Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Methods Enzymol. Author manuscript; available in PMC 2010 November 30.
Published in final edited form as:
PMCID: PMC2994264

In Vivo Mutagenesis


Mutation of bacteria in vivo by chemical and physical means is a powerful method by which to generate genetic variants. When searching for mutants in an unknown gene or in uncharacterized regions of a cloned gene, chemical and/or physical mutagenesis can be combined with insertional mutagenesis to obtain the maximum number of alleles for phenotypic analysis. In general, base substitutions are the most useful mutations with which to investigate both the in vivo activities and the structure–function relationship of the gene product of interest. By using the appropriate mutagen and bacterial strain, it is even possible to generate specific classes of mutations. The methods outlined below are designed to aid the geneticist as well as those who wish to investigate mutagenic and DNA repair processes in Escherichia coli and Salmonella typhimurium.

Bacterial Strains

Escherichia coli and S. typhimurium have mutagenic and accurate repair processes that act on DNA lesions. After DNA damage both types of repair are induced as part of the SOS response.1 In general, mutagens can be divided into those which do not require the SOS response to be mutagenic, such as SN1 alkylating agents (e.g., N-methyl-N’-nitro-N-nitrosoguanidine) and base analogs (e.g., 2-aminopurine), and those which do, such as SN2 alkylating agents (e.g., methyl methane sulfonate) and mutagens that produce bulky lesions in the DNA (e.g., UV light). Escherichia coli is proficient for SOS mutagenesis unless mutated in recA or umuDC, but SOS mutagenesis in S. typhimurium is considerably weaker. This deficiency can be overcome by using strains with plasmid pKM101,2 which carries mucAB+, a more active version of the umuDC operon.3 Derivatives of this plasmid that are deleted for conjugal and slow-growth functions4 also can be used to enhance SOS mutagenesis in E. coli. Indeed, with some mutagens, such as aflatoxin B1 and benzo[a]pyrene, the yield of mutations even in E. coli is low unless mucAB+ is present.5 It is likely that the error-prone replication process elicited by the SOS response inserts adenines opposite noninformational DNA lesions6; thus, increasing the activity of the mutational process should amplify certain classes of mutations. From the limited data available, enhancing SOS mutagenesis by prior induction, use of SOS-constitutive mutants, or by supplying mucAB+ activity increases the proportions of GC to TA transversions and all mutations at AT sites, although the specific effect obtained varies among mutagens and treatment.7-12

Strains mutant in various accurate DNA repair pathways are widely available and can be used to increase the yield of mutations.13 Table I lists several such repair pathways and mutations that inactivate them. Strains defective in UvrABC excision repair are particularly useful to increase mutagenesis by a variety of mutagens. Some agents that make bulky DNA adducts, such as aflatoxin B1 and benzo[a]pyrene, are barely mutagenic at all unless this repair pathway is inactivated. In contrast, the mutagenicity of DNA cross-linking agents such as mitomycin C and cis-diamminedi-chloroplatinum(II) (cisplatin) is greatly enhanced by UvrABC excision repair.14,15 The photoreactivation pathway is only active against pyrimidine dimers, but, because these lesions induce the SOS response, photoreactivation can result in a virtual elimination of UV-induced mutation.16 Since photoreactivation requires visible light, it can be circumvented by keeping UV-irradiated cells in the dark. Repair pathways can also be manipulated to yield specific classes of mutations. For example, after exposure to an alkylating agent, strains defective in O6-alkylguanine DNA alkyltransferase (ada) have high frequencies of GC to AT transitions.17 However, when strains defective in 3-methyladenine glycosylase (alkA) are induced for O6-alkylguanine DNA alkyltransferase by exposure to low levels of an alkylating agent, a high proportion of mutations are AT transversions.18

DNA Repair Pathways

Assays for Mutagenesis

Drug resistances are the most versatile assays for mutagenesis in most genetic backgrounds. Resistances to rifampicin (100 mg/liter), streptomycin (200 mg/liter), and nalidixic acid (40 mg/liter) are due to base substitutions.19 Reversions of the amino acid auxotrophies common to many E. coli strains are particularly convenient for monitoring base substitutions (hisG4, argE3) and, in some cases, frameshift mutations (trpE9777).20 In S. typhimurium a variety of revertible mutations in the his operon are available.21 Specific classes of mutations can also be monitored. A set of revertible mutations in the lacZ gene of E. coli can detect each of the six possible base changes.22 By using simple screens, the base changes that revert specific his alleles in S. typhimurium can also be identified.23 A plasmid is available which will allow GC to TA transversions to be monitored by the induction of ampicillin resistance.24

Although only a few mutagenic events can give rise to drug resistances or revert amino acid auxotrophies, a wide variety of mutagenic events lead to loss of gene function. However, assays for gene knockout are generally less convenient. In E. coli LacI mutants can be selected for by growth on the noninducing substrate phenyl-β-d-galactoside. More commonly, loss of gene function, for example, loss of the ability to metabolize a carbohydrate, must be screened for on indicator plates or identified by replica plating. A variety of additional mutational assays are given in Eisenstadt20 and Miller.19

General Methods

The sensitivity and reproducibility of mutagenesis experiments are greatly enhanced by using exponentially growing (log-phase) cells. This is easily accomplished by diluting a saturated culture l : 50 into fresh Luria broth (LB)19 and growing the cells at 37° for 2–3 hr. The cell density should be about 6 × 107 cells/ml, which corresponds to Klett 65 or OD600 0.5. The cells are chilled, centrifuged, washed once, resuspended in the appropriate buffer (see below), and kept on ice until treatment. Most of the methods given below are designed so that disposable microcentrifuge tubes can be used for mutagen exposure. To generate dose–response curves, either the mutagen concentration or the time of exposure can be varied. After exposure to the mutagen, cells are diluted with cold buffer, centrifuged, washed twice, resuspended in cold buffer, and kept on ice. Appropriate dilutions (in 0.85% NaCl plus 0.001% gelatin to prevent clumping) are plated on LB plates to obtain the level of survival relative to untreated control cells.

Mutations are “fixed” into the DNA by replication; thus, a period of outgrowth under nonselective conditions is necessary after mutagen exposure to obtain the maximum number and an unbiased yield of mutations. As soon as possible after treatment, the washed cells are diluted into LB and grown to saturation at 37° with good aeration. These cultures are then titered and plated for mutants. The mutation frequency is the number of mutants divided by the total number of viable cells plated (assuming that all the cells, including the mutants, had equal growth rates in the nonselective medium).

If too few viable cells are inoculated for outgrowth, the progeny of a small number of mutants can dominant the resultant population, giving a large proportion of siblings. To avoid this problem, conditions are adjusted so that at least 106 viable cells are inoculated. If high mutation rates are desired for a general mutant search, or if a particularly sensitive strain is being used, the number of cells treated and the volume of culture for outgrowth must be increased to compensate for the low cell survival.

Reversions can be assayed after outgrowth, as above, or immediately after mutagen exposure by plating the cells on selective plates with a small amount of the required nutrient.25 For example, to assay for reversion of an amino acid auxotrophy, approximately 107 viable cells are plated on a minimal plate containing 100 nmol of the required amino acid. The cells will grow on the plate until the amino acid is exhausted; in 48 hr revertant colonies will appear on the background lawn. Since with this procedure every mutant arises independently, the mutation frequency is the number of mutants divided by the number of viable cells originally plated (although a “cell density artifact” can be introduced26).

“Plate mutagenesis” can be used to quickly test the mutagenic response of a strain or to check the potency of a mutagen. This procedure works well when the mutagenic assay is reversion of an amino acid auxotrophy. Both the cells and the mutagen are added to top agar and plated on minimum medium with a limiting amount of the required amino acid (see above). Alternatively, a small volume of the mutagen is added to a filter-paper disk placed on the surface of the plate.25

Some agents are only mutagenic if activated to short-lived, reactive intermediates. Bacteria do not have cytochrome P-450 enzymes; thus, when using compounds that require these enzymes, activation must be done externally. Aroclor-induced S9-fraction liver microsomes work well for most mutagens and are commercially available. The procedure is described by Ames et al.25 The reaction mixture is 0.1 M sodium phosphate buffer, pH 7.4, with 8 mM MgCl2, 33 mM KCl, 5 mM glucose 6-phosphate, 4 mM NADP, and 20–80 μl/ml microsomes. Small aliquots of glucose 6-phosphate (1 M) and microsomes are kept at − 20° and − 80°, respectively, thawed on ice, and discarded after use. NADP (0.1 M) is also kept at − 20° and thawed on ice, but it can be refrozen and used several times. Cells and mutagen are added to the reaction mixture and incubated for 1 hr at 37°. A range of concentrations of microsomes should be tested at each mutagen concentration to obtain the best combination. The S9 microsome mixture can also be incorporated into the top agar when doing plate mutagenesis (see above).

Specific Methods


A stock solution of N-methyl-N’-nitro-N-nitrosoguanidine (MNNG) is prepared at 1 mg/ml (7 mM) in 100 mM citrate buffer, pH 5.5 (23 mM citric acid, 77 mM sodium citrate). MNNG is inactivated at higher pH and is unstable in phosphate buffer. The solution can be warmed briefly to 37° to dissolve the MNNG. The stock is then dispensed in small aliquots, stored at − 20°, used once, and discarded.

Mid log-phase cells are centrifuged, washed once with 1/2 volume cold citrate buffer, and resuspended at 10 × in the citrate buffer. MNNG is added to 100 μl of cells at 5–100 μg/ml for 5 min at room temperature. One-half milliliter of cold 100 mM sodium phosphate buffer, pH 7 (39 mM NaH2PO4, 61 mM Na2HPO4), is added, and the cells are centrifuged, washed twice, and resuspended in 0.5 ml of the phosphate buffer. After plating for survival, the entire 0.5 ml is added to 5 ml of LB and grown overnight. With reversion assays, mutation frequencies of 10−5 per viable cell with little killing are typical.

Because the major mutagenic lesion induced by MNNG, O6-methyl-guanine, is not very lethal, high mutation frequencies can be produced with this mutagen. For example, exposure to 1 mg/ml for 15 min gave a survival of 10% and a mutation frequency to Rifr of about 10−4 per viable cell. For a mutant search, this level of mutagenesis will give a good probability of recovering mutations in the desired gene. However, MNNG-induced mutations tend to cluster at the replication point, giving a high frequency of double mutants.27 In addition, the mutations that result will be almost exclusively GC to AT transitions.

The adaptive response to alkylating agents is induced by low, nonmutagenic concentrations of MNNG.28 For example, we achieved good induction by diluting overnight cultures grown in Vogel–Bonner minimal medium29 1 : 5 into fresh minimal medium containing 0.5 to 1.0 μg/ml MNNG and incubating the cells with gentle aeration for 4 hr at 32°. Wild-type cells were not mutated by this treatment, but cells mutant in alkA were, with a high proportion of the mutations occurring at AT sites.18

Methyl Methane Sulfonate

Methyl methanesulfonate (MMS) is a volatile liquid (11.8 M when pure). It is stored at 4° in the dark and used only in a chemical hood.

Mid log-phase cells are centrifuged, washed once with 1/2 volume cold E salts29 (57 mM K2HPO4, 9.5 mM citric acid, 17 mM NaNH4HPO4, 0.8 mM MgSO4, pH 7; this buffer is used for convenience—an equivalent 0.1 M phosphate buffer can be used), and resuspended at 10 × in the E salts. To 100 μl of cells, 1–10 μl of MMS is added, mixed well, and incubated for 5 min at room temperature. Alternatively, 5–25 μl of a 1 : 100 dilution of MMS in distilled water is added to 100 μ1 of cells for 30 min at 37°. After exposure, 0.5 ml of cold E salts is added, and the cells are centrifuged, washed twice, and resuspended in 0.5 ml of E salts. The MMS is not inactivated, only diluted, so the entire procedure should be as rapid as possible. Plating for survival and outgrowth are as above. With reversion assays, mutation frequencies of 10−6 per viable cell at 30% survival are typical. MMS induces all classes of base substitutions with a high proportion of GC to TA transversions.18,29 Unlike MNNG, mutagenesis by MMS is SOS-dependent.9

Ethyl Methane Sulfonate

Ethyl methane sulfonate (EMS) is also a volatile liquid (9.2 M when pure). It should be handled as is MMS (see above).

The following method is from Cupples and Miller.22 Mid-log cells are centrifuged, washed twice in cold A buffer19 [60 mM K2HPO4 33 mM KH2PO4, 7.6 mM (NH4)2SO4, 1.7 mM sodium citrate, pH 7], and resuspended at 2 × in cold A buffer. EMS is added at 1.4% to aliquots of the cell suspension in culture tubes, the tubes are sealed with tape, and the cultures are incubated at 37° with gentle aeration. After various times up to 60 min, the cells are centrifuged, washed twice with A buffer, and resuspended in the same volume of A buffer. After plating for survival, 0.5 ml is inoculated into 10 ml of LB and grown overnight. With a 30-min exposure, this procedure gave a mutation frequency of 4 × 10−4 Rifr per viable cell with 56% survival.22 Like MNNG, EMS is mutagenic in the absence of SOS activity14 and induces predominately GC to AT transitions.30

UV Light

Germicidal (shortwave) UV lamps vary greatly in their intensity, but a new 8-W bulb gives a fluence of about 1 J/m2/sec at 20 inches. The UV fluence can be measured with a UV light meter, but it is often easier to generate a dose–response curve with the strain to be used. To stabilize the light fluence, the lamp must be turned on at least 20 min before use.

Mid log-phase cells are centrifuged, washed once with E salts, and resuspended at 1 × in the E salts. One-milliliter aliquots are distributed to sterile 50-mm glass petri dishes; it is important that the bottom of the dish be completely covered with a thin layer of the cell suspension. Working in the dark, the top of each dish is removed and the dish placed under the lamp for an appropriate time (a sheet of foil can be used as a shield before and after exposure). To prevent shading the cells are gently swirled or stirred during exposure. For uvr strains, 2–5 J/m2 corresponds to approximately 50% killing, whereas 10 times this fluence is required for uvr+ strains. Immediately after exposure and for all subsequent manipulations, the cells must be protected from visible light. No washing is necessary. Plating for survival and outgrowth are as above except plates and cultures are incubated in the dark. With reversion assays, mutation rates of 10−5 at 50% survival are typical. UV light induces predominantly GC to AT transitions but also induces all other base changes, frameshifts, and deletions.31 Mutagenesis by UV light is SOS-dependent.26

Depending on the mutational target, S. typhimurium may be poorly mutated by this method.32 For reversion of the hisG46 allele, good mutagenesis was achieved by plating 108 mid log-phase cells on minimal plates containing 100 nmol histidine, incubating the plates for 2 hr at 37°, and then exposing the plates to UV light.12

Aflatoxin B1

Aflatoxin B1 (AFB) is inactivated by water and light. The stock is prepared by adding the appropriate volume of dichloromethane (CH2Cl2) to the vial of AFB as received to give a 10 mM (3.12 mg/ml) solution (this is safer than weighing out an aliquot). Twenty-microliter aliquots are dispensed, evaporated under inert gas (N2 or Ar) or vacuum, and stored at −20° under dry N2 or Ar in tightly capped tubes. Immediately before use 40 μl of dimethyl sulfoxide (DMSO) is added to an AFB aliquot, resulting in a 5 mM solution. This working stock is further diluted in DMSO to the appropriate concentrations. If immediately stored at −20° under dry gas, the 5 mM stock can be used several times, but more dilute stocks are discarded after one use. Gas is dried by passing it through a column of CaSO4; dichloromethane and DMSO are purged with and stored under the dry gas. Although extremely light-sensitive, AFB can be handled under yellow light. It fluoresces blue under UV light, which property can be used to check for contamination.

Mid log-phase cells are centrifuged, washed with E salts, and resuspended at 10 × in cold E salts. Cells are diluted 1 : 5 into S9 mixture (see above) and a 0.5-ml aliquot dispensed for each dose to be used. Five microliters of an appropriate AFB dilution in DMSO is added to give 5–100 μM final concentration. For each treatment, including the control, the same amount of DMSO (1% of the total volume) is added to the cells. The cells are then incubated 60 min at 37°, diluted with 0.5 ml of cold E salts, washed twice, and resuspended in 0.5 ml of cold E salts. Plating for survival and outgrowth are as above. Mutation rates vary greatly among different targets, but a frequency of 10−5 for Rifr at 50% killing is typical for a uvr strain carrying a mucAB+ plasmid (see above). AFB induces primarily GC to TA transversions.33


ICR-191, a chlorinated alkylacridine, can be obtained from Raylo Chemicals (8045 Argyll Road, Edmonton, Alberta, Canada T6C 4A9). It is dissolved in distilled water at 1 mg/ml (2.2 mM) and stored at −20° in the dark. It can be thawed and refrozen several times.

The following method is from Miller.19 Approximately 104 cells from a saturated culture are inoculated into 3-ml aliquots of minimal medium supplemented with 2% LB. Then 5–20 μg/ml ICR-191 is added (protect from light), and the cultures are grown with aeration overnight at 37° in the dark. The cells are then titered and plated for mutants.

Initially, a dose–response curve must be generated. The maximum yield of mutations occurs at a concentration that gives a half-saturated culture after 12 hr. Since many of the mutants will be siblings, several cultures at the optimum dose are grown and one mutant from each selected to obtain independent mutants. True mutation frequencies cannot be obtained with this method, but a ratio of mutants to nonmutants of 1 : 100 is possible using a forward mutation assay. ICR-191 induces frameshift mutations.19


2-Aminopurine (2-AP) is an analog of adenine. To make a 0.6 mg/ml stock solution from the free base, 1.2 mg/ml is dissolved in 0.1 N HCI and titrated to pH 7 with NaOH. An equal volume of 2 × LB (but 1 × NaCl) is added, and the stock is filter sterilized and stored at 4°.

Cells are mutated when growing in the presence of 2-AP, and the method is essentially the same as for ICR-191.19 A dose–response curve is generated by inoculating 102–103 cells from a saturated culture into LB with 0–600 μg/ml 2-AP. After the cultures reach saturation (which may take 48 hr), the cells are titered and plated for mutants. As with ICR-191, several cultures at the optimum dose are then grown to generate independent mutants. Mutation frequencies 100-fold above spontaneous are typical. 2-AP induces GC and AT transitions30; since 2-AP is a DNA base analog, 2-AP mutagenesis is SOS-independent.


1,2-Dibromoethane (EDB) is an extremely volatile, light-sensitive liquid. It is stored at room temperature in the dark and used only in a chemical hood.

The highest levels of mutation are obtained by mutating the cells with gaseous EDB. The following method is adapted from Rosenkranz.34 Using a reversion assay, 108 cells are plated in top agar on minimum medium containing a limiting amount of the required nutrient (see above). A filter paper disk is fixed to the top of the inverted petri dish with a small amount of top agar. Working in a chemical hood, 1–20 μl of EDB is added to the disk, and the plate is closed and immediately sealed with Parafilm. The time elapsed between adding the EDB and sealing the plate is a critical variable and should be as short as possible. Mutants are scored after 2 days of incubation at 37°.

With this method we obtained up to 6000 EDB-induced mutants per plate, which was 10- to 100-fold higher than levels achieved with more conventional treatment methods.35 EDB is poorly soluble in aqueous solutions unless dispersed in a solvent such as DMSO. Because of its lipid solubility, it disrupts cell membranes.36 Thus, its acute toxicity to cells may be unrelated to DNA damage, and the gas-phase treatment may be successful because it allows long exposures to subtoxic doses. Although this method is not widely applicable, it may be useful for other lipid-soluble volatile mutagens. EDB induces predominantly transitions at GC and AT sites, most of which are SOS-independent.35

Mutator Strains

Mutator strains can be used to generate mutations in plasmid-borne genes. The most powerful mutators are those with defects in methyl-directed mismatch repair (mutH, mutL, and mutS), in G–A mismatch repair (mutT and mutY), or in the proofreading activity of DNA polymerase (dnaQ/mutD). Table II lists the increases in mutation frequency and the classes of mutations induced by these defects. To generate mutants, plasmid-containing strains of the mutators are grown to saturation from a small inoculum (103–104 cells/ml), and the DNA is isolated and transformed into another strain.

Mutator Alleles

The mutation rates conferred by mutD5 and dnaQ49, both alleles of dnaQ, which encodes the ε subunit of DNA polymerase III, are dependent on growth conditions. When grown in rich medium, mutD5 strains have mutation rates 10- to 100-fold greater than when grown in minimal medium.37 The mutation rates of dnaQ49 strains are similarly increased by temperature and by growth in salt-free rich medium.38 It has been shown for mutD5, and is likely to also be true for dnaQ49, that the increase in mutation rate is in part due to saturation of the methyl-directed mismatch repair pathway.39 Thus, the specificity of these mutators will tend to change from transversions to transitions as the mutation rates increase.


All mutagens are potential carcinogens, and many are also acutely toxic. The safest procedure is to confine all work with mutagens to a chemical hood with designated pipettors, centrifuges, water baths, etc. Gloves, laboratory coats, and safety glasses (UV opaque for work with UV light) should be worn. After cells are washed free of mutagen, work can be continued at the laboratory bench. Disposal of contaminated material should be as designated in the safety sheet supplied by the manufacturer.


To assemble the methods described here, I drew upon the experience and knowledge of E. Eisenstadt, J. H. Miller, C. Mark Smith, and J. Cairns, all of whom I thank. The work was supported by U.S. Public Health Service Grant CA37880 awarded to the author by the National Cancer Institute.


1. Walker GC. Microbiol Rev. 1984;48:60. [PMC free article] [PubMed]
2. Mortelmans KE, Stocker BAD. J Bacteriol. 1976;128:271. [PMC free article] [PubMed]
3. Walker GC, Dobson PP. Mol Gen Genet. 1979;172:17. [PubMed]
4. Langer PL, Shanabruch WG, Walker GC. J Bacteriol. 1981;145:1310. [PMC free article] [PubMed]
5. Foster PL, Groopman JD, Eisenstadt E. J Bacteriol. 1988;170:3415. [PMC free article] [PubMed]
6. Loeb L. Cell (Cambridge, Mass) 1985;40:483. [PubMed]
7. Fowler RG, McGinty L, Mortelmans KE. J Bacteriol. 1979;140:929. [PMC free article] [PubMed]
8. Fowler RG, McGinty L, Mortelmans KE. Genetics. 1981;99:25. [PubMed]
9. Todd PA, Monti-Bragadin C, Glickman BW. Mutat Res. 1979;62:226. [PubMed]
10. Miller JH, Low KB. Cell (Cambridge, Mass) 1984;37:675. [PubMed]
11. Couto LB, Chadhuri I, Donahue BA, Demple B, Essigmann JM. J Bacteriol. 1989;171:4170. [PMC free article] [PubMed]
12. Eisenstadt E, Miller JK, Kahng L-S, Barnes WM. Mutat Res. 1989;210:113. [PubMed]
13. Kushner SR. In: Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. Neidhardt FC, Ingraham JL, Magasanik B, Low KB, Schaechter M, Umbarger HE, editors. American Society for Microbiology; Washington, D.C.: 1987. p. 1044.
14. Kondo S, Ichikawa H, Iwo K, Kato T. Genetics. 1970;66:187. [PubMed]
15. Brouwer J, van de Putte P, Fichtinger-Schepman AMJ, Reedijk J. Proc Natl Acad Sci U S A. 1981;78:7010. [PubMed]
16. Brash DE, Haseltine W. J Bacteriol. 1985;163:460. [PMC free article] [PubMed]
17. Schendel PF, Robins PE. Proc Natl Acad Sci U S A. 1978;75:6017. [PubMed]
18. Foster PL, Eisenstadt E. J Bacteriol. 1985;163:213. [PMC free article] [PubMed]
19. Miller JH. Experiments in Molecular Genetics. Cold Spring Harbor Laboratory; Cold Spring Harbor, New York:
20. Eisenstadt E. In: Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. Neidhardt FC, Ingraham JL, Magasanik B, Low KB, Schaechter M, Umbarger HE, editors. American Society for Microbiology; Washington, D.C.: 1987. p. 1016.
21. Hartman PE, Aukerman SL. In: Mechanisms of DNA Damage and Repair. Simic MG, Grossman L, Upton AC, editors. Plenum; New York: 1985. p. 407.
22. Cupples CG, Miller JH. Proc Natl Acad Sci U S A. 1989;86:5345. [PubMed]
23. Levin DE, Ames BN. Environ Mutagen. 1986;8:9. [PubMed]
24. Foster PL, Dalbadie-McFarland G, Davis EF, Schultz SC, Richards JH. J Bacteriol. 1987;169:2476. [PMC free article] [PubMed]
25. Ames BN, McCann J, Yamasaki E. Mutat Res. 1975;31:347. [PubMed]
26. Witkin EM. Bacteriol Rev. 1976;40:869. [PMC free article] [PubMed]
27. Guerola N, Ingraham JL, Cerdá-Olmedo E. Nature (London) 1971;230:122. [PubMed]
28. Samson L, Cairns J. Nature (London) 1977;267:281. [PubMed]
29. Vogel HJ, Bonner DM. J Biol Chem. 1956;218:97. [PubMed]
30. Coulondre C, Miller JH. J Mol Biol. 1977;117:577. [PubMed]
31. Miller JH. J Mol Biol. 1985;182:45. [PubMed]
32. Smith CM, Eisenstadt E. J Bacteriol. 1989;171:3860. [PMC free article] [PubMed]
33. Foster PL, Eisenstadt E, Miller JH. Proc Natl Acad Sci U S A. 1983;80:2695. [PubMed]
34. Rosenkranz HS. Environ Health Perspect. 1977;21:79. [PMC free article] [PubMed]
35. Foster PL, Wilkinson WG, Miller JK, Sullivan AD, Barnes WM. Mutat Res. 1988;194:171. [PMC free article] [PubMed]
36. Brem H, Coward JE, Rosenkranz HS. Biochem Pharmacol. 1974;23:2345. [PubMed]
37. Cox EC, Horner DL. Genetics. 1982;100:7. [PubMed]
38. Horiuchi T, Maki H, Sekiguchi M. Mol Gen Genet. 1978;163:277. [PubMed]
39. Schaaper RM. Proc Natl Acad Sci U S A. 1988;85:2432. [PubMed]