My recent post about a striking new experiment in evolution (E. coli evolving the ability to eat a new kind of food) is still drawing lots of commenters and links. Very cool! Not so cool are the claims that this experiment is evidence of creationism, made by people who have not actually read the paper itself. Unfortunately, the paper is behind a subscription wall at the journal. Fortunately, the scientists have posted it on their own web site (pdf link). So go, read, and digest.
I'm also hoping that Zachary Blount, the grad student who pored over the trillions of E. coli in this experiment, will have time to respond to the many comments in the next few days. Stay tuned.
This is a fantastic paper, and I'm glad it's getting the attention it deserves. I do have one question for Zachary Blount (if you're still around)...
In the paper, you say you'll be identifying the mutations in question by whole-genome sequencing. Is there any particular reason you're not trying to map the mutations by genetic crosses? At least for the second mutation (from Cit- to Cit+), you've got a positive selection, and crossing those two strains against each other should (if I'm thinking about this right) let you narrow down the relevant region of the chromosome pretty quickly. (I'm assuming that your strains are still P1 phage-sensitive, here.) Heck, putting a genomic library of the Cit+ strain into the Cit- ancestor ought to work, too.
The other two mutations are trickier to map without that strong positive selection, but it seems like old-fashioned mapping might be cheaper and easier to interpret than whole genome comparisons, without quite so many background mutations around to muddy the waters.
Regardless, terrifically cool work. I love this stuff.
I wrote a blogpost about Michael Behe's reaction to this. Let me know what you guys think:
Thank you for your compliment. It is good to here after so much blood, sweat, toil, and tears went into the four or so years of work the paper describes.
As to your question, we decided on whole genome sequencing because it is fairly cheap and time-saving at this point while giving us a lot of information that genetic crosses and transduction experiments would not. I actually did some preliminary work with P1 to try to transduce (for the lay person: this is the use of a virus, P1 in this case to move a selectable mutation from one genome to the next - a very valuable tool in molecular biology. Wikipedia has an article on it at: http://en.wikipedia.org/wiki/Transduction_%28genetics%29) the immediate underlying Cit+ mutation to a Cit- background. I failed to get any Cit+ transductants, however. This suggests the possibility of a more complex underlying genetic basis than a point mutation (which wouldn't be surprising given that the third contingency experiment's results imply that two mutations are required to become Cit+ even with the potentiating mutation). I meant to follow up on this work fairly quickly at the time (late 2006), but I got involved in the large scale plating experiments that are described in the paper as contingency experiment #3. As you might imagine from the description in the paper, experiment #3 basically consumed my life for a while, and I worked on it to the exclusion of other aspects of the project for a number of months (January through June of 2007). By the time I was finished, MSU had gotten in the equipment to do 454 pyrosequencing (for lay people this is a technology that allows rapid genome sequencing for comparatively low cost), and it was just easier to go the genome sequencing route. We hope to get the genetics figured out in the next several months and published sometime next year.
As time goes on, though, will likely do experiments that may involve using P1 to move what mutations we find to ancestral and Cit- backgrounds to assess and verify their effects, so we won't totally ignore that very useful phage!
For those interested in the use of phage P1 in work that moved mutations from one genome to another, a good example is one by Lenski and Elena (Elena, S., and Lenski, R. 1997. Test of synergistic interactions among deleterious mutations in bacteria. Nature, 390: 395 398.)that is available in pdf format on Dr. Lenski's webpage (https://www.msu.edu/user/lenski/) as publication #84 in the complete list. For those interested about the long term evolution experiment in general and the findings that have out of it, I would also recommend publication #147 as the best summary.
Cool. I love that whole-genome sequencing has become the "easier" route to anything!
I don't envy you the task of dissecting out which differences cause your phenotype, but you certainly don't flinch at hard work. "Experiment #3" was a beast.
Good luck, and I look forward to reading the next paper.
P.S. On further thought, maybe I wasn't clear in my original question, actually. I meant to ask if you had thought about trying to map the _second_ mutation (which made the strain originally Cit+) by transducing it into the strain that already carries the _first_, potentiating mutation.
Thanks for taking the time to share this stuff with the world. It's fun.
Yes, the march of technology has been very nice in terms of bringing down the time and cost of genomics. This is a blessing and a curse. Genome sequencing allows much more rapid answering of many questions, including those we could not answer before. However, it also means that we have to fight two other problems: 1. genome sequencing yields so much data, it is easy to drown in them, and 2. with easy access to so much data, there is always an impulse that must be fought to proceed without taking time to formulate crisp, clear experimental questions and hypotheses. This second issue is dangerous because, for science to work, clear questions, hypotheses, and testable predictions are an absolute must. Without them research is blind and typically ineffective. Of course, the cost of sequencing is still significant (especially to someone whose basis of understanding cost comes from living on a graduate student's stipend), so that keeps us in the lab focused and thinking clearly.
As to the transduction clarification, the recipient strains I used were potentiated Cit- clones from Ara-3. This was necessary given that, without knowing the Cit+-causing mutation, I could not exclude the possibility that the potentiating mutation was required for the manifestation of the Cit+ phenotype once the final mutation had occurred. This epistatic hypothesis is opposed by the alternate hypothesis that potentiation involved setting up the final mutation itself. Without knowing the final mutation and determining its phenotypic effect on its own, I cannot and could not exclude either hypothesis. That is another nice thing about the genome sequencing: it doesn't matter which of those two hypotheses are correct for use to eventually identify the mutation. However, once we do identify the mutation, we can then use transduction of the final mutation and other traditional genetic approaches to differentiate between those two hypotheses, as they go to a feature of the Cit+ story that is interesting and needs resolution. Again, stay tuned for the eventual genetics paper!
I would like to thank Carl, and Zachary, and all involved in this. This is truly exciting stuff to read about. I cannot say enough! Thanks!!!
Carl Zimmer said in the original post,
Not so cool are the claims that this experiment is evidence of creationism, made by people who have not actually read the paper itself.
Why single out the creationists? A lot of Darwinists have also been jumping to conclusions about the meaning of the study.
IMO there is nothing wrong with commenting about something that one hasn't read if that something is long, complicated, and/or of limited (until now) accessibility -- e.g., this paper. But I think it is wrong to jump to conclusions. IMO more thought and discussion are needed to sort things out here.
I am especially concerned about mutations being lost in the experiment because (1) only one percent of old populations were used to start new populations and (2) there were very few generations per population.
Here is more evidence of the level of scholarship that those who would argue against serious research are willing to engage:
"IMO there is nothing wrong with commenting about something that one hasn't read if that something is long, complicated, and/or of limited (until now) accessibility -- e.g., this paper."
Again, this reads like the complaints of an undergrad--it's "long" and "complicated" so they won't read it, but they certainly are willing to comment about it. No one said that serious scholarship was going to be easy and commenting about articles you have not read is to take the arrogance of ignorance to a new level.
Actually, Tyrone, I think that sounds like something a lazy high school student would say. Just knowing that Larry is worried about "lost" mutations because "(1) only one percent of old populations were used to start new populations and (2) there were very few generations per population," makes me believe he has never taken a genetics course in his life let alone ever done any research. Poor guy should go back to school.
Just knowing that Larry is worried about "lost" mutations because "(1) only one percent of old populations were used to start new populations and (2) there were very few generations per population," makes me believe he has never taken a genetics course in his life let alone ever done any research. Poor guy should go back to school.
You stupid fathead, even Zachary Blount, a co-author of the paper, expressed concern about "lost" mutations. He wrote in a comment in another thread,
. . . one cannot ignore the possibility of the loss of even weakly beneficial mutations during transfer (especially of a new mutation that had occurred during the last generation of the previous day) . . .
tyrone slothrop said,
commenting about articles you have not read is to take the arrogance of ignorance to a new level.
Presumably most people who have commented on this paper have not read it.
The mutations lost question had me going for several seconds when i first read the blog post. It's a real question.
I assume that the answer is that the vast majority of each new generation are the ones that out compete the others. After all, those are the ones who reproduce faster.
Clearly, this is a selection for fast breeders. That isn't what you'd always get in the wild. This is a new environment. Does the artificialness of the new environment mean much?
I assume that the answer is no. Any new environment is one that evolution should be able to adapt to.
My son wants a pet. Mom is allergic to dogs and cats. I want it to be something easy to take care of, and if it dies, no one will be too sad. E. Coli is just the thing.
But there will be trillions of them. And he'll want to name them individually.
When at low frequency (such as when they first occur) or in a population sufficiently small, even beneficial mutations may be lost at random due to genetic drift. (Wikipedia has a good entry on drift if you are interested in more details on it, and its role in experimental evolution experiments is dealt with in publication #145 on Dr. Lenski's webpage. You might also want to consult any decent college evolution or genetics text for more.) Drift, the change in allele frequencies due to chance (one major component of this in the LTEE is the random manner in which a few million cells of each population are transferred each day to new medium), is not a confounding factor in the long term evolution experiment, as it is a part of the core evolutionary process, and has been recognized as such since Sewell Wright first wrote about it in the 1920's. As such, loss of mutations, even beneficial ones, has been as much a part of the evolution of the twelve long term E. coli populations as has natural selection. Each population thus has an evolutionary history within the experiment in which mutations of a variety of effects arose at random, and were then either lost at random by drift or non-randomly by natural selection against detrimental mutations, or even non-randomly rose to high frequencies if beneficial and at a selective advantage, provided they became frequent enough to escape drift. Add to this that, because the bacteria are asexual, there is a phenomenon called "clonal interference", and it involves instances in which two different beneficial mutations arise in the population, but in different clones. Because there is no sex, there is no way for the two mutations to come to be in the same chromosome by recombination. As a consequence, competition between the two mutations can result in the loss of the weaker one. Again, all of these factors, the occurrence of mutations, and their subsequent fates due to drift, natural selection, and clonal interference are a part of evolution.
It just happened that Ara-3's history saw the occurrence, rise, and persistence of the potentiating mutation, whatever it might be. Could the potentiating mutation have occurred in the other eleven populations? It is possible. Could it have arisen and been lost in those populations? It is possible. Could it be present in any of those populations now? It is possible. For that matter, what was its fitness effect? Aside from it being clear that the potentiating mutation was not detrimental (if it were detrimental it would not have persisted in the population for the 11,000+ generations that it did), we won't know that until we can identify it and construct strains of E. coli that vary only in that one mutation and compete them. Indeed, once we identify it, we will then be able to answer those above questions. Unfortunately we can't right now.
I hope that answers you question about lost mutations. I will see about including a bit about drift in the upcoming post (which should be better than these "on the fly" comments).
As to your question about what has been selected for, I recommend that you read publication #147, as it goes in that in detail, and there are a number of other papers on Dr. Lenski's webpage that go into aspects of that question. I will say in regard to the environment, that the experiment is really a matter of us providing the bacteria with an environment. The environment then selects among the variation that arises by random mutation in the populations, with the added complications I discussed above. This is just like what happens in nature. Publication #145 is a fairly easy to read that reviews experimental evolution using microbial model systems, and I think you would find much there of interest.
Thank you, and I hope that helps!
I almost forgot about the pet thing: 1. keep in mind that you and your son, and your wife all have E. coli already in you, and thus you can be said to already be keeping them as pets. 2. Be forewarned that E. coli cultures are not the most pleasant-smelling of things. 3. They aren't that snuggly, and they don't come when called, though they don't have to be housebroken. Still, it is an interesting idea!
[E. coli] don't come when called,
I suppose you could put a streak of glucose in the direction that you want them to go...
They might not come themselves, but they'll send their kids.
Larry, did you really read the comment you linked to? It just goes to show you really do need to take a course on population genetics. I never said mutations couldn't be lost during a transfer or that such instances of drift were unimportant. You simply seem to be dwelling on something that really doesn't have much relevance to the conclusions of this paper. Take Zachary's advice and read the papers he referenced regarding drift.
Carl wrote: So go, read, and digest.
GROAN. Carl. And the rest of you being so interested in your discussions as to ignore this awful pun, for shame :P
"2. Be forewarned that E. coli cultures are not the most pleasant-smelling of things."
Depends on their media. Aerated, glucose-minimal media cultures don't smell unpleasant, even in late-log phase. I wonder how fast-growers smell on citrate media.
>>"2. Be forewarned that E. coli cultures are not the most pleasant-smelling of things."
Depends on their media. Aerated, glucose-minimal media cultures don't smell unpleasant, even in late-log phase. I wonder how fast-growers smell on citrate media.<<
That's a good point. I was thinking of the smell of rich medium cultures (LB in specific). I thought an E. coli pet-owner might like to see more turbidity.
The Cit+ cultures growing in DM25 smell like Cit- cultures. If there is a difference, I can't tell. I don't know if this is just me, but the thiamine in DM25 gives the medium and the cultures an odor that has always reminded me of popcorn...
Smell is one way of noticing what is happening in cultures (from build-up of excreted metabolites), but it's not something that gets described often in papers. It becomes part of the undocumented oral tradition for the field instead. Sometimes you can tell when cultures switch over their metabolism to alternate sources or if cultures get contaminated just by sniffing.
S. cerevisiae smell great (well, they're bread/brewing yeast) in early, aerobic growth but let an overgrown culture sit around a few days and... whew.
Not that you'd necessarily want to try this sort of sniff-the-culture thing with V. cholerae...