The lock with no key — competitive µ-opioid blockade
IUPAC: 17-(cyclopropylmethyl)-4,5α-epoxy-3,14-dihydroxymorphinan-6-one · C20H23NO4 · MW 341.4 g/mol (freebase); 377.9 (·HCl) · CAS 16590-41-3 · ATC N07BB04 · ChEMBL CHEMBL19019
Naltrexone (ReVia, Depade oral; Vivitrol depot IM; low-dose "LDN"). A morphinan antagonist — the pharmacological inverse of every other opioid on this site. Structurally it is naloxone's long-acting oral cousin: the N-allyl of naloxone swapped for an N-cyclopropylmethyl, which buys oral bioavailability and a real duration of action. It slots into the µ-opioid pocket, contacts the same residues an agonist would, and then does nothing — no G-protein coupling, no reward, just occupancy. Used for opioid use disorder and alcohol use disorder, because it also blocks the endogenous opioids alcohol recruits. Not a rescue drug — that job belongs to naloxone.
Naltrexone is a competitive, reversible, surmountable antagonist at all three classical opioid receptors, with highest affinity at the µ-opioid receptor (MOR). It binds the same orthosteric pocket an agonist like morphine or fentanyl would occupy — deep in the 7-transmembrane bundle, salt-bridging its protonated amine to Asp1473.32 — but the morphinan scaffold, with its bulky N-cyclopropylmethyl and 6-keto/14-hydroxyl substitution, fails to trigger the TM6 outward swing that opens the G-protein cavity. The result is pure occupancy: no Gαi/o coupling, no cAMP suppression, no GIRK activation, no reward. In receptor-reserve terms, it is a spanner in the door.
Because the block is competitive, it can in principle be overcome by a large enough agonist dose — but naltrexone's sub-nanomolar MOR affinity (Ki ≈ 0.2 nM, human recombinant, ChEMBL) means the concentrations required are enormous and dangerous. At clinical exposure it functionally silences MOR signaling for hours (oral) to weeks (depot). Reported constitutive activity at MOR makes some of naltrexone's effect look inverse-agonist-like in dependent tissue, which is part of why it can precipitate withdrawal rather than merely fail to relieve it.
Primary target. Ki ≈ 0.2 nM (human recombinant, [³H]DAMGO displacement). Occludes the orthosteric site, preventing β-endorphin, enkephalin, morphine, heroin, and fentanyl from coupling Gi/o. This is the anti-reward and the withdrawal-precipitating action.
Ki ≈ 0.4 nM ([³H]U-69,593 displacement). Naltrexone blocks dynorphin/KOR signaling almost as tightly as MOR. KOR antagonism is dysphoria-relieving and is one hypothesized contributor to the anti-craving effect in alcohol use disorder.
Ki ≈ 10.8 nM — real but ~50× weaker than MOR. Enkephalin/DOR block rounds out the "pan-antagonist" profile. Naltrexone is not selective; it lowers the ceiling on the entire endogenous opioid system.
Alcohol triggers β-endorphin release into the ventral tegmental area / nucleus accumbens. That endogenous opioid tone is a major driver of alcohol's reward and priming. MOR blockade severs this loop: drinking becomes less reinforcing, reducing heavy-drinking days.
The dominant circulating species after oral dosing is 6β-naltrexol — itself a peripherally-weighted MOR antagonist with a longer half-life (~13 h). It extends and smooths the blockade well past the parent's clearance.
At micromolar exposure — and in stereochemistry-independent fashion — naltrexone antagonizes TLR4 on microglia, dampening glial pro-inflammatory signaling. This non-opioid action underpins the low-dose naltrexone (LDN) hypothesis (§05). Evidence is early.
Orally, naltrexone is nearly completely absorbed but hammered by first-pass metabolism — systemic bioavailability of the parent is low and highly variable (~5–40%). The saving grace is its major metabolite: hepatic dihydrodiol dehydrogenase (DD / AKR1C) reduces the 6-keto group to 6β-naltrexol, an active antagonist that circulates at far higher concentrations than the parent and carries a longer half-life. Metabolism is largely CYP-independent, so classic CYP drug interactions are minimal; both parent and metabolite are then glucuronidated and renally cleared.
The extended-release IM depot (Vivitrol, 380 mg) sidesteps first-pass entirely: a PLGA microsphere suspension releases naltrexone over roughly a month, giving a durable, adherence-independent blockade. This is the formulation's whole point — a person cannot simply skip a dose to "get around" the block. But that same durability is the source of its most dangerous failure mode: as the depot wanes near the end of the interval, blockade fades while opioid tolerance stays low (§04).
Naltrexone is FDA-approved (since 1984 for opioid dependence; 1994 for alcohol) for two indications that look unrelated until you trace them both back to the µ-opioid receptor.
With MOR occupied at sub-nanomolar affinity, exogenous opioids simply do not work: no high, no analgesia, no respiratory depression. There is nothing to reinforce. Unlike methadone or buprenorphine (agonist/partial-agonist maintenance), naltrexone is a pure antagonist maintenance — it does not treat withdrawal or craving directly, it removes the payoff. The depot's month-long blockade makes it viable precisely because oral adherence in OUD is notoriously poor.
Alcohol's reward is partly opioid-mediated: ethanol drives β-endorphin release onto MOR in the mesolimbic system, disinhibiting VTA dopamine neurons. Block MOR and you blunt that surge — drinking feels less rewarding and the priming effect of the first drink weakens. Clinically this shows up as fewer heavy-drinking days and reduced relapse to heavy drinking (COMBINE trial). Naltrexone does not cause an aversive reaction like disulfiram; it quietly lowers the reinforcement.
An evidence-based AUD protocol: take naltrexone ~1 h before drinking, so every drinking episode occurs under MOR blockade. Over weeks the opioid-reinforced drinking behavior undergoes pharmacological extinction, and consumption falls. It is a genuine, mechanism-honest use of the antagonist — reinforcement learning run in reverse.
Naltrexone is paired with bupropion as Contrave (weight management: MOR block disinhibits POMC neurons in the hypothalamus) and formulated with morphine (Embeda) or oxycodone (Troxyca) as an abuse-deterrent sequestered core that only releases antagonist if the pill is crushed. It is not an analgesic and not an emergency reversal agent.
Naltrexone has two failure modes that are not side effects — they are direct, predictable consequences of the pharmacology, and both have killed people. Any honest account has to lead with them.
You must be opioid-free before the first dose. If any opioid is still occupying MOR, a sub-nanomolar antagonist will displace it off the receptor in minutes, converting a slow natural taper into an abrupt, severe, precipitated withdrawal — vomiting, diarrhea, agitation, autonomic storm — far worse than spontaneous withdrawal and not reversible by re-dosing the opioid (it's outcompeted). Standard practice: 7–10 opioid-free days for long-acting agonists (methadone), 3–7 days for short-acting (heroin, oxycodone), often confirmed with a naloxone challenge or a documented negative window before induction.
Sustained MOR blockade means opioid tolerance falls during treatment — receptors are not being stimulated, so the adaptations that let a tolerant user survive high doses regress. When the blockade ends — a missed oral dose, or the tail of a depot cycle as the depot wanes — the person is functionally opioid-naïve but often returns to their old dose. That dose is now potentially fatal. This is one of the best-documented overdose-death mechanisms in antagonist-maintenance treatment, and it is why depot naltrexone requires a relapse/overdose plan and take-home naloxone.
Low-dose naltrexone (typically 1.5–4.5 mg, versus 50 mg for AUD/OUD) is a genuinely different pharmacological regime, and it has attracted a large, enthusiastic, and evidence-thin following for chronic pain and autoimmune conditions. Two non-mutually-exclusive mechanisms are proposed:
A brief, low-dose MOR blockade may trigger compensatory upregulation of endogenous opioid tone (β-endorphin, enkephalins) and receptor density. As the short blockade lifts, the rebound endorphin signaling is proposed to be analgesic and immunomodulatory. Mechanistically plausible; direct human evidence is limited.
Naltrexone (and its (+)-enantiomer, which has no opioid activity) antagonizes TLR4 on microglia, reducing release of pro-inflammatory cytokines and reactive glial signaling implicated in central pain sensitization. This is a real, opioid-independent action observed in preclinical models.
LDN is used off-label for fibromyalgia, Crohn's disease, multiple sclerosis, complex regional pain syndrome and more. The supporting data are mostly small, short, single-centre or pilot trials (e.g. small crossover fibromyalgia studies from Stanford; small open-label Crohn's series) plus a large body of patient-reported benefit. Signals are promising and the safety/cost profile is excellent, but there are few adequately powered, placebo-controlled RCTs, and systematic reviews consistently conclude the evidence is insufficient to establish efficacy. LDN is best framed as a low-risk, plausibly-active intervention that is still experimental — not a proven therapy. Anyone selling it as a miracle cure is ahead of the data.
The two are structural twins and both are µ-opioid antagonists, but they are not interchangeable, and confusing them in an overdose can cost a life.
N-allyl morphinan. Almost no oral bioavailability, so it's given IV/IM/intranasal. Onset in 1–3 minutes, duration only 30–90 minutes — deliberately short, so it's a rescue tool that snatches an agonist off MOR to restore breathing right now. Because it's short-acting, a person revived from a long-acting opioid can re-sedate and needs monitoring/re-dosing.
N-cyclopropylmethyl morphinan. Orally active, long-acting (parent ~4 h, 6β-naltrexol ~13 h, depot ~28 days). It is for preventing relapse over weeks, not for reversing an overdose in progress — its onset and route make it useless as an acute antidote. If someone is overdosing: give naloxone and call emergency services. Naltrexone is the drug you were on before; naloxone is the drug that brings you back.
The morphinan cage is rigid. Naltrexone's 4,5-epoxymorphinan core is a fused pentacyclic
scaffold with essentially one low-energy conformation; only the N-cyclopropylmethyl and the hydroxyls carry
any rotational freedom. In FlexAID∆S terms this means a very small ligand ΔS_conf penalty on
binding — the molecule is pre-organized, arriving at MOR already shaped like the bound state. Almost all of
the −RT·ln(Ka) budget is available for enthalpic contacts (the Asp1473.32 salt bridge,
the phenolic hydroxyl H-bond, aromatic stacking) rather than being spent paying an entropy tax.
The interesting part is the receptor's entropy. An agonist collapses the MOR conformational
ensemble and pays it forward — the closed→open TM6 transition opens a new, ordered G-protein interface,
propagating the entropy collapse into the transducer. An antagonist collapses the ensemble and stops there:
naltrexone reduces H_pocket by locking the orthosteric site, but it stabilizes the inactive
TM6 arrangement, so no downstream ordering (no G-protein cavity, no coupled ensemble) ever forms. The entropy
collapse is real and local; it just never gets transduced into signal. That is the thermodynamic signature of
antagonism — high-affinity ordering with zero efficacy.
Rough thermodynamics: Ki ≈ 0.2 nM at MOR corresponds to ΔG ≈ −RT·ln(Ka) ≈ −13.2 kcal/mol at 298 K — a very tight binder, consistent with the FlexAID∆S expectation for a rigid, pre-organized morphinan making an ionic anchor plus aromatic contacts. The near-equal MOR (0.2 nM) and KOR (0.4 nM) affinities, with δ ~50× weaker (10.8 nM), reflect how conserved the orthosteric anchoring is across the µ/κ pair versus the more divergent δ pocket — a selectivity pattern that falls out of the differential rigidity and electrostatics of the three binding sites, not of the ligand.
| Target | Affinity | Rel. | Action |
|---|---|---|---|
|
MOR (µ)
µ-opioid receptor · OPRM1
|
Ki = 0.2 nM
Primary target
|
Antagonist | |
|
KOR (κ)
κ-opioid receptor · OPRK1
|
Ki = 0.4 nM
Near-equal to MOR
|
Antagonist | |
|
DOR (δ)
δ-opioid receptor · OPRD1
|
Ki = 10.8 nM
~50× weaker
|
Antagonist | |
|
TLR4
Toll-like receptor 4 (glial)
|
µM range
Non-opioid · LDN
|
Antagonist |