The Tangled Tree

by David Quammen

  Gogarten’s relations with Ford Doolittle, as the two men converged on the subject of horizontal gene transfer, were more sunny. Their interaction began when Doolittle invited Gogarten to Halifax in 1994 for private discussions and to give a seminar. Gogarten talked about gene transfer from archaea to bacteria, among other things, and the idea that a “net of life” might better represent evolutionary history than a tree of life. Two years later, they both attended a big microbiology meeting held at the University of Warwick, in England. Doolittle’s talk to that meeting stands less vividly in his own memory than the one by Gogarten, who spoke again about horizontal gene transfer: genes moving sideways across the great microbial divide between bacteria and archaea. There seemed to be so much of such transference, newly discovered, Gogarten said, repeating his Halifax point to a wider audience, that the phylogeny of species didn’t look like a tree anymore. Not during the early phase of life on Earth, anyway. It looked more like a net. Evolution was “reticulate” as well as branching. Ford Doolittle was listening closely and inclined to agree.

  Back in Halifax, Jim Brown coaxed Doolittle in the same direction, and the flow of new sequencing data tended further to tangle the tree. In 1997 Doolittle and Brown did a big tree-building effort together, looking at sixty-six different proteins that are essential to all forms of life, and at the different variants of those proteins as reflected in more than 1,200 different gene sequences, from a wide variety of bacteria, archaea, and eukaryotes. Most of these sequences were publicly available; Brown and Doolittle downloaded them from databases and subjected them to comparative analysis. They constructed an individual tree for each of the sixty-six proteins, showing how it had evolved into distinct variants within different lineages of creatures. Each protein had a name, of the sort that you or I could scarcely pronounce, let alone remember: tryptophanyl-tRNA synthetase, for instance. One version of that exists in humans, another in cows, another in the bacterium Haemophilus influenzae—each distinct but fundamentally the same protein. Why is the stuff so universal? Because it’s a very basic tool, necessary to all forms of life for the role it plays in linking amino acids with code triplets during the translation process. The other sixty-five proteins chosen by Brown and Doolittle included some involved in DNA repair, some involved in respiration, some devoted to metabolism, structural proteins for ribosomes, and more. Brown and Doolittle compared the variants, constructing an independent tree of descent for each. They printed all sixty-six trees in their published paper, so that the paragraphs of text seemed to exist within a forest—or at least a very green suburb. This exercise yielded a telling point: the trees didn’t match.

  There was a lot of disagreement. Many of them were incongruent with one another—sprouting different branches in different places—and incongruent also with the supposedly canonical 16S rRNA tree of Carl Woese. The logical conclusion was that genes have their individual lineages of descent, not necessarily matching the lineage of the organism in which they are presently found. It was the same thing Robert Feldman would soon tell the reporter Elizabeth Pennisi: “Each gene has its own history.” How was that possible? Horizontal gene transfer. While the creatures replicated themselves vertically—humans producing humans, yeast producing yeast, Haemophilus influenzae producing more Haemophilus influenzae—sometimes the genes went sideways. They had their own selfish interests and opportunities.

  63

  By 1998, Ford Doolittle had become one of the go-to experts for science journalists seeking comment on new developments in this field. He had a distinguished research record, he knew all the issues and most of the players, he would answer his phone, and he could turn a phrase. Reporters from the journal Science particularly favored him, including the one who covered Craig Venter’s first whole-genome sequencing, of a bacterium, and the one who reported on the archaean sequencing a year later. Elizabeth Pennisi too, in her piece headlined “Genome Data Shake Tree of Life,” quoted Doolittle several times. Then in 1999 he heard from the editors of Science with a slightly different request.

  They were preparing a special issue on the subject of evolution. It would offer a package of articles, each one a wide-angle review of some realm of biology, seen from an evolutionary perspective, and the authors would include eminent figures such as Stephen Jay Gould and David Jablonski, as well as lesser-known scientists. Their reviews would cover a range of topics and scales, from nucleic acids structure to dinosaurs. The editors asked Doolittle if he would kindly recommend someone to write on evolution and microbiology. He replied: How about me?

  They agreed, not knowing quite what would come. Doolittle himself thinks that the Science editors were caught by surprise, and that they said yes because saying no would have seemed rude. Grudgingly, they accepted what he calls “my self-promotion.”

  “I wanted to write about this topic,” he told me years later—“this topic” meaning horizontal gene transfer and the tree of life. He felt impelled to produce a sort of manifesto. “And they weren’t so keen on that. But they didn’t really . . . I think they didn’t see how they could back down.” If he was authority enough to comment on articles and recommend authors, why not authority enough to write this review himself, and to choose those aspects of microbiology that were currently most interesting and important?

  So the issue of Science for June 25, 1999, contained a review article by Doolittle entitled “Phylogenetic Classification and the Universal Tree.” It became the most provocative paper he ever published. It put horizontal gene transfer in the center of a new discussion. And part of what made it arresting was not just Doolittle’s words but also his drawings. He later told me that, while he was glad the editors of Science published his self-invited paper, he was pleasantly surprised that they accepted also those hand-sketched trees. “Normally, journals wouldn’t do that. It must have been a flat day at the office, or something.”

  He began his text with a long view. “The impulse to classify organisms is ancient,” he wrote, “as is the desire to have classification reflect the ‘natural order.’ ” Histories of biology told that story, going back to Aristotle and forward to Linnaeus. But putting “natural order” in quotes was Doolittle’s first hint of how ambiguous such an order could be. His purpose in this paper was to follow those ambiguities to their logical extreme.

  He described the rise of evolutionary phylogenetics and reprinted the branching figure from Darwin’s Origin, noting that this (and Darwin’s passage about the simile of “a great tree”) was what brought tree imagery into evolutionary thinking. He noted the big change in modern phylogenetics: the change from morphology to molecular evidence, yielding a whole new dimension of discoveries. He mentioned endosymbiosis theory and its two key tenets—about mitochondria and chloroplasts entering the eukaryotic lineage as captured bacteria—and that those tenets had been confirmed by molecular data. He highlighted the role of Woese, who seized on ribosomal RNA as the sole basis for a three-domain tree of life. Doolittle even drew a picture: a cartoon of the Woesean tree, with thick branches rising upward in three main clusters, labeled “Bacteria,” “Eukarya,” and “Archaea.” Each branch was topped as an arrow, aimed vertically into the future. In addition to those vertical arrows, two diagonal ones jabbed sideways: from the early bacteria into the early eukaryotes, representing those two momentous instances of endosymbiosis, the origin of chloroplasts and the origin of mitochondria. He called it the “current consensus” model rather than simply “Woese’s.” Then Doolittle asked: How true is it?

  His answer: probably not true enough. And the problem was horizontal gene transfer. Microbiologists had long been aware of HGT, going back through Joshua Lederberg to Oswald Avery and others; but for phylogeneticists, drawing their pictures and diagrams of vertical descent, it presented greater difficulty. Doolittle was largely addressing the latter group, scientists concerned with tracing phylogeny. If the new evidence proved correct, he wrote, and HGT was not a rarity but a rampant phenomenon, at least among bacteria, archaea, and
the early eukaryotes, then Woese’s tree—now the consensus model—was badly wrong and incomplete.

  He mentioned some of that new evidence, citing Peter Gogarten, James Lake, Sorin Sonea, and others. For instance, two researchers looking at the “molecular archeology” of E. coli, the most-studied bug in biology, had just reported an unexpected finding: that its genome contains at least 755 genes acquired by horizontal transfer, accounting for 18 percent of its chromosomal DNA. Those transfers had occurred not during early evolution, furthermore, but more recently, giving E. coli adaptations it wouldn’t otherwise possess. William F. Martin, the brilliant and blunt American based at Heinrich Heine University in Düsseldorf, had noted that there was “something quite ominous” about these E. coli results. If so many “relatively recent” transfers had gone into one bacterium, Martin wondered, then how many horizontal transfers had occurred among the entire bacterial domain throughout the depths of geological time? The rough answer was “countless.” Martin made his comments in a paper titled “Mosaic Bacterial Chromosomes: A Challenge En Route to a Tree of Genomes,” published not long before Doolittle’s review. All this sideways transfer raised a challenge, Martin warned, to delineating a genomic tree of life. Doolittle, following similar logic and embracing the challenge, asked: What might the tree look like? Again he drew a sketch, and, to his surprise, the editors of Science printed it.

  He called this one “a reticulated tree.” It was a tangle of rising and crossing and diverging and converging limbs. It had its antecedent (as Doolittle acknowledged) in a somewhat similar figure offered by Martin in his “Mosaic” paper. Martin’s tree resembled a sea fan, one of those delicate structures built by a colony of coral-like animals on the ocean floor. Its long limbs and branches rose wavily, diverging from a simple base. It was done in color, pastel shades, and some of the slim branches also converged: turquoise and lavender joining to form purple. Doolittle’s sketch, by contrast, was in black and white and bulkier, as drawn freehand with a blunt pen; more robustly entangled from the bottom up, it resembled a mangrove thicket, if mangrove limbs could inosculate. It was so intricate and yet so fluid, so quirky, that it seemed almost funny. It looked like something John Krubsack might have grown, on a bet, from a cluster of box elder saplings in a field near Embarrass, Wisconsin.

  Doolittle’s reticulated tree, as drawn by Doolittle, 1999.

  And yet Doolittle’s second figure was still much too simple: another cartoon. It rose by multiple trunks from multiple roots, then split into multiple limbs, but not so very many, and the roots weren’t identified. It conveyed paradox, but it lacked detail. It was suggestive, not literal. It was eloquently weird. Maybe, Doolittle said in his text, as well as with this drawing, the history of life just can’t be shown as an ordinary tree.

  64

  The 1999 paper was a landmark in several ways, one being its influence in getting horizontal gene transfer taken seriously. “That had a huge effect,” William F. Martin said later. “It broke the dam.” Suddenly HGT seemed a mainstream idea, an ongoing process of major importance at least in microbial evolution—something that had to be considered and discussed—rather than a hallucination, an artifact, or a quirk.

  Another way of seeing Doolittle’s 1999 paper, more personal, was that it marked the end of the friendship between him and Carl Woese. That friendship had already become strained several years earlier over a single word: prokaryote. Doolittle persisted in using it, to label the category in which Bacteria and Archaea could be grouped together, and Woese hated it, because his greatest discovery implied that those two never should be grouped together. So when Doolittle wrote about prokaryotes versus eukaryotes, in the old sense that Roger Stanier had used those terms, to Woese it seemed like an insult and a taunt. Then came this review paper of 1999, with Doolittle in his speculative tone posing an even more direct challenge to the central premises of Woese’s career. Was it true that 16S rRNA (and its equivalent in eukaryotes, 18S) is a uniquely stable molecule, too fundamental in its role within each cell to be subject to horizontal transfer? And was it true, therefore, that those rRNA molecules constitute evidence for a uniquely definitive tree of life? Woese said: Yes. Doolittle said: Hmm, possibly not, and, in fact, maybe there is no definitive tree of life.

  The historian Jan Sapp, so close to Woese in the later years, told me that Woese felt betrayed, harboring darkly imagined theories about why Doolittle had turned against him. Woese had known Doolittle when Ford was a young postdoc in Urbana, after all, and they had drunk beer together; Woese had sent Linda Bonen, with her crucial skills, to Doolittle in Halifax, making possible some of his best early work; the two men had shared many curiosities and ideas. This amicable history seems to have brought a bit of “Et tu, Brute?” into Woese’s response. Science sometimes gets petty and emotive, as I’ve noted, and maybe especially so if you perceive yourself an isolated, neglected genius at a state university in downstate Illinois. Then again, if there was a “Woese’s Army,” in that phrase Lynn Margulis favored (and Woese detested), Doolittle now counted himself AWOL. But the reason for his departure from those ranks, notwithstanding Woese’s dark theories about jealousy and betrayal, was simple: new data. New genomic evidence of horizontal gene transfer blurred the picture of Woese’s three domains.

  The rift between the two men was further exacerbated less than a year later, when Doolittle published a popularized version of his recent thinking in Scientific American. That magazine carried less authority than the journal Science and served a different purpose—not announcing discoveries but explaining them for a mostly lay readership. Doolittle’s draft article was heavily edited and rewritten by Sci Am editors, so that it contained his ideas (and ideas from some of his colleagues, such as William F. Martin) but not his voice. And not his drawing hand: in place of Doolittle’s pen-sketched cartoons of tangled limbs and roots, Scientific American offered smoothly professional (and more sterile) airbrushed figures. As for the title, that was iconoclastic even by Doolittle standards: “Uprooting the Tree of Life.”

  Horizontal gene transfer was “rampant,” he told the readers, and it had affected the course of evolution “profoundly.” Scientists already knew that bacterial genes sometimes move sideways, yes, carrying “the gift of antibiotic resistance” or other special adaptive traits. Bacterial geneticists such as Martin and his mentors were well familiar with that phenomenon. But most researchers concerned with evolutionary history and phylogeny, such as Woese and his followers, for whom HGT seemed rather startling, had assumed that a stable core of other genes—genes essential to cell survival, basic genes for metabolism and replication—remain firmly embedded within their original lineages and are inherited vertically, rarely if ever traded horizontally. “Apparently,” Doolittle wrote, “we were mistaken.”

  “By swapping genes freely,” he wrote, early cells “had shared various of their talents with their contemporaries.” Eventually this ragout of changeable cells and interchangeable genes coalesced and differentiated into three major domains as we know them today. (He meant Bacteria and Eukarya and Woese’s third, Archaea.) Even after that differentiation, horizontal transfer continued throughout another couple billion years of history and into the present, especially within each domain but even sometimes between one domain and another. “Some biologists find these notions confusing and discouraging,” Doolittle granted. “It is as if we have failed at the task that Darwin set for us: delineating the unique structure of the tree of life. But in fact, our science is working just as it should.” How so? Because the tree itself was always just an “attractive hypothesis,” Darwin’s hypothesis, for the shape of life’s history. Scientists were now testing that hypothesis against fresh data, genomic data, and if need be, Doolittle concluded cheerily, they would reject it and find a new one.

  The Scientific American article was published in February 2000. During the next few years, as evidence of horizontal gene transfer continued to pile up, Doolittle became ever more engrossed in trying to gau
ge its significance. He read the papers announcing new data, and the analyses, and talked with (or argued against) other scientists at meetings. He found common ground with three men in particular: Peter Gogarten, William F. Martin, and Jeffrey Lawrence, a genome biologist at the University of Pittsburgh. At the end of his 1999 paper, he had thanked Gogarten and Martin for “persuading me of the importance” of HGT. In 2002 he invited Gogarten and Lawrence up to Halifax for a weekend of brainstorming. While they were in town, Doolittle “locked” the three of them in his laboratory, where they sat at the big wooden table in his office and wrote a paper together.

  All three believed that horizontal gene transfer was the new elephant in the room. They had agreed on that much during earlier conversations. If you weren’t thinking about HGT, Jeffrey Lawrence told me, when I visited his lab in Pittsburgh, “you weren’t thinking about what you were seeing.” But no one had deeply considered its implications. “You have to move beyond the collecting phase,” Lawrence said, meaning the phase of amassing data without analysis. “Or, ‘Here’s a case, here’s a case, gosh, there’s a lot of horizontal gene transfer.’ ” The big question, he said, was: What might it mean in the history of evolution?