The Edge of Evolution

by Michael J Behe

  As for a marooned fellow exploring an island, so, too, for biologists probing the hidden corners of life. In the past half century science has made enormous strides in understanding the molecular basis of life. In terms of the island illustration above, in the past few decades science has surmounted that final hill and spied stunning examples of design where it hadn’t been expected, in the cell. For those who don’t rule it out from the start, design is as evident in such sophisticated systems as the cilium as it is for the castaway in the wrecked ship. Once design has been established for such luminous cases, it then becomes a possible explanation for other, less overpowering examples. There will always be hard cases in the middle, but using the same principles as the stranded gent, we can go back and reappraise many features of life on earth. If a cellular feature has some discernible function, and if it seems to be beyond what is biologically reasonable to expect of chance, then with varying degrees of confidence we are justified in chalking it up to design.

  BEYOND MOLECULES

  Design dominates the molecular level of life. But what of higher levels of biological organization, beyond the cell? What of animal body shapes? Mammals versus fish? Individual species?

  9

  THE CATHEDRAL AND THE SPANDRELS

  HOW DEEP GOES DESIGN?

  Up until now we have examined molecular structures and processes and have drawn a tentative line marking the molecular edge of Darwinian evolution. Most protein-protein interactions in the cell are not due to random mutation. Since cells are integrated units, it’s reasonable to view cells in their entirety as designed. But keep in mind that accidents do happen, so there are Darwinian effects, of some degree, everywhere. For example, just as automobiles may accumulate dents or scratches over time, or have mufflers fall off, but nonetheless are coherent, designed systems, so, too, with cells. Some features of cells of course result from genetic dents or scratches or loss, but the cell as a whole, it seems, was designed.

  Now it’s time to look at higher levels of biological organization. There are several major classes of cells, which include the simpler prokaryotic cells of bacteria and the more complex eukaryotic cells of creatures ranging from yeasts to humans. Were just the simpler, prokaryotic cells designed? Could the more complex eukaryotic cells have evolved from them over time by unintelligent processes? In other words, given the simpler, designed cells in the distant past as a starting point, is it biologically reasonable to think that random mutation and natural selection could reach the more complex cells?

  No. Eukaryotic cells contain a raft of complex functional systems that the simpler prokaryotes lack, systems that are enormously beyond Darwinian processes. For example, the cilium discussed in Chapter 5, which contains hundreds of protein parts, and IFT, the system that constructs the cilium from the ground up, both appear in eukaryotic cells, but not in prokaryotic cells. And the cilium isn’t the only difference. As the evolutionary developmental biologists Marc Kirschner and John Gerhart exclaim in The Plausibility of Life, “enormous innovations attended the evolution of the first single-celled eukaryotes one and a half to two billion years ago.”1 The innovations include such fundamental features as sexual reproduction (meiosis and recombination), the organization of DNA into chromatin, and the provisioning of a cellular protein “skeleton.” Of course, the two kinds of cells share a number of similar systems, such as the genetic code. Nonetheless, just as it’s reasonable to view a motorcycle as a different sort of system from a bicycle, because eukaryotic cells contain multiple complex systems that prokaryotes do not, it’s reasonable to view eukaryotes as integrated, designed systems in their own right,

  So design extends beyond the simplest cells at least to more complex cells, which is the biological level of “kingdom.” Does it go further? Although prokaryotes are single-celled organisms, not all eukaryotes are. Eukaryotes include not only single-celled organisms such as yeast and malaria, but also multicellular organisms: plants, and animals from jellyfish to insects to humans. So does design stop at the eukaryotic cell, or does it extend to multicellular organisms? More pointedly, given a generic, designed, eukaryotic cell in the distant past, is it biologically reasonable to think that over time the rest of life developed from it entirely by unintelligent processes? This chapter answers that question.

  Before we begin, I should be clear that the arguments of this chapter will necessarily be more tentative and speculative than for previous chapters, which dealt with molecules and the cell. The reason is simply that, although rapid progress is being made, much less is known about what it takes to build an animal than about what it takes to build a protein machine. No experiments like those of Greg Winter exploring the shape space of proteins have been done to, say, thoroughly explore the shapes of animals. What’s more, to be secure in our conclusions about life—even about large animals—we have to understand the relevant biology at the molecular level. The inflexible fact is that all of physical life is built of molecules, whose intricate interactions make possible such things as plants and animals. Like a computer, whose overall shape is visible to the naked eye but whose basic workings take place in microscopic circuits, animals live or die depending on the workings of invisible molecular machines. So to locate the edge of evolution, we have to understand the molecular differences between levels of life.

  Even just twenty years ago such a project would have been impossible, since little was known then about the molecular basis of animal life. But especially in the past decade an avalanche of information about the embryonic development of higher organisms has exploded into view. The information in hand isn’t yet enough to allow us to draw definitive, quantitative conclusions. Nevertheless, Darwinian defenders have already begun using the new work to speculate freely about how their theory might still be salvaged (at least for higher levels of biology, beyond the cell). An entire field of inquiry has arisen in the past decade, appropriating the spectacular findings of developmental biology for evolutionary theory. It is the Darwinists’ latest line of defense. Yet, as we’ll see, the new work offers further evidence of design, extending up past animal body plans and the major branches of life.

  A MOLECULAR SWITCH

  Although Charles Darwin was a perceptive man, the molecular basis of animal development was hidden from him, as it was from all scientists of his age. When Darwin mused about how a bear might turn into a whale, or a light-sensitive spot into a full-fledged eye, he did so unhindered by knowledge of what would be needed for such transformations to occur. For a century after Darwin died, only inklings of the process arose as biologists investigated life. Reports of misshapen animals with missing or extra limbs or organs titillated scientific curiosity, but the beginnings of genuine understanding awaited the discovery of the molecular foundations of life. Once the molecular structure of DNA was unveiled in the 1950s, some of the necessary conceptual foundation was laid. The fog was gradually lifting; now science understood somewhat more clearly how molecules went about performing the necessary tasks of life.

  A huge breakthrough in understanding how proteins control DNA and life came with the work of François Jacob and Jacques Monod in the 1960s. It was known then that bacteria could digest different types of sugars, including the most common kind, called glucose, as well as another, much less common sugar, called lactose, which is found in milk. Intriguingly, when bacteria were grown in the presence of glucose, they couldn’t use lactose. Only in the absence of glucose and the presence of lactose could they digest the milk sugar. When glucose was missing, the bacteria made proteins that could pull lactose into the cell and metabolize it, but when no lactose was around, the bacteria didn’t make those proteins. This was a very clever trick that made great biological sense, since in normal conditions the bacterium would waste energy if it manufactured proteins that could metabolize only a rarely encountered sugar. The interesting question was, How did the bacteria “know” when to switch on the genes for making the proteins?

  Jacob and Monod discovered a defective mut
ant bacterium that made lactose-using proteins all the time, even in the absence of lactose. It was lacking a control mechanism. The French scientists reasoned that the bacteria contained another, hidden protein, which they called a “repressor.” They conjectured that the repressor would ordinarily bind to a specific sequence of DNA near the genes that generated the lactose-using proteins, switching them off. In the presence of lactose, the milk sugar would bind to the repressor itself, changing the protein’s shape enough to make it fall off the DNA, switching back on the previously blocked genes. Jacob and Monod surmised that the mutant bacteria had a broken repressor.

  Their model turned out to be exactly correct, earned them a Nobel Prize, and blazed the path for understanding how the genetic program contained in the DNA of all organisms is controlled. There are three critical lessons of the Jacob-Monod model, which we now know apply not just to bacteria but to all of life: First, the genes for many proteins in the cell aren’t on all the time—they have to be turned on or off at some point. Second, it is the job of some proteins to control when the genes for other proteins are turned on and off. The control proteins do little else in the cell other than to act as molecular fingers to flip genetic switches. And third, there are regions of DNA—usually close to the genes for the proteins that they control—to which the control proteins bind. The physical association of the control proteins to the DNA regions constitutes the flipping of the switch.

  FIGURE 9.1

  A simple genetic switch. (A) A repressor binds tightly to the control region (c) of a gene, physically excluding the polymerase (which “transcribes” the gene) from binding.(B) An activator (the small shape marked a) binds to the repressor, distorting its shape and causing it to fall off the gene, which allows the polymerase to bind and begin transcription. (For simplicity, the role of glucose and the CAP protein are not pictured.)

  BLAST FROM THE PAST

  Bacteria are one thing, animals another. Or are they?

  The tiny fruit fly Drosophila melanogaster is an unprepossessing creature. Multifaceted eyes stare out from an antennaed head, its body like a horizontal stack of tires chopped into clearly defined insect segments, a pair of wings coming up from one segment, a nubby pair of stumps from another. Yet Drosophila has enchanted biologists since the early twentieth century, when the great geneticist Thomas Hunt Morgan used the flies to establish the chromosome theory of heredity. The fly is so easy to breed in the lab, and its body so visibly divided into discrete regions, that it has long attracted developmental biologists and embryologists curious about how a distinctly shaped animal body is built from a nondescript fertilized egg.

  FIGURE 9.2

  The fruit fly Drosophila melanogaster.(Modified from Plate V in The University of Texas Publication No. 4313: April 1,1943, Studies in the Genetics of Drosophila III. The Drosophilidae of the Southwest, Directed by J. T. Patterson, Professor of Zoology, The University of Texas. Courtesy of FlyBase.net.)

  By crossbreeding a very large number of fruit flies in the 1970s, the Cal Tech geneticist Edward Lewis showed that the DNA in one region of one of Drosophila’s chromosomes contained a number of genes that appeared to regulate the development of different regions of the body of the fly. Curiously, the genes appeared to be arranged on the chromosome in the same order as the segments of the fly that they helped control, ranging from genes controlling development of head parts at the leftmost, genes for the thorax in the middle, and genes for the abdomen at the right. Mutations in these genes sometimes had bizarre effects, including the formation of flies with four wings instead of two, or flies that had legs emerging from their heads where antennae should have been. Such monstrous alterations, which caused different sections of the animal’s body to be mixed up, were dubbed “homeotic” mutations. The important biological point was that one or a few mutations could cause big mix-ups in the body plan of the animal.

  In the 1970s and 1980s the German Christiane Nüsslein-Volhard and American Eric Wieschaus used chemicals to mutate flies, and in a heroic effort studied tens of thousands of different mutant flies. From these they discovered there were more than a hundred genes essential to fly development. Mutations in these genes didn’t cause a fly to just keel over and die. Rather, they caused big mix-ups in the basic shape of its body. In the case of some mutations, whole organs such as eyes were missing. With other mutations, the poor fly embryo had only half as many body segments as usual. Clearly these genes were not ones that coded for ordinary proteins like hemoglobin. Apparently, the genes controlled long chains of events leading to the building of large, discrete chunks of the fly’s body.

  But what exactly were those genes? By the mid-1980s biologists could routinely determine the nucleotide sequence of fragments of DNA. If the piece of DNA was part of a gene coding for a protein (as opposed to “junk” DNA), the amino acid sequence of the protein could be deduced directly from it and compared to the sequences of other proteins. One of the first homeotic fruit fly genes sequenced, in fact, coded for a protein that resembled the bacterial repressor protein that Jacob and Monod studied in the 1960s2

  That was a strong clue that, like the bacterial gene, the fly gene also acted as part of a switch to turn other genes on or off. Surveys of other organisms, ranging from worms to people, unveiled a whole new class of such proteins, all containing a region of about sixty amino acids similar to the repressor protein and very similar to one another. The segment of the genes that coded for the sixty-amino-acid region of a homeotic protein is called the “homeobox.” The proteins are dubbed Hox proteins.

  In subsequent years homeotic proteins, and other classes of control proteins, have proven to be master regulators of developmental programs in animals. Although they resemble the repressor protein that Jacob and Monod discovered decades earlier, in that they bind near a gene to turn it on or off, the regulatory systems of animals are much, much more complex than bacterial systems. The bacterial lactose system was turned on or off by a single protein. In animals, a master switch sets in train a whole cascade of lesser switches, where the initial regulatory protein turns on the genes for other regulatory proteins, which turn on other regulatory proteins, and so on. Eventually, after a pyramid of control switches, a regulatory protein activates a gene that actually does some of the construction work to build an animal’s body. But there’s another complication. A gene in an animal cell might be regulated not by just one or a few proteins, as bacterial genes are, but by more than ten. What’s more, there may be dozens of sites near the gene at which the regulatory proteins might bind, with multiple separate sites for some regulatory proteins.

  TO BUILD A FLY

  Why such enormous complexity, far beyond that of bacterial cells? The reason is that animal bodies contain many different kinds of cells that have to be positioned in definite relationships to other cells, in order to be formed into organs, and to connect to other parts of the body. Cal Tech biologist Eric Davidson emphasizes what the task of building an animal demands:

  The most cursory consideration of the developmental process produces the realization that the program must have remarkable capacities, for development imposes extreme regulatory demands…Metaphors often have undesirable lives of their own, but a useful one here is to consider the regulatory demands of building a large and complex edifice, the way this is done by modern construction firms. All of the structural characters of the edifice, from its overall form to minute aspects that determine its local functionalities such as placement of wiring and windows, must be specified in the architect’s blueprints. The blueprints determine the activities of the construction crews from beginning to end.3

  In other words, the molecular developmental program to build an animal must consist of many discrete steps and be profoundly coherent. As we’ve seen throughout this book, random mutation cannot take multiple coherent molecular steps. Therefore, like a castaway re-evaluating structures on an island in light of the knowledge that some things there were designed, we should already suspect that to some
extent animal forms were designed. But to what degree?

  For a flavor of the careful planning that goes into building even a relatively simple animal, let’s look briefly and sketchily at some of what’s been learned from studies of Drosophila development in recent years.4 The mother fly starts the process off by depositing in the egg, at the end that will become the head, a concentration of the instructions5 to make one kind of protein, called “bicoid,” and, at the end that will become the tail, a second kind of protein, called “nanos.” The bottom of the embryo is marked by the mother fly in a somewhat different way. The genes coding for proteins that specify the sides of the egg (front, back, top, bottom) are called “egg-polarity” genes. Critically, the proteins (or other proteins they affect) can stray in the egg, drifting away from their source; as they do they become more diffuse. As the egg initially divides into many cells, the high concentration of signal protein at one end of the fly turns on one set of control genes, the middle concentration turns on a different set of control genes in the middle portion of the embryo, and the lowest concentration activates a third set.

  Once the front, back, top, and bottom are marked (caution—it’s critical to keep in mind that the signal genes don’t actually form the structures found in those regions of the developing fly; they simply mark the location of a cell, like a surveying crew mapping out land for a construction project), positions are further refined with other control proteins. Several groups of proteins controlled by “segmentation genes” subdivide the embryo further. One group of about six so-called “gap genes” is switched on, marking chunks of segments; if one of these control proteins is defective, several neighboring segments of the embryo will be missing. Oddly, another group of eight genes, called “pair-rule genes,” affect alternate segments. If one of these is broken, a fly embryo will have only half its normal complement of segments. Finally, a group of ten “segment-polarity” genes helps differentiate each segment. Although in a normal fly the front of each segment looks a bit different from the back, in some segment-polarity mutants the two ends look the same.