Table 1. Symptoms of mice lacking clotting factors.
The probable explanation is straightforward. The pathological symptoms of only-plasminogen-deficient mice apparently are caused by uncleared clots. But fibrinogen-deficient mice cannot form clots in the first place. So problems due to uncleared clots don't arise either in fibrinogen-deficient mice or in mice that lack both plasminogen and fibrinogen. Nonetheless, the severe problems that attend lack of clotting in fibrinogen-deficient mice continue in the double knockouts. Pregnant females still perish.
An important lesson exemplified by Bugge et al. (1996) is that it can be worse for the health of an organism to have an active-but-unregulated pathway (the one lacking just plasminogen) than no pathway at all (the one lacking fibrinogen, which exhibited fewer overt problems). This emphasizes that model scenarios for the evolution of novel biochemical systems have to deal with the issue of regulation from the inception of the system. Most important for the issue of irreducible complexity, however, is that the double-knockout mice do not merely have a less sophisticated but still functional clotting system. They have no functional clotting system at all. They are not evidence for the Darwinian evolution of blood clotting. Therefore my argument, that the system is irreducibly complex, is unaffected by that example.
b. Gene Duplication is not a Darwinian Explanation
I believe that the point about the knockout mice is quite important because it helps illustrate the serious shortcomings of simple invocations of gene duplication as evolutionary explanations. Appeal to gene duplication has been quite common among scientists reviewing my book. Besides Russell Doolittle, it has been invoked by Allen Orr (Orr 1996), Douglas Futuyma (Futuyma 1997), Neil Blackstone (Blackstone 1997), Robert Dorit (Dorit 1997), a committee of the National Academy of Sciences (National Academy 1999), and others. The typical argument goes something like this: Modern biology has recognized that the sequences and structures of some proteins are quite similar to others (an example is hemoglobin vs. myoglobin), and this similarity is normally interpreted in terms of duplication and divergence of an ancestral gene. Many proteins in complicated pathways, such as the blood clotting cascade, are also similar to other proteins, consistent with the idea that they descended from a relatively few ancestor proteins. We can assume, then, (the argument continues) that although we don't know the details, most complicated pathways were built by natural selection using gene duplication.
A recent publication of the National Academy of Sciences nicely illustrates the argument in action:
Modern-day intelligent design proponents argue that . . . molecular processes such as the many steps that blood goes through when it clots, are so irreducibly complex that they can function only if all the components are operative at once. . . .
Complex biochemical systems can be built up from simpler systems through natural selection. . . . Jawless fish have a simpler hemoglobin than do jawed fish, which in turn have a simpler hemoglobin than mammals . . . .
Genes can be duplicated, altered, and then amplified through natural selection. The complex biochemical cascade resulting in blood clotting has been explained in this fashion.
(National Academy of Sciences 1999, 21-22)
But the reaction to Bugge et al. (1996) is a paradox for the typical argument. On the one hand, structural and sequence evidence for gene duplication and domain swapping in the clotting cascade is very clear. So clear, in fact, that cascade proteins are used as textbook examples of those processes (Li 1997). On the other hand, if there were indeed a robust Darwinian explanation for the origin of blood clotting by natural selection, or if sequence analyses had demonstrated how gene duplication might have produced the cascade, it would be difficult to understand why one would point to the knockout mice as exemplifying Darwinian possibilities, when in reality they only underscore the serious problems facing the evolution of irreducibly complex systems. Detailed knowledge of the sequence, structure, and function of the proteins of the clotting cascade did not prevent a wholly unviable model from being proposed as a potential evolutionary intermediate. Why not?
The predicament is easily resolved when a critical point is recalled: EVIDENCE OF COMMON DESCENT IS NOT EVIDENCE OF NATURAL SELECTION. Homologies among proteins (or organisms) are the evidence for descent with modification--that is, for evolution. Natural selection, however, is a proposed explanation for how evolution might take place--its mechanism--and so must be supported by other evidence if the question is not to be begged. This, of course, is a well-known distinction (Mayr 1991). Yet, from reviewers' responses to my book, the distinction is often overlooked. Knowledge of homology is certainly very useful, can give us a good idea of the path of descent, and can constrain our hypotheses. Nonetheless, knowledge of the sequence, structure, and function of relevant proteins is by itself insufficient to justify a claim that evolution of a particular complex system occurred by natural selection. Gene duplication is not a Darwinian explanation because duplication points only to common descent, not to the mechanism of evolution.
c. What Would an Explanation Look Like?
If homology is not sufficient to justify a Darwinian conclusion, what is? The required amount of justification depends on the complexity of the system under consideration. For example, the task of getting from a simple oxygen-binding protein such as myoglobin, with one chain, to hemoglobin, with four chains that bind oxygen, does not appear to present substantial problems, as I discussed in Darwin's Black Box. In both cases the proteins simply bind oxygen, with more or less affinity, and neither globin has to interact critically with other proteins in a complex system. There seems to be a straightforward pathway of association leading from a simple myoglobin-like protein to a more complex hemoglobin-like one. In fact, its relative simplicity is probably the reason it is a favorite example in discussions of evolution by gene duplication.
Like hemoglobin/myoglobin, many proteins of the clotting cascade are similar to each other, and also similar to non-cascade proteins. So they too appear to have arisen by some process of gene duplication. I agree this is a good hypothesis. But does gene duplication lead straightforwardly to the blood clotting cascade? No. The important point to keep in mind is that a duplicated gene is simply a copy of the old one, with the same properties as the old one--it does not acquire sophisticated new properties simply by being duplicated. In order to understand how the present-day system got here, an investigator would have to explain how the duplicated genes acquired their new, sophisticated properties.
With clotting, however, the task of initiating and adding proteins to the cascade appears to be quite problematic. With one protein acting on the next, which acts on the next, and so forth, duplicating a given protein doesn't yield a new step in the cascade. Both copies of the duplicated protein would have the same target protein which they activate, and would themselves be activated by the same protein as before. In order to explain how the cascade arose, therefore, an investigator would have to propose a detailed route whereby a duplicated protein turns into a step in the cascade, with a new target, and a new activator. Furthermore, because clotting can easily go awry and cause severe problems when it is uncontrolled, a serious model for the evolution of blood clotting would have to include such things as: a quantitative description of the starting state, including tangentially interacting systems; a description of the initial regulatory mechanisms; a quantitatively-justified proposal for a step-by-step route to the new state; a detailed plan for how regulatory mechanisms accommodated the changes; and more.
An alternative to presenting an exhaustively detailed model would be an experimental demonstration of the capability of natural selection to build a system whose complexity rivals that of the clotting cascade. In fact, experimental evidence is much preferred to mere model building, since it would be extremely difficult for models to predict whether proposed changes in complex systems might have unforeseen detrimental effects.
I pointed out in Darwin's Black Box that scenarios for the origin of biochemical systems lack essential detail. But since I am a proponent of an alternative explanation, some Darwinists have accused me of setting the evidentiary standard so high that it is impossible for them to meet it. The evidentiary standard, however, is set not by me, but by the complexity of the biochemical systems themselves. If malfunctioning of the blood clotting cascade or other complex system can cause a severe loss of fitness, then a Darwinian scheme for its evolution must show how this could be avoided. And if the system can malfunction when small details go awry, then the scheme has to be justified at least to the level of those details. Unless that is done, we remain at the level of speculation.
In noting that not much research has been done on the Darwinian evolution of irreducibly complex biochemical systems, I should emphasize that I do not prefer it that way. I would sincerely welcome more investigation of their supposed Darwinian origins. I fully expect that, as in the field of origin of life studies, the more we know, the more difficult the problem will be recognized to be.
III. Kenneth Miller's Criticism
a. "From that point on . . ."
If an eminent scientist and expert on blood clotting such as Russell Doolittle does not know how blood clotting arose, nobody knows. Nonetheless, it is instructive to look at how several other scientists have addressed the issue. In this section I examine Kenneth Miller's writing.
In the chapter of Finding Darwin's God (Miller 1999) which defends Darwinism from my criticisms, Professor Miller devotes the largest part--fully nine pages--to blood clotting. During the first five pages he gives an overview of how the blood-clotting cascade works, as well as noticing that bleeding can be slowed by platelet aggregation, which is not a part of the clotting cascade. In the next two pages he writes of the sequence similarity of clotting factors and the phenomenon of gene duplication--facts well known to Russell Doolittle. In the final several pages he writes of a totally unrelated clotting system, that of lobsters. In other words, Miller spends almost all of the space writing about things other than how the vertebrate blood clotting cascade may have arisen step-by-Darwinian-step.
His proposed model for the evolution of the vertebrate cascade is confined to just one paragraph. After postulating that, when a blood vessel breaks and they enter the new environment of a tissue, some blood proteins might be non-specifically cut by serine proteases and non-specifically aggregate, Miller writes:
What happened next? . . . A series of ordinary gene duplications, many millions of years ago, copied some of these serine proteases. One of these duplicate genes was then mistargeted to the bloodstream, where its protein product would have remained inactive until exposed to an activating tissue protease--which would happen only when a blood vessel was broken. From that point on, each and every refinement of this mechanism would be favored by natural selection. Where does the many-layered complexity of the system come from? Again, the answer is gene duplication. Once an extra copy of one of the clotting protease genes becomes available, natural selection will favor slight changes that might make it more likely to activate the existing protease. An extra level of control is thereby added, increasing the sensitivity of the cascade. (Miller 1999, 156-157)
Let's start with the last half of the paragraph ("From that point on . . ." and forward). The first thing to notice is that it's terminally fuzzy--too sketchy for much criticism. As I explained above, simply chanting "gene duplication" does not show how a complex system can be built, since duplication does not explain how new enzyme properties and targets arise. Russell Doolittle knew all about gene duplication, and yet postulated as a model for an evolutionary intermediate mice that turned out to be severely disabled. Professor Miller simply tries to use the term "gene duplication" as a magic wand to make the problem go away, but the problem does not go away. Miller's assertion that natural selection would favor each additional step is made quite problematic by the fact that each step in clotting has to be strictly regulated or else it is positively dangerous, as noted by Torben Halkier in the opening quotation of this document. In other words, what Halkier calls the "central issue" of regulation is ignored by Miller. Miller's statement does not even say what the newly duplicated proteases are envisioned to be acting on--whether the tissue protease, the original mistargeted circulating protease, plasma proteins, or everything at once.
Such a brief story is of no use at all in understanding how the irreducible complexity of the clotting cascade could be dealt with by natural selection. It strikes me that the main purpose of the paragraph is not to actually contribute to our understanding of how clotting actually may have arisen, but to persuade those who aren't familiar with biochemical complexity to believe Darwinism has the problem under control. It doesn't.
b. Problems from the get-go
Now let's look at the beginning of Miller's scenario. It turns out that as soon as he tries to get past the simple postulated beginning (that is, the nonspecific aggregation of proteins that have been nonspecifically degraded when a blood vessel is broken) his scenario runs into severe problems.
Miller's first step postulates a potentially deadly situation: a non-regulated zymogen circulating in the bloodstream with clottable proteins. Although we don't have access to Miller's imaginary organism to test the effects of this situation, to understand what it might mean we can look to several cases situations where regulatory proteins are missing from modern organisms: 1) Because the condition is very likely lethal in utero, no cases have been reported in the medical literature (Scriver 1989, 2213) of human patients missing antithrombin, a prominent regulator of the clotting cascade; 2) As described earlier, knockout mice missing the gene for plasminogen to remove blood clots suffer severe thrombosis and increased mortality, as well as other debilitating symptoms. (Bugge et al. 1995) It seems quite likely that Miller's hypothetical organism would experience the unregulated zymogen not as an improvement, but as a severe genetic defect.
Regulation is the key issue in clotting, but Miller does not address it at all. Miller's brief scenario does not even address potentially fatal difficulties--it ignores them. However, while Darwinists telling just-so stories can ignore difficulties, real organisms can't.
Here are several more problems with the brief scenario. First, it should be noted that the problem the scenario is trying to solve--hemostasis--can't initially be severe, because the starting point is a living organism, which must already be quite well adjusted to its environment. Second, the protein that Miller postulates to be mistargeted to the bloodstream would then no longer be doing its initial job; that would be expected to be detrimental to the organism. Third, Miller begins by postulating the mistargeting of a non-specific protease-precursor (a zymogen) to the bloodstream of some unfortunate organism. (Miller writes that if his scenario is correct, then "the clotting enzymes would have to be near-duplicates of a pancreatic enzyme . . . ." (Miller 1999, 157) Pancreatic enzymes, which have to digest a wide variety of protein foodstuffs, are among the most nonspecific of enzymes). Now, that would pose a severe health threat to the mutant organism even greater than just an unregulated clotting cascade. For example, if the digestive enzyme precursor trypsinogen were mistargeted to the bloodstream, the potential for disaster would be very large. In the pancreas, misactivation of trypsinogen is prevented by the presence of trypsin inhibitor. In Miller's scenario one cannot plausibly suppose there to be a trypsin inhibitor fortuitously circulating in the plasma. If the mistargeted enzyme were accidentally activated, it would most likely cause generalized damage in the absence of a regulatory mechanism. It would not be a viable evolutionary intermediate.
Problems of regulation aside, it is difficult to see the advantage of the protease mistargeted to the bloodstream in the first place. While Miller's initial cellular or tissue protease would by necessity be localized to the site of a cut, a circulating zymogen would not. In modern organisms thrombinogen has a vitamin K-dependent gla-domain which allows it to localize to cell surfaces. In order to be effective before membrane-binding features had been acquired, it would seem that the postulated circulating protease would have to be present at rather high concentrations, exacerbating the regulatory difficulties discussed above.
As I wrote in Darwin's Black Box (Behe 1996, 86), the problem of blood clotting is not in just forming a clot--any precipitated protein might plug a hole. Rather the problem is regulation. The regulatory problems of the clotting cascade are particularly severe since, as pointed out by Halkier (1992, 104), error on either side--clotting too much or too little--is detrimental. As irreducible complexity would predict, Kenneth Miller's scenario has no problem postulating a simple clot (the initial nonspecific aggregation) but avoids the problem of regulation.
The take-home lesson is that Miller doesn't even try to deal with the problem of irreducible complexity and other obstacles that I pointed out in Darwin's Black Box--he just ignores them.
IV. Keith Robison's Proposal
Soon after Darwin's Black Box was published Keith Robison posted some criticisms on talk.origin (), one of which concerned the blood clotting cascade. I think his proposed scheme for adding steps to cascades, while it doesn't work, is the most serious and interesting one I have come across (grad students often come up with the best ideas). It spurred me to think more about the situation and has led me to formulate the concept of irreducible complexity in more explicitly evolutionary terms.
Robison's proposal was not focused on blood clotting per se, so he doesn't worry about forming a clot. Rather he concentrates on how new steps might be added to a cascade such as occurs in blood clotting, as well as other systems such as complement. His starting point is a rather complex one, which I will grant for purposes of argument. He postulates a protein X which already has three pertinent properties: 1) it is activated by some external factor (perhaps by tissue trauma); 2) the activated protein X* then can activate more X by hydrolysis; and 3) activated X* cleaves some additional target. It's pretty much a cascade all by itself.
Fine, let's start there. Now begins the interesting scenario that I'll contest. To build a new step in the cascade, Robison then postulates several further steps. First is an initial gene duplication. Both genes make X, and the X from either gene when activated can activate the other. The second postulated event is a mutation in just one of the X genes that causes it to lose the ability to interact with the target. Nonetheless, it retains the ability to activate itself and the X coded by the original gene. The third step is loss of the yet-unmutated protein's ability to either respond to the external factor or activate itself and the other protein. At the end we have one of the proteins responsive to the external factor and able to activate both itself and the second protein, and just the second protein is able to cleave the target. Replication of the scenario yields more steps in the cascade, building irreducible complexity.
I argue that, while Robison's scenario does indeed build a new step in the cascade, it doesn't do it by Darwinian means. Rather, it does so by Robison's intelligent direction. Here are a couple pertinent quotes from the several steps of his scenario (Robison 1996): "This arrangement is neutral; the species has gained no advantage."; "Again, this genotype is neutral; it is neither beneficial nor detrimental."; "The initial steps are neutral, neither advantageous nor disadvantageous." "The final step locks in the cascade. It is potentially advantageous . . . ." (my emphasis--the "potentially" advantageous final step would require a further mutation to make it actually advantageous, so before that happens it is neutral.).
Thus his scenario postulates four successive, very specific steps: 1) gene duplication of the particular multi-talented enzyme; 2) the first loss of function step; 3) the second loss of function step; 4) a step to take advantage of the situation. As Robison emphasized, the first three steps are neutral; that is, they do the organism neither harm nor good. Only when the fourth step is completed is there a selective advantage. Now, it must be remembered that the Darwinian magic depends on natural selection. If a trait is advantageous, it will take over a population, thus providing a large base from which the next advantageous mutation might arise. However, if a trait is neutral, providing no advantage, it is far, far less likely to spread, so the odds of a second mutation appearing that depends on the first are not improved at all--they're pretty much the same as luckily getting the two specific mutations simultaneously. In the final analysis Robison's scenario is completely non-Darwinian. It postulates an already-functioning system that wasn't justified in Darwinian terms, and it then goes through three neutral, non-selected steps. Only at the very end is there a selectable property that wasn't postulated at the beginning.
To get a flavor of the difficulties Robison's scenario faces, note that standard population genetics says that the rate at which neutral mutations become fixed in the population is equal to the mutation rate. Although the neutral mutation rate is usually stated as about 10^-6 per gene per generation, that is for any random mutation in the gene. When one is looking at particular mutations such as the duplication of a certain gene or the mutation of one certain amino acid residue in the duplicated gene, the mutation rate is likely about 10^-10. Thus the fixation of just one step in the population for the scenario would be expected to occur only once every ten billion generations. Yet Robison's scenario postulates multiple such events.
V. A Modest Conclusion
I would like to pause here for a moment to point out that all three scientists who tried to meet the challenge to Darwinian evolution of blood clotting--Russell Doolittle, Kenneth Miller, and Keith Robison--foundered on exactly the same point, the point of irreducible complexity. Yet they foundered in three different ways. Doolittle mistakenly thought that even the current cascade might not be irreducibly complex, but experimental results showed him to be wrong. Miller either proposed unregulated steps or just waved his hands and shouted “gene duplication”, avoiding the problem by obfuscation. Robison directly attacked a piece of the problem, but failed to see he was intelligently guiding events in a distinctly non-Darwinian scenario. Perhaps we may be allowed to conclude that when three scientists, highly intelligent and strongly motivated to discredit it, all come up empty, that irreducible complexity is indeed a big hurdle for Darwinism.
VI. An Evolutionary Perspective on Irreducible Complexity
In Darwin’s Black Box I defined the concept of irreducible complexity (IC) in the following way.
By irreducibly complex I mean a single system which is composed of several well-matched, interacting parts that contribute to the basic function, and where the removal of any one of the parts causes the system to effectively cease functioning. (Behe 1996, 39)
While I think that’s a reasonable definition of IC, and it gets across the idea to a general audience, it has some drawbacks. It focuses on already-completed systems, rather than on the process of trying to build a system, as natural selection would have to do. It emphasizes “parts,” but says nothing about the properties of the parts, how complex they are, or how the parts get to be where they are. It speaks of “parts that contribute to the basic function”, but that phrase can, and has, been interpreted in ways other than what I had in mind (for example, talking about whole organs that contribute to complex functions such as “living”), muddying the waters in my view. What’s more, the definition doesn’t allow for “degree” of irreducible complexity; a system either has it or it doesn’t. Yet certainly some IC systems are more complex than others; some seem more forbidding than others.
While thinking of Keith Robison’s scenario, I was struck that irreducible complexity could be better formulated in evolutionary terms by focusing on a proposed pathway, and on whether each step that would be necessary to build a certain system using that pathway was selected or unselected. If a system has to pass through one unselected step on the way to a particular improvement, then in a real evolutionary sense it is encountering irreducibility: two things have to happen (the mutation passing through the unselected step and the mutation that gives a selectable system) before natural selection can kick in again. If it has to pass through three or four unselected steps (like Robison’s scenario), then in an evolutionary sense it is even more irreducibly complex. The focus is off of the “parts” (whose number may stay the same even while the nature of the parts is changing) and re-directed toward “steps.”
Envisioning IC in terms of selected or unselected steps thus puts the focus on the process of trying to build the system. A big advantage, I think, is that it encourages people to pay attention to details; hopefully it would encourage really detailed scenarios by proponents of Darwinism (ones that might be checked experimentally) and discourage just-so stories that leap over many steps without comment. So with those thoughts in mind, I offer the following tentative “evolutionary” definition of irreducible complexity:
An irreducibly complex evolutionary pathway is one that contains one or more unselected steps (that is, one or more necessary-but-unselected mutations). The degree of irreducible complexity is the number of unselected steps in the pathway.
That definition has the advantage of promoting research: to state clear, detailed evolutionary pathways; to measure probabilistic resources; to estimate mutation rates; to determine if a given step is selected or not. It allows for the proposal of any evolutionary scenario a Darwinist (or others) may wish to submit, asking only that it be detailed enough so that relevant parameters might be estimated. If the improbability of the pathway exceeds the available probabilistic resources (roughly the number of organisms over the relevant time in the relevant phylogenetic branch) then Darwinism is deemed an unlikely explanation and intelligent design a likely one.