by Matthew Cobb
To put all this speculation into a theoretical context, Crick explained to his audience that there were three ways of understanding the processes involved in protein synthesis – ‘the flow of energy, the flow of matter, and the flow of information.’20 He focused on the final, most elusive, and most radical aspect – the flow of information. In both his lecture and the Scientific American article that appeared at the same time, Crick used a memorable term to describe a fundamental feature of genes: he outlined what he called ‘the central dogma’ of genetics. Crick explained this dogma as follows:
once information (meaning here the determination of a sequence of units) has been passed into a protein molecule it cannot get out again, either to form a copy of the molecule or to affect the blueprint of a nucleic acid. The idea is not universally accepted, however. In fact, Sir Macfarlane Burnet, the eminent Australian virologist, persuasively argued another point of view in a very interesting little book which he published recently.21
On the basis of the available molecular evidence, Crick was arguing that there were four kinds of information transfer that were likely to take place routinely: DNA → DNA (as in DNA duplication), DNA → RNA (as in the first step of protein synthesis), RNA → protein (as in the second step of protein synthesis) and RNA → RNA (presumed to exist because of the existence of RNA viruses such as TMV, which used RNA both to store information and to synthesise proteins and were able to copy themselves). Crick agreed that there were also two possible information transfers that might conceivably take place, but for which there was no evidence: DNA → protein (this would occur if protein synthesis took place directly on the DNA molecule, which seemed unlikely) and RNA → DNA (there was no evidence for this, nor any biological process that seemed to require it, but it was not seen as structurally impossible).
There were several genetic information transfers that Crick and his colleagues considered to be impossible: protein → protein (this had been disproved by the work of Avery, Hershey and Chase and others), protein → RNA and, above all, protein → DNA. There was no evidence for any of these information flows, nor was there any conceivable mechanism for the sequence of amino acids being back-translated into a DNA or RNA code. As Crick later put it, ‘I decided, therefore, to play safe, and to state as the basic assumption of the new molecular biology the non-existence of [these] transfers’.22 This was the ‘central dogma’: once information had gone from DNA into the protein, it could not get out of the protein and go back into the genetic code.
*
Crick had first come up with the ‘central dogma’ phrase and its underlying concept in October 1956, in a set of notes entitled ‘Ideas on protein synthesis’.23 These were not circulated – not even to the RNA Tie Club – but they formed the basis of his discussions with his colleagues and his thinking over the following months. In those original notes – but not in either of the published forms of his talk – Crick included a little diagram to show what he meant.
On the original note, Crick playfully wrote ‘The doctrine of the Triad’ (DNA, RNA and protein), but he soon coined the more dramatic term ‘central dogma’. As was evident from his presentation of the idea in 1957, it was not, strictly speaking, a dogma (a fundamental belief that cannot be questioned). It was instead a hypothesis, and rather than being based on any a priori position, it was simply based on the available data. Crick later recalled:
5. Crick’s first outline of the central dogma, 1956, http://profiles.nlm.nih.gov/ps/access/SCBBFT.pdf
I called this idea the central dogma, for two reasons, I suspect. I had already used the obvious word hypothesis in the sequence hypothesis, and in addition I wanted to suggest that this new assumption was more central and more powerful. I did remark that their speculative nature was emphasized by their names. As it turned out, the use of the word dogma caused almost more trouble than it was worth. Many years later Jacques Monod pointed out to me that I did not appear to understand the correct use of the word dogma, which is a belief that cannot be doubted. I did apprehend this in a vague sort of way but since I thought that all religious beliefs were without any serious foundation, I used the word in the way I myself thought about it, not as most of the rest of the world does, and simply applied it to a grand hypothesis that, however plausible, had little direct experimental support.24
As well as predicting the results of future experiments, Crick was unwittingly supporting two of the central tenets of twentieth-century biology. First, there was the assumption, suggested by August Weismann in the 1890s, that in animals the organism contains two entirely separate cell lines, one devoted to the development of the body (the somatic line) and the other to reproduction and the transmission of hereditary characters (the germ line) (no such division exists in plants or, obviously, in single-celled organisms; it is also absent in some animals). According to Weismann, these two cell lines did not interact. As a result, it was impossible for any character that was acquired during an animal’s life and which affected the somatic line, to have any effect on the germ line, on heredity. In Crick’s language, information could not go in the direction protein → DNA.
By providing a molecular basis for Weismann’s position, Crick was also reinforcing the widespread opposition to the suggestion by the nineteenth-century naturalist Lamarck (and also Darwin) that characters acquired by an organism during its lifetime could have an effect on its offspring by altering its hereditary constitution. In Weismann’s model, there was simply no route for this to occur. During the 1920s and 1930s, Weismann’s division of cell types became a cornerstone of what was known as the neo-Darwinian synthesis as genetics and evolutionary theory fused, changing the way in which scientists looked at evolution by natural selection. In 1957 Crick said essentially the same thing as Weismann, but in the latest language and with a far greater import, because it applied to all organisms, not just animals: once information had got into the protein, it could not get out again. It could not go back into the DNA. Although not a dogma, this was a very strong assertion. As Crick later reflected: ‘In looking back I am struck … by the brashness which allowed us to venture powerful statements of a very general nature.’25
Crick did not consciously set out to support the neo-Darwinian position – that would indeed have been dogmatic. Instead he developed his ideas on the basis of the experimental data, which revealed no potential mechanism for information to go from protein to DNA.26 Crick recalled:
6. Burnet’s view of the pathway from DNA to protein, from Burnet (1956). PR = protein, F.PR = functional proteins, such as enzymes.
Nobody tried to go from protein sequence back to nucleic acid, because that wasn’t on. You see. But I don’t think it was ever discussed.27
As Crick explained in both his Scientific American article and his lecture, he was struck by Macfarlane Burnet’s adoption of a rather different position in Enzyme, Antigen and Virus. Burnet later described this as ‘a rather bad over-ambitious book’, primarily because he very soon changed his views on its main subject – the way in which antibodies are created.28 Whatever the book’s virtues, it contains a diagram outlining his view that may have been the inspiration for Crick’s sketch.
Given that Burnet stated that he was trying to apply information theory to cell communication, it is notable that his arrows do not show the flow of information, but instead represent what he called specific pattern (a rather vague concept that seemed to mean something like specificity). In the accompanying text, Burnet described how the system was capable of ‘transferring coded information in pattern on one medium to a different coding of pattern on another medium’.29 Burnet’s main difference with Crick, apart from the systematic importance he gave to enzymes (given in the diagram as ‘F.PR = functional proteins’), was that the gene product can alter the genetic message, in the shape of the RNA. In Burnet’s schema, information can get from the protein into RNA, and hence into the way in which genetic information is represented. Although Crick implied that Burnet argued that proteins could affect DN
A, this was not in fact what Burnet stated. Burnet’s explanation of his hypothesis was complex and was based on his then-current model of how the body is able to generate vast numbers of different molecules that are used in the immune response as a way of differentiating self from non-self. Burnet argued that foreign proteins or antigens prompted the body to produce a specific antibody by becoming incorporated into the RNA; this was the nature of the arrow going from ‘F.PR’ to RNA.30 Within three years, Burnet had abandoned this view in favour of his new clonal theory of antibody generation; a year after that he won the Nobel Prize in Physiology or Medicine for his work on viruses and the immune response.
In the 1960s, Crick’s central dogma was rendered truly dogmatic by Jim Watson, who included a simplified version in his textbook Molecular Biology of the Gene, converting it to the form first outlined by Boivin in 1949 and Dounce in 1952: DNA → RNA → protein.31 Crick’s initial view, which allowed for the unlikely possibility of information transfer from RNA to DNA, was largely forgotten. Crick’s dramatic and mistaken use of words ended up undermining his aim – for some scientists, the central dogma became a dogma, and not merely a hypothesis.32
*
Like Burnet, Crick was later critical of his own work. In his autobiography, Crick described ‘On protein synthesis’ as ‘a mixture of good and bad ideas, of insight and nonsense’.33 As with Burnet, Crick’s self-criticism was aimed at those detailed areas of the mechanism that he got wrong. On the big picture, Crick was absolutely right. And in one area, both he and Burnet were positively visionary. Although neither Burnet nor Crick were evolutionary biologists, they each had insights into the way that evolution affects genetic information, insights that are still valid today. In his 1957 talk, Crick pointed to the handful of proteins that had thus far been sequenced in more than one organism and made a leap of the imagination that eventually transformed how we study evolution:
Biologists should realise that before long we shall have a subject which might be called ‘protein taxonomy’ – the study of the amino acid sequences of the proteins of an organism and the comparison of them between species. It can be argued that these sequences are the most delicate expression possible of the phenotype of an organism and that vast amounts of evolutionary information may be hidden away within them.34
Crick was right. Today, protein fragments from the depths of time, such as bits of collagen from Tyrannosaurus rex, can be used to study evolution.35
Burnet’s contribution to evolutionary biology was less dramatic but equally insightful. In the pages of Enzyme, Antigen and Virus, Burnet described the Watson–Crick model of the genetic code – that is, of a relation between a sequence of four nucleotide bases in the nucleic acid and the near-infinite structure of proteins – as ‘faintly unsatisfactory’.36 What exactly was ‘unsatisfactory’ – apart from the fact that the genetic code was still unbroken – Burnet did not explain. Instead he did some quick back-of-the-envelope calculations about the amount of DNA in an average cell and came up with a problem. Assuming that a human cell contained 40,000 genes, each composed of 3,000 bases, that would still account for only 1 per cent of the DNA that was estimated to be present in the nucleus of a human cell; this raised the question of what the other 99 per cent was doing.37 Although Burnet’s guesstimate that an average gene was 3,000 bases long was entirely gratuitous (genes are in fact often much, much bigger), his question was entirely valid – much of the DNA in our cells seems to be doing nothing. By the 1970s, when it became obvious that this really was a problem, Burnet’s insight had long been forgotten.
Crick’s 1957 speech and the articles that accompanied it were enormously influential, in terms of both the ideas they contained and the words they used. Crick’s framework – seeing genes and proteins in terms of information flow – rapidly became the accepted way of understanding the fundamental processes of cells. It might be expected that this shift would have been accompanied by a flourishing application of information theory in the realm of molecular biology – Crick’s information could surely be studied using Shannon and Wiener’s equations. In fact, at the same time as Crick came up with the idea of the central dogma, it began to become evident that information theory was not going to transform biology.
*
In 1942, 3,000 people who lived and worked on a plateau near Oak Ridge in the Appalachian Mountains were ordered to leave their homes. Within a year, the US government had built a small city on the site, devoted to the production of weapons-grade uranium and plutonium as part of the drive to build the atomic bomb. Thousands of workers were involved in this top-secret work, which cost more than $1bn to set up. After the war, Oak Ridge National Laboratory focused on the development of nuclear reactors and on the production of radioisotopes for use in medicine and in research, as well as occasional crazy projects such as a plan to build a nuclear-powered aircraft. New groups were set up at the laboratory, including a Biology Division to study the effects of radiation and a Mathematical and Computing Section, armed with the latest von Neumann computers, to analyse the data that the site generated. Researchers at Oak Ridge had a broad range of scientific interests, and in 1950 some of them organised the meeting on the chemistry of nucleic acids at which Chargaff showed that DNA was not a boring molecule.
At the end of October 1956, scientists at the laboratory held a symposium on the links between information theory and biology, following in the footsteps of Henry Quastler’s 1952 meeting. The conference, hosted by the Manhattan Project physicist Hubert Yockey, was entitled ‘A Symposium on Information Theory in Health Physics and Radiobiology’, and half of the talks focused on how radiation and ageing affected biological tissues and processes.38 Some of the other presentations ambitiously tried to find evidence for negative feedback in liver regeneration, or studied the role of protein synthesis and information transfer in the development of the chick embryo, but none came to any real conclusion. The main issues that were discussed in the autumn Appalachian air related to the application of information theory to the genetic code. Gamow and Yčas were there, although they were the only members of the coding community who attended – none of the experimentalists who were involved in trying to crack the code made the trip.
Yockey began uncontroversially by outlining what he called the coding chain: protein specificity was encoded by the order of amino acids, which in turn was encoded by the order of base pairs, which in turn implied the existence of a code that translated from four ‘letters’ (A, C, G and T) to twenty ‘words’ (the amino acids). Then, with the confidence of a theoretician unhindered by experimental facts, Yockey suggested that cracking the code was a problem for mathematicians, not biologists:
Thus by following the logical consequences of purely biological, or perhaps biochemical, problems, one is led directly to a problem purely mathematical in character.39
Yockey claimed that ‘the central ideas of this paper are independent of much of the detail embodied in Watson and Crick’s papers’ – the coding problem had no necessary link with biochemical reality, he seemed to be suggesting. Nonetheless, he boldly claimed that it was possible that ‘the role information theory will play in biology will parallel that played by thermodynamics in physics and chemistry.’40
Gamow and Yčas were less optimistic. Yčas looked at the information content of proteins, treating them as a text and applying cryptographic techniques to work out the genetic code that lay behind it. To Yčas’s dismay, there was no apparent consistency in the organisation and length of protein molecules, and an audacious attempt to predict protein diversity from RNA diversity failed dismally.41 In a joint paper with Yčas, Gamow summarised the molecular model of gene function that flowed from Watson and Crick’s work, but ultimately could only express his weary conviction that as more protein sequences became known, ‘this problem will be solved in one way or another’. The only way in which Gamow could conceive of cracking the code was by continuing to treat it as a mathematical puzzle that would eventually reveal a solution if
probed with sufficient ingenuity.
In contrast, the scientists at the meeting who were most directly involved in information theory were beginning to question whether the informational approach to biology was the right one at all. In so doing, they began to undermine some of the main points of Shannon’s original vision. It was widely known that Shannon was not enthusiastic about the application of his ideas to fields other than the strict realm of communication. Henry Quastler, who had organised the 1952 meeting and had been the intellectual driving force behind the biological application of Shannon’s ideas, countered this criticism by showing that information theory was an appropriate tool for biology: biological systems involve control, control depends on communication, and communication depends on information. This neat summary tied in with cybernetics, but brought in another problem: meaning.
The key point in cybernetics was that certain signals – for example negative feedback – had greater significance than others in determining how a system functioned. Meaning had been anathema to Shannon, who had insisted that information had to be considered at a completely abstract level for his theory to exercise its full power. But it seemed that in biological systems it was impossible to avoid meaning. In 1953 this point had been the focus of a highly critical paper presented to a meeting on cybernetics by the philosopher Yehoshuua Bar-Hillel.42 At the 1956 Oak Ridge meeting, the physicist Leroy Augenstine took up the argument and pointed out that not all ‘bits’ of information were equivalent – their meaning and context had to be taken into account: