PMC

Selection for dispersed cell pellet morphologies. Six examples of heritable changes in cell pellet morphology from plates BYS1 and BYB1 are shown. Imaged cell pellets from all 144 populations on these two plates are shown in Figure S3.

The waiting time to the first observation of sterile mutations, τ_up1. (A) The distribution of τ_up1 shows that sterile mutations were initially observed earlier in the small N_e populations (blue bars) than in big N_e populations (yellow bars). (B) Late-occurring steriles spread slightly slower than early-occurring steriles (P = 0.0016, Spearman's rank correlation). The black line is the best-fit linear regression (Pearson's) to the data.

A schematic of a clonal interference trajectory illustrating the parameters measured for each trajectory. The initial spread of steriles is characterized by the time at which steriles initially reach 0.1% (τ_up) and the initial exponential (linear in log space) rate of increase (s_up). Similarly, the purging of steriles is characterized by s_down and τ_down. The maximum height of the trajectory is %_max. Not all parameters were extracted from each trajectory, and for some trajectories, multiple values for a given parameter were obtained where possible and are denoted by subscripts. The full list of extracted parameters is available in Figure S1 and Table S1.

The distribution of fates of spontaneous sterile mutations. (A) The number of lines in each propagation regime in which sterile mutations were observed during the course of the experiment. Steriles were observed more frequently in small populations and when they confer a larger selective advantage. “No data available” refers to populations that were lost or that changed in a way that prohibited the detection of steriles. (B) The frequency of different types of dynamics among the lines where steriles were observed. In the larger populations, more of the trajectories remain unresolved over the 1000 generations of our experiment. This reflects the longer timescales for the spread of mutants in larger populations, but the trends in these unresolved trajectories are consistent with the rest of the data.

The maximum frequency attained by sterile mutations, %_max. (A) The distribution of %_max across the four regimes shows that sterile trajectories attained greater frequency in smaller populations and when sterile mutations confer a larger selective advantage. (B) There is a strong positive correlation between s_up and %_max (P < 10⁻⁵, Spearman's rank correlation on the combined data). The black line is the best-fit linear regression (Pearson's) to the data. Separating the data from RM and BY lines still produces a significant positive correlation (P = 0.0008 and P < 10⁻⁵ for s_ste = 0.6% and 1.5%, respectively, Spearman's rank correlation). However, the best-fit line to the s_ste = 1.5% data is shifted to a higher %_max relative to the best-fit line to the s_ste = 0.6% data. Sterile mutations with a higher fitness advantage, therefore, attain a higher %_max.

Measuring the fraction of sterile cells within a population. (A) Binding of the mating pheromone (αF) is signaled through the mating pathway, ultimately resulting in a cell-cycle arrest and a transcriptional response mediated by the transcription factor Ste12. To detect mating competency, we put the fluorescent reporter yEVenus under the control of a Ste12-responsive promoter. A mutation in any one of 10 genes within the mating pathway (green) results in sterility and eliminates αF-induced expression of yEVenus. In the presence of αF, mating-competent cells arrest and induce the fluorescent reporter. Sterile cells, however, remain dark and continue dividing, thereby amplifying low-frequency steriles within a population. (B) Standard curve showing the amplification of steriles following αF inductions of 4, 6, 8, and 12 hr. For the experiments described here, we used a 6-hr induction. (C) Examples of flow cytometry profiles for cultures with no steriles (left) and with 1% sterile individuals (right) following a 6-hr αF induction.

Rate of mean fitness increase. (A) A schematic of a clonal interference trajectory relating the parameters s_up, s_down, and %_max in terms of the fitness of the sterile subpopulation relative to the mean fitness of the population. The average speed of mean fitness increase is (s_up + s_down)/τ_transit, where τ_transit = τ_down – τ_up. If the speed of mean fitness increase is constant, the trajectory will be symmetrical with s_up = s_down and %_max occurring equidistant between τ_up and τ_down. Asymmetry indicates either the slowing down (s_down < s_up) or speeding up (s_down > s_up) of the speed of mean fitness increase. (B) The average rate of mean fitness increase during the transit time of the sterile lineage, as inferred from the 51 high-quality clonal interference trajectories, color coded by propagation regime (top) and acceleration of mean fitness increase (bottom). (C) The rate of adaptation for 12 of the 51 populations for which standard (competitive fitness assay) measurements of mean fitness over time were available. This independent method of calculating the speed of mean fitness increase is consistent with the estimate of 1%/100 generations as estimated from the trajectories. (D) Relationship between s_up and s_down in populations in which we observed 0.6%-effect sterile mutations; note the excess of trajectories in which the speed of mean fitness increase is slowing down. (E) Relationship between s_up and s_down in populations in which we observe 1.5%-effect sterile mutations; these are not biased in terms of changes in mean fitness increase.

Spontaneous sterile mutations experienced one of four general fates. The top row illustrates a hypothetical scenario that could produce each of the four types of dynamics. Below each illustration are five representative examples of each dynamic. For each experimental population, the frequency of sterile mutants is shown as a function of time and the populations are identified by their systematic names (Figure S1). Note the logarithmic y-axis. (A) The simplest case is a selective sweep. (B) More commonly, sterile mutations are outcompeted by more fit lineages, a process known as clonal interference. (C) In some cases we observed the reemergence of steriles after clonal interference. This could occur either by a second sterile mutation (as shown in the illustration at the top) or by an additional beneficial mutation arising in the original sterile lineage. (D) Long-term maintenance of sterile mutations at a given frequency could indicate the action of frequency-dependent selection, where the difference in fitness between the sterile and nonsterile subpopulations is a function of the frequency of the subpopulations. This phenomenon is currently being explored and will be reported elsewhere.

Schematic of a selective sweep. (A) A schematic of a selective sweep showing that, unlike with clonal interference, the sterile mutation initially occurs in a highly fit individual and is never outrun by another lineage. If, at the time the sterile mutation arose, there was variation in fitness within the population, the relative fitness of the sterile lineage will decrease during the sweep as the mean fitness of the nonsterile subpopulation increases. By the end of the sweep, the sterile lineage will be competing against a population similar in fitness to the background in which the sterile mutation arose. (B) An example of a selective sweep, shown as the change in the log ratio of steriles to nonsteriles over time fit to a linear model (corresponding to a constant relative fitness of the sterile lineage) and a power law model (corresponding to a decreasing relative fitness of the sterile lineage). Similar plots for the other seven sweep populations are shown in Figure S2. (C) The change in the relative fitness of the sterile lineage relative to the population mean fitness can be evaluated at any point by calculating the derivative of the best power-law fit to the data. The initially fast increase slows to a value similar to the fitness provided by the sterile mutation alone, s_ste, by the end of the sweep. Populations 1–8 are BYS1-A07, -A08, and -F05 and BYS2-C03, -C06, -D06, -D07, and -E03, respectively.

The distribution of s_up and the role of genetic variation. (A) A three-dimensional representation of a sterile trajectory (in this case, involving clonal interference) showing that s_up is a measure of the fitness of the sterile lineage relative to the population mean fitness at the time the sterile mutation arose. That is, s_up is the sum of the fitness of the background in which the sterile mutation occurred and of the fitness advantage provided by the sterile mutation itself. (B) The distribution of fitness effects of the sterile mutation alone in the ancestral background (s_ste, gray bars) centers on 0.6% (top) and 1.5% (bottom) in the RM and BY lines, respectively. In all four regimes, however, the initial rate of spread of sterile mutations (s_up, orange bars) is faster than can be explained by the effect of the sterile mutation alone, s_up > s_ste (P = 2.5 × 10⁻⁷, 2.7 × 10⁻³, 4.4 × 10⁻⁹, and 2.6 × 10⁻³, for 0.6% small N_e, 1.5% small N_e, 0.6% big N_e, and 1.5% big N_e, respectively, two-tailed t-test, unequal variance). Surprisingly, the distribution of s_up is independent of the fitness advantage of the sterile mutation itself (P = 0.66 and P = 0.13 for the hypothesis that s_up depends on fitness for big and small N_e, respectively, two-tailed t-test, unequal variance). The eight populations in which steriles swept to fixation are indicated by open blue circles; three of these have s_up > 4%. The open circles and horizontal bars indicate the means and standard deviations of the distributions.

Selection for cell aggregation. (A) Examples of cell morphologies in four populations after 1000 generations. The top panels correspond to light images of each population and the bottom panels are fluorescence images of the same populations following 6 hr of αF induction. Population RMS1-A05 retained the ancestral phenotype with respect to sterility and aggregation, BYS2-E03 fixed a sterile mutation, and BYS2-A11 and RMB1-A02 evolved aggregation. Aggregation was observed in 52 populations across the four regimes (note that many of these populations were excluded from the analysis of the dynamics of sterile mutations and fall into the “no data available” category in since aggregation interferes with the quantification of sterile frequencies). Nevertheless, we observed sterile aggregates in 6 populations. (B) Two examples of sterile aggregates. Both the top images (light microscopy) and the bottom images (fluorescence) were taken on the same field of cells following αF induction. There are two ways in which sterile aggregates can arise: sterility first or aggregation first. On the basis of the presence of three cell types (ancestral, aggregates, and sterile aggregates) in population RMB1-D10 we can infer that the mutation conferring aggregation occurred first and that the sterile mutation arose in this background. Conversely, for population RMB1-G05, the presence of steriles and sterile aggregates (but not nonsterile aggregates) indicates that the sterile mutation occurred first and that the aggregate mutation occurred in the sterile background. (C) For 4 of the 6 populations where sterile aggregates were observed, the aggregate mutation occurred first.