Insights
Long-form deep-dives into how peptide hormones actually work — receptors, second messengers, and physiology, traced from the source. Mechanistic depth, not marketing.
Featured — start here
What does GLP-1 do? How the hormone actually works
From an intestinal cell to a closed potassium channel: the receptor, the second messenger, and why the whole system only fires when glucose is high.
Read the deep-diveOral GLP-1: can you take semaglutide as a pill?
A peptide is food — swallow it and your gut digests it before it works, which is why this class has lived on the needle. Yet the pill is arriving, by two philosophically opposite routes: smuggle the fragile peptide across the gut wall, or stop using a peptide at all and rebuild the signal from a rugged small molecule. The second one is the real revolution, and it comes down to a quiet truth about receptors — they read the message, not the messenger.
Read the deep-diveWhat is retatrutide? The triple agonist explained
The newest metabolic peptides don't mimic one hormone — they play three at once. Why the field moved from a single signal to a chord, and why glucagon, of all things, earns a seat.
Read the deep-diveMuscle loss on GLP-1: how to keep muscle while losing weight
The next leap in weight loss isn't losing more — it's losing better. Pair a GLP-1 drug with myostatin inhibition and you can strip fat while sparing, even building, muscle. Inside the TGF-β biology and the combination therapies engineering it.
Read the deep-diveMore deep-dives
Mechanism is not proof
Every deep-dive here ends the same way — bullish on the science, sceptical on the page — and it isn’t a hedge. It’s a claim that two things people routinely merge are separate and graded on different evidence: the science is the mechanism (a receptor, a pathway, a plausible chain of cause and effect), and the page is the benefit (what the molecule actually does in a person). A compound can have landmark-grade science and an empty page, and most of the peptide world lives in that gap. This is the map of it: the ladder the whole reference grades against, told as an argument rather than a standard. Evidence comes in rungs — plausible mechanism, cell data, animal efficacy, human surrogate markers, controlled human outcome, independent replication — and each rung up is rarer and more predictive, so a claim’s confidence should track its highest real rung, not its most flattering sentence. Three failures recur: the surrogate trap, where a marker moves but the outcome doesn’t follow or worsens (CAST’s arrhythmia drugs raised mortality; MK-677 lifted IGF-1 and lean mass but not strength); the cross-trial trap, where numbers from separate studies get ranked as if on one ruler (amycretin’s 24%); and the missing replication rung, where a striking single-source result is unaudited, not proven (epitalon) — and below even that, the headlines that confuse a trial that has enrolled with one that has reported. None of this is dismissal: low-rung evidence is real science and where discovery lives; the sin is selling it as proof. One question cuts through every confusing claim — what is the best evidence, and which rung is it on? Bullish on the science, sceptical on the page. No dosing.
Keep one clause
A hormone is rarely a single instruction. Growth hormone grows tissue, raises IGF-1, pushes glucose and mobilizes fat; α-MSH colors skin, curbs appetite and calms inflammation; EPO builds red cells and protects tissue; ACTH makes cortisol and is separately neuroactive. One molecule, several clauses — and 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, armored against the enzymes that would destroy it. This site has met that move again and again; here it is as one playbook. Done well it produces real medicine — teriparatide is PTH(1-34), tesamorelin and sermorelin are GRF fragments, cosyntropin is ACTH(1-24). Applied to the research frontier it fans into very different verdicts: KPV keeps α-MSH’s anti-inflammatory clause (real but lightly evidenced), ARA-290 splits EPO’s tissue protection from its red cells (elegant, unproven), semax keeps ACTH’s neuroactivity (registered abroad, thin), and AOD-9604 executed the split perfectly and kept a clause that was empty in humans (failed trials). The design has two halves of unequal trust: the subtraction — you dropped the other activities — is easy and verifiable; the retention — the kept clause still works in people — is the hard, trial-dependent half, and the only one that decides whether you have a drug. ‘None of the side effects’ describes the easy half. Judge the clause you kept, never the elegance of the cut. Bullish on the method, exact about the verdicts. No dosing.
The body keeps time
Most of physiology speaks in amounts — a gland releases more or less, a receptor reads the concentration, the response scales. But a surprising share of the endocrine system, and nearly all of its master controls, speaks a second language the amount model misses: time. The same molecule, at the same total dose, means different and sometimes opposite things depending on its rhythm. This reference has met the idea four times in separate corners — GnRH, growth hormone, PTH, cortisol — and here it is as one principle. It rests on a single mechanism: receptor desensitization. Pulsed with gaps, a receptor recovers between beats and answers each one; bathed in a continuous high level, it downregulates and goes silent — so a steady maximal input is not a louder ‘on’ but the erasure of the pattern, which many systems read as ‘off’. That one fact unifies the paradoxes: why pulsatile GnRH switches the reproductive axis on while continuous GnRH switches it off, why a brief daily PTH spike builds the bone that continuous PTH dissolves, why injected growth hormone is a worse signal than the gland’s own nightly bursts, why even cortisol runs on ultradian pulses and insulin on five-minute portal bursts. Evolution encodes in time because a frequency resists noise an absolute level can’t, carries two messages on one line, and can’t be spoofed by flooding. And modern endocrinology has learned to write to the clock: the drug is often less a molecule than a schedule. Bullish on the science — biology over benefits at its clearest. No dosing.
The half that did nothing
Growth hormone does several things at once — grow tissue, raise IGF-1, push glucose up, mobilize fat — and AOD-9604 is its tail, the C-terminal fragment hGH(176-191), isolated on a clean and appealing theory: that the fat-burning lives in that fragment and can be cut away from everything else. A fat-loss drug with none of growth hormone’s baggage. It is the same ‘keep one clause’ move that gave KPV and ARA-290 their appeal — and the instructive case where it didn’t pay off. In the narrow sense the separation worked: in obese mice AOD-9604 drove lipolysis through the β3-adrenergic pathway, with no IGF-1 rise and none of the insulin-sensitivity harm chronic GH can cause. The fragment really had been split from the rest of GH biology. But the kept clause was empty in people: controlled human obesity trials showed no significant weight loss over placebo, and development as a weight-loss drug was discontinued. The rodent effect didn’t translate — you had cleanly isolated a clause that, in humans, said almost nothing. And yet it still sells, marketed for fat loss on exactly the mechanism story its own pivotal trial failed to support: the sharpest small case of the gap between a molecule’s mechanism and its proven effect. Bullish on the method, sceptical on this result. No dosing.
The gene the parents fight over
IGF-1 has a sibling, nearly identical in shape and name — IGF-2 — and by most measures it is the stranger and in some ways the more important of the two: the dominant growth factor before birth, and the more abundant IGF in adult human blood, yet rarely discussed because it answers to none of the growth-hormone machinery the IGF-1 story runs on. It is governed instead by something older and odder, a genetic tug-of-war between the two parents written into the gene itself. IGF2 is imprinted: in most tissues only the father’s copy is switched on and the mother’s is silenced — proven in mice, where losing the paternal copy stunts growth and losing the silent maternal one does nothing. The leading explanation is a conflict of interest: a father’s genes favor a larger, more demanding offspring, a mother’s favor restraint, and the conflict is literally encoded as two oppositely imprinted genes — the paternally expressed IGF2 accelerator and the maternally expressed IGF2R clearance receptor, a brake whose only job is to destroy IGF-2. It signals growth through a borrowed insulin receptor as well as IGF-1R, and when the balance tips it reads out as overgrowth (Beckwith-Wiedemann), growth restriction (Silver-Russell), or tumor-driven hypoglycemia. The sibling that grew first is the clearest case in the body of growth governed by genetic conflict. Bullish on the science, which here is genuinely strange. No dosing.
Read by the brain
For a decade the incretin story was a metabolic one — hold blood sugar, quiet appetite, lower weight. Brenipatide takes the same sentence and asks whether the brain can read it. Mechanically it is close to familiar ground: a single peptide that agonizes both GIP and GLP-1, the same receptor pair as tirzepatide, engineered for a far longer, roughly once-monthly reach. What is unfamiliar is the direction — Eli Lilly is leading its development not with obesity but with alcohol use disorder and major depression, on the bet that incretin signaling touches the brain’s reward and mood circuitry. The rationale is real: GLP-1 receptors sit in the mesolimbic dopamine pathway, and in animals GLP-1 agonists cut self-administration of alcohol and other drugs. The first human signal is real too, and comes with a built-in asterisk — semaglutide lowered alcohol quantity and craving in a small trial, but drinking days didn’t change and weight fell, leaving the field’s central question open: a specific effect on craving, or general consumption suppression wearing a new label? And brenipatide itself has zero published efficacy, yet its RENEW program goes straight to large Phase 3 trials in psychiatry, reading out around 2028. A well-reasoned, high-conviction, unproven bet — the incretin sentence carried to the clinic on the wager that the brain can read it. Bullish on the science, sceptical on the page. No dosing.
The chain of command
The hypothalamus doesn’t run the body’s slow glands by wire — it runs them through a chain. A tiny releasing hormone tells the pituitary to release a tropic hormone, which tells a peripheral gland to do its work: three tiers, each commanding the next. This site has walked two of them, the reproductive axis under GnRH and the growth axis under GHRH; here are the other two, which complete the set. The stress axis: CRH drives ACTH drives adrenal cortisol. The thyroid axis: TRH drives TSH drives the thyroid. The architecture earns its complexity twice over — it amplifies on the way down, so a whisper of CRH becomes a flood of cortisol, and it reports back on the way up, with the output hormone feeding back to silence its own commanders, the governor that holds a setpoint. ACTH turns out to be cut from POMC, the same precursor as α-MSH; TRH is a three-residue tripeptide that runs an entire axis, proof that potency is about fitting a receptor, not size. And because each axis is a feedback cascade, medicine tests it by perturbing the top — dexamethasone suppression flags Cushing’s, cosyntropin stimulation flags Addison’s — turning the releasing and tropic hormones into diagnostic reagents. Amplify down, report up, diagnose by provocation. Bullish on the science, plainly told. No dosing.
EPO without the blood
Erythropoietin is famous for one thing — making red blood cells, the property that treats anemia and powers blood doping. But across the 2000s the lab kept seeing EPO protect tissue, brain, heart and nerve from injury in ways the red-cell effect couldn’t explain. The resolution is that EPO does two jobs through two receptors: a homodimeric EPO receptor builds red cells, while a separate ‘innate repair receptor’ — an EPO-receptor joined to the β-common chain that appears only on injured, inflamed tissue — carries a tissue-protective, anti-inflammatory signal. ARA-290 (cibinetide) is what you get when you keep the second job and discard the first: an eleven-residue peptide modeled on EPO’s helix-B face that engages the repair receptor but not the red-cell one, delivering the protection with no rise in hematocrit. The receptor story is genuinely elegant, and the ‘keep one clause’ design echoes KPV. The clinical record is much thinner: the most-cited result is a 22-patient pilot in sarcoidosis small-fiber neuropathy, with later corneal-nerve gains, all small, mostly single-group, and never confirmed in a phase 3 before development stalled. Elegant mechanism, promising pilot, not proven at scale — held honestly. Bullish on the science, sceptical on the page. No dosing.
Two residues apart
Vincent du Vigneaud synthesized oxytocin in 1953 — the first peptide hormone ever made in a lab, a Nobel-winning feat — and then vasopressin, and the two turned out to be almost the same object: nine amino acids, the same disulfide ring, both released from the posterior pituitary, differing at only two positions out of nine. Yet they share almost nothing in what they do. Oxytocin drives labor, milk let-down and the famous bonding story; vasopressin runs water balance and blood pressure — the cleanest proof in the body that peptide specificity is written in small sequence differences, not size. And here is the twist: the quieter twin became by far the more useful drug. Where oxytocin’s bonding sprays mostly failed the replication record, vasopressin spun off a whole shelf of real medicines, each tuned to one of its receptor subtypes — desmopressin at V2 (diabetes insipidus, bedwetting, and, via clotting-factor release, von Willebrand disease), the vaptans blocking V2 for hyponatremia and polycystic kidney disease, and vasopressin itself as a V1a pressor in shock. Its brain arm is real too — V1a drives vole pair-bonding — but a vole is not a person, and the human behavioral drugs stay thin. Why the molecule that reads less romantic is the one you’re far more likely to be prescribed. Bullish on the science, sceptical on the page. No dosing.
The door swings both ways
A molecule’s biology is climate — its sequence and receptor don’t change from one quarter to the next. What you can actually get is weather, and in 2026 the weather turned in two opposite directions at once. One door swung shut: semaglutide and tirzepatide came off the FDA shortage list (December 2024 and February 2025), the enforcement grace periods for 503A pharmacies and 503B outsourcing facilities expired through spring 2025, and in April 2026 the agency proposed dropping all three leading incretins from the 503B bulks list outright — the approved blockbusters walled off from the compounding the shortage years had made routine. Another door cracked open: the research peptides the FDA had restricted in 2023 got their day in front of the Pharmacy Compounding Advisory Committee on 23–24 July 2026, which voted 8–6, against the agency’s own review staff, to recommend adding six of them — BPC-157, KPV, TB-500, MOTS-c, epitalon, semax — to the 503A bulk list, rejecting a seventh. Read quickly that sounds like six peptides became legal; read correctly, almost nothing changed, because a committee advises and rulemaking decides. Meanwhile the powder’s source tightens under China’s 2026 drug-law reforms. A delisting is a fact that already bit, a proposed rule a signal that hasn’t, a vote a forecast — three tenses the field keeps reading as one. Bullish on the science, sceptical on the paperwork. No dosing.
Teriparatide, backwards
Parathyroid hormone is the body’s fast defense against low blood calcium, and its reputation in bone is fearsome: let it run continuously, as a parathyroid tumor does, and it mines the skeleton for calcium — the bone-wasting disease of hyperparathyroidism. So the drug built from it is one of the strangest objects in endocrinology. Teriparatide is PTH(1-34), the hormone’s active fragment, given as a once-daily injection, and it does the opposite of what its parent’s reputation predicts: it builds bone, the first anabolic osteoporosis drug, where every older option only slowed the loss. The whole paradox resolves to one variable — the same one that runs the reproductive axis. Bone reads how long PTH stays, not whether it is present: a brief daily spike switches on the osteoblast formation program and opens an ‘anabolic window’ before resorption catches up, while a permanently high level keeps driving osteoclasts through RANKL. Same PTH1 receptor, opposite outcome, written by the shape of the exposure in time. PTHrP obeys the identical law — intermittent as abaloparatide it builds bone, continuous from a tumor it causes hypercalcemia of malignancy. The drug isn’t a new molecule; it’s a new schedule. Bullish on the science, sceptical on the page. No dosing.
The off switch
Most hormones say a specific thing to a specific tissue. Somatostatin says one thing to almost everything — stop. It shuts down growth hormone and TSH from the pituitary, insulin and glucagon from the islet, and a long list of gut secretions, through five receptor subtypes, and in the growth-hormone axis it is the brake the accelerators have to overcome. It is also the rare hormone whose entire drug strategy is inverted: you don’t supply it to boost a signal, you rebuild it to last and install it as a brake. Native somatostatin is useless as a drug — a two-minute half-life and rebound hypersecretion — so octreotide and lanreotide were engineered to last weeks and narrowed to grip mainly SSTR2, enough to silence the growth hormone of acromegaly and the hormones of neuroendocrine tumors without flattening every secretion in the body; the CLARINET trial showed they slow the tumors too, not just quiet them. Pasireotide goes broader — SSTR1, 2, 3 and 5 — which unlocks Cushing’s disease through SSTR5, but the same SSTR5 on the beta cell drives hyperglycemia in most patients: the breadth that makes it work is the breadth that costs. Why receptor-subtype selectivity, not strength, is the whole game. Bullish on the science, sceptical on the page. No dosing.
The hormone discovered twice
In 1994 a lab cloned the gene a famously fat strain of mouse was missing and found leptin — the hormone fat tissue uses to report its own size to the brain, proof that adipose is an endocrine organ and not inert storage. The logic was irresistible: a hormone that signals fat stores and curbs appetite should, given in excess, melt weight away. It emptied the fat mouse and was licensed as the coming cure for obesity. Then it barely moved ordinary obese humans in trial, and leptin had to be understood a second time — not as a satiety brake but as a starvation alarm, a signal of plenty whose most powerful action is its own absence, in a body that defends the fat it has. Common obesity runs on high leptin the brain has stopped hearing (leptin resistance), which is why more of it does little; replacement works only where leptin is genuinely missing, as in congenital deficiency and lipodystrophy. Its second act is stranger and more current: amylin can partly re-sensitize the brain to it, the thread that runs straight into today’s amylin era. Why the landmark hormone was never the drug. Bullish on the science, sceptical on the page. No dosing.
The clock you can't audit
Epitalon is four amino acids — Ala-Glu-Asp-Gly — carrying one of the largest claims in peptide science: that so small a molecule can switch on telomerase, the enzyme that rebuilds the caps on our chromosomes, and bend the course of ageing. It is the defined tetrapeptide distilled from epithalamin, a crude pineal-gland extract, by a St. Petersburg school — the same ‘white box from a black box’ move behind cartalax and thymalin. The mechanism is a pleasure to trace, and a cell-culture study really did show telomerase induction and telomere elongation in human fibroblasts. But read honestly, the lifespan data is thinner than the brochure: in mice epitalon left mean lifespan unchanged while nudging maximum lifespan and cutting some tumors, and the striking human mortality claim rests on a single non-randomized report. The weakness isn’t that the mechanism is absurd — it’s that almost every result traces to one research network and none of the lifespan findings has been replicated by an outside lab. Telomerase activation is also a hallmark of cancer, so the hoped-for upside and the feared downside are the same event. Not proven, not debunked — unaudited. Bullish on the science, sceptical on the page. No dosing.
Written for the brain
Semax and selank are real registered medicines — just not where most readers live. Both were developed and are prescribed in Russia, both are near-unknown to Western medicine, and both were engineered the same way: take an endogenous human peptide, keep the fragment that does the one thing you want, and armor it against the enzymes that would destroy it. Semax is ACTH(4-10) with the cortisol-releasing activity stripped out and a peptidase-resistant tail added, repointed toward cognition through the BDNF/TrkB system; selank is a modified tuftsin fragment used as an anxiolytic — the same ‘keep one clause’ move as KPV. The biology is plausible and in places elegant, the domestic use is decades deep, and the independent Western-standard evidence is thin to absent: semax’s BDNF mechanism is animal-grade, and selank’s best result is a single home-grown trial calling it comparable to a benzodiazepine without the sedation. DSIP — named for a sleep effect that never reliably replicated — is the standing warning that a brain peptide can be real, named and studied for fifty years and still have no confirmed job. Real mechanism, real use, missing audit, held honestly. Bullish on the science, sceptical on the page. No dosing.
The molecule that is both halves
CagriSema proved the GLP-1-plus-amylin bet with two engineered peptides in two pens. Amycretin asks the obvious next question — why two? — and answers it with a single molecule that agonizes the GLP-1 receptor and the amylin receptors at once, the unimolecular version of the same combination. In June 2025 two trials landed together in The Lancet: a subcutaneous phase 1b/2a in which the top dose took off a mean 24.3% of body weight by week 36 with the curve still falling, and a first-in-human study of an oral form — a daily pill — that reached ~13% in twelve weeks. The numbers are large and the biology is sound; the asterisks are the whole point. These are small, short, early-phase studies with no head-to-head arm, so the tempting cross-trial ranking against tirzepatide or retatrutide is exactly the comparison the data cannot support. And writing a combination as one word costs something a two-drug stack keeps free: the GLP-1-to-amylin ratio, welded fixed in a single chain. Why amylin belongs next to GLP-1, why the pill is the real story, and what the unimolecular bet trades away. Bullish on the science, sceptical on the page. No dosing.
Two ways to ask for growth
Growth hormone comes out of the pituitary in pulses, and there are exactly two ways to ask for more of it without injecting the hormone itself: push the accelerator the hypothalamus uses — GHRH, and its analogs sermorelin, tesamorelin and CJC-1295 — or press the second button ghrelin discovered, the GHS-R1a receptor the GHRPs and ipamorelin hit. The whole case for secretagogues over straight HGH is that they ask the gland to fire its own pulse instead of flattening it, and leave IGF-1 feedback intact so the loop self-limits. Both halves of that case are real, and both get complicated fast. The only one of these molecules with an approved label and an outcome trial is tesamorelin, for visceral fat in HIV; the oral ghrelin mimetic MK-677 raises IGF-1 and lean mass for a year but moves no strength and nudges glucose up; and the long-acting CJC-1295-with-DAC trick quietly erases the very pulsatility that was the argument. Why you’d press both buttons at once, why the synergy is more than additive, and where the surrogate markers stop and the evidence runs out. Bullish on the science, sceptical on the page. No dosing.
The note without the incretin
In December 2025 The Lancet published a 48-week trial in which a once-weekly injection took off a mean 20% of body weight, and the GLP-1 receptor was nowhere in the molecule. Eloralintide is Lilly’s amylin analog: 37 residues on the hormone insulin is co-secreted with, the native disulfide swapped for a methylene-thioacetal ring that cannot be opened, backbone methyls the ribosome cannot make, and a C20 fatty diacid on an engineered lysine that binds albumin for a fortnight. It was tuned to prefer the amylin-1 receptor twelve-fold over the bare calcitonin receptor it is built from, which is the whole argument for calling it gentler, and the only evidence for that is a rat taste-aversion assay. We read the structure off the public registry record, walk the three monotherapy trials and the EloraTZP combination that reached 23.3% in type 2 diabetes in September 2026, and separate what selectivity has proved from what a press release with a 27% discontinuation rate has not. Bullish on the science, sceptical on the page. No dosing.
The trial that hasn't reported
In September 2026 the phrase “first human study of BPC-157” went around again. It wasn’t one. We checked the primary record: PubMed for August–September 2026 holds reviews and a critical re-reading of the rodent ischemia-reperfusion literature; ClinicalTrials.gov holds one randomized trial still recruiting and one not yet open. The entire published human record is three uncontrolled reports from a single Florida clinic – sixteen knee-pain patients surveyed by phone, twelve cystitis patients who all scored 100%, two volunteers infused for three days – thirty people and no placebo. The first trial that could actually answer the question, 120 hamstring strains randomized and blinded against placebo with MRI endpoints, began in February 2026 and reads out in 2027. And the July FDA advisory vote that gets reported as a verdict was 8–6 against the agency’s own scientists, on access grounds, with no new data in the room. What exists, what is still missing, and how to read the next headline. Bullish on the science, sceptical on the page. No dosing.
Wrong end of the hormone
Somewhere between the catalog page and the group chat, KPV picked up a claim it never earned: that it does something for low sexual desire. The confusion is understandable, because the melanocortin family really does contain an approved drug for hypoactive sexual desire disorder – bremelanotide, PT-141 – and both molecules descend from the same thirteen-residue hormone, α-MSH. But they were cut from opposite ends of it. Bremelanotide keeps the middle four residues that bind the MC4 receptor and drive the whole desire story; KPV is the last three, which bind no melanocortin receptor at all and were kept precisely because they don’t. There is no study of KPV and sexual function in any species. And the drug that does work is a smaller story than its reputation: a third of a point on a desire scale, no difference in satisfying sexual events, nausea in four of ten. Where the erection in a tanning trial came from, why it can’t be in the tail, and what the record actually says. Bullish on the science, sceptical on the page. No dosing.
The famine without the slowdown
For ninety years the oldest result in ageing biology came with a tidy explanation: cut an animal's calories and it lives longer because its metabolism slows down. In September 2026 a Berkeley lab put twenty-month-old mice on semaglutide – the age at which rapamycin earned its reputation – and ran calorie restriction beside it. Both arms lived about 12% longer, both aged better on nearly every readout, and only one of them turned the engine down. The GLP-1 mice kept their metabolic rate and got the dividend anyway. That is the interesting finding, not the headline: a controlled experiment in which the famous slowdown is separated from the benefit it was supposed to cause. What the paper actually measured, why the muscle result reads the opposite way in mice and people, the only human mortality number we have, and how much of a female-only, single-lab lifespan claim survives the study design. Bullish on the science, sceptical on the page. No dosing.
The lizard that ate twice a year
The best-selling molecules in the catalog trace back to a venom. A Gila monster gorges a few times a year and fasts between meals, and in the late 1980s a Bronx VA endocrinologist wondered what kept its pancreas intact through that feast-and-famine — a curiosity that pulled a 39-residue peptide, exendin-4, out of the animal's saliva. It was two-thirds a match for a human gut hormone nobody could yet turn into a drug, and, by an accident of one amino acid, immune to the enzyme that destroyed the human version in two minutes. That single substitution became exenatide, the first GLP-1 drug, and the template for everything after it: liraglutide's fatty-acid tail, semaglutide's weekly dose, tirzepatide's second receptor, retatrutide's third, cagrilintide's borrowed amylin. This is the whole lineage — the incretin idea that took seventy years to pay off, the tangents that didn't, and how a lizard's meal schedule ended up coded on a vendor's shelf as GLP-1, GLP-2, GLP-3. Bullish on the science, sceptical on the page. No dosing.
Two ways to bottle a gland
The thymus trains the immune system and then, from puberty on, quietly dissolves itself — so for fifty years two schools have tried to bottle what it does. One is thymalin: a crude polypeptide complex boiled out of calf thymus in Soviet-era Leningrad, never fully characterised, still sold across the former USSR, and hung on a startling claim — that a course of it in the old cut six-year mortality by half. The other is thymosin α1: the same tissue run all the way down to a single defined 28-residue sequence, synthesised, patented, and sold in more than thirty countries as Zadaxin, with four decades of hepatitis-B and sepsis trials behind it and a real receptor — TLR9 on the dendritic cell — under it. Same gland, same starting extract, opposite epistemics: the black box you cannot audit versus the white box you can. A reference-grade read on what each one actually is, why the defined peptide got a drug label and the extract did not, and how much of the longevity headline survives contact with the study design. Bullish on the science, sceptical on the page. No dosing.
Distilled to three letters
Cartalax is what you get when a whole-cartilage extract is run through a mass spectrometer and asked what, exactly, is doing the work: a single tripeptide, Ala-Glu-Asp, three letters pulled from a complex of thousands. The St. Petersburg school that isolated it makes the boldest claim in peptide science — that a fragment this small carries tissue-specific instructions and acts by binding the chondrocyte's own DNA, turning the apoptosis trigger p53 down and the proliferation marker PCNA up, so cartilage cells that had stopped dividing start again. The mechanism is a genuine pleasure to follow; the human evidence is open-label capsules, subjective pain scores, and nothing moving on the X-ray, almost all of it from one institute. Both halves are true. Bullish on the science, sceptical on the page. No dosing.
Asking the cells that didn't die
Humanin wasn't designed; it was overheard. In 2001 a lab screened the surviving neurons of an Alzheimer's brain for whatever was keeping them alive and pulled out a 24-residue peptide written not in the nuclear genome but inside mitochondrial DNA — a message from the powerhouse that tells the cell not to kill itself. It reads two ways at once: secreted, it behaves like a hormone, docking a three-part cytokine receptor to switch on STAT3; kept inside, it physically grabs the apoptosis trigger Bax before it can reach the mitochondrion. The blood level falls with age and runs high in the children of centenarians. A reference-grade read of the mitochondrial genome's most surprising export — and an honest account of how little of it has been tested in people. Bullish on the science, sceptical on the page. No dosing.
SLU-PP-332 vs BAM15: burning fuel is not the same as getting fit
One switches on an exercise-associated gene program. The other makes mitochondria spend more fuel for a given amount of ATP. Both have meaningful mouse data; neither has established human benefits or a validated stack. What the studies actually show, why neither compound is a peptide, and why the most revealing human uncoupler trial lost fat without improving exercise capacity. Now including SLU-PP-915, the orally bioavailable successor whose best result came from adding it to training rather than replacing it — and the anti-doping assay built for it before any human trial.
Bigger, but not stronger
For fifteen years the drugs that release muscle's brake kept failing the same strange way — not by doing nothing, but by working. ACE-083 grew its target muscle by double digits on the scan and left patients no stronger; ACE-031 built lean mass and had to be pulled for nosebleeds. Buried in both is the fact the field kept relearning: making a muscle bigger and making it work are different problems. The same ligand-trap platform that failed in muscle quietly became two approved drugs for the blood and the lung — and in 2026 apitegromab finally moved a functional endpoint in spinal muscular atrophy, but only by being added on top of therapy that had already repaired the nerve. Why size was never the finish line, and what all of it means for the muscle everyone is now racing to save from GLP-1 weight loss.
The rest of the cascade
Men on testosterone sometimes add hCG and report feeling steadier, clearer, more themselves — benefits the testosterone number alone never predicted. The reason is less exotic than a mood drug and more interesting. Testosterone replacement fixes one hormone while switching off the gland that made it: the pituitary goes quiet, intratesticular testosterone collapses by about 94 percent, and the testis stops producing the whole steroid cascade around testosterone — pregnenolone, progesterone, estradiol, the brain-active ones included. hCG is an LH signal that relights that factory. The honest read of the 'mental benefits' is estradiol you can measure on a lab draw, plus a broader neurosteroid milieu you mostly can't, plus a good deal of reassurance and expectancy. The same lens sorts the hormones and the protocols: hCG stands in for LH and wakes only the Leydig (steroid) arm, while hMG and FSH wake the Sertoli (fertility) arm, and the felt distance from testosterone alone up to a full restart comes down to how much of the gland you keep online. Why the fertility claim is settled and the mood claim is still thinly evidenced — and why this exact molecule has overpromised wellbeing before. Bullish on the science, sceptical on the page.
The gut's full report
The gut is the body's largest endocrine organ, and after every meal it files a report: a timed sequence of peptides that tells the brain what arrived and how digestion should run. CCK fires first from the duodenum. GLP-1 and PYY are released together, further down, from the very same L-cells. Secretin manages acid; motilin runs the housekeeping between meals. The blockbuster GLP-1 drugs are one line of that report, amplified. The proof that the whole report, not any single peptide, governs appetite came from bariatric surgery, which raises the entire post-meal chorus at once. The frontier now is rebuilding that chorus in a vial: PYY analogs, GLP-1 paired with PYY, tri-hormone infusions built to imitate what the gut does after a bypass. A reference-tier read of the enteroendocrine system the incretin era only partly borrowed. No dosing.
How to reconstitute peptides: the vial is not a recipe
Adding water to a peptide vial looks like the simplest step in the process. It is where four separate systems collide: product identity, sterile technique, concentration, and stability. Get any one wrong and a clear solution can still be contaminated, mismeasured, or chemically compromised. A label-first guide to what reconstitution actually does, why sterile water and bacteriostatic water are not interchangeable, why there is no universal storage clock, and which shortcuts turn a preparation problem into a patient-safety problem.
The pulse is the message
The reproductive axis does not read how much GnRH is present. It reads the rhythm. GnRH leaves the hypothalamus in pulses, and the pituitary decodes their frequency: delivered in pulses the same molecule switches the axis on, delivered continuously it switches the axis off, by desensitizing its own receptor. That one fact explains the strangest drug in endocrinology, a GnRH agonist like leuprolide that suppresses the very system it stimulates, and its cleaner mirror, the antagonists that block without the flare and are now going oral. Upstream sits kisspeptin, the gate that generates the pulse and the gene whose failure hides puberty, now being turned into a gentler fertility tool. A reference read of the one axis where frequency, not concentration, is the signal. No dosing.
A switch, not a supply
GHK-Cu isn't a nutrient you top up — it's a signal that launches a program. Each dose starts the same phased wound-repair sequence a fibroblast runs after injury: activation, then collagen and matrix synthesis, then MMP-driven remodeling, then resolution and quiet. A continuous injected signal never lets that program reach its last phase; it pins the tissue in perpetual build-and-remodel, which is the opposite of what settled, mature matrix looks like. Why the mechanism — not any trial — is the real argument for cycling it, why that logic is different from the receptor-tolerance story behind growth-axis peptides, why an injectable needs the break more than a cosmetic topical does, and why the specific 'weeks on / weeks off' numbers everyone repeats are a reasonable range rather than a settled fact.
The heart is a gland
The heart is not only a pump. It is a gland. When its walls stretch under too much blood, cardiac muscle secretes hormones, ANP from the atria and BNP from the ventricles, that tell the kidney to shed salt and water and the vessels to relax, unloading the very stretch that released them. That one circuit explains three things at once: why a blood test for BNP is how heart failure is diagnosed, why the best modern heart-failure drug does not supply the hormone but blocks the enzyme that destroys it, and why the third family member, CNP, became a treatment for a form of dwarfism. A reference read of the peptides the heart uses to manage its own load, and of why, in this family, protecting the signal beat replacing it. No dosing.
The touch deficit
Oxytocin is the bonding hormone — when it is released where bonding happens. The 'love hormone' industry sells it as a spray or an injection, but the molecule that writes pair-bonds in mammals barely crosses from blood into brain, survives roughly five minutes in plasma, and the famous behavioral studies behind it have largely failed to replicate. What couples-bonding research actually rests on is oxytocin released by touch — the one thing no bottle can substitute for. Why the needle misses the site, what the replication failures mean, where the delivery fixes that genuinely exist — better devices, tuned timing, the RAGE transporter — stop at the replication record, and why the melanocortin agonist bremelanotide — the drive half of the same circuit — is the half of a couples strategy with real clinical evidence.
The half-life, backwards
IGF-1 LR3 is a well-defined molecule with a poorly-defined human profile. The biochemistry is solid: an 83-residue, ~9,100 Da IGF-1 analog re-engineered to slip past the binding proteins while keeping full IGF-1R activity. Almost nothing downstream in humans is — starting with the half-life, where the most-repeated claim has the mechanism backwards. Evading the binding proteins doesn't lengthen IGF-1's persistence; it shortens it. A reference-tier walk through the structure, the half-life the vendor pages get wrong, the mecasermin label that's the closest thing to human data, the IGF-1/cancer evidence, WADA status, and what a certificate of analysis actually has to prove. No dosing.
The peptide exercise writes
MOTS-c isn't a pre-workout. It's a signal your muscle writes during hard training — an exercise-induced peptide transcribed from the mitochondrial genome that tells the cell to adapt. That reframes the whole timing question: the stimulus comes first, the peptide follows. What the exercise-mimetic data actually shows, why lifestyle is the real dose, where an injected protocol could plausibly fit around training (and why sport has already banned it), and how the cell's other mitochondrial peptides — the cardiolipin-protector SS-31, the cytoprotectant humanin — play a completely different game.
The last three words
α-MSH is a thirteen-residue hormone that says several things at once — colour the skin, curb the appetite, calm the inflammation. Cut it down to its final three residues and you get KPV, which still says exactly one of them: enough. It keeps the anti-inflammatory clause, drops the pigment and the appetite, and — stranger still — delivers it by breaking the rule that a peptide has to knock on a receptor from the outside. The shortest sentence in the melanocortin system, and why it still parses.
The other pedal
Muscle mass is set by two opposing forces, and this site has only told half the story — the brake, myostatin. The other half is the accelerator: the IGF-1 axis growth hormone works through, the local repair pulse a worked muscle splices for itself, and the research peptides — IGF-1 LR3, MGF — built to push it from outside. Why the accelerator is the older, stronger lever, why it's also the harder and riskier one to pull, and what the honest evidence actually supports.
Stopping Ozempic: what happens when you stop GLP-1 drugs
The people who struggle most to come off a GLP-1 are usually the ones it helped most - and that is not weakness, it is physiology doing exactly what it evolved to do. The body defends the weight it has lost, the drug holds that defense at bay, and stopping lets it back in. What the withdrawal trials actually show, why 'chronic' is the honest word for obesity, and how to think about time off: maintenance, tapering, and an off-ramp with a return ticket.
Born switched off
Myostatin, the body's brake on muscle, is manufactured pre-disabled: folded shut around its own off-switch and released into the blood inert. Arming it takes two molecular cuts by two different enzymes. That safety-catch design isn't a quirk - it's the exact vulnerability the most selective muscle drugs are built to exploit.
One receptor, a whole family
Myostatin isn't the only brake on muscle. Activin A and GDF-11 sign in through the very same receptor - and GDF-11 is so nearly identical to myostatin that a famous "rejuvenation" claim couldn't reliably tell the two apart. Why the muscle brake has a built-in backup, and why that redundancy, not any single molecule, decides how you drug it.
Where trust starts to mean something
Short synthetic peptides are cheap to make and, more to the point, cheap to verify — reversed-phase HPLC and a mass spec settle what they are in an afternoon, and a competitive market keeps them honest. Trust and availability only become live variables higher up the complexity ladder, at the folded biologics — antibodies, ligand traps, the myostatin inhibitors — where correctness lives in the fold, not the sequence, and only a bioassay can confirm it. Why complexity, not fraud, governs what you can actually source.
Cataloged vs. reachable
A reference will tell you what a molecule is down to the receptor. None of them tell you the thing you actually act on: whether you can get it, in what form, this month. Availability is the frontier's real bottleneck — and unlike a sequence, it moves. In July 2026 it moved in public: an FDA advisory committee voted, over its own scientists' objections, to recommend six research peptides for pharmacy compounding — and changed nothing yet. Why what's reachable is itself information, what a vote does and doesn't do to it, and how a reference can publish it as data without becoming a store.
How long does semaglutide last? Two minutes to seven days
Natural GLP-1 is destroyed about two minutes after the gut releases it. Semaglutide — the same signal, essentially the same shape — survives a week. That gap is a five-thousand-fold engineering feat, and it wasn't achieved by making the peptide tougher. It was achieved by attaching it to something the body has already decided to keep. Inside the three tricks that turned a disposable hormone into a weekly drug, and why the next cadence is monthly.
GIP agonist vs antagonist: why both cause weight loss
Tirzepatide turns the GIP receptor on. Maridebart cafraglutide turns the same receptor off. Both drive weight loss. That should be impossible under a "more signal is better" model of pharmacology — and the fact that it isn't is one of the most revealing puzzles on the metabolic frontier. How a receptor can be pushed either way and land in the same place.
What is amylin? Insulin's forgotten twin
Every time the pancreas releases insulin, it releases a second hormone from the same granule — amylin. For a century insulin took all the credit while its partner went unnamed. Amylin runs a receptor it borrowed rather than built, sabotages itself by clumping into the amyloid that scars a diabetic pancreas, and is now the metabolic frontier's newest axis. The story of the twin the body never forgot.
Is there a GLP-4?
Single agonist, then dual, then triple — so the next breakthrough must be a bigger number, right? There is no GLP-4. GLP-1's "1" was never a version number, and seeing why reveals where the field is actually going: not up the same axis, but sideways — into amylin, into GIP antagonism, into the pill and the quality of the weight you keep.
Why are peptides injected? The delivery problem
A peptide is, chemically, food — swallow it and your gut digests it; rub it on your skin and it never gets past the surface. The needle isn't a preference, it's physics. Inside the barriers that block every other route, why most "oral" peptide claims are marketing, and the narrow cases where a nasal spray genuinely works.
Where the powder comes from
Most of the world's finished, lyophilized peptide traces back to a handful of Chinese manufacturing lines — and China's 2026 drug law just pushed that supply chain toward real traceability. The strategic read: "American made" often marks more hand-offs, not fewer, and in lyophilized peptide every hand-off is an irreversible chance to lose purity. Fewest transfers, finished direct, wins.
The arm that points at the liver
Every co-agonist so far added receptors like verses — one, then two, then three. Survodutide adds only one to GLP-1, and picks the strange one: the glucagon receptor, the same hormone a diabetes drug is built to fight. The bet is that glucagon's other job — spending energy and burning the liver's own fat — is worth its cost in glucose, so long as a GLP-1 arm is there to hold the line. It is why survodutide's headline result is not its ~19% weight loss but a class-leading liver-disease signal and a Breakthrough Therapy tag in MASH. Now it is reaching the grey market ahead of its phase 3 verdict, so here is what the near-future buyer is actually reaching for: why glucagon earns a seat, why the liver and not the scale is the story, why the dose has to be crept up rather than started at target, and how far availability has run in front of the evidence. Bullish on the science, sceptical on the page. No dosing.
The community found it first
Long before GLP-1 went mainstream, the community running peptide protocols was the field's informal R&D — and their demand shaped the catalogs of synthesis companies and compounding pharmacies. Their core instinct, foundation first, is exactly what the metabolic era is proving right.
From insulin to the peptide boom
A hundred years of peptide medicine — the first miracle, the families it built, and how the molecules quietly running your body became a 2026 cultural phenomenon.