The Tangled Tree

by David Quammen

  I had asked Gogarten a few months earlier: How do three people write a paper in a weekend? Does one sit at the computer while the other two . . .

  “No, we all sat at computers,” he said. They had their laptops. “We discussed things. We had a kind of discussion outline, a draft. And then, yeah, we each went to our computers and wrote sections on it. Did some computations.” Gogarten’s computations led him to generate a few trees—trees of descent for individual genes, and one tree sketching the history of whole organisms. Lawrence nixed the tree of organisms. “We can’t do that,” he said, by Gogarten’s recollection. “We can’t write a paper on gene transfer and use a tree to illustrate that.” Lawrence’s point being that, with genes flitting sideways, there was no tree that could depict the history of whole organisms. It wouldn’t just be wrong. It would be self-contradictory; it would be nonsense. So they included no tree in this manuscript—not Martin’s colorful sea fan, not the Scientific American sort of thing, not even one of Doolittle’s fluidly hand-drawn tangles.

  They focused on “prokaryote” evolution, using the old word for bacteria and archaea, lumped together, that so aggravated Woese. Among such simpler microbial creatures, they wrote, horizontal gene transfer is far more important, in quantity and consequences, than imagined previously. Its impacts could be understood in four ways. First, new genes received by sideways transfer, from a different lineage or species, may allow a population of microbes (the recipient bug and its offspring) to colonize an entirely new ecological niche. Second, it may allow organisms to acquire a new sort of adaptation abruptly, without passing through the dangerous stage of being only half adapted to one situation or another. Third, this transformation happens fast compared with incremental mutation, which proceeds slowly. Fourth, HGT is a “font of innovation,” bringing drastically new genetic possibilities, new supplies of variation, on which natural selection can act. All four kinds of impact are interrelated and represent overlapping perspectives on the same phenomenon.

  Take those four together, and you have a strong case, they wrote, that horizontal gene transfer might be the “principal explanatory force” in prokaryote evolution. Darwin’s natural selection is still there, but it’s operating on a much different supply of variation from a much different source than imagined previously. What the three authors meant to show in this paper, they stated, was that recognizing the role of horizontal gene transfer led toward “a broad and radical revision” of the old paradigm. The old paradigm they meant was that microbes conform to Darwin’s theory. But this should be a revision, they emphasized, not an outright rejection. They proposed “a synthesis” of old and new perspectives, acknowledging that gene transfer is horizontal as well as vertical, that life’s history looks both “weblike and treelike,” and that adaptations can evolve by “many modes,” not all of which were discernible to Charles Darwin in 1859. Their paper appeared in December 2002.

  In the meantime, that fourth figure, William F. Martin, had continued his own challenge to the conventional tree of life. Ford Doolittle sometimes speaks jokingly of “the four horseman” of horizontal gene transfer—the four scientists who most loudly declaimed its importance around the turn of the twenty-first century—and when he does that, he counts Bill Martin, along with Gogarten and Lawrence and himself. (Jeffrey Lawrence asks: Which horseman am I, Pestilence?) But for some reason, or no reason, Martin didn’t participate in the “locked-in-a-lab” coauthorship weekend. Was he not invited, not interested, not available? Was there bad chemistry with one of the other horsemen? Martin, as I’ve mentioned, has a reputation for being smart, confident, forceful with his views, and sometimes startlingly brusque when disputing points of science in public. Then again, sheer geography might have been enough to keep him away from that Halifax confab. Martin lives and works in Düsseldorf, an ocean away. That’s where I went to see him, curious for his thoughts, respectful of his work, but wondering whether I’d get a blast of his famous gruffness.

  65

  Bill Martin’s lab is in the Department of Molecular Evolution at Heinrich Heine University, in the heart of Düsseldorf, not far from a great bend of the Rhine. I arrived early, was ushered through by his secretary, and sat waiting in his inner office while Martin finished participating in a PhD examination. That left me time to browse the books on his shelves, gaze at the arcane scribblings on a large standing flipchart, and notice the poems, cartoons, and other whimsical décor on the back of his door, including photographs of notable predecessors and colleagues, such as Constantin Merezhkowsky and Ford Doolittle. There were also framed photos of two smiling young daughters. Martin appeared at ten thirty, a large man, large enough to have played lineman during his time at Texas A&M University, though he did not. He greeted me with a robust Texas handshake, we sat at his table, and he talked for two hours, almost without need of prompting by questions from me.

  The beginning of it all, for him, was endosymbiosis—hence the photo of Merezhkowsky. Martin was a young student in botany, hoping to run a nursery someday, and not yet aware that, as he told me, “if you want to have your own nursery, you don’t study botany, you study business.” He soon discovered, anyway, that he was more interested in research than in selling plants. “I just wanted to know about things.” When he took microbiology, in 1978, the professor made two statements that gripped Martin’s attention. “The first thing he said was, ‘We used to isolate insulin from pig pancreas, and now we can take the gene, put it in E. coli, and make it in buckets.’ ” That was a glimpse of the prospects of genetic engineering. But the real fascination, for this student, lay in pure science. “The other thing he said is, ‘And there are some people who believe that chloroplasts used to be free-living cyanobacteria.’ ” That was Martin’s introduction to the theory of endosymbiosis—the idea that complex cells arose by capturing bacteria and converting them to internal organelles.

  The theory was still in disrepute. Lynn Margulis had revived it, from the writings of Merezhkowsky and Ivan Wallin and others; Ford Doolittle and his colleagues had delivered some of the first molecular evidence to confirm it; further gene sequencing, of the slow and laborious sort, was adding support; but widespread acceptance hadn’t yet come. As Martin reminded me, back in 1925 endosymbiosis had been called “too fantastic for present mention in polite biological society,” and to many biologists, it still seemed too fantastic. Young PhDs of Martin’s generation were warned not even to mention the theory when they auditioned for academic jobs. “It was just completely taboo,” he told me. Until suddenly it wasn’t—when the new molecular data assembled by Doolittle and others showed that those cell organelles, mitochondria and chloroplasts, carried genomes reflecting their bacterial origins. Within a few years, even polite biological society had accepted it. Around that time, in the middle 1980s, Bill Martin began his own work.

  He had discontinued his studies at Texas A&M, worked as a carpenter, traveled through Europe, and become fluent in German, until his scientific interests resurfaced, and he enrolled at a university in Hanover. By 1988, he had finished a doctorate, in Cologne, on molecular genetics and plant evolution. During that period, Martin became familiar with bacterial genetics and bacterial geneticists, for whom HGT was common knowledge. His earliest research project, an investigation of chloroplast enzymes, led him straight into the endosymbiosis theory and the role of HGT within it. He already knew that the theory was probably correct, and he learned one other important thing: the genomes found in chloroplasts and mitochondria are tiny—far too tiny to code for all the enzymes and other proteins that those organelles need to function. They are only minuscule samples of their original bacterial genomes. The other genes, manufacturing those hundreds of other proteins, must be still present in the cell, somewhere, but not in the organelles. Where else could they be? In the one place where genes live protected: the cell nucleus. These missing organelle genes must have been transferred from the captured bacteria into the nuclei of their respective cells—maybe one gene at a
time over the long eons of eukaryote evolution—becoming patched into the cell’s own nuclear genome. “So gene transfer was a part of the picture from the get-go,” Martin told me. Not just interspecies transfer, but transfer across the big boundary, between domains. “That is the world that I grew up in.”

  As his research progressed, he found more evidence that this sort of gene transfer—from the captured bacteria into the nuclei of complex cells—had been common and widespread throughout time and across a range of creatures. It occurred at different rates, with different results, in different lineages of animal, plant, and fungus. One plant that Martin looked at, a small flowering thing related to cabbage, has a nuclear genome of which 18 percent is bacterial. Another of his studies showed that yeast, a fungus, contains 850 genes from bacteria and archaea. The human nuclear genome includes more than 263,000 base pairs (letters of code) of what was originally bacterial DNA, transferred from our mitochondria. In these cases and many others, genetic material has escaped from the organelles, leaked into the cell nuclei, and gotten integrated into the chromosomes. This is a profoundly consequential process: the transit of DNA from organelles of bacterial origin into the chromosomes; alien genes becoming incorporated over millions of years into the deepest cellular identity of plants, fungi, and animals. And no one knows, not yet, just how it has occurred. Martin eventually coined a term for the phenomenon: endosymbiotic gene transfer.

  It’s horizontal in a slightly different sense: sideways passage of genes from one domain of life to another, yes, but within the confines of a single cell, and therefore even subtler than other forms of HGT. It began when our remote single-cell ancestors incorporated those fateful bacteria.

  You could think of it this way: as domestication and transfer of duties. Wolves find their own food. Dogs, being domesticated, rely on humans to feed them. By mutual agreement, over the course of fifteen thousand years, wolf descendants have transferred their food-gathering functions (and responsibilities) to us. It began with bones and meat scraps at the edges of human campfires, probably. It progressed to all manner of extremes. In exchange for food and other emoluments, canines now offer love, bark at mailmen, herd sheep, point at pheasants, and chase Frisbees. Likewise mitochondria: they are domesticated bacteria in your cells. They have transferred many of their genes to your nuclear genome, and they rely on that genome to send back proteins enabling them to exist and do their work. Instead of chasing Frisbees, they manufacture ATP, that battery-like molecule I’ve mentioned before, the one that offers portable energy for fueling your metabolism.

  Of course, all this discovery of endosymbiotic gene transfer, Martin explained, led to further problems for Carl Woese’s tree of life. There were too many branches going every which way, including many that branched from one major limb and then inosculated with another. Further complicating matters was the fact, revealed as more genomes were sequenced and compared, that those bacteria captured early in eukaryotic cells—the ones that became mitochondria and chloroplasts—had themselves been recipients of horizontal gene transfer, from different kinds of bacteria, before their capture. This meant parts of genomes existed within other genomes before becoming parts of still other genomes, including yours. It was all a snarl. It was a mess. It was a plate of spaghetti. It was wonderful.

  “It’s not a tree, though,” Martin said.

  As we talked away the morning, he sometimes interrupted himself to jump up and grab a book, or to find the file of a paper on his computer and print me a copy, or he tilted back his head to ruminate and launch a new topic. Or he paused to refocus, asking “What were we talking about?” At one point, after a flurry of erudite speculation on the origin of photosynthesis, the varied methods of nitrogen fixation, and the adaptive value of sex, he said, “Sorry, I’m just rambling on.”

  “No, no. That’s good,” I said. But yes, I did want to get back to HGT and the tree. “How much time do we have?”

  “We have all day. I planned you in all day.”

  “Bless you,” I said, and then a half hour later, he suggested we break for lunch. Exhausted by the flow of ideas and information, I turned off my recorder.

  Did I like Japanese food? he asked. Sushi? Yes indeed I did. Good, Martin said, because I’m supposed to be losing weight, but you’re my excuse. We drove to his favorite Japanese restaurant, where he ordered us a huge and very excellent meal, ate voraciously, and allowed me to pay for nothing. Over the noodle soup and the combo plates, we talked more about the origin of eukaryotic cells, the origin of the cell nucleus, and the ultimate origin of life. I shared with him, somewhat shamefacedly, the fact that I had watched the Super Bowl broadcast from two to four that morning, streaming it live on my computer at the hotel, then went back to bed for a few hours’ rest before meeting him. Patriots over Seattle on Malcolm Butler’s last-second interception, worth every minute of lost sleep. I know, I watched it too, he said. All this time, I kept wondering: Where is the famously brusque, intolerant-of-fools, formidably combative Bill Martin?

  After lunch and tea, we returned to his office for more talk. At the end of the afternoon, he offered to walk me back toward my lodgings, not far, so I wouldn’t get lost. A little exercise was better than another taxi. It was a chilly winter day, and he wore a casual black jacket and a watch cap, striding the pavement like John Cleese disguised as a stevedore. At a major crossing, he pointed: straight down that road to your hotel. Himself, he would go back now to the lab for more work. We shook hands. Let’s get together again and talk, Martin said. This was fun, he said.

  66

  Bill Martin did his part, back around the turn of the millennium, to integrate horizontal gene transfer into evolutionary thinking and to consider how it strained the idea of the tree. Although he didn’t participate in the Halifax writing weekend with Doolittle and Lawrence and Gogarten, he published a series of his own papers, alone or with coauthors, in a similar vein. The first of them, “Is Something Wrong with the Tree of Life?,” preceded Ford Doolittle’s manifesto by three years. Another, as already mentioned, discussed the “mosaic” character of bacterial genomes. Given the evidence of relatively recent and abundant horizontal transfer in a familiar bug such as E. coli, with almost a fifth of its genome acquired from other bacteria, Martin found it “quite ominous” to contemplate how HGT must have played out across the depths of time. He meant “ominous” in a retrospective sense: a portent that the past, not the future, still held some wild surprises. This paper on mosaic genomes included Martin’s distinctive tree illustration, the multicolored sea fan, showing HGT by way of merging lines and blended hues. “That’s a nice tree. I mean, that’s what it looks like,” he told me with some small pride. “I drew that all by myself.” Very different from Ford Doolittle’s freehand style, but making the same three points: the shape of the tree is important, and counterintuitive, and here it is.

  In a later coauthored paper titled “The Tree of One Percent,” Martin described how endosymbiotic gene transfer, his favorite kind, continues to move genes of bacterial origin from mitochondria and chloroplasts into the nuclear genomes of complex creatures. He noted that only 1 percent of the genes in an average bacterial or archaeal genome—and maybe far less than 1 percent in the genome of a eukaryote—are so deeply and complexly essential to the organism that they couldn’t be swapped by HGT. The 1 percent figure came from work by another group of scientists, who studied thirty-one select proteins of the roughly three thousand proteins coded in the genome of an average prokaryote, and offered those “universal” proteins as a basis for phylogenetic analysis. Martin, with a coauthor, turned that group’s logic against them. Sure, you might well use your preferred, stable genes to define a single tree. (It’s what Woese had done with 16S rRNA, though Martin didn’t say so explicitly.) But if you did that, your tree of life would be really just “the tree of 1 percent” of those genomes—a small selection, unambiguous but not necessarily representative. And if your tree captured the story of only 1 percent of each genom
e, what was the point? You’d be better off depicting life’s history with graphs and theories than drawing any such marginal tree.

  Carl Woese, in Urbana, wasn’t oblivious to all this. He saw the unfolding discoveries of horizontal gene transfer and grasped the challenge they raised against his Big Tree. While the four horsemen of HGT galloped across the field of battle, their banners flying, Woese took the phenomenon into account and made his own sense of it. He formulated a fallback position that allowed him to celebrate the importance of HGT, while reconciling—or anyway, seeming to reconcile—the new data on sideways genetics with his own older ideas and trademark discoveries. During the late 1990s and early 2000s, he published a series of conceptual articles on early evolution, the origins of cellular life-forms, and what he began calling “the universal phylogenetic tree.” By that last phrase, he meant his tree of life, the one derived from comparing sequences of ribosomal RNA.

  This group of papers, four of them, have been called his “millennial series,” and as one expert commentator noted after Woese’s death, they were peculiar in several ways. They didn’t report any new research. But they weren’t review articles in the usual sense. They contained almost no data from his own work or anyone else’s. They were too serious and dogmatic to come across as essays or opinion pieces. The commentator, a brilliant Russian-American biologist named Eugene Koonin, who was quite respectful toward Woese, found them hard to categorize except maybe as “treatises” or “tracts.” Call them whatever, they were pronouncements on early evolution from a man with great personal confidence that he understood that vast, blurry subject better than anyone else. What makes the millennial series relevant here is how Woese tried to resolve the tension between rampant horizontal gene transfer and his beloved universal tree. He did it by pushing HGT into the very distant past.