Microcosm

by Carl Zimmer

  For the flagellum, Behe offered evolutionary biologists an idea for an experiment to overturn irreducible complexity. “To falsify such a claim, a scientist could go into the laboratory, place a bacterial species lacking a flagellum under some selective pressure, for mobility, say, grow it for 10,000 generations, and see if a flagellum, or any equally complex system, was produced. If that happened, my claims would be neatly disproven.”

  Behe was cross-examined by Eric Rothschild, one of the lawyers for the Dover parents. Rothschild pointed out the inconsistencies riddling his testimony. Behe’s proposal for evolving a flagellum in the lab revealed an indifference to the scale of evolution. A 10,000-generation experiment might last two years, whereas bacteria have been evolving for well over 3 billion years. In a typical experiment a scientist might study several billion microbes. But the world’s population of microbes is inconceivably larger. A microbe’s failure to evolve in a laboratory would offer no evidence of intelligent design.

  While Behe issued absurd demands to evolutionary biologists, he demanded little of himself. He felt no need to offer his own step-by-step account of how an intelligent designer created the flagellum (or when, or where, or why). Intelligent design, he informed the court, “does not propose a mechanism in the sense of a step-by-step description of how those structures arose.” The only feature that Behe needed to find in those structures to call them intelligently designed was the appearance of design. “When we see a purposeful arrangement of parts, we have always found that to be design,” he testified. “What else can one go with except on appearances?”

  This sort of testimony persuaded Judge Jones that intelligent design was scientifically empty. In December 2005, he ruled that Of Pandas and People had no place in the Dover classroom. “The evidence at trial demonstrates that ID [intelligent design] is nothing less than the progeny of creationism,” Jones declared in his decision. He chose the flagellum as an illustration of how seamlessly creationism and intelligent design were connected. “Creationists made the same argument that the complexity of the bacterial flagellum supported creationism as Professors Behe and Minnich now make for ID,” he wrote.

  The Dover trial was a creationist disaster. The Dover School Board members who had brought Of Pandas and People into the school were defeated by a slate of opponents to the policy even before the trial was over. Other intelligent design–friendly board of education members have lost their seats in Kansas and Ohio. Judge Jones’s decision was so thorough that it will probably set a precedent for any future cases on the teaching of creationism in whatever guise it takes.

  Remarkably, though, creationists still love E. coli. Access Research Network, another organization that promotes intelligent design, has plastered its flagellum on T-shirts, aprons, beer steins, baseball jerseys, coffee mugs, calendars, greeting cards, calendars, tote bags, and throw pillows. All these creationist items can be purchased on a Web site. The site declares: “The output of this mechanism is used to drive a set of constant torque proton-powered reversible rotary motors which transfer their energy through a microscopic drive train and propel helical flagella from 30,000 to 100,000 rpm. This highly integrated system allows the bacterium to migrate at the rate of approximately ten body lengths per second. Would you please find out who filed the patent on this thing?”

  The message you’ll actually get on your flagellum apron will be far simpler. Above the picture of the flagellum it reads, “Intelligent Design Theory.” And below: “If it looks designed, maybe it is.”

  THE FLAGELLUM AFTER DOVER

  It was a delicious coincidence that the Dover trial, which brought E. coli’s flagellum to the world’s attention, took place right around the time scientists were starting to get a good look at the flagellum’s evolution. They began to trace the history of its genes by finding related genes both in E. coli and in other microbes. Together those genealogies are beginning to add up to a history of the flagellum—and an illustration of how life can produce a complex trait.

  The most important lesson of this new research is that it’s absurd for creationists to talk of the flagellum. From species to species there’s a huge amount of variation in flagella. Even within a single species different populations of microbes may make different kinds of flagella.

  Flagella vary at all levels, from their finest features to their biggest. Take flagellin, the protein that E. coli uses to build the tail of its flagella. Scientists have identified forty kinds of flagellin in various strains of E. coli, and they expect to find many more as they expand their survey. And from species to species, flagellins vary even more. In 2003, a ship of microbiologists and geneticists trawled microbes in the Sargasso Sea and analyzed their genes. They discovered 300 genes for flagellins.

  These patterns make eminent sense in light of evolution. A single ancestral flagellin gave rise to many new flagellins through gene duplication and mutations. As different species adapted to different environments—from feeding inside the human gut to swimming the Sargasso Sea—their flagellins evolved as well. After E. coli emerged tens of millions of years ago, its flagellins continued to evolve. The variation in its flagellins was probably driven by the need to evade the immune system of its host, which recognizes intruders by the proteins on their surface, such as flagellin. If a mutation makes the outer surface of flagellin harder for an immune system to recognize, it may be favored by natural selection. And just as you’d expect, the most variation found in flagellins in E. coli lies in the parts that face outward. The parts that face inward—and have to lock neatly into the other flagellins—are much more similar to one another. Natural selection does not look kindly on mutations that disturb their tight fit.

  Flagella also vary in other ways. E. coli drives its motors with protons, but some species use sodium ions. E. coli spins its flagella through a fluid. Other species make flagella for slithering across surfaces. Scientists have observed some species of bacteria that can make either kind, depending on what sort of swimming they have to do.

  In 2005, Mark Pallen of the University of Birmingham in England and his colleagues discovered a set of genes for building slithering flagella in an unexpected place: the genome of E. coli. E. coli cannot actually build these slithering flagella, because the switch that turns on the genes was disabled by a mutation. In some strains, scientists have found all forty-four genes necessary for building all the parts of the slithering flagellum—its hooks, its rings, its filament. In other strains, some of the genes have disappeared entirely. In K-12 only two badly degraded genes remain.

  Pallen’s discovery makes ample sense if flagella are the product of evolution, and it makes no sense at all if they are the result of intelligent design. A complex feature evolves and is passed down from ancestors to descendants. In some lineages it falls apart. Darwin described many rudimentary organs, from the flesh-covered eyes of a cave fish to the stubby wings of ostriches. If natural selection no longer favored their use, Darwin argued, individuals would be able to survive well enough even if the organs no longer served their original function. E. coli carries vestiges as well, like ancient passages hidden in a palimpsest.

  E. coli also carries clues to how its flagellum evolved in the first place. As Kenneth Miller pointed out in the Dover trial, the needle that delivers flagellin across the microbe’s membrane corresponds, protein for protein, to the type III secretion system for injecting toxins and other molecules. The resemblance speaks to a common ancestry. The type III secretion system is far from the only structure that is related to parts of flagella. Proteins in the motor, for example, are related to proteins found in other motors that E. coli and other bacteria use to pump out molecules from their interior.

  Scientists are now developing hypotheses from this evidence to explain how flagella evolved. Pallen and Nicholas Matzke, now a graduate student at the University of California, Berkeley, offered one hypothesis in 2006. Before there were flagella, Pallen and Matzke argued, there were simpler parts carrying out other functions. Gene duplica
tion made extra copies of those parts, and mutations caused the copies to be combined into the evolving flagellum. Today flagella serve one main function: to swim. But their parts did not start out that way.

  The flagellum’s syringe may have begun as a simple pore that allowed molecules to slip through the inner membrane. A proton-driving motor became linked to it, allowing it to push out big molecules. This primitive system may have allowed ancient bacteria to release signals or toxins. Two kinds of structures eventually evolved from it: the type III secretion system and the needle that injects pieces of the flagellum across the membrane.

  The next step in the evolution of flagella may have come when the needle began squirting out sticky proteins. Instead of floating away, these proteins clumped around the pore. Bacteria could have used these sticky proteins as many species do today, to allow them to grip surfaces. The microbes added more proteins to produce hairs, which could reach out farther to find purchase.

  In the next step, this sticky hair began to move. A second type of motor became linked to it, which could make the hair quiver. Now the microbe could move. Its crude, random movement may have allowed it to disperse during times of stress. Over time this protoflagellum became fine-tuned. Gene duplication allowed the proteins making up the filament to become a flexible hook at the base and stiff, twisted fibers along the shaft. And finally bacteria began to steer. One of their chemical sensing systems became linked to their flagella, allowing them to change their direction.

  This hypothesis is not the unveiling of absolute truth. Scientists don’t have that power. What scientists can do is create hypotheses consistent with previous observations—in this case, observations of the variations in flagella, the components that play other roles in bacteria, and the ways in which evolution combines genes for new functions. Pallen and Matzke’s hypothesis may well prove to be flawed, but the only way to find out is to search the genomes of E. coli and other microbes for more clues as to how the flagellum was assembled, to study how intermediate structures work, and perhaps even to genetically engineer some of the intermediate steps that have disappeared. A better hypothesis may emerge along the way. But it is a far superior hypothesis to one built on nothing but appearances and a personal sense of disbelief.

  NETWORKS UNDER CONSTRUCTION

  In order to build a flagellum, E. coli does not simply churn out all the proteins in a blind rush. It controls the construction with a sophisticated network of genes. Only when it detects signs of stress does it switch on the flagella-building genes, and it uses a noise filter to avoid false alarms. It turns the genes on step by step as it gradually builds up the flagellum, then it turns them off. And like the flagellum, E. coli’s control networks have an ancient history of their own.

  In 2006, M. Madan Babu, a biologist at the University of Cambridge, and his colleagues published a major investigation of how E. coli’s circuitry evolved. They began by searching for E. coli’s genetic switches—the proteins that grab on to DNA and turn on, turn off, or otherwise influence other genes. They ended up with more than 250 of them. They then combed through the scientific literature to figure out which genes these switches controlled. All told, Babu and his colleagues mapped a dense web of 1,295 links joining 755 genes.

  The map Babu’s team drew looks a lot like the hierarchy of a government or a corporation. A few powerful genes sit at the top, each directly controlling several other genes. Those middle-manager genes control many other genes in turn, which may control still others. This organization allows E. coli to cope with changes in its environment with swift, massive changes to its biology. Babu’s map also let him survey E. coli’s network down to its smallest circuits.

  Once Babu had finished his map of E. coli’s network, he could reconstruct its history. He compared it with the networks in 175 other species of microbes. Babu discovered a network core shared by all of them, made up of 62 genetic switches controlling 376 genes, for a total of 492 links. This core, Babu concluded, existed in the common ancestor of all living things.

  This core network offers some hints of what that common ancestor was like. It already had sensors, which allowed it to detect different kinds of sugar and monitor its own energy levels. It could detect oxygen, not to breathe it—since the atmosphere was nearly oxygen free—but probably to protect itself from its own toxic oxygen-bearing waste. This ancestral microbe was already using genetic switches to control iron-scavenging genes, to create the building blocks for proteins and DNA. It was, in other words, a fairly supple little bug.

  From that common ancestor every living thing today evolved. Along the way its network evolved as well. The lineage that led to E. coli gained new circuits to sense and feed on new sugars, for example. Experiments on living E. coli have helped shed light on how mutations and natural selection rewired its network. One of the simplest means by which E. coli’s network can be rewired is the accidental duplication of a chunk of DNA. In some cases, the duplication may create two copies of the same switch. If the gene for one of those switches mutates, it may begin to control a different gene. In other cases, extra copies of genes created by duplications are controlled by the switch that turned on the original gene.

  The ancestors of E. coli rewired their networks as they adapted to new ways of life. Sometimes only minor tinkering with a circuit would produce an important adaptation—adding an extra switch to a gene, for example, or taking one away. One of these tinkered circuits allows E. coli to sense a drop in oxygen and switch its metabolism over to oxygen-free pathways. It is almost identical, gene for gene, to an oxygen-sensing circuit in Haemophilus influenzae, a species of bacterium that infects the bloodstream. In H. influenzae one switch turns on two genes, which then activate all the other genes required to shift the microbe to an oxygen-free metabolism. It’s a fast circuit, which suits H. influenzae well since it lives in the blood and experiences rapid drops in oxygen as it moves from arteries to veins.

  E. coli, on the other hand, does not make snap decisions about oxygen. Living in the relatively stable environment of the gut, it does not experience the sudden, long-term drops in oxygen that H. influenzae does. A slight fluctuation might be a false alarm, which would cause E. coli to invest a lot of energy making new enzymes that would be of no use. And that fact of life is reflected in E. coli’s oxygen circuit. It is identical with H. influenzae’s circuit but for one extra gene, called NarL:

  In H. influenzae, Fnr immediately switches on FrdB and FrdC. But in E. coli those genes also need a signal from NarL. It takes time for Fnr to drive the level of NarL high enough to give the two genes both the signals they need. A minor dip in oxygen won’t provide them with enough time to prime the pump.

  As E. coli’s network evolved, it became impressively robust. The growth of man-made networks offers some clues to how that happened. The Internet did not suddenly appear one morning, ready to send your e-mail anywhere in the world. It began in 1969 as a crude link between computers at the University of California in Los Angeles and the Stanford Research Institute in Palo Alto, California. Other institutions joined the network over the years, and more links were added between them. The Internet became robust thanks to its overall architecture. But no one wrote down the design specifications for the entire Internet in 1969. They emerged along the way. Computer engineers focused their attention on how well each small part of the network performed. They worried about the cost of long-range connections between servers, and so they kept the links short.

  E. coli’s network grew in a similar way. As genes were accidentally duplicated, the network grew more complex. Mutations rewired some of the new genes so that they interacted with other genes. Natural selection then selected the favorable mutations and rejected the rest. As efficient small-scale components evolved, a robust network emerged as a by-product.

  At the Dover intelligent design trial, creationists revealed a fondness for analogies to technology. If something in E. coli or some other organism looks like a machine, then it must have been designed inte
lligently. Yet the term intelligent design is ultimately an unjustified pat on the back. The fact that E. coli and a man-made network show some striking similarities does not mean E. coli was produced by intelligent design. It actually means that human design is a lot less intelligent than we like to think. Instead of some grand, forward-thinking vision, we create some of our greatest inventions through slow, myopic tinkering.

  FIRST WORDS

  Scrape away E. coli’s new genes—the arrivistes carrying resistance to penicillin and other drugs. Peel back the older genes that E. coli evolved after splitting off from other bacteria millions of years ago. Strip off the deeper layers, the ones that build E. coli’s flagella and the ones that have been destroyed beyond use. Strip away the genes for its peptidoglycan mesh, its sensors for rewards and dangers, its filters and amplifiers. Get rid of the genes that encode the proteins that were carried by the last common ancestor of all living things some 4 billion years ago.

  You are not left with a clean sheet. A scattered collection of enigmatic chunks of DNA remains. These are not typical genes. E. coli uses them only to make RNA, and that RNA is never used to make proteins. These RNA genes are the oldest level of the palimpsest. Scientists suspect that they are vestiges of some of the earliest organisms that existed on the planet, from a time before DNA.

  Life’s raw materials are no different from lifeless matter. Stars made the carbon, phosphorus, and other elements in our bodies. If you travel the solar system, you will encounter meteorites and comets with ample supplies of amino acids, formaldehyde, and other compounds found in living things. The Earth incorporated many of these molecules as it formed 4.5 billion years ago, and showers of space dust and the occasional impact of a bigger hunk of rock or ice brought in fresh supplies. The planet acted like a chemical reactor, baking, mixing, and percolating these molecules, probably producing still more molecules essential to life before life yet existed. The great mystery that attracts many scientists is how this reactor gave rise to life as we know it, complete with information-encoding DNA, its single-stranded counterpart RNA, and proteins.