Activation of the bgl Operon by Adaptive Mutation Barry G. Hall Biology Department, University of Rochester In growing Escherichia coli K12 cells, the cryptic bgl operon is activated 98% of the time by insertions of IS1 or IS5 into the control region, designated bglR. The activated bgl operon permits utilization of the b-glucoside sugar arbutin as a sole carbon and energy source. The bgl operon is also activated by late-occurring mutations during prolonged selection on arbutin. The late-occurring mutations that occurred during prolonged carbon starvation in the presence of arbutin were ‘‘adaptive mutations’’ because they were specific to the presence of arbutin, and they did not occur during prolonged starvation in the absence of arbutin. The spectrum of late-arising mutations differed from that of early-arising, growth-dependent mutations in that 20% of the late-arising mutants resulted from mutations at the hns locus. This provides the first direct evidence for adaptive mutagenesis mediated by the insertion of IS elements. Because no special genetic background is required to select Bgl1 mutants, this affords the opportunity to study IS-element-mediated adaptive mutagenesis in a variety of genetic backgrounds, including the backgrounds of natural isolates of E. coli.

Introduction The bgl operon is one of at least four cryptic, or silent, systems for the uptake and metabolism of b-glucoside sugars in Escherichia coli (Prasad and Schaefler 1974; Kricker and Hall 1984; Parker and Hall 1988, 1990; Hall and Xu 1992). The bgl operon encodes all of the functions required for the transport and hydrolysis of the b-glucoside sugars arbutin and salicin (Schnetz, Toloczyki, and Rak 1987), but in order to be expressed at a level that will permit growth on those sugars, it must be activated by mutations that greatly increase the activity of a normally weak promoter. The bgl operon is maintained in a silent state by silencer elements that are located both upstream and downstream of the promoter and the CAP binding site (Schnetz 1995). The operon can be activated by mutations that disrupt those flanking silencer element sequences. The most common disruptions involve insertions of IS1 or IS5 into that 223-bp region (Schnetz and Rak 1988), but rare IS2, IS10, or TN1000 insertional activation, and activation by deletion of portions of the sequence upstream of the CAP binding site, have also been reported (Schnetz and Rak 1992; Schnetz 1995). The bgl operon can also be activated by either of two base substitution mutations in the CAP binding site (Schnetz and Rak 1992). In addition to these cis-acting mutations, the bgl operon can be activated by transacting mutations in the gyrase genes gyrA and gyrB (DiNardo et al. 1982), by mutations in hns, which encodes a highly abundant histone-like nucleoid-associated protein that is a major component of the bacterial chromatin (Defez and deFelice 1981), or by an IS10R insertion that permits transcription of bglJ, whose product is not yet understood (Giel, Desnoyer, and Lopilato 1996). Adaptive mutations are mutations that occur in nondividing or very slowly dividing cells during proKey words: cryptic genes, adaptive mutations, bgl operon, insertion sequences, E. coli. Address for correspondence and reprints: Barry G. Hall, Biology Department, River Campus, University of Rochester, Rochester, New York 14627. E-mail: [email protected]. Mol. Biol. Evol. 15(1):1–5. 1998 q 1998 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038

longed nonlethal selection, and are specific to the challenge of the selection in the sense that the only mutations that can be detected are those that provide a growth advantage to the cell (Hall 1997). Two lines of evidence indicate that adaptive mutations arise by different processes than the processes that generate spontaneous mutations in growing cells. First, the processes differ in that they generate different mutational spectra both for base substitution mutations (Hall 1991) and for reversion of a lacI-lacZ fusion frameshift mutation (Foster and Trimarchi 1994; Rosenberg et al. 1994). Second, lesions in the excision-repair pathway genes uvrA, uvrB, and uvrC increase the adaptive reversion rate of trpA46 but do not affect the growth-dependent spontaneous mutation reversion rate of the same allele (Hall 1995). Because it has been reported that 98% of all spontaneous Bgl1 mutants are due to transposition of IS5 or IS1 into the 223-bp region flanking the bgl promoter (Schnetz, Toloczyki, and Rak 1987), the bgl operon appeared to be a good candidate to study the effects of adaptive mutagenesis on mutations that result from transposition of mobile elements. Materials and Methods E. coli K12 Strains Strain SJ134 is F2 ebgA51 rpsL DlacZ4680. Strain JF201 is F2 DlacZx74 D(bgl-pho) ara gyrA thi. Media Mineral salts medium consisted of 423 mg sodium citrate, 100 mg MgSO4 · 7H2O, 1 g (NH4)2SO4, 540 mg FeCl3, 1 mg thiamine, 3 g KH2PO4, and 7 g K2HPO4, with either glucose (0.1 or 2.0 g per liter) or arbutin (1 g per liter) as carbon sources. Plates were solidified with 15 g purified agar (Sigma) per liter. L-agar consists of LB-agar (Miller 1972) plus 1 g per liter glucose. MacConkey media were prepared from MacConkey Agar Base (Difco) and the indicated sugar according to manufacturer’s instructions. 1

2

Hall

FIG. 1.—A, Number of viable cells per plate during selection on arbutin. B, Accumulation of Bgl1 mutant colonies during selection on arbutin.

Results and Discussion Strain SJ134 was grown in 0.01% (w/v) glucose mineral salts medium as 120 individual 200-ml cultures to a density of 8.5 3 106 cells per culture. Each culture was spread onto an arbutin mineral salts plate, and the plates were incubated at 308C. Each day for 4 days cells were washed off of two plates, suitably diluted, and plated onto L-agar to determine the number of viable cells per plate (fig. 1A). Once Bgl1 colonies began to appear on day 2, on a subset of plates those colonies were immediately eliminated, with little disturbance of surrounding cells, through the use of a diathermy probe (Hyfrecator Plusy Model 7–796, Birtcher Medical Supplies), an electrosurgery device that delivers an intense spark which kills the cells in the colony. Those treated plates were used to estimate the number of viable cells. By day 1, the population had grown at the expense of trace contaminants from 8.5 3 106 cells to 1.2 3 108 cells per plate. In other experiments, this maximum population density was achieved in about 8 hours (data not shown). The population grew no further, and on day 3

began to die exponentially with a first order rate constant of 21.23 day21. Colonies first appeared on day 2 at an average of 2.2 colonies per plate, and additional colonies continued to appear for the next several days (fig. 1B). The colonies that appeared on day 2 are presumed to have arisen during the growth of the cultures, and because each plate was inoculated with an independently grown culture, the distribution of those colonies can be used to estimate the growth-dependent mutation rate from a Luria-Delbru¨ck fluctuation test (Luria and Delbru¨ck 1943). The Stewart et al. method for analysis of fluctuation tests (Stewart, Gordon, and Levin 1990), as implemented by Stewart’s DataFity program, permits one to estimate both the average number of mutations that occurred prior to plating during the growth of each culture and the average number of mutations that occurred after plating. In this experiment, an average of 0.3 mutations occurred during the growth of each culture prior to plating. The average mutation rate to Bgl1 during growth of the cultures was calculated as 0.3/8.5 3 106 (the mean number of cells per culture), and was 3.6 3 1028 mutations per cell division. There are four different cryptic operons that can be activated to permit arbutin utilization by E. coli K12 including the bgl operon (Prasad and Schaefler 1974), the cel operon (Kricker and Hall 1984; Parker and Hall 1990), the asc operon (Parker and Hall 1988; Hall and Xu 1992), and the arbT transport gene (Kricker and Hall 1987). These operons can be distinguished phenotypically on MacConkey salicin and MacConkey cellobiose agar. Bgl1 colonies are strongly positive (dark red) on MacConkey salicin but are negative (white) on MacConkey cellobiose. Cel1 colonies are strongly positive on MacConkey salicin and on MacConkey cellobiose (Kricker and Hall 1984), ArbT1 colonies are negative on both salicin and cellobiose (Hall and Xu 1992, Kricker and Hall 1987), while Asc1 colonies are very weakly positive (light pink) on salicin and negative on cellobiose (Hall and Xu 1992). Each day, several colonies were restreaked onto MacConkey salicin in order to obtain pure cultures. In order to be sure that all mutants were of independent origin, on any given day only one Bgl1 mutant was isolated from any single plate. All isolated colonies produced dark red (strongly positive) colonies on MacConkey salicin. The isolated colonies were patched onto MacConkey cellobiose medium and all proved to be cellobiose negative. All of the mutants were thus identified as Bgl1 mutants. A reconstruction test was used to determine whether the late-appearing Bgl1 mutants were the result of slow growth or the result of mutations that occurred after population growth had ceased. Cultures were grown from 10 Bgl1 mutants that had appeared on days 2–5, and approximately 100 cells of each culture were spread onto arbutin mineral salts medium either alone or together with about 108 JF201 (Dbgl) scavenger cells, and the plates were incubated at 308C. In each case, both with and without scavenger cells, there were about 100 colonies visible on the plates after 2 days’ incubation. Although colonies on plates with scavenger cells were

Adaptive Activation of bgl

smaller than those on plates without scavenger cells, the observations that the plating efficiencies were the same under both conditions and that colonies appeared in 2 days indicate that the late-arising colonies were the result of late-occurring mutations. Based on the assumption that each colony appeared 2 days after the mutation occurred, the adaptive mutation rate on each day was calculated by dividing the average number of viable cells per plate into the average number of colonies that appeared 2 days later per plate. The average adaptive mutation rate over days 1–3 was 3.4 3 1027 mutations per viable cell per day. The key feature that distinguishes adaptive mutations from random, growth-dependent mutations is their specificity, i.e., that only useful mutations can be detected under the selective conditions employed. Carbon starvation per se is not generally mutagenic (Hall 1997), but it is well known that tranposition of IS elements is particularly active in stored stab cultures (Naas et al. 1994). If the late-arising Bgl1 mutants were the result of adaptive mutations, then their appearance should be specific to the presence of arbutin in the medium. To assess that specificity, 107 strain SJ134 cells were spread onto mineral salts medium containing 10 mg/liter glucose. Cells were washed off of 6 plates 6 h after plating on day 0, and on days 1–4 during incubation at 308C. Ten microliters of each 10-ml cell suspension was diluted and spread onto L-agar to determine the number of viable cells, and the remainder of the suspension was concentrated and spread onto a single mineral salts arbutin plate. Each arbutin plate was incubated for 2 days, and the number of colonies was counted to estimate the number of Bgl1 mutants that were present at the time of plating. The population grew to 7 3 107 cells per plate by 6 h and reached a maximum density of 7 3 108 cells per plate (fig. 2A). Since the population grew from 1 3 107 cells to 6 3 107 cells in 6 h, I estimate that glucose was exhausted and the maximum density was achieved after 14 h incubation. The frequency of Bgl1 mutants remained roughly constant over the course of the experiment (fig. 2B). If late-appearing Bgl1 mutants on arbutin medium were the result of random IS transpositions or other random mutations, then their frequency should have increased at the rate of about 3.4 3 1027 cell21 day21 in the absence of arbutin just as it did in the presence of arbutin. Figure 2B shows the observed frequencies of Bgl1 mutants and the frequencies expected if Bgl1 mutations were unrelated to the presence of arbutin (nonadaptive mutations). The expected frequencies if nonadaptive were based upon the observed frequency at 6 h, the increase due to additional growthdependent mutations at the rate of 3.6 3 1028 mutations per cell division, growth of both new and preexisiting mutants until glucose was exhausted, and a subsequent mutation rate of 3.4 3 1027 mutations per cell per day. The results indicate that the late-arising Bgl1 mutants on arbutin medium were the result of adaptive mutations. To determine the nature of the mutations that had activated the bgl operon, genomic DNA was prepared from 48 day 2 mutants, 24 day 3 mutants, 24 day 4

3

FIG. 2.—A, Number of viable cells per plate during incubation on limiting glucose medium. B, Observed and expected frequencies of Bgl1 mutants during 4 days of carbon starvation on limiting glucose medium. See text for explanation of expected values.

mutants, and 24 day 5 mutants. A 379-bp region flanking the bgl promoter was amplified, and the sizes of the products were determined by agarose gel electrophoresis. If no insertion is present the product is 0.38 kb; if IS1 is inserted, it is 1.15 kb; and if IS5 is inserted it is 1.57 kb. All insertion sequences other than IS1 that are native to E. coli K12 are between 1.2 and 1.4 kb, and because transposition of other elements into the bgl promoter, although very rare, has been reported, all insertions that yielded amplification products .1.2 kb were sequenced to determine which element had inserted. All such insertions proved to be IS5 elements that had inserted between bp 358 and bp 376 of the bgl operon. Deletions would either produce a product smaller than 0.38 kb or fail to produce an amplification product; however, no deletions were detected in this experiment. The mutants that appeared on day 2 are presumed to have arisen from preexisting Bgl1 cells that were present at the time of plating. Among the 48 day 2 mu-

4

Hall

tants, 47 resulted from insertions. This result is consistent with previous reports (Schnetz, Toloczyki, and Rak 1987) that 98% of Bgl1 mutants arise from insertions. Among the 48 late-arising mutants from days 4 and 5, 37 had an insertion, while 11 had no insertion. By Fisher’s exact test for 2 3 2 contingency tables (Finndy 1948) the distribution among day 4 and 5 mutants is significantly different from the distribution for the earlyarising day 2 mutants (P 5 0.002, where P is the probability that the two distributions differ by chance alone). The distribution for the 24 day 3 mutants was 4 noninsertion and 20 insertion mutants, which is not significantly different from the day 4–5 distribution (P 5 0.76). Day 3 mutants were therefore pooled with day 4 and 5 mutants and considered to be late-arising. For all 72 late-arising mutants, the distribution was 15 non-insertion mutants and 57 insertion-generated mutants, i.e. 79% of late-arising Bgl1 mutants were the result of insertions. The probability that the distributions of the early-arising and the late-arising mutants do not differ is 0.0023. The vast majority (91%) of the insertions were IS1, and there was no significant difference between the early-arising mutants and the late-arising mutants. That distribution is surprising for two reasons. First, only 40% of the insertions whose positions have previously been reported were IS1 (Schnetz and Rak 1992). Second, E. coli K12 carries 11 copies of IS5, but only 6 copies of IS1 (Deonier 1996). If the insertion frequencies were proportional to copy number, then we would expect only about 35% of the insertions to be IS1. To determine the nature of the 16 noninsertion Bgl1 mutants, the sites of the mutations were mapped by P1vir cotransduction with Tn10 elements located near the bglBFG operon and near hns (Tn10::zch-3060). Bacteriophage P1vir lysates were prepared from the donor strains CAG18499 (Tn10::zid-501) and CAG12016 (Tn10::zch-3060) (Singer et al. 1989) according to Miller (1972). In each case, following infection with the P1vir lysate, cells were spread onto MacConkey-salicin medium (Difco) containing 15 mg/ml tetracycline, and cotransductants between the Tn10 element and the wildtype allele (Sal2 phenotype) appeared as white colonies. Only one noninsertion mutant mapped to the bgl operon, with a cotransduction frequency of 71% between Tn10: :zid-501 and bgl1 (Sal2 phenotype), a value that is virtually identical to the cotransduction frequency of 73% when an IS1 insertion mutant was the recipient. That mutant presumably arose from a base substitution in the CAP binding site, as has previously been reported (Schnetz and Rak 1992). The remaining 15 noninsertion mutants were the result of mutations in hns, as has previously been reported (Defez and deFelice 1981). With those recipients, the cotransduction frequencies between Tn10::zch-3060 and hns1 (Sal2 phenotype) ranged from 30% to 45%, with a mean of 35.4 6 6.2%. Adaptive mutations have previously been shown to arise from base substitutions (Cairns, Overbaugh, and Miller 1988), frameshifts (Cairns and Foster 1991), excision of mobile elements (Shapiro 1984; Hall 1988, 1994), and genomic rearrangements (Thomas et al.

1992). This study adds insertion of mobile elements to that list and supports the notion (Mittler and Lenski 1992; Foster 1993) that there are multiple mechanisms and pathways for adaptive mutagenesis. The observation that about 20% of adaptive Bgl1 mutants arose via mutations at hns, whereas 98% of growth-dependent Bgl1 arose from insertions of IS1 or IS5 into the bglR region, further supports the notion that the spectra of growthdependent and adaptive mutations are quite different. Although about 20% of adaptive Bgl1 mutants are not mediated by insertion elements, Bgl activation is still a useful tool for studying insertion-element-mediated adaptive mutagenesis. Because no particular genetic background is required, as is the case when adaptive mutagenesis is studied via reversion of known mutations, Bgl can easily be used to study insertion-elementmediated adaptive mutagenesis in a variety of genetic backgrounds, including natural isolates such as those that constitute the ECOR collection (Ochman and Selander 1984). Acknowledgments This study was supported by grant #NP-932 from the American Cancer Society. I am grateful to Jacqueline Toner for expert technical assistance. LITERATURE CITED

CAIRNS, J., and P. L. FOSTER. 1991. Adaptive reversion of a frameshift mutation in Escherichia coli. Genetics 128:695– 701. CAIRNS, J., J. OVERBAUGH, and S. MILLER. 1988. The origin of mutants. Nature 335:142–145. DEFEZ, R., and M. DEFELICE. 1981. Cryptic operon for b-glucoside metabolism in Escherichia coliK12: genetic evidence for a regulatory protein. Genetics 97:11–25. DEONIER, R. C. 1996. Native insertion sequence elements. Pp. 2000–2011 in F. C. NEIDHARDT, R. CURTISS III, J. L. INGRAHAM, E. C. C. LIN, K. B. LOW, B. MAGASANIK, W. S. REZNIKOFF, M. RILEY, M. SCHAECHTER, and H. E. UMBARGER, eds. Escherichia coli and Salmonella. ASM Press, Washington, D.C. DINARDO, S., K. A. VOELKEL, R. STERNGLANZ, A. E. REYNOLDS, and A. WRIGHT. 1982. Escherichia coli DNA topoisomerase I mutants have compensatory mutations in DNA gyrase genes. Cell 31:43–51. FINNDY, D. J. 1948. The Fisher-Yates test of significance in 2 3 2 contingency tables. Biometrika 35:145–156. FOSTER, P. L. 1993. Adaptive mutation: the uses of adversity. Annu. Rev. Microbiol. 47:467–504. FOSTER, P. L., and J. M. TRIMARCHI. 1994. Adaptive reversion of a frameshift mutation in Escherichia coli by simple base deletions in homopolymeric runs. Science 265:407–409. GIEL, M., M. DESNOYER, and J. LOPILATO. 1996. A mutation in a new gene, bglJ, activates the bgl operon in Escherichia coli K-12. Genetics 143:627–635. HALL, B. G. 1988. Adaptive evolution that requires multiple spontaneous mutations. I. Mutations involving an insertion sequence. Genetics 120:887–897. . 1991. Spectrum of mutations that occur under selective and non-selective conditions in E. coli. Genetica 84: 73–76.

Adaptive Activation of bgl

. 1994. On alternatives to selection-induced mutations in the Bgl operon of Escherichia coli. Mol. Biol. Evol. 11: 159–168. . 1995. Genetics of selection-induced mutations: I. uvrA, uvrB, uvrC, and uvrD are selection-induced specific mutator loci. J. Mol. Evol. 40:86–93. . 1997. On the specificity of adaptive mutations. Genetics 145:39–44. HALL, B. G., and L. XU. 1992. Nucleotide sequence, function, activation, and evolution of the cryptic asc operon of Escherichia coliK12. Mol. Biol. Evol. 9:688–702. KRICKER, M., and B. G. HALL. 1984. Directed evolution of cellobiose utilization in Escherichia coli. Mol. Biol. Evol. 1:171–182. . 1987. Biochemical genetics of the cryptic gene system for cellobiose utilization in Escherichia coliK12. Genetics 115:419–429. LURIA, S. E., and M. DELBRU¨CK. 1943. Mutations of bacteria from virus sensitivity to virus resistance. Genetics 28:491– 511. MILLER, J. H. 1972. Experiments in molecular genetics. Cold Spring Harbor Laboratories, Cold Spring Harbor, N.Y. MITTLER, J. E., and R. E. LENSKI. 1992. Experimental evidence for an alternative to directed mutation in the bgl operon. Nature 356:446–448. NAAS, T., M. BLOT, W. M. FITCH, and W. ARBER. 1994. Insertion sequence related genetic variation in resting Escherichia coli K-12. Genetics 136:721–730. OCHMAN, H., and R. H. SELANDER. 1984. Standard reference strains of Escherichia coli from natural populations. J. Bacteriol. 157:690–693. PARKER, L. L., and B. G. HALL. 1988. A fourth E. coli gene system with the potential to evolve b-glucoside utilization. Genetics 119:485–490. . 1990. Mechanisms of activation of the cryptic cel operon of E. coli K12. Genetics 124:473–482.

5

PRASAD, I., and S. SCHAEFLER. 1974. Regulation of the b-glucoside system in Escherichia coli K12. J. Bacteriol. 120: 638–650. ROSENBERG, S. M., S. LONGERICH, P. GEE, and R. S. HARRIS. 1994. Adaptive mutation by deletions in small mononucleotide repeats. Science 265:405–407. SCHNETZ, K. 1995. Silencing of Escherichia coli bgl promoter by flanking sequence elements. EMBO J. 14:2545–2550. SCHNETZ, K., and B. RAK. 1988. Regulation of the bgl operon of Escherichia coli by transcriptional anti-termination. EMBO J. 7:3271–3277. . 1992. IS5: a mobile enhancer of transcription in Escherichia coli. Proc. Natl. Acad. Sci. USA 89:1244–1248. SCHNETZ, K., C. TOLOCZYKI, and B. RAK. 1987. b-glucoside (Bgl) operon of Escherichia coli K12: nucleotide sequence, genetic organization, and possible evolutionary relationship to regulatory components of two Bacillus subtilis genes. J. Bacteriol. 169:2579–2590. SHAPIRO, J. A. 1984. Observations on the formation of clones containing araB-lacZ cistron fusions. Mol. Gen. Genet. 194: 79–90. SINGER, M., T. A. BAKER, G. SCHNITZLER, S. M. DEISCHEL, M. GOEL, W. DOVE, K. J. JAACKS, A. D. GROSSMAN, J. W. ERICKSON, and C. A. GROSS. 1989. A collection of strains containing genetically linked alternating antibiotic resistance elements for genetic mapping of Escherichia coli. Microbiol. Rev. 53:1–24. STEWART, F. M., D. M. GORDON, and B. R. LEVIN. 1990. Fluctuation analysis: the probability distribution of the number of mutants under different conditions. Genetics 124:175– 185. THOMAS, A. W., J. LEWINGTON, S. HOPE, A. W. TOPPING, A. J. WEIGHTMAN, and J. H. SLATER. 1992. Environmentally directed mutations in the dehalogenase system of Pseudomonas putida strain PP3. Arch. Microbiol. 158:176–182.

JULIAN P. ADAMS, reviewing editor Accepted September 15, 1997