The Tangled Tree

by David Quammen

  PART  VI

  Topiary

  60

  The town of Embarrass, Wisconsin, is halfway between Wausau and Green Bay. In 1907, on a plot of land thereabouts, a man named John Krubsack planted some box elder saplings in a careful arrangement and began growing them into the shape of a chair.

  Krubsack was a banker who also farmed (or a farmer who also ran a bank) and built furniture from driftwood as a hobby. He had decided to cultivate his “living chair” on a whim that amounted to a self-dare. His son later remembered him telling a friend: “Dammit, one of these days I am going to grow a piece of furniture that will be better and stronger than any human hands can build.” In 1908 he began bending, shaping, tying, and grafting the trunks and branches of the young box elders into the configuration he wanted. The grafts took. The stems grew together in a crisscrossing pattern. Krubsack pruned away whatever growth was extraneous to the blueprint in his mind. After four years, he removed all the rooted trunks except four—the four legs of his emerging chair—and, despite that truncation, the grafted-in sections continued to meld and grow. The legs and the crosspieces and the sweeping back and the arms thickened. The structure got stronger. In 1914 he cut it free of the ground. Presumably he sat in it and enjoyed a moment of satisfaction. Success. A year later, Krubsack’s chair was on display in San Francisco at the Panama-Pacific International Exposition—the 1915 world’s fair. Robert Ripley, who wrote a syndicated newspaper column titled Believe It or Not!, eventually featured the horticultural chair. Someone offered Krubsack $5,000 for the thing, but he declined. It stayed in the family. In due time, it became the totem, in a Plexiglas case, of a furniture company back home in Embarrass, Wisconsin.

  John Krubsack hadn’t invented any novel or esoteric techniques in creating his chair, though the concept and the execution were clever. Grafting was routine in the horticultural realm, and it’s still practiced today. A fruit tree is generally grown from rootstock of one type, onto which an upper section (or scion) of another type is grafted. The upper is fitted into the lower like a splicing of ropes. You make the cuts, insert one stem into another so that their cambium layers (containing the vascular plumbing) are in contact, wrap the area with tape, and wait. The rootstock might be selected for hardiness, to resist drought or disease, or maybe for dwarfism, so that the tree doesn’t grow too tall. The scion is selected for the kind and quality of its fruit. Grapefruit may be grown on rootstock of orange; commercial pears are often grown on rootstock of quince. When the cambium layers make contact, and the vascular systems merge, the graft has been successfully achieved. Water and nutrients from the rootstock can now flow up into the branches. Carbohydrates produced by photosynthesis can now flow from the leaves to the rootstock. Two trees have become one.

  A form of natural grafting occurs even in the wild—though rarely. It’s called inosculation. From a Latin verb, meaning “to kiss.” When the limbs or the trunks of two trees rub together, scraping away bark, creating two raw spots, cambium to cambium, sometimes those layers smooch and fuse. Dense growth, competition, and wind-driven rubbing can be the causes. It’s unusual, but it happens. What doesn’t happen, or only with very greatest rarity, is that two branches on the same tree inosculate. Branches on real trees diverge, reaching outward and away for light. Limbs diverge. I’ve said it already, and I’ll repeat: not everything that rises must converge. Limbs on an oak don’t. Branches on a cottonwood don’t. Twigs on a sycamore don’t.

  That’s the difference between actual trees and phylogenetic depictions. And so the tree of life concept became ever less satisfactory, ever more challenged, as new evidence of horizontal gene transfer continued to accumulate during the 1990s, because “tree” just didn’t suggest the right shape. There’s something spooky and unnatural about any tree whose limbs grow together rather than branching apart. Believe it or not.

  61

  The earliest tree of life reflecting any such anomaly was probably Constantin Merezhkowsky’s sketch of evolution via symbiogenesis, published as an illustration to one of his papers, in 1910. Yes, the crazy Russian pedophile again. This tree depicted the crossover of bacteria into the eukaryotic lineage, adding chloroplasts to complex cells, as a stream of dotted diagonal lines, from one limb to another. Lynn Margulis suggested the same thing, also with dotted lines, in the cartoony sketch that served as frontispiece to her 1970 book Origin of Eukaryotic Cells. Besides the chloroplast crossover, she showed two others, representing the bacterial origin of mitochondria (about which she was right) and of undulipodia (the little tails, about which she was probably wrong). Although these cases of endosymbiosis weren’t horizontal gene transfers in the narrow sense—the sense in which HGT is now mainly understood—in a broader sense they were, carrying whole bacterial genomes into the eukaryotic lineage, where they assumed new functions and created new possibilities. But those fateful endosymbioses (whether you count two or three) represented anomalous events in the very distant past. They were rare conjunctions, not instances of an ongoing process; and whether for that reason or others, they didn’t cause scientists to reconsider the whole tree metaphor.

  Carl Woese, for instance, ignored the crossovers entirely in his “universal phylogenetic tree,” offered with Kandler and Wheelis in 1990. Basing that tree on his favorite molecule, 16S rRNA, and on its eukaryotic equivalent, 18S, he felt no need to consider where and when other genes might have moved sideways. So the Woesean tree portrayed divergence without any hint of convergence. But then, during the later 1990s, ideas about horizontal gene transfer in evolutionary history changed drastically—and so did the illustrations by which those ideas were portrayed.

  One factor driving such changes was the dramatic improvement in DNA sequencing methods and tools, which brought an explosion of new genome data. The dangerous, toxic, and laborious steps by which Woese and his team had deduced the sequences of a few handfuls of RNA fragments back in the 1970s, ingenious as they were, looked like a Stone Age campfire compared with the streamlined, automated operations of the mid-1990s. The Human Genome Project, begun in 1990 and conducted by that huge government-and-university consortium I’ve mentioned, helped catalyze the technological improvements with big money, medical incentives, and the endless fascination of us humans with ourselves. The race against Craig Venter’s private group helped too, as competition put a premium on speedy and (for Venter’s group) cost-effective production of genome data. This led to fancy new machines and clever shortcuts. (Carl Woese himself acquired one of the new machines—an ABI 370A, made by Applied Biosystems, state of the art in 1986—but Woese’s lab people could never make it work.)

  Applying those methods and tools to the sequencing of nonhuman genomes, both as practice exercises and for the sake of pure science, was a side benefit. And with every passing year, genome sequencing became not only faster and more accurate but also cheaper. Another constraint on the huge task of sequencing genomes, besides technical difficulty and cost, was computer capability. Speedy computers with lots of processing power were necessary for assembling a large genome and analyzing what was there. As computers became so much faster, and methods of applying them to genome assembly improved, that constraint fell away too.

  Microbial genomes are much smaller than the human genome, and in those early days of automated sequencing, small genomes seemed less daunting but offered proof of principle. Venter’s method, known as whole-genome shotgun sequencing, involved detecting the sequences of randomly grabbed fragments, enough when totaled to comprise the whole genome, and then putting those fragments together according to how their sequences overlapped. It was the jigsaw-puzzle approach: Look, here’s a piece of blue sky, and it seems to match that piece of blue sky, so let’s see if they fit together. Yes! This was faster than the Consortium’s method, and so in 1995 Venter and his colleagues at TIGR, along with other partners from Johns Hopkins University and elsewhere, published the first complete genome of a free-living organism (that is, something larger and more complex tha
n a virus). It was the bacterium Haemophilus influenzae, the same bug that I saw at Porton Down in a strain cultured from Alexander Fleming’s nose. Venter and his group found that genome to be 1,830,137 letters long, and, to the best limits of their method, they identified every letter. Their report made the cover of Science.

  Another big event came the following April, when a different team announced its success in reading the genome of brewer’s yeast. Brewer’s yeast may not sound thrilling to you or to me—it’s not charismatic megafauna—but it is a eukaryote, and no complete eukaryotic genome had ever before been sequenced. The brewer’s yeast genome was therefore closer to the human genome than any other whole-genome sequence yet produced. It was also somewhat larger than the average genome of bacteria. Those distinctions, plus the nervous atmosphere of the race toward the human genome, with Venter’s gang in one lane and the Consortium in another, may account for why this third team, a far-flung international group, made its announcement by press release well before the formal report could appear in a scientific journal. Hey look, we’ve done a eukaryote!—and we’ll publish details pretty soon. The pace of discoveries and the fervor of competition were picking up.

  Just four months later, in August 1996, Venter and a large team of collaborators captured attention again, publishing the first complete genome of a member of the Archaea, the third of Carl Woese’s three domains. This bug was Methanococcus jannaschii, a heat-loving and methane-producing microbe first isolated from a sediment sample taken at the bottom of the Pacific Ocean. That sample was scooped up by a robot submersible driven along the sea floor near a thermal vent, more than eight thousand feet deep near the Eastern Pacific Ridge. Like most other archaea known at that time, it was a strange little creature from an extreme environment. Woese himself appeared as an honored senior coauthor (second from last in a list of forty, just before Venter) on the paper announcing the achievement, again in Science. He had persuaded Venter to undertake this piece of work. Gary Olsen, one of his young and ingenious collaborators at Urbana, was also among the coauthors, but otherwise the list comprises mostly Venter people from TIGR. It must have been a bittersweet event for Woese, when the first archaean genome came from Venter’s institute, not his little lab, with its nonfunctioning ABI 370A.

  The M. jannaschii genome ran to 1,739,933 letters, including 1,738 sections that seemed to be genes. Of those genes, more than half were entirely new to science, with no equivalents ever before seen in any other form of life. That degree of uniqueness went far to confirm what Woese had been saying since 1977 and what some hidebound scientists had resisted: that these archaea were a separate form of life. The old two-kingdom paradigm had now been “shattered,” according to one eminent microbiologist, asked to comment by Science for a news story accompanying the report. “It’s time to rewrite the textbooks.”

  Ford Doolittle agreed. “This completes that basic set,” he told Science, meaning the triad of whole-genome sequences for a bacterium, a eukaryote, and an archaeon, “and so it will certainly have a major impact.”

  One impact it helped deliver, like the blade of an axe, was on the very idea of the tree—particularly on the tree as Carl Woese had drawn it, using ribosomal RNA as the definitive signal of life’s ever-diverging history. Whole-genome sequencing of that first bacterium, then that first archaeon, and then other organisms revealed more and more instances of horizontal gene transfer, confusing the picture and inosculating limbs. Within another two years, by 1998, more than a dozen microbial genomes had been sequenced and another from a eukaryote, a nematode worm.

  Scientists who inspected those genomes found puzzling mixtures of bacterial genes and archaeal genes within single genomes, as though part of a deck of tarot cards had been shuffled into a poker deck. Occasionally a few bacterial or archaeal genes even turned up in eukaryotes. Another reporter from Science, Elizabeth Pennisi, picked up the story and described the growing perplexity. By that point, even Woese was pondering horizontal gene transfer, though in his view, it was a phenomenon limited mostly to the earliest era of evolution, when cellular life was just taking shape and there weren’t yet any distinct lineages or species. A blurry time, and HGT back then was part of the blur. Pennisi spoke with Woese for her report, quoting him to the effect that “you can’t make sense of these phylogenies because of all the swapping back and forth.” He probably meant “all the swapping” during that earliest time, but the distinction didn’t make it into Pennisi’s article.

  Another source, a molecular geneticist named Robert Feldman, who had just helped sequence another bacterium and found its phylogenetic relationships ambiguous, voiced his dawning distrust of Woese’s faith in an rRNA tree. Feldman noted that “you get different phylogenetic placements based on what gene is used”—different genes yield different trees of relatedness—and he explained why: “Each gene has its own history.” If each gene has its own history, Woese was wrong to draw such grand conclusions from a single molecule, however fundamental, and sketching the course of evolution as a single neat image was impossible. Pennisi saw that. Her piece ran in May 1998 under the headline “Genome Data Shake Tree of Life.”

  62

  Ford Doolittle absorbed this new line of thought gradually. He was skeptical of the notion that horizontal gene transfer might have played—might still be playing—a huge, unsuspected role in the history of life. Yes, he could see, it was a mechanism for the spread of antibiotic resistance from one kind of bacteria to another. But beyond that? Did it explain why certain other genes, more basic to the functioning of single-called organisms, were turning up on what seemed the wrong limbs of the tree of life? Those anomalies multiplied as more individual genes were sequenced, and then whole genomes, but alternate explanations existed for why a bacterial gene, or what looked like one, might show up in an archaeon, and vice versa. The alternate explanations didn’t require such extraordinary leaps. They weren’t so dramatic and counterintuitive. Horizontal gene transfer still seemed deeply improbable, a rare sort of event, and Doolittle remembers calling it “the last resort of the impoverished imagination.”

  Unlike some scientists, Ford Doolittle abides in a zone of detached bemusement and pure curiosity that leaves him comfortable admitting when he was wrong. It’s good scientific etiquette, arguably even the scientific ideal: you hypothesize, you test against data, you correct your views where necessary, regardless of ego; you hypothesize again. When you’ve goofed, and you need to backtrack, you admit it. Doolittle practiced that. He started revising his view of HGT under the influence of two colleagues and all the new data those colleagues helped him consider. One of the two was a postdoc in his own lab.

  James R. Brown came east to Halifax after finishing a PhD at Simon Fraser University in British Columbia, where he worked on molecular evolution and the population genetics of sturgeon. Brown had always been a fish-loving kid, growing up in Ontario with aquariums full of cichlids and angelfish and an interest in marine biology. He snorkeled cold Ontario waters during summer vacations, watched Jacques Cousteau on TV, and read books about the sea. After his undergraduate degree in marine biology, he worked as a marine field technician and scuba diver in the Great Lakes and the Arctic for the Canadian government before going back to school. The group of sturgeon he studied for his PhD dissertation includes some fascinating fishes: long-lived animals with primitive traits and recognizable ancestors dating back more than two hundred million years. Brown focused his doctorate on mitochondrial DNA as a gauge of genetic diversity among their populations. In the process, he learned a bit about using molecular data to draw phylogenetic trees. Bringing those skills to Doolittle’s lab, he joined his boss on a series of projects and publications, during the 1990s, that involved not sturgeon populations but the molecular phylogenetics of bacteria, archaea, and eukaryotes.

  Brown’s project involved the question of how to root the universal tree: Should it be midway between the bacterial limb and the archaea limb, with eukaryotes branching secondarily from the archa
ea? Or not? And could it be done on the basis of a single fundamental gene? They investigated one gene, for instance, that seemed to have gotten into certain bacteria by sideways transfer from archaea. Other genes, they knew, gave other answers to the rooting question, and they looked at several related genes for a possible clue. They recognized that horizontal gene transfer, with genes leaping sideways from limb to limb, might be what had made the rooting question (among others) so hard to answer. “Extensive gene transfer,” Brown and Doolittle wrote, with two other colleagues, “may have played such an important role in early cellular evolution as to jeopardize the very concept of cellular lineages.” Meaning: Damn it, Maybe there is no single tree of life. Or if there’s a picture of life’s history, maybe it doesn’t look like a tree. Ford Doolittle was coming around.

  Another influence on him was a colleague named Peter Gogarten, a German-born scientist who had trained in plant physiology, come to America in 1987, and shifted into molecular studies of early evolution. Gogarten, since his arrival in the States, had his own interesting experience with Carl Woese and the three domains. As a postdoc in California, he devised a method, in collaboration with his lab leader and others, of determining where the tree of life should be rooted. Their answer: at the base of a trunk that rose into two major limbs, one representing bacteria, the other leading to everything else. Either for that reason (two major limbs, not three) or others, Woese didn’t care for Gogarten’s paper and omitted it pointedly (despite its clear relevance) from the citations in his landmark 1990 paper with Kandler and Wheelis. Too bad for young Peter Gogarten, who was a new assistant professor at the University of Connecticut, needing some further publications and recognition if he were to get tenure.