The Tangled Tree

by David Quammen

  This wild thought became reality a year later, in a paper by Woese with Otto Kandler and one other scientist (but not Zillig) as his coauthors. For publication, Woese turned to the Proceedings of the National Academy of Sciences, a journal in which—as a member now of the Academy—he could be a bit more speculative and face less rigorous peer review than he would at Nature or Science. The paper, published in June 1990 and titled “Towards a Natural System of Organisms,” made several main assertions. First, any system of classification should be strictly “natural,” as the title suggested—meaning phylogenetic, reflecting evolutionary relationships, and not compromised by convenience for memory and teaching (as Whittaker and Margulis preferred). Second, there should be three major divisions of life, higher in rank than Whittaker’s kingdoms, higher also than Haeckel’s kingdoms or Copeland’s or Lake’s, and those divisions should be known as domains. Three domains, recognized above the old kingdoms rather than replacing them: it was ingenious strategically, transcending rather than rejoining the battle of the kingdom keepers. But incidentally, it led to a small conflict between Woese and Kandler regarding the third coauthor, Mark L. Wheelis.

  Wheelis was a younger microbiologist at the University of California, Davis, not previously known for work in phylogenetics. He had trained under Roger Stanier at Berkeley in the 1960s, encountered Woese passingly during a postdoc at Urbana, and become friendly with Mitch Sogin while teaching molecular evolution in Sogin’s summer course at Woods Hole. How those contacts or other factors may have brought him back to Woese’s attention, in the late 1980s, even Mark Wheelis doesn’t know.

  “It’s always been sort of a mystery to me,” Wheelis said when I reached him by telephone. He had been interested in what he called “the kingdom problem”—the glaring illogic of treating plants and animals each as a distinct kingdom coequal with all bacteria, a much more overarching group—and he may have mentioned in conversations that a new, higher category of classification was needed; but he hadn’t published on that subject. And then, “out of the blue,” he received a draft of what became the 1990 paper. “This manuscript showed up in my mailbox one day, with Carl asking if I had comments.” Wheelis made suggestions, including the one about a higher category, and returned the draft. Woese incorporated many of Wheelis’s notes, raised other questions, and the manuscript flew back and forth four or five times. At that point, as Wheelis recalled, he asked Woese to consider adding him as a coauthor.

  Woese agreed, notifying Otto Kandler in Germany. Kandler was surprised at the addition of a third collaborator, but he put on a game face and acquiesced. “I have no objections if Mark becomes a coauthor,” he wrote Woese, “although I am not convinced that it is fully justified.” Kandler noted that the Wheelis suggestion about a higher category merely “reactivated” an idea that Woese had already entertained. Wheelis was nevertheless “brought on board,” according to Jan Sapp, his contributions rewarded with coauthorial credit. This is the decorous version of Woese’s decision process, anyway, as told by Sapp in his superb book The New Foundations of Evolution. In private, over a glass of wine, Sapp gave me a slightly different version.

  Woese, fond as he was of Kandler, didn’t want it to appear that his German colleague had codiscovered the archaea. That was Woese’s great distinction, shared with George Fox and, to a lesser degree, Wolfe and Balch, but no one else. He wanted his full measure of glory. Sapp himself loved Woese like a brilliant uncle and worked closely with him while researching New Foundations, but he was clear eyed about Woesean foibles. “I know this man.” If the three-domains paper was a milestone, with just two coauthors, “then he and Kandler would be seen as codiscoverers. But he puts Wheelis in, and it waters Kandler down.”

  “That’s bizarre,” I said. Yet it’s how science sometimes works—and human nature, often.

  “Yeah,” said Sapp. “That’s Carl.”

  Last of the paper’s main points was that these three domains should henceforth be known as the Bacteria, the Eucarya, and . . . the Archaea. The word archaebacteria should now disappear, the authors argued. So should the word prokaryote. Prokaryotes didn’t exist as a phylogenetic category—it was a false unit—because Archaea and Bacteria stood so utterly distinct from each other.

  The “Natural System” tree of Woese, Kandler, and Wheelis, 1990.

  And, of course, there was a tree. It was drawn in straight, simple lines, but it was rich and provocative nonetheless. Unlike his rootless 1987 tree, this one was rooted, using a complicated technique (involving duplicated genes traced back into deep time, but never mind those details) that Woese had come upon more recently in work by some Japanese researchers. Its trunk rose vertically from a single origin, then split into two big limbs, and then one limb split again. The left limb was Bacteria. The two limbs on the right were Archaea and Eucarya (that spelling later corrected to Eukarya, as I’ve mentioned, a better transliteration of the Greek root). This arrangement asserted what Woese’s 16S rRNA data showed: that we humans, and all other animals, all plants, all fungi, all eukaryotes, have arisen from an ancestral lineage that was unknown to science before 1977. It was the last of the great classical trees: authoritative, profound, completely new to science, and correct to some degree. But it entirely missed what was coming next.

  PART  V

  Infective Heredity

  49

  What came next was an exploding awareness of the role played by horizontal gene transfer in this whole story. That explosion occurred during the 1990s but had deep precedents. The long, bizarre history of the HGT phenomenon goes back about four billion years, in fact, and the first recognition by science that any such thing might be possible dates to 1928. It grew from work published that year by an Englishman named Fred Griffith, though no one at the time, not even Griffith himself, saw the implications of what he had found.

  Griffith was born in a small town in North West England and educated in Liverpool, just down the Mersey estuary, with a degree in medicine and a fellowship in pathology. He worked for a while in a local laboratory and a hospital, picked up a diploma in public health at Oxford, and did research on tuberculosis before joining Britain’s Ministry of Health in London during the Great War. He had a long, straight nose and a steady gaze. As a medical officer at the ministry’s Pathological Laboratory on Endell Street, just north of Big Ben, he shared an upstairs lab space and kitchen with two technicians and his colleague William M. Scott; downstairs was the post office.

  Griffith was a fastidious bench scientist and a consummate civil servant, narrow of focus, wary of speculation, described as “very shy and aloof and difficult to get to know,” whose assigned topic was bacterial pneumonia. Scott’s work is less known, but they became friends and remained side by side to the end, in 1941. By one account, in their little underfunded laboratory, Griffith and Scott “could do more with a kerosene tin and a primus stove than most men could do with a palace.”

  The bacterium Pneumococcus pneumoniae (now called Streptococcus pneumoniae), which Griffith studied, was a dangerous bug that could cause severe, often fatal, pneumonia. During the 1918–19 influenza pandemic, this kind of pneumonia took hold as a secondary infection in many patients and probably killed more millions of people than the flu virus itself. Antibiotics didn’t exist then. The best treatment was antiserum therapy, using blood serum rich with antibodies drawn from inoculated horses; such serum could bolster a patient’s immune response and help clear the bacterial infection. But there were at least four different types of pneumococcus—Types I, II, and III, plus a catchall type known with sublimely unnecessary confusingness as Group IV—and therefore several different sera, which were specific for type. When giving treatment, you wanted to know which pneumococcal type a patient was inflicted with, so as to pick the right serum. That’s the sort of thing medical bacteriologists did. Griffith’s work during the 1920s involved telling one type from another, investigating their properties, and tracking the prevalence of different types in different pneumo
nia outbreaks around the country. He studied almost three hundred cases between 1920 and 1927 and saw, among other things, that pneumococcus was rife in Smethwick district, just west of Birmingham, but that Type II was giving way there to Group IV. Such intelligence was useful for tracking outbreaks and judging which sera to prepare and ship. Griffith got his data by examining sputum coughed from the lungs of the ill. He had an ice chest full of hacked-up gobs.

  In 1923 Griffith discovered something important: that in addition to different types of the pneumococcus bacterium, there existed two different forms within each type—one that was ferociously virulent, one that was mild. The virulent bugs clustered in colonies that looked smooth under microscopy, so he labeled that form S. The nonvirulent form clustered in rough colonies, so that was R. Sometimes the S form might transmogrify into the R form, he noticed. He didn’t know why. Maybe it was mutation and natural selection, possibly in response to serum. Or not. Then he made a second discovery, far more surprising even to him: under certain experimental circumstances, the R form of, say, Type II bacteria could change into the S form of, say, Type I. What? It seemed as though the pneumococcus had morphed into a different species. But Griffith was firmly grounded in the Ferdinand Cohn school of bacteriology, which held (as your keen memory will tell you) that bacterial species are fixed, stable, knowable—not protean and capable of shape-shifting magically from one into another. Yet here was a change. Griffith doubted his own lab technique and tested again. Same result. Having taken such care to exclude the possibility of contamination, he wrote later, admitting his skepticism, that “there seems to be no alternative to the hypothesis of transformation of type.” And so that’s what he called the phenomenon: transformation. One form of bacteria transformed into another. It was mystifying.

  These transformations happened, for Griffith, in the bodies of his laboratory mice. His experimental procedure involved injecting mice with one type or another of pneumococcus (say, Type I or Type II), one form or another (S or R), and sometimes also a dose of serum, then watching to see whether the mouse died or survived. If the mouse died, he would extract a blood sample, culture the bacteria that had been raging within the animal, and by microscopy determine its type and its form. In the most revealing of his many experiments, he gave each mouse two forms of bacteria, injecting a dose of dead form S (virulent, if it had been alive) and living form R (mild). He had killed the dose of virulent bacteria (but not entirely destroyed its biochemical components) by heating it at a carefully chosen temperature. His first interesting result from this method was that a mixture of dead S and live R was capable of killing a mouse.

  Even more surprising were the results when he mixed types as well as forms. In one experiment, for instance, he injected heat-killed S form (virulent) of Type I bacteria along with living R form (mild) of Type II bacteria into five mice. When all five keeled over within a few days, and Griffith drew blood, he found living Type I that was virulent. Note again this change: dead virulent I, plus living mild II, becomes . . . living virulent I. Something weird had happened. It sounded like zombie bacteria. Either the mixing had brought the virulent Type I back to life, or else the dead Type I had somehow transformed the living Type II into a version of itself. This wasn’t a sci-fi movie, and neither of those options was supposed to be possible.

  Griffith himself struggled to explain it when he published a long paper on his pneumococcus work. It seemed to him that the living bacteria “actually make use of the products of the dead culture” for generating their own capacity to be virulent. How could that work? Well, maybe some part of the wreckage of the killed, virulent bacteria served as a kind of “pabulum” from which living, nonvirulent bacteria built up their own virulence. Pabulum, British spelling for our pablum, simply means a pureed and easily absorbable food. The groping vagueness of that suggestion reflects the difficulty of interpreting Griffith’s very peculiar experimental results, even for Griffith. Still, vague or not, it was roughly right.

  In the summary of his pneumococcus paper, Griffith declined to speculate about the “pabulum” at all, merely listing his results like the steady, empirical government employee he was. At the top of his list: Type II pneumonia has declined in Smethwick, though the incidence of Type I has held steady. Near the end of his list: oh, and by the way, dead bacteria seem to be capable of transforming live bacteria from one type to another.

  Griffith never pursued this topic further. By one account, he seemed uninterested in the phenomenon of transformation, possibly even irritated by it (because it seemed to contradict the fixity of species) and happy to leave it “up to the chemists” to explain. He shifted his research from pneumococcus to other bacteria. His paper, with its passing notice of transformation buried within forty-seven pages of precisely described experiments, became widely read and influential—one historian has called it “a bombshell which fell into a fused situation”—as other researchers (and not just “the chemists”) picked up where he had stopped. He never knew, never lived to see other scientists discover, what the mysterious pabulum was.

  As for himself, he had no taste for drama or scientific limelight. He rarely went to meetings or gave talks. When the International Congress of Microbiology met in London in 1936, Griffith “had to be practically forced into a taxi” to get him to the event, though he was committed to speak. He never married, and at one point, he shared his London flat with William M. Scott, his friend and colleague since the early days at the Pathological Lab. He died on a night in April 1941, during the Blitz, and so did Scott, when that apartment took a direct hit. His two-paragraph obituary, as it ran in the British Medical Journal, didn’t mention transformation.

  50

  Griffith’s long paper appeared in January 1928. His reputation for rigor was excellent, but his discovery seemed so bizarre that other bacteriologists remained unsure, at least for a year or two, whether to swallow it. A few worked to redo his experiments and confirm, or not, his findings. They confirmed them: yes, there was no mistake, transformation occurred. But what was the stuff, Griffith’s pabulum, that came from dead bacteria and caused live bacteria to change identity? What might it reveal about the very nature of heredity?

  These questions arose in a context—the context of early-twentieth-century genetics—in which the word gene was used freely, but no one really knew what it meant. A gene was an abstraction made concrete only by the word itself, that nice little unit of jargon coined in 1909 to stand for an entity of some sort that determines hereditary traits.

  Geneticists suspected by then that genes reside on the chromosomes in a cell, and chromosomes could be seen through a microscope, but the genes themselves couldn’t. Were they physical realities, discrete chemical units arranged on a chromosome like beads on a string? Or was each “gene” just the net effect of some measured-out quantity or fluctuating process, as Darwin had (wrongly) guessed? As late as 1934, the eminent American geneticist Thomas Hunt Morgan, who pioneered studies of mutation and heredity using fruit flies, said in his Nobel Prize acceptance lecture: “There is no consensus of opinion amongst geneticists as to what the genes are—whether they are real or purely fictitious.” It didn’t matter, Morgan added, because “at the level at which the genetic experiments lie,” the results were the same either way. He meant genetic experiments like his, concerned with relative positions of genes on chromosomes, and how such positioning influences new combinations during sexual reproduction.

  But as Oswald Avery, a medical researcher at the Rockefeller Institute, and others looked more closely at phenomena such as transformation in nonsexual creatures such as bacteria, it damn well did start to matter whether a gene was a material reality or a cloud of influences. If there was a pabulum that carried heritable change, what was the recipe? If the hereditary substance was a physical thing, a chemical entity, which molecule or molecules composed it?

  Once this line of inquiry began, the leading hypothesis held that genes are made of protein. A protein, remember, is a long mo
lecule consisting of different amino acids linked as a chain, their sequence highly variable from one protein to another, and such variation offers vast possibilities for encoding biological traits as linear information. (You heard about this back when I discussed Francis Crick and his 1958 paper, suggesting the use of proteins to chart phylogeny.) In fact, “vast possibilities” is an understatement. Start with the twenty amino acids of life; mix and arrange them in every possible way to make a molecule that’s, say, three hundred amino acids long; do the math; and you get a gazillion possible sequences. That’s enough for genetics.

  An alternate hypothesis was that nucleic acids might be somehow involved. Nucleic acids, which we know as DNA and RNA, comprise one of the four major categories of molecule in living creatures—the other three being fats, carbohydrates, and proteins. But nucleic acids just didn’t seem to have enough chemical complexity and variability to serve as an all-purpose alphabet.

  DNA, for instance: its structure wasn’t known in the early twentieth century, but its components were, and those components—the sugar ribose, a bit of phosphate, and the four bases designated by their initials, A, C, G, and T—didn’t appear to offer a gazillion structural options. In fact, well into the century, the prevailing supposition about DNA’s structure was radically simplistic. It presumed that all four bases were present in equal amounts and repeated themselves in a fixed sequence, such as ACTG-ACTG-ACTG-ACTG, on and on. That’s why DNA was considered a “boring molecule” or a “stupid molecule” by some of the smartest whips in biology, even into the 1930s and 1940s. DNA was underrated, mistakenly assessed, like Albert Einstein as a sixteen-year-old high school dropout, when his father urged him to buck up and become an electrical engineer.