A watch clearly has an artificer—because a watch is a brittle design. It performs one specific task, it only works if all of its parts work, and it cannot cope with any influence of chance or chaos. In short, a watch is not really that complicated. As a matter of fact—apologies to my watchmaker father—watches are Tinkertoys compared with even the smallest organelle of a cell. On the other hand, many natural entities, which even Paley would have excluded from being designed, are vastly complex: stars, planets, mountains, volcanoes, weather patterns, and, yes, even pebbles.
Paley was similarly misguided on the topic of reproduction. He mused about the possibility of one day finding that the watch could make another watch, “similar to itself.” Would that not be proof that the creator was even more sophisticated than we thought? The watch’s ability to reproduce would only add to the complexity of the watch and would therefore make it even more likely that an artificer had created it. But this reasoning is flawed: If a watch could make another watch, it clearly would not need a creator. A watch would simply be the result of another watch. And if the new watch, as Paley argues, were merely “similar” to the original one, then why couldn’t the new watch be a little bit better than the old one? Interestingly, Paley introduces chance into the argument when he says that the new watch is similar, but not identical. How similar? What determines what is the same and what is different in the offspring? What if there were millions of such watches, reproducing and exchanging information about their construction, passing every improvement to the next generation— would these watches not improve over time?
Before you object that watches don’t have babies, let me remind you that I am simply following Paley’s argument. Obviously, a watch that could reproduce itself would stop being a watch. A watch is an artifact made for an external purpose—to tell the time. Once the watch starts reproducing itself, it acquires an internal purpose—efficient reproduction. If we still had an external agent who selects watches for how well they tell time and only lets good timepieces reproduce, they would, over the generations, become better and better watches. But in the absence of an external agent, the watches would stop just being watches, because efficient reproduction would become their new raison d’être. After some time, they would radiate into many different machines—based on which could reproduce the best.
Now let’s go one step further. Nothing works in the absence of energy. A watch needs to be wound up to work. A living organism must eat. Thus, reproducing watches would need to take in energy. They would need to find ways to beat competing watches in the race to sequester enough energy to reproduce. This would require them to find new ways of making a living. Biologists call these roles niches. Before long, we would not recognize our watches anymore. Few would still tell time. Their wheels would now be used for digestion or locomotion. The hands and the dial would be used for attracting a suitable partner with which to exchange information. Maybe, a glow-in-the-dark hour hand would drive the opposite sex wild. This is evolution—this is life.
Evolution
How do molecules evolve? Despite the histrionic debates in various American school boards, the mechanism of evolution is, as we saw in Chapter 1, quite obvious when contemplated with an open mind. This observation prompted Darwin’s supporter Thomas Huxley to lament his not having thought of it first. Molecules are subjected to the same natural selection that applies to the more macroscopic parts of an organism. As a matter of fact, the evolution of proteins is a good way to see how evolution works, because there is a direct relationship between the protein’s amino acid sequence and the information encoding this sequence in DNA. Every molecular innovation—every new molecular machine that transports cargo a bit faster or makes fewer mistakes when transcribing DNA— will give an advantage to the organism it inhabits. Consequently, better molecular machinery will become more prevalent in a population. Or when conditions change, new machinery will emerge to deal with the changed conditions.
A famous example of the evolution of proteins was the 1975 discovery of a strain of Flavobacterium in a wastewater pond at a Japanese nylon factory. The bacteria in this pond had evolved to eat chemicals associated with nylon manufacturing—chemicals that do not exist in nature. On further investigation, researchers isolated three enzymes that had evolved inside these bacteria and that helped the bacteria break down nylon. None of these enzymes existed in Flavobacterium strains that were not raised in the nylon pond. How did the bacteria invent the new enzymes? Bacteria multiply very fast and exist in large numbers. Therefore, they can evolve very rapidly. In this case, the DNA replication machinery of a few Flavobacterium cells apparently made a mistake. The machinery read off a DNA sequence from the wrong starting point, leading to a so-called frame-shift mutation. It so happened that the resulting protein was helpful in breaking down nylon, which is helpful when you live in a pond full of the stuff. Is such serendipity really believable? Absolutely. Just consider that a human body contains 1014 (a hundred thousand billion) bacteria. The Japanese pond must have contained much more than that. A typical time for bacteria to multiply is twenty to sixty minutes. Assuming the slower time and assuming that the nylon factory was in production ten years, the bacteria would have gone through 87,600 generations of gazillions of bacteria. Considering this enormous number of bacteria and the many generations they pass through, a rather unlikely mutation now moves into the realm of definite possibilities. But we are not done yet: The first enzyme may not have been good at digesting nylon, but a bad nylon-digesting enzyme was certainly better than none at all. Once the bad nylon-digesting enzyme spread through the bacterial population, it evolved and improved rapidly.
The glacial, step-by-step, and somewhat unpredictable process of evolution makes it difficult for people to believe that this mechanism could have led to the sophisticated machinery of our cells. But as the above example shows, sometimes evolution happens in a few years. Remarkably, the lion’s share of the history of life (almost three-quarters of life’s history, or three billion years) consisted of the evolution of single-celled organisms. Multicellular organisms only appeared in the last billion years. Why did it take so long for multicellular life to appear? When we look at the complicated machinery of our cells, an answer suggests itself: It took billions of years of evolution to turn the first primitive enzymes into our modern sophisticated cellular machinery. Multicellular organisms became possible only when a minimum degree of efficiency and sophistication was reached. This view of life’s early evolution is supported by the observation that on a fundamental level, all multicellular animals (and all plants) are the same. Humbling as it may sound, at the nanometer scale little distinguishes a human from a fungus. The basic cellular toolkit is the same. The complexity of this kit justifies the length of time it took for it to develop. Once the toolkit was in place, evolution was free to create ever more amazing multicellular creatures, from octopi to redwood forests. In some sense, the real mystery of life lies at the molecular scale. This is where all the real work of evolution was done. The rest is icing on the cake.
The fossil ancestors of our molecular machines are, for the most part, gone forever. Proteins do not keep for over three billion years, and bacteria with primitive machinery would have been eaten a long time ago. Even the so-called archaea microorganism, which have been found to be significantly different from bacteria, are not really archaic. In some sense, bacteria and archaea are more evolved than we are. After all, they had a lot more time, and they reproduced much faster. With this in mind, is there anything that can be done to determine how molecular machines may have evolved?
Trying to figure out the exact evolutionary steps leading to the ribosome or a kinesin molecular motor is like trying to solve a crime hundreds of years after it happened. Who was Jack the Ripper? It is impossible to tell. The trail has gone cold. Yet, using the few reports and other scant evidence that remains, we can make some plausible arguments about what kind of person he may have been. In much the same way, when it comes to the evolution
of molecular machines, we have to look at the few remaining clues and try to come up with a plausible story. In this case, plausible means that the story matches the evidence and is in accordance with known physics and chemistry. Once a plausible story has been hypothesized, parts of it can be tested in the laboratory. If it passes these tests as well, we end up with a likely story, but we will never get a proven story. Jack will remain at large.
An instructive example is the evolution of the ribosome. In Chapter 7, I suggested that RNA is believed to be a more ancient molecule than either DNA or proteins. In the chicken-and-egg problem of what came first—DNA, which encodes protein, or proteins, which are needed to read the DNA—the answer is clearly neither. RNA contains information and can catalyze reactions. It is a kind of egg on feet, which can lay its own eggs. No chicken needed.
RNA’s ability to catalyze reactions is a fairly recent discovery. In 1982, Tom Cech and coworkers at the University of Colorado–Boulder discovered that a certain RNA strand in a bacterium was able to splice parts of itself and reconnect the RNA strand, without any protein-based enzymes. This was the first indication that RNA could act as an enzyme. It took another ten years before Harry Noller and his group at the University of California– Santa Cruz demonstrated that the RNA in the ribosome also has catalytic properties. Over the years, it became clearer that all the hard work in the ribosome is done by RNA. When researchers removed the protein components, the RNA was still able to process messenger RNA and produce an amino acid chain, although at some loss to efficiency and fidelity.
The finding that the ribosome needed its RNA, but not its proteins, suggested that catalytic RNA may have been the basic constituent of early life. In a paper in 2010, researchers from the Weizman Institute of Science in Rehovot, Israel, and the European Molecular Biology Laboratory in Heidelberg, Germany, suggested that the RNA pocket where the amino acid chain is assembled is universal in all ribosomes and may constitute the original ribosome precursor. If RNA really was first, and it could catalyze its own evolution through splicing and reshaping, it may have eventually hit on a structure that could produce proteins. After that, proteins that assisted the primitive organism by being better catalysts than the RNA could have formed. This first protein-producing RNA may not have been able to create well-controlled protein products, but less control would have also led to more mutations and possibly faster evolution. Eventually, the combined forces of RNA and proteins invented DNA, and the modern cell was on its way. It is a likely story.
Another way to consider the evolution of molecular machines is to look at family relationships. In the previous chapter, we mentioned that kinesin and myosin share many similarities in their motor domain. This suggests that they may have evolved from a common ancestral protein. Myosin and kinesin share certain loops in their switch domains, which are associated with shape changes upon binding of ATP. Intriguingly, they also share these structures with so-called G proteins, which are not machines, but molecular switches. Molecular switches communicate chemical signals from the outside of the cell to the inside. In Chapter 6, we speculated that molecular motors may have evolved from enzymes that could change their shape when they bound a control molecule. This allosteric effect is what makes molecular switches work.
The details of how G proteins connect to kinesin and myosin is lost in the fog of billions of years of evolution. Nevertheless, that motor proteins would have evolved from molecular switches is very plausible, and the relationship with G proteins confirms this idea. The relationship between kinesin, myosin, and G proteins shows that in evolution, similar parts in different molecules can often serve different purposes. The evolution of a sophisticated machine like kinesin does not require that each part be invented from scratch or that all parts come into existence simultaneously. When it comes to evolution, almost anything goes.
Let us imagine two proteins A and B, encoded by certain genes in our DNA. Proteins A and B perform different functions. What if part of protein B could help make protein A work better, or what if the combination of parts from A and B were to create a new protein with a completely new function? No problem. Sometimes, whole protein sequences are translocated in our genome, either through copying errors or by viruses. This can lead to the combination of different proteins and the creation of an entirely new line of molecular machines. An example is the nylon-eating enzyme of Flavobacterium.
Ever since kinesin and myosin came into existence eons ago, they have evolved into many different forms, forming large superfamilies. We have seen an example for such a family tree in Chapter 7. The same is true for almost any molecular machine. Every protein is part of a family of related proteins whose jobs are often quite different, but which are clearly descendants of a common ancestor. Evolution never ends; it is ongoing. Once evolution discovers a new trick, such as a walking molecular motor, it soon creates many variants, all fulfilling specialized functions.
A common objection of creationists is that some biological structures are “irreducibly complex.” What they mean is that a structure has many interdependent parts, so that if you remove just one, the whole thing could not work. For example, how could a car evolve? The engine could not evolve without already having a whole car in place. But the car could not evolve without an engine. All the parts of a car must be designed to fit together. No part can be left out. Thus, goes the argument, molecular machines must be designed, just as a car is designed. This is because (following their argument), a molecular machine is only functional when all the parts are in place. There can be no intermediate evolutionary steps. Every previous version of the molecular machine—without all the necessary parts in place—would have been utterly useless.
There are a number of problems with this superficially persuasive idea. First, as we have seen, structures are often put together from parts that previously served a completely different purpose. Take the car example. Clearly, different parts of the car can be developed independently of the whole car. An engine can drive a stationary machine. The Cardan shaft of today’s automobiles, as we saw in Chapter 2, was invented for a water pump. Pistons come from air pumps. Gears from watches. And so on. There are countless examples of such versatility in evolution. The bones in our middle ear evolved from a jaw bone of an ancestral amphibian. The evolution of motor proteins from molecular switches, like G protein, is another example.
The second problem with the irreducibility argument is that incomplete structures are not as useless as one might think. Take the eye. Is an eye without a lens really useless? It sure beats no eye at all. Almost all intermediate stages of eyes exist in nature, from mere light-sensitive spots of some microorganisms to the sophisticated eyes of mammals. The same applies to motor proteins. We have seen that two-legged motor proteins can be processive. However, one-legged motor proteins can work as well, although with much less efficiency and highly reduced processivity, using a pure Brownian ratchet mechanism. Indeed, as we have mentioned, there are one-legged kinesins—although the jury is still out as to whether they can pair to form a two-legged kinesin. Nevertheless, at least theoretically, there is no physical reason why such a one-legged kinesin would not work. It may not be as good as kinesin-1, but it’s better than no kinesin at all.
Many biologists consider evolutionary changes of DNA the most important events in the history of life. One proponent of this view is evolutionary biologist Richard Dawkins, famous for his idea of the “selfish gene.” While there is, of course, much to be said in favor of this view, biologists often underestimate the role of physical law. The DNA-centered view therefore emphasizes chance over necessity. This has been exploited by creationists who like to abuse the concept of chance in evolution to claim that evolution is random and that randomness alone could not have created the complexity of life.
If evolution were truly random, the probability of creating just one functional protein would be astronomically small. Calculating such probabilities is a common parlor game among creationists. But these probabilities are irrel
evant. Evolution is not random: It is the collaboration between a random process (mutation) and a nonrandom, necessary process (selection). It is the result of the balance of chance and necessity. This is not unusual—all of nature is the result of this balance. If not, nature would be either a featureless structure that is the same everywhere (if necessity wins) or a random “mush” with no structure at all (if chance wins). The exquisite order and the amazing variety we see in nature at every level—from galaxies to molecules—is the result of the fruitful interaction of chance and necessity. What is the probability of Earth or a pebble? It’s a meaningless question. Similarly, the question about the random assembly of a protein is also meaningless. Evolution is not random.
Another favorite question of creationists is “How did all the information get into DNA?” At first glance, questions about information in DNA seem legitimate. This is because the message in DNA has meaning. It encodes the structure of a protein or regulates the development of an organism. But is meaning the same as information? Information is measurable; meaning is not. Creationists conflate these two terms to suit their own ends. In information theory, a message has more information the more random it is. As we mentioned before, a perfectly ordered message contains little information. AAAAAAA contains no information, while ACTTGATTC contains information. But does ACTTGATTC have meaning?
Life's Ratchet: How Molecular Machines Extract Order from Chaos Page 25