The details can be mind-numbing, but the shape of the process is important: from egg-polarity genes to gap genes to pair-rule genes to segment-polarity genes, and we still aren’t ready to build the fly. The lifespan of all of the proteins coded by these control genes is brief, but they turn on genes for the more permanent Hox proteins, and thus permanently mark the position of cells in the developing animal embryo.
Similar processes subsequently lay out compartments at finer and finer levels of the fly. For example, as a wing is built, the front, back, top, and bottom are marked by control genes, sometimes the same control genes that earlier marked various regions of the entire embryo. But now, working in a defined region of the developing animal, they mark the divisions and edges of the subcompartment. Remember, individual control genes don’t by themselves embody the instructions to build a wing—they just mark areas of the fly, and signal other genes to turn on or off.
This short description leaves out many, many known details of the developmental process, including other means of cell-cell communication and the mechanics of how a signal is received and interpreted. But it at least gives a taste of how the body plan of a simple organism is set in motion.
FLY BOY
The discovery of master regulatory molecules such as Hox proteins that controlled whole body sections of Drosophila was surprise enough. But researchers were absolutely astounded when the proteins were compared with those of distantly related organisms such as, say, people. Every Hox gene seen in the fruit fly has a very similar counterpart in humans! The similarities went well beyond the amino acid sequences of the proteins. The human counterparts even controlled the development of analogous sections of the developing human embryo. That is, the human counterpart to the fruit fly gene that controls the growth of insect head parts directs construction of regions near mammals’ heads (the genes of all mammals are similar to those of humans). The tail end of humans is built under the direction of the mammalian counterpart of the master fly regulatory gene that directs the arrangement of the insect’s hindquarters. Even more strangely, as with the fly, the genes in mammals were still lined up with body segments, with the leftmost gene coding for head regions, middle genes for middle-body regions, and rightmost gene for tail-end sections.
It seemed that life was imitating art with a vengeance. In the 1986 remake of the classic horror movie The Fly, a scientist accidentally mixes his DNA with that of the insect, and over the length of the film slowly and dramatically turns into a fly. The discovery that humans and Drosophila share the same master regulatory genes conjured visions of a person under a full moon sprouting wings and antennae.
The spooky dreams took a step toward reality in the 1990s when the Swiss biologist Walter Gehring isolated the corresponding gene from the fruit fly that was known to affect the development of eyes in vertebrates. Using clever lab techniques, he inserted a mouse gene into different spots in a developing fly embryo, and eyes grew in those spots!6 There were eyes on antennae, eyes on legs. To everyone’s relief, the eyes at least weren’t mouse eyes—they were the regular compound eyes of an insect. This re-emphasized that the master regulatory genes are simply switches, turning on the cellular hardware that does the actual construction of the organ. Just as the same kind of light switch can be used to turn on either an incandescent light or a fluorescent light, whose structures and mechanisms are considerably different, the eye gene just switches on the construction program in an animal that builds an eye.
Nonetheless, the result vividly brought home two points. First, animals as disparate as mice, flies, and worms rely to a very surprising degree on similar developmental programs that use similar components. (As a great practical benefit, this makes it possible to study the development of lower animals and use those results to infer biological facts about humans, where such experiments would be morally problematic.) Second, genetic programs to build organs such as eyes, limbs, and body segments seem to occur in discrete modules. After all, it took just one gene on some fly’s leg to trigger the building of an eye where it shouldn’t be. The rest of the genetic program clearly was already there, waiting to be activated.
This finding has two implications for Darwinism. First, it offers yet more confirmation of common descent. If mammals and flies use the same switching genes, it is reasonable to think that they inherited them from the same ancestor or ancestors. Second, it is possible for single mutations to have very large effects on animal bodies, rearranging whole regions in one fell swoop. So if under some odd circumstance it would be beneficial for a fruit fly to have an extra pair of eyes on its antennae, the eyes wouldn’t have to be built from scratch, one tiny mutation at a time, first changing one protein in the antenna to something like rhodopsin, then changing another protein to start to form a lens, and so on. Maybe instead the gene for the master eye regulatory protein might by accident simply be switched on in an antenna cell, allowing a mutant animal to form extra eyes in a single generation. Or, less dramatically, perhaps extra legs or wings could be grown on body segments where they normally are missing, or suppressed where they usually occur.
With the discovery of master genetic regulatory programs for animal body modules, it seemed a viable path had opened up around Darwin’s tedious insistence that evolution must always be gradual. Instead of changing letter by letter, now monkeys could rearrange whole chapters at a time. Now random mutation and natural selection could work by leaps and bounds.
MODULATING DARWIN
The recent exciting advances in understanding the genetic basis of animal embryology have helped spark a new field of inquiry dubbed “evolutionary developmental biology” or “evo-devo,” for short. Evo-devo looks both at how animals are built in each generation and at how they might have evolved over millennia. Proponents of evo-devo typically whistle gingerly past questions of how basic cellular machinery may have come about by unintelligent processes at the start. But, given a generic eukaryotic cell that has been endowed with what’s been styled a “tool kit” of regulatory genes, they imagine they can scout a path for mutation and selection to go from such humble creatures as flatworms, past insects and arachnids, up through fish, all the way to cats.
The dominant theme of the new thinking is “modularity.”7 As proponents admit, the concept can be pretty fuzzy. Roughly, a module is a more-or-less self-contained biological feature that can be plugged into a variety of contexts without losing its distinctive properties. A biological module can range from something very small (such as a fragment of a protein), to an entire protein chain (such as one of the subunits of hemoglobin), to a set of genes (such as Hox genes), to a cell, to an organ (such as the eyes or limbs of Drosophila). Some thinkers even apply the concept to mind, art, and culture.8 In the next few sections I’ll concentrate on the sustained discussions of modularity that I think are the most evolutionarily relevant. The bottom line is that, while great progress has been made toward understanding how animals are made, and has revealed unexpected, stunning complexity, no progress at all has been made in understanding how that complexity could evolve by unintelligent processes.
MANY SWITCHES, NO EXPLANATIONS
In Endless Forms Most Beautiful, University of Wisconsin biologist and leading evo-devo researcher Sean Carroll delivers a vivid, enthusiastic, firsthand account of the pioneering work of his own lab and others on fruit flies and butterflies. After discussing the discovery and action of Hox proteins and other regulatory proteins, Professor Carroll concentrates on what he believes to be the key to understanding animal evolution, which is the short DNA regions to which the regulatory proteins bind, which he calls “switches.” Switches can be considered modules that can be placed next to any gene. Because each different kind of regulatory protein has a unique, relatively short (about six to ten nucleotides) “signature” sequence of DNA to which it binds near a gene that it helps to turn on, Carroll proposes that genes can be turned on and off over evolutionary time just by random mutations in the DNA region next to the gene.
The way it might work in evolution is something like the following. Suppose it would be beneficial for a developing structure (say an incipient wing or claw) in some evolutionarily promising creature to have a particular one of the ten thousand or so proteins in its genome turned on or off, or even just turned on more or less strongly (perhaps that would make a protein that on balance would strengthen the appendage). To do so it wouldn’t have to evolve a brand-new protein just for the novel appendage. Instead, the region of DNA near the gene would just have to mutate a few nucleotides to form a switch region that could bind the correct one of the hundreds of regulatory proteins the animal’s genome codes for. When the correct regulatory protein bound, perhaps the gene would be turned on or off—not in the whole animal, which might be damaging, but just in the subset of cells that form the appendage.
That sounds easy enough, and Carroll generally stops the story there. But, since one change surely would not give a different new structure, let us continue thinking along the same lines. Suppose a second protein would help push the process along. Well, then, like the first, just the right one of ten thousand genes would again have to develop just the right one of hundreds of possible switch regions. But what if, in the meantime, it would “help” to break a gene, as in thalassemia, which would occur hundreds of times more frequently than specific point mutations? Or what if a momentarily helpful but disconnected change popped up, as in hereditary persistence of fetal hemoglobin? Or what if a coherent change would require passing through a detrimental mutation, as with chloroquine resistance in malaria? What is the likelihood of those looming brick walls?
The scenarios in Carroll’s book seem persuasive because they focus on a single switch or protein. Like considering just one short sentence (“Call me Ishmael”) of a much longer literary work, zeroing in on just one aspect of a difficult evolutionary problem reduces what in reality is a very rugged landscape to one that apparently consists of a single gentle evolutionary hill. From a broader perspective, however, the evo-devo process looks as if it has as much potential for incoherence—with successive evolutionary steps jumbled and disconnected from each other—as traditional Darwinian schemes.
It turns out that, because the regions they bind are so small, developing a binding site for a regulatory protein is too easy. By chance, any particular six-nucleotide sequence should occur about once every four thousand or so nucleotides.9 Given the enormous length of DNA, there is a great chance that a binding site might already be near a gene. What’s more, the likelihood of having a site that matched five out of six positions—so that only one mutation would be needed to change the last position to make a perfect match—is even better. There should be one of those every few hundred nucleotides. Further, since there are so many regulatory proteins with different binding sites, potential binding sites that are one or two mutations away from binding some regulatory protein or other should be packed pretty much cheek by jowl in DNA.
Several studies have shown that is indeed the case. J. R. Stone and G. A. Wray have calculated that the likelihood of forming a new binding site for a given regulatory protein near a given gene, by random mutation in newborn organisms, is very high, about one in seventy.10 Out of a million individuals in a generation, over ten thousand would have a shiny new site for any given control protein. In one person sick with malaria, for example, there would be ten billion new sites produced in a few days! In other words, an embarrassment of riches: There are so many potential binding sites that it’s hard to conceive how they could be the chief factor determining whether a gene was turned on.11
You might object at this point that I seem to be impossible to please: Mutations are too rare, when we look at chloroquine resistance to malaria, but now they are too common, when we look at the theoretical possibilities for all these genetic switches. Here’s the problem: So many kinds of switches are so common that, if they were the most important factor in determining whether a gene was turned on, the organism would be an incoherent mess. Instead of a fly or sea urchin or frog, a developmental program might at best produce a blob of tissue. In fact, as Stone and Wray explain, many, many other factors besides nucleotide sequence are required to be in place before a gene is activated.12
Remember the pyramid of gene switches, from egg-polarity genes to gap genes to pair-rule genes, and so on. The evo-devo hope is that such overarching control structures provide a way for zillions of simple mutations to toggle switches, making evolution somehow easier. Yet even with the new discoveries, a Darwinian path to the typical very complexly regulated eukaryotic gene would still have to be long and tortuous:
The promoter regions of eukaryotic genes are complex and include approximately a dozen to several dozen transcription factor binding sites. The likelihood of a dozen binding sites evolving simultaneously without selection is infinitesimally small…. We envision instead that complex regulatory systems are the result of long and complex evolutionary histories involving stepwise assembly and turnover of binding sites.13
Modularity was supposed to make evolutionary changes simple—to smooth out a rugged evolutionary landscape. But, except for the unexpected complexity of genes and development, what exactly has changed? How have coherent changes been made easier?
In his review of Endless Forms Most Beautiful for Nature, University of Chicago evolutionary biologist Jerry Coyne is unimpressed by evo-devo claims.
The evidence for the adaptive divergence of gene switches is still thin. The best case involves the loss of protective armour and spines in sticklebacks, both due to changes in regulatory elements. But these examples represent the loss of traits, rather than the origin of evolutionary novelties. Carroll also gives many cases of different expression patterns of Hox genes associated with the acquisition of new structures (such as limbs, insect wings and butterfly eyespots), but these observations are only correlations. One could even argue that they are trivial…. We now know that Hox genes and other transcription factors have many roles besides inducing body pattern, and their overall function in development—let alone in evolution—remains murky.14
In his book Carroll does not actually spell out how a novel structure would be built by evo-devo manipulations. Although he beautifully describes and illustrates fly embryology, he provides no specifics on how particular structures would evolve by random mutation and natural selection. In a typical passage, Carroll speculates about the evolution of insect wings.15 He points to research showing that two certain control proteins found in wings are also found in crustacean gills, and concludes that the best explanation for this is that the organs are homologous—that is, the same body part in different forms in two different animals.
Like myriad biologists before him, Carroll confuses evidence for common descent with evidence for random mutation. Although, as he argues, the occurrence of the same control proteins in crustacean gills and insect wings may point to their common ancestry, it says absolutely nothing about how gills could be converted to wings by a Darwinian process.16 In the same way, although one gene may flip the switch to trigger eye development, that tells us nothing about how unintelligent mechanisms could evolve an eye. Although studies of the genetics of embryology have unveiled breathtaking elegance and complexity, the ruminations of evo-devo proponents have—in my view—contributed little to the understanding of the evolution of complex structures.
THE FACILITATORS
Another new book by stellar researchers that trades heavily on the concept of modularity (which they call “compartmentation”) is The Plausibility of Life: Resolving Darwin’s Dilemma. Authors Marc Kirschner of Harvard and University of California–Berkeley’s John Gerhart pick up where Sean Carroll left off (the jacket carries an appreciative blurb by Carroll and a handful of other high-powered scientists). They, too, recount the work of Monod and Jacob, the role of switches in controlling genes, Hox genes in Drosophila, and fruit fly development. They, too, emphasize that an animal’s body can be subdivided into compartments by master regulatory genes, and tha
t to a surprising extent the compartmentation is the same from fly to mammal. Unlike Carroll, however, they aim to fill in the blanks of Darwin’s mechanism. They write that neither Darwin nor any of his contemporaries had a clue as to the underlying mechanisms needed to generate the variation from which nature could select. The tiny, random changes Darwin envisioned would have been grossly inadequate, think Kirschner and Gerhart. But evo-devo and modularity make random mutation more effective.
They christen their novel proposal “facilitated variation,” signifying the idea that control genes make it relatively easy for organisms to vary in ways that might be evolutionarily helpful. If a complex system can be turned on by one simple trigger, and anything that pushes the button will work, then complexity is not necessarily an obstacle to Darwinism. Any input that flips the master regulatory gene for eye development in Drosophila will turn on the system and build an eye. Because links connecting genetic modules are “weak,” they argue, systems and subsystems apparently can be disconnected, switched around, and reconnected pretty easily. Surely that would generate as much variation as Darwin could ask for. Dilemma resolved.
The Edge of Evolution Page 20