Arrival of the Fittest: Solving Evolution's Greatest Puzzle

by Andreas Wagner

  If regulation matters because it avoids waste, then it should be everywhere. And indeed it is. Think of a metabolism with its hundreds of reactions—lactase catalyzing only one of them—as a sophisticated interconnected network of pipelines. Into this network flow nutrients, out of it flow biomass molecules. Each pipe has a dedicated pump, an enzyme that propels materials through it. A cell can regulate each pump according to its needs. If new nutrients turn up in a patch of soil—a fallen apple, a rotting carcass—the soil bacteria turn up the pumps through which these molecules flow. Once the nutrients are gobbled up, they shut these pumps down. And if more of some nutrients and less of others become available, cells can fine-tune the pumps to the right speed.

  The beta-gal gene is repressed by a regulator, but other genes are regulated in the opposite way: Cells leave them off by default and activate them only when needed—through transcriptional regulators that help rather than prevent DNA polymerase from transcribing a gene.11 And even though the regulation of transcription is the most important kind of regulation, there are many others. Cells regulate how fast they manufacture proteins from transcribed RNA, how active these proteins are, how long-lived, and on and on. This is perhaps the most convincing evidence for regulation’s importance: Life has invented a dozen different ways of doing it.

  Imagine the kitchen of a high-end restaurant. The pantry is well stocked with all manner of vegetables, meats, fruit, fish, cooking oils, spices, and flavorings. There are enough ingredients to create every conceivable dish, from the mundane to the exotic, each one nutritious and delicious. The executive chef wants a complete supply of everything, all the time. One of regulation’s roles in metabolism is that of a penny-pinching manager, requisitioning just the right amount of ingredients, anxious not to squander money on even as little as an extra potato.

  But regulation is much more than that. It also provides the recipes, the kind that call for cooking a cup of beans with two cups of chicken stock and a dash of salt in a 350-degree oven for thirty minutes. Each recipe is a sophisticated gene expression program encoded in the genome. It tells cells just when and how much to make of each protein ingredient in an organism.

  If comparing a blue whale to a soufflé seems offensive, consider the complexity of life’s recipes. Each cell type needs many more ingredients than even the most complex dish, thousands of them, in amounts so finely tuned and so exquisitely timed that not even the most skilled five-star cook could hope to follow life’s recipes. What is more, evolution has untiringly created new recipes that brought forth ever-new dishes, innovations in cells, tissues, and organs, as well as entirely new kinds of bodies that emerge through the shifting and enormously complex patterns of regulation.

  Regulatory recipes are the business of developmental biology, a branch of biology that studies the near-magical process of creating a body from a single cell. It ponders how the cells in a body form not just an amorphous, shapeless blob, but organs like brains, livers, hearts, and lungs in animals, and stems, leaves, roots, and flowers in plants. Each organ has a highly specialized task, and each contains many specialized types of cells. Your heart, for example, includes contracting cells that make the cardiac muscles pump, connective tissue cells that hold them together, and pacemaker cells that coordinate the heartbeat via electric signals, like a drummer beating the rhythm of the oarsmen on a galley. How do these cell types form from a single fertilized egg cell, and why do they form at exactly the right time and place? How does a cell know to become a pacemaker cell rather than a neuron or a liver cell?

  The answer lies in regulation, which guides the development of all organisms. The specialized cells in multicellular organisms get their identity from the proteins they manufacture. Each cell contains complete copies of all our genes, but cells differ in which of these genes they express as proteins.12 Muscle cells express motor proteins, tiny molecular machines that allow them to contract, which is about all muscles do. Cells in your eye make transparent proteins whose purpose is to transmit light and focus it on the light-sensing retina. Cartilage cells express collagen and elastin, proteins that cushion the bones in your joints and prevent them from rubbing against each other.13

  The relationship between specialized cells and distinctive proteins seems a simple one, but while cells do express specific proteins, they don’t do so exclusively. Any one protein, in fact, can be expressed in multiple cell types. The vitreous body of your eyes—the clear gel between the lens and the retina—produces the same collagen as cartilage, the muscle cells in your biceps express the same molecular motors as the heart, and so on.14 What gives a cell its identity is not one molecule, but a molecular fingerprint, a combination of hundreds of proteins that is unique to a cell. And the identity of an innovative cell type is really a new fingerprint, caused by a new pattern of gene regulation.15

  Because identity-shaping genes can be expressed in multiple cell types, they have not just one but multiple on-off switches, symbolized in figure 14 through several small rectangular boxes. Each of these boxes stands for a different word that can bind its own regulator (open shapes). An example is the gene encoding crystallin, a protein in the fingerprint of your eye lens that also helps your eyes focus (more about this gene in chapter 6). No fewer than five regulators, including a protein called Pax6, bind near it and determine its expression.16 Some such regulators bind DNA strongly and also influence transcription strongly, others bind weakly and influence transcription weakly.17 Like a cabinet of counselors who jockey for influence over a king, these regulators jockey for influence over the polymerase’s “decision” to express a gene. Some favor repression, others activation, some are influential, others less so, and the sum of their influences determines gene expression.

  FIGURE 14. One gene, multiple regulators

  And what regulates the regulators? Simple: more regulators. Keep in mind that the regulators of the gene in figure 14 are proteins, and like all proteins, they are encoded by genes that are usually also regulated. The Pax6 regulator of lens crystallin is itself expressed not just in the lens but also in the cornea, the pancreas, and the developing nervous system, where multiple regulators guide its expression. And what about these regulators, how are they regulated? Through other regulators. And their regulators? Through yet other regulators. All these regulators often form daisy chains, regulation cascades like that shown in figure 15.

  FIGURE 15. A regulation cascade

  That would seem enough complexity for one day, but alas, regulation can get much more complicated: Regulators form not just linear chains but complex regulator circuits where regulators regulate each other. Figure 16 illustrates this idea with a regulator circuit of five genes labeled A through E, each symbolized by a rectangle. To keep it simple, the figure no longer shows where regulators bind DNA, but only which regulators regulate each other, through black arrows indicating that a regulator activates a gene, or through gray lines ending in crossbars for regulators that repress a gene. (As a verbal shorthand, we can say that genes activate or repress one another.) The dashed lines hint at yet another complication: Each regulator can twiddle the knobs of other genes—up to hundreds—outside the circuit. The Pax6 gene is a member of such a circuit, and mutations in it illustrate the power of regulator circuits: Defects in the human Pax6 gene can cause blindness through aniridia—a missing iris—as well as an opaque lens and a degenerate retina. The gene has counterparts with similar roles in many animals, including mice, fish, and even fruit flies—despite the great architectural differences between their compound eyes and ours. The fruit fly version of Pax6 is known as eyeless, for the simple reason that flies lacking eyeless form no eyes. But even more striking is what happens when flies get too much of a good thing, when biologists turn on eyeless in unusual places while a fly embryo develops: In that case, flies can sprout eyes on their antennae, legs, and even wings.18

  FIGURE 16. A regulator circuit

  Figure 16 looks a bit like a wiring diagram an engineer might create. This metaph
or, of genes being wired, is useful, even though no real wires run between them. A wiring diagram is a compact way of writing down a circuit’s genotype, the DNA that encodes its regulators and the keywords that each regulator binds.19 It allows you to read at a glance that gene A in figure 16 activates B and E, but represses D, B activates C and D, but is repressed by E and C, D represses C, and so on. Inside a cell these influences create a symphony of mutual activation and repression, where each instrument in the orchestra of genes responds to the melodic cues of others with its own notes, until the circuit reaches an equilibrium—like a polyphonic closing chord—where the expression of circuit genes no longer changes and their mutual influences have reached a balance. At this point of balance, some circuit genes are turned off whereas others are turned on. As a hypothetical example, genes A and C in figure 16 might be on, whereas B, D, and E might be off. This state of a circuit (e.g., “on,” “off,” “on,” “off,” “off”) is the gene expression pattern that ultimately creates a cell’s molecular fingerprint, because each of the circuit genes regulates many other genes.20 This pattern is also the circuit’s phenotype. It is another one of those phenotypes that can’t be perceived directly but needs to be measured with sophisticated instruments. Ultimately, however, it creates the most visible phenotype of all—a body and its shape. And new gene expression patterns are needed to create new, innovative types of bodies.

  Regulatory circuits help build the bodies of organisms as different as the tiny fruit flies that congregate on rotting fruit, the small weedy plant known as the thale cress, and the zebrafish, a striped freshwater fish less than ten centimeters long. None of these organisms is eye-catching, but they have other qualities that make them favorite examples for researching development: They are small and they develop fast, thus allowing us to study many of them in little time.

  One lesson they taught us is that regulation circuits can shape a body incredibly rapidly. The larvae of fruit flies, Drosophila melanogaster, hatch a mere fifteen hours after a fruit fly has laid its eggs, pupating and metamorphosing into adult flies after as little as seven days. Fifteen hours to produce a complex organism that can feed, crawl, and navigate the world—no wonder that thousands of scientists spend their careers trying to figure out how that recipe works.

  A fly’s body has three major parts that are further subdivided into fourteen segments: a head, a thorax with three segments, and an abdomen with eleven more, where individual segments have special tasks like walking or reproducing. They are not, to most people, especially beautiful or elegant: nothing like the perfection of a bird’s wing, or the majesty of a giant redwood. However, those fourteen segments and their special tasks, as distinctive to a fly as a flying buttress is to a Gothic cathedral, or a Doric column to the Parthenon, have produced as much scientific enlightenment as any aspect of the living world. They are still studied by multitudes, from high school biology students to Nobel laureates. That’s because they are full of lessons about regulation, many of which apply to all animals.21

  Before laying an egg, a fruit fly deposits several chemical signals in the egg, tiny snippets of the complicated recipe to form a larva. One of these signals is an RNA copy of a gene called bicoid, from which the egg can make a bicoid protein. (Yes, fly biologists use strange names.) Deposited near the egg’s front, where the head will later form, bicoid disperses through the egg like a small glob of syrup through a glass of water. Its concentration remains highest at the front and decreases toward the rear.

  In addition to bicoid, the mother fly also deposits RNA transcripts of several other genes at the front, and yet other transcripts at the rear, where they spread toward the front. After the mother’s job is done, each region of the embryo is bar-coded with a combination of regulators in amounts unique to that region.

  After a sperm has fertilized the egg, the egg begins to divide into many cells, and the developing embryo synthesizes proteins from the deposited RNA molecules. In any one cell, the amount of each synthesized protein depends on how much RNA the mother deposited nearby.22 These proteins are—you guessed it—regulators that can turn other genes on or off, depending on their amount. If a gene’s activator, for example, is abundant in the front like bicoid, the gene would be turned on by bicoid in the front, but nowhere else.

  Among the many genes these regulators control, some are special, because they encode further regulators, which turn yet other genes on, some of which also express regulators, and so on. What’s more, these regulators then regulate each other. They form a complex regulation circuit with more than fifteen different genes.23 This circuit performs the choreography of mutual regulation I described earlier, creating a pattern of gene expression in which some regulators are expressed and others are not.

  An especially remarkable protein in this circuit is called engrailed. Guided by the circuit’s choreography, it becomes expressed in seven highly regular stripes across the embryo. Seven regions that produce engrailed alternate with seven other regions that don’t, and the stripes of engrailed expression demarcate the nascent fourteen segments of the fly. Engrailed and other regulators then control many further genes that specify each segment’s identity, specifying whether a segment will carry legs, or support wings, or be part of the abdomen.24

  All this and more happens in a matter of hours, but it’s not only their speedy development that made fruit fly embryos a boon to developmental biology: Until segmentation is well under way, the embryo’s cells are not separated by walls. This means that molecules can drift freely through the growing embryo. In most other species, cells are walled off right after fertilization, which makes communication between them more challenging.

  Not impossible, though. The male reproductive apparatus in humans—the penis and scrotum—is an instructive example. When a male fetus is about eight weeks old, clusters of so-called Leydig cells release chemical signals called androgens near the area where sex organs will eventually form. These signals are hormones like testosterone that are crucial for sculpting sexual organs. Released from the Leydig cells, they instruct nearby cells to specialize in forming the penis, the scrotum, and, much later, sperm cells. Once an androgen hormone molecule has been released by a Leydig cell, it drifts through the space between cells, and because its chemical structure allows it to penetrate the cellular membrane, it eventually enters another cell. Inside, an androgen receptor, a special protein that can recognize the hormone’s shape, is already waiting for it. And when the two connect, the receptor protein changes its shape—yet another turning lock. Its new shape allows the protein to bind specific words on DNA and activate nearby genes. The androgen receptor turns on many genes, some of which make regulators that eventually arouse hundreds of genes to create unique cellular identities within the male sexual organs.25

  From flies to humans, hundreds of similar signals crisscross every tiny sliver of tissue during every moment of an embryo’s development. This unimaginably complex communication process instructs cells about their location and fate, like the bicoid-expressing cells that “know” they are near the front. These signals also command cells to divide, move, swell, shrink, flatten, and, eventually, to acquire an identity and shape a body. And they are involved whenever cells acquire new identities that lead to innovative shapes and body plans.

  If we could predict the regulatory dance that shapes embryos from flies to humans, we could predict how organs, tissues, and cells form, and why different organisms have very different body plans. That would be quite a feat. Unfortunately, a circuit’s expression pattern can be extremely complicated, even for a simple circuit like that of figure 16. If A activates B and C represses B, whereas B activates C, and D represses C, it’s not immediately obvious which gene expression pattern would result. And worse, many circuits contain many more genes than this one, dozens of regulators that form a dense filigree of interwoven regulation, complex beyond our mind’s grasp. But there is hope, in the form of computers that can describe a circuit’s choreography through mathemati
cal equations, process these equations in their silicon brains, and predict a circuit’s gene expression patterns.

  One remarkable computational scientist who devotes his life to this task is John Reinitz. I first met John when I was a graduate student at Yale in the 1990s. He was a few years my senior, and what they call a character, chain-smoking filterless cigarettes at a time when smokers were already ostracized, and dressing in ways that made casual Friday look like a formal banquet. He drove a fossilized Volkswagen Beetle, whose rear seats had disappeared beneath a landfill of discarded fast-food packaging. John’s nonconformism surely was an asset in his research, because he swam with bold strokes against the mainstream.

  At the time, many scientists studied fly embryos, but they used computers for little more than writing their research papers. Instead they studied regulation by changing the DNA text of a gene, or manipulating a regulator in the laboratory, and measuring how such changes altered segmentation. Their experiments were productive: Among other things, they established which among thousands of genes formed the fruit fly’s embryo. But their efforts to understand a circuit’s entire expression patterns, one gene at a time, were doomed to fail, because a whole circuit is so much more than the sum of its parts. Today this is widely accepted, but in the early 1990s John’s efforts to emulate fly development in a computer made him an outsider. His work was ignored by many and belittled by some.