Paper #9 

Paper: Denton

Respondent: Wilcox


Denton: At every level from a protein up, there is a considerable amount of potential for variation and change around a particular point. You have to take that sort of evidence, however, in conjunction with other evidence, like the recent discovery of the gene which is involved in segmentation--the same protein product, as demonstrated by immunological evidence, is involved in switching the fate of specific neuroblasts. There is an increasing amount of evidence like that--an increasing number of genes now that are found to be overlapping. So you have two completely different functions in the same DNA sequence.

As I say, I don't believe we have an objective knowledge of the significance of these odds and ends of things. We can't really measure the final objective significance of these findings. The only place we can be objective about the potential for change, dramatic change, is at the molecular level when we start doing genetic engineering. This is why I am withdrawing to some extent from a lot of evolutionary debate: I am increasingly aware of the fact that all these things cannot be fit into a coherent pattern of real knowledge, though there are observations that are very interesting.

For instance, the experiment you (Wilcox) described: You can graft a part of a frog's head into a newt's head when it is developing, and it will grow there and become part of the frog. This is certainly suggestive of plasticity. Other evidence seems to impose a sort of Cuvierian constraint on change, like the evidence of increasing numbers of genetic systems where you have a gene inside an intron; or the fact that you get many genes, particularly in the nematode, where you have a gene involved in switching the fates of all sorts of different cells. What the final answer to all this is I don't know. We haven't got an objective science of biology because we have no real way of measuring the significance of these things. If you are an evolutionist you can take certain evidence and say that means things are very plastic. Also you could take certain evidence from protein engineering and say it looks as if proteins can't be changed. You have site-directed mutagenesis of substrate binding, you can change the heat stability by putting in more H-bonds, things like this. I've heard these sorts of arguments over and over again. They suggest that perhaps you can change proteins, perhaps there is a network of continuous change which you go with little steps all the way down the line. The bottom line is that there just is not enough evidence. Just circumstantial, non-objective evidence. I'm sorry to say this, but I think it actually is the case.

Thaxton: In your paper you referred to the difference between building something from the bottom up versus building it from the top down. Would you address that?

Denton: There is an interesting editorial by Maddox in Nature, May 5, 1988, in which he raises the question, Is reductionism beginning to fail in biology? Suzuki, the Canadian geneticist, came to a meeting in Sydney. His keynote address was "The Failure of Reductionism in Molecular Biology." This is a very interesting situation which relates to the possibility of a genetic engineering, and the possibility of an objective biology. Can I say something about this, I think it is an interesting point. Sorry, Dave (Wilcox), we could have talked about this for ages, but that's just my general feeling about these things.

Wilcox: I was just raising...

Denton: Yes, and I think your objection is valid. Most of my life as a medic in human genetics and molecular biology, I have been relentlessly reductionistic. That's a confession. All the time I've been mechanistic and reductionist in my thinking. That means I have basically held the following assumption: that living things are combined of elements like this. (Slide). These things have emergent properties but everything is predictable from where you start.

But there are problems coming from genetic engineering, and this is why there is this wave of wondering about whether or not the reductionist's prospect is going to work. It is the old concept of an organic unity--that as things amalgamate and coalesce into the form, they completely change their structure and behavior in unpredictable ways. This is the concept of an organic unity as opposed to a mechanical unity, where the behavior and the emergent properties of the whole are all predictable. That's how we send a space ship to Mars. Already in protein engineering this is fundamentally the difficulty. A protein is a very simple thing. It is only 100 amino acids long. A space ship has four million components, yet they have predictable properties and you can send them to Mars. A protein is only a hundred amino acids long and yet still we can't tell how it's going to fold. The reason is that it is formed in sort of modular areas, bits of alpha helices, but when they coalesce you get completely unpredictable structures coming out of it.

Paul (Nelson) was showing me a paper yesterday about how pentapeptides in proteins have generally completely different conformations. So, you might have a pentapeptide in this protein with a beta-sheet conformation, but in another protein it has an alpha helical conformation. In other words the whole is determined in the form and composition of the parts in a way that is very difficult to predict.

I've said I believe protein engineering is possible because although proteins are certainly analogous to what Aristotle and Cuvier thought of as an organic unity, I think in practice it is possible to solve the problem of organic unity at this level. But what worries me, and is worrying more and more molecular biologists, is when you get to complex systems, where you are putting hundreds and hundreds of things together, and they are going through all sorts of complex conformational changes, you wonder if that's ever going to be predictable. You wonder whether the reductionist thesis will succeed. The fact that living things coalesce in complex ways presents a very difficult problem for predicting what they are going to be from the linear sequences.

I have an example showing the conformational change in a component of the T-phage. It's dramatic. It's the connection between the head and the tail. It's composed of the two proteins which form a weird star formation which suddenly jumps into a tight collar. It's predicting these sorts of interactions between thousands of atoms which is going to be difficult because the form of the collar is not obvious from the two proteins that make it up. This is the problem of an organic unity as opposed to a molecular unity. I've said that genetic engineering can make a biology objective but there are doubts as to how easy it is going to be predicting things from the one-dimensional form. This is an extremely challenging area of science now. It may be a long time before we have theories like you have in physics which can predict thousands of millions of years of cosmic evolution. It is much more difficult I think to predict how these things are going to fold because you get into the problem of infinite computability.

Yockey: Could you give that reference again about the failure of reductionism?

Denton: Nature, May 5, 1988. Maddox, who wrote the editorial, has been known as a hard-headed reductionist mechanist for the last 30 years. "Not seeing the wood for the trees" is the title.

Yockey: I have addressed the topic of reductionism in the beginning of my paper, and the reason is that I talk about entropy. Entropy is measurable but it is not material. What I am saying is that you add that to the physics and chemistry which you have to obey, something which the reductionist may not like because it's not material. But it's got to be there. Otherwise, you have to reject Watson and Crick. And I don't think people are going to do that. So I am extremely interested in anything you know about this area. In other words, how controversial reductionism is in biology. As a physicist I'm not in close contact with that.

Denton: I think for the last 30 years the reductionist viewpoint has burnt like a furnace. Nobody has ever questioned it. Its explanatory power has been enormous. But suddenly now in genetic engineering you're faced with going up from a linear sequence to three dimensions and it's more difficult. You can see it, and describe it happening. But when you try to predict it happening it's very, very difficult.

Olson: If it takes Herculean human effort to produce new functioning proteins, is it proper to conclude, as I think you did early in your talk, that new proteins will necessarily appear over time in a natural world having no human input?

Denton: It's going to take a Herculean effort to solve the problem of prediction. But it might turn out when we have a good protein science that in fact functional proteins are quite common. I don't think that myself. But it would be quite improper of me to say that can be decided yet. I'm inclined to think the evidence points in favor of the idea that functional conformations are relatively discontinuously separated.

The Embden Myerhof pathway I showed you: It's funny but successive enzymes in that pathway handle the same substrates but have quite different conformations. That strikes me as being very significant. The fact that proteins go through very tortuous pathways to get to the stable form strikes me as being significant. It strikes me as being significant that in all the massive numbers of artificial changes made in proteins, you either conserve the form or you have nothing.

Yockey: Doesn't this blow Sidney Fox's proteinoids out of the water?

Denton: I have to keep covering myself. This is my prejudice. And this is how I interpret the evidence for the past five years. In ten years I might have to change my mind, though. There is a lot of circumstantial evidence that suggests that good stable folds are pretty uncommon. But I don't know what is going to turn out eventually. The approach to protein engineering at the moment is designing from the bottom up. You make an alpha helix, and you try to stick another alpha helix onto it. Then you try to put a beta-sheet onto it. You build up slowly in a modular fashion. In my interpretation, this approach is failing. What happens is that once you get to about sixty amino acids long, properties become unpredictable--either it aggregates or it becomes a random coil, or something like this. All the circumstantial evidence is pointing to the fact that proteins are pretty rare. But it will be a fantastic day when we have a protein science and we can answer that question.

Yockey: You don't start out with a mixture of d- and l-, do you?

Denton: No.

Yockey: But he (Fox) does, and he cooks them up and he says, ah, they go into little droplets and ...

Denton: But that's irrelevant.

Yockey: I know it's irrelevant. But the point I'm making has to do with what you're saying about the complicated folding of these things: Fox's damn things don't fold at all. They're just globs.

Denton: To show whether any of his proteinoids are functional or not... mind you, some of his proteinoids have some vague catalytic activity. This doesn't surprise me because anything can be a catalyst. To show that he didn't make a protein that folded you have to isolate all the various protein-like species in his tars, and then characterize them to see if any of them did fold, right? You see what I am getting at? Some of them might have biological properties, I don't know. Certainly the mixture has catalytic properties. The fact is that doing an experiment like that now is dated in the context of genetic engineering. I think that's a silly way to go about it. Let's find out what makes a stable alpha helix. Let's see how many alpha helices can go into proteins. Find out what the constraints are that you have to satisfy.

Rust: Why did you say it would be disturbing for the creationists if the space of all the protein conformations would be densely populated with viable configurations?

Denton: Because it means that you could have a continuity of molecular forms separated by only small steps.

Rust: So what?

Denton: As Dawkins showed in The Blind Watchmaker, if you have continuity you can have evolution. That's fundamental. If you have a continuity of forms you can go from anything to anything else.

Rust: You can have evolution with creation.

Denton: You could have. But if you have continuity you can find anything you want by going in tiny little steps from any starting point. The most parsimonious explanation would then definitely be natural selection. That's why Darwin was so accepted--he presented an argument which is plausible. If in fact Nature is continuous, as Dawkins shows in The Blind Watchmaker, then you can get anything. As he says, I can get anything from anything else if I can go continuously in little steps. So that if it turns out that proteins are common... and I'm sorry I don't know what the answer is going to be, but my suspicion is that the evidence points more toward discontinuity than continuity. If it turns out to be continuous it's a disaster for creationism because you can produce anything from anything else. There is effectively no longer a problem. It is the old Muller idea: that just by making little steps you can get to an incredible end.

Rust: I think it still doesn't say anything about creation.

Denton: If the forms of nature are largely as continuous as most evolutionists have assumed from Empedocles right through to Darwin and people like Dawkins, you don't need any Creator.

Rust: We don't need a Creator for creation only.

Denton: I'm just saying the scientific community would find this very convenient if this turned out to be correct.

Mills: Let me mention in regard to hemoglobin mutants (of course hemoglobins are tetramers rather than monomers), just a single change in many cases markedly increases the instability. The instability ultimately leads to precipitation. Whether it is a randomization of the structure that causes this I don't know, but that's one of the primary means by which you can identify the unstable variants of hemoglobin and other products. In other words, the natural conformation is the most stable and only a slight change in it leads to the instability.

Denton: That's possibly what happens when it precipitates, it becomes a random coil. I have a selection of papers here from the Journal of Protein Engineering from the past few months. They make the same points I am making here but in a slightly more sophisticated form. I'm a human geneticist making recombinant DNA probes, doing linkage work on the human gene. I'm trying to isolate genes. When I got into this work four years ago, I was unaware of the dramatic effect genetic engineering was going to have on biology. I don't think the claims of people like David Baltimore, Brenner, Crick, Watson--The Asilomar Conference--are exaggerated, though I did tend to think that before I started working with the molecules.

What's impressed me is we have a genetic word processor which allows us to make any DNA sequence, any change, in any organism we want. It's not true of complex metazoa yet, but I think it's coming rapidly. It stems out of the chromosomal structure of complex organisms. It's almost providential. It is almost as if man was allowed to play this game. It looks as if in fact it is going to be simple even to make artificial chromosomes. Because you have to have the telemeric sequence correct, the centromeric sequence correct, then I think you can cut bits out and put bits in. Already this is starting in yeast. This is an eucaryote. I don't think there is overwhelming reason why we shouldn't eventually by the end of the century have the technology available.

If somebody had told me in 1965 when I was in the middle of a medical course, that we would fiddle at will with DNA sequences and contemplate a real science of biology like this I would never have believed it. It is incredible what has happened. And now I would say that by the end of the century man will have the power--the genetic word processor--which will allow him to change any DNA sequence in any organism at will, and insert and take out genes from any organism he wants. Prometheus will be truly unbound to do what he wants. What he'll do is difficult to tell. What science will come out of it I don't know. But I do think this is a historic point in the genuine sense of the word. This is like the Lavoisier era of chemistry. Biology is being dragged into a quantitative or objective science. This is because you can do anything now that you want. Instead of having to just take what nature is giving you, you can make anything you want. That's getting at the great question of continuity or discontinuity in the form of nature.

Hearn: I've read some interesting notes lately about the effect of our mathematical training on the way we look at nature. I wonder if Dave Bossard would comment on this. I read some notes from The American Mathematical Association saying that if they hadn't invented calculus, which is a continuous view of nature, and we had just digital computers and so forth, we might think quite differently. Is there anything to this?

Denton: A lot of people are thinking about this question actually. Yes, I think that in fact continuous processes are much easier to handle than discontinuous processes. Would this be true in mathematics?

Bossard: Discrete processes, yes.

Denton: We like continuity anyway, a priori, because it is easier to explain. Evolution is easier to explain if things are continuous. That's the whole problem with the Cuvierian view. They couldn't give a natural explanation. Even Gould has said nature is so incredibly discontinuous but

I can't think of a natural explanation for it if it's discontinuous. So what you're saying, Walt, is that we like simple continuous processes because they are easy to model.

Bossard: The complexities that biologists are now discovering in trying to predict the three-dimensional structure say from the one-dimensional linear structure is well known to physicists. It's a classic N-body problem which has been around since the 1700s. Actually it is a combination of two mind-boggling problems: the N-body problem and three-dimensional global minimization, where you have a whole slew of local minima, by which we mean the smallest of a specified set of real numbers.

Actually, in terms of the discrete realization that you talk about mathematically, what you picture is a bunch of valleys between hills. That's discrete by definition. Local minima are not continuously connected to each other necessarily. So, mathematically you're still talking about a continuous process--I mean you're not talking about a quantum biology in a sense. You're talking about continuous transitions but global minima which are very deep wells, and there are only a few of these deep wells.

I view the problems you are facing as problems not of mathematical impossibility, but levels of mathematical computational complexity. If in fact you can give me the equations that would do all this, then it's just a matter of solving them. Recently there was a book on the New York Times' best-seller list called Chaos by James Gleick. One of the pictures in there is the trajectory of a 3-body problem, where you have two equal masses and a third mass orbiting those two. That's only 3 bodies but it is amazingly complex. Now, we're talking about hundreds, not point bodies. It may not be impossible but the computational complexities are enormous.