Chapter 20: Endorphins, Enkephalins, and the Opioid Peptides — Your Body's Natural Painkillers

"Pain — has an Element of Blank — It cannot recollect When it begun — or if there were A time when it was not —" — Emily Dickinson, c. 1862

Overview

Almost everything the public knows about this system arrived through two channels: a poppy and a jogging magazine. Neither is a good teacher.

The poppy gave us morphine, and morphine gave us a set of assumptions — that opioids are a foreign chemistry we borrowed from a plant, that the relief they produce is an intrusion into normal physiology, that the receptors they occupy are a kind of unfortunate lock that happened to fit a dangerous key. The jogging magazine gave us "endorphins," a word that now functions less as a class of molecules than as a vague synonym for feeling good after effort, invoked to explain runner's high, chocolate, laughter, spicy food, and cold plunges, usually with no evidence attached to any of them.

Both stories get the direction of causation wrong, and the correction is the most important idea in this chapter. Your opioid receptors did not evolve to bind poppy alkaloids. They evolved to bind peptides your own cells make. Morphine works on you because it resembles something you already are.

That reframe pays off immediately. Once you see the system as an endogenous one, the interesting questions change. Not what does the drug do to me, but what is this system doing all day, and what happens when a drug shoves it. Not do endorphins make runners happy, but what would it take to show that, and does anybody have it. And, in what I think is the single most elegant experiment in this book, not is the placebo effect real, but can we block it with a receptor antagonist — a question that turns a philosophical argument about reporting bias into a pharmacology experiment with an answer.

This chapter also has to handle opioid tolerance, dependence, and addiction, and it will do that as receptor biology and clinical epidemiology, not as a morality play. Two of those concepts are routinely confused in public discussion, and that confusion has caused documented harm to people living with pain. Untangling them is worth doing carefully.

In this chapter, you will learn to:

• Explain why the receptor came first and the drug came second, and what that reframe buys you
• Trace three peptide families from three precursor genes, including one precursor you already met in a completely different chapter
• Predict what mu, delta, and kappa activation each produce, and why one of them makes you feel bad
• Describe descending inhibition and say what "pain is actively regulated" actually means mechanically
• Give an honest account of the runner's high, including what the famous measurement cannot show
• Separate physical dependence from addiction cleanly enough to explain the difference out loud
• Read the naloxone-reversal experiments and state precisely what they establish and what they do not
• Dismantle an "endorphin-boosting" supplement claim in four moves
• Explain why a genuinely effective peptide analgesic has to be injected into spinal fluid

Learning Paths

All five paths should read §20.7. The placebo-and-naloxone material is the best evidence in this book that an endogenous peptide system produces a real, measurable human experience — and it is the cleanest demonstration of a study design you will use everywhere else.

💊 GLP-1 — §20.2 (POMC) connects directly back to Chapter 13's appetite circuitry; §20.6 is the general theory of what sustained agonism does to any receptor, GLP-1 included. 🏋️ Performance — §20.5 and §20.8 are yours. Between them they cover the two claims you will meet most often, and both turn on the same measurement error. 🔬 Science — read straight through. §20.3 and §20.4 are the mechanistic core; §20.9 is the medicinal-chemistry payoff. 💄 Cosmetic — the lightest chapter in the book for you, but §20.8's structure — magnitude, duration, existence proof, delivery — is exactly the argument Chapter 30 makes about topical peptides. Learn it here where the physiology is simpler. 🏥 Clinical — §20.6 and §20.7 are the two sections most likely to change a conversation you have with a patient. §20.6's dependence/addiction distinction in particular.

20.1 Morphine mimics us, not the reverse

Here is the sentence that reorganizes the whole field: the receptors were found before the ligands.

In the early 1970s, several laboratories independently demonstrated that nervous tissue contains specific, saturable, high-affinity binding sites for opiate drugs — sites that bound morphine and its relatives with the stereochemical fussiness characteristic of a real receptor rather than a nonspecific stickiness. The binding was blocked by naloxone, an antagonist. The distribution of the sites across brain regions tracked, roughly, the regions you would expect to matter for pain and mood.

That result immediately created a problem, and the problem is the good part. Evolution does not build elaborate, stereochemically selective receptor systems on the off chance that a human being will one day scrape a seed pod. If the brain has a high-affinity binding site for a plant alkaloid, the overwhelmingly likely explanation is that the site exists for something else — something the body makes — and the alkaloid is a coincidental fit.

Within a few years, that something else was found. The first were two pentapeptides isolated from brain tissue, named enkephalins — from the Greek for "in the head." Others followed: a thirty-one-residue peptide named β-endorphin, a contraction of "endogenous morphine," and later a family named dynorphins for their unusual potency. By the end of the decade the picture had inverted. Opiate drugs were no longer foreign chemistry acting on a mysteriously receptive brain. They were impostors — molecules that happen to present the same chemical face as a peptide the body already uses.

This is not a semantic distinction. Three concrete things follow from it.

First, the side effects are not accidents. Respiratory depression, constipation, itch, euphoria, sedation, and tolerance are not contaminants of an otherwise clean analgesic. They are what happens when you activate the mu receptor everywhere at once, because the mu receptor does all of those things natively in different places. A drug that hits the receptor cannot decline the parts of its job that you did not want. Chapter 2's discussion of receptor distribution predicted exactly this before you knew what receptor we were talking about.

Second, the endogenous system is already running. You are not opioid-naive in any meaningful sense. Enkephalins are being released in your spinal cord right now, modulating signals that will never reach your awareness. Whatever an exogenous opioid does, it does on top of an existing tone — which is why withdrawal is not simply the absence of a drug but the exposure of a system that adapted to the drug's presence (§20.6).

Third, "endorphin" is not a synonym for "the feeling." β-endorphin is a specific 31-residue peptide with a known precursor, receptor preference, and distribution. When a wellness article says an activity "releases endorphins," it is almost never reporting a measurement of that peptide. It is asserting a mechanism to explain a mood — and Chapter 5's discipline applies at full force: knowing how something would work is not evidence that it does.

🧬 The Molecule — why a peptide and an alkaloid look the same to a receptor

Morphine and met-enkephalin do not resemble each other on paper. One is a rigid five-ring alkaloid; the other is a floppy five-residue peptide. A chemist shown both without context would not group them.

A receptor is not looking at the paper. It is looking at a small number of contact points arranged in space. Both molecules present the same essential pair: an aromatic ring bearing a hydroxyl group (the phenol of morphine's A ring; the phenol of the enkephalin's N-terminal tyrosine) held a certain distance from a basic nitrogen that is protonated at body pH (morphine's tertiary amine; the enkephalin's free N-terminal amino group). Everything else about the two molecules differs. That pharmacophore does not.

This is why the N-terminal tyrosine is nearly untouchable in every endogenous opioid peptide. Remove it, or acetylate its amino group, or swap it for phenylalanine, and opioid activity collapses. §20.2 shows you the shared motif written out, and one member of the extended family that does start with phenylalanine — with exactly the consequence you would predict.

Notice what this argument is and is not. It explains a fit. It does not, on its own, establish that any particular molecule does any particular thing in a person. Structural plausibility is a reason to run an experiment, never a substitute for one — and in the peptide marketplace, a story about shape standing in for a trial is the single most common failure mode you will meet.

20.2 Three families, three precursors

Endogenous opioid peptides are not synthesized as such. Like nearly every peptide in Part II, they are cut out of larger precursor proteins by processing enzymes, and the cutting pattern differs by tissue. Three genes produce three precursors, and each precursor yields a family.

THREE GENES, THREE PRECURSORS, THREE FAMILIES

  GENE     PRECURSOR PROTEIN          PRINCIPAL OPIOID PRODUCTS
  ──────────────────────────────────────────────────────────────────────────
  POMC  →  proopiomelanocortin    →   β-endorphin (31 residues)
           …which ALSO yields:        ACTH · α-MSH · β-lipotropin · γ-MSH · CLIP
                                      └── a stress axis and an appetite system,
                                          from the same precursor (Ch 13)

  PENK  →  proenkephalin          →   met-enkephalin (5 residues) ×several copies
                                      leu-enkephalin (5 residues)
                                      plus extended enkephalin fragments

  PDYN  →  prodynorphin           →   dynorphin A (17 residues)
                                      dynorphin B · α- and β-neoendorphin

  ── a fourth, structurally related, conventionally kept separate ──
  PNOC  →  prepronociceptin       →   nociceptin / orphanin FQ (17 residues)
                                      acts at the NOP receptor
                                      NOT blocked by naloxone

The processing is tissue-specific, which matters more than it sounds. The same POMC precursor produces adrenocorticotropic hormone in one pituitary cell type and α-melanocyte-stimulating hormone plus β-endorphin in another, depending on which prohormone convertases that cell expresses. One gene, several cell types, several entirely different physiological outputs. This is a general lesson about peptide biology and not a curiosity: asking "what does this gene do" is often the wrong question. The right one is "what does this gene do in this cell."

Now look at the sequences, because the family resemblance is visible at a glance.

THE OPIOID MESSAGE — the first four residues are the family resemblance

  met-enkephalin     Tyr–Gly–Gly–Phe–Met
  leu-enkephalin     Tyr–Gly–Gly–Phe–Leu
  β-endorphin        Tyr–Gly–Gly–Phe–Met–Thr–Ser–Glu–Lys–Ser–…      (31 residues)
  dynorphin A        Tyr–Gly–Gly–Phe–Leu–Arg–Arg–Ile–Arg–Pro–…      (17 residues)
                     └────── shared ──────┘└────── differs ──────┘
                          "the message"          "the address"

  nociceptin/OFQ     Phe–Gly–Gly–Phe–Thr–Gly–Ala–Arg–Lys–Ser–…      (17 residues)
                      ↑
                      tyrosine replaced by phenylalanine — one missing hydroxyl,
                      and the peptide no longer activates classical opioid
                      receptors or responds to naloxone

This is the classic message-address arrangement, and it is one of the most useful organizing ideas in peptide pharmacology. The shared N-terminal Tyr-Gly-Gly-Phe carries the message: it is the part that says "opioid" to a receptor, and it is the part morphine imitates. Everything C-terminal to it is the address: it determines which of the three receptors the peptide prefers, how long it survives, and where it can travel. Dynorphin A's string of arginines and lysines gives it a strong positive charge and a kappa preference. β-endorphin's long tail gives it high affinity for mu and delta and a much longer half-life than an enkephalin, which is degraded within seconds.

And then there is nociceptin. Same architecture, same receptor superfamily, one substitution at position 1 — phenylalanine where the others have tyrosine — and the peptide is functionally outside the opioid system. It has its own receptor, NOP, and naloxone does essentially nothing to it. That single difference is the message pharmacophore of §20.1 demonstrated by subtraction.

🧬 The Molecule — one precursor, two unrelated physiologies

Chapter 13 spent a long time on α-MSH, the melanocortin-4 receptor, and the appetite circuitry that setmelanotide targets. This chapter is about pain and mood. They have nothing in common functionally. They come from the same protein.

Proopiomelanocortin is a single precursor, roughly 240 residues, that is chopped into a startling assortment of products: ACTH (adrenal stress axis), α-MSH (pigmentation and appetite), β-endorphin (opioid), β-lipotropin, γ-MSH, and CLIP. Which products a cell makes depends on which convertases it expresses. Anterior pituitary corticotrophs favor ACTH. Hypothalamic arcuate neurons and the intermediate lobe favor α-MSH and β-endorphin.

Two consequences are worth carrying forward.

First, a mutation in one gene can produce a syndrome that looks like several unrelated diseases at once. POMC deficiency presents with adrenal insufficiency, red hair, and severe early-onset obesity — three findings from three different products of one broken precursor. Chapter 13 covered the obesity arm.

Second, "the POMC system" is not a unit of analysis, and neither is "the opioid system" in the loose sense. The peptides that come out of a precursor are not a team. They are separate signals that happen to share a manufacturing origin, and they can be regulated together, separately, or in opposition. This is exactly the problem the Dossier section at the end of this chapter is built to handle.

🔍 Check Your Understanding

1. Nociceptin has the sequence Phe-Gly-Gly-Phe at its N-terminus rather than Tyr-Gly-Gly-Phe. Using §20.1, predict what that single change does to its interaction with the mu receptor, and why.
2. A paper reports that a cell line "expresses POMC." What have you learned about which peptides that cell actually secretes, and what would you still need to know?

20.3 Mu, delta, kappa — three receptors, three different jobs

All three classical opioid receptors are G protein-coupled receptors of the kind Chapter 2 dissected, and all three couple through inhibitory G proteins. The downstream consequences are consistent across the family: adenylyl cyclase is inhibited, potassium channels open and hyperpolarize the neuron, and voltage-gated calcium channels close so that less neurotransmitter is released.

Opioid signaling is, at the cellular level, almost entirely a braking system. It does not excite neurons. It makes them harder to fire and quieter when they do. Every effect discussed in this chapter — analgesia, euphoria, constipation, respiratory depression — is downstream of applying that brake in a specific place.

Where the three receptors differ is in where they are expressed, which peptides prefer them, and therefore what braking those particular circuits produces.

THREE RECEPTORS, THREE JOBS — all inhibitory GPCRs (Ch 2)

  RECEPTOR    PREFERRED ENDOGENOUS      WHAT ACTIVATION PRODUCES
              LIGANDS
  ────────────────────────────────────────────────────────────────────────────
  MU  (μ)     β-endorphin               strong analgesia (brain + spinal cord)
  OPRM1       enkephalins               EUPHORIA
                                        RESPIRATORY DEPRESSION
                                        constipation, itch, miosis, sedation
                                        tolerance and dependence
                                        ← the target of morphine and of nearly
                                          every clinical opioid analgesic

  DELTA (δ)   enkephalins               analgesia, generally more modest
  OPRD1       β-endorphin               mood effects; antidepressant-like in
                                        animal models
                                        seizure risk with some agonists

  KAPPA (κ)   dynorphins                analgesia (real, spinal especially)
  OPRK1                                 DYSPHORIA and aversion — NOT euphoria
                                        sedation, diuresis, hallucination
                                        minimal respiratory depression
                                        low abuse liability

  NOP         nociceptin / orphanin FQ  context-dependent; anti-reward in some
  OPRL1                                 circuits; naloxone-insensitive

The mu receptor is where the clinical action is, and it is a package deal. The same receptor that produces the analgesia produces the respiratory depression that kills people in overdose. This is not a manufacturing defect that a cleverer molecule has yet failed to fix; decades of medicinal chemistry have gone at the problem and the separation has proven extraordinarily stubborn. Chapter 2's Case Study 2 examined the biased-agonism approach to that problem at the mu receptor in detail, and I will not retell it here — but keep its conclusion in view when you read §20.9.

Delta is the least clinically developed of the three. Delta agonists produce analgesia and antidepressant-like effects in animal models and have repeatedly run into convulsant liability in development. It remains a live target rather than a delivered one.

Kappa is the strange and instructive one. Kappa agonists are genuinely analgesic — this is not a mechanism-only claim. They also produce dysphoria: a reliable, unpleasant, sometimes frankly distressing altered state, along with sedation and, at higher exposures, hallucination. Dynorphin signaling has been implicated in stress responses and in the aversive components of withdrawal.

From a drug-development standpoint this makes kappa maddening. An analgesic that people actively do not enjoy has, almost by construction, low abuse liability — you cannot get a compulsive-use disorder around a molecule that makes users feel worse. That is the exact property the field has spent forty years wanting. It comes packaged with the exact property that makes patients refuse the drug. The central lesson of §20.1 keeps reappearing: you cannot pick and choose among a receptor's effects, because the receptor does not distinguish between the ones you wanted and the ones you did not.

💊 In the Clinic — what the receptor map predicts about a real prescription

Take a patient started on a conventional mu agonist for postoperative pain. The receptor map above predicts, before you have seen the patient, most of what will happen.

Analgesia, because mu receptors sit in the periaqueductal gray, the rostral ventromedial medulla, and the spinal dorsal horn (§20.4). Constipation, because mu receptors densely populate the enteric nervous system and inhibiting those neurons slows gut motility — and, notably, this effect shows little tolerance, which is why constipation persists for months while analgesia fades. Itch, which is mu-mediated and not an allergy, a distinction that gets patients incorrectly labeled as morphine-allergic for life. Miosis, small pupils, useful clinically as a sign of exposure. Respiratory depression, dose-dependent and the mechanism of fatal overdose. Nausea, via the chemoreceptor trigger zone.

Two things about that list are worth stating plainly. It is derivable from receptor distribution alone — this is Chapter 2 doing real predictive work. And different effects develop tolerance at very different rates, which is why a stable dose can drift over weeks into a state where the analgesia has diminished but the constipation has not.

None of this is a treatment recommendation, and nothing in this book should inform a decision about anyone's analgesia. It is here so that a conversation with a clinician can be a more useful one.

20.4 Pain modulation and descending inhibition

The intuitive model of pain is a wire. Tissue is damaged, a nerve fires, a signal travels to the brain, and the brain reports what arrived. On that model, the amount of pain is a readout of the amount of damage, and any discrepancy between them is a psychological failure of some kind.

That model is wrong, and it has been known to be wrong for sixty years. The signal is not merely transmitted. It is modulated at its first synapse by circuitry descending from the brain, and the endogenous opioid peptides are a principal currency of that modulation.

DESCENDING INHIBITION — pain is regulated on the way in

   PERIPHERY               SPINAL CORD                 BRAIN
   nociceptor  ─────────→  dorsal horn  ───────────→   thalamus → cortex
   (tissue damage)         (FIRST SYNAPSE)             (perception)
                                ▲
                                │  descending
                                │  inhibition
                                │  (endogenous opioids act HERE)
                           ┌────┴────┐
                           │   RVM   │   rostral ventromedial medulla
                           └────┬────┘
                                ▲
                           ┌────┴────┐
                           │   PAG   │   periaqueductal gray
                           └────┬────┘
                                ▲
                    input from hypothalamus, amygdala,
                    and prefrontal cortex — that is, from
                    CONTEXT, THREAT APPRAISAL, and EXPECTATION

Read that diagram from the bottom up, because the bottom is the surprising part. The periaqueductal gray receives input from regions that handle context, emotional salience, and anticipation. It projects to the rostral ventromedial medulla. The RVM projects down the spinal cord to the dorsal horn, where the incoming nociceptive signal makes its first synapse. Opioid peptides released in that region suppress transmission — presynaptically, by reducing neurotransmitter release from the incoming fiber, and postsynaptically, by hyperpolarizing the projection neuron. Both are the inhibitory mechanisms of §20.3 applied at a specific address.

The result is that the brain has a volume control on its own pain input, and the control is wired to appraisal. A signal from a damaged toe does not arrive at consciousness with a fixed magnitude. It arrives having already passed through a gate whose setting depends on what else is happening.

Two lines of experimental evidence anchor this, and both are old and robust.

Electrical stimulation of the periaqueductal gray produces analgesia. In classic animal work, stimulation in this region produced antinociception profound enough to permit surgery without anesthesia, in an animal that remained otherwise alert and responsive. The analgesia was regionally specific and reproducible. This is not a subtle modulation; the descending system can, when driven hard, largely close the gate.

That stimulation-produced analgesia is reduced by naloxone. Give an opioid receptor antagonist, and the analgesia produced by stimulating the brain diminishes. That result is the linchpin: it says the analgesia is not merely electrical disruption but is mediated by endogenous opioid transmission. The brain, when stimulated in the right place, releases its own opioids, and blocking the receptor they act on blocks the effect.

Hold onto that experimental logic, because §20.7 runs the identical design on a much stranger input than an electrode.

Descending modulation is not purely inhibitory, which is a detail that gets flattened in popular accounts and matters clinically. The RVM contains cell populations that facilitate as well as populations that inhibit; the system can turn pain transmission up as well as down. Descending facilitation has been implicated in the maintenance of some chronic pain states and in opioid-induced hyperalgesia, a phenomenon in which sustained opioid exposure paradoxically increases pain sensitivity. The gate is not a dimmer that only goes one way.

The clinical translation of this section is short and consequential. Pain is not a passive readout of tissue damage; it is actively regulated, and the regulation is real biology, not attitude. A patient whose pain is disproportionate to their imaging is not necessarily reporting badly. They may have a descending modulatory system that is set differently, which is a physiological statement, not a psychological accusation.

🔍 Check Your Understanding

1. An opioid drug given systemically reaches receptors in the periaqueductal gray, in the RVM, and in the spinal dorsal horn simultaneously. Using the diagram, explain why that produces stronger analgesia than acting at any one of those sites alone.
2. Naloxone reduces the analgesia produced by electrically stimulating the periaqueductal gray. What exactly does that establish that the stimulation result alone does not?

20.5 The runner's high — where the evidence is more interesting than the story

Here is the popular account, which you have heard: prolonged intense exercise causes the release of endorphins, and endorphins cause the euphoric, floaty, pain-tolerant state that distance runners report. It is repeated in fitness journalism, in medical school lecture halls, and in the copy of approximately every wellness product ever sold.

The state is real. Many people do experience a distinctive mood change during or after prolonged exercise, and it is measurable on standard mood instruments and in pain thresholds. The question is what causes it, and the honest answer is that we do not fully know, for reasons that make this one of the most instructive case studies in the book.

Where the story came from. In the late 1970s and 1980s, investigators measured β-endorphin in the blood of exercising subjects and found that it rose with intense or prolonged exertion. The finding was robust, replicated, and correctly reported. It coincided with a running boom and with the recent, exciting discovery of endogenous opioids, and the inference practically made itself: exercise raises endorphins, endorphins are the body's morphine, runners feel euphoric, therefore endorphins cause the runner's high.

Why that inference does not hold. The problem is the compartment.

β-endorphin measured in plasma is largely of pituitary origin, released alongside ACTH as part of a stress response. β-endorphin is a 31-residue peptide, and peptides of that size do not freely cross the blood-brain barrier — Chapter 22 is entirely about why, and it is the single most important constraint in this part of the book. Circulating β-endorphin therefore cannot be assumed to reach the brain regions that would produce euphoria, and central β-endorphin release is not measured by a blood draw. Two pools, one name, and no established quantitative relationship between them.

This is not a technicality. It is the same error, in a different costume, that Chapter 4 warned about with oral peptides and Chapter 15 warned about with growth-hormone secretagogue claims: a peptide measured where it is easy to measure is not evidence about a peptide where it acts. Blood is easy. Brain is hard. The measurement that made the story famous is the one that establishes the least.

What better evidence exists. Two lines, pointing in interestingly different directions.

Brain imaging has been used to look for central opioid release directly. Work published in 2008 used positron emission tomography with an opioid receptor ligand in trained athletes before and after a long endurance run, and reported reduced ligand binding in frontal and limbic regions afterward — consistent with the receptors being occupied by something the brain had released — with the reduction correlating with self-reported euphoria. That is far better evidence than a blood level. It is also a small study with a design that shows association rather than causation.

Meanwhile, a separate literature has implicated the endocannabinoid system. Endocannabinoids such as anandamide rise with exercise, and — critically — they are small lipid molecules that cross the blood-brain barrier much more readily than a 31-residue peptide. In rodent work, the anxiety-reducing and pain-threshold effects of voluntary wheel running were abolished by blocking cannabinoid CB1 receptors and were not abolished by opioid antagonism. A human study administering an opioid antagonist before running reported that exercise-induced euphoria persisted.

🔬 Read the Study

```text FIGURE 20.1 — "What actually causes the runner's high?" [contested]

THE STUDY Not one study — three lines of work that disagree productively. (a) PET imaging with an opioid-receptor ligand in a small group of trained athletes, scanned before and after roughly two hours of endurance running (published 2008). (b) Rodent work on voluntary wheel running, testing whether the resulting anxiolysis and raised pain thresholds survive blockade of cannabinoid CB1 receptors versus blockade of opioid receptors (published 2015). (c) A human experiment giving an opioid receptor antagonist or placebo before a run and measuring euphoria and anxiety afterward (published 2021).

THE QUESTION Are endogenous opioids necessary for the mood change that follows prolonged exercise — or merely present during it?

WHAT IT SHOWS (a) Reduced radioligand binding in frontolimbic regions after running, consistent with endogenous opioid release, correlating with reported euphoria. Something opioid-like does happen centrally. (b) In mice, the effects were cannabinoid-dependent and survived opioid blockade — evidence that the endocannabinoid system is sufficient to produce the phenomenon in that model. (c) In humans, opioid blockade did not abolish exercise-induced euphoria.

WHAT IT DOESN'T (a) Cannot establish causation. Displacement of a tracer plus a correlation with a mood rating is an association, in a small sample of unusually trained subjects. (b) Is a mouse, and "runner's high" in a mouse is an operational proxy — anxiety-like behavior and a paw-withdrawal threshold, not euphoria. (c) Is a null result in one paradigm at one antagonist exposure. A null does not exclude a partial contribution, and it does not generalize to every exercise mode or duration.

THE VERDICT Endogenous opioids are released during prolonged exercise and are plausibly involved in some of what follows. They have not been shown to be necessary, and the strongest causal evidence currently available points at the endocannabinoid system. Both may contribute; they are not exclusive.

THE LESSON The evidence everybody quotes — rising blood β-endorphin — is the weakest item on this list, because it measures the wrong compartment. The evidence that would settle it is an antagonist experiment, which is exactly the design §20.7 uses to spectacular effect on a different question. When a peptide cannot be measured where it acts, block its receptor and see what changes. ```

The honest position, then, is a three-part sentence that no headline will ever print: endogenous opioids are probably involved in the exercise-induced mood state; the peripheral measurement that made the story famous does not establish it; and the endocannabinoid system is a serious competing or complementary explanation with better causal evidence behind it.

Notice that this conclusion is unsatisfying in a specific and healthy way. Nothing here says the runner's high is fake. Nothing says endorphins are irrelevant. What it says is that a confident causal claim has been circulating for forty years on the strength of a measurement that cannot support it — and that the field's own better experiments have complicated rather than confirmed it.

📊 Evidence Rating

Claim: Endorphins cause the runner's high — that is, endogenous opioid peptides are the primary mediator of the euphoric, analgesic mood state that follows prolonged intense exercise in healthy adults. Rating: ⚠️ Promising but preliminary (as of 2026) Why: Central opioid release during prolonged exercise has real imaging support, but the famous supporting evidence — rising plasma β-endorphin — measures a compartment that does not communicate freely with the brain, and antagonist experiments in rodents and in humans have failed to abolish the effect while cannabinoid blockade did. What would change it: Adequately powered human studies using opioid receptor antagonism, with pre-registered mood and pain-threshold endpoints, across multiple exercise modalities — ideally with a factorial design that blocks opioid and cannabinoid signaling separately and together. A consistent abolition under opioid blockade would move this to ✅; consistent survival under opioid blockade with abolition under CB1 blockade would move it to ❌.

20.6 Reward, tolerance, and dependence

Chapter 2 established the general principle in §2.7 and §2.8: sustained agonism at a G protein-coupled receptor triggers adaptive responses. The receptor is phosphorylated, arrestin proteins are recruited, signaling is uncoupled from the G protein, receptors are internalized, and with prolonged exposure the total number of receptors at the surface falls. The cell also adjusts downstream — in the opioid case, by upregulating the adenylyl cyclase machinery that the receptor was inhibiting, so that the same receptor activation now produces less net effect.

Opioid tolerance is that mechanism at its most clinically consequential. There is nothing exotic about it. It is the same desensitization that makes a continuous GLP-1 signal behave differently from a pulsatile one (Chapter 3), and the same reason chronic agonist exposure is generally a worse pharmacological idea than intermittent exposure. It simply happens to be attached to a system where the consequences of needing more drug are severe.

Three distinct phenomena result, and they are routinely conflated even in clinical settings.

Tolerance is diminished effect at a constant exposure. It develops at different rates for different effects — quickly for euphoria and analgesia, slowly or not at all for constipation and only partially for respiratory depression. That last asymmetry is a large part of why escalating exposure is dangerous: the effect you are chasing fades faster than the effect that can kill you.

Physical dependence is a state in which the body has adapted to the drug's presence such that abrupt removal produces a withdrawal syndrome. In the opioid case that syndrome is well characterized — autonomic arousal, gastrointestinal distress, muscle and bone aching, insomnia, profound restlessness, anxiety — and it is generated in substantial part by the very adaptations that produced tolerance, now unopposed. It is a predictable pharmacological consequence of sustained exposure.

Addiction — the clinical entity is opioid use disorder — is a different thing. It is defined by a pattern of behavior: compulsive use, loss of control over use, craving, and continued use despite harm. It involves reward and reinforcement circuitry, particularly mu receptor effects on dopaminergic signaling in the mesolimbic system, along with a large set of genetic, environmental, and social contributors.

The distinction between the second and third is not academic, and getting it wrong causes documented harm.

Physical dependence occurs in essentially anyone maintained on opioids long enough, including patients taking them exactly as prescribed for a legitimate indication. A person with cancer pain on stable long-term therapy is physically dependent. So, for that matter, is a patient on long-term beta-blockers or certain antidepressants, in the sense that abrupt cessation produces a discontinuation syndrome — a comparison worth making precisely because nobody describes those patients as addicted.

Dependence is not addiction. It is not a moral status, a diagnosis, a prediction of future behavior, or evidence of anything about a person's character. It is a receptor adaptation.

When the two are conflated — in clinical documentation, in policy, in family conversations — the result is a set of specific harms: patients with severe pain undertreated because dependence was read as addiction; abrupt discontinuation of long-term therapy, which produces withdrawal and in some cases worse outcomes than continuation; patients avoiding disclosure to clinicians; and stigma attached to people whose only relevant feature is that their receptors adapted the way receptors do.

🩺 Safety and Risk — three words that are not synonyms

Tolerance. The same exposure produces less effect. A receptor-level adaptation. Develops at different rates for different effects, which is itself a safety issue.

Physical dependence. Cessation produces withdrawal. Occurs in anyone on sustained therapy, including patients using their medication correctly. Not a behavior, not a diagnosis, not a character finding.

Addiction (opioid use disorder). A clinical diagnosis defined by compulsive use and continued use despite harm, involving reward circuitry and a wide set of biological and social contributors.

A patient can have all three, any two, or — commonly, in long-term therapy — the first two without the third. A person can also develop the third without ever having been prescribed anything.

Why this section exists: the conflation of dependence with addiction has caused real, documented harm to people living with pain — undertreatment, abrupt discontinuation, and stigma. It has also obscured the actual clinical picture for people who do have opioid use disorder, which is a treatable medical condition for which effective medications exist and for which outcomes are substantially better with treatment than without.

This book does not evaluate, recommend, or comment on anyone's analgesic regimen. Decisions about opioid therapy, tapering, or treatment for opioid use disorder belong with a clinician who knows the person's history — and, for the last of those, with clinicians who specialize in it. If you are reading this because of a situation in your own life, that is the correct next step, and it is the only one this chapter can honestly point you toward.

There is one more piece worth stating, because it bears directly on §20.8. The endogenous system does not produce this problem. Enkephalin release in the dorsal horn is phasic, local, spatially restricted, and terminated within seconds by peptidases. Tolerance and dependence are consequences of sustained, whole-body, supraphysiological receptor occupancy — a pattern that no endogenous release produces. That is Chapter 3's central thesis, and the opioid system is its most vivid demonstration: the difference between a hormone and a drug is often not the molecule. It is the schedule.

20.7 Placebo analgesia and naloxone

This is the best experiment in the chapter and possibly in the book, and it deserves to be understood precisely rather than admired vaguely.

The setup. Placebo analgesia is well documented: administer an inert substance with the expectation that it will reduce pain, and a substantial fraction of people report less pain. This has been known for as long as controlled trials have existed, and for most of that time it was regarded, by serious people, as an interpretive nuisance — a mixture of reporting bias, politeness toward the investigator, regression to the mean, and natural fluctuation in symptoms. Under that reading, nothing happens in the patient. Only the report changes.

The experiment. In 1978, investigators studying postoperative dental pain took the patients who had responded to a placebo with genuine pain relief, and gave them naloxone — an opioid receptor antagonist with no analgesic activity of its own. If placebo analgesia were merely a change in reporting, naloxone should do nothing to it; it is a receptor blocker, and there is no receptor involved in politeness.

Naloxone reduced the placebo analgesia.

Sit with the logic of that for a moment, because it is doing something remarkable. It converts a question about psychology into a question about pharmacology and then answers it. If a drug that blocks a specific receptor removes an effect, that effect was being produced through that receptor. The expectation of relief was not merely changing what patients said. It was engaging the descending inhibitory system of §20.4, causing release of endogenous opioid peptides, and producing measurable analgesia through the same receptors morphine occupies.

🔬 Read the Study

```text FIGURE 20.2 — "Naloxone reversal of placebo analgesia" [landmark]

THE STUDY A line of work spanning four decades. (a) The original: patients with postoperative pain following third-molar extraction, published in The Lancet in 1978. Placebo responders were given naloxone, an opioid receptor antagonist. (b) The dissection: experiments published in 1999 separating placebo analgesia induced by verbal expectation from placebo analgesia conditioned by prior exposure to an actual drug, and testing each for naloxone sensitivity. (c) The imaging: PET studies published from 2005 onward measuring mu-opioid receptor availability during placebo analgesia.

THE QUESTION Is placebo analgesia a real neurochemical event, or a change in what people report?

WHAT IT SHOWS (a) Naloxone substantially reduced the pain relief produced by placebo. An opioid receptor antagonist should have no purchase on a reporting bias. (b) Placebo analgesia induced by verbal suggestion was naloxone-reversible. Placebo analgesia conditioned by prior administration of a NON-opioid analgesic was largely NOT naloxone-reversible — the conditioning had recruited a different system. Expectation and conditioning are dissociable mechanisms. (c) Placebo administration under expectation of relief was accompanied by changes in mu-opioid receptor availability in pain-regulatory regions, and the magnitude tracked reported analgesia.

WHAT IT DOESN'T · Does NOT show that placebo can substitute for treatment. The effects are real, generally modest, variable between people, and studied mostly in acute experimental and post-surgical pain. · Does NOT show that all placebo effects are opioid-mediated. Several demonstrably are not — the conditioned non-opioid placebo in (b); dopamine release in placebo response in Parkinson's disease; nocebo hyperalgesia, which has been linked to cholecystokinin rather than to opioid signaling. · Naloxone is not a perfectly clean probe. It can alter pain sensitivity on its own in some conditions, which complicated the early interpretation and required later designs — hidden administration, balanced placebo — to address.

THE VERDICT Expectation of relief engages the endogenous opioid system, and blocking that system removes a substantial part of the resulting analgesia. Placebo analgesia is, at least in part, a pharmacology.

THE LESSON The strongest evidence in this chapter that endogenous opioid peptides produce real human experience does not come from measuring a peptide. It comes from blocking a receptor and watching an experience change. When you cannot measure a peptide where it acts — which, per Chapter 22, is most of the time in the brain — the antagonist experiment is often the only honest test available. Learn to look for it. ```

Three implications, each of which people routinely overreach on.

What this establishes. Expectation produces a real neurochemical event. A belief about what is about to happen is transduced, through prefrontal and limbic input to the periaqueductal gray, into opioid peptide release in the descending pain-modulatory system, which measurably changes pain transmission. The route from "I expect this to help" to "less pain signal reaches my cortex" is now a traced anatomical and chemical path, not a metaphor.

What this does not establish. It does not establish that placebo can substitute for treatment. Placebo analgesia is generally modest, highly variable between individuals and settings, and best characterized in acute experimental and postoperative pain. Nobody has shown it is an adequate alternative to analgesia in severe pain, and the endogenous system's magnitude ceiling (§20.8) is precisely the reason to doubt it could be.

It also does not establish that all placebo effects run through opioids. The conditioning experiments in Figure 20.2 showed a placebo analgesia that naloxone did not touch. Placebo responses in Parkinson's disease involve dopamine. Nocebo hyperalgesia — the mirror phenomenon, where negative expectation increases pain — has been linked to a different signaling system entirely. "The placebo effect" is not one thing. It is a family of expectancy- and conditioning-driven physiological responses that recruit whichever system is relevant to the outcome being measured. That plurality is a finding, not a hedge.

📊 Evidence Rating

Claim: Endogenous opioid peptides mediate a substantial component of placebo analgesia in adult humans with acute or postoperative pain. Rating: ✅ Strong clinical evidence (as of 2026) Why: The effect is abolished or substantially reduced by an opioid receptor antagonist across multiple independent laboratories and paradigms, has a mechanistically coherent route through the descending inhibitory system, and has been corroborated by receptor imaging. What would change it: Well-powered replications finding no attenuation of expectation-induced placebo analgesia under adequate opioid blockade, or a demonstration that the naloxone effect is fully explained by naloxone's own action on pain sensitivity rather than by blockade of placebo-released opioids. Note the scope: this rating covers acute and postoperative pain in adults. Extending it to chronic pain, to non-pain placebo effects, or to any claim that placebo substitutes for treatment requires separate evidence and would receive a separate rating.

That last note demonstrates the rating rule you will use most often. One molecule, many ratings — and one system, many ratings. The endogenous opioid system has a ✅ for mediating placebo analgesia in this population and endpoint, a ⚠️ for the runner's high, and, as the next section shows, an ❌ for a set of products claiming to enhance it.

20.8 Why endorphin supplements cannot work

Search for "endorphin supplement" and you will find capsules, powders, tinctures, and sprays claiming to raise your endorphin levels, boost your natural painkillers, or support endogenous opioid function. Some contain DL-phenylalanine. Some contain proprietary blends. Some contain "endorphin" on the label without further explanation.

Chapter 13 built a three-joint test for any "boost your own hormone" claim. It applies here almost unmodified, and it is worth running slowly because the shape of the argument transfers to a dozen other claims in Part III.

Joint 1 — Magnitude. Ask what the endogenous system produces at full tilt, and compare it to the effect being promised. The clinical analgesia produced by an opioid drug is a supraphysiological receptor occupancy — a level of mu receptor activation, sustained across the whole body, that no physiological release pattern generates. Endogenous release is local, brief, and modest by comparison. Even the dramatic examples — stimulation-produced analgesia, stress-induced analgesia — operate within a range the body can produce and stop, and a supplement that genuinely enhanced endogenous release could not exceed that ceiling, because the ceiling is set by what the cells can make and release.

Joint 2 — Duration. Endogenous opioid release is phasic and tightly terminated. Enkephalins are degraded within seconds by ectopeptidases — principally neprilysin and aminopeptidase N — sitting in the synaptic space for exactly that purpose. The system is built to produce brief, local signals and shut them off. This is Chapter 3's pulsatile-versus-sustained argument again: even if you increased release, you would be increasing the amplitude of a signal that is designed to be terminated immediately, not creating a sustained analgesic state.

Joint 3 — The existence proof. This is the one that closes the argument, and it is embarrassingly simple. Every human being already has a fully functional endogenous opioid system, and it does not produce pharmacological analgesia. If having endorphins were sufficient for the effects these products promise, nobody would ever have needed morphine. The system is not idle, waiting to be switched on. It is running continuously, and its output is the baseline that you experience as ordinary. Any product claiming to unlock it has to explain why a system that has been optimizing this for several hundred million years is leaving so much performance on the table, and why the specific thing that unlocks it is available without a prescription.

And then, the delivery argument, which is fatal on its own. Chapter 4 laid out the five barriers a swallowed peptide meets: gastric acid and pepsin, pancreatic proteases, brush-border peptidases in the intestinal lining, an epithelial wall that a molecule of that size and polarity does not readily cross, and hepatic first-pass metabolism. A capsule of β-endorphin, or of any opioid peptide, is protein in a pill. It is food.

And β-endorphin has an additional barrier that the GLP-1 analogs of Part II do not. Even if it survived the gut and reached the bloodstream intact, it would then have to cross the blood-brain barrier to reach the sites that produce analgesia and mood change. Chapter 22 is devoted to why peptides of that size do not do that. Two impassable barriers in series is not a challenging delivery problem. It is a physical impossibility argument.

⚠️ Hype Check — "boost your natural endorphins"

The claim, in its usual form:

"Instead of masking pain with drugs, support your body's own painkillers. Our formula provides the precursors and cofactors your body needs to produce endorphins naturally — no side effects, no tolerance, no dependence."

What's true in it. The endogenous opioid system is real, it genuinely modulates pain, and pharmacologically enhancing it is a legitimate research program (§20.9). The complaint about drug side effects is fair. The instinct — work with the system rather than overwhelming it — is the same instinct behind serious drug development in this area.

Where it fails. Four places, and you can now check each one.

Magnitude. The therapeutic effect being implied is supraphysiological. The endogenous system, by definition, is not.

Precursor logic. Endorphins are not built from free amino acids you swallow. β-endorphin is cut out of POMC, a protein made by ribosomes from a gene transcript. Enkephalins come from proenkephalin the same way. In an adequately nourished person, peptide hormone output is limited by transcription, processing, and release signals — not by the availability of phenylalanine. "Provides the building blocks" is a category error about how peptide hormones are made.

Delivery. Any intact peptide in the capsule is digested. Anything that survives faces the blood-brain barrier.

"No tolerance, no dependence." This is the most interesting failure, because it is the one that reveals the seller has not thought about mechanism at all. A receptor does not know where its ligand came from. If a product genuinely produced sustained elevation of mu receptor activation, the adaptations of §20.6 would follow, because those adaptations are responses to receptor occupancy, not to a drug's provenance. The claim that a compound both meaningfully activates this system and is immune to its adaptive responses is internally inconsistent.

Verdict: the argument fails on magnitude, on how peptides are synthesized, on delivery, and on its own internal logic. And the existence proof sits underneath all four: everybody has this system, and nobody gets pharmacological analgesia from having it.

The specific case of DL-phenylalanine. The most common named ingredient in this category is DL-phenylalanine, usually marketed on the theory that it inhibits "enkephalinase" and thereby prolongs the life of endogenous enkephalins.

Give the theory its due: the target is real. Enkephalins are degraded by neprilysin and aminopeptidase N, and inhibiting those enzymes to prolong endogenous enkephalin signaling is a genuine, continuing pharmaceutical research program (§20.9). This is not an invented mechanism.

But mechanism is not evidence, and the rating rule is that we never upgrade on mechanism. The human clinical evidence for DL-phenylalanine as an analgesic consists of a small number of small, mostly old studies, several without adequate controls, with inconsistent results, and no modern adequately powered randomized trial. Rate it on its own evidence, which is thin — which is a statement about what has been shown, not about what is possible.

📊 Evidence Rating

Claim: Orally administered "endorphin" or "endorphin-boosting" supplements — including DL-phenylalanine marketed as an enkephalinase inhibitor — produce clinically meaningful analgesia or mood benefit in adults. Rating: ❌ Hype outpaces evidence (as of 2026) Why: Intact opioid peptides taken orally are digested and could not cross the blood-brain barrier if they survived; the magnitude and duration of endogenous release cannot produce the promised effect; every human already has this system and does not get pharmacological analgesia from it; and for the one ingredient with a real mechanistic rationale, DL-phenylalanine, the human trial evidence is sparse, dated, and inconsistent. What would change it: For the enkephalinase-inhibitor class specifically, adequately powered randomized placebo-controlled trials in a defined pain population showing a clinically meaningful effect on a pre-specified endpoint, with pharmacokinetic evidence that the compound reaches and inhibits the target enzyme at achievable exposures. That is a realistic study to run, and it is what the pharmaceutical programs in §20.9 are attempting. For oral intact β-endorphin, nothing short of a demonstrated delivery technology would move this rating, and the ❌ describes the evidence, not the molecule.

20.9 Peptide-based analgesics and the difficulty of drugging this system

Everything in this chapter converges on a practical question. We have a natural analgesic system governed by peptides. We have receptors that are well characterized and druggable. We have known endogenous ligands with known sequences. Why is it so hard to build a good drug here?

Four obstacles, in ascending order of stubbornness.

The peptides themselves are unusable as drugs. Enkephalins are degraded within seconds. β-endorphin lasts longer but is still cleared quickly and still cannot enter the brain from the bloodstream. Administering the natural ligand is not an option, which is why the field has always worked with analogs, mimetics, or entirely different chemistry.

The blood-brain barrier. Central analgesia requires reaching the central nervous system. For a peptide, that is Chapter 22's problem in its most acute form, and it drives the field toward one of two strategies: build something small and lipophilic that crosses (at which point you are no longer doing peptide chemistry), or deliver the peptide past the barrier physically.

Mu selectivity does not solve the selectivity problem. A perfectly mu-selective agonist still produces respiratory depression, because mu receptors mediate both analgesia and respiratory depression. Separating them has been the field's central ambition for decades. Chapter 2's Case Study 2 covered the biased-agonism approach at this receptor in detail, and its outcome is the reason this section is written with such restraint.

The alternatives that avoid mu bring their own problems. Kappa agonists avoid respiratory depression and abuse liability, and produce dysphoria instead (§20.3). Delta agonists have struggled with convulsant liability.

Against that, here is a genuine success, and its design is a lesson in itself.

Ziconotide is a synthetic equivalent of a peptide from the venom of a marine cone snail. It is twenty-five amino acids long, held rigid by three disulfide bonds, and it is not an opioid. It does not touch mu, delta, or kappa. It blocks a specific voltage-gated calcium channel — the N-type channel — on presynaptic terminals in the spinal dorsal horn, preventing the release of the neurotransmitters that carry the pain signal across its first synapse.

Look back at §20.3 and notice what that means. When a mu receptor is activated, one of its principal downstream actions is inhibition of exactly this class of calcium channel. Ziconotide does directly, at the channel, what an opioid does indirectly, through a receptor. It arrives at the same endpoint by a different road — which is why it produces analgesia without opioid tolerance, without opioid dependence, and without respiratory depression.

It is approved for severe chronic pain in patients for whom other therapies are inadequate or intolerable. And it must be administered intrathecally — delivered directly into the cerebrospinal fluid, by an implanted pump — because it is a 25-residue peptide and cannot cross the blood-brain barrier.

That administration requirement is the most eloquent fact in this chapter. This is a real, approved, effective peptide analgesic, and the only way to get it where it works is to place it there surgically. The route of administration is itself the evidence for how absolute the barrier is. Ziconotide is not exotically delivered because the developers lacked imagination. It is delivered that way because there is no other way, and a company with an approved drug and every commercial incentive to find one has not found one.

Ziconotide also carries real risks — a narrow therapeutic window, and a boxed warning for severe psychiatric symptoms and neurological impairment — which is why it is reserved for a specific population under specialist management. It is a demonstration of possibility, not a template.

📊 Evidence Rating

Claim: Ziconotide, administered intrathecally, produces clinically meaningful analgesia in adults with severe chronic pain for whom other therapies, including intrathecal morphine, are inadequate or intolerable. Rating: ✅ Strong clinical evidence (as of 2026) Why: Randomized placebo-controlled trials supported regulatory approval for this population, the mechanism is established, and the safety profile — including its serious neuropsychiatric risks and narrow therapeutic window — is characterized well enough to define who should and should not receive it. What would change it: Not much for this indication; it is approved with trial support. Note the scope carefully. This rating says nothing about ziconotide by any other route, in any other population, or for any other indication — and its intrathecal-only requirement means no oral, transdermal, or subcutaneous claim about it could ever inherit this rating. If you encounter a product invoking "cone snail peptide" technology in any non-intrathecal form, this ✅ does not transfer to it, and the burden is entirely on the seller.

Two more approaches are worth knowing, because they show the field's cleverest current moves.

Using the barrier as a design feature rather than fighting it. If a peptide cannot cross into the brain, that limitation can be converted into selectivity. Difelikefalin is a small synthetic peptide kappa agonist built from D-amino acids and deliberately designed not to cross the blood-brain barrier, so it engages peripheral kappa receptors without producing the central dysphoria that has blocked every previous kappa program. It is approved — for pruritus, the severe itch associated with chronic kidney disease in patients on hemodialysis, not for pain — and it is the clearest existing proof that peripheral restriction is a usable strategy. The same logic runs in reverse in the peripherally restricted mu antagonists used for opioid-induced constipation: block the gut receptors, leave the brain receptors alone.

Enhancing the endogenous system pharmacologically — done properly. This is the legitimate version of §20.8's failed claim. Rather than supplying opioid peptides, inhibit the enzymes that destroy them, so that enkephalins released naturally at active synapses survive longer. Because the enhancement occurs only where and when the body is already releasing, the approach would in principle preserve the system's spatial and temporal specificity — the very properties a systemic agonist destroys. Dual inhibitors of neprilysin and aminopeptidase N have been pursued for years, and a human peptide called opiorphin that inhibits these enzymes has been characterized. This work is 🔬 Frontier: the mechanism is sound, and no approved analgesic has yet emerged from it. It is the right place to watch, and the right place to be skeptical of anyone claiming the problem is already solved in capsule form.

The overall picture is sobering and worth stating without decoration. This is one of the best-understood signaling systems in human biology, and we have not been able to turn that understanding into a safe, orally available, non-addictive analgesic. Mechanistic knowledge, even excellent mechanistic knowledge, does not convert automatically into a drug. If you needed a single argument for why this book rates claims on evidence rather than on plausibility, this system is it.

📋 Your Evidence Dossier

This chapter fills Field 3 — but for a system rather than a molecule.

Chapter 2 introduced Field 3 as Mechanism: what the peptide binds, what happens downstream, and what the effect is. That works when there is one ligand and one receptor. It falls apart here.

Consider trying to write a Field 3 entry for "endorphins." Which peptide — β-endorphin, the two enkephalins, four or five dynorphins? Which receptor — three of them, with different preferences and partly overlapping ligands? What effect — analgesia at all three, euphoria at one and dysphoria at another? Any single answer you write will be wrong, and the wrongness will be invisible because the entry will look complete.

When a peptide belongs to a family with multiple ligands, multiple receptors, and internally opposing effects, the unit of analysis is the system, not the molecule. Here is the version of Field 3 that handles that case.

FIELD 2 — MECHANISM (system version)

  Unit of analysis      state explicitly that this entry covers a SYSTEM
  Ligand roster         which peptides, from which precursor genes
  Receptor roster       which receptors, what G protein, inhibitory or excitatory
  Ligand × receptor map who prefers what — including overlaps
  Opposing effects      where two arms of the system pull AGAINST each other
  Compartments          central / spinal / peripheral — and can they communicate?
  Regulation            release pattern, and how the signal is terminated
  How to interrogate it agonist? antagonist? measurement? imaging? which exist?
  Measurement trap      is the accessible measurement in the compartment that matters?

Worked demonstration — the endogenous opioid system

FIELD 2 — THE ENDOGENOUS OPIOID SYSTEM            [worked demonstration]

  Unit of analysis   A system. "Endorphins" is not a molecule and should not be
                     entered as one.

  Ligand roster      β-endorphin (31 aa) ................ from POMC
                     met- and leu-enkephalin (5 aa) ..... from proenkephalin
                     dynorphins A, B, neoendorphins ..... from prodynorphin
                     [related, kept separate: nociceptin/OFQ, from PNOC]

  Receptor roster    mu (OPRM1), delta (OPRD1), kappa (OPRK1)
                     all GPCRs, all inhibitory G proteins, all reduce neuronal
                     excitability and neurotransmitter release
                     [related: NOP (OPRL1), naloxone-insensitive]

  Ligand × receptor  β-endorphin → mu, delta
                     enkephalins → delta (preferred), mu
                     dynorphins  → kappa
                     NOT a clean one-to-one map. Overlap is the norm.

  Opposing effects   *** THE CRITICAL FIELD FOR THIS SYSTEM ***
                     mu activation    → euphoria, reward, reinforcement
                     kappa activation → DYSPHORIA, aversion
                     Two arms of one system, opposite hedonic sign, both
                     analgesic. Any sentence beginning "endorphins make you
                     feel good" is false about a third of the system.

  Compartments       central (brain), spinal (dorsal horn), peripheral (gut,
                     immune cells, sensory terminals), plus pituitary
                     β-endorphin secreted into blood.
                     *** These compartments do NOT freely communicate. ***
                     Blood-brain barrier (Ch 22) separates the accessible one
                     from the interesting one.

  Regulation         phasic, local, terminated in seconds by ectopeptidases
                     (neprilysin, aminopeptidase N). Nothing endogenous
                     resembles sustained systemic occupancy (Ch 3).

  Interrogation      Agonists: yes, many, clinically approved.
                     Antagonists: yes — naloxone, naltrexone. THIS IS THE KEY
                       TOOL. Because peptide levels cannot be measured where
                       they act, blockade is the workhorse causal test.
                     Imaging: yes, PET with opioid receptor ligands. Powerful,
                       small-n, expensive, correlational.
                     Direct measurement: plasma β-endorphin — available, cheap,
                       and largely uninformative about central activity.

  Measurement trap   *** The compartment error. *** Plasma β-endorphin is the
                     easiest measurement and the least meaningful one for any
                     central claim. Forty years of "endorphin" claims rest on
                     it (§20.5). Whenever you meet a peptide claim built on a
                     blood level, ask where the peptide is supposed to act.

Ratings on a system

Because the unit is a system, the ratings multiply. A single entry now carries several, and they do not average:

• Mediates a substantial component of placebo analgesia (acute/postoperative pain, adults) — ✅
• Causes the runner's high — ⚠️
• Can be meaningfully enhanced by oral supplements — ❌
• Can be enhanced by pharmacological enkephalinase inhibition to produce approved analgesia — 🔬

One system, four ratings, no contradiction. If you ever find yourself wanting to give a system a single overall rating, that is the moment to notice you have compressed away the information you needed.

Your turn

Convert one system-level entry in your own dossier to this format. Good candidates: the incretin system if you track GLP-1 drugs (ligands GLP-1, GIP, and glucagon; several receptors; effects that oppose each other in places); the melanocortin system from Chapter 13 (one precursor, several products, agonists and an endogenous antagonist); or the oxytocin/vasopressin pair, which Chapter 21 takes up next.

For whichever you choose, fill the Opposing effects and Measurement trap rows first. Those two rows are where system-level entries earn their keep, and they are where almost every confident public claim about a peptide family goes wrong.

Conclusion

The receptors came first. Everything else in this chapter follows from that.

Because the receptors are endogenous, morphine's effects are not an invasion but an imitation, and the side effects are not impurities but the receptor doing its other jobs. Because three precursor genes produce three peptide families with a shared N-terminal message and divergent addresses, the system has internal structure — including a kappa arm whose hedonic sign is opposite to the mu arm's, which makes "endorphins make you feel good" a sentence that is false about part of its own subject. Because the system's principal action is descending inhibition onto the first synapse of the pain pathway, and because that descending system takes input from regions that handle context and expectation, pain is regulated rather than merely reported — and a belief about relief can become a measurable neurochemical event.

That last point is the chapter's center of gravity. Placebo analgesia is not a courtesy that patients extend to investigators; it is, at least in substantial part, endogenous opioid pharmacology, and we know that because an opioid receptor antagonist takes it away. The limits are real too: it does not mean placebo substitutes for treatment, and it does not mean every placebo effect is opioid-mediated, because several demonstrably are not.

Set that against the runner's high, where the popular claim is far more confident and the evidence far weaker. The difference is not the plausibility of the mechanism — both are plausible. The difference is the study design. One rests on blocking a receptor and observing a change; the other rests largely on measuring a peptide in blood that cannot reach the place the effect is supposed to happen. When a peptide acts somewhere you cannot sample, the antagonist experiment is often the only honest test, and its absence is the most informative thing about a claim.

And the practical takeaways sit at the two ends. Nothing you can swallow will meaningfully boost your endorphins, for four independent reasons any one of which would be sufficient. Meanwhile, the one unambiguously effective peptide analgesic on the market has to be pumped directly into cerebrospinal fluid, because a 25-residue peptide has no other way in.

Chapter 21 turns to oxytocin and vasopressin, where the science is subtler, the public claims are even larger, and the central discipline is the same one you just practiced: separating what a peptide does from what people need it to mean.

Key Terms

Endogenous opioid peptide — any of the peptides produced by the body that act at opioid receptors, comprising the endorphin, enkephalin, and dynorphin families.

β-endorphin — a 31-residue peptide derived from proopiomelanocortin, with high affinity for mu and delta receptors. The molecule most people mean, incorrectly, when they say "endorphins."

Enkephalin — either of two pentapeptides, met-enkephalin and leu-enkephalin, derived from proenkephalin. Prefer the delta receptor; degraded within seconds.

Dynorphin — a family of peptides derived from prodynorphin, acting principally at the kappa receptor and associated with dysphoria and stress responses.

Proopiomelanocortin (POMC) — the precursor protein that yields β-endorphin, ACTH, α-MSH, and several other products, processed differently in different cell types.

Proenkephalin — the precursor protein that yields met- and leu-enkephalin.

Prodynorphin — the precursor protein that yields the dynorphins.

Mu receptor (μ, OPRM1) — the opioid receptor responsible for the analgesia, euphoria, respiratory depression, constipation, tolerance, and dependence associated with clinical opioids.

Delta receptor (δ, OPRD1) — an opioid receptor mediating more modest analgesia and mood effects; clinically undeveloped, partly because of convulsant liability.

Kappa receptor (κ, OPRK1) — an opioid receptor mediating analgesia along with dysphoria and aversion rather than euphoria; low abuse liability, poor tolerability.

Nociceptin / orphanin FQ — a 17-residue peptide structurally related to the opioid peptides but beginning with phenylalanine rather than tyrosine; acts at the NOP receptor and is not blocked by naloxone.

Naloxone — a short-acting opioid receptor antagonist. Used clinically to reverse overdose and experimentally as the standard probe for whether an effect is opioid-mediated.

Naltrexone — a longer-acting opioid receptor antagonist used clinically and in research.

Antagonist — a molecule that occupies a receptor without activating it, preventing agonists from acting there. The antagonist experiment is the workhorse causal test in this chapter.

Descending inhibition — modulation of incoming pain signals by circuitry projecting from brain to spinal cord, running from the periaqueductal gray through the rostral ventromedial medulla to the dorsal horn.

Periaqueductal gray (PAG) — a midbrain region central to descending pain modulation, receiving input from regions handling context, emotion, and expectation.

Rostral ventromedial medulla (RVM) — a brainstem relay in the descending pathway, containing cell populations that both inhibit and facilitate pain transmission.

Dorsal horn — the region of the spinal cord where incoming nociceptive fibers make their first synapse, and where descending inhibition is principally applied.

Nociception — the neural detection and transmission of potentially damaging stimuli. Distinct from pain, which is the conscious experience.

Analgesia — reduction of pain. Antinociception is the corresponding term when the measurement is a reflex or threshold rather than a report.

Dysphoria — an unpleasant, distressed mood state. Characteristically produced by kappa receptor activation.

Tolerance — diminished effect from a constant exposure, arising from receptor desensitization and downregulation. Develops at different rates for different effects of the same drug.

Physical dependence — a state in which abrupt cessation produces a withdrawal syndrome. Occurs in anyone on sustained therapy, including patients using medication appropriately. Not addiction.

Addiction (opioid use disorder) — a clinical diagnosis defined by compulsive use, loss of control, and continued use despite harm. Distinct from physical dependence, and a treatable medical condition.

Opioid-induced hyperalgesia — increased pain sensitivity arising from sustained opioid exposure, associated with descending facilitation.

Placebo analgesia — pain relief produced by an inert intervention plus the expectation of benefit; substantially reduced by opioid receptor antagonism.

Nocebo — the mirror phenomenon, in which negative expectation worsens symptoms. In pain, linked to signaling systems other than opioid.

Neprilysin — an ectopeptidase that degrades enkephalins, among other peptides. The principal "enkephalinase" target.

Ziconotide — a synthetic 25-residue peptide equivalent to a cone snail venom peptide; an N-type calcium channel blocker, not an opioid; approved for severe chronic pain and administered intrathecally.

Difelikefalin — a peripherally restricted synthetic peptide kappa agonist, designed not to cross the blood-brain barrier; approved for pruritus in hemodialysis patients.

Intrathecal — administered directly into cerebrospinal fluid, bypassing the blood-brain barrier.

Spaced Review

1. (Ch 2 + Ch 20) All three classical opioid receptors couple to inhibitory G proteins and reduce neuronal excitability. Yet mu activation produces euphoria and kappa activation produces dysphoria. Explain how the same cellular mechanism produces opposite subjective effects, and state what that implies about predicting a peptide's effect from its signaling pathway alone.

2. (Ch 3 + Ch 20) Enkephalins are released phasically and degraded within seconds; a clinical opioid produces sustained, whole-body receptor occupancy. Using Chapter 3's framework, explain why tolerance and dependence are consequences of the second pattern and not of the first — and why "it's the same receptor being activated either way" is not a rebuttal.

3. (Ch 5 + Ch 20) The endogenous opioid system in this chapter received a ✅, a ⚠️, an ❌, and a 🔬. State the claim attached to each, and explain why issuing a single overall rating for "the endogenous opioid system" would destroy information a reader needs.

4. (Ch 20 + Ch 22 preview) Two claims: (a) intense exercise raises plasma β-endorphin; (b) intense exercise raises brain β-endorphin activity. The first is well established. Explain precisely why the first does not establish the second, and name the experimental design that could.

5. (Ch 2 + Ch 5 + Ch 20) A product is marketed as a "natural enkephalinase inhibitor" for pain relief, citing published research on neprilysin inhibition. Write two sentences: one granting exactly what the mechanistic literature supports, and one stating what would still have to be shown before the product's claim could be rated above ❌. Then say which rating rule you just applied.