Life's Greatest Secret

by Matthew Cobb

  At 3.00 a.m. on Saturday 27 May 1961, Heinrich Matthaei began one of the most significant experiments in the history of biology. Matthaei was a 32-year-old German researcher in Marshall Nirenberg’s laboratory at the National Institutes of Health in Bethesda, Maryland, and he was about to crack the genetic code. Together with Nirenberg, a biochemist only two years his senior, Matthaei was studying protein synthesis in a test tube. In the middle of the night, he took his protein-synthesising mix, added two radioactive amino acids, phenylalanine and tyrosine, to different tubes and then introduced a long string of man-made RNA that was composed of just one kind of base – uracil (U – this replaces the T found in DNA sequences). The RNA molecule therefore read ‘UUUUUUUU …’ and was known as poly(U). By seeing which radioactive amino acid was turned into a protein chain by poly(U), Matthaei hoped he would be able to read the first word in the genetic code. It did not matter whether the code used sets of one, two, three, four or more U bases: the ‘cell-free’ protein synthesis system in the test tube would be able to read the message.

  When Gordon Tomkins, the laboratory head, came in at around 9 a.m., Matthaei had the answer. Radioactive protein had been produced in the test tube containing the amino acid phenylalanine. This must mean that a combination of Us coded for phenylalanine. On 27 May 1961, Heinrich Matthaei had read the first word of the book of life.

  *

  Nirenberg and Matthaei’s discovery transformed the study of the genetic code, both because it was successful and because it used a radically different approach. Until this breakthrough, the campaign to crack the code had begun to look listless. In 1959, at a meeting held at the Brookhaven Laboratory, Crick had summarised what he called ‘the present position of the coding problem’. He divided coding research into three phases: the vague phase (up to 1954), the optimistic phase (opened up by Gamow) and the ‘present position’, which Crick called ‘the confused phase’.1

  The situation was confused because none of the theoretical models matched the increasingly complex experimental findings. For example, a study of nineteen different species of bacteria showed that they had very different ratios of bases in their DNA, but their RNA and amino acid compositions were essentially similar.2 Crick outlined a number of ‘unattractive’ explanations of this finding, including the possibility that the genetic code was not universal, or that only part of the DNA in an organism codes for protein, with the rest being ‘nonsense’. Crick was optimistic, however, and with Brenner was trying to create viral mutants that would give them an insight into the code. This approach had been given a boost in the summer of 1960, when Heinz Fraenkel-Conrat at Berkeley described the amino acid sequence of the tobacco mosaic virus, and began making mutations with the hope of observing amino acid changes. Although this approach would take a long time and had yet to provide any concrete insight into the code, the press had got interested and was hyping up the idea that the code could soon be cracked. In May 1960 Time magazine published two articles on the subject: the first was entitled ‘Close to the mystery’, and the second proclaimed that Fraenkel-Conrat’s work represented the ‘genetic Rosetta stone’.

  Meanwhile, there were still mathematicians who were convinced that it was possible to crack the code simply by thinking. Just six weeks before Matthaei completed the decisive experiment, there was a symposium in New York on ‘Mathematical problems in the biological sciences’, which included molecular biologists such as Max Delbrück and Alex Rich. One of the speakers was Solomon Golomb, a mathematician from the Jet Propulsion Lab in Pasadena, who had previously worked with Delbrück. Golomb described various theoretical schemes that might correspond to the actual genetic code, before concluding, ‘It will be interesting to see how much of the final solution will be proposed by mathematicians before the experimentalists find it.’3 The answer, worked out over the following seven years or so, was simple: not one single part of it.

  Nirenberg and Matthaei’s radical experimental approach to the coding problem was all the more notable because they were complete outsiders, unconnected with any of the groups that had been struggling with the coding problem over the previous eight years. They were not part of the golden trio of Cambridge, Harvard and Paris, which had produced all of the main discoveries thus far.4 Nirenberg was so unknown that his application to attend the June 1961 Cold Spring Harbor meeting had been rejected. Ironically, while the great and the good of molecular biology were talking about the genetic code, Nirenberg and Matthaei were cracking it.

  *

  Nothing in Nirenberg’s early career suggested that he would be the man who would crack the genetic code. In 1951 he obtained his MSc for a study of caddis-fly biology, then he changed subject and did a PhD in biochemistry. Next he held a two-year postdoctoral fellowship at the National Institute of Arthritis and Metabolic Diseases (part of the National Institutes of Health) in Bethesda, a few miles northwest of Washington DC. After François Jacob and Joshua Lederberg both rejected his applications to work with them, Nirenberg became an NIH research biochemist in the Section of Metabolic Enzymes at Bethesda, led by the charismatic jazz fanatic Gordon Tomkins, who at 35 years old was barely older than Nirenberg.

  During those initial years at Bethesda, Nirenberg tried to induce the synthesis of an enzyme called penicillinase in Bacillus cereus bacteria. Nirenberg kept detailed lab diaries in which he noted his thoughts and ideas, giving an insight into how he approached his experiments.5 These diaries reveal that he had been thinking about his revolutionary approach to cracking the genetic code for more than two years before he was finally able to see the result. At the end of November 1958, he described his idea of getting protein synthesis to work in a test tube and he gave an outline of the ideal experiment he wanted to perform:

  would not have to get polynucleotide synthesis very far to break the coding problem. Probably 30 nucleotides & equal number of AA would do it. Could crack life’s code!6

  Nirenberg’s approach was predicated on recent results from the laboratories of Paul Zamecnik and Severo Ochoa. In the early 1950s, Zamecnik had achieved a technical tour-de-force by making protein synthesis take place in a test tube. Zamecnik’s ‘cell-free’ system, based on the contents of rat liver cells, used radioactive amino acids to show that a new protein was synthesised.7 The second element of Nirenberg’s approach came from the research of Spanish-born Severo Ochoa, who worked in New York and won the 1959 Nobel Prize in Physiology or Medicine for the discovery of polynucleotide phosphorylase – an enzyme involved in the metabolism of RNA.8 The isolation of polynucleotide phosphorylase meant that Ochoa was able to create artificial RNA molecules by incubating the enzyme with the four RNA bases (A, C, G and U). It was not possible to determine the order in which the nucleotides were strung together, but it was relatively straightforward to create a monotonous RNA molecule composed only of one base – known as poly(A), poly(U) and so on.

  For most people, it was not immediately obvious what could be done with such molecular freaks of nature – nothing like them had ever been observed in any cell – but Nirenberg glimpsed the opportunity. Nirenberg was clever and he was lucky: Ochoa was synthesising poly(A), poly(U) and so on together with Leon Heppel – the head of biochemistry at Bethesda. Heppel’s lab began to produce synthetic RNA molecules in collaboration with another Bethesda researcher, Maxine Singer.9 Nirenberg found himself in one of the two places in the world that was making these unearthly strings of RNA.

  Despite his understandable desire to attack the genetic code immediately, Nirenberg tried to keep his focus on the research he was supposed to be doing. As he reminded himself in an entry in his lab diary from spring 1959: ‘My main aim is not to crack protein synthesis but to have everything ready to study enzyme induction.’10 At the spring 1960 meeting of the Federation of American Societies for Experimental Biology (FASEB), Nirenberg gave a brief talk on his work on induction.11 His aim was to see whether the same gene was involved in the synthesis of two very similar inducible enzymes, or, as he put it, ‘whether a
portion of one gene contains information for the synthesis of a protein subunit which might be an integral part of two or more enzymes’. To Nirenberg’s disappointment, there was no proof of what he termed ‘shared genetic information’.

  Although this finding was not particularly interesting (the talk has never been cited), the way in which Nirenberg approached the problem is significant because it highlighted the two ways of looking at life that were now happily coexisting in laboratories all over the world. Everyone thought that genes contained information, but that abstract quality also had a concrete form: it was a nucleotide sequence that made something happen. In this case it was what Nirenberg called ‘information for the synthesis of a protein subunit’. Whatever the advantages of the new way of thinking about genes as information, in the end those ideas would have to be translated into detailed, dirty biochemistry.

  Nirenberg’s outlook changed completely in August 1960, when Zamecnik showed that it was possible to get protein synthesis occurring in a test tube containing the contents of Nirenberg’s favoured organism, Escherichia coli.12 Nirenberg immediately began to try the experiment at Bethesda. He wrote in his diary: ‘Hurry up exps. Shouldn’t take 1 week to know whether system will work. Work-Work-Work.’13 But it would not work. Then he had two strokes of luck: first, Alfred Tissières and François Gros at Harvard published a refinement of Zamecnik’s system that was much easier to use, and then a lanky, prematurely balding German called Heinrich Matthaei joined his lab.14 Matthaei had obtained a NATO fellowship to work on cell-free protein synthesis, using radioactively labelled amino acids.15 He had initially been working on carrots but various things had gone wrong, and he eventually found himself assigned to Nirenberg. Matthaei had exactly the technical skills that Nirenberg needed.

  Shortly after Matthaei’s arrival, Nirenberg abandoned his ideas of looking at induction in bacterial cells and threw himself into an exploration of protein synthesis in cell-free E. coli-based systems. Within weeks, the pair had made a technical breakthrough as they were able to make bulk enzyme extracts and then store them, rather than having to make fresh extracts for every experiment. This soon led to a steep increase in the number of experiments they could do.16

  By end of November 1960, Nirenberg’s diaries were full of discussions about cell-free systems, the importance of messenger RNA, and the use of synthetic RNA as a key: ‘Can you swamp system with messenger RNA?’ he wrote.17 This is striking, because Nirenberg was writing several months before the May 1961 publication of the Nature papers by Brenner, Jacob and Crick and by Gros and Watson, which first publicly used the term messenger RNA. ‘Cytoplasmic messenger’ had been used in the 1959 PaJaMo paper, and by the end of 1960 Jacob and Monod had arrived at the concept of messenger RNA, at which point the term was being bandied about in conferences.18 But the exact phrase had not yet been used in print. Even though Nirenberg was not part of the inner circle of molecular biology, he had picked up on the term before there was conclusive evidence that the substance existed.

  In one respect, this relative distance from the intellectual centres of the work on the genetic code proved an advantage. Nirenberg was blissfully unaware of the debates among those working on the coding problem over the structural restrictions imposed by what was called a commaless code. In 1957, Crick, Leslie Orgel and J. S. Griffith had theorised that if the code was composed of words made of three bases, as many people thought, and there were no bases between the triplets that acted as commas indicating the separate words, then triplets composed of the same base (AAA or UUU, for example) were forbidden because the cellular machinery would not know where to start reading. With various other ad hoc restrictions, partly based on chemistry, Crick’s theoretical scheme allowed merely twenty combinations out of the sixty-four possible combinations of bases. As there were twenty naturally occurring amino acids, this was aesthetically very pleasing, but it was entirely speculative. As Crick noted at the time, this ‘gave the magic number – twenty – in a neat manner’ but the ‘arguments and assumptions’ behind the theory were ‘too precarious for us to feel much confidence in it’.19 A few months later he admitted:

  I find it impossible to form any considered judgement of this idea. It may be complete nonsense, or it may be the heart of the matter. Only time will show.20

  Time showed it to be complete nonsense. More importantly, the idea may have restricted what researchers thought was possible, in particular because it ruled out triplets composed of the same base. For most people trying to crack the code, investigating the effect of a polynucleotide containing only one kind of base would have been pointless.

  *

  By the end of 1960, Nirenberg and Matthaei were working through the night on protein synthesis in E. coli extracts. In mid-January, one of Nirenberg’s diary entries was headed ‘Idea. Approach to code’ and outlined the use of poly(A), poly(U), poly(C) or poly(G) and of poly(AG), etc.; poly(AG) would be composed of equal amounts of A and G, in an unknown sequence. Nirenberg’s aim was to put polynucleotides into his cell-free protein synthesis system and use the output to understand the nature of the genetic code, starting with establishing the number of bases involved in coding for an amino acid:

  Might be able to get enough info to establish limits of a code … If you need all 4 bases, could not be triplet code.21

  At the FASEB Meeting in February 1961, Matthaei and Nirenberg gave a brief talk describing the way their system incorporated a 14C-labelled amino acid (valine) into a protein.22 A few weeks later, on 22 March, they submitted an article on the topic to Biochemical and Biophysical Research Communications.23 This journal had been set up in the previous year to respond to the increasingly fierce competition in the area by offering rapid publication of short articles – it used camera-ready typed copy provided by the authors rather than traditional typesetting, thereby speeding up the whole publishing process.

  The article described how the presence of ribosomal RNA in their cell-free system was essential for amino acid incorporation, which ‘had many characteristics expected of protein synthesis’, and that ‘all of the activity appeared to be associated with RNA’.24 Exactly what kind of RNA Nirenberg and Matthaei were referring to was not entirely clear. They concluded, somewhat confusingly ‘It is possible that part or all of the ribosomal RNA used in our study corresponds to template or messenger RNA.’25 By ribosomal RNA they did not mean the RNA that makes up the ribosome itself, but rather an RNA molecule that was attached to the ribosome. The ambiguity contained in the term ‘template or messenger RNA’ is not simply a matter of uncertainty over which word to choose. As Lily Kay has pointed out, Nirenberg and Matthaei’s use of ‘template or messenger RNA’ shows that their language was poised at a cusp between the old, physical, way of thinking of specificity – as a structural template – and the new, abstract idea of information being transferred by a messenger.26 Understandably, these semiotic niceties were not noticed at the time, and the paper made no impression – it was not cited until 1963, by which time all the dust had settled.

  In early May 1961, Nirenberg and Matthaei decided to add RNA from the tobacco mosaic virus (TMV) to see if they could get the cell-free system to synthesise TMV protein. It worked like a dream. As Nirenberg recalled in the 1970s, the results were ‘superb … beautiful … It was superbly active’.27 He realised that they would need to collaborate with the Berkeley TMV expert Fraenkel-Conrat if they were to fully exploit this novel approach. In the meantime, they continued to crank through the effects of the various synthetic RNA molecules that they were able to borrow from Leon Heppel’s laboratory next door.

  9. Matthaei’s notebook showing the results of the crucial experiment – reproduced from Kay (2000)

  In the middle of May, Nirenberg left to spend a month in Fraenkel-Conrat’s lab in Berkeley, getting himself up to speed with TMV. Back in Bethesda, Matthaei started a set of experiments to study the response of the cell-free system when it was seeded with artificial RNAs. On 15 May (his 32nd birthday), Mattha
ei began an experiment testing the effect of poly(A) (AAAAA…), poly(U) (UUUUU…), poly(2A)U (a ratio of two A nucleotides to one U, randomly distributed through the RNA molecule), and poly(4A)U (a randomly distributed ratio of four A nucleotides to one U). When all twenty amino acids added to the test tube were radioactive, Matthaei obtained a twelvefold increase in radioactivity in the protein product after incubation with poly(U), a small increase with poly(AU), and barely any change with poly(A). Something was going on in the poly(U) tube, which could explain how genetic information leads to a particular protein being created. To detect which radioactive amino acid had been incorporated into the protein that was produced by the cell-free set-up, Matthaei had to test all twenty amino acids systematically. He did this by putting ten radioactive amino acids into a test tube – the other ten amino acids added were the usual ‘cold’ versions. He then did the poly(U) experiment again. If there was an increase in radioactivity, then clearly one of the ‘hot’ amino acids was involved. By repeating this process, Matthaei was finally able to narrow the effect down to one of two amino acids: phenylalanine or tyrosine.

  On Saturday 27 May at 3.00 a.m., Matthaei began the final experiment. This involved ten test tubes and was labelled ’27-Q’ in his lab book. In tube number 3 he had nineteen unlabelled amino acids together with radioactive phenylalanine, and in tube number 8 he had nineteen unlabelled amino acids together with radioactive tyrosine. The remaining eight tubes contained various controls to prove that the effect was due to the combination of the poly(U) and one of the two radioactive amino acids. Matthaei allowed the mixture to incubate for one hour at 36°C; then he began the tedious task of isolating the protein produced by the reactions and measuring the radioactivity that had been incorporated. Tube 3, which contained radioactive phenylalanine, produced a protein with a radioactivity level that was more than twenty times higher than the control tubes; the protein from the tyrosine tube showed no increased radioactivity. When Gordon Tomkins came into the lab a few hours later, Matthaei told him the news: poly(U) coded for phenylalanine. The first word in the genetic code had been read.