Adaptive amplification and point mutation are independent mechanisms: evidence for various stress-inducible mutation mechanisms

Abstract

"Adaptive mutation" denotes a collection of processes in which cells respond to growth-limiting environments by producing compensatory mutants that grow well, apparently violating fundamental principles of evolution. In a well-studied model, starvation of stationary-phase lac(-)Escherichia coli cells on lactose medium induces Lac(+)revertants at higher frequencies than predicted by usual mutation models. These revertants carry either a compensatory frameshift mutation or a greater than 20-fold amplification of the leaky lac allele. A crucial distinction between alternative hypotheses for the mechanisms of adaptive mutation hinges on whether these amplification and frameshift mutation events are distinct, or whether amplification is a molecular intermediate, producing an intermediate cell type, in colonies on a pathway to frameshift mutation. The latter model allows the evolutionarily conservative idea of increased mutations (per cell) without increased mutation rate (by virtue of extra gene copies per cell), whereas the former requires an increase in mutation rate, potentially accelerating evolution. To resolve these models, we probed early events leading to rare adaptive mutations and report several results that show that amplification is not the precursor to frameshift mutation but rather is an independent adaptive outcome. (i) Using new high-resolution selection methods and stringent analysis of all cells in very young (micro)colonies (500-10,000 cells), we find that most mutant colonies contain no detectable lac-amplified cells, in contrast with previous reports. (ii) Analysis of nascent colonies, as young as the two-cell stage, revealed mutant Lac(+)cells with no lac-amplified cells present. (iii) Stringent colony-fate experiments show that microcolonies of lac-amplified cells grow to form visible colonies of lac-amplified, not mutant, cells. (iv) Mutant cells do not overgrow lac-amplified cells in microcolonies fast enough to mask the lac-amplified cells. (v)lac-amplified cells are not SOS-induced, as was proposed to explain elevated mutation in a sequential model. (vi) Amplification, and not frameshift mutation, requires DNA polymerase I, demonstrating that mutation is separable from amplification, and also illuminating the amplification mechanism. We conclude that amplification and mutation are independent outcomes of adaptive genetic change. We suggest that the availability of alternative pathways for genetic/evolutionary adaptation and clonal expansion under stress may be exploited during processes ranging from the evolution of drug resistance to cancer progression.

Figure 1. Colony Morphologies

(A) Point-mutant Lac⁺ colony showing solid blue color (the pale colonies are derived from Lac⁻ cells). (B) lac-amplified colonies showing sectoring caused by the instability of the amplified array. Cells from these colonies grow either into sectored blue colonies or, if they have lost the amplification, into white colonies, a phenotype that we call “unstable.” (C) A sectored colony that is not unstable in that it was found to contain only stable blue and stable white cfu upon retesting. (D) An example of a microcolony of the sort used in this work. The visible colony on the lower edge of the field has a diameter of 1.4 mm (>10⁸ cells). (E and F) Phase contrast (E) and green fluorescence (F) of the same field, showing two of 30 SMR6039 cells fluorescing.

Figure 2. Stringent Analysis of Whole Microcolonies: Analyzing All Cells in a Microcolony and Reducing Contamination by Unrelated Neighboring Bacteria

To analyze all cells in a microcolony, very young microcolonies were harvested (10³–10⁴ cfu/microcolony; see Figure 1D; Table 1). To reduce contamination with neighboring Lac⁻ bacteria and unrelated Lac⁺ microcolonies from the minimal-lactose selection plate, first, a minority of the Lac⁻ cells plated carried an antibiotic-resistance marker, and only resistant microcolonies were analyzed, and, second, sectored colonies observed were retested for instability (to eliminate sectored colonies that are not unstable, shown in Figure 1C, which may result from accidental overlap of blue and white cfu). Procedures described in Materials and Methods, “whole microcolony analysis.” Cam^R, chloramphenicol resistant; Kan^R, kanamycin resistant; Tet^R, tetracycline resistant. Streptomycin resistance (not shown) was also used.

Figure 3. Temporal Distribution of lac-Amplified Microcolonies

Data from six adaptive mutation microcolony experiments were pooled to give the distribution of point-mutant and lac-amplified phenotypes. Squares, the mean percentage of Lac⁺ microcolonies with unstable phenotype (± SEM; n = 3 experiments on day 3 and n = 4 on days 4, 5, and 6); diamonds, the percentage of lac-amplified visible colonies on the same days included for comparison (data from Hastings et al. 2000). Note that the observed proportion of unstable microcolonies is higher than the actual proportion because lac-amplified clones are slower growing than point mutants (Hastings et al. 2000), and so spend more time as microcolonies before becoming visible colonies. Thus, the lack of unstable microcolonies on early days is even more severe than the data show.

Figure 4. Fate of Microcolonies Developing In Situ

Microcolonies on days 4 and 5 of separate adaptive mutation experiments were sampled by touching with a needle, then the sample was spread on rich X-gal medium and scored for point-mutant or sectored Lac⁺ phenotype (see Materials and Methods). The microcolonies were then left to grow into visible colonies (∼10⁷ cells), and the colonies sampled again and samples plated to score for sectoring (hatched bars, day 4 microcolony samples; black bars, day 5 microcolony samples). The numbers of each type of colony observed are shown next to the bars. One single stable microcolony produced a mixed visible colony, which may have been caused by overlap with another colony or microcolony.

Figure 5. Pol I Is Required for Adaptive Amplification and Not Point Mutation

Strains were plated on lactose-minimal medium and Lac⁺ colonies counted daily (see Materials and Methods). The plots are cumulative, showing the mean of 3–4 cultures with one SEM. Strains used: FC40 dinB⁺ polA⁺ fadAB::Tn10Kan, SMR3490 (squares); FC40 dinB⁺ polA12(Ts)fadAB::Tn10Kan, SMR3491 (diamonds); FC40 dinB10 polA12(Ts)fadAB::Tn10Kan, PJH308 and PJH309 (triangles, inverted triangles); and FC40 dinB10 pol⁺ fadAB::Tn10Kan, PJH310 (circles). All cultures were grown at 30 °C and the experiments conducted at 37 °C. (A) An example of the effect of polA(Ts) on the yield of lac-amplified colonies, showing a partial requirement for polA at a semi-permissive temperature. (B) and (C) show the effect of the dinB10 mutation on adaptive lac-amplification and point mutation, respectively. The dinB polA(Ts) cells display the decreased lac amplification of the polA mutant (B), and the decreased point mutation characteristic of the dinB mutant (C), demonstrating that the decrease in lac amplification rate does not result from channeling of lac-amplified cells into a point mutation pathway. These data (C) also show that the absence of Pol I increases point mutation (reported previously, Harris 1997) in a completely DinB-dependent manner. This could occur via the absence of Pol I leading to SOS induction (Bates et al. 1989) and more DinB/Pol IV, or via relief of a competition between high-fidelity Pol I and error-prone Pol IV at the replisome. Neither of these ways should affectlac amplification, which is Pol IV–independent, as observed (B).

Figure 6. DNA Amplification Does Not Induce the SOS Response

(A) Known lac-amplified and point-mutant (control) derivatives of SMR6039 were grown in liquid M9 medium containing either lactose or glycerol. Mid-logarithmic phase cells were harvested and scored microscopically for GFP fluorescence, using an Olympus BL51 microscope mounted with a mercury lamp UV source and a High Q Endow GFP emission fluorescence filter cube. Some 1,000–2,000 cells from each of 4–10 fields were scored per determination. Error bars indicate one SEM for 8–13 cultures as indicated. (B) Microcolonies were harvested and suspended in 500 μl of buffer, 50 μl of which was spread on LBH X-gal rif solid medium to determine sectoring in resulting colonies. The remainders were concentrated and examined microscopically for GFP fluorescence, counting 60–400 cells per microcolony. These experiments were plated at very low cell density to avoid significant numbers of background Lac⁻ cells being harvested with the microcolonies. Open bars, fraction of stable Lac⁺ isolates (32 microcolonies); black bars, fraction of sectored isolates (11 microcolonies). The two distributions do not differ (p = 0.8 by Student's t-test). These experiments were repeated, giving similar results.

Figure 7. Stress-Response Models for Adaptive Amplification and Point Mutation

Modified from Lombardo et al. (2004) and Rosenberg and Hastings (2004a). DSBR, DSB repair; hypermutation, increased mutation at lac and elsewhere. The origins of DSEs during starvation on lactose medium in this assay system are unknown, and possibilities are reviewed elsewhere (Rosenberg 2001). On the F′ plasmid carrying the lac gene, DSEs may be frequent and may be derived from chronic single-strand nicks at the origin of transfer. However chromosomal mutation during starvation on lactose medium in these cells also requires DSB repair proteins (Bull et al. 2001), and so probably results from (lower levels of) DSEs generated in the chromosome, independently of the transfer origin, or perhaps by transient integration of the F′ into the chromosome (Hfr formation). Both adaptive point mutation and amplification are proposed to be outcomes of RpoS-dependent stress response, which both mechanisms require (Lombardo et al. 2004), and to result from alternative error-prone ways of repairing DSBs.