Tuesday, December 03, 2013

Lenski's long-term evolution experiment: the evolution of bacteria that can use citrate as a carbon source

Richard Lenski set up twelve flasks of E. coli B back in 1988. They were allowed to grow overnight in minimal medium containing 139μM glucose and 1,700 μM citrate. The glucose was the only carbon source that the parent strain could use and it limited growth of the cultures. The citrate was present as a standard chelating agent. The bacteria could not take up citrate and use it as an additional carbon source.

Every day the culture was diluted by transferring one ml to 99 ml of fresh medium (1/100). There were 6.64 generations per day or 2,423 generations per year (slightly more in a leap year).

These cultures were under strong selective pressure. Individual bacteria that could grow faster would out-compete other bacteria in the culture and take over. Lenski expected that each culture would show a variety of different solutions to the selective pressure with many common mutations and many that could be unusual.

A spontaneous cit+ mutation arose in the Ara-3 culture at about 31,000 generations and cit+ cells dominated the culture at around 33,000 generations. (Blount et al., 2008) (Cit- cells remained in the culture at low concentrations.) This is 14 years into the long-term evolution experiment. The Ara-3 culture became very cloudy since the concentration of cells increased several fold after 24 hours (when the cultures were diluted 1/100).

The cit+ phenotype depended on one or more prior mutations since it could only be reproduced starting from later ancestors. Lenski and his students interpreted this result in terms of historical contingency. Here's what they say in the introduction to Blount et al. (2008) ...
At its core, evolution involves a profound tension between random and deterministic processes. Natural selection works systematically to adapt populations to their prevailing environments. However, selection requires heritable variation generated by random mutation, and even beneficial mutations may be lost by random drift. Moreover, random and deterministic processes become intertwined over time such that future alternatives may be contingent on the prior history of an evolving population. For example, multiple beneficial mutations will arise in some unpredictable order, and those that are substituted first may differ from others in their pleiotropic effects and epistatic interactions, thus constraining some evolutionary paths while potentiating other outcomes. These accidents of history may even determine the survival or extinction of entire lineages, given the capricious and sudden nature of some environmental changes.

Stephen Jay Gould maintained that these historical contingencies make evolution largely unpredictable. Although each change on an evolutionary path has some causal relation to the circumstances in which it arose, outcomes must eventually depend on the details of long chains of antecedent states, small changes in which may have enormous long-term repercussions. Thus, Gould argued that contingency renders evolution fundamentally quirky and unpredictable, and he famously suggested that replaying the "tape of life" from some point in the distant past would yield a living world far different from the one we see today. Simon Conway Morris countered that natural selection constrains organisms to a relatively few highly adaptive options, so that "the evolutionary routes are many, but the destinations are limited". He and others point to numerous examples of convergent evolution as evidence that selection finds the same adaptations despite the vagaries of history. Evolution may thus be broadly repeatable, and multiple replays would reveal striking similarities in important features, with contingency mostly confined to minor details.

Of course, replaying life's tape on the planetary scale is impossible, but careful experiments can examine the role of contingency in evolution on a more modest scale. To address the repeatability of evolutionary trajectories and outcomes, the long-term evolution experiment (LTEE) with Escherichia coli was started in 1988 with the founding of 12 populations from the same clone.
The authors recognize that they are dealing with a different scale from what Gould imagined but, nevertheless, their result might be relevant "on a more modest scale."

So, what did they conclude about the evolution of cit+ strains in one (and only one) of their cultures? Here's the bottom lines of the abstract.
Thus, the evolution of this phenotype was contingent on the particular history of that population. More generally, we suggest that historical contingency is especially important when it facilitates the evolution of key innovations that are not easily evolved by gradual, cumulative selection.
Contrast that statement with that or Elizabeth Pennisi who said that ....
Lenski's humble E. coli have shown, among other things, ... that Gould was mistaken when he claimed that, given a second chance, evolution would likely take a completely different course.
I discussed her strange conclusion in another post [Elizabeth Pennisi writes about Richard Lenski's long-term evolution experiment].

Graduate student Zachary Blount was assigned the task of finding out how the Ara-3 strain was able to utilize citrate.1 In the beginning, they didn't know which mutations were responsible for the phenotype. All they knew was that several "potentiating" mutations were present in the ancestral Ara-3 cultures before 31,000 generations. These "potentiating" mutations made it possible for weak cit+ mutations to arise and increase the growth rate. They could reproduce these "actualization" mutations by starting with the "potentiated" strain. Finally, there were additional mutations that enhanced the ability to utilize citrate. These were "refinement" mutations.2

Back in 2008, this was really frustrating for those of us who wanted to know the molecular details. I'm sure this was frustrating for Blount and Lenski and their colleagues as well. We had no idea how difficult it would be to sort out the mutations. Now we know, and the photo of Zachary Blount (above) with all his Petri dishes drives home the point.

Here's what he discovered. It was published in Nature last year (Blount et al. 2012). The phylogenetic history of the Ara-3 culture is shown in the figure on the left. There were some early unsuccessful clades (UC) but most of the culture consisted of three main clades, C1, C2, and C3. (Remember that these are clones.) Clade 1 died out at about the same time that the first cit+ bacteria arose in clade 3. Clade 2 (cit-) persisted in the culture even after the evolution of efficient citrate utilizing bacteria. The authors do not know why the cit- bacteria continue to be present in the culture. Perhaps they contain an unidentified beneficial mutation.

The first cit+ bacteria were detected at 36,000 generations but the mutation probably occurred a few thousand generations earlier as shown in the figure. The surviving cit+ clone carries a mutation in the mutS gene (methyl directed mismatch repair) making this a mutator strain. These are fairly common in the other cultures and it's not thought to be significant in the evolution of the cit+ phenotype.


Recall that there are three steps in the evolution of the cit+ clade: potentiation, actualization, and refinement. The actualization step involved a tandem duplication of part of the genome that included the citT gene. This is the gene that encodes the citrate transporter (citrate/succinate antiporter), the protein responsible for taking up citrate from the external medium. Normal wild-type E. coli do not express this gene under aerobic conditions so they are unable to utilize citrate. The citT gene is part of an operon that's under the control of an upstream promoter to the left of the citG gene shown in the figure below.

The duplication results in the fusion of the 3′ end of the citG gene to the 5′ end of the rnk gene. This brings the new copy of the citT gene under the control of the rnk promoter resulting in constitutive expression of CitT (citrate transporter) and uptake of citrate from the medium. Thus, the mutant bacteria are able to use citrate as a carbon source under aerobic conditions.

The selective advantage of the initial tandem duplication was weak. The new cit+ phenotype conferred only a 1% growth advantage over the parental cit- strain. Previous results from the long-term evolution experiment suggest that this is not sufficient to take over the culture under the conditions of the experiment.


The actualization step was "refined" by additional mutations that increased the level of the citrate transporter resulting in a much bigger growth advantage. The refinement step occurred around 32,000 generations. It involved additional mutations resulting in multiple tandem copies of the new citT operon. Some clones had four copies of the citT operon and one had nine copies.


Blount moved the new operon into several other strains to see if it would confer a cit+ phenotype. In most cases expression of CitT did NOT result in significant use of citrate. However, when transferred into ancestors of clades 1, and 2 from the Ara-3 culture there was a weak ability to grown on citrate. The effect was even stronger in the ancestors of clade 3 from before the evolution of the Cit+ phenotype.

This indicates that earlier mutations in these clones enhanced the ability to utilize citrate. These are the "potentiators" and there had to be at least two of them; one in the ancestor of clades 1 and 2 and another in the ancestor of clade 3. These mutations, which are still unidentified, could be making chromosomal rearrangements easier leading to an increased rate of tandem duplications at the citT locus. Alternatively, they could be exerting an epistasis effect where the products of the potentiating genes interact with, or enhance, the citrate transporter.

In order to distinguish between these two possibilities, Blount et al. (2008) decided to run "replay" experiments where they took ancestral clones out of the freezer and subjected them to further rounds of evolution. The first replay experiment was started on the 3rd anniversary of Stephen Jay Gould's death (May 20, 2005) and ended on the 66th anniversary of his birth (Sept. 10, 2007) (~3,700 generations). Other replay experiments ran for shorter lengths of time. They also did plating experiments where they selected for colonies that could grow on citrate.

Blount et al. (2012) sequenced 19 cit+ clones from the replay experiments. Eight (8) of them had tandem duplications similar to the original clone but with different boundaries. Six (6) had IS3 insertions just upstream of citT creating a new promoter for expression of citT (see figure below).

Two (2) mutants had larger deletions/insertions at the same locus and two (2) others had mutations that brought citT under control of a different prompter. One mutant couldn't be resolved.

These result show that the cit+ phenotype can arise in a number of different ways providing evidence that the potentiating mutations exert an epistasis effect and not an effect on mutations.

The data shows that the evolution of the cit+ phenotype was a complex process involving a number of separate mutations. The final mutations (refinement) were contingent on earlier mutations (actualization) and the effectiveness of mutations at the citT locus was contingent on earlier potentiating mutations. These potentiating mutations occurred, by chance, in the Ara-3 culture and not in any of the other cultures.

Here's how Blount et al. (2012) summarize the results.
Before a new function can arise, it may be essential for a lineage to evolve a potentiating genetic background that allows the actualizing mutation to occur or the new function to be expressed. Finally, novel functions often emerge in rudimentary forms that must be refined to exploit the ecological opportunities. This three-step process—in which potentiation makes a trait possible, actualization makes the trait manifest, and refinement makes it effective—is probably typical of many new functions.

[Photo Credits: Science: The Man Who Bottled Evolution]

1. This is a long-standing tradition. When you've got a difficult problem and you don't know how to solve it, give it to a graduate student. It worked for me several times.

2. This may have led to confusion on the part of some people, including Elizabeth Pennisi. They may have thought that because replaying the final steps showed that the same mutations occurred again this meant that Gould was wrong. Lenski and his students clearly didn't think that way.

Blount, Z.D., Borland, C.Z. and Lenski, R.E. (2008) Historical contingency and the evolution of a key innovation in an experimental population of Escherichia coli. Proceedings of the National Academy of Sciences 105:7899-7906. [doi: 10.1073/pnas.0803151105]

Blount, Z.D., Barrick, J.E., Davidson, C.J. and Lenski, R.E. (2012) Genomic analysis of a key innovation in an experimental Escherichia coli population. Nature 489:513-518. [doi: 10.1038/nature11514]


  1. That's a really nice overview of the of an important scientific work. I hope that you do more of these, Larry. They are quite helpful.

  2. Simon Conway Morris makes the important point. Convergent conclusions in nature means there is limited options at the end. In fact it makes unlikely evolution as a haphazard mutation driven thing ABLE to create like looking or working traits.
    They must always invoke convergent evolution to explain such common design/conclusions.

    Interesting experiment.
    Yet it hints there is problems explaining results of changes. Mutations from mutations and then selection seems to be a messy trail. it seems to be very minor attrition perfectly fine with creationism(S).

    1. Simon Conway Morris was completely wrong on convergent evolution.

      The sole reason he presented that argument was the need to make humans an inevitable outcome of evolution. Otherwise theistic evolution becomes one of these:

      1) impossible
      2) equivalent to intelligent design
      3) incompatible with theology (there are scenarios that sort of work and are compatible with both evolution and theology but they invariably do away with at least one (often more than) of the omni-s in God's description.

      The problem is, of course, that you can only claim convergent evolution when you see it, and we do not see it everywhere. It is not the case that we see each and every adaptive solution arrived at independently by more than one lineage. Crucially, that includes humans - intelligence only arose once, as far as we can tell, therefore it is a complete fallacy to claim that convergent evolution makes sure human were inevitable.

      It's actually worse than that when you think about it. Imagine that there were in fact more than one species that evolved intelligence through convergent evolution (note that they need not coexist in the same time, in fact it is quite possible that 100 million years from now there will be no traces of our existence as most likely the period of time in which we will exist on this planet will only be a few tens of thousands of years, of which most have already passed). What are the theological implications of that? Did each of them sin separately? Ddid Jesus come to save each of them separately? Were both of them made in the image of God? And a lot more...

  3. Larry,

    Could you please comment on Carl Zimmer's elaboration on this relatively recent paper? Especially in regards to the following statement:

    "Barrick and his colleagues suspect that the evolution of a new kind of dctA gene allowed the bacteria to keep up a supply of succinate, which they needed on hand in order to feed on citrate. Together, the mutations to citT and dctA turned the mutant microbes into winners.

    Which leaves the role of all the other mutations shrouded mystery. In the new study, none of the mutations that came before generation 31,500 proved to be vital for being a full-blown citrate feeder. They didn’t lay the groundwork in any essential way. And yet the previous research clearly indicated that things were afoot before generation 31,500.

    If the statement in bold is true, then as far as I understand, this means that there actually were no potentiation mutations that are absolutely essential for acquiring the new function. Is my understanding correct, or am I missing something?

    Here's the full article at Zimmer's blog: Evolution Hidden in Plain Sight