Festooning The Tree Of Life

By Carl Zimmer | July 17, 2008 2:42 pm

Darwin treeCan we make a picture of history? Charles Darwin recognized that all the different species on Earth share a common ancestry, having branched apart over millions of years. He pictured the history of life as a tree, as he sketched here in one of his notebooks. He later published a more formal illustration of the tree of life–or at least the branching of evolution–in the Origin of Species. Later scientists put actual names to the branches. They proposed that birds and crocodiles share a close common ancestry, for example, based on traits they share in common in their skeleton and skin. Later studies–including comparisons of DNA–supported many of those relationships. Today scientists are comparing entire genomes in order to illuminate some of the murkier nooks in the tree.

At the same time, however, a number of scientists have argued that a tree is only a limited metaphor at best for the history of life. It works best with species–like ourselves–that pass down their DNA from parents to children. This transfer–called vertical gene transfer–means that the flow of genes over time tracks the flow of generations. How could it be otherwise, you might wonder?

Well, imagine you’re E. coli.

E. coli, a resident of your gut, is also the workhorse of biotechnology because scientists have figured out how to plug genes into it with relative ease. Even a kid can do it now. So if you want to sell insulin to diabetics, you don’t need a warehouse full of pig pancreases. You just need to insert the human insulin gene into E. coli, feed it some sugar, and wait for it to do your chemistry for you.

As I explain in my book Microcosm: E. coli and the New Science of Life, scientists figured out how to carry out genetic engineering on E. coli by mimicking nature. They knew that it traded rings of DNA called plasmids. Bacteria can also trade genes using viruses as their proxy. Some species can even just slurp up naked DNA. Foreign genes don’t pose much risk to bacteria, it turns out. Last year a team of scientists injected 246,045 genes into E. coli and then waited to see if the engineered bacteria could continue to grow. Only 1402 genes (about 1 in 200) were toxic.

This process is known as lateral gene transfer, and it turns out to be rampant on Earth. The genome of E. coli and other bacteria are loaded with foreign DNA their ancestors picked up over millions of years. The chance that any one microbe picks up a foreign gene is tiny, and the chance that natural selection favors it over other microbes is also small. But over geological time, lateral gene transfer can have huge effects. It also raises the possibility that we must upgrade from the Tree of Life to some more powerful metaphor.

Animals still have a tree-like pattern of evolution, because we rarely pass genes around. (But perhaps we do it more than we realize! See this new paper on an antifreeze protein that cold-water fish may have traded. It’s possible that when male fish flood the ocean with their sperm, some of the DNA leaks out and is taken up by the eggs of other species.) But lateral gene transfer matters a lot to life as a whole. Bacteria and other single-celled microbes make up much more of life’s genetic diversity, and they were around for three billion years before animals showed up for the party. So much of the history of life may not fit the tree metaphor very well any more. No longer can we assume that the genes in a species all have the same history. Some of them may have leaped from species to species.

So how should we picture the history of life then? The newest assault on this tough question just came out in the Proceedings of the National Academies of Sciences. Tal Dagan, a biologist at the University of Dusseldorf, and her colleagues have festooned the tree of life with lateral gene transfer. They analyzed 539,723 genes from the completely sequenced genomes of 181 species of microbes. To begin making their new picture of evoluiton, they drew a tree showing how those 181 species are related. They used a gene that doesn’t seem to have been traded around much, and which therefore reflects the common descent of the microbes.

Tree only

Next, the scientists sorted the genes in these species into networks, connecting similar genes to one another. In many cases, these networks matched up nicely with the branches of the tree. In other words, they were passed down from ancestors to descendants. But in other cases, some of the genes didn’t line up well with this tree. These were genes that showed signs of having moved between the branches.

Dagan and her colleagues then added three sets of new lines to the tree. The first set, colored green below, are links between distantly related groups of microbes that share a gene. These links represent ancient lateral gene transfers billions of years ago.

Int2IntNext, the scientists identified genes that are found in a tuft of branches, as well as an unrelated single living species. These transfers probably happened more recently, after the ancestors of the 181 species branched off from each other.

Ext2IntAnd finally, the scientists added links to show pairs of species that shared genes found in none of the others. These are probably the most recent leaps, marked below in red.

Ext2ExtPut them altogether, and here’s what you get. (I’ve also uploaded the full-resolution picture here: All edges)All edge

Analyzing this tree bush mangrove thicket Gordian knot, Dagan and her colleagues found a fascinating interplay between vertical and lateral gene transfer. If you look at any one of the 181 genomes, 81% on average of its genes experienced lateral gene transfer at some point in its history. So clearly lateral gene transfer is rampant. But once genes made the jump, they tended not to make another one–in fact, Dagan and her colleagues conclude that most became trapped in vertical descent.

This new picture is a far cry from Darwin’s sketch, and thank goodness for that. A science that doesn’t move forward for 150 years isn’t much of a science at all. But we may need some new metaphors to help us catch up with it.

[Tree images courtesy of Tal Dagan. Darwin tree from Wikipedia.]


Comments (24)

  1. EastwoodDC

    It looks a bit like a Sno Cone to me (http://www.abfnj.com/extras.htm)
    “The Sno Cone of Life” just doesn’t have that nice ring to it though, and I doubt it will catch on. 😉

    Now a serious question: Are there any similar estimates for viral transfer of DNA with eukaryotes? In your book you mention this is a rare sort of gene transfer. How rare is it?

  2. Doug

    Hi Carl,

    Horizontal gene transfer is advocated by Carl Woese. Wikipedia uses a diagram of a “Phylogenetic tree showing high rates of horizontal gene transfer between organisms.”

  3. Assuming genes aren’t being traded backwards in time, it’s not a bush, or a thicket, or a Gordian knot. It’s a directed acyclic graph, or DAG. Of course “The Great DAG of Life” doesn’t quite have the same ring to it.

  4. Greg

    You heard it here first. Microbial evolution is a “cloud” of life rather than a “tree” of life. Heh.

  5. Sven DiMilo

    Please help me out with this one:

    Next, the scientists identified genes that are found in a tuft of branches, as well as to a single living species.

    From context, I’m guessing it should read “…as well as in a single living species on a distant branch,” or something like that. If not, please explain.

    [Carl: You inferred rightly. I’ve tweaked the post. I hope that’s clearer now.]

  6. The concept of lateral gene transfer reminds me of David Quammen’s recent article in Harper’s about contagious cancer (http://www.harpers.org/archive/2008/04/0081988). For instance, about Canine Transmissible Venereal Tumor in domestic dogs, Quammen writes:

    “Tumor cells in Dog B, Dog C, Dog D, and Dog Z are more closely related to each other than those cells are to the dogs they respectively inhabit.”

  7. amphiox

    What could possibly trap a gene in vertical descent after it has made one jump? Is it something to do with the mechanisms by which genes make the jump? Or is it a statistical thing, that is, the probability of a gene jumping and surviving (both being retained by the recipient, and the recipient lineage surviving to the present day) low enough that on average each gene observed to jump has only jumped once.

    Are there any genes at all that have been confirmed to have jumped more than once?

  8. Ah yes ;
    the web of life on this planet :
    It’s still a ” primal soup “of all kinds of genes ?

  9. amphiox – there is a theory called the Core Genome Hypothesis, also known as the “housekeeping genes” hypothesis, that a set of microbial genes are crucial to the very functioning of the microbe and hence are hard to swap and hard to lose. Although there are exceptions, this seems robust enough to explain why some microbial species remain coherent.

  10. Torbjörn Larsson, OM

    this tree bush mangrove thicket Gordian knot

    Yeess… but if the most recognizable feature (that of tending to a single lateral transfer) is of the process, not of its outcome, it would IMHO be best if the new metaphor describes that instead.

    Janne is then closer with his DAG description, but it is general as a metaphor.

    As a process this reminds me of parachute canopy formation (formation involving a state change of the parachute, somewhat analogous to the lateral transfer). (And as a picture it reminds me of tree canopies mixing, keeping some connection to the earlier logic, as well as the earlier metaphor specifically.)

    So the “canopy [formation] of life”, perhaps?

  11. “Assuming genes aren’t being traded backwards in time, … It’s a directed acyclic graph, or DAG. ”

    Note that this change of metaphor also requires changing the mapping from metaphor to reality. In Darwin’s tree, each node (or equivalently, each node’s parent branch) represents a species or clade. In the DAG, if each node represented a species, the graph would be cyclic, since species A can transfer genes to species B and vice versa. In order to have a DAG as you suggest, each node must represent one end of a gene-transfer event, i.e. we must have a new node (at least on the recipient side?) wherever a gene transfer occurs. It doesn’t appear that Dagan et al have constructed their drawing that way… the y axis only represents descent, and time by implication.

    An interesting prediction of this model would be that a DAG *exists*, that can account for all the evident gene transfers, i.e. without going backward in time. Whether that’s a trivial condition to meet or not, I don’t know.

  12. On a related note…

    One of the pieces of evidence for evolution, claimed as a successful prediction of the “new synthesis”, has been the tree of life, in the sense that phylogenies derived from both genetic and morphological data are supposed to match each other.

    In the case of a mangrove/DAG/Gordian knot of horizontal gene transfer, is there still a prediction that can be made, or does it become an unfalsifiable mess?

  13. Torbjörn Larsson, OM

    Life intervened, but finally returning to old threads FWIW:

    is there still a prediction that can be made, or does it become an unfalsifiable mess?

    I think that is debated, but molecular phylogenies are perhaps testable.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

The Loom

A blog about life, past and future. Written by DISCOVER contributing editor and columnist Carl Zimmer.

About Carl Zimmer

Carl Zimmer writes about science regularly for The New York Times and magazines such as DISCOVER, which also hosts his blog, The LoomHe is the author of 12 books, the most recent of which is Science Ink: Tattoos of the Science Obsessed.


See More

Collapse bottom bar