The Hormone Found by Looking for Something Else

Imagine you have spent a decade building an assay to answer one question. You add a fraction of brain extract to pituitary cells, you measure how much growth hormone comes out, and a positive result means you are closer to the messenger you are hunting. Now imagine that a fraction comes back and the number is lower than the control. Lower than nothing. The cells released less hormone in the presence of your precious extract than they did when left alone. What do you write in the notebook?

The honest answers available in 1968 were unflattering. Your fraction is toxic. Your column has leaked something. Your cells are unhappy. An assay built to detect release is not, in any obvious sense, an instrument for detecting the absence of release, and a negative number in a positive assay looks far more like a fault than a finding. The reason somatostatin exists as a named molecule is that more than one group decided the number was real, and one of them was eventually forced to admit it was the answer to a question nobody had asked.

An assay that kept saying no

The first clear published account of the problem came from Samuel McCann's laboratory in 1968. Working with purified hypothalamic extracts on rat pituitary tissue, Ludwik Krulich and colleagues found that different fractions did opposite things: some increased growth hormone release, others suppressed it, and the two activities separated on the column and localised to different regions of the hypothalamus 1. They drew the obvious conclusion — that growth hormone secretion is controlled by two opposing signals rather than one — and gave the suppressing activity a provisional name.

That is, on the face of it, a discovery. It sat in the literature for five years without becoming one. Partly this is because a named activity is not a molecule, and in this field the gap between the two was routinely measured in years of work and tonnes of tissue. Partly it is because the whole community was pointed the other way. The prize everyone wanted was the releasing factor, the positive signal, the counterpart to the releasing hormones already being pulled out of the hypothalamus. An inhibitor was, at best, a complication in the purification of the thing that mattered.

It is worth being precise about what kind of obstacle it was. If a fraction contains both a releasing activity and an inhibiting one, then your assay reports their sum, and your purification is being scored on a number that two different molecules are pulling in opposite directions. Fractions that should have looked promising looked flat. Progress that should have been visible was cancelled out. The inhibitor was not merely an oddity in the margin of the search; it was actively corrupting the search.

It helps to remember what these assays cost. A hypothalamic peptide is present in the tissue in quantities that are barely quantities at all, so purification meant collecting glands by the hundred thousand from slaughterhouses, freezing them, dissecting out a structure the size of a sugar lump, and running the pooled material down column after column while assaying every fraction in living tissue. A single purification sequence could occupy a laboratory for a year and end in a few milligrams. That economics is why an ambiguous readout was not a minor annoyance. A misleading assay does not waste an afternoon; it wastes a year's supply of brains.

Naming the negative

Roger Guillemin's laboratory at the Salk Institute had spent the better part of two decades on hypothalamic extracts, processing animal hypothalami on an industrial scale to isolate quantities of material that could be weighed only with difficulty. By the early 1970s the group had the methods, the tissue pipeline and the assays. It also had the same recurring inhibition, and by 1972 the decision had been taken to stop treating it as interference and purify it deliberately.

The 1973 paper reporting the result is short and carefully titled: a hypothalamic polypeptide that inhibits the secretion of growth hormone 2. Fourteen residues, with a disulfide bridge closing most of the chain into a ring, isolated from something on the order of half a million sheep hypothalami. The name assembled itself out of the function — a stopper of growth hormone — and within a very short time the molecule had been synthesised, which both confirmed the structure and made further work possible without another lorry-load of brains.

Guillemin's own retrospective account, written thirty-five years later, is unusually frank about the sequence of events and about what came next 3. The inhibitory activity was found while looking for something else. The something else — a genuine hypothalamic releasing factor for growth hormone — remained unfound for another decade, and when it finally turned up in 1982 it did so not in the hypothalamus but in a pancreatic tumour, which had produced enough of it to be workable. The releasing factor that started the entire programme was thus isolated from the wrong organ, nine years after the inhibitor it had been obscured by.

The inhibition would not stay put

The molecule then did what the interesting ones always do, which is refuse the boundaries of the experiment that found it. Given to animals, it suppressed insulin. It suppressed glucagon. It suppressed gastrin and much of the secretory activity of the gut and pancreas. It was found in the pancreatic islets themselves, in the gut wall, and in nerve cells throughout the brain — a hypothalamic hormone that was, on the evidence, barely hypothalamic at all 4.

For the pharmacologist this was both the opportunity and the trap. A molecule that turns down half a dozen secretory systems at once is a superb research probe and a poor medicine, because a therapeutic effect on one system arrives packaged with effects on all the others. And the natural peptide had a second disqualification: injected into the bloodstream it disappears within a couple of minutes, so that any clinical use would mean a continuous infusion for as long as the effect was wanted.

The escape route was to abandon the molecule and keep the message. Systematic dismemberment showed that most of the fourteen residues were structural rather than informational, and that a much smaller ring carrying the essential few could be made to do the job — with unnatural amino acids inserted at the points where enzymes preferred to cut, so that the shortened analogue survived in circulation for hours rather than minutes. The octapeptide reported in 1982 was the first of these to become a medicine, and it remains the template for the class 5.

What the story says about screening

There is one more twist in the middle of all this that is easy to skip past. The 1973 result arrived in a field already primed for a fight about credit, since Guillemin's laboratory and Andrew Schally's had been competing over hypothalamic hormones for the best part of twenty years, and the 1977 Nobel Prize in Physiology or Medicine would eventually be shared between them. Somatostatin belongs to that rivalry, but it is the odd item in the list: the releasing hormones were the objects of a deliberate hunt, and this one was the thing the hunt kept tripping over. The most consequential molecule to come out of that whole programme, measured by what it became clinically, is the one nobody set out to find.

The moral is not that scientists should be more open-minded, which is the moral usually attached to stories like this and is close to useless as advice. The moral is structural. A screen is a question with a fixed grammar, and the grammar determines the range of answers it can return. Build an assay whose readout is release, and inhibition arrives as a negative number that resembles a failure. Build a purification scored on that readout, and any molecule doing the opposite is not simply invisible but actively costly, because it degrades the score of every fraction that contains it.

Two things had to happen for somatostatin to exist. Someone had to notice that the inconvenient result was reproducible and regionally specific rather than random, which is the observation Krulich and colleagues made. And someone with the industrial capacity to purify a hypothalamic peptide had to be willing to spend that capacity on a molecule nobody had asked for. Neither is open-mindedness. Both are the willingness to treat an anomaly as an object rather than as noise, and to pay the full price of doing so.

There is a second, quieter lesson about the shape of biological control. The programme began with an assumption that the brain issues commands to the pituitary, and that finding those commands meant finding releasing factors. Growth hormone turned out to be governed by an opposed pair, with the balance between them doing the regulating. Once the field had absorbed that, it started looking for inhibitors deliberately, and found them. The accident was not that an inhibitor existed. The accident was that the field's instruments were built on the assumption that it would not.