Natural Endocannabinoid-System Modulators: the Master Table

How to read the table

Three columns do the real work. Target names the molecule the compound acts on, because that is what transfers to a preparation not in the table. Reported potency gives the number with its units and its kind, since an IC50, an EC50 and a Ki are not interchangeable quantities and comparing them across papers is indicative at best. Assay level is the honesty column: isolated enzyme or in vitro means the compound met a purified or cellular target in a dish; cell means an intact-cell uptake or signalling assay; animal means a whole-organism effect was measured; human means a person. Read down the assay-level column before reading anything else. The number of rows that reach animal is small, and the number that reach human, for an endocannabinoid endpoint, is zero.

The table contested in vitro

Potencies are as reported in the cited source. Where a figure is carried from the operator corpus without a traced primary measurement, the citation column says so and the figure should be treated as unverified. Two synthetic compounds are included at the bottom purely as scale references; they are not botanicals and no preparation route for them appears anywhere on this shelf.

Contested — caveat. Potencies come from different laboratories using different assay formats, enzyme sources and substrate concentrations; cross-paper comparison is indicative only. Three rows (gingerols and shogaols, spilanthol, and the macamide uptake figure) rest on claims carried by the operator corpus that could not be traced to a primary source and are labelled unverified in the table itself.

Sources: Tung MC 2021 · King AR 2009 · Gertsch J 2008 · Ligresti A 2012 · Nicolussi S 2014 · Wu H 2013 · Thors L 2010 · Thors L 2008 · Primary reference not identified during compilation* · Primary reference not identified during compilation* · Raduner S 2006 · Chicca A 2017 · Reynoso-Moreno I 2021 · Long JZ 2009 · Van Kush Family Research Institute 2026*

THE CAVEAT: a micromolar IC50 is not a dietary effect in vitro

This is the section that separates a research reference from supplement marketing, and it is placed here in the body of the page on purpose. An IC50 of 9.5 µM means that half of the enzyme activity was inhibited when the compound was present at that concentration in a dish, at the enzyme, for the duration of the assay. To translate that into a human effect you need every one of the following to be true, and for most of the compounds in the table above at least one of them is unknown: that the compound survives the gut; that it survives first-pass hepatic metabolism; that enough of it enters the circulation to matter (oral bioavailability); that it reaches the tissue where the enzyme is, which for MAGL and FAAH in the brain means crossing the blood-brain barrier; that free, unbound concentration at the enzyme — not total plasma concentration, since these are lipophilic, protein-bound molecules — reaches the micromolar range; and that it stays there long enough. A polyphenol present at a few milligrams in a cup of tea, with single-digit-percent bioavailability and rapid glucuronidation, is not delivering micromolar free concentrations to a brain enzyme. Two compounds in the table have animal behavioural data showing the translation actually occurred for them (guineensine and β-caryophyllene). None has human data for an endocannabinoid endpoint. Stating an in-vitro IC50 and letting a reader infer a dietary effect is the specific move this section exists to block.

• In vitro IC50 is a property of the compound-plus-enzyme pair, not of a food.
• The chain from mouth to enzyme has at least six independent failure points, and for most of these compounds several are unmeasured.
• Total plasma concentration overstates the relevant figure for lipophilic, highly protein-bound compounds; free fraction is what meets the enzyme.
• An animal behavioural effect is the minimum evidence that the translation happened at all.
• No compound in the table has human efficacy data for an endocannabinoid-mediated endpoint.

Sources: Nicolussi S 2014 · Gertsch J 2008 · Anand P 2007 · Huggins JP 2012

Where a compound has a known pharmacokinetic problem, say so human data

Two entries need their own warnings. 8-Prenylnaringenin is the most potent phytoestrogen so far identified in hops and beer: Milligan and colleagues characterised its estrogenic activity and it exceeds that of the better-known soy isoflavones by a wide margin. That is not a side note to be buried under the MAGL number. A compound proposed for endocannabinoid modulation that is also a potent estrogen receptor agonist has a real pharmacological consequence in anyone with hormone-sensitive tissue, and it is also extensively conjugated on first pass, so the systemic exposure that reaches an enzyme is not the dose swallowed. Curcumin, which appears in the operator corpus as a CYP-inhibiting bioavailability enhancer and a putative anandamide elevator, has the most notorious bioavailability problem in the entire natural-products literature: poor absorption, rapid intestinal and hepatic glucuronidation and sulfation, and rapid systemic elimination, reviewed by Anand and colleagues. That is why it is always co-administered with piperine — Shoba and colleagues measured the piperine effect on curcumin pharmacokinetics in humans — and co-administering a CYP and P-glycoprotein inhibitor to fix a bioavailability problem is itself a drug-interaction hazard, which is the subject of the piperaceae page and the cyp450 shelf.

Sources: Milligan SR 1999 · Anand P 2007 · Shoba G 1998 · Volak LP 2008 · Bhardwaj RK 2002 · Thors L 2008

Correcting one claim that circulates widely contested in vitro

8-Prenylnaringenin is frequently described, including in the operator datasheet this shelf draws on, as the most potent natural MAGL inhibitor identified. That is not correct as stated. Its reported IC50 is approximately 9.5 µM, while King and colleagues reported pristimerin and euphol as reversible natural-product MAGL inhibitors with potencies in the nanomolar range — roughly two orders of magnitude more potent. The claim is recorded here with its correction rather than repeated, because a shelf that propagates a convenient error is not usable as a research reference. The 8-PN finding is real and worth having; the superlative attached to it is not.

Contested — caveat. The correction rests on cross-paper IC50 comparison, which is indicative rather than quantitative; the direction of the difference (two orders of magnitude) is nonetheless too large to be an artefact of assay format.

Sources: Tung MC 2021 · King AR 2009 · Van Kush Family Research Institute 2026*

What would actually move any of this forward human data

For any compound in the table, the missing evidence is the same and is specific: a human pharmacokinetic study measuring free plasma and, where the target is central, cerebrospinal-fluid concentration after a realistic oral dose, alongside a measured target-engagement biomarker — circulating anandamide, 2-AG, or the fatty-acid amides, which is exactly what the pharmaceutical FAAH and MAGL programmes measured. That study design is unglamorous and it is the only thing that converts a row in this table from a chemistry fact into a pharmacology fact. Until then the honest description of this table is: a map of molecular targets that botanical compounds are known to hit in vitro, with two compounds that demonstrably reach the target in an animal, and a research agenda.

Sources: Huggins JP 2012 · Nicolussi S 2014 · Cisar JS 2018