by Matthew Cobb
Recent developments in automation have led to the use in industry of machines capable of performing operations that have been compared with certain types of human activity. In the internally regulated machine, as in the living organism, processes are controlled by one or more feedback loops that prevent any one phase of the process from being carried to a catastrophic extreme. The consequence of such feedback control can be observed at all levels of organization of a living animal’.11
For Umbarger, the relatively simple system of bacterial biosynthesis provided an opportunity to explore the molecular mechanisms involved in negative feedback, and he was sure that the example he described – the biosynthesis of isoleucine – was just one case among many.
Although all these researchers contributed to the shift in thinking about how protein synthesis worked, the people who linked the ideas of feedback and genetic information, changing our view of life and of the genetic code, were Jacob and Monod, the new Paris team.
Monod had realised that for a full understanding of protein synthesis he needed the help of a geneticist, so he contacted a colleague at the Institut Pasteur in Paris, François Jacob. Jacob, a physician who had signed up with de Gaulle’s Free French in 1940 and had been severely wounded after D-Day, joined André Lwoff’s laboratory at the Institut Pasteur in 1950 to study the interactions between bacteriophages and their single-cell hosts.12 Jacob’s skills in bacterial genetics and his wide-ranging philosophical interests formed a perfect complement to Monod’s more biochemically centred approach and his interest in existentialism. The result was an intellectual partnership that rivalled that of Watson and Crick and which in some ways surpassed it in terms of providing a model to young researchers around the world – Crick himself called it ‘the great collaboration’.13
Jacob and Monod’s joint research, which began in 1957, took place in the heart of Paris, in an Institut Pasteur attic laboratory that was nicknamed the grenier (loft). Jacob and Élie Wollman had been studying how bacterial mating (‘conjugation’) affected the growth of bacteriophage viruses; with Monod, Jacob now used conjugation to explore the genetic basis of induction in bacteria. They carried out these experiments in late 1957 and early 1958 with one of many US visitors to the Institut, Arthur Pardee – the studies became widely known as the PaJaMo (Pardee, Jacob and Monod), or, more colloquially, Pajama (or even Pyjama), experiments.14
Pardee had arrived in Paris in September 1957, and began studying how one of Jacob and Monod’s bacterial mutants responded to induction.15 Normal bacteria could digest lactose by producing induced β-galactosidase. These cells were called lac+ – lac (short for lactose) referred to a region of the bacterial chromosome containing several genes involved in this complex phenomenon, and the ‘+’ indicated that this was the normal, or wild, type. Mutant bacteria – known as lac– – could not grow on lactose unless they acquired the relevant genes by mating with a lac+ individual. Pardee showed that when the z+ gene, which produced the β-galactosidase enzyme, was transferred into a lac– individual, it became active within minutes. This implied that there was an immediate chemical signal that passed directly from the introduced gene to the host cell’s protein synthesis system. Over the next year or so, the Paris group became focused on the nature of this mysterious messenger molecule, which they called X.16
The PaJaMo experiments also investigated bacteria that continually produced β-galactosidase in the absence of an external inducer molecule. These were known as constitutive strains because they produced the enzyme as part of their constitution. A single gene called i seemed to be involved: i+ bacteria were inducible, whereas i– individuals were constitutive (the system also involved another gene, y, which controlled the action of an enzyme, permease, that allowed lactose into the cell). Things got interesting when the group explored the interaction between the i gene and the z gene that allowed bacteria to produce β-galactosidase. When Pardee introduced both i+ and z+ genes into bacteria carrying the i– and z– forms, the bacteria initially produced high, constitutional, levels of β-galactosidase, showing the action of the z+ gene. But then something odd happened: β-galactosidase production declined rapidly. The i+ gene seemed to start repressing the activity of the z+ gene.17 To the untutored eye, this complex set of results seems either bewildering or boring, or both. But what Jacob and Monod did with these data – the way in which they interpreted them – altered our understanding of what genes do.
The first step forward in this new view of life came at the beginning of 1958, when Leo Szilárd was visiting Paris as part of a tour of laboratories during which he presented his negative feedback model of protein synthesis.18 Five years earlier, Szilárd had proposed to Monod that a form of negative feedback might explain induction; now he made a mind-twisting suggestion that expressed the idea at an even more complex level. Perhaps, he said, ‘induction could be effected by an anti-repressor rather than repression by an anti-inducer?’19 Szilárd was proposing that induction might work by stopping the action of a molecule (a ‘repressor’) inside the bacterial cell that normally repressed enzyme formation. The effect would be a bit like releasing a brake. This could be understood as an example of two negatives producing a positive or, as Monod put it in his 1965 Nobel Prize lecture, a ‘double bluff’.20
Szilárd had not come up with this idea himself. In April 1957, the bacteriologist Werner Maas had given a lecture in Chicago that Szilárd had attended; Maas, who was studying inducible enzymes, had hypothesised that:
inducers which enhance the formation of an enzyme when added to a growing bacterial culture may perhaps be capable of doing so only because there is a repressor present in the cell, and that the inducer might perhaps do no more than inhibit some enzymes.21
Maas later stated that on hearing this, Szilárd ‘jumped on the hypothesis … and became quite excited’.22 Szilárd wanted to publish the idea straight away, but Maas refused, because he had no evidence to back it up. At the same time, Henry Vogel of Yale University submitted an article in which he suggested that a common theoretical framework could understand both induction and negative feedback inhibition of biosynthesis – they involved what Vogel called regulator molecules.23 Although Maas’s idea of a repressor and Vogel’s regulatory framework were both based on interactions between proteins, not genes, the ideas of regulation and repression were in the air.
By the time that Szilárd visited Paris less than a year later, he was clearly convinced that induction might not be a positive effect, but rather a ‘de-repression’. The Paris group were intrigued by this suggestion and they briefly outlined the concept alongside the first publication of the PaJaMo results (in a French journal in May 1958, where it was described as an ‘initially surprising hypothesis’) and again in a lecture by Jacob in June 1958. Even though the team were prepared to go public with the idea, they had made no experimental test of the hypothesis and it was still not certain that there were any general implications beyond the narrow world of bacterial genetics.24
Jacob later recalled that the decisive moment came on a Sunday afternoon late in July 1958. All his colleagues were on holiday, while he remained in Paris with his wife, Lise, preparing for a lecture he had to give in New York on how phage viruses hijack the genetic apparatuses of their host. Unable to work, Jacob went to the cinema with his wife. He could not concentrate on the film, so he closed his eyes and suddenly in ‘a flash’, he realised that the two experiments he had been thinking about – the PaJaMo experiment and his own work with Élie Wollman on phage reproduction – were in fact fundamentally identical. He now understood that they both involved the modulation of gene activity by directly affecting the DNA. Jacob later described his almost mystical experience as he realised the connection between his two problematic experiments:
Same situation. Same result. Same conclusion. In both cases, a gene governs the formation of a cytoplasmic product, of a repressor blocking the expression of other genes and so preventing either the synthesis of the galactosidase or the multiplica
tion of the virus. … Where can the repressor act to stop everything at once? The only simple answer, the only one that does not involve a cascade of complicated hypotheses is: on the DNA itself! … These hypotheses, still rough, still vaguely outlined, poorly formulated, stir within me. Barely have they emerged than I feel invaded by an intense joy, a savage pleasure. A sense of strength as well, of power. As if I had climbed a mountain, attained a summit from which I saw in the distance a vast panorama. I no longer feel mediocre or even mortal. I need air. I need to walk.25
It was over a month before Jacob could share his insight with Monod. When the two men finally met up, in September 1958, Monod was initially unconvinced. The link between the two experiments was tenuous, and the idea that the repressor acted directly on DNA seemed outlandish. Genes had previously been seen as pure and abstract entities, solidly placed in the background of the cell, passive repositories of information and nothing more – as Monod later put it, they were thought to be as inaccessible ‘as the material of the galaxies’.26 According to Jacob’s new view, genes were intimately involved in the messy reality of cellular processes. Over weeks and months of increasingly intense argument and repeated cycles of experimentation, Jacob and Monod explored their ideas, testing hypotheses by creating new mutants and predicting how they would behave if the model were correct. Jacob later recalled their discussions ‘moved at top speed, in bursts of brief retorts, like a ping-pong match’. As they scribbled feverishly on the blackboard, the model and its predictions became clearer.27 The language they were using changed, too.
Monod began to describe gene function in terms of information transfer, and he explicitly embraced cybernetic thinking by describing protein synthesis in terms of patterns of control. This shift took its clearest shape in a plan for a book entitled Essays in Enzyme Cybernetics, which he began writing with Melvin Cohn of Washington University towards the end of 1958.28 Even though the project was never completed, its very existence was extremely revealing. Ephrussi and Watson’s weak satirical joke that had fooled the editors of Nature six years earlier had become a reality: cybernetics was being used to understand biology. This was not a reflection of some fashionable trend; it was a powerful interpretative approach that had very real advantages for understanding biological processes. However, as with information theory, it was the general framework, rather than the precise mathematical detail, that was being employed. For biologists, cybernetics was becoming an analogy, a metaphor, a way of thinking about life in terms of flows of control and information, a way of thinking about how genes worked.
Jacob used a military analogy to explain the genetic ‘circuit’ he had discovered with Monod, unwittingly returning to Wiener’s original recognition of the role of feedback through his work on anti-aircraft fire control:
We saw this circuit as made up of two genes: transmitter and receiver of a cytoplasmic signal, the repressor. In the absence of the inducer, this circuit blocked the synthesis of galactosidase. Every mutation inactivating one of the genes thus had to result in a constitutive synthesis, much as a transmitter on the ground sends signals to a bomber: ‘Do not drop the bombs. Do not drop the bombs.’ If the transmitter or the receiver is broken, the plane drops its bombs. But let there be two transmitters with two bombers, and the situation changes. The destruction of a single transmitter has no effect, for the other one will continue to emit. The destruction of a receiver, however, will result in dropping the bombs, but only by the bomber whose receiver is broken.29
A few months later, Monod gave a lecture in Germany and described the model in more precise terms, adopting the vocabulary of information and control: ‘We could imagine that the z locus contains genetic information relating to the structure of the galactosidase protein, while the i locus determines the conditions under which this information is potentially transferred to the cytoplasm.’ By implication, Monod was arguing that that the ‘one gene, one enzyme’ idea – less than 20 years old – could not explain the complex reality of protein synthesis. Instead, he suggested there were two kinds of genes: structural genes, which contained the information necessary to make proteins, and regulator genes, which determined when that information was employed, by synthesising a specific repressor that inhibited the expression of the structural gene.30
In March 1959, the final version of the PaJaMo experiment was submitted to a new academic publication, which had a title that was a manifesto for the new science: the Journal of Molecular Biology. In the paper, which was more developed than the original French publication, the trio acknowledged that they were ‘much indebted to Professor Leo Szilárd for illuminating discussions’. The paper presented the dense details of their experiment and outlined the repressor hypothesis, suggesting that it was a general model for protein synthesis and showing the parallels with the phage work of François Jacob and Élie Wollman. Above all, they highlighted two points that they could not answer: the nature of the repressor and how it worked.31 As they explored these issues, Jacob and Monod changed the vocabulary that was used for talking and thinking about genes and what they – and the code they contain – might do.
*
For much of the 1950s, scientists had felt uncomfortable about the word ‘gene’. In 1952, the Glasgow-based Italian geneticist Guido Pontecorvo highlighted the existence of four different definitions of the word that were regularly employed by scientists and which were sometimes mutually contradictory. A gene could refer to a self-replicating part of a chromosome, the smallest part of a chromosome that can show a mutation, the unit of physiological activity or, finally, the earliest definition of a gene – the unit of hereditary transmission.32 Pontecorvo questioned whether the gene could any longer be seen as a delimited part of a chromosome, and suggested instead that it was better seen as a process and that the word gene should therefore be used solely to describe the unit of physiological action.
Although Pontecorvo’s suggestion was not taken up, scientists recognised the problem. The debate over words and concepts continued at the Johns Hopkins University symposium on ‘The Chemical Basis of Heredity’, which was held in June 1956. By this time it was generally accepted as a working hypothesis that all genes in all organisms were made of DNA and that the Watson–Crick double helix structure was also correct. Joshua Lederberg, a stickler for terminology, declared audaciously that ‘“gene” is no longer a useful term in exact discourse’.33 He would no doubt be surprised to learn that it is still being used, more than half a century later. At the same meeting, Seymour Benzer came up with a solution:
The classical ‘gene’, which served at once as the unit of genetic recombination, of mutation, and of function, is no longer adequate. These units require separate definition.34
Since 1954, Benzer had been studying the structure of genes, focusing on the rII genetic region of the T4 phage virus.35 From today’s perspective, it is noteworthy that at the 1956 meeting Benzer was still hedging his bets over the nature of the genetic material in T4: all he was prepared to say at that point was that in this virus DNA ‘appears to carry the hereditary information’.36 By screening hundreds of mutants in the rII region of the virus’s DNA and then carrying out thousands of crosses to see whether they were part of the same functional unit, Benzer was able to construct an extremely detailed genetic map, down to a single pair of nucleotides. Like Rutherford showing that the atom had an internal structure through his work in Manchester at the beginning of the twentieth century, Benzer was able to show that the gene was not a single, indivisible unit.37 Rather than the classic image of the gene as a bead on a string, Benzer saw the gene as a one-dimensional stretch of DNA, which would reveal its secrets by dissecting it down to the molecular level. He found that not all parts of the gene were equal – some areas were much more prone to spontaneous mutations, and mutations in different areas produced different effects. Long before the development of DNA sequencing, Benzer’s pioneering and painstaking study showed that genes have an internal structure.
On the basis o
f this work, Benzer came up with new words, focusing on what genes actually did: the unit of genetic recombination was a ‘recon’, the smallest unit of mutation was a ‘muton’, and the unit of function was a ‘cistron’.* Although only ‘cistron’ survived into common scientific usage for a while (it is now very much on the wane), Benzer’s attempt to reconceive the gene in molecular terms was highly influential. Widely praised at the time – at the Johns Hopkins symposium, George Beadle described it as ‘very beautiful work’ – Benzer’s approach helped fuse the structural insights of Watson and Crick with the traditional approaches of genetics, creating the new subject of molecular genetics.38
This new field was reinforced in 1957 by two PhD students, Matthew Meselson and Frank Stahl, who carried out what has been described as ‘the most beautiful experiment in biology’.39 One of the problems raised by the double helix structure was how the DNA molecule copied itself. The complementarity of the base pairs on the two strands suggested that the cell used each strand as a template to create two identical double helices, but how this worked was unclear. At the same 1956 Johns Hopkins meeting at which Benzer spoke, Max Delbrück outlined three models for DNA replication – ‘conservative’, in which the original DNA double helix remained intact and was entirely copied into a completely new molecule; ‘semi-conservative’, in which one strand of each molecule was copied, producing two daughter molecules, each of which had one old and one new strand (this was the model suggested by Watson and Crick); and Delbrück’s preferred view, ‘dispersive’ replication, whereby bits of each DNA strand were copied and then reassembled, producing two double helices, each strand of which was made up of a mixture of old and new.40 Meselson and Stahl’s experiment was designed to distinguish between these three hypotheses.