IUPAC: 7-methoxy-1-methyl-9H-pyrido[3,4-b]indole · C₁₃H₁₂N₂O · MW 212.25 g/mol · CAS 442-51-3
Harmine. The lead β-carboline (harmala) alkaloid of Banisteriopsis caapi — the ayahuasca vine — and of Syrian rue (Peganum harmala). A reversible, competitive MAO-A inhibitor (RIMA): it is the enabling half of ayahuasca, blocking the enzyme that would otherwise destroy orally-dosed DMT in the gut wall and liver. Separately, one of the most-cited chemical-biology probes for the kinase DYRK1A. Also known as: banisterine, telepathine, yageine, leucoharmine.
Harmine is not a psychedelic in its own right — it is an enzyme inhibitor. Its defining action is reversible, competitive inhibition of monoamine oxidase A (MAO-A), the flavoenzyme that oxidatively deaminates serotonin, noradrenaline, dopamine and dietary/xenobiotic amines. In the co-crystal structure (PDB 2Z5X), the flat, aromatic β-carboline stacks in the substrate cavity in front of the FAD cofactor, sterically occluding the catalytic site without forming the covalent flavin adduct that defines irreversible MAOIs. Reported human MAO-A affinities cluster around Ki ≈ 5 nM (individual assays span ~1–17 nM; IC50 ≈ 4 nM), with roughly 7,000-fold selectivity for MAO-A over MAO-B — the mirror image of deprenyl.
The planar β-carboline nucleus π-stacks against the re-face of FAD in the ~550 ų MAO-A substrate cavity, blocking access of monoamine substrates. Contact residues include Tyr407/Tyr444 (the "aromatic cage"), Phe208 and Ile180 — the pocket that determines A-vs-B substrate selectivity.
Unlike phenelzine or tranylcypromine (which form permanent covalent adducts to FAD), harmine binds non-covalently and dissociates. MAO-A activity recovers as harmine clears — hours, not the ~2 weeks of enzyme resynthesis irreversible MAOIs demand. This is the pharmacological basis of the "RIMA" class.
Human MAO-B Ki ≈ 121 µM vs MAO-A Ki ≈ 5–17 nM — a ~10³–10⁴-fold preference. Harmine therefore leaves the dopamine/PEA-metabolising MAO-B largely intact while shutting down serotonin/tyramine clearance, shaping both its ayahuasca role and its interaction profile.
Harmine is the prototypical ATP-competitive inhibitor of DYRK1A (IC50 ≈ 30–80 nM), a proline-directed kinase on chromosome 21 implicated in tau phosphorylation, NFAT signalling and neurogenesis. It also hits the related CLK dual-specificity kinases. This is an entirely MAO-independent activity.
By inhibiting intestinal + hepatic MAO-A, harmine spares co-ingested DMT from first-pass oxidative deamination — the single reason oral ayahuasca is orally active. Without the vine's β-carbolines, swallowed DMT is destroyed before it reaches the systemic circulation.
Harmine has weak, non-primary affinity at 5-HT2A/2C, I₂ imidazoline and benzodiazepine-associated sites (all µM range). These contribute mild subjective "buzz," nausea and tremor from caapi alone but are dwarfed by the nanomolar MAO-A and DYRK1A actions.
The therapeutic and cultural logic of ayahuasca is a two-drug pharmacokinetic hack: the vine supplies MAO-A inhibitors (harmine, harmaline, tetrahydroharmine) and a companion plant — Psychotria viridis (chacruna) or Diplopterys cabrerana (chaliponga) — supplies the actual psychedelic, DMT. Neither is orally active alone at brew doses; together, the RIMA opens a metabolic window for DMT to reach the brain and drive 5-HT2A partial agonism (see DMT #003).
Harmine has low and highly variable oral bioavailability owing to extensive first-pass metabolism — paradoxically, it is itself a good CYP substrate even as it inhibits MAO. Peak plasma is reached within roughly 1–2 hours after a brew, and the elimination half-life is short (~1–3 h). Because the MAO-A block is reversible and harmine clears quickly, the enzyme inhibition is essentially co-terminous with the ayahuasca experience rather than a two-week commitment — a key difference from prescription irreversible MAOIs.
Metabolism cascade: harmine is O-demethylated at the 7-position — chiefly by CYP2D6 — to the phenolic metabolite harmol, which is then conjugated (glucuronidation / sulfation) and excreted.
Two pharmacokinetic quirks matter for real-world dosing. First, CYP2D6 is polymorphic: poor metabolisers (~7–10% of Europeans) clear harmine more slowly and reach higher, longer β-carboline exposure from a standard brew — and, since serotonergic drugs like paroxetine and fluoxetine also inhibit CYP2D6, co-use both raises harmine levels and stacks serotonin toxicity risk. Second, harmine's short half-life means the MAO-A block relaxes over the same hours the DMT effect fades, so re-dosing to extend a session compounds both the psychedelic and the pressor/serotonergic hazard rather than simply prolonging it.
The companion alkaloids ride the same window. Harmaline (3,4-dihydroharmine) is also a nanomolar reversible MAO-A inhibitor and adds sedation, nausea and tremor. Tetrahydroharmine (THH) is a much weaker MAO inhibitor but a modest serotonin-reuptake inhibitor in its own right — a second, independent serotonergic pressure point that is easy to forget when reasoning about interactions.
Because harmine works by removing a brake rather than directly driving a receptor, its "psychopharmacology" is best read as a set of downstream systems it unmasks — plus a genuinely separate kinase story that has nothing to do with monoamines.
Shutting down MAO-A raises synaptic and intracellular serotonin, noradrenaline and dopamine, and — critically — abolishes the gut/liver clearance of dietary and drug-derived amines. On its own this produces mild activation, warmth, and at higher caapi doses nausea and tremor; combined with serotonergic drugs or tyramine it becomes the mechanism of the two classic MAOI emergencies (serotonin syndrome; hypertensive crisis). This is the same enzyme that clears DMT — hence the ayahuasca synergy.
Harmine, harmaline and THH together convert a non-orally-active tryptamine into a 4–6 hour oral psychedelic. Clinical ayahuasca research (for treatment-resistant depression, for example) leans on exactly this pairing; the antidepressant signal is usually attributed to DMT's 5-HT2A-driven plasticity, with the β-carbolines cast as pharmacokinetic enablers that also contribute their own monoaminergic and possibly neurotrophic effects.
Independently of MAO, harmine is a benchmark DYRK1A inhibitor. DYRK1A sits on chromosome 21 (triplicated in Down syndrome) and phosphorylates tau, NFAT and cell-cycle regulators. Harmine's DYRK1A blockade has made it a workhorse probe in adult neurogenesis, tau/Alzheimer's, and — most prominently — the only well-validated small-molecule trigger of human pancreatic β-cell proliferation, a live diabetes-regeneration lead. These programs use harmine as a scaffold precisely because its MAO activity is a liability to engineer out.
Isolated in the 19th century and briefly trialled in the 1920s–30s as "banisterine" for post-encephalitic Parkinsonism (an early, crude MAO-inhibitor rationale), harmine has cycled through neurology, psychiatry and now oncology/endocrinology. Its modern interest is bimodal: as the indispensable MAOI component of ayahuasca-assisted therapy research, and as a kinase-inhibitor chemotype — two research communities that rarely cite each other.
The through-line for harm reduction is simple and non-negotiable: whatever else harmine is doing, it is a real MAO-A inhibitor at the doses people actually take. "Reversible" and "plant-based" do not exempt it from MAOI pharmacology while it is in the body.
Evidence-based, non-moralistic. Harmine is broadly low-toxicity on its own, but as an MAO-A inhibitor it creates the exact same drug and food interactions as any MAOI. The most common real-world ayahuasca casualties are not from DMT — they are from combining the brew with an SSRI/SNRI, or from an undisclosed serotonergic drug.
| Target | Affinity | Rel. | Mechanism |
|---|---|---|---|
|
MAO-A
Monoamine oxidase A (MAOA)
|
Ki ≈ 5 nM
IC₅₀ ≈ 4 nM · range 1–17 nM
|
Reversible inhibitor | |
|
DYRK1A
Dual-specificity kinase 1A
|
IC50 ≈ 30–80 nM
|
ATP-competitive inh. | |
|
CLK
CDC-like kinases (CLK1/4)
|
Low nM (potent)
|
Inhibitor | |
|
MAO-B
Monoamine oxidase B (MAOB)
|
Ki ≈ 121 µM
~7,000× weaker than MAO-A
|
Weak inhibitor |