Experimental Analysis of Molecular Events During Mutational Periodic Selections in Bacterial Evolution

Corresponding author: Thomas Ferenci, Department of Microbiology G08, University of Sydney, NSW 2006, Australia., t.ferenci{at}microbio.usyd.edu.au (E-mail)

Communicating editor: R. MAURER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A fundamental feature of bacterial evolution is a succession of adaptive mutational sweeps when fitter mutants take over a population. To understand the processes involved in mutational successions, Escherichia coli continuous cultures were analyzed for changes at two loci where mutations provide strong transport advantages to fitness under steady-state glucose limitation. Three separate sweeps, observed as classic periodic selection events causing a change in the frequency of neutral mutations (in fhuA causing phage T5 resistance), were identified with changes at particular loci. Two of the sweeps were associated with a reduction in the frequency of neutral mutations and the concurrent appearance of at least 13 alleles at the mgl or mlc loci, respectively. These mgl and mlc polymorphisms were of many mutational types, so were not the result of a mutator or directed mutation event. The third sweep observed was altogether distinct and involved hitchhiking between T5 resistance and advantageous mgl mutations. Moreover, the hitchhiking event coincided with an increase in mutation rates, due to the transient appearance of a strong mutator in the population. The spectrum of mgl mutations among mutator isolates was distinct and due to mutS. The mutator-associated periodic selection also resulted in mgl and fhuA polymorphism in the sweeping population. These examples of periodic selections maintained significant genotypic diversity even in a rapidly evolving culture, with no individual "winner clone" or genotype purging the population.

Akey hypothesis in bacterial evolution is that period- ic selections reduce diversity by displacing less-adapted bacteria (SELANDER et al. 1987 ; MILKMAN 1997 ). Purging by periodic sweeps is one of the explanations used to explain the clonal nature of extant bacterial populations (SELANDER et al. 1987 ; MILKMAN 1997 ). Despite this significance, some basic aspects of the periodic selection hypothesis were never tested directly in experimental populations, namely: (1) Is a periodic sweep due to a single winner genotype and (2) does it reduce genetic diversity? Results with long-term cultured populations are increasingly inconsistent with the idea that mutational diversity is low in evolving experimental populations (ROSENWEIG et al. 1994 ; FINKEL and KOLTER 1999 ; NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ; PAPADOPOULOS et al. 1999 ). It was therefore essential to investigate a central feature of the periodic selection hypothesis, namely whether any particular periodic selection event is due to a sweep by a single winner genotype.

Bacteria carrying neutral mutations constitute a fluctuating proportion of growing cultures. The fluctuations are attributed to periodic selection of fitter clones, with each successive sweep replacing less fit members of the population, including those with neutral mutations (NOVICK and SZILARD 1950 ; ATWOOD et al. 1951 ; NOVICK 1958 ; KUBITSCHEK 1974 ; LEVIN 1981 ; DYKHUIZEN 1990 ; BERG 1995 ). The frequency of neutral mutations can also change in clonal populations as a consequence of hitchhiking with favorable mutations. Although evolutionary theories for hitchhiking are persuasive (MAYNARD-SMITH 1991 ), experimental evidence of such events has relied on detective work with natural populations and their uncertain histories (GUTTMAN and DYKHUIZEN 1994 ; NURMINSKY et al. 1998 ). The experimental system described here revealed for the first time the hitchhiking of neutral mutations to T5 resistance with a strongly selected mgl mutation in an experimental population.

Another important question addressed in this study is the role of random spontaneous mutation and mutator-induced mutations in evolving populations. There is ample evidence that the presence of mutators with reduced DNA repair is an advantage in continuous cultures (CHAO and COX 1983 ). Mutator mutations have been observed in long-term recycled batch cultures (SNIEGOWSKI et al. 1997 ; TADDEI et al. 1997 ) and occur at a finite level in natural bacterial populations (LECLERC et al. 1996 ). There is both an advantage and a cost in maintaining high mutation rates, so theoretical considerations suggest mutators causing a 100-fold increase in mutation rates may appear transiently in populations (TADDEI et al. 1997 ). Fortuitously, we found that one of the periodic changes in T5 resistance was associated with the transient appearance of a mutator mutation in one of the populations. We describe here the time course of appearance of the mutator mutation in an experimental population and its association with the generation of advantageous mutations.

The major hurdle in testing the nature of bacterial adaptive events is in identifying which of several thousand genes to analyze in an evolving population, which is something of a needle in a haystack problem (LENSKI et al. 1998 ). The recent analysis of mutations leading to improved fitness of Escherichia coli under glucose limitation has identified several loci universally mutated under a particular selection condition (MANCHE et al. 1999 ; NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ). Three regulatory loci (mlc, mglD/O, and malT) were commonly affected within 280 generations of aerobic, glucose-limited continuous culture. Finding these target genes opens the way for a molecular analysis of periodic selection processes since regulatory mutations in mlc, mglD/O, and malT genes are identifiable by phenotypic tests. Sequence changes at these loci can then be analyzed during the progress of periodic selection events in glucose-limited continuous cultures.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Bacterial strains and culture conditions:
All bacterial strains used in this study are derivatives of E. coli K12. BW2952, an MC4100 derivative with a malG-lacZ transcriptional fusion (NOTLEY and FERENCI 1995 ) was used as the starting strain in all long-term chemostat experiments. The basal salts medium used in all experiments was minimal medium A (MMA; MILLER 1972 ) supplemented with glucose, lactate, or glycerol as specified for each experiment. Batch cultures contained 0.4% sugar and were harvested during midexponential growth. Tryptone agar (TA) plates contained 5 g/liter NaCl, 10 g/liter tryptone, and 1.5% agar.

Glucose-limited chemostats (80 ml) were set up as previously described (DEATH et al. 1993 ). The glucose concentration in the feed medium was 0.02%. The chemostats were maintained for several weeks at a dilution rate of 0.3 hr^-1. Populations were sampled every 5–10 generations (once or twice daily). Culture (10 ml) was collected aseptically from the overflow port and subjected either directly to T5-resistance assay or stored in 40% glycerol at -70° for later use. Samples were streaked on nonselective nutrient agar and randomly separated colonies were purified and numbered (e.g., sampling day 1, A; day 2, B; etc.) for further testing.

T5 phage resistance assays:
The frequency of T5-resistant mutants in a population was determined by the method of HELLING et al. 1987 . Samples from the chemostat population were mixed directly with a suspension of bacteriophage T5 (kindly supplied by K. Heller) at a multiplicity of 50, and CaCl₂ present at 5 mM. Generally, 100 µl of chemostat sample (2–3 x 10⁷ bacteria) was mixed directly with 100 µl of T5 phage (10⁹ pfu), except when the proportion of T5-resistant mutants rose, in which case a 10-fold dilution of the chemostat sample was required. The mixed suspensions were held on ice for 15 min to allow absorption before spreading 50-µl aliquots in duplicate onto TA plates containing 5 mM CaCl₂ and incubating overnight at 37°.

Mutator assay:
For initial qualitative screening of elevated mutation rates in isolates, a single colony of each isolate to be tested was inoculated into 4 ml Luria broth (LB). After overnight growth at 37°, cultures were centrifuged for 10 min at 4000 x g and resuspended in 0.5 ml LB. A total of 10 µl of the suspension (10⁸ cells) was spotted (9 spots per plate) on LB agar containing rifampicin at 100 µg/ml. The plates were incubated overnight at 37° and the number of rifampicin-resistant colonies in each spot was counted. Isolates producing <10 rifampicin-resistant colonies per spot were considered to have wild-type mutation rates, while >50 colonies were indicative of a mutator phenotype. The latter isolates as well as those showing intermediate numbers were retested at least twice to exclude the possibility of random jackpot events. A more quantitative estimate of mutation rate was obtained by spreading 250 µl of the overnight cultures of individual isolates (or less in the case of high mutation rates) onto rifampicin plates as well as on nutrient agar plates at the appropriate dilutions to determine total counts. Rate of mutation to rifampicin resistance was calculated per 5 x 10⁸ bacteria plated.

Phenotypic tests for mgl, mlc, and malT mutations:
To detect mgl-con mutations, the initial rate of uptake of 1 µM [U-¹⁴C]galactose by glycerol-grown chemostat isolates was determined using bacteria resuspended to an A₅₈₀ of 0.2 as described previously (DEATH and FERENCI 1993 ).

For mlc changes, the increased sensitivities to the PtsG and PtsM substrates, methyl--glucoside and 2-deoxyglucose, respectively, were determined by plating lactate-grown isolates onto glycerol minimal agar plates (CURTIS and EPSTEIN 1975 ) overlaid with a 6-mm disk containing 10 µl 20% w/v methyl--glucoside or 20% w/v 2-deoxyglucose. Zones of inhibition were measured after overnight incubation at 37°.

For malT, ß-galactosidase activity of the malG-lacZ fusion was measured on glycerol-grown isolates treated with chloroform and SDS by the method of MILLER 1972 .

Mutation analysis:
PCR amplification of the mglD/galS sequence and the mgl operator used primers GalSF1 (5'-GCAAT CTCATAACAGGTAGTG-3') and GalSR2 (5'-ATGACGCTGT TACCTCGGC-3'). The reaction profile consisted of 34 cycles of denaturation at 94° for 30 sec, followed by annealing at 58° for 30 sec, and extension at 72° for 1.5 min in a DNA thermal cycler (Perkin-Elmer Cetus, Norwalk, CT). PCR amplification of the mlc gene used two primers: MlcF1 (5'-CTGAATGCTCT CAGGTGAGG-3') and MlcR1 (5'-CTCCACCGTTATGCTTC AC-3'). The reaction profile was similar to that for mgl except primers were annealed at 60°. The fhuA gene was sequenced using four primers: FhuAF1 (5'-CCCATCTAAGATATTA ACCC-3'), FhuAF2 (5'-CATTATCTGGCACGTAAATAC-3'), FhuAR1 (5'-CGCATACGCATAAAGTCGAC-3'), and FhuAR2 (5'-CGATAACAGCCAACTTGTGA-3'). The reaction profile was as above except primers were annealed at 59°, while the extension interval was increased to 2 min. PCR products were purified directly with Wizard PCR preps DNA purification system (Promega Corp., Sydney, Australia). The nucleotide sequences were determined using the above primers and dye-terminator sequencing reactions and were analyzed on a catalyst robotic workstation. Mutations were located in mutant sequences by aligning with the known galS(mglD), fhuA, and mlc sequences in the E. coli genome database using software available in the Australian National Genomic Information Service (ANGIS), Sydney, Australia.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

As shown in Fig 1A and Fig B, for two independent glucose-limited chemostat populations, sharp fluctuations in the proportion of T5-resistant bacteria were observed in the first 200 generations of growth, very much in line with earlier findings (KUBITSCHEK 1974 ; HELLING et al. 1987 ). T5 resistance is used to follow neutral mutations in such studies since the alteration in resistant bacteria does not affect fitness under the culture conditions used (NOVICK and SZILARD 1950 ; KUBITSCHEK 1970 ). Mutations of E. coli to T5 resistance affect fhuA, one of several redundant iron transporters (BRAUN and KILLMANN 1999 ). The frequency of T5-resistant mutants increases steadily due to recurrent mutation to T5 resistance, but beneficial mutations occur predominantly in the majority, T5-sensitive, portion of the population. At the points shown in Fig 1A and Fig B, 10–20 randomly chosen members of chemostat populations were assayed for changes at mglD/O, malT, and mlc by measuring rates of galactose transport, ß-galactosidase activity (malG-lacZ fusion activity), or sensitivity to glucose analogues (-methyl-glucoside and 2-deoxyglucose), respectively (NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ). Isolates with increases in assayed levels relative to wild type were further analyzed for gene sequence changes after amplification by PCR.

View larger version (27K): In this window In a new window Download PPT slide   Figure 1. Phenotypic sweeps during periodic selection events in E. coli populations. To monitor periodic selection events in chemostats, the strain BW2952 (NOTLEY and FERENCI 1995 ) was inoculated into glucose-limited chemostats (glucose at 0.02% w/v) and fed with medium at a dilution rate of 0.3 hr-1 (10 generations per day) as described previously (DEATH et al. 1993 ). Growth was maintained with 2–3 x 1010 bacteria for 250–400 generations over a 3- to 5-week period. The populations were sampled every 5–10 generations and streaked onto LB-agar plates. The fluctuation in the population of bacteria carrying a T5-resistance mutation was determined by mixing the chemostat sample directly with T5 phage and plating on TA plates. At selected time points, 10–20 randomly chosen isolates were assayed for changes to rates of galactose transport, ß-galactosidase activity (malG-lacZ fusion activity), or sensitivity to -methyl-glucoside and 2-deoxyglucose, as described previously (NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ). (A) The profile of population 20 [one of a series of chemostat populations used in this laboratory (NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B )] and the increase in the proportion of mgl mutants between points D, E, and F. (B) Another population, population 26, was found to change in the proportion of mutS mutators (), mgl mutations (), and mlc mutations () at the points labeled in capital letters.

Identification of sweeps involving mgl and mlc mutations:
Fig 1A shows the time course of fixation of mgl-con mutations in one population (chemostat 20). In this example, the increase in rates of galactose transport was found to occur in isolates tested during the first adaptive shift. The timing and rate of decrease in the T5-resistance frequency between points D and F in Fig 1A corresponded directly to the increase in proportion of mgl-con mutants in the population, which went from 1/20 to 19/20 in the same period. This changeover was due to the mgl mutation increasing in frequency in the T5-sensitive majority members of the population, displacing T5-resistant mutants. None of the tested mgl mutants was resistant to T5 in this population.

In another culture (population 26), the spread of mgl mutations was also detected early in the lifetime of the population, as shown in Fig 1B. Between points E and F, isolates became predominantly mgl-con mutants. However, a major difference to the mgl sweep in Fig 1A was that it was accompanied by an increase in T5 resistance between points E and F in population 26. On isolating and testing individual members of this population, it was evident that 100% of T5-resistant isolates carried an mgl mutation by point F. In population 20, this was clearly not the case at point F in Fig 1A. Hence the simultaneous increase in resistance and mgl constitutivity in population 26 was due to the hitchhiking of T5 resistance with the selectively useful mgl mutation. These changes in population 26 represent a fortuitous trapping of a hitchhiking event.

Also striking was the steeper rate of increase of T5 resistance between points E and F in Fig 1B, faster than any other rate of increase in populations 20 or 26. This result raised the possibility of increased mutation rates in the population, most likely caused by the presence of mutator mutants in the culture. Hence the mutator phenotype of individual isolates from the culture was tested. As also shown by the plot in Fig 1B, mutators as a proportion of the population reached 33% at point F, concurrently with the spread of mgl (and T5-resistant hitchhikers) in population 26.

The appearance of mutator activity was transient and the downward sweep after point F led to the rapid elimination of the mutator from being a detectable member of the population in subsequent samples. Between points F and G in population 26, the sweep of a T5-sensitive, mgl-con, mutS⁺ organism with an additional strongly advantageous mutation limited the numbers of hitchhikers, which never became a dominant proportion of the culture. The height of the T5-resistance peak at point F was 10-fold higher than at the corresponding point in population 20 but still <1 in 10⁴ of the general population. One of the reasons why the T5-resistant, mgl-con double mutants did not become predominant was because mgl mutations also spread in the T5-sensitive members of the population (by point F, 66% of T5-sensitive isolates were altered in mgl). So the hitchhiker and mutator numbers were kept in check by these T5-sensitive, mutS⁺, mgl-con members arising in the bulk population.

Population 26 later acquired a second identifiable mutation, between points I and M in Fig 1B. The appearance of mlc mutants as frequent members of the population was associated with the third adaptive shift in population 26. At point I, there were 0/20 isolates containing mlc mutations, while 10/20 isolates harbored mlc mutations at point L, and finally all 20 isolates carried an mlc mutation by point O. As with the mgl sweep in population 20, the rate of drop in T5 resistance mirrored the increase in mlc mutant frequency in the population (checked by linear regression between points I and M, result not shown). The rate of takeover by mlc mutants was slower than for mgl mutants, as was the rate of decrease of T5 resistance. The proportion of T5-resistant mutants in the population already began to increase in the mlc portion of the population before the full takeover by mlc mutants. These properties would suggest mlc mutations were selectively weaker than mgl-con mutations in the respective genetic backgrounds. It is worth stressing that the mlc sweep occurred when all tested members of the population were already mutated at mgl. The results with the mlc sweep are consistent with the periodic selection model, insofar as the decrease in the frequency of T5-resistant mutants correlated with the occurrence and fixation of beneficial mutations in the T5-sensitive bulk population. Mutators were not detectable during the mlc sweep.

Besides the mgl and mlc sweeps discussed above, it is evident from the data in Fig 1 that numerous other periodic selections occurred in chemostats 20 and 26. The advantageous mutations causing other periodic selections in Fig 1 remain to be identified.

Molecular analysis of mutations in mgl and mlc associated with periodic selection events:
The mgl and mlc genes were amplified by PCR and sequenced from isolates displaying phenotypic changes to galactose transport or methylglucoside sensitivity. Five isolates with wild-type mgl phenotype were also sequenced with the mgl primers to test the fidelity of this approach, and all five gave wild-type sequences. The identified sequence changes in isolates at each tested time point are shown in Table 1 for the two mgl sweeps and for mlc mutations, respectively.

Most strikingly, the results in Table 1 show that the fixation of mgl and mlc mutations was not due to the spread of a single winner genotype. Rather, a polymorphic population of fitter isolates displaying different alleles was responsible for each of the shifts. In population 20, 13/19 different mgl-con alleles were found after the first adaptive shift at point F in Fig 1A. The alleles in Table 1 include mutations that result in slightly different transport effects; as shown in a previous study (NOTLEY-MCROBB and FERENCI 1999A ), mgl operator mutants exhibit higher transport than repressor mutations in mglD, as was also confirmed for the mutants in Table 1 (result not shown).

Similarly, in population 26, 4/10 different mlc alleles were found at point L, while 13 alleles coexisted among 20 tested at point O (Table 1). Only two of the alleles at point L were common to those at point O, hence 15 different mlc alleles were present in the population. Given the small number of isolates sampled relative to the size of the population, these 15 must be a subset of an even greater number of alleles in the culture as a whole and indicate a remarkable mgl and mlc polymorphism in these populations. The results are consistent with previous data on long-term experimental populations that showed allele diversity at each fitness locus after 280 generations of selection (NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ).

The nucleotide changes occurring in each sweep are also listed in Table 1. In the two sweeps not involving a mutator, a mixture of point mutations (mainly transversions), frameshifts, and short deletions, as well as insertion sequence-mediated disruptions, was observed. These were characteristic of random spontaneous mutations in E. coli cultures (TAKIMOTO et al. 1997 ). The polymorphisms in mgl and mlc could easily arise from the large pool of spontaneous mutations occurring during replication of the 10¹⁰ bacteria constantly in the population. At least in two of the sweeps, the synchrony of mgl or mlc mutations was not due to a single wave of directed or mutator-inspired mutations of a uniform nature.

In contrast to the two sweeps described above, the mgl spread in population 26 associated with the mutator was of a more restricted mutational spectrum. Among the mutator isolates noted in Table 1, the mgl mutations comprised purely transitions and frameshifts. Despite the small number of mutations characterized, the spectrum was suggestive of mutS-associated mutagenesis (COX et al. 1972 ). To test if this mutator gene was indeed affected in the chemostat isolates, the transductional mapping of the mutator was undertaken. The mutator mutation was linked by cotransduction to both srl and rpoS, consistent with the position of mutS (20% cotransduction with srl::Tn10 and almost 100% with rpoS::Tn10). Quantitative estimates of the mutation rates of selected isolates (26Ea18 and 26Fa3) were tested and gave rates of mutation to rifampicin-resistance of 2330 ± 553 and 2240 ± 577 per 5 x 10⁸ bacteria, respectively, compared to the wild-type rate (in strain BW2952) of 10 ± 2 rifampicin (Rif)-resistant colonies per 5 x 10⁸ bacteria. The resultant 250-fold increase in the rate of spontaneous Rif resistance is in line with previous studies of mutS null mutations (MILLER 1996 ).

A common feature of all the sweeps studied here, including the hitchhiking event, was the multiplicity of alleles at the target genes. Even the mgl sweep associated with the mutator in population 26 led to the detection of 9 different alleles from the population among 60 sequenced, as shown in Table 1.

Molecular analysis of the hitchhiking event in population 26:
Mutations at three separate loci (affecting mgl, fhuA, and mutS) started to become common in population 26 at point E in Fig 1B. To reveal the pattern of association between the three classes of mutation, the detailed characteristics of 18 random isolates from point E are shown in Fig 2A for the bulk population and in Fig 2B for the T5-resistant subpopulation.

View larger version (30K): In this window In a new window Download PPT slide   Figure 2. Profile of isolates during the mutator sweep at point E in population 26. Isolates were obtained by random selection of colonies obtained from the whole population (A) or from colonies obtained from T5-containing (-resistant) plates (B). The mutator phenotype was determined qualitatively in each isolate, the mgl phenotype assayed, and the mgl-constitutivity mutations sequenced as shown for each isolate with elevated galactose transport.

Most importantly, the mutator mutation was present at point E in isolates lacking either of the other two mutations. The presence of isolates 9 and 17 (Fig 2A) suggests the mutS mutation arose independently and did not become common by hitchhiking on the back of the mgl advantage. Presumably, the high frequency of mutators was due to hitchhiking with another advantageous mutation not identified in this study. It is worth noting that the mutation causing the first sweep at point C in Fig 1B was not identified and the likeliest possibility is that the mutS mutation arose in a bacterium containing the unknown advantageous mutation.

At point E, mgl mutations were still undetectable in the bulk population (Fig 2A), yet half the T5-resistant subpopulation was already mutated in mgl (Fig 2B). By point F, 17/18 T5-resistant isolates tested were mgl-con (not shown). Significantly, all the T5-resistant/mgl mutants at point E were also mutators, explaining the increased coincidence of multiple mutations. Also indicative of multiple events was the presence of two different mgl frameshift mutations among the T5-resistant bacteria at point E. This suggests there was more than one progenitor of mgl/fhuA double mutants.

To further explore the diversity of genotypes in the population, the fhuA gene encoding the receptor for phage T5 (HELLER 1992 ) was sequenced from T5-resistant mutants at point E. The analysis revealed that multiple alleles of fhuA were present together with the same mgl allele. The mgl G306 frameshift mutation was associated with four different fhuA alleles at point E (isolates 2, 14, and 16 contained a C > T change, causing a stop codon at Q133; isolate 7 contained a C > T substitution, causing an S170F substitution; isolate 8 contained a T frameshift in codon 365; isolate 13 contained a +C frameshift in codon 156; and isolate 3 contained a T at codon 191). Hence the hitchhikers themselves were genotypically heterogeneous and the peak of T5 resistance in Fig 2 was due to multiple combinations of mgl and fhuA mutations. The fhuA change in the mutS/mgl isolates was distinct from those in T5-resistant mutants analyzed from nonmutators; the majority of mutS⁺ isolates contained IS insertions, though at distinct sites and with different IS elements (results not shown).

To understand the genealogy of population 26, it is also relevant that the G306 mgl frameshift mutation common in the T5-resistant bacteria was also detected among T5-sensitive bacteria (at point F in Table 1). The likeliest scenario, shown in Fig 3, is that the mglD114 frameshift mutation originally arose in a mutS/fhuA⁺ background and acquired several independent mutations to T5 resistance. Most likely, the independent T5-resistance changes arose early in the mglD114/mutS background, permitting hitchhiking gains before point E, when the bulk population was still mostly wild type for mgl.

View larger version (25K): In this window In a new window Download PPT slide   Figure 3. Emergence of mutations in population 26. xxxA is the unidentified gene affected in the first sweep in population 26, between points C and E in Fig 1. The other alleles are described in Table 1, except for the fhuA mutants; fhuA100 was found in T5-resistant isolates 2, 14, and 16 at point E (Fig 2) and contained a C > T change causing a stop codon at Q133; fhuA101 was in isolate 7 and contained a C > T substitution causing an S170F substitution; fhuA102 was in isolate 3 and contained a T frameshift in codon 191; fhuA103 was in isolate 8 and contained a T frameshift in codon 365; fhuA104 was in isolate 13 and contained a +C frameshift in codon 156.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Aside from the pioneering work of Adams and co-workers (TREVES et al. 1998 ), these studies permitted for the first time an identification of molecular events associated with population shifts in experimental cultures. By focusing on loci known to be mutated in glucose-limited populations, three sweeps were characterized, two associated with spontaneous mutations and a third influenced by a mutator.

An important conclusion from all studied events was that a closed asexual population is not taken over by a single genotype when mutations sweep the population, particularly when the mutations arise as random spontaneous events. The mgl and mlc sweeps ensured a takeover by constitutive mutants with advantageous phenotypes (MANCHE et al. 1999 ; NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ) but with diverse alleles coexisting in the population. Each of the different spontaneous mgl or mlc mutations must have arisen in independent genomes and will hitchhike these along with the favorable mutation. Hence the 10¹⁰ original wild-type genomes were succeeded by a subset (minimally 13–15, probably dozens) of winner bacterial genotypes. The purging of the initial population was considerable, but not by a single winner clone. In effect, the observed sweeps were phenotypic and due to a synthetic allele.

It needs to be remembered that the population size and mutation rate will determine the number of alleles generated in each sweep (DYKHUIZEN 1990 ). Perhaps when population sizes are small and mutation rates match population size, individual advantageous mutations may still be rare enough to sweep as individual clones. Our experimental results are in accord with the views of KORONA 1996 , who stated, "The classical model of periodic selection is probably correct when adaptive mutations are rare. It may less accurately describe evolution in large populations where beneficial mutations are relatively abundant, genetically diverse and similarly effective." Our system has all the attributes recognized by Korona as leading to diversity in a bacterial population.

The mgl and mlc mutations of advantage in glucose-limited populations were both loss-of-function changes (NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ) and were likely to appear in 10⁴ bacteria at each round of replication of the large populations studied here. In natural niches, the population size of E. coli in an individual intestine is of the order of 10⁸–10¹³ bacteria (SAVAGE 1977 ), but the effective population size of E. coli is subject to debate. Nevertheless, whatever the precise number, polymorphic mutational sweeps must be common in nature. It needs stressing that the processes studied here were mutational sweeps, excluding rarer lateral gene transfer events that possibly play an even more important role in the history of extant bacterial populations (JAIN et al. 1999 ; LAWRENCE 1999 ; PROZOROV 1999 ). The rarity of these events in comparison to mutational sweeps may indeed lead to unique winner clones even in large populations.

In line with predictions (TENAILLON et al. 1999 ), the large population size also contributed to the establishment of mutators in the culture leading to one of the events analyzed here. A transient appearance of a mutS mutation coincided with one of the mgl sweeps. Uniquely, this population shift was associated with an increased incidence of neutral T5 resistance. In the other two described events, the mgl or mlc mutations appeared as more classic sweeps in the T5-sensitive members of the culture. The presence of mgl mutations in T5-resistant isolates in population 26 provided direct evidence for hitchhiking of the neutral mutation. The chance of the mgl-T5-resistance double mutation occurring in the same cell was enhanced by the high mutation rates in the mutS bacteria and indeed occurred with two independent mgl mutations, so even this sweep was not monoallelic with respect to mgl. The fhuA mutations were also distinct within the T5-resistant population, so the T5-resistant hitchhikers were definitely not of a single genotype.

Not all the events involving the mutS sweep can be fully defined on the evidence available but a fairly detailed scenario can be assembled and is shown in Fig 3. A mutator is selectively advantageous only through the mutations it causes (COX and GIBSON 1974 ). Hence, as predicted in modeling studies (TENAILLON et al. 1999 ), the mutS mutation must itself have hitchhiked on the back of favorable mutation(s) to become a significant proportion of population 26. The presence of mutS mutants without mgl mutations at point E suggests the mutator was not initially hitchhiking with mgl. The simplest scenario up to point E is that:

1. The first periodic sweep in population 26, at points C–E in Fig 1B, led to accumulation of unidentified favorable mutation(s) (in xxxA in Fig 3). The first sweep was not due to mutS per se, as the frequency of mutator was still relatively low at point E.

2. A mutator appeared in the T5-sensitive population containing the unidentified mutation and hitchhiked on the back of this mutation.

3. After mutators became relatively common but before point E, the higher mutation rates permitted at least two independent occurrences of mgl mutations in the mutator bacteria.

4. Several independent fhuA mutations arose in the mutS/mgl mutants and the growth advantage of the mgl mutation resulted in the initial enrichment of the T5-resistant bacteria.

5. A variety of mgl mutations arising in mutators as well as nonmutators (perhaps directly in the xxxA mutants after point C but before mutS was common) in the T5-sensitive majority prevented mgl/fhuA bacteria from entirely taking over the population.

6. After point E, yet another unidentified advantageous mutation spread in the T5-sensitive bacteria at point F, resulting in the reduction of T5 resistance and mutators by point G.

The mutator sweep was present in one of two populations analyzed here, but transient mutator sweeps were recently found in three out of three other similarly grown chemostat populations (L. NOTLEY-MCROBB, unpublished results). Of the four "mutator" populations, two contained mutY and two, including population 26, contained mutS. The frequent occurrence of mutators helps to explain the skewed mutational spectrum of previously studied populations that were monitored less frequently (weekly) and that may have missed transient mutator sweeps of the type demonstrated in population 26 (NOTLEY-MCROBB and FERENCI 1999A , NOTLEY-MCROBB and FERENCI 1999B ).

Despite contributing to the rapid spread of mgl mutations, the 250-fold increase in mutation rates was not sustained in members of a chemostat population. The mutS mutation must have disappeared extremely rapidly from the culture because the mlc sweep apparent after only 30 further generations in population 26 had all the characteristics of random spontaneous, rather than mutator-inspired, mutations. The results of this study provide direct evidence that highly elevated mutation rates are counterselected in long-term populations (TROBNER and PIECHOCKI 1984 ; TADDEI et al. 1997 ; RAINEY 1999 ; TENAILLON et al. 1999 ).

In conclusion, it is evident that fluctuations in the frequency of neutral mutations are not necessarily linked with purges in bacterial populations. The frequency of neutral mutations was also influenced by the transient appearance of mutators in chemostat 26 and there is a complex relationship between favorable and neutral mutations in large bacterial populations. Nevertheless, the identification of mgl and mlc as mutational targets has led to the possibility of molecular analyses of fundamental events in bacterial evolution described 50 years ago (NOVICK and SZILARD 1950 ). Further experimental studies of the occurrence of rare gain-of-function sweeps, as well as variations of population size currently underway, will help to further our understanding of periodic selection effects in evolution.

	ACKNOWLEDGMENTS

We thank K. Heller for the supply of bacteriophage T5 and the Australian Research Council for grant support.

Manuscript received May 23, 2000; Accepted for publication August 16, 2000.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ATWOOD, K. C., L. K. SCHNEIDER, and F. J. RYAN, 1951  Selective mechanisms in bacteria. Cold Spring Harbor Symp. Quant. Biol. 16:345-355[Abstract/Free Full Text].

BERG, O. G., 1995  Periodic selection and hitchhiking in a bacterial population. J. Theor. Biol. 173:307-320[Medline].

BRAUN, V. and H. KILLMANN, 1999  Bacterial solutions to the iron supply problem. Trends Biochem. Sci. 24:104-109[Medline].

CHAO, L. and E. C. COX, 1983  Competition between high and low mutating strains of Escherichia coli.. Evolution 37:125-134.

COX, E. C. and T. C. GIBSON, 1974  Selection for high mutation rates in chemostats. Genetics 77:169-184[Abstract/Free Full Text].

COX, E. C., G. E. DEGNEN, and M. L. SCHLEPPE, 1972  Mutator gene studies in Escherichia coli: the mutS gene. Genetics 73(Suppl.):551-567.

CURTIS, S. J. and W. EPSTEIN, 1975  Phosphorylation of D-glucose in Escherichia coli mutants defective in glucosephosphotransferase, mannosephosphotransferase and glucokinase. J. Bacteriol. 122:1189-1199[Abstract/Free Full Text].

DEATH, A. and T. FERENCI, 1993  The importance of the binding-protein-dependent Mgl system to the transport of glucose in Escherichia coli growing on low sugar concentrations. Res. Microbiol. 144:529-537[Medline].

DEATH, A., L. NOTLEY, and T. FERENCI, 1993  Derepression of LamB protein facilitates outer membrane permeation of carbohydrates into Escherichia coli under conditions of nutrient stress. J. Bacteriol. 175:1475-1483[Abstract/Free Full Text].

DYKHUIZEN, D. E., 1990  Experimental studies of natural selection in bacteria. Annu. Rev. Ecol. Syst. 21:373-398.

FINKEL, S. E. and R. KOLTER, 1999  Evolution of microbial diversity during prolonged starvation. Proc. Natl. Acad. Sci. USA 96:4023-4027[Abstract/Free Full Text].

GUTTMAN, D. S. and D. E. DYKHUIZEN, 1994  Detecting selective sweeps in naturally occurring Escherichia coli.. Genetics 138:993-1003[Abstract].

HELLER, K. J., 1992  Molecular interaction between bacteriophage and the gram-negative cell envelope. Arch. Microbiol. 158:235-248[Medline].

HELLING, R. B., C. N. VARGAS, and J. ADAMS, 1987  Evolution of Escherichia coli during growth in a constant environment. Genetics 116:349-358[Abstract/Free Full Text].

JAIN, R., M. C. RIVERA, and J. A. LAKE, 1999  Horizontal gene transfer among genomes: the complexity hypothesis. Proc. Natl. Acad. Sci. USA 96:3801-3806[Abstract/Free Full Text].

KORONA, R., 1996  Genetic divergence and fitness convergence under uniform selection in experimental populations of bacteria. Genetics 143:637-644[Abstract].

KUBITSCHEK, H. E., 1970 Introduction to Research With Continuous Cultures. Prentice-Hall, Englewood Cliffs, NJ.

KUBITSCHEK, H. E., 1974  Operation of selection pressure on microbial populations. Symp. Soc. Gen. Microbiol. 24:105-130.

LAWRENCE, J. G., 1999  Gene transfer, speciation, and the evolution of bacterial genomes. Curr. Opin. Microbiol. 2:519-523[Medline].

LECLERC, J. E., B. G. LI, W. L. PAYNE, and T. A. CEBULA, 1996  High mutation frequencies among Escherichia coli and Salmonella pathogens. Science 274:1208-1211[Abstract/Free Full Text].

LENSKI, R. E., J. A. MONGOLD, P. D. SNIEGOWSKI, M. TRAVISANO, and F. VASI et al., 1998  Evolution of competitive fitness in experimental populations of E. coli—what makes one genotype a better competitor than another. Antonie Leeuwenhoek 73:35-47.

LEVIN, B. R., 1981  Periodic selection, infectious gene exchange and the genetic structure of E. coli populations. Genetics 99:1-23[Abstract/Free Full Text].

MANCHE, K., L. NOTLEY-MCROBB, and T. FERENCI, 1999  Mutational adaptation of Escherichia coli to glucose limitation involves distinct evolutionary pathways in aerobic and oxygen-limited environments. Genetics 153:5-12[Abstract/Free Full Text].

MAYNARD-SMITH, J., 1991  The population genetics of bacteria. Proc. R. Soc. Lond. Ser. B Biol. Sci. 245:37-42[Abstract/Free Full Text].

MILKMAN, R., 1997  Recombination and population structure in Escherichia coli.. Genetics 146:745-750[Medline].

MILLER, J., 1972 Experiments in Molecular Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

MILLER, J. H., 1996  Spontaneous mutators in bacteria—insights into pathways of mutagenesis and repair. Annu. Rev. Microbiol. 50:625-643[Medline].

NOTLEY, L. and T. FERENCI, 1995  Differential expression of mal genes under cAMP and endogenous inducer control in nutrient stressed Escherichia coli.. Mol. Microbiol. 160:121-129.

NOTLEY-MCROBB, L. and T. FERENCI, 1999a  Adaptive mgl-regulatory mutations and genetic diversity evolving in glucose-limited Escherichia coli populations. Environ. Microbiol. 1:33-43[Medline].

NOTLEY-MCROBB, L. and T. FERENCI, 1999b  The generation of multiple coexisting mal-regulatory mutations through polygenic evolution in glucose-limited populations of Escherichia coli.. Environ. Microbiol. 1:45-52[Medline].

NOVICK, A., 1958 Some chemical bases for evolution of micro-organisms, pp. 531–546 in Perspectives in Marine Biology, edited by A. A. BUZZATI-TRAVERSO. University of California Press, Berkeley.

NOVICK, A. and L. SZILARD, 1950  Experiments with the chemostat on spontaneous mutations of bacteria. Proc. Natl. Acad. Sci. USA 36:708-719[Free Full Text].

NURMINSKY, D. I., M. V. NURMINSKAYA, D. DE AGUIAR, and D. L. HARTL, 1998  Selective sweep of a newly evolved sperm-specific gene in Drosophila.. Nature 396:572-575[Medline].

PAPADOPOULOS, D., D. SCHNEIDER, J. MEIER-EISS, W. ARBER, and R. E. LENSKI et al., 1999  Genomic evolution during a 10,000-generation experiment with bacteria. Proc. Natl. Acad. Sci. USA 96:3807-3812[Abstract/Free Full Text].

PROZOROV, A. A., 1999  Horizontal gene transfer in bacteria: laboratory modeling, natural populations, and data from genome analysis. Microbiology 68:551-564.

RAINEY, P. B., 1999  Evolutionary genetics: the economics of mutation. Curr. Biol. 9:R371-R373[Medline].

ROSENWEIG, R. F., R. R. SHARP, D. S. TREVES, and J. ADAMS, 1994  Microbial evolution in a simple unstructured environment: genetic differentiation in Escherichia coli.. Genetics 137:903-917[Abstract].

SAVAGE, D., 1977  Microbial ecology of the gastrointestinal tract. Annu. Rev. Microbiol. 31:107-133[Medline].

SELANDER, R. K., D. A. CAUGANT and T. S. WHITTAM, 1987 Genetic structure and variation in natural populations of Escherichia coli, pp. 1625–1648 in Escherichia coli and Salmonella typhimurium. Cellular and Molecular Biology, Ed. 1, edited by F. C. NEIDHARDT. ASM Press, Washington DC.

SNIEGOWSKI, P. D., P. J. GERRISH, and R. E. LENSKI, 1997  Evolution of high mutation rates in experimental populations of E. coli.. Nature 387:703-705[Medline].

TADDEI, F., M. RADMAN, J. MAYNARD-SMITH, B. TOUPANCE, and P. H. GOUYON et al., 1997  Role of mutator alleles in adaptive evolution. Nature 387:700-702[Medline].

TAKIMOTO, K., A. TACHIBANA, H. AYAKI, and K. YAMAMOTO, 1997  Spectrum of spontaneous mutations in the cyclic AMP receptor protein gene on chromosomal DNA of Escherichia coli.. J. Radiat. Res. 38:27-36.

TENAILLON, O., B. TOUPANCE, N. H. LE, F. TADDEI, and B. GODELLE, 1999  Mutators, population size, adaptive landscape and the adaptation of asexual populations of bacteria. Genetics 152:485-493[Abstract/Free Full Text].

TREVES, D. S., S. MANNING, and J. ADAMS, 1998  Repeated evolution of an acetate-crossfeeding polymorphism in long-term populations of Escherichia coli.. Mol. Biol. Evol. 15:789-797[Abstract].

TROBNER, W. and R. PIECHOCKI, 1984  Selection against hypermutability in Escherichia coli during long term evolution. Mol. Gen. Genet. 198:177-178[Medline].

This article has been cited by other articles:

				L.-M. Chevin, S. Billiard, and F. Hospital Hitchhiking Both Ways: Effect of Two Interfering Selective Sweeps on Linked Neutral Variation Genetics, September 1, 2008; 180(1): 301 - 316. [Abstract] [Full Text] [PDF]

				R. Maharjan, S. Seeto, L. Notley-McRobb, and T. Ferenci Clonal Adaptive Radiation in a Constant Environment Science, July 28, 2006; 313(5786): 514 - 517. [Abstract] [Full Text] [PDF]

				R. Woods, D. Schneider, C. L. Winkworth, M. A. Riley, and R. E. Lenski Tests of parallel molecular evolution in a long-term experiment with Escherichia coli PNAS, June 13, 2006; 103(24): 9107 - 9112. [Abstract] [Full Text] [PDF]

				T. King, S. Seeto, and T. Ferenci Genotype-by-Environment Interactions Influencing the Emergence of rpoS Mutations in Escherichia coli Populations Genetics, April 1, 2006; 172(4): 2071 - 2079. [Abstract] [Full Text] [PDF]

				S. Zhong, A. Khodursky, D. E. Dykhuizen, and A. M. Dean Evolutionary genomics of ecological specialization PNAS, August 10, 2004; 101(32): 11719 - 11724. [Abstract] [Full Text] [PDF]

				D. E. Dykhuizen and A. M. Dean Evolution of Specialists in an Experimental Microcosm Genetics, August 1, 2004; 167(4): 2015 - 2026. [Abstract] [Full Text] [PDF]

				S. de Weert, L. C. Dekkers, I. Kuiper, G. V. Bloemberg, and B. J. J. Lugtenberg Generation of Enhanced Competitive Root-Tip-Colonizing Pseudomonas Bacteria through Accelerated Evolution J. Bacteriol., May 15, 2004; 186(10): 3153 - 3159. [Abstract] [Full Text] [PDF]

				A. C. Shaver and P. D. Sniegowski Spontaneously Arising mutL Mutators in Evolving Escherichia coli Populations Are the Result of Changes in Repeat Length J. Bacteriol., October 15, 2003; 185(20): 6076 - 6082. [Abstract] [Full Text] [PDF]

				O. Michelsen, M. J. Teixeira de Mattos, P. R. Jensen, and F. G. Hansen Precise determinations of C and D periods by flow cytometry in Escherichia coli K-12 and B/r Microbiology, April 1, 2003; 149(4): 1001 - 1010. [Abstract] [Full Text] [PDF]

				L. Notley-McRobb, S. Seeto, and T. Ferenci Enrichment and Elimination of mutY Mutators in Escherichia coli Populations Genetics, November 1, 2002; 162(3): 1055 - 1062. [Abstract] [Full Text] [PDF]

				A. C. Shaver, P. G. Dombrowski, J. Y. Sweeney, T. Treis, R. M. Zappala, and P. D. Sniegowski Fitness Evolution and the Rise of Mutator Alleles in Experimental Escherichia coli Populations Genetics, October 1, 2002; 162(2): 557 - 566. [Abstract] [Full Text] [PDF]

				L. M. Wick, H. Weilenmann, and T. Egli The apparent clock-like evolution of Escherichia coli in glucose-limited chemostats is reproducible at large but not at small population sizes and can be explained with Monod kinetics Microbiology, September 1, 2002; 148(9): 2889 - 2902. [Abstract] [Full Text] [PDF]

				L. Notley-McRobb, R. Pinto, S. Seeto, and T. Ferenci Regulation of mutY and Nature of Mutator Mutations in Escherichia coli Populations under Nutrient Limitation J. Bacteriol., February 1, 2002; 184(3): 739 - 745. [Abstract] [Full Text] [PDF]