Aneuploidy underlies rapid adaptive evolution of yeast cells deprived of a conserved cytokinesis motor

Abstract

The ability to evolve is a fundamental feature of biological systems, but the mechanisms underlying this capacity and the evolutionary dynamics of conserved core processes remain elusive. We show that yeast cells deleted of MYO1, encoding the only myosin II normally required for cytokinesis, rapidly evolved divergent pathways to restore growth and cytokinesis. The evolved cytokinesis phenotypes correlated with specific changes in the transcriptome. Polyploidy and aneuploidy were common genetic alterations in the best evolved strains, and aneuploidy could account for gene expression changes due directly to altered chromosome stoichiometry as well as to downstream effects. The phenotypic effect of aneuploidy could be recapitulated with increased copy numbers of specific regulatory genes in myo1Delta cells. These results demonstrate the evolvability of even a well-conserved process and suggest that changes in chromosome stoichiometry provide a source of heritable variation driving the emergence of adaptive phenotypes when the cell division machinery is strongly perturbed.

Figure 1. Adaptive evolution and phenotypic characterization of myo1Δ e-strains

(A) Schematic representation of the adaptive evolution of myo1Δ strains. (B) MYO1/myo1Δ meiotic products after 3 days of growth. (C) Zymolyase-treated myo1Δ cells derived from spore 23a at passage 1 (P1) and 15(P15). Scale bars: 5 μm. (D) Phenotype of 45 myo1Δ strains during the evolution experiment. Passage numbers are on the left; spore names are listed in the top row; and strain numbers above row P3. Cytokinesis index (CI) was evaluated by comparing with the average CI of the wild type (x̄ = 0.55, n = 20). “Heterogeneous” refers to colonies of different sizes within a streak. Gray: undetermined. (E) PCA of phenotypic data of the 29 stabilized e-strains. For details, see Experimental Procedures.

Figure 2. Microscopy characterization of cytokinesis in 10 fittest myo1Δ e-strains

(A–C) Re-introduction of Myo1-GFP to wild type and e-strain cells, which also express GFP-Tub1. A) Schematic representation of the localization of Myo1-GFP and GFP-Tub1. B) Representative images of wild type and 7a-2 e-strain. Scale bar: 5 μm. C) Quantification of the Myo1-GFP localization at the bud neck. 3–7 different clones and >100 cells/clone were analyzed for each strain. Shown are average and standard error of the mean (SEM) across the different clones. (D) Septum morphologies observed in wild type (top row) and e-strains (morphology I-IV) by TEM. Arrows point to growing septal structures. Scale bars: 1 μm. (E) Frequency of four septum morphologies in wild type and e-strains. Bold numbers: predominant morphology for each strain. Note that morphology I for wild type refers to normal septum morphology (see text). (F) Membrane dynamics during cytokinesis observed with Ras2-GFP. >5 time-lapse movies were recorded for each strain. Representative frames are shown for each strain. Inset numbers represent time (min) relative to spindle breakdown (t = 0). Arrows point to tip of inward membrane protrusions during cytokinesis. Scale bars: 1 μm.

Figure 3. Transcriptome changes underlying the evolved phenotypes

(A) Schematic diagram of the transcriptome analysis (see text for details). (B) Subset of FatiScan Gene Ontology (GO) analysis applied to the correlation profiles (see Figure S6 for complete results). (C) Microarray expression data for HSP82 and HSC82 genes across the ten fittest e-strains, color-coded based on preferred septum morphology (black: I; red: III; blue: IV). Data are represented as mean ± SEM. (D) Growth of wild type and e-strains on 1% DMSO YEPD plates with or without radicicol (60 μM). wt_i: wild type at P1; wt_f: wild type at P20. (E) Effects of radicicol on cytokinesis. Cultures of wild type and e-strains were diluted to OD₆₀₀ = 0.1 and grew in YEPD with or without radicicol (80 μM) over 24 hours, and cytokinesis defect (percentage cells with ≥3 bodies, represented in the heat map) was determined at 0, 6, 12 and 24 hrs. (F) Growth of wild type and e-strains on YEPD plates with or without cycloheximide (0.1 μg/ml).

Figure 4. myo1Δ e-strains are characterized by polyploidy and aneuploidy

(A) Ploidy of the 29 stable myo1Δ e-strains inferred from flow cytometry data presented in Figure S8. (B) Correlation between polyploidization and restoration of cytokinesis. Data was taken from Table S3. Restored cytokinesis: CI >0.46; not restored cytokinesis: CI <0.46; P value determined by Fisher’s exact test. (C) Comparison of cytokinesis defect (as in Figure 3E legend) during parallel adaptive evolution of myo1Δ diploids (7) and haploids (22) at 12h after germination and at passage 2, 3 or 4. Symbols and error bars respectively represent average and SEM across all strains of a given genotype. (D) Cytokinesis defect in myo1Δ haploids and diploids at passage 2. Histograms represent number of strains with different degrees of cytokinesis defect. P value was determined by a one-tailed, two-sample t-test, assuming unequal variances. (E) aCGH of ten fittest myo1Δ e-strains and the wild-type haploid at P20. Hybridization intensity from all 16 chromosomes and the mitochondrial genome are plotted as log₂ ratios to that of the wild type at P1. Gray lines: x-axis. Upper and lower boundaries are truncated at a log₂ ratio of +1 and −1, respectively. (F) Inferred karyotypes of the 10 fittest myo1Δ e-strains.

Figure 5. Effects of aneuploidy on gene expression

(A) Side-by-side comparison of the heat maps of chromosome median log₂ ratios (aCGH data, Figure 4E) and chromosome median gene expression changes (microarray data, Figure 3A) in the 10 fittest e-strains. (B) Linear regression of chromosome median gene expression stoichiometry vs. chromosome copy number stoichiometry. Stoichiometries were determined as described in Supplemental Experimental Procedures. The correlation value r and the P value from a Pearson’s correlation test of the linear regression through the plotted points are shown. (C) An example of individual gene expression changes along a single chromosome in an e-strain. Genes are color-coded based on the number of standard deviations their expression change deviates from the chromosome average gene expression change. Black, gray and red respectively denote genes with expression change <1, 1–3 and >3 standard deviations from chromosome average change. (D) Two hypotheses to link outlier gene expression to karyotype. (E) Distribution of outlier genes between euploid and aneuploid chromosomes in the ten fittest e-strains, normalized to chromosome gene content. Histograms and error bars represent averages and SEM, respectively. (F) Correlation between expression of inlier genes and expression of outlier genes. For all possible (45) non-redundant pairs of e-strains, the Euclidean distance between the expression profiles of the two strains was calculated for those genes that were detected as outliers in at least one of the 10 e-strains (outlier genes, y-axis) and those genes that were not outliers in any of the 10 fittest e-strains (inlier genes, x-axis). A linear regression is superimposed on the plot. The correlation value and the P value from a Pearson’s correlation test are shown. (G) Shown are percentages of outlier (black bars) or non-outlier genes (white bars) that have upstream TFs encoded on aneuploid chromosome(s) over total number of each gene type in e-strains; P values evaluated by Fisher’s exact tests. (H) Shown are percentage of aneuploid chromosome-encoded TFs of outlier (black bars) and non-outlier genes (white bars) over total TFs of each gene types in e-strains; P values evaluated by Fisher’s exact tests.

Figure 6. Effects of increased copy numbers of RLM1 and MKK2 in myo1Δ cells

(A) Increased expression of RLM1 targets that were identified as outlier genes in e-strains 23a-1 and 23b-1. Histograms show average fold change and standard deviation across three biological replicates, measured by microarray (black) or real-time PCR (white). (B) Increased viability of myo1Δ spores with extra copies of RLM1 or MKK2 or both, compared to the viability of myo1Δ spores; P value determined by a Fisher’s exact test. (C) Comparison of cytokinesis defects of freshly generated (viable colonies grown after tetrad dissection) myo1Δ strains with or without extra copies of RLM1 and MKK2. Histograms represent number of strains with different degrees of cytokinesis defect. P value was determined using a one-tailed, two-sample t-test, assuming unequal variances. (D–E) Equal-representation (by OD₆₀₀) pools of 7 viable myo1Δ strains and 19 viable myo1Δ strains with extra RLM1 and MKK2 were analyzed by TEM. D) Representative micrographs are shown for each pool. Scale bars: 1 μm. E) Quantification of septum morphologies in myo1Δ strains with or without extra RLM1 and MKK2 (>200 randomly chosen cells analyzed for each pool); P value determined by a Fisher’s exact test.

Figure 7. Conceptual model of myo1Δ adaptive evolution

The scheme represents a conceptual model explaining evolvability of myo1Δ yeast. Solid lines represent the pathway predominantly observed in this study; dashed lines represent other potential contributing events.