Arrival of the Fittest: Solving Evolution's Greatest Puzzle (15 page)

João computed the answer and quickly found that not one, not two, not three, but hundreds of
E. coli
’s neighbors are viable on glucose. This discovery contained a simple but vital lesson: The uniqueness of this phenotype is but a deeply flawed prejudice.
⁴²
The neighborhood of any one text contains many other viable texts like it. But nothing had prepared us for what came next, when we began to venture further.

João used
E. coli
as a starting point for deep probes of the metabolic library that led further and further away from the starting text. The objective was to learn how far we could travel—hopping from one viable text to a viable neighbor, to the neighbor’s neighbor, and so on—without losing viability on glucose. How radically could a metabolic text be edited without losing its meaning? When João showed me the answer, my first reaction was disbelief. The furthest viable metabolism he found—the one with the highest
D—
shared only 20 percent of its reactions with
E. coli
. We had walked, computationally speaking, almost all the way through the library—80 percent of the distance that separates the furthest volumes—before we were finally unable to find a glucose-viable text by taking a single step.

Worried that this might be a fluke, I asked João for many more random walks, a thousand more, each preserving metabolic meaning, each leading as far as possible, each leaving in a different direction—possible only because the library has so many dimensions. When the answer came back, I was stunned once again. These random walks had led just as far away as the first one. Each of them led to a metabolism that differed in almost 80 percent of its reactions from
E. coli
. They had found a thousand metabolic texts that shared very little with
E. coli
, except that all of them could produce everything a cell needs from the carbon and energy stored in glucose. If we had kept on walking, we would have found even more texts, too many even to count, although later we were able to estimate their number in parts of the library.
⁴³
For example, the number of metabolisms with two thousand reactions that are viable on glucose exceeds 10
⁷⁵⁰
.

The number of texts with the same meaning is itself hyperastronomical. The metabolic library is packed to its rafters with books that tell the same story in different ways.

While we surely had not expected this, our explorations had revealed an even more bizarre feature of the library. The thousand random walks did not end in a few stacks of the library, where texts with similar meanings might huddle in small groups—groups of metabolisms with similar sets of reactions. These texts were just as different from each other as they were from that of
E. coli
—they encoded metabolisms with very different sets of chemical reactions. The library does not have clearly distinct sections, like rooms that separate all texts on history from those on science.
⁴⁴

And even more surprising was what we found when we started new random walks from these texts—as we had done from
E. coli—
and walked toward other texts without ever changing viability. We always succeeded in reaching them, no matter how far away they were from our starting point. Every single time. This taught us that a connected network of paths linking texts with the same meaning extends throughout the library. I call this network a
genotype network
. It might look a bit like the network of straight lines in figure 9, where the large rectangle stands for the metabolic library, and the lines connect neighboring texts (circles) with the same meaning. Pictures like this are wobbly visual crutches—two dimensions instead of five thousand, a handful of texts instead of unimaginably many—but they are all we have to visualize places as strange as this.

In an ordinary public library, you might find biographical information about Charles Darwin in one text on a shelf in the history section, and another in biography. In a large research library using the Library of Congress’s classification system, you might find some such texts in section QH (for “Science: Natural History, Biology”), but others in sections DA (“World History, Great Britain”), GN (“Anthropology”), PR (“English Literature”), and even BL (“Religion, Mythology, Rationalism”). But you would find nothing that resembles the organizing principles of the metabolic library. You would not find a network of meaning-preserving paths connecting the Darwin biography in HM (“Sociology, General”) with another in BT (“Doctrinal Theory”). You would not be able to walk from one book to its neighbor, to the neighbor’s neighbor, and so on, almost all the way through the library, without ever being farther than one book away from another that told Darwin’s life story in different words.

In the metabolic library, though, that’s exactly what a browser can do. Its myriad texts with the same meaning could be like stars in our universe, islands separated by vast expanses of dark empty space. But they are not. You can travel between them on a network of well-lit paths.

Thus far, we had cataloged only the volumes of one subject area—viability on glucose—but many other subject areas exist. There are metabolisms that are viable on ethanol, acetate, and dozens of other fuels. And we mapped them, using the same random browsing strategy: different random walks whose editing steps preserved the phenotype—viability on ethanol, for example—and stopped only when we could walk no further. We did that for eighty different fuels, and each time we saw the same pattern. Viable metabolisms can have very different texts—they share as little as 20 percent of their reactions—and form a vast connected genotype network in the metabolic library.

Emboldened by this general pattern, we began to map metabolisms viable on multiple different fuels like ethanol, glucose, and acetate, able to synthesize all biomass molecules from each of them. (The advantage of this ability is obvious: It permits survival when the supply of any one fuel runs out.) Because that metabolic skill would be more difficult to achieve, perhaps only a few metabolisms might have it, all of them shelved in one corner of the library? We were again proven wrong. We studied metabolisms viable on five, ten, twenty, and up to sixty different fuel molecules. Each meaning-preserving random walk starting from one of them led far away. Even some metabolisms viable on sixty different fuels shared fewer than 30 percent of reactions. And the metabolisms with the same phenotype—countless trillions in each subject category—again formed a connected genotype network.
⁴⁵

At this point, I was close to ecstatic. We had stumbled upon fundamental principles that govern the metabolic library’s organization. First, many metabolisms are viable on the same fuel molecules—it matters little which fuels you choose. Organisms can assemble biomass building blocks in many ways, through many different sequences of reactions. Second, many of these metabolisms are very different from each other, sharing only a minority of reactions. Third, the viable metabolisms we found were connected in a gigantic network—a genotype network. This genotype network reached far through the space of metabolism.
⁴⁶
Each subject area has one such genotype network, and these networks form a densely woven fabric in the metabolic library.

We had accomplished all this with modest means, because our computing power was puny when compared to the number of texts in the metabolic library. We had crudely mapped a world vast beyond imagination. We had crossed an ocean in a bathtub.

Myriad metabolic texts with the same meaning raise the odds of finding any one of them—myriad-fold. Even better, evolution does not just explore the metabolic library like a single casual browser. It crowdsources, employing huge populations of organisms that scour the library for new texts. Every time gene transfer alters the metabolic genotype of an organism, it takes a step through the library. Different readers—billions of them—walk off in different directions to explore the library.

Evolution’s library exploration also differs in another way from how we humans would browse a library. Imagine a hapless organism that steps off the path connecting viable metabolisms by encountering a change—perhaps a gene deletion—that disrupts the metabolic instructions for manufacturing a key molecule. That organism would be dead, courtesy of natural selection. In the metabolic library readers die (and others get born) in an exploration that unfolds over generations.

Viewed from afar, the library’s explorers, from bacteria to blue whales, might appear like giant clouds of dust grains—dwarfed by the library itself—drifting this way and that, from one stack to the next, endlessly meandering swirls of living things that try new combinations of chemical reactions over and over and over again. Some die. Others survive, and pass innovative combinations on to subsequent generations. This churning mass of life is evolution in action.

That action would vanish if genotype networks did not exist. If only one text could confer viability on any one fuel, then all members of a population would have to share that text, crowding around it in the library. Whenever a member stepped aside to sneak a peak at another volume, it would die. If only a few similar texts were viable, the population could only explore a tiny section of the library. But because of genotype networks, evolving populations can explore the library far and wide.

Genotype networks are the first of two keys to innovability. And now for the second: the immense diversity of the neighborhoods where these explorations begin.

Imagine a patch of soil where billions of bacteria can thrive as long as new food arrives occasionally—a leaf blown in from afar, a rotting carcass, or perhaps a ripe apple dropping from a tree. Many molecules in these foods are nutritious, but they would be useless unless one of the microbes had acquired the right combination of enzymes, the right metabolic text for transforming the food into biomass. That very text could be lifesaving once the billions of its soil-mates, all of them hungry, have consumed and exhausted other fuel molecules. This text, a metabolic innovation, could give one microbe a new lease on life.

Even if there were only a hundred fuel molecules, some 10
³⁰
metabolic phenotypes would exist, and finding this text is to find a specific one of them. There is simply no way you could pack 10
³⁰
texts into a small neighborhood of the library. Each neighborhood only has enough space for a few thousand texts, whose meanings can encompass a tiny fraction—one in 10
²⁶
—of all possible phenotypes. It’s as if you borrowed a few volumes at random from the New York Public Library to fill your nightstand and hoped to find
The Origin of Species
among them—almost impossible. But these odds change if a crowd of readers can browse the library along a genotype network that extends far through the library. Because genotype networks are so large, the population could explore thousands of neighborhoods and increase the odds of finding the new lifesaving phenotype.

You may have noticed a hidden premise here: Different neighborhoods must contain different novel phenotypes. Near a volume on photovoltaics, you would find others on medieval French literature, twentieth-century architecture, and Italian cooking, whereas near
another
volume on photovoltaics—in a different part of the library—books on toy trains, World War II, and astrophysics would be shelved. In metabolic terms, you might find metabolisms viable on acetate and ethanol and citrate in one neighborhood, and metabolisms viable on sucrose and fructose in another.

To find out whether this bizarre library organization really exists, we chose pairs of metabolic texts that had the same phenotype (viability on glucose) but that were otherwise very different. The two metabolisms, A and B, were located in different parts of the library—they did not share many reactions—yet both were part of the same genotype network. We then examined the phenotypes of all their five-thousand-odd neighbors, and found that some of them were likewise viable on glucose—they belonged to the same genotype network—while others had lost a critical chemical reaction, which spells death. Yet other neighbors—those we were really interested in—could live on a new combination of fuels, such as ethanol or fructose. For these networks we asked: Do the neighbors of metabolic genotype A—those texts that differ from A in only a single reaction—contain metabolic innovations different from those of the neighbors of metabolic genotype B? If the neighborhood of A contained metabolisms viable on the new fuels ethanol and fructose, would the neighborhood of B contain metabolisms viable on, say, acetate and sucrose?