Mutation: Where Variation Comes From

Grow ten small cultures of the same E. coli to the same size, spread each one on a plate covered with a bacterial virus, and you may find 52 resistant colonies on one plate while five plates have none.

The cells were identical, the virus was identical, and the plates were treated alike. Why such different numbers? The answer decides an old question in biology: does the environment that selects a useful variant also create it? To answer it you first need words for what a variant is.

The idea: a mutation is a change in sequence, sorted two ways

A mutation is a heritable change in the DNA sequence. A point mutation changes one base pair, or adds or removes a few. You can sort point mutations in two independent ways.

By the change in DNA. A transition swaps a purine for a purine (A and G) or a pyrimidine for a pyrimidine (C and T). A transversion swaps a purine for a pyrimidine or the reverse. Each base has one transition partner and two transversion partners, so there are 4 possible transitions and 8 possible transversions. Even so, in most organisms studied, transitions are the more common kind. An insertion or deletion (an indel) adds or removes bases.

By the effect on the protein. Inside a coding sequence:

• A synonymous (silent) change gives a codon for the same amino acid.
• A missense change gives a different amino acid. Some missense changes leave the protein working; the label does not tell you the severity.
• A nonsense change turns an amino-acid codon into a stop codon.
• A frameshift is an indel whose length is not a multiple of three, so every codon downstream is read in the wrong groups.
• An in-frame indel adds or removes a multiple of three bases. The frame downstream is kept, and the protein gains or loses amino acids.

Think of a sentence written in three-letter words: THE CAT ATE THE RAT. Change one letter and one word changes. Delete one letter and everything after it becomes nonsense. Delete a whole word and the rest still reads.

“Silent” describes the codon, not the consequence. Exons also carry signals that guide splicing and affect how stable an mRNA is and how fast it is translated, so a synonymous change can still lower the amount of protein. How the code is read is the subject of the genetic code lesson.

Does selection cause the mutation? Luria and Delbrück

When E. coli B is flooded with phage T1 (called virus α in the paper), nearly every cell dies, but a few give resistant colonies. Two explanations were on the table in the early 1940s. Either the virus induced resistance in a few cells it touched, or resistant mutants had arisen by chance during growth, before the virus arrived.

Salvador Luria and Max Delbrück saw that the two ideas predict different patterns of numbers. In their 1943 paper in Genetics, Luria did the experiments and Delbrück the theory.

• If the virus induces resistance, every cell on the plate has the same small chance of becoming resistant. Counts from separate cultures should scatter only as much as random sampling allows, like raindrops on paving stones. The variance should be close to the mean.
• If mutation comes first, a mutation early in a culture’s growth leaves a large clone of resistant descendants (a “jackpot”), a late one leaves a few, and many cultures have none. The variance should be far larger than the mean.

They compared many independent cultures with repeated samples from a single large culture, which measures plating noise alone. The independent cultures scattered far beyond sampling error, and they concluded that, in their system, resistance arose independently of the virus. Later work has shown that mutation rates can change with a cell’s condition (pulses of mutagenesis after alkylation damage have been seen in single E. coli cells), and the debate over stress and mutation continues. None of it reverses the 1943 result for phage resistance.

Worked problem. Ten independent 0.2 mL cultures and one bulk culture of E. coli each grow to about 2 x 10^8 cells per culture volume. Each independent culture, and ten 0.2 mL samples of the bulk culture, are plated on T1. Assume every mutant survives plating and is counted, mutants grow like the parent, no mutations arise after plating, and the number of divisions in a culture roughly equals its final cell number.

Independent cultures: 0, 0, 1, 0, 3, 0, 0, 52, 2, 2. Bulk samples: 4, 8, 6, 3, 9, 5, 7, 6, 8, 4.

1. Means. Both sets sum to 60, so both means are 6.0.
2. Variances. For the independent cultures, the squared deviations from 6 are 36, 36, 25, 36, 9, 36, 36, 2,116, 16 and 16, summing to 2,362. Divided by 9, the sample variance is about 262. For the bulk samples the squared deviations sum to 36, so the variance is 36/9 = 4.0.
3. Read it. The bulk samples have a variance close to their mean, as Poisson sampling predicts. The independent cultures have a variance about 44 times their mean. Mutation before exposure fits; induction does not.
4. Rate. Five of ten cultures have no mutants, so p0 = 0.5. If mutation events per culture follow a Poisson distribution with mean m, then p0 = e^-m.

m = -ln(p0) = -ln(0.5) = 0.69 mutation events per culture

rate = m / divisions = 0.69 / (2 x 10^8) = 3.5 x 10^-9 per cell per division

Now compare mutant frequency, the fraction of cells that are mutant. In the jackpot culture it is 52 / (2 x 10^8) = 2.6 x 10^-7, about 75 times the rate, from what was probably a single early event. Frequency counts cells; rate counts events per division. Confuse them and one lucky culture looks like a mutator.

Replica plating: seeing the mutants before the test

The fluctuation test is an argument from statistics. In 1952 Joshua and Esther Lederberg gave a direct one. They pressed sterile velvet onto a master plate of colonies grown without any selecting agent, then printed the velvet onto fresh plates, each receiving the colonies in their original positions. Printed onto plates with phage or streptomycin, E. coli K-12 strains gave resistant colonies at matching positions on separate prints.

Matching positions point back to particular spots on the master. The Lederbergs took samples from those spots, from cells that had never touched the phage or the drug, and through repeated rounds of growth and re-plating purified resistant cultures from them. The mutants existed before the selection that revealed them. Replica plating also became a workhorse for finding mutants that fail to grow on one medium, the start of many screens like those in complementation and pathways.

Where mutations come from: spontaneous and induced

Spontaneous mutations arise from the cell’s own chemistry and copying, with no applied mutagen.

• Replication errors. DNA polymerase sometimes inserts the wrong base (see DNA replication). Rare alternative forms of the bases, called tautomers, were proposed as one cause of mispairing; that model is historical and now hedged.
• Deamination. Cytosine loses its amino group and becomes uracil, which repair enzymes recognize as foreign. 5-Methylcytosine deaminates to thymine, a normal base, so the G:T mismatch is harder to resolve correctly and CpG sites, where cytosines are often methylated in vertebrates, are hot spots for C:G to T:A transitions. The chemistry is in nucleotide metabolism.
• Depurination. The bond between a purine and its sugar breaks, leaving a site with no base to copy.
• Slippage. In runs of repeated bases, the new strand can loop out or slip back during copying, adding or removing repeat units.

Induced mutations come from outside agents, called mutagens.

• Base analogs such as 5-bromouracil are built into DNA in place of a normal base and mispair, giving transitions.
• Alkylating agents such as ethyl methanesulfonate (EMS) add groups to bases. EMS chiefly gives G:C to A:T transitions.
• Intercalators such as proflavine slip between base pairs and favor single-base insertions and deletions, so frameshifts.
• Ultraviolet light joins adjacent pyrimidines into dimers. A dimer is damage, not yet a mutation; whether it becomes one depends on repair and on how polymerases copy past it.
• Ionizing radiation breaks DNA strands and damages bases.

Induced mutation began with Hermann Muller, who reported in Science in 1927 that X-rays greatly raise the mutation rate in Drosophila, in experiments that counted new lethal mutations on the X chromosome. He received the Nobel Prize in 1946. How cells fix damage before it becomes mutation is the subject of DNA repair and recombination.

The Ames test and what it can and cannot say

Bruce Ames’s test uses Salmonella strains that cannot make histidine because of a mutation in a histidine gene. On a plate with only a trace of histidine, they divide a few times and stop, unless a new mutation restores histidine synthesis. Each revertant colony marks such an event, sometimes at the original site and sometimes at a second, compensating site. Strain TA100 reverts mainly by base-pair substitution; TA98 reverts mainly by frameshift. Plates are run with and without S9, a rat-liver enzyme fraction, because many carcinogens become reactive only after liver metabolism. In 1973 Ames and colleagues reported that liver homogenate converted 18 tested carcinogens into frameshift mutagens.

Two limits matter. A positive result shows that a chemical, or a metabolite, mutates bacteria; it does not measure cancer risk in people. A negative result does not prove safety, because some carcinogens act without mutating DNA or through processes bacteria lack. In McCann and colleagues’ 1975 set, 156 of 174 tested carcinogens were positive, a figure for that set of chemicals.

Design problem

TA100 gives about 120 spontaneous revertants per plate and TA98 about 30. Compound X is tested at three rising doses. TA100 without S9: 125, 118, 130. TA100 with S9: 240, 480, 950. TA98 with S9: 30, 28, 35. What does X do, and what does the result not show?