The Extended Phenotype

by Richard Dawkins

  That eukaryotes, including ourselves, may not be insulated from this hypothetical genetic traffic is suggested by the fast-growing success of the technology of ‘genetic engineering’, or gene manipulation. The legal definition of gene manipulation in Britain is ‘the formation of new combinations of heritable material by the insertion of nucleic acid molecules, produced by whatever means outside the cell, into any virus, bacterial plasmid or other vector system so as to allow their incorporation into a host organism in which they do not naturally occur but in which they are capable of continued propagation’ (Old & Primrose 1980, p. 1). But of course human genetic engineers are beginners in the game. They are just learning to tap the expertise of the natural genetic engineers, the viruses and plasmids that have been selected to make their living at the trade.

  Perhaps the greatest feat of natural genetic engineering on the grand scale is the complex of manipulations associated with sexual reproduction in eukaryotes: meiosis, crossing-over and fertilization. Two of our foremost modern evolutionists have failed to explain to their own satisfaction the advantage of this extraordinary procedure for the individual organism (Williams 1975; Maynard Smith 1978a). As Maynard Smith (1978a, p. 113) and Williams (1979) both note, this may be one area where, if nowhere else, we shall have to switch our attention away from individual organisms to true replicators. When we try to solve the paradox of the cost of meiosis, perhaps instead of worrying about how sex helps the organism we should search for replicating ‘engineers’ of meiosis, intracellular agents which actually cause meiosis to happen. These hypothetical engineers, fragments of nucleic acid which might lie inside or outside the chromosomes, would have to achieve their own replication success as a byproduct of forcing meiosis upon the organism. In bacteria, recombination is achieved by a separate fragment of DNA or ‘sex factor’ which, in older textbooks, was treated as a part of the bacterium’s own adaptive machinery, but which is better regarded as a replicating genetic engineer working for its own good. In animals, centrioles are thought to be self-replicating entities with their own DNA, like mitochondria, although unlike mitochondria they often pass through the male as well as the female line. Although at present it is just a joke to picture chromosomes being dragged kicking and screaming into the second anaphase by ruthlessly selfish centrioles or other miniature genetic engineers, stranger ideas have become commonplace in the past. And, after all, orthodox theorizing has so far failed to dent the paradox of the cost of meiosis.

  Orgel and Crick (1980) say much the same about the lesser paradox of variable C values and about the selfish DNA theory to account for it: ‘The main facts are, at first sight, so odd that only a somewhat unconventional idea is likely to explain them.’ By a combination of fact and daydreamed extrapolation, I have tried to set a stage on to which selfish DNA itself can enter almost unnoticed; to paint a backdrop against which selfish DNA will appear, not unconventional but almost inescapable. DNA that is not translated into protein, whose codons would spell meaningless gibberish if they ever were translated, may yet vary in its replicability, its spliceability, and its resistance to detection and deletion by the debugging routines of the cellular machinery. ‘Intragenomic selection’ can therefore lead to an increase in the amount of certain types of meaningless, or untranscribed, DNA, littered around and cluttering up the chromosomes. Translated DNA, too, may be subject to this kind of selection, although here the intragenomic selection pressures will probably be swamped by more powerful pressures, positive and negative, resulting from conventional phenotypic effects.

  Conventional selection results in changes in the frequency of replicators relative to their alleles at defined loci on the chromosomes of populations. Intragenomic selection of selfish DNA is a different kind of selection. Here we are not dealing with the relative success of alleles at one locus in a gene-pool, but with the spreadability of certain kinds of DNA to different loci or the creation of new loci. Moreover, the selection of selfish DNA is not limited to the time-scale of individual generations; it can selectively increase in any mitotic cell division in the germ-lines of developing bodies.

  In conventional selection, the variation on which selection acts is produced, ultimately, by mutation, but we usually think of it as mutation within the constraints of an orderly system of loci: mutation produces a variant gene at a named locus. It is therefore possible to think of selection as choosing among alleles at such a discrete locus. Mutation in the larger sense, however, includes more radical changes in the genetic system, minor ones such as inversions, and major ones such as changes in chromosome number or ploidy, and changes from sexuality to asexuality and vice versa. These larger mutations ‘change the rules of the game’ but they are still, in various senses, subject to natural selection. Intragenomic selection for selfish DNA belongs in the list of unconventional types of selection, not involving choice among alleles at a discrete locus.

  Selfish DNA is selected for its power to spread ‘laterally’, to get itself duplicated into new loci elsewhere in the genome. It does not spread at the expense of a particular set of alleles, in the way that, say, a gene for melanism in moths spreads in industrial areas at the expense of its alleles at the same locus. It is this that distinguishes it, as a ‘laterally spreading outlaw’, from the ‘allelic outlaws’ which were the subject of the previous chapter. Lateral spreading to new loci is like the spread of a virus through a population, or like the spread of cancer cells through a body. Orgel and Crick indeed refer to the spreading of functionless replicators as a ‘cancer of the genome’.

  As for the qualities which are actually likely to be favoured in the selection of selfish DNA, I would have to be a molecular biologist in order to predict them in detail. One does not have to be a molecular biologist, however, to surmise that they might be classified into two main classes: qualities that make for ease of duplication and insertion, and qualities that make it difficult for defence mechanisms of the cell to seek them out and destroy them. Just as a cuckoo egg seeks protection by mimicking the host eggs which legitimately inhabit the nest, so selfish DNA might evolve mimetic qualities that ‘make it more like ordinary DNA and so, perhaps, less easy to remove’ (Orgel & Crick). Just as a full understanding of cuckoo adaptations is likely to require a knowledge of the perceptual systems of hosts, so a full appreciation of the details of selfish DNA adaptations will require detailed knowledge of exactly how DNA polymerase works, exactly how snipping and splicing occur, exactly what goes on in molecular ‘proof-reading’. While full knowledge of these matters can only come from detailed research work of the kind that molecular biologists have so brilliantly succeeded in before, it is, perhaps, not too much to hope that molecular biologists might be aided in their research by the realization that DNA is not working for the good of the cell but for the good of itself. The machinery of replication, splicing, and proof-reading may be better understood if it is seen as in part the product of a ruthless arms race. The point may be emphasized with the aid of an analogy.

  Imagine that Mars is a Utopia in which there is complete trust, total harmony, no selfishness and no deceit. Now imagine a scientist from Mars trying to make sense of human life and technology. Suppose he studied one of our large data processing centres—an electronic computer with its associated machinery of duplication, editing and error-correction. If he made the assumption—natural to his own society—that the machinery had been designed for the common good, he would go a long way towards understanding it. Error-correcting devices, for instance, would clearly be designed to combat the inevitable and non-malevolent Second Law of Thermodynamics. But certain aspects would remain puzzling. He would make no sense of the elaborate and costly systems of security and protection: secret passwords and code numbers that have to be typed in by computer users. If our Martian examined a military electronic communication system he might diagnose its purpose as the rapid and efficient transmission of useful information, and he might therefore be baffled by the trouble and expense to which the system seems to go
in order to encode its messages in a way which is obscure and hard to decode. Is this not wanton and absurd inefficiency? Brought up as he is in a trusting Utopia, it might require a major flash of revolutionary insight for our Martian to see that much of human technology only makes sense when you realize that humans distrust each other, that some humans work against the best interests of other humans. There is a struggle between those who wish to obtain illicit information from a communication system and those who wish to withhold that information from them. Much of human technology is the product of arms races and can only be understood in those terms.

  Spectacular as their achievements have been, is it possible that molecular biologists have hitherto, like biologists at other levels, been in something like the position of our Martian? By assuming that the cell is a place where molecular machinery whirrs for the good of the organism, they have come a long way. They may go further if they now cultivate a more cynical view and countenance the possibility that some molecules may be up to no good from the point of view of the rest. Obviously they already do this when they contemplate viruses and other invading parasites. All that is needed is to turn the same cynical eye on the cell’s ‘own’ DNA. It is because they are starting to do just this that I find the papers of Doolittle and Sapienza and Orgel and Crick so exciting, in comparison with the objections of Cavalier-Smith (1980), Dover (1980) and others, although of course the objectors may be right in the particular detailed points they make. Orgel and Crick summarize the message well:

  In short, we may expect a kind of molecular struggle for existence within the DNA of the chromosomes, using the process of natural selection. There is no reason to believe that this is likely to be much simpler or more easy to predict than evolution at any other level. At bottom, the existence of selfish DNA is possible because DNA is a molecule which is replicated very easily and because selfish DNA occurs in an environment in which DNA replication is a necessity. It thus has the opportunity of subverting these essential mechanisms to its own purpose.

  In what sense is selfish DNA an outlaw? It is an outlaw to the extent that organisms would be better off without it. Perhaps it wastes space and molecular raw materials, perhaps it wastefully consumes valuable running time on the duplicating and proofreading machinery. In any event we may expect that selection will tend to eliminate selfish DNA from the genome. We may distinguish two kinds of ‘anti-selfish DNA’ selection. Firstly, selection might favour positive adaptations to rid organisms of selfish DNA. For example, the already discovered proofreading principle might be extended. Long sequences might be examined for ‘sense’ and excised if found wanting. In particular, highly repetitive DNA might be recognized by its statistical uniformity. These positive adaptations are what I had in mind in my above discussion of arms races, ‘mimicry’, etc. We are talking, here, about the evolution of anti-selfish DNA machinery which could be as elaborate and as specialized as the antipredator adaptations of insects.

  There is, however, a second kind of selection that could act against selfish DNA, which is much simpler and cruder. Any organism that happened to experience a random deletion of part of its selfish DNA would, by definition, be a mutant organism. The deletion itself would be a mutation, and it would be favoured by natural selection to the extent that organisms possessing it benefited from it, presumably because they did not suffer the economic wastage of space, materials and time that selfish DNA brings. Mutant organisms would, other things being equal, reproduce at a higher rate than the loaded down ‘wild-type’ individuals, and the deletion would consequently become more common in the gene pool. Note that I am not now talking about selection in favour of the capacity to delete selfish DNA: that was the subject of the previous paragraph. Here we are recognizing that the deletion itself, the absence of the selfish DNA, is itself a replicating entity (a replicating absence!), which can be favoured by selection.

  It is tempting to include under the heading of outlaws somatic mutations that cause cells to out-reproduce the non-mutant cells of a body, to the ultimate detriment of the body itself. But although there is a kind of quasi-Darwinian selection that can go on in cancer tumours, and Cairns (1975) has ingeniously drawn attention to what appear to be bodily adaptations to forestall such within-body selection, I think that to apply the outlaw concept here would not be helpful. Not, that is to say, unless the mutant genes concerned somehow manage to propagate themselves indefinitely. They could do this either by getting themselves transported in virus-like vectors, say through the air, or by somehow burrowing into the nuclear germ-line. In either of these two cases they would qualify as ‘germ-line replicators’ as defined in Chapter 5 and the title of outlaw would be appropriate.

  There has been one startling recent suggestion that genes which are beneficiaries of somatic selection might indeed burrow into the germ-line, though in this case they are not cancerous and not necessarily outlaws. I want to mention this work because it has been given publicity as a would-be resuscitator of the so-called ‘Lamarckian’ theory of evolution. Since the theoretical position adopted in this book is fairly describable as ‘extreme Weismannism’, I am bound to see any serious revival of Lamarckism as undermining my position. It is therefore necessary to discuss it.

  A Lamarckian scare

  I use the word ‘scare’ because, to be painfully honest, I can think of few things that would more devastate my world view than a demonstrated need to return to the theory of evolution that is traditionally attributed to Lamarck. It is one of the few contingencies for which I might offer to eat my hat. It is therefore all the more important to give a full and fair hearing to some claims made on behalf of Steele (1979) and Gorczynski and Steele (1980, 1981). Before Steele’s (1979) book was available in Britain, The Sunday Times of London (13 July 1980) printed a full-page article about his ideas and ‘astonishing experiment which seems to challenge Darwinism and resurrect Lamarck’. The BBC have given the results similar publicity, in at least two television programmes and several radio programmes: as we have already seen, ‘scientific’ journalists are constantly on the alert for anything that sounds like a challenge to Darwin. No less a scientist than Sir Peter Medawar forced us to take Steele’s work seriously by doing so himself. He was quoted as being properly cautious about the need to repeat the work, and as concluding: ‘I have no idea what the outcome will be, but I hope Steele is right’ (The Sunday Times).

  Naturally any scientist hopes that the truth, whatever it is, will out. But a scientist is also entitled to his innermost hopes as to what that truth will turn out to be—a revolution in one’s head is bound to be a painful experience—and I confess that my own hopes did not initially coincide with Sir Peter’s! I had my doubts about whether his could really coincide with those attributed to him, until I recalled his, to me always slightly puzzling (see page 22), remark that ‘The main weakness of modern evolutionary theory is its lack of a fully worked out theory of variation, that is, of candidature for evolution, of the form in which genetic variants are proffered for selection. We have therefore no convincing account of evolutionary progress—of the otherwise inexplicable tendency of organisms to adopt ever more complicated solutions of the problems of remaining alive’ (Medawar 1967). Medawar is one of those who have, more recently, tried very hard, and yet failed, to replicate Steele’s findings (Brent et al. 1981).

  To anticipate the conclusion I shall come to, I now view with equanimity, if with dwindling expectation (Brent et al. 1981; McLaren et al. 1981), the prospect of Steele’s theory being upheld, because I now realize that, in the deepest and fullest sense, it is a Darwinian theory; a variety of Darwinian theory moreover which, like the theory of jumping genes, is particularly congenial to the thesis of this book, since it stresses selection at a level other than that of the individual organism. Though pardonable, the claim that it challenges Darwinism turns out to be just a journalistic gloat, provided Darwinism is understood in the way that I think it ought to be understood. As for Steele’s theory itself, even if the fac
ts do not uphold it, it will have done us the valuable service of forcing us to sharpen our perception of Darwinism. I am not qualified to evaluate the technical details of Steele’s experiments and those of his critics (a good evaluation is given by Howard 1981), and will concentrate on discussing the impact of his theory, should the facts eventually prove to support it.

  Steele forges a threefold union of Burnet’s (1969) theory of clonal selection, Temin’s (1974) provirus theory, and his own attack on the sanctity of Weismann’s germ-line. From Burnet he gets the idea of somatic mutation leading to genetic diversity among the cells of the body. Natural selection within the body then sees to it that the body becomes populated by successful varieties of cells at the expense of unsuccessful varieties. Burnet limits the idea to a special class of cells within the immune system, and ‘success’ means success in neutralizing invading antigens, but Steele would generalize it to other cells. From Temin he gets the idea of RNA viruses serving as intercellular messengers, transcribing genes in one cell, carrying the information to another cell and reverse-transcribing it back into DNA in the second cell using reverse transcriptase.

  Steele uses the Temin theory, but with an important additional emphasis on germ-line cells as recipients of reverse-transcribed genetic information. He wisely limits most of his discussion to the immune system, although his ambitions for his theory are larger. He cites four studies on ‘idiotypy’ in the rabbit. If injected with a foreign substance, different individual rabbits combat it by making different antibodies. Even if members of a genetically identical clone are subjected to the same antigen, each individual responds with its own unique ‘idiotype’. Now, if the rabbits really are genetically identical, the difference in their idiotypes must be due to environmental or chance differences, and should not, according to orthodoxy, be inherited. Of the four studies cited, one gave a surprising result. The idiotype of a rabbit turned out to be inherited by its children, although not shared by its clone-mates. Steele stresses the fact that in this study the parent rabbits were exposed to the antigen before mating to produce the offspring. In the other three studies the parents were injected with antigen after mating, and the offspring did not inherit their idiotypes. If idiotype were inherited as a part of an inviolate germ-plasm, it should not have made any difference whether the rabbits mated before or after injection.