black. Analogous plots for all insertion mutations are shown in fig. S7. Error bars represent standard errors (21).

Figure 4.. Genetic determinants of fitness effects.

(A) Histogram of regression slopes between fitness effect and background fitness for each mutation. Significant negative and positive correlations are shown in blue and yellow, respectively. (B) For each mutation, the standard deviation of fitness effect across segregants and the square root of the variance explained by each of the three models. For each mutation, the variance explained by models that were not significantly better than a no-epistasis model are not plotted. Mutations shown in red or black are insertions in or near the corresponding gene respectively; stars indicate the mutations shown in Figure 3. Only mutations with fitness effect measurements in at least fifty segregants are shown.