Bacteria take the Chance out of Evolution
Contents Updated: Saturday, April 15, 2000
- Evolution by Natural Selection
- Luria and Delbruck
- Cairns’ Experiments
- Cairns’ Deciding Experiment
- A Mechanism Needed
Evolution by Natural Selection
Bacteria can mutate in ways that specifically enhance their survival, according to a paper published in Nature (vol 335, p 142). This discovery, by John Cairns, Julie Overbaugh and Stephan Miller at The Harvard School of Public Health in Boston, challenges one of the cherished tenets of Darwinian natural selection—that mutation is spontaneous and random.
Evolution by natural selection has been considered to be a two stage process. First, mutations crop up with no regard for what is happening to the organism. They occur at random. If one of those mutations confers a reproductive advantage on its holder, that mutation will be favoured by natural selection and the genetic make-up of the population wlll change. It is selection that adds the directional component to evolution by filtering out almost all of the randomly-produced mutations. In Cairn’s experiments, by contrast, bacteria produced mutations in dircet response to a change in their environment.
Luria and Delbruck
The apparent proof that mutation occurs independently of selection came from the so called slot-machine experments conducted in the 1940s by Salvador Luria and Max Delbruck. They used bacteria which are normally sensitive to bacteriophages—viruses that generally kill the bacteria they infect. Sometimes, however, a culture of bacteria infected with phage will throw up bacteria which are resistant to the phage. These grow into visible colonies even in the presence of phage. Luria realised that the statistical distribution of resistant bacteria could tell him when these bacteria acquired the mutation that protected them from phage.
If the bacteria mutated specifically in response to phage, the probability of getting a resistant colony would depend only on the number of bacteria exposed to the phage. In several repetitions of the same experiment, the probability of getting a resistant colony would follow a Poisson distribution.
If, by contrast, the bacteria mutated at random and were then selected by adding the phage, the number of resistant colonies would depend on when, in the life of the culture, the mutation had taken place. Usually, the mutation would occur late in the experiment, when there were many more bacteria present to mutate. The mutant would produce only a few descendants before being challenged by the addition of phage, and so would give rsse to only a few colonies. Very occasionally, however, the mutation would happen early on and the mutant would produce many more offspring
Luria and Delbruck grew cultures of E. coli under ideal conditions and then exposed the culture to bacteriophages. They spread the infected culture on agar plates and counted the number of bacterial colonies that grew. Each colony represented the offspring of a single bacterium resistant to phage. In the event, the number of resistant colonies in some expenments was 100 times greater than in others. There was no hint of a Poisson distribution. Luria and Delbrock’s result thus firmly supported the idea that mutations arise at random and may then be selected. And further experiments on resistance to antibiotics confirmed this view.
Cairns Experiments
Calrns, however, points out that these classic experienents have a flaw. Although they prove that some mutations do arise spontaneously and at random, they do not rule out the possibility that others might be non-random and directed at meeting a particular environmental challenge. Phage, for example, kllls all bacteria that are not resistant. The bacteria have no chance to "try" to become resistant. To overcome this objection, Calrns and his colleagues studied less drastic selection, where mutants are rewarded by better growth, but non-mutants survive so that, in Cairns’s words, "they can at least have the opportunity to perform directed mutation".
In the first experiment, the team looked at bacteria unable to use the sugar lactose. They have a mutation which prematurely stops the decling of the gene for beta-galactosidase, one of the enzyymes involved in lactose fermentation. When a culture of these bacteria was tranferred to a medium that contained lactose, colonies appeared of mutants that were now able to use the lactose. Some of these were clearly the result of spontaneous mutations that had occurred while the cultures were growing. But others took much longer to appear, suggessing that the mutation took place late in the experiment, once the bacteria had been forced to try to use lactose. The number of these "late" mutants followed a Poisson distribution, as would be expected if they occurred in response to selection.
In this first experiment, then, E. coli somehow produces the most appropriate mutation, one that changes the full stop in the beta galactosidase gene into a readable codon for the correct amino acid. The replication of DNA is an error-prone process at the best of times—with special proofreading enzymes that correct the many mistake—and so, as Cairns admits, this one result does not seem too hard to believe. He followed it up by looking for much larger mutations which would seldom, if ever, arise by chance.
Cairns and his colleagues used a strain of E. coli that contains an extra piece of DNA from a bacteriophage called Mu. This insert knocks out two enzymes, one for using arabinose and one for using lactose. Mutants able to ferment these two sugars do arise, but slowly and only when those are the only sources of energy available. Mutants have never been deleted in the absence of selection. If these bacteria are forced to use lactose and arabinose, however, and mutant colonies start to appear after three to four days.
In this case, the bacteria get rid of the interfering Mu DNA only when they need to. As Cairns says, these bacteria "either can prevent useless mutations from occurring or can destroy unsuccessful mutants soon alter they arise", because those mutations are never present in the population, waiting for circumstances to change and select them.
Cairns’ Deciding Experiment
Cairns added lactose either immediately after spreading the bacteria on agar plates, 24 hours later or 72 hours later. The graph shows the number of colonies that appeared, on the left in real time and on the right relative to the time at which the lactose was added. Lactose causes the late mutants to appear Delaying the lactose delays their appearance, but does not affect the number of mutants.
The final experiment that Cairns and his colleagues report comes even closer to the sort of selection one might expect in the real world. It exploits a technique developed early this century for classfying bacteria according to their ability to use cetain sugars. Wild E. coli, for example, can ferment lactose whereas Shigella and Salmonella cannot. Some species, however, are so-called "late" fermenters of certain sugars. That is, it may take a week or more belore bacteria start using an unusual food source. Shigella sonnei, for example, is a late fermenter of lactose.
In fact, bacteria possess several such cryptic genes, which are brought into play only when needed. The mechanism of activation varies. Sometimes, another piece of DNA is inserted upstream of the desired gene and switches that gene on.
In other cases, The DNA sequence needs several specific changes before the cryptic gene will function properly. Cairns and his group studied one such cryptic gene which allows E. coli to ferment lactose even when its beta-galactosidase gene is not working.
The gene is called ebg, and it needs at least two mutations to turn it on. The first is a change in the repressor, a DNA sequence which codes for a protein that normally keeps ebg inactive. The second is a change to ebg itself. The enzyme produced by the usual version of the gene cannot, in fact, break down lactose. It needs a mutation to make it effective. Under normal circumstances, each of these two mutations happens roughly once in every 100 million generations. Both mutations are needed, which would happen by chance roughly once every 10 million billion generations.
"That such events ever occur seems almost unbelievable," says Cairns, yet colonies do appear after about two weeks. That they do, without at the same time gathering a lot of neutral and outright harmful mutations, suggests to Cairns that bacteria must have aocess "to some reversible process of trial and error".
A Mechanism Needed
Cairns stresses that the main purpose of his paper is "to show how insecure is our belief in the spontaneity [randomness] of most mutations". But he realises that the experiments he reports are not going to settle the issue. What seems to be missing, at the moment, is a mechanism that would use what we already know about the workings of the cell to achieve the sorts of directed mutations that the group at Harvard has demonstrated.
Cairns suggests that the cell might make a set of vaiiable RNA messages—which carry the genetic instructions from the DNA to the machinery that makes proteins according to those instructions—and reverse-transcribe the most effective of these back into DNA.
It would need some way of monitoring "effective" RNA, but if it reverse- transcribed only those messages present when it staated to grow again, it would most likely capture the message that had indeed enabled growth to resume.
Critics of Darwinism have already leapt on Cairns’s work to support their belief that there is something rotten in the state of evolutionary biology. Biologists, however, are being more circumspect. They would like to see further demonstrations of the phenomenon (and preferably a mechanism too) in bacteria before they embrace it wholeheartedly. They also doubt that it could apply when the biochemical link between adaptation and gene is longer and more complex than that between an enzyme and its substrate.
Blog Back
Here you can give short responses and suggestions. Considered contributions, criticisms and discussion can be made privately via email[†]
