Keep one clause

A hormone is rarely a single instruction. Growth hormone grows tissue, raises IGF-1, pushes glucose up, and mobilizes fat. α-MSH colors the skin, curbs the appetite, and calms inflammation. Erythropoietin builds red cells and protects injured tissue. ACTH drives cortisol and is, separately, neuroactive. One molecule, several clauses — and most of the time you only want one of them. The side effects of a hormone drug are usually just its other clauses, firing when you didn’t ask.

So a recurring move in peptide design is to cut the hormone down to the fragment that carries the clause you want, and throw the rest away. This site has met that move again and again, always locally. Here it is as one playbook — the engineering recipe, the real drugs it has produced, and the honest reason the same smart design sometimes delivers a medicine and sometimes a molecule that does nothing. Bullish on the method, exact about the verdicts.

The design rests on a fact about receptors: they often read only a small patch of a hormone’s surface, not the whole molecule. The active face that triggers one receptor can be a handful of residues, and the parts that engage the hormone’s other receptors can sit elsewhere on the chain. Cut out the active patch and you have a fragment that still speaks to one receptor while having lost its grip on the others — one clause, kept.

There is a second, mechanical step. A naked fragment is usually destroyed in minutes, so the practical version of the trick adds a protecting group — a cap, a resistant residue, a tail the body’s peptidases cannot cleave. The engineered result is a short, durable peptide that does one of the parent’s jobs. That is the whole recipe: find the clause, cut it out, armor it. The interesting question is always what happens next.

Done well, the playbook produces real medicine. The clearest case is teriparatide, which is simply PTH(1-34) — the active N-terminal fragment of parathyroid hormone, the part that fires the receptor — and which became the first drug that builds bone rather than merely slowing its loss (Neer et al., NEJM, 2001). The growth-hormone-releasing hormone drugs tesamorelin and sermorelin are fragments of GRF, kept to its first 29 residues. And cosyntropin, the synthetic ACTH(1-24) used in the adrenal stimulation test, is a fragment repurposed as a diagnostic.

These are not research-peptide curiosities; they are approved, routine tools of endocrinology. The lesson they carry is that fragment engineering is a legitimate and productive strategy — when the kept clause is both real and clinically useful, cutting the hormone down to it is exactly the right thing to do.

The same design, applied to the research-peptide frontier, fans out into very different verdicts, and reading which is which is the whole skill. KPV, the last three residues of α-MSH, keeps the hormone’s anti-inflammatory clause and drops its pigment and appetite ones — a genuine, if still lightly-evidenced, retained activity. ARA-290, an 11-residue slice of erythropoietin’s helix-B face, separates EPO’s tissue protection from its red-cell effect as cleanly as the design ever manages (Brines et al., PNAS, 2008) — elegant mechanism, but a thin, unproven clinic. Semax keeps ACTH(4-10)’s neuroactivity while shedding its cortisol-releasing role — registered and used in Russia, barely evidenced anywhere else.

And then the cautionary one. AOD-9604, the lipolytic tail of growth hormone, executed the split perfectly: it drove fat metabolism in rodents with no IGF-1 rise and no glucose penalty. The separation succeeded completely — and the kept clause turned out to be empty in people, producing no significant weight loss in human obesity trials. Four molecules, one design, four different places on the spectrum from “real drug” to “hollow fragment.”

Line those cases up and a clean rule emerges. The fragment design has two halves, and they are not equally trustworthy. The subtraction — you dropped the parent’s other activities — is the easy, verifiable half. You can measure it directly: no IGF-1, no cortisol, no rise in hematocrit. The retention — the clause you kept still works, still matters, still holds up in a human — is the hard half, and the only one that decides whether you have a drug. A clean cut proves what you removed. It proves nothing about what remains.

That asymmetry is exactly what a marketplace exploits. “None of growth hormone’s side effects” is a true and verifiable statement about the subtraction, deployed to imply a benefit the retention side never established. The honest reader’s move is to ask which half a claim actually rests on: a fragment peptide that leads with everything it doesn’t do is usually quiet about whether the one thing it is supposed to do has survived a trial.

Fragment engineering is one of the most elegant ideas in peptide medicine: a hormone is a sentence, and you can often keep just the word you need. When the word still carries meaning and that meaning is useful, you get teriparatide. When it doesn’t, you get AOD-9604 — a grammatically perfect cut that says nothing. The method is sound across the whole spread; the verdicts are not, and they turn entirely on evidence for the piece you kept, not admiration for the piece you dropped.

• Hormones are pleiotropic; side effects are the other clauses. GH, α-MSH, EPO and ACTH each do several things. Fragment engineering cuts the molecule to the one activity you want and discards the rest.
• Receptors read short motifs, so fragments work. The active face is often a small patch; a fragment can engage one receptor while losing the others. Add a peptidase-resistant cap and you have a candidate.
• The playbook produces real drugs. Teriparatide is PTH(1-34); tesamorelin and sermorelin are GRF fragments; cosyntropin is ACTH(1-24). Approved medicine, not just research peptides.
• The same design also fails. KPV (real but light), ARA-290 (elegant, unproven), semax (registered abroad, thin), AOD-9604 (clean split, empty clause, failed trials) — one method, a spectrum of verdicts.
• Judge the retention, not the subtraction. Dropping the other activities is easy and verifiable; keeping a useful effect is the trial-dependent hard part. 'None of the side effects' describes the easy half, not the benefit.