Adaptive Mutation Requires No Mutagenesis— Only Growth Under Selection: A Response

IN a bacterial system devised by Cairns and Foster The letter from Cairns and Foster (2003, this issue) (1991), a lac mutant population, starved in the pressuggests that our calculations are in error because we ence of lactose, gives rise to 100 revertant colonies extrapolated from rates observed or inferred for genes over 6 days. In this time, the plated population is not on an F plasmid to predict the number of mutations growing and is not experiencing general mutagenesis. that would be caused in the chromosome. Their arguThree models have been proposed to explain this. The ment is based on three assumptions: (1) that mutation directed mutation model (DMM) suggests that stress rates are 100-fold lower in the chromosome than on induces mutagenesis that is focused on the relevant the F , (2) that stress induces mutagenesis regardless target (lac, or the F plasmid that carries it) to the excluof where lac is located, and (3) that reversion is not sion of the chromosome at large (Cairns et al. 1988; detectably enhanced when lac is in the chromosome, Cairns and Foster 1991; Foster and Cairns 1992). simply because the basal reversion rate of a lac allele at The hypermutable state model (HSM) suggests that that position is too low. We do not accept these assumpstress induces genome-wide mutagenesis in a subset of tions. We have moved the lac allele used in the Cairns the population (105 cells); this generates Lac revertants experiment to 34 different sites in the chromosome and but kills the rest of the mutagenized population (Hall compared unselected reversion rates to that of the same 1990; Torkelson et al. 1997). According to the HSM, allele on the F plasmid. At all chromosomal sites the mutation appears directed because only Lac revertants unselected reversion rates clustered around the 10 8 survive mutagenesis. We have proposed (see below) the value found for lac on F 128 (Slechta et al. 2002a; S. amplification mutagenesis model (AMM), in which Slechta, unpublished data). The chromosomal revergrowth under selection increases revertant number with sion rates were not affected by the presence of F 128 no required change in the rate or target specificity of (Slechta et al. 2003). The 100-fold lower mutation rate mutation (Andersson et al. 1998; Hendrickson et al. suggested for chromosomal lac in the Cairns-Foster let2002; Slechta et al. 2003). ter would be reasonable for a typical chromosomal The article under discussion (Roth et al. 2003) examframeshift mutation (Bull et al. 2001), but the lac allele ines some quantitative predictions of the HSM and finds used in this system is not typical. That allele has a constelthat this model requires an implausibly high intensity lation of three mutations—an IQ promoter, a 1 of genome-wide mutagenesis—vastly higher than that frameshift in lacI, and a deletion that fuses the lacI and estimated experimentally (Rosche and Foster 1999; lacZ genes (Cairns and Foster 1991). This combinaSlechta et al. 2002b). If realized, this mutation rate would tion provides a leaky Lac phenotype that can be coradd so many lethal mutations that one could not recover rected by any 1 mutation (or other mutation having the number of lac mutations that are observed. We arthe same effect on reading frame) in 100 bp (about gued that the behavior of the Cairns-Foster system must one-tenth of the lacI gene sequence). This allele would be explainable by some means other than the HSM. therefore be expected to revert at 100 times the rate of a typical 1 frameshift, and our measurements suggest that it does. There seems to be no effect of gene 1Corresponding author: Section of Microbiology (DBS), 1 Shields Ave., position on reversion in the absence of selection. University of California, Davis, CA 95616. E-mail: [email protected] When selection is imposed, a position effect is seen. 2Present address: Microbiology Section (DBS), University of California, Davis, CA 95616. When lac is on the F plasmid, 100 revertants accumulate