The Longest Chain Anyone Had Built

In 1955 the state of the art in peptide synthesis was a molecule nine residues long, and it had just won a Nobel Prize. In the same year the sequence of adrenocorticotropic hormone was published, and it was thirty-nine residues long. Between those two numbers lies one of the more revealing episodes in the history of the field: a decade in which chemists set out to build a chain more than four times longer than anything that had worked, discovered along the way that they did not need to, and in the process changed what everyone thought a hormone actually was.

The reason it is revealing is that length was not a detail. Length was the entire problem. Every fact about how peptides were made before automation follows from a single unforgiving piece of arithmetic, and the ACTH story is the clearest place to watch a generation of chemists collide with it.

Thirty-nine residues

The hormone itself was already a commercial and clinical story before it was a chemical one. Pituitary extracts that stimulated the adrenal cortex had been in use since the 1930s, and by the late 1940s corticotropin preparations were being used to treat inflammatory disease, which meant industrial interest, which meant money for purification. Two efforts pushed hardest: a group at the Armour laboratories in Chicago, and Choh Hao Li's laboratory in Berkeley, which had already made its name isolating pituitary hormones.

Both got there in the middle of the decade. Paul Bell at Armour reported the purification and structure of the preparation from pig pituitary in 1954 1; Li and colleagues published the sequence of the sheep hormone in Nature the following year 2. The two structures were not identical, which was itself informative: the differences between species clustered in a stretch towards the tail of the molecule, while the front portion was the same in every species examined. Nobody yet knew what to make of that, but it is the thread the whole story hangs on.

What everybody did know was the number. Thirty-nine. Set against nine residues for oxytocin, and against the eighteen mostly-glycine residues that had been the record for a synthetic chain since 1907, thirty-nine was not an increment. It was a different order of task, and it was the obvious next mountain.

Why length was the barrier

Building a chain in solution means a repeating cycle: block the reactive groups you do not want involved, join the next residue, then separate the desired product from unreacted starting material, from reagent by-products and from chains that failed to extend. That separation is the expensive step, and it gets harder as the chain grows, because a thirty-residue chain and a twenty-nine-residue chain are very similar objects with very similar solubility. The purification that was straightforward at the start of the sequence becomes nearly impossible near the end of it.

Losses therefore compound. A cycle that recovers ninety per cent of the material is a good cycle, and thirty-eight of them in a row leave you about two per cent of what you started with. Drop the recovery to eighty per cent and thirty-eight cycles leave under one part in five hundred. And that is only the mass. Purity degrades in parallel, because every failed coupling leaves behind a chain one residue short that will follow the real product through the rest of the synthesis and be almost impossible to remove at the end.

The standard workaround was fragment condensation: build short pieces separately, purify each one properly while it is still small enough to be purified, then join the pieces. It reduces the number of steps the whole molecule has to survive, and it was the strategy behind every long synthesis of the period. It brings its own problem, though, because joining two fragments means forming a bond at an activated residue that sits in the middle of a chain, and those conditions are exactly the ones that can flip a residue into its mirror-image form. A single such flip produces a molecule that is chemically almost identical to the target and biologically inert, and in 1958 there was no practical way to detect it.

There was one more constraint, easy to overlook and quietly decisive: solubility. Protected peptide fragments are greasy things. The blocking groups that keep the chemistry orderly also strip the molecule of the charges that would otherwise keep it dissolved, and beyond a certain length a protected chain becomes a substance that will not go into any solvent you can then run a reaction in. Chemists of the period write about intermediates that simply stopped being workable — not because the coupling failed but because nothing would dissolve. That failure mode does not appear in any yield calculation, and it set a hard practical limit on how long a fragment could usefully be.

The fragments that worked

Two laboratories pushed hardest at ACTH: Klaus Hofmann's at Pittsburgh and Li's at Berkeley, with Robert Schwyzer's group at Ciba in Basel coming up behind them. All three built fragments, as everyone did, and all three assayed the fragments along the way — partly to check that the chemistry was sound, partly because there was nothing else to do with an intermediate. This is where the surprise arrived.

In 1961 Li and colleagues reported a nineteen-residue chain corresponding to the front of the hormone that was biologically active 4. In the same year Hofmann's group reported a twenty-three-residue chain that possessed, in the title's careful phrase, essentially the full biological activity of natural ACTH 3. Not a fraction of it. Essentially all of it. Sixteen residues of a molecule that had taken years to sequence and would take years more to build could be left off entirely without the adrenal cortex noticing.

This was not what anyone expected. Oxytocin had taught the field the opposite lesson: a nine-residue hormone where removing or altering almost anything cost you most of the activity, and where the ring closure was structurally essential. The natural inference was that hormone molecules were tightly economical, every residue earning its place. ACTH said something quite different — that a hormone could be mostly scaffolding, with the instruction confined to one end.

What a hormone turned out to be

The species comparison suddenly made sense. The front of the chain, identical across every species examined, was the part being read. The variable tail was doing something else — contributing to stability in circulation, to how quickly the molecule is cleared, to how it is recognised by the immune system of an animal receiving hormone from another species. Those are real functions, but they are not the message. The hormone had a message segment and a carrier segment, and the carrier could be edited freely.

Narrowing continued from there. A much shorter stretch near the front turned out to be the core recognised by the receptor family, and the same short motif appears in the pigment-controlling hormones, which explained a long-standing curiosity: why preparations of corticotropin also darkened skin. The molecules are relatives sharing an active core, and everything else about them is context. That insight is the direct ancestor of the whole melanocortin field, and of the modern practice of designing a peptide drug around a minimal active motif rather than a natural sequence.

The economics of a long chain

The complete molecule was finally assembled by Schwyzer and Peter Sieber, and reported in Nature in 1963 5. It was a genuine landmark — the longest defined chain built to that point, and a vindication of fragment condensation carried out with enormous care. It was also, in the most practical sense, uneconomic. The work took years, in a well-resourced industrial laboratory, and produced quantities suitable for characterisation and assay rather than for use.

It is also worth registering how close the finish was. The Nature paper appeared in July 1963; the paper describing peptide assembly on an insoluble bead appeared in an American chemistry journal a few weeks earlier the same year. Two approaches to the same problem, published within a season of each other, one of them the summit of a fifty-year tradition and the other the thing that would retire it. Nobody reading either at the time could have said which was which.

This is the condition of the field on the eve of its transformation. Chain length was rationed. A synthesis of any real length was a career-scale commitment, which meant that only molecules with a strong commercial or clinical case ever got attempted, which in turn meant that nobody could afford to make a series of variants and compare them. Structure-activity work of the kind that is routine now — build twenty analogues, test them all, see which residue matters — was simply out of reach when a single analogue cost a year.

Then the arithmetic changed. Within a couple of years of the Schwyzer synthesis, the practice of purifying every intermediate was abandoned in favour of anchoring the chain to an insoluble support and rinsing the failures away, and the cost per residue collapsed. It is worth noticing what that did to the ACTH result, though. The fragment discovery was made under scarcity, by people assaying intermediates because intermediates were all they had. Abundance would have got to the same answer eventually, but it was the constraint that forced the question, and the answer it produced — that a hormone is a message on a carrier — outlasted every method used to reach it.