Fig 1.
Schematic illustration of the quantification of phenotypic penetrance.
Fig 2.
(A) Phenotypic penetrance (mean ± SE; n = 6) over time for three antibiotic resistance mutations. Gray dashed lines: time at 50% phenotypic penetrance. (B) Frequency of homozygous mutants among all mutants (orange) for the three resistance mutations assessed by lacZ reporter constructs (rpoB-lacZ, gyrA-lacZ, rpsL-lacZ), overlaid with their respective phenotypic penetrance. (C) Genotypic mutant frequency for the resistance mutations. (D) Phenotypic penetrance of the lactose prototrophy (rpsL-lacZ) mutation. (E) Colonies founded by homozygous (blue) and heterozygous (sectored) lac+ mutants. The numerical data for panels A to D can be found in S1 Data. MIC, minimum inhibitory concentration.
Fig 3.
Single-cell analysis of fluorescent mutants.
(A) Overlay of phase-contrast and fluorescence images showing a microcolony containing fluorescent mutants (see also S1 Movie). Yellow arrow: the first cell showing significantly higher fluorescence than background in the given frame. Accounting for the time required for YFP protein folding and maturation, the ssDNA integration is estimated to have happened before the first division of the labeled cell. (B) Genealogy of the aforementioned microcolony. The yellow arrow indicates the cell in A, while the remaining arrows indicate three lineages in which fluorescence was quantified. (C) YFP expression history of three lineages in B showing fully, transiently, and non-fluorescent phenotypes (green, blue, and red, respectively). Yellow dashed line: onset of fluorescence. Black and grey dashed lines correspond to the black and grey arrows in B. Black: emergence of the first homozygous mutant; grey: its first division. (D) Distribution of time to form 34 homozygous mutant lineages from 25 microcolonies. The data are obtained by directly analyzing genealogies as in B and compiled from two separate experiments. The dashed grey line indicates the estimated generations to form half of the homozygous mutant lineages. (E) Photo of a microcolony with one filamentous fluorescent cell. (F) The distribution of number of generations to form homozygous mutant lineages sorted by the presence/absence of filamentation. The numerical data for panels C, D, and F can be found in S1 Data. ssDNA, single-stranded DNA; YFP, yellow fluorescent protein.
Fig 4.
Chromosomal location effects on recombineering.
Resistance target genes rpoB, gyrA, and rpsL are at 5.5%, 34.2%, and 9.7% genomic distance from the replication origin, respectively. Strong negative correlation (quantified by Pearson’s correlation coefficient, R) exists between the distance from the origin and the initial frequency of genotypic mutants induced by recombineering, for both (A) the resistance mutations and (B) their lacZ reporters. (C) The inferred minimal ploidy from the cell physically treated by recombineering at the time of mutant ssDNA integration shows a similar distribution for all constructs, regardless of their distance from the origin (n = 31 colonies examined per construct). Eight is the median and the most common inferred minimum ploidy, consistent with previous estimates of ploidy in E. coli [10,22]. Higher estimates of 16, 32, and 64 could have resulted from delayed ssDNA integration or ssDNA integration in a filamentous cell. The numerical data can be found in S1 Data. ssDNA, single-stranded DNA.
Fig 5.
Reconciling Luria-Delbrück fluctuation test with phenotypic delay by effective polyploidy.
(A) The original Luria-Delbrück mutation model disregards polyploidy. For instance, a phenotypic delay of two generations results in four mutants appearing at once. (B) The observation of many one-mutant (“singleton”) populations was interpreted as evidence against the existence of a delay [1,3]. (C) With polyploidy considered, cells with four genome copies require two divisions to generate a homozygous mutant that expresses a selectable recessive phenotype. Therefore, a delay of two generations can generate just one mutant. (D) Heterozygous cells containing recessive mutations will not survive selection, leading to an underestimation of mutational events.
Fig 6.
Mutation rate estimates (MLEs, filled squares; and 95% confidence intervals, error bars) from 50 simulated parallel cultures at each ploidy (c), with constant mutant interdivision time, assuming either a recessive (A) or dominant (B) mutation. The lower solid line and upper dashes indicate the per-copy (μc = 3 × 10−10) and per-cell (c·μc) mutation rates, respectively, used for simulation. For c = 4 and recessive (C) or dominant (D) mutations, the observed mutant count distribution (histogram) is compared to that predicted by the standard model parameterized by the MLE mutation rate (connected points). This figure can be reproduced using code and simulated data deposited on Dryad (http://dx.doi.org/10.5061/dryad.8723-t). MLE, maximum likelihood estimate.
Fig 7.
Schematic of the model used to evaluate SGV and rescue, illustrated for ploidy c = 4.
(A) Flow diagram of all events. A cell is represented by a grey oval, containing four chromosomes (complete or partial, so long as they contain the gene of interest). These chromosomes are colored blue if wild-type at the gene of interest or red if mutant. A cell either divides to produce two daughter cells with type-specific probability pj or otherwise dies. These probabilities pj differ between the old environment (to model SGV) and the new environment (to model rescue). Upon type 0 division, mutation (producing type 1) occurs with probability in each daughter cell; otherwise, the daughter is also type 0. In the remaining types, chromosome segregation determines the types of the daughter cells. (B) A mechanistic view of chromosome replication and segregation, illustrated for the production of one type 2 and one type 0 daughter cell from a type 1 mother cell. On each chromosome, the black dash indicates the origin of replication. SGV, standing genetic variation.
Table 1.
Approximate mutant frequencies at mutation–selection balance.
Ploidy is c = 2n (n ≥ 1) in the polyploid cases, is the per-copy mutation rate, and s is the cost of the mutation in homozygotes (in heterozygotes, the cost is masked in the recessive case but expressed in the dominant case).
Fig 8.
Impact of effective polyploidy on the probability of evolutionary rescue.
(A, C) The probability that at least one mutation from the SGV survives in the new environment (PSGV; linear scale) and (B, D) the probability that at least one mutation arises in the new environment and survives (PDN; log scale), plotted as a function of ploidy, for a recessive mutation (A, B) or a dominant mutation (C, D). The different colored curves indicate probability of division before death of phenotypically wild-type (sensitive) cells in the new environment, pS, varying from 0 (black) to 0.45 (magenta) in increments of 0.05. The remaining parameters are fixed: probability of division before death of phenotypically mutant (resistant) cells, pR = 0.9; mutational cost in the old environment, s = 0.1; population size at the time the environment changes, N = 2 × 108; and per-copy mutation rate, . This figure can be reproduced using code deposited on Dryad (http://dx.doi.org/10.5061/dryad.8723t). SGV, standing genetic variation.