plus accompanying article: "Mutations and Misconceptions" B. Determing Mutation Rates in Bacteria.......................................................................p.4 plus accompanying article: "Multiplying Mutants" " " sample problems, Poisson distributionC. The Luria-Delbrück Fluctuation Test.........................................................................p.6 Background..............................................................................................................p.6 Experimental Protocol........................................................................................…p.7 Other Points.............................................................................................................p.9 Extensions................................................................................................................p.11 respreading protocol replica plating Supplementary Material Mutation (Selected excerpts from Chap. 6 of Stent's "Molecular Genetics" Mutations--Preexisting or Acquired?Added: Spontaneous Mutations in Yeast (from Manney and Manney manual)D. Supplementary Material: The Ames Test…………………………………………..p.12

INTRODUCTION One of the fundamental observations regarding heredity is that organisms belonging to the same species generally differ from one another with respect to a given character or trait. Indeed, genetic analysis would not be possible without phenotypic variants. Genetic variants arise because all organisms have an inherent tendency to undergo change from one hereditary state to another; the process, called mutation, is one in which a gene changes becoming a new allele . Mutations arise spontaneously (e.g. due to rare, but finite frequency of persistent errors in DNA replication and/or spontaneous alterations of DNA structure) or can be induced by a variety of xenotoxic agents (i.e. mutagens: chemicals, radiation). Some mutations can be harmful or even lethal to the individual organism and others may end up being beneficial for the species; most are neutral. In this section, we will explore some basic properties of both spontaneous and induced mutations in both prokaryotes (bacteria) and eukaryotes (yeast), using well-characterized protocols that are easily adapted to classroom use. The following experimental activities are included:1. Prokaryotes a. Isolation and characterization of bacterial mutations (2 papers by Thomas Corner on "Mutations and Misconceptions", from The Science Teacher (Dec. 1992; Jan. 1993)b. Distinguishing between mutations as spontaneous, random events vs. an adaptive response to the environment 1.) The Luria-Delbrück Fluctuation Test 2.) The Lederberg replica plating procedure 3.) The Newcombe respreading procedurec. Detection of the mutagenic action of xenotoxic agents: The Ames Test (background material only)

A. THE ISOLATION AND STUDY OF MUTANT BACTERIA(From: Thomas R. Corner, "Mutations and Misconceptions", The Science Teacher, Dec. 1992) PROTOCOL FOR THE ISOLATION OF STREP RESISTANT BACTERIA FROM THE SOIL1. Add 1 g soil to 100 ml sterile dist water; shake extensively (e.g. 25X).2. Using a sterile Q-tip or inoculating loop, streak the suspension out on 6 LB agar plates.4. Pick out several colonies from the plates in step 2 and grow up overnight in 10 ml LB broth (37oC, with shaking) 5. Use inocula from these tubes to streak plates along the gradient, as diagrammed below (use an overnight culture of MM294 as a control)

B. DETERMINING MUTATION RATES IN BACTERIA(From: Thomas R. Corner, "Multiplying Mutants", The Science Teacher, Jan. 1993)

PROTOCOL 1. Isolate different colonies of the same bacterial culture from plates; add to 10 ml LB broth and incubate overnight with shaking at 37oC.2. Subculture 3-4 x by diluting the culture 102 in fresh LB broth and growing overnight. Alternatively, use cultures of different bacteria such as those isolated in the first experiment3. Start experiment with: broth culture of the test bacterium 30 tubes with 10 ml LB broth 1 500 ml flask with 100 ml LB broth (to serve as a sterile blank) 2 ml samples of streptomycin at: 1, 10, and 20 mg/ml4. To the first set of 10 tubes of LB broth, add 0.1ml 1mg/ml streptomycin ( final concentration: 10µg/ml) " " second " " " " " " " " 10mg/ml " " " 100µg/ml " " third " " " " " " " " 20mg/ml " " " 200µg/ml5. Dilute the broth culture 102 in 0.85% NaCl and add 0.1ml to each tube.6. Incubate at 37oC until growth in some tubes is visible; record the number of tubes in each set with growth and no-growth.7. From step 5, continue to dilute the broth culture to give a total dilution of 107; plate 0.1ml in triplicate on LB agar plates.ANALYSIS1. From step 7, determine the average number of bacteria/ml in the test culture (Alternatively, you may wish to follow the method described by Corner in the paper.)2. To calculate the mutation rate, you must be familiar with the Poisson distribution. Given below is a brief sketch of the essentials needed; you also have the description given in the paper by Corner. Finally, included with this activity are some additional examples of Poisson calculations taken from: "Discovering Molecular Genetics" by Jeffrey H. Miller (Cold Spring Harbor, N.Y. 1996) We know that if we flip a coin, half the time it will come up heads (or tails). This understanding comes both from our experience as well as the recognition that there are only two outcomes, each equally likely. Calculating the probability of getting a heads (or tail) is not difficult. But when the probability of two alternate events, p and q, can each range from 0 (impossible) to 1 (certainty), and there are N cases to evaluate in order to know the overall probability of the outcome, the calculations get more complicated. Indeed, one can show there is a distribution of outcomes, each with its own frequency of occurrence. The most general distribution is the binomial distribution: (p + q)N. In general, the binomial distribution is applicable to situations, as in the toss of a coin, in which p and q are equal or nearly so. (When N approaches infinity, the mathematics allow a representation of this distribution by a continuous function, the result of which is the familiar "normal" or Gaussian distribution.) When p is much greater than q (or vice-versa), the binomial distribution takes on a different mathematical form. This form is called the Poisson distribution. Before discussing this, let's do a thought (or, "gedanken") experiment: suppose we were able to break up a 1 ml bacterial culture into small chunks, each having the volume of a bacterium; what is the chance that if we were to pick one of these chunks, it would contain a bacterium? First, we need to know the volume of a bacterium. The simplest way to determine this is to approximate a "typical" bacterium as a sphere of volume 1µm3 (a reasonable approximation, given the known dimensions of many bacteria), or (10-4cm/µm)3 = 10-12cm3 = 10-12ml/bacterium. If we further assume that an overnight culture has a cell density of 109bacteria/ml, then there are (109)(10-12) = 10-3 cell volumes per ml of culture....so you see that even in a dense culture, very little (1/1000) of the 1 ml is actually occupied by cells. Now let's break the 1 ml up into the 1012 chunks, each having a volume of 10-12ml: what's the chance that any given chunk will be a bacterium? The answer, of course, is very small (10-3). This is a condition in which the Poisson distribution applies: the probability of selecting no cell is much greater that the probability of selecting a cell. The general expression for the Poisson distribution is: where: P(r) is the probability of obtaining the outcome, r m is the mean value of the outcomeApplying this to our gedanken experiment, r = 0 is the most likely outcome; the probability, P(0) = e-m (since m0=1 and 0!=1). This makes the probability of actually getting a cell, 1-P(0), since all the individual probabilities must sum to 1. The essential point is m is known, P(0) can be calculate; conversely, if P(0) is known, m can be determined. It is this point that we will use in this exercise.Let's suppose that in the experiment, the set that had some growth was the 10µg/ml streptomycin set, and there were 7 tubes that showed no-growth. Here, r is the number of cells/ml in each culture. From the discussion above, if we know the number of subcultures that produced 0 cells (no growth), we can calculate m, the mean number of bacteria/ml (keep in mind that we added 0.1ml of the original culture, so we must eventually take that into account when calculating m). So, r = 7/10 = 0.7; and P(0) = 0.7 = e-m. This can be determined from tables or calculator; m = 0.36. Thus, in the original culture, there must have been 0.36/0.1 = 3.6 mutants resistant to 10 µg/ml streptomycin. We now know N, the number of bacteria/ml in the test culture, and m, the average number of mutants in the culture. Therefore, we're able to calculate the rate of mutation. The rate of mutation is given simply by:m = number of mutations in the original culture = (mutations/cell/generation)(cells/culture)(generations/culture) = a x N x n where a is the mutation rate per cell, and n is the number of generations (assuming the culture started out, on the average, with a single cell.)So, a = (m/N)n. According to:N = No2n, if we assume that N = 109 and No = 103; then n ≈10.

So, a = (3.6/109)(10) = 3.6 x 10-8 mutations/bacterium/generation (i.e. after 10 divisions, roughly 4 mutations will have occurred; or, in a population of 109 cells, there will be about 4 mutants.)

A. BACKGROUND

When virulent bacteriophage are added to a turbid culture of bacterial cells that are sensitive to the phage, after a few hours the culture becomes clear due to multiplication of the virus inside the cells which ultimately burst (i.e. lyse) to release the particles of phage progeny. After further incubation, the culture may again become turbid due to the growth of bacteria which are resistant to the bacteriophage.

Two hypotheses have been advanced to account for the origin of these resistant variants:

1. The adaptation hypothesis: according to which every cell has a small probability of being induced by the phage itself so that it can survive and grow in the presence of the phage, this adaptation then being passed on to its descendants .

2. The spontaneous-mutation hypothesis: which states that every cell has a small probability of mutation during its life-time from phage-sensitivity to phage-resistance, whether phage is present or not. The progeny of such a resistant cell will also be resistant unless back-mutation occurs.

In 1943, Luria & Delbrück devised the fluctuation test to decide between these two hypotheses . An important difference between the two alternative ideas is that, according to the adaptation hypothesis, the bacterial population is homogeneous before the phage is added. Whilst, according to the mutation hypothesis, the population is not homogeneous, since mutation to resistance may occur at any time during the growth of the culture before the phage is added. The number of bacteria resistant to the phage will thus depend upon whether the first mutation to phage resistance occurred early or late in the growth of the culture.

Thus, according to the adaptation hypothesis the probability of any bacterium becoming resistant after contact with the phage should be the same for all the bacteria in the culture. The adaptation hypothesis therefore predicts that there will be no large fluctuations in the numbers of resistant bacteria from culture to culture in a parallel series to which phage is added. The fluctuations should in fact be no greater than those encountered in a series of samples all taken from the same culture.

On the other hand, according to the hypothesis of spontaneous mutation, the time of occurrence of a mutation in a series of parallel cultures will be subject to random variation. Cultures in which a mutation occurs early will contain large clones (large numbers) of resistant cells, while cultures in which mutation occurs late will contain small clones (very few) resistant cells. In other words, the mutation hypothesis predicts that resistant bacteria will arise as clones in the culture, whilst the adaptation hypothesis does not. The mutation hypothesis thus leads to the prediction that there will be larger fluctuations in the numbers of resistant mutants from culture to culture in a parallel series, than from a series of samples taken from the same culture. The experiment carried out by Luria and Delbrück measured resistance to phage T1 in E. coli B, and the results analyzed mathematically . Basically, the argument relied on an analysis of the variance of the plating results: whether the variance (i.e the "spread" on the different plates was narrow and close to the mean of all the plates, or was much greater than the mean. According to their argument, a large variance would be consistent with the spontaneous mutation hypothesis, whereas a small variance would support the adaptation hypothesis. Their results clearly were consistent with the former hypothesis.

The fluctuation test has been applied to investigate the origin of a large number of different types of bacterial variant in several bacterial species. The characters investigated include resistance to bacteriophages of various kinds, resistance to various antibiotic drugs (e.g. streptomycin, penicillin, sulfonamide) and to radiation, independence of growth-factor requirements, and the ability to ferment various carbohydrates. In each case the variants have shown a 'clonal' distribution indicative of mutation. In addition, other investigators approached the same question with more direct (but certainly not more elegant) approaches and came to the same conclusion.

The following experiment will allow you to repeat this classic experiment (but use the easier-to-handle, phage T4). Data from the original paper are included as well as a diagrammatic analysis to aid in understanding the method of analysis. In addition, other approaches are outlined in accompanying diagrams and protocols suitable for the classroom.

References

Luria, S.E. & M. Delbrück (1943) Genetics , 28, 491.

B. EXPERIMENTAL PROTOCOL

1. Dilute an overnight culture of E. coli (K12, strain MM294; T4 sensitive) 10-7 in LB broth

a. Add 1.0 ml of the 10-7 dilution to each of the 10 tubes.

b. Add 10 ml of the 10-7 dilution to the 50 ml flask.

2. Incubate overnight at 37oC, with shaking.

3 Next day:

c. Also dilute the flask suspension (i.e. the "batch" culture) 10-5 and add 0.1 ml to soft

d. Pool the 10 tubes, dilute 10-5 and plate 0.1 ml on soft agar in triplicate.

e. Incubate plates overnight at 37oC.

C. ANALYSIS

1. A convenient statistical measure of the degree of fluctuation of an observed parameter about its mean value is the variance. This is given by:

Where xi is the number of T4 resistant colonies in the ith tube ( i goes from 1 to 10), is the mean of the observed numbers of T4 resistant colonies, and N is the number of observations (10.) If the plating method is reliable, fluctuation among different samples should be due to random sampling only, and the variance from a series of samples should be close to the mean.

Look at the data from the Luria-Delbrück paper ( Tables 1, 2, and 3): in every experiment, the fluctuation of the numbers of T1 resistant bacteria in the separate cultures is much higher than could be accounted for by the sampling errors -- in striking contrast to the results of plating from the same (batch) culture ( Table 1.)

2. Another way to look at this is in the accompanying schematic diagram of this experiment (see Figure 6-4 in the accompanying xeroxed material).

a. Both parts A and B of the figure represent a hypothetical example of 4 cultures of bacteria, each grown from a single T1 sensitive founder bacterium to a final number of 16 bacteria (giving a grand total of 64) upon spreading each culture on plates with T1 phage. (Luria-Delbrück used T1 rather than T4 -- this makes no difference in the argument.)

Upper Diagram: assuming Tonr (i.e. " T-one resistance") is induced upon contact with T1 in a fraction of cells; each culture yields one plate with the number of resistant colonies equal to the number of shaded cells. Summed over the total progeny, there is an average mutation frequency of 10/64 = 0.15, where the 10 colonies are randomly distributed over the 4 plates if all the Tonr bacteria are equally eligible to acquire resistance when exposed to phage. The variation in the number of resistant cells about the mean value, = 10/4 = 2.5, should be small (no greater than that characteristic of random sampling processes.) If the variance is equal to the mean, , then:

Variance = 1

Putting this to the test, we see that:

= 1.1

(where, according to the diagram, 2.5 is the mean number of shaded cells in the four cultures, and the numbers, 3,1,5,1, are the actual number of shaded cells in cultures 1,2,3,4, respectively)

Lower Diagram: assuming Tonr arises by spontaneous mutation during Tons growth before exposure to T1. Here, only 2 Tons to Tonr mutations have occurred during the 64 - 4 = 60 cell divisions that produced the 64 bacteria in the 4 cultures from the 4 founder cells. This corresponding to a probability of 2/60 = 0.03 Tonr mutations/cell generation. One mutation occurred in the last generation of culture 1, giving rise to 2 Tonr cells; the other occurred in the fourth generation of culture 3 growth, giving rise to 8 Tonr bacteria. The other two cultures produced no Tons to Tonr mutations. The final result is the same as part A: a total of 10 Tonr bacteria, The mean is the same also: (2+0+8+))/4 = 2.5 Tonr bacteria per culture. But:

Moreover, the difference between the two situations increases progressively as the culture grows since, with more bacterial multiplication in the parallel cultures there is an opportunity for the appearance of large mutant subclones in a few of the cultures. In other words, the occurrence of spontaneous mutations, and hence of clones of mutant cells at various stages of growth of individual cultures, would result in much greater fluctuations in the observed number of Tonr colonies per culture than the random induction of the resistance following contact with the phage at plating.

D. OTHER POINTS

Results of these experiments also allow calculation of the mutation rate (probability of mutation per cell per generation). One method makes use of the fact that the number of mutations in a series of similar cultures should be distributed according to Poisson statistics; the average number of mutations per culture can be calculated from the proportion of cultures containing no resistant bacteria at the moment of the test, according to:

where: r = average number of mutants per culture

a = spontaneous mutation rate per cell per generation

N = number of cells per culture

so that, aN = number of mutations per culture

For r = 0 (no mutations on the assay plate),

Inspection of the individual culture data (see Table 6-1 below the diagram in the attached xeroxed material) reveals that 11 of the 20 cultures contained no Tonr mutant bacteria (i.e. did not produce the Tons-->Tonr mutation). Hence P(0) = 11/20 = 0.55, and since N = 0.2 x 108 bacteria, it follows from the above equation that a = (-ln 0.55)/0.2 x 108 = 3 x 10-8 mutations per cell per generation.

Try this calculation with your results.

The final point is somewhat ironic. The theory upon which this paper is based is, of course, quite solid, and so is the experimental approach. The choice of phage, however, we now know was crucial -- and lucky. They chose to work with "virulent" bacteriophage -- those that infect and kill their host directly and rapidly. Had they happened to pick a phage type that later became known as "temperate" (establishes a latent infection which can be provoked to kill the host later on), Luria and Delbrück would have had to conclude that the bacterial variants acquire their resistant character by contact with the antibacterial agent on the test agar plate, and this would have contributed, willy nilly, to the fortification of the last stronghold of Lamarckism

mutation frequency:

EXTENSIONS

Included below are two experimental protocols based that address the same question asked by Luria and Delbrück; they are based on classical papers published several years after the fluctuation technique and approaches that provided a more direct, visual answer (rather than one that required mathematical inference). These might be more useful for 8th and 9th graders, than trying to have them understand both the biology and the math in the previous protocol.

The first is the "respreading" experiment of Newcombe; a copy of a protocol designed for high school students is included. Briefly, this involves spreading a phage-sensitive strain of E.coli on growth agar and incubating a time sufficient for the cells to have divided several times (but not enough to form a visible colony. These invisible colonies on half the plates are then respread (to scatter out the progeny of the initial divisions so that they become new foci for colony growth) and a suspension of phage is sprayed evenly over the plate; these are compared to plates left undisturbed and sprayed with phage. According to the spontaneous mutation hypothesis, each colony on the undisturbed plate can give rise to only one resistant colony (no matter how many resistant descendants are in the colony). In contrast, the respread plate should contain many more resistant colonies since the progeny were separated. The alternate hypothesis would predict the same number under the two conditions.

The second is the replica plating experiment of Lederberg. Again, a phage-sensitive strain is plated and incubated as above, except macrocolony formation is allowed to occur. At that time, these plates are replica-plated onto plates already spread with the phage. A few phage-resistant colonies will be seen; the spatial distribution of the colonies is the same on alLB plates, showing that there existed colonies on the master plate that had never been exposed to the phage that had arisen spontaneously (see attached diagram).

Since it is easier to assay for mutations to streptomycin resistance than phage resistance in these experiments (it is difficult to spray the phage on evenly and reproducibly), we shall describe protocols for these two techniques using strep resistance mutations.

1. Spread about 5 x 104 strs E.coli on each of 12 LB agar plates; incubate 5 hrs. at 37oC.

2. Respread the microcolonies on 6 of the plates.

3. Overlay all 12 plates with L-soft agar containing 10µg/ml streptomycin and incubate at 37oC for another 12 hrs. (until macrocolony formation is observed).

Replica plating

1. Spread about 107 strs E.coli onto 2 LB agar plates and incubate overnight at 37oC.

2. Replica plate each onto 3 LB agar plates containing 10µg/ml streptomycin.

3. Incubate overnight at 37oC.

D. SUPPLEMENTARY MATERIAL

BACKGROUND

To understand the mechanisms of gene mutation requires analysis at the level of DNA and protein molecules. The table below draws together this information.

TYPE OF MUTATION RESULT/EXAMPLE

Forward Mutation

at the level of DNA:

transition Pur replaced by another Pur or Pyr replaced by another Pyr

AT==>GC GC==>AT CG==>TA TA==>CG

transversion Pur replaced by a Pyr or Pyr replace by a Pur

AT==>CG AT==>TA GC==>TA GC==>CG

TA==>GC TA==>AT CG==>AT CG==>GC

at the level of protein:

silent mutation triplet codes for same amino acid

AGG==>CGG; both code for Arg

neutral mutation codon specifies different but functionally equivalent amino acid AAA==>AGA; changes basic Lys to basic Arg

missense mutation codon specifies a different, non-functional amino acid

nonsense mutation codon signals chain termination

CAG==>UAG; changes codon for Gln to an amber terminator

single nucleotide pair addition Any addition/deletion that isn't a multiple of 3 changes the

or deletion (frameshift) or multiple reading frame, resulting in different amino acids from that point on

nucleotide addition/deletion

Reverse mutation forward reverse

exact reversion AAA(Lys)==>GAA(Glu)==>AAA(Lys)

wild type mutant wild type

equivalent reversion forward reverse

UCC (Ser)==>UGC(Cys)==>AGC(Ser)

wild type mutant wild type

CGC(Arg, basic)==>CCC(Pro, not basic)==>CAC(His, basic)

wild type mutant pseudo wild type

Intragenic suppressor mutations CAT CAT CAT CAT CAT CAT

frameshift of opposite sign at + - (bases added and deleted)

second site within gene resulting reading frame:

CAT XCA TAT CAT CAT CAT

ok bad bad ok ok ok

second-site missense mutation A second distortion that restores a more or less wild type protein

conformation after a primary distortion

Extragenic suppressor mutations

nonsense suppressors A gene (e.g. for Tyr tRNA) undergoes a mutational event in its

anticodon region that enables it to recognize and align with a

mutant nonsense codon (e.g. amber, UAG) to insert an amino acid(Tyr) here and permit completion of translation

missense suppressors Usually caused by a change in tRNA anticodon (e.g. an abnormal tRNA that carries Gly but inserts it in response to an Arg codon; in this case, all Arg codons are mistranslated, but since the abnormal substitution is usually very inefficient the observed mutation is often not lethal

frameshift suppressors A 4 ntd anti-codon in a single tRNA can read a 4 letter code caused by a single bp insertion (very rare)

physiological suppressors A defect in one biochemical pathway is circumvented by another mutation (e.g one that permits more efficient transport of a compound produced in smaller quantities due to the original mutation)

1. Errors in DNA replication (in both cases, K ≈ 10-4)

a. tautomeric shift in base structure: keto=====>enol; amino====>imino; results in different H-bonding pattern causing base mispairing (transition mutations)

c. frameshifts due to looping out of regions of DNA single strands during replication

d. deletions and duplications can occur a repeated sequences; the mechanism isn't well understood but probably arises from slippage and mispairing

2. Spontaneous lesions

b. oxidatively damaged bases (O2•, H2O2, OH•, normal products of aerobic

metabolism, can result in altered Pur and Pyr, resulting in mutation and have been

implicated in a number of human diseases)

3. Induced lesions: Mutagens induce mutagenesis by:

a. Incorporation of base analogs: some chemical compounds are sufficiently similar to

c. Intercalating agents: Certain aromatic, planar molecules (e.g. proflavine, acridine

The modern environment exposes each individual to a wide variety of chemicals in drugs, cosmetics, food preservatives, pesticides, compounds used in industry, pollutants, and so on. Many of these compounds have been shown to be carcinogenic and mutagenic (e.g. the food preservative AF-2, the food fumigant ethylene dibromide, the antischistosome drug hycathone, several hair-dye additives, and the industrial compound vinyl chloride----all are potent and some have subsequently been subjected to government control). However, hundreds of new chemicals and products appear on the market each week. How can such vast numbers of new agents be tested for carcinogenicity before much of the population has been exposed to them?

Many test systems have been devised to screen for carcinogenicity. These tests are time-consuming, typically involving laborious research with small mammals. More rapid tests do exist that make use of microbes and test for mutagenicity rather than carcinogenicity. The most widely used test was developed in the 1970s by Bruce Ames, who worked with Salmonella typhimurium. This Ames test uses two auxotrophic histidine mutations, which revert by different molecular mechanisms. Further properties were genetically engineered into these strains to make them suitable for mutagen detection. First, they carry a mutation that inactivates the excision-repair system. Second, they carry a mutation that eliminates the protective lipopolysaccharide coating of wild-type Salmonella, to facilitate the entry of many different chemicals into the cell.

Bacteria are evolutionarily a long way removed from humans. Can the results of a test on bacteria have any real significance in detecting chemicals that are dangerous for humans? First, the genetic and chemical nature of DNA is identical in all organisms, so that a compound acting as a mutagen in one organism is likely to have some mutagenic effects in other organisms. Second, Ames devised a way to simulate the human metabolism in the bacterial system. In mammals, much of the important processing of ingested chemicals occurs in the liver, where externally derived compounds normally are detoxified or broken down. In some cases, the action of liver enzymes can create a toxic or mutagenic compound from a substance that was not originally dangerous. Ames incorporated mammalian liver enzymes in his bacterial test system, using rat livers for this purpose.

Of course, chemicals detected by this test can be regarded not only as potential carcinogens (sources of somatic mutations) but also as possible causes of mutations in germinal cells. Because the test system is so simple and inexpensive, may laboratories throughout the world now routinely test large numbers of potentially hazardous compounds for mutagenicity and potential carcinogenicity.

The experiment is based on the fact that histidine-requiring mutants of S. typhimurium cannot grow unless they are provided with the amino acid, histidine (thus, they are incapable of growing on nutrient agar that lacks histidine). The histidine requirement can be eliminated by reversion of the mutation that produced it. Therefore, one can screen for compounds that are mutagenic by looking for the appearance of non-histidine-requiring revertants from the histidine-requiring mutants after treatment with suspected mutagens.

REFERENCES:

Ames et al. (1973) "Carcinogens are mutagens: a simple test system combining liver homogenates for activation and bacteria for detection" Proc. Nat. Acad. Scie (U.S.) 70, 2281.

Ames et al (1975) "Methods for detecting carcinogens and mutagens with the Salmonella/mammalian microsome mutagenicity test." Mutation Res. 31, 347.

McCann et al (1975) "Detection of carcinogenous mutagens in the Salmonella/microsome test; assay of 300 chemicals" Proc. Nat. Acad. Sci. (U.S.) 72, 5135

Maron and Ames (1983) "Revised Methods for the Salmonella Mutagenicity Test" Mutation Res. 113, 173-215.