Life Finds a Way

by Andreas Wagner

  Among these innovators was the late Pim Stemmer, a Dutch biochemist and serial entrepreneur with dozens of patents to his name. Stemmer’s fame in biotech circles comes from his 1994 invention of DNA shuffling, a biochemical technique that uses an enzyme called DNA polymerase to copy huge numbers of DNA molecules that encode one or more genes.

  DNA polymerase is by no means an exotic enzyme, a laboratory creation of biotechnologists. It occurs in every living cell and is essential to making a copy of a cell’s DNA whenever the cell divides. However, biotechnologists use an engineered form of this enzyme and exploit a property that is crucial for DNA shuffling.22 Starting from one end of a DNA string, DNA polymerase glides along the string and copies it letter by letter. While it does, it can hop from that string to another nearby DNA string—biochemists say it switches templates—and continue copying the other string. It’s as if you had two similar English texts side by side, and sometime after you started copying one of them, you moved over to the other text and continued copying that one. The polymerase’s jump makes no difference to the copy’s letter sequence if the two DNA strings are identical, but if they are different, the resulting copy is a chimaera, starting with the letter sequence of the first DNA text and ending with that of the second.

  With DNA shuffling, biochemists can shuffle mixtures of many molecules, each with a different letter sequence. The DNA they shuffle can also be highly diverse, more so than the typical genes of two amorous bacteria—it could come from organisms as different as marmots and marigolds. What is more, while replicating any one of these molecules, polymerase can switch templates multiple times, such that the final copy contains text snippets from multiple DNA strings.

  Think of DNA shuffling as group sex for molecules.

  The long jumps through the adaptive landscape enabled by DNA shuffling proved their mettle when researchers in Stemmer’s laboratory tested the power of DNA shuffling to improve on nature’s handiwork in creating efficient enzymes.23 They focused on one of those enzymes that disarms antibiotics like cefotaxime, starting with a pool of genes from four different bacteria that encoded different variants of the enzyme. Very different variants, I should add: up to 40 percent of the letters in their DNA text differed among them. To begin, the researchers asked how much they could improve the starting enzymes by walking through the landscape one step at a time, changing individual DNA letters one by one. The answer: eight-fold. In other words, the resulting enzyme could cleave eight times more antibiotic molecules in the same amount of time. Not too bad, you might say. But it is nothing compared to what happened when they forced the molecules into a foursome of DNA shuffling. That foursome created an enzyme five-hundred-fold better than its parents. Other experiments with DNA shuffling created enzymes that remove dirt from clothes faster, cleave new kinds of molecules, or detoxify arsenic-laced mining waste.24

  DNA shuffling, promiscuous bacteria, and species hybrids teach us that teleportation in genetic landscapes is crucial to nature’s creative powers. We would therefore expect to find it everywhere on the vast tree of life, and that’s indeed the case. Almost. Some of the tree’s million-plus species apparently do not recombine their DNA. These include some salamanders whose females propagate through unfertilized eggs, and flowering plants whose seeds develop without a pollen grain.25 But what’s telling about these species is that almost all of them form tiny twigs on life’s tree. No major branch of animals or plants reproduces asexually. That observation, as innocuous as it seems, makes a profound statement about the importance of sex. Asexual species did not just lose sex, they lost it recently in their evolutionary history, otherwise larger branches of life’s tree would be sex-free. Species that lost sex many millions of years ago are no longer around. They suffered evolution’s ultimate death penalty: extinction.

  The message could not be clearer: lose sex and you are not long for this world.

  But here is a mystery, an apparent exception to this rule—a tiny fraction of species known as ancient asexuals. They seemingly made do without sex for millions of years. Among them are some three hundred species of tiny freshwater animals called the bdelloid rotifers, which originated more than thirty million years ago.26 While no amount of searching turned up any evidence for hanky panky in these critters, a recent analysis of their genome’s DNA revealed something even more remarkable than their apparent asexuality: more than three thousand genes in their genome are not their own.27 These genes do not even come from other multicellular animals. They have been transferred into their genome from who-knows-where.

  We have no idea how the bdelloid rotifers do it, but they clearly utilize the same kind of horizontal gene transfer perfected by bacteria to leap through an adaptive landscape. In other words, these asexuals are not so asexual after all, even though their teleportation machinery still awaits discovery. And perhaps other ancient asexuals are like them? Perhaps they also are secretly sexual, harboring genomic signatures of unusual sexuality? It’s good to know that in the twenty-first century there are still biological mysteries to be solved and important discoveries to be made.

  Recombination’s near-universality testifies that genetic teleportation has been essential to life’s ascent. But it also raises a vexing question. Why does Captain Kirk always land on the mothership and never in outer space? Why do recombination’s blind leaps rarely end up in some deep valley of the landscape, producing a mashed-up genome and a broken organism? Or perhaps they do, and the teleportation machine kills most of those who enter it? Regrettably, that’s not easy to find out, because the organisms whose genomes leapt off a cliff were never even born, so we cannot examine them.28 But we can do something else. We can use computers to simulate the long-distance jumps of recombining molecules and genomes.

  Different researchers around the world use computers to do just that, and they are coming up with similar answers. Among them is Allan Drummond from the University of Chicago, who has asked where genes land in an adaptive landscape after recombination—closer to a peak or nearer to a valley? More precisely, he has asked whether recombination leaves the protein encoded by these genes unharmed. Likewise, some researchers in my laboratory study recombination in the DNA that encodes the chemical reactions of metabolism. They predict whether a metabolism can still support life after a long-distance jump through its adaptive landscape. Yet other researchers study recombination among the regulators and circuits that help build new bodies. They ask whether recombination leaves the sophisticated recipes for creating whole organisms intact.

  All of these scientists compare the harm from the long-distance jumps of recombination with the harm that comes from traveling the same distance, but like a pedestrian, through many steps of random single-letter changes.29 And they all come up with similar answers: recombination is much more likely to preserve life—up to thousands of times more likely—than random mutation is. To be sure, recombination does have the potential to destroy—just think of those sterile hybrids. However, this destructive potential is much smaller than that of random mutation. It is no match for recombination’s enormous creative potential.

  The reason? When nature recombines genomes, and when biotechnologists recombine molecules, they do not make completely haphazard changes to DNA. Instead, they take organisms or molecules that already work well—we know, because they have survived to this day—and mix up their parts. It’s as if you exchanged the pages of two texts that tell a similar story but in different words. Such recombination will not always improve the text, but it will usually not destroy its meaning completely, and could even create unexpected twists or new plotlines. Not so if you just “mutated” the text through millions of typographical errors. You’d be almost certain to garble its meaning.

  Another way to understand recombination’s potential is offered by the high-dimensional nature of adaptive landscapes and by the spiderwebs of high-altitude ridges. The very existence of these ridges means that long-distance jumps can land in a region of high elevation and preserve adapta
tion. What is required is that they actually do land on a ridge. And recombination’s reshuffling helps them achieve that soft landing because it recombines parts of molecules that already work well together.

  Together with genetic drift, DNA recombination and the sprawl of adaptive ridges counterbalance natural selection’s shortsightedness. They temper selection’s compulsive ascent of the nearest hill in an adaptive landscape. Whereas drift takes modest steps—downhill as much as uphill—recombination causes giant leaps through such a landscape. And adaptive ridges do not only permit a soft landing after recombination, they also permit diversity in a population, which opens new vistas and enables some individuals to ascend new, even higher adaptive peaks.

  Nature has come up with multiple ways to temper natural selection, which tells us how important such tempering must be. As it turns out, the need to temper selection has analogs in the human realm. We have seen hints of it in the mental journey that the physicist Hermann von Helmholtz described using when he solved difficult problems, and we will encounter it again later in multiple forms, such as the meandering lives of eminent creators that allow the cross-fertilization behind scientific revolutions.

  Carnivorous caterpillars, desert sunflowers, and toxin-gobbling bacteria are only a few of the myriad organisms that recombination and drift helped natural selection create. Together with the spiderweb ridges of adaptive landscapes, these mechanisms of evolution are essential for nature’s creative powers. As we shall see next, these powers are so formidable and far-reaching, they even extend to the inanimate world, where shortsighted hill-climbers don’t get very far either.

  Chapter 5

  Of Diamonds and Snowflakes

  Geodesic domes may be the biggest triumph of architecture since the Gothic spire. Buildings like the Montréal Biosphère or Spaceship Earth at Florida’s Walt Disney World are built from a latticework of struts that form a hollow cage prized for its lightness and stability. The name comes from the large circles—geodesics—that revolve around a sphere’s center and that are traced by the struts. Geodesic domes were invented by the German engineer Walther Bauersfeld after World War I, but they became popular only when the American architect and inventor Buckminster Fuller touted them as a solution to the world’s housing problems.1

  These gravity-defying structures could be Exhibit A for human creativity and its unique powers were it not for an annoying little fact: light years away, in the infernal cauldron of ancient stars and nebulae, nature has been churning out miniature versions of them for eons.

  We have to thank chemist Harold Kroto and a team of collaborators for this discovery. In 1985, they were puzzled by data from spectrometers that hinted at the existence of complex carbon molecules with more than a dozen atoms near distant stars and in interstellar space.2 Wondering how such molecules could form in the hostile environment of outer space, they tried to create them in the lab at the high temperatures found near stars. In a now-famous experiment performed at Rice University, they shot a focused laser beam at a piece of graphite, which created infernal temperatures above ten thousand degrees and instantly vaporized the graphite into atomic carbon.3

  As these atoms cooled in a jet of helium, they formed molecules even more complex—and beautiful—than those the scientists had been after. In each molecule, sixty carbon atoms bonded to form a highly regular spherical cage with thirty-two faces—twenty hexagons and twelve pentagons (Figure 5.1). It’s the shape of a regulation soccer ball, or, if you like mathematical jargon, of a truncated icosahedron.

  Figure 5.1.

  In honor of Buckminster Fuller’s geodesic domes, Kroto called these molecules buckminsterfullerenes, but they soon came to be known—easier on the tongue and more affectionately—as bucky-balls. Their discovery would win Kroto and two colleagues the 1996 Nobel Prize in chemistry.4

  Bucky-balls far surpassed the complexity of the interstellar carbon molecules that had prompted Kroto’s experiment. But, as it turns out, they do indeed exist in outer space, even though it took decades to discover them there. In 2010, another team of scientists found that bucky-balls assemble by the trillions in carbon-rich shells of old stars and interstellar nebulae.5 So great are their numbers that they can blot out the light of nearby stars.6

  Nature created the structure of today’s universe long before we came along, and even long before life came along, when the remnants of the Big Bang assembled into atoms, and those atoms assembled into swirling galaxies, whose gas clouds assembled into trillions of suns and even more planets—all by themselves. But nature’s creative power is nowhere more evident than in beautiful molecules like bucky-balls. And understanding how such molecules self-assemble holds important lessons about all creativity.

  Two carbon atoms bonded together in a buckminsterfullerene—or in any other molecule—are a bit like two tiny balls on a spring. When you pull them apart, you need to expend energy. This energy gets transferred to the spring and stored there in a form physicists call potential energy. Think of it as the atoms’ ability—their “potential”—to move closer once you release them. The stronger you pull, the more potential energy they accumulate. And the same thing happens when you push the atoms together: they store potential energy and will release it as soon as you stop pushing and allow the atoms to move farther apart.

  Once the two atoms are left to their own devices, they eventually come to rest at some intermediate distance, where their potential energy—or, more precisely, that of the molecule they form—is smallest. That’s the lowest point on the parabola of Figure 5.2, which describes the potential energy of a two-atom molecule. Push them together, and they move upward—their potential energy increases—along the left wall of the parabola. Pull them apart, and they also move upward—their potential energy increases again—but now along the right wall. The harder you push or pull them, the farther they move uphill, and the more energy the molecule stores.

  Figure 5.2.

  Viewed through a different lens, that parabola is also a simple two-dimensional landscape, so simple that it has only one valley. Chemists call it the potential energy landscape of a two-atom molecule. If you were to drop a marble on the hillsides enclosing the valley, the marble would slide down, roll around near the bottom for a while, and eventually come to rest. The marble’s location in the landscape is analogous to the distance between the two atoms. As the marble slides, the atoms move—farther apart or closer together—until they have reached their resting point of lowest potential energy.

  Physical laws act on linked atoms like gravity acts on a marble. This principle applies to not just two carbon atoms joined by a covalent bond—the kind of strong chemical bond that holds bucky-balls together; it holds for any two atoms and for any kind of physical attraction between them. That includes the attraction between positively and negatively charged ions, like those of sodium and chlorine in table salt. It also includes various kinds of weaker forces that can bond atoms to each other, such as the van der Waals force that supports the three-dimensional shapes of proteins.7 All these forces are variations on the same theme, like balls of different size linked by springs of different stiffness.

  What is more, the same principle applies to more than two atoms. And that’s where things get interesting.

  To describe your place in the two-dimensional landscape of Figure 5.2, you would need to know only one quantity—the distance between two atoms. That information would immediately tell you the elevation you are at. It’s no longer so simple when three instead of two atoms are linked. Three balls can be connected by three springs, which would form a triangle. Each of these springs can be pushed or pulled, and it can store or release potential energy. In other words, you would need three numbers—the length of each spring—to describe how far apart the three atoms are. And you would need a fourth number to describe the potential energy of this atomic configuration. In landscape terms, the first three numbers specify a location, namely that of the three-atom molecule on its potential energy landscape. The f
ourth number specifies the elevation at this location—the molecule’s potential energy. In other words, describing a three-atom molecule already requires a landscape in four dimensions—one more than in our familiar three-dimensional space.

  As the number of atoms increases, the number of springs increases too, just faster. Four atoms connect via six springs, five via ten springs, six via fifteen, and so on—the number of springs increases like the number of possible pairings between an increasing number of tennis players. And so do the energy landscape’s dimensions. A further complication is that atoms are usually not confined to a two-dimensional plane like that of Figure 5.2. In the three-dimensional world of bucky-balls, you would need three numbers to describe the location of each atom. For a bucky-ball’s 60 atoms, 180 coordinates are necessary to describe the location of all atoms. Combine these 180 coordinates with the bucky-ball’s potential energy—one additional number—and you have a landscape in 181 dimensions.

  Whether a landscape has three dimensions or three hundred, its topography could be as dreary as the endless plains of Texas or as simple as the single crater of Figure 5.2. Indeed, up to five atoms, a molecule’s potential energy landscape can be that simple: a single valley in which the atoms form a shape called a triangular bipyramid, the only stable molecule they can form.8 But for six atoms, you get two valleys—two possible stable molecules. For seven atoms, you get six stable molecules; for eight atoms, sixteen stable molecules; for nine atoms, seventy-seven stable molecules; and for ten atoms, 393 stable molecules—each one corresponding to an atomic arrangement where the atoms can come to rest, having expended their potential energy.