The Three-Way Race to Make Insulin

Consider the position of a chemist in 1960 who wants to make insulin. The molecule is fifty-one residues arranged in two separate chains, stitched together by two bridges of sulfur with a third bridge folded inside one of the chains. It is roughly six times longer than the largest peptide hormone anyone has yet synthesised, and that synthesis took a Nobel laureate's laboratory a year. Nobody has ever made a protein. There is no reason to believe it can be done, and three excellent reasons to believe it cannot. And yet at that moment groups in Aachen, in Pittsburgh and in Shanghai are all quietly working on it, none of them entirely sure what the others are doing.

Within five years all three would claim success, in terms that did not quite match and in journals that did not quite reach each other. The result was the first genuine priority dispute of the peptide era, and one whose settlement still depends on which account you read.

A sequence is a starting gun

None of this was thinkable before Cambridge. Frederick Sanger spent the decade after the war working out the order of the amino acids in insulin, beginning with one chain and a set of partial hydrolysates that had to be reassembled by overlap into a single consistent order 1. He finished the second chain and then the placement of the sulfur bridges, and by the mid-1950s insulin was the first protein anyone could write down residue by residue.

A sequence changes the character of a problem. Before it, making a protein is not a task, because there is nothing specific to aim at. After it, the target is defined, the failure is detectable, and the whole thing becomes a question of technique and endurance. Every laboratory that took up the challenge cites the same starting point, and each of them read the sequence as an invitation.

There was a deeper motivation than difficulty. If a protein assembled from bottled chemicals turned out to be biologically identical to the natural article, then the sequence really was the whole of the information — no vital residue of the living cell, no unspecified extra factor supplied by the pancreas. That question was still live in 1960, and insulin was the obvious molecule on which to settle it.

Aachen, Pittsburgh, Shanghai

Helmut Zahn's group worked at Aachen, in an institute for textile chemistry. That sounds like an odd address for a protein synthesis until you remember what wool is: keratin, a protein cross-linked by exactly the sulfur bridges that make insulin difficult. Zahn had come to insulin through the chemistry of disulfides in fibres, and his group had been publishing on insulin chains since the late 1950s.

Panayotis Katsoyannis worked in Pittsburgh, and his training was the most direct of the three: he had been part of the team that synthesised oxytocin, one of the named authors on the paper that founded the field. He knew the solution methods intimately and applied them to a target six times the size, publishing a long series of papers under the plain title of insulin peptides.

The Chinese effort was different in kind. Approved as a national project at the end of 1958, it pooled three institutions — the chemistry department of Peking University, the Shanghai Institute of Biochemistry and the Shanghai Institute of Organic Chemistry — into a single programme with dozens of researchers, and it ran for nearly seven years through a period of severe national hardship and considerable political turbulence. It was, by design, a collective effort in a way neither Western group was.

The problem was never the chains

All three groups took the same strategic decision, because it was the only one available: build the two chains separately by fragment condensation — make short pieces, purify them, then join the pieces — and only afterwards try to persuade the finished chains to find each other. Making the chains was laborious but tractable. Joining them was where the project lived or died.

The difficulty is combinatorial. Insulin has six sulfur atoms available for bridging, and they must pair in one specific arrangement: two bridges holding the chains together, one folded within the shorter chain. Reduce the natural hormone to its separate chains, then let it re-oxidise, and the sulfurs will pair more or less at random — with themselves, with the wrong partner on the other chain, or with a molecule of the same chain, producing insoluble polymer. Only a small fraction of the material finds the arrangement that is insulin. Work at the beginning of the 1960s had established that recombination was possible at all, which was the crucial encouragement, but the yields were dismal and stayed dismal.

Because the combination step returned so little, everything turned on what you could prove about the small quantity you recovered. Was it insulin, or insulin-like material contaminated with something active? Did it have the full potency of the natural hormone, or a fraction of it? Could you crystallise it — the classical demonstration that a protein preparation is homogeneous — or only demonstrate activity in an assay? Those questions, and not the synthesis itself, are what the priority dispute was about.

Priority, 1963 to 1965

Zahn's group at Aachen got into print first, with a short note received in the last week of December 1963 reporting the synthesis of both insulin chains and their combination into preparations with insulin activity 2. Katsoyannis published his own chain syntheses across 1963 and 1964, with the paper describing the synthetic B chain combined with A chain — natural or synthetic — to generate insulin activity appearing in early 1964 3. Both were real achievements and both were reported with the honesty the yields demanded: these were active preparations, obtained in very small amounts, characterised as far as the methods of the day allowed.

The Shanghai and Beijing collaboration reported crystalline synthetic bovine insulin in 1965, from material obtained on 17 September of that year, with the full account appearing in Scientia Sinica 4. The distinction matters: crystallisation was the accepted proof of homogeneity for a protein, and the crystals were shown to be indistinguishable from those of natural insulin, with full biological activity. On the strictest reading of what total synthesis of a protein means — the complete molecule, purified to crystallinity, fully active — that was the first.

That the Chinese work is under-cited in Western histories is a documented observation rather than a provocation, and the reasons are mostly mundane. Scientia Sinica had thin circulation in Europe and North America. The work was published under a collective institutional identity rather than a small set of individual names, which fitted the norms of Chinese science at the time and fitted the citation habits and prize conventions of Western science badly. Contemporary Western readers who did notice it — the result was reported in Science within months — had no easy way to follow the primary literature behind it. And within a year the Cultural Revolution had closed off much of the scientific exchange through which a claim normally establishes itself 5.

There is a further wrinkle worth stating plainly, because it is the usual objection. The Chinese team obtained natural A and B chains for its combination experiments partly through international supply channels and worked for years on the recombination conditions before the fully synthetic run, which is precisely what the other two groups also did. Optimising recombination with natural chains and then applying it to synthetic ones was the standard method, not a shortcut. The record supports treating all three efforts as belonging to the same collective achievement, arrived at from three directions within twenty-four months.

What the race settled

It settled the scientific question completely. A protein assembled from bottled amino acids, folded and paired correctly, is the same substance as the protein taken from an animal. There is nothing else in it. That had been the reasonable doubt behind a decade of argument about what proteins were, and after 1965 it was no longer available.

It settled nothing at all about supply. Not one gram of insulin used by a patient has ever come from a total chemical synthesis. The yields were far too small, the labour far too great, and animal pancreas remained the source until recombinant production arrived in the late 1970s and replaced the whole approach. In that sense the race was a demonstration rather than a manufacturing breakthrough — the most expensive proof of principle in the history of the field, and worth every hour of it, since the thing being proved was that the sequence is the molecule.

It also marked the end of an era in method. All three groups worked in solution, purifying every intermediate, in the tradition that ran unbroken from the first dipeptides at the start of the century. In the very year Zahn's note appeared, a chemist in New York published a technique that made the whole approach obsolete, and by the end of the decade a machine would assemble longer chains unattended overnight. The insulin race is the last great achievement of the old way of doing things, and the fact that it succeeded at all is the most impressive thing about it.