by Andrew Brown
One of the substances that Bernal studied in 1931, at Astbury’s behest, was the ring-shaped diketopiperazine molecule. He had concluded, correctly, that diketopiperazine is ‘built from centro-symmetrical, almost flat, hexagonal molecules linked together in ribbons by their residual electrical forces’.10 Corey confirmed Bernal’s finding that the molecule was indeed flat, and then extended the earlier work by accurately determining the lengths and angles of the inter-atomic bonds. He went on to establish that the C–N peptide bond in all the structures he analysed was planar – the atoms adjacent to peptides bonds had to lie in the same plane, a crucial constraint on any protein structure.11
Pauling’s intention had been to use his ‘knowledge of structural chemistry to predict the dimensions and other properties of a polypeptide chain and then to examine possible conformations of the chain, to find one that would agree with the X-ray diffraction data’.12 Despite many hours of strenuous thought, the solution eluded Pauling until he was laid low by sinusitis, while a visiting professor to Oxford in March 1948. Bored with reading detective stories, he asked his wife for a pencil, a piece of paper and a ruler. He set out to sketch a polypeptide chain that would be spiral in shape, yet would show no rotation on either side of a peptide bond. The whole structure would be stabilized by hydrogen bonds on the outside of the spiral. Pauling rolled his paper up and found that these conditions would all be met by a spiral containing about 3.7 amino acid residues per twist, in a pattern that repeated every 5.4 Å. The idea of a non-integral pattern was novel, but not impossible: the larger objection lay in the miniscule discrepancy between 5.4 Å, and Astbury’s well-established 5.1 Å.
Work on protein structure in the British schools of X-ray crystallography was a dispiriting business in the late 1940s. At Oxford, the wartime work on penicillin had diverted Dorothy Hodgkin away from insulin; her next pick was the vitamin B12 molecule – medically important and much smaller than insulin. At Birkbeck, Bernal had brought Carlisle a tube containing several hundred crystals of ribonuclease from a Rockefeller chemist, Moses Kunitz, who had been the first to crystallize the enzyme. The enzyme was recovered from the pancreas of cows and appeared to be chemically stable and rather small (molecular weight 13,000), making it a promising candidate for X-ray analysis. As it turned out, Carlisle and his group would take twenty years to obtain the definitive structure of ribonuclease.13 Even at Cambridge, the redoubtable Perutz was reluctantly facing ‘the stark truth that the years of tedious labour, the many nights of interrupted sleep and the appalling strain of measuring the intensities of thousands of little black spots by eye had brought me no nearer to the solution of the structure of haemoglobin, and that I wasted some of the best years of my life trying to solve a seemingly insoluble problem’.14
In 1949, Perutz did find evidence of an Astbury-type 5.1 Å repeat in crystals of horse haemoglobin which led him to conclude ‘rashly that the structure of haemoglobin consisted of a set of close-packed, α-keratin-like polypeptide chains’.15 His colleague, John Kendrew, who had just started studying the smaller myoglobin molecule from horse muscle, soon made the same interpretation of that protein. Perutz and Kendrew were the only two permanent members of the MRC Unit at the Cavendish Laboratory, where Sir Lawrence Bragg was still the chief. One day Bragg arrived in their lab ‘armed with a broomstick into which he had hammered a helical pattern of nails repeating at intervals of 5.1 cm along its axis, representing what he suspected was the repeat of amino acid residues along the fibre axis of α-keratin’.16 Kendrew and Perutz attempted to build models, with either two, three or four amino acid residues per turn of the helix, but none was a good fit. Even without knowing Corey’s finding that the peptide bond was always planar, the angles of the bonds in their models seemed to be under strain. Bragg remained optimistic that they were on the verge of a breakthrough and he wrote to Sage about it:
Perutz, Kendrew and I have been rather excited lately because we feel we are getting somewhere with the nature of the actual protein chain.… If we accept Perutz’s results about the spacings along the chains and the distances between them, it implies that we have got to fit 3 amino acid residues per 5 Å of length, and the possibilities boil down to very few… Incidentally, Perutz has tried to make models of the Astbury chain, and they do not seem to work too well.17
When the Cambridge group eventually published the inconclusive results of their modelling work, Pauling read the paper and immediately saw its flaws. Bragg’s group had not been restrictive enough in applying modern chemical concepts, and he now felt the time had come to publish his own model even though he had no experimental data of his own to support it. He alerted Bernal to the new model in a letter dated 13th June 1951:
Have you seen the series of 8 papers that Professor Corey and I have got out on the structure of proteins? They are in the April and May issues of the Proceedings of the National Academy of Sciences. I think that there is very little doubt about the correctness of these structures. So far we have not found any configurations of polypeptide chains in proteins other than those described in these papers.18
Sage was not entirely convinced.
I have read your papers with Corey with the greatest interest and I certainly think that you have made the point that it is not necessary that the residues should follow each other along the chain with any regular crystallographic repeats. On the other hand, I am rather doubtful on the basis of Carlisle’s work here, that any structure with as many as 3, let alone 3.7, residues per turn can be fitted into the data for some crystalline proteins such as ribonuclease and chymotrypsin. The number we get by estimation of the number of chains and the periodicities along the chain, in this case about 5.4 Å, indicates a figure much more like 2.19
To Max Perutz, who had been so frustrated by trying to build models for Bragg that would comply with Astbury’s concept of α-keratin, Pauling’s announcement of his α-helix with a fractional number of amino acids per complete turn was revelatory. He was ‘thunderstruck by Pauling and Corey’s paper. In contrast to Kendrew’s and my helices, theirs was free of strain… the structure looked dead right.’20 But there were no experimental data to support it. Perutz thought about the Pauling model in terms of a spiral staircase with each amino acid residue as a separate step. Each step would be 1.5 Å high and ‘according to diffraction theory, this regular repeat should give rise to a strong X-ray reflection of 1.5 Å spacing from planes perpendicular to the fibre axis’.21 Why had no one, Astbury in particular, ever seen these reflections? The answer came in a flash – Astbury always orientated his protein fibres with their long axis perpendicular to the X-ray beam (in fact, an oblique angle was required) and his flat plate camera had too narrow an aperture to capture the image.
In mad excitement, I cycled back to the lab and looked for a horse hair that I had kept tucked away in a drawer. I stuck it on a goniometer head at an angle of 318 to the incident X-ray beam; instead of Astbury’s flat plate camera I put a cylinderical film around it that would catch all reflections with Bragg angles of up to 858.22
After exposing the horse hair in the X-ray beam for about two hours, Perutz developed the film, with ‘his heart in his mouth’, and there were the lines he predicted. A week after Bernal had written his letter expressing reservations to Pauling, Nature published a letter from Perutz offering the first experimental substantiation of the α-helix.23 In this letter, Perutz included some preliminary results showing that even the globular haemoglobin molecule contained stretches of α-helix.
Perutz continued to worry about the discrepancy between Astbury’s 5.1 Å and Pauling’s 5.4 Å. The answer was supplied by his PhD student, Francis Crick, in 1952 when he came to the lab ‘with two rubber tubes around which he had pinned corks with a helical repeat of 3.6 corks per turn and a pitch of 5.4 cm. He showed me that the two tubes could be wound around each other to make a double helix such that the corks neatly interlocked. This shortened the pitch of the individual chains, when projected onto the fibre axis, from 5.4 to 5.1 c
m, as required by the X-ray pattern of α-keratin.’24 In 1952, Crick referred to this arrangement as a ‘coiled coil’25 – the double helix would be next year’s model.
In the summer of 1951, there was a meeting of the leading British protein researchers at the Cavendish. Crick was going to make his first presentation in such exalted company and was understandably nervous. His supervisor, Perutz, spoke before him and described the latest findings on haemoglobin. When Bernal rose to comment on Perutz’s talk, Crick was astonished by his mild manner:
I regarded Bernal as a genius. For some reason I had acquired the idea that all geniuses behaved badly. I was therefore surprised to hear him praise Perutz in the most genial way for his courage in undertaking such a difficult and, at that time, unprecedented task and for his thoroughness and persistence in carrying it through. Only then did Bernal venture to express, in the nicest possible way, some reservations.26
When Crick did speak, his contribution was daring. His theme ‘broadly speaking, was that they [the X-ray crystallographers] were all wasting their time and that… almost all the methods they were pursuing had no chance of success’.27 The barrier to progress that Crick analysed more rigorously than some of his teachers was the old problem of the phases of the diffracted X-rays. This is a complicated mathematical concept, but a crude analogy would be with setting the correct focus of a light microscope. The spots in an X-ray diffraction pattern gives some measure of intensity, but without knowledge of the peaks and troughs of the reflected X-ray waves, Crick realized that no worthwhile three-dimensional representation of protein molecules would ever be possible. A solution to the phase problem had been successfully applied in the 1930s to solve the relatively simple structure of an organic dye by introducing a heavy atom, nickel, into its crystal. The principle of what became known as isomorphous replacement is that by adding a heavy atom (which gives a strong diffraction) to an identical position in all unit cells of a crystal, a measurable change will be produced in the diffraction pattern. Comparison of this new heavy atom pattern with that of the native crystal allows the phase of each spot to be determined mathematically: the added heavy atom will make each spot in the pattern stronger or weaker depending on phase.
Bernal had raised the possibility of applying the isomorphous replacement method to protein structure in his 1939 talk at the Royal Institution:
Unfortunately, however, direct analysis of these X-ray photographs [of protein crystals] is rendered impossible by the fact that we can never know the phases of the reflections corresponding to the different spots. The ambiguity introduced in this way can only be removed by some physical artifice, such as the introduction of a heavy atom or the observation of intensity changes on dehydration, which have not hitherto been carried out in practice.28
Crick, despite his junior status, was more forceful at the 1951 Cavendish meeting. He had confided in Kendrew what he was going to say, and Kendrew suggested that he title his talk ‘What mad pursuit’, which he did.
Bragg was furious. Here was this newcomer telling experienced X-ray crystallographers, including Bragg himself, who had founded the subject and been in the forefront of it for almost forty years, that what they were doing was most unlikely to lead to any useful result. The fact that I clearly understood the theory of the subject and indeed was apt to be unduly loquacious about it did not help.29
At the time, Crick felt that Bernal was not paying attention, but in later years he found that Bernal was the only one to remember that he had given his colleagues ‘a very necessary jolt’ in the right direction. A few months later, the leading British protein investigators decided to invite Pauling to London to defend the α-helix model to all at a Royal Society meeting. The date was set for 1st May 1952, but the guest of honour was prevented from attending by the US State Department, who refused to grant him a passport because they believed him to be a communist. In his introductory talk as chairman of the meeting, Astbury clung tenaciously to the primacy of his 5.1 Å repeat distance and was concerned that the ‘Pauling–Corey spiral’ failed to conform to this experimental fact. A pair of papers from Caltech followed, but Pauling’s assistant, Eddie Hughes, who delivered the second one in Pauling’s stead was flustered by lack of preparation and the grandeur of the venue; for the rest of the afternoon, he had to endure British scientists ‘telling us what was wrong’30 with Pauling’s model. Bernal, in his talk,31 expressed some scepticism about Pauling’s claim ‘that purely chemical considerations do, in the case of proteins, limit almost to one model the possible configuration of the protein chain. It can only be said at this stage that however well this model can account for the data for fibrous proteins – natural and synthetic – it has not yet been fitted satisfactorily to any globular protein.’ Whatever their reservations about Pauling’s model of protein structure, the British scientists were unanimous in their resentment that he had been barred from coming to England, and Bernal sent a letter of protest to the State Department signed by thirty other scientists.
In time, Sage came to view Pauling’s irrational (non-integer) α-helix as a revolutionary event in the field of molecular biology, and thought that the only thing wrong with the hypothesis was the unstated implication that ‘the α-helix was an important structural feature of all globular proteins. If it had been stated as some globular proteins, it would have been correct as well as illuminating.’32 Again Sage gave credit to Crick for pointing out the notion he shared with most other crystallographers in the early 1950s – that globular proteins consisted of rods of polypeptides arranged parallel to each other – was demonstrably false. Sage gave his own perspective on these historical developments in a letter to Kendrew towards the end of his life.
The term ‘Molecular Biology’ was introduced by W.T. Astbury, the founder of protein analysis, for his own X-ray studies of wool structure. His genius, going far ahead of his observations, enabled him to divine – and to divine correctly – the basic structure of a linear polymer, thus leading on to the structures of all proteins, globular and chain alike, and those of the helices of nucleic acids. This is the keystone of the Crick– Watson discovery. Astbury was prevented from reaching this degree of comprehension by his obstinate adherence to two rather than three-dimensional models. The Astbury fold is now forgotten in favour of the Pauling alpha-helix.
This invasion of three-dimensional thinking into molecular structure was to do to biology what Pasteur had done in founding stereochemistry. Both are examples of the ultimate simplification of the complication of biological structures in terms of metrical distances and angles, of nature copying art, the molecular hypothesis.33
Crick was lured away from the exacting tedium of protein crystallography by Kendrew’s post-doctoral student, Jim Watson, to build models of DNA. It was left to Perutz to exploit the isomorphous replacement method. When Bernal had first suggested the idea of introducing a heavy atom into proteins for the purpose of X-ray analysis, he had not bothered to calculate whether it would have a measurable effect in such large molecules. There was also the practical difficulty of trying to grow crystals with the added heavy atom. Sage had suggested to Dorothy Hodgkin in 1935 that she substitute cadmium for zinc in her insulin crystals34 – but she was unable to obtain any crystals at all. Even if she had, she would have been disappointed because cadmium was not heavy enough to make a measurable difference. Perutz made a crucial observation:
A chance experiment made me realize that in the diffraction pattern of haemoglobin, the scattering contributions of most of the atoms were extinguished by interference. So that for any one reflection fewer than one per cent of the electrons contributed. Substitution with heavy atoms to solve the phase problem was nothing new, but we all thought one would never be able to measure the contribution of the heavy atom among the thousands of light atoms in the protein. But I measured the fraction of the incident beam that is scattered by a haemoglobin crystal, and this measurement led me to realize that most of the scattering contributions of light atoms were extingui
shed by interference; but if I had a heavy atom, all its electrons would be concentrated at a point and they would all scatter in phase, and so they would make a measurable contribution.35
Then Perutz had a stroke of luck. In the spring of 1953 he received reprints of a paper from a researcher at Harvard, Austin Riggs, who was studying sickle cell anaemia. Riggs had introduced mercury atoms into haemoglobin and found that it made virtually no difference to its oxygen carrying capacity: Perutz immediately ‘realized that attachment of the mercury atom must leave the structure intact’. He now had his heavy atom and a way of incorporating it into haemoglobin. The experimental work was completed in about a month. Perutz demonstrated ‘that attaching two mercury atoms to a molecule of haemoglobin would produce subtle differences in the intensities of the diffracted rays, and from these differences [he] was able to derive the positions of the mercury atoms and then the phases of one set of reflections in haemoglobin. This really was the discovery which opened the whole field of protein structure.’36 Unlike Watson and Crick with the double helix, Perutz did not feel the breath of scientific rivals on his neck and carefully wrote a set of full-length papers for the Proceedings of the Royal Society. He felt ‘there was not the remotest possibility of anybody else doing the same thing: there were so few crystallographers around and the astonishing thing was that when the papers appeared, nobody else had a go.’37
Perutz had succeeded in obtaining the first accurate picture of the haemoglobin molecule, but it was only a two-dimensional image: the result ‘was a triumph but unfortunately it was totally uninterpretable’.38 Such a complex molecule gives a tangle of overlapping elements when projected onto a plane. It would take Perutz several more years of innovative effort before the first three-dimensional model of haemoglobin emerged. Kendrew, using analogous techniques for the simpler myoglobin molecule, achieved the distinction of the first detailed protein structure in 1958. They were jointly honoured with the Nobel Prize for chemistry in 1962. They shared the platform with Crick, Watson and Maurice Wilkins who were in Stockholm to collect their medicine or physiology prizes.
