Grayanotoxin I vs Grayanotoxin III: What Is the Difference?

Grayanotoxin I and grayanotoxin III are the two compounds that mad honey testing actually measures. They are the two with commercially available analytical reference standards, the two the European Food Safety Authority built its 2023 risk assessment around, and the two that appear as separate lines on a competent certificate of analysis. Readers who encounter both names reasonably assume the distinction carries weight: that one is the real active compound and the other is a lesser relative, or that a sample reporting one is meaningfully different from a sample reporting the other.

The short answer is that the two molecules differ by a single acetyl group, and that the difference in effect is smaller than the naming implies. EFSA assigned both a potency factor of 1, meaning it treated them as similarly toxic and assessed them as a summed total rather than separately. That decision was made under acknowledged uncertainty rather than on the strength of a direct human comparison, and this page states where that uncertainty sits rather than presenting the equivalence as settled fact.

The practical consequence runs the other way from what most people expect. Because the two are assessed together, a result that reports only one of them is not a more focused measurement. It is an incomplete one. That is why both belong on a lab report, and why a claim that singles out one isoform as the meaningful number should be treated as a claim about the testing rather than the honey.

Contents

1. The structural difference: One acetyl group
2. Both act at the same site on the voltage-gated sodium channel
3. Relative potency: What the evidence actually supports
4. Side by side: Grayanotoxin I and Grayanotoxin III
5. Do they occur together in honey?
6. Where Grayanotoxin II fits
7. What this means when you read a lab report
8. Sources

The structural difference: One acetyl group

Grayanotoxins are diterpenes built on the grayanane skeleton, produced by plants in the Ericaceae family, which includes Rhododendron species whose nectar produces mad honey. More than twenty-five grayananes have been described. Only a small number occur in honey at concentrations studied toxicologically, and the isoform numbering reflects the historical order in which they were isolated and characterised, not a ranking of potency. Grayanotoxin I is not the first because it is the strongest.

Grayanotoxin I is grayanotoxane-3,5,6,10,14,16-hexol 14-acetate. Grayanotoxin III shares that skeleton and that pattern of hydroxyl groups, but carries a free hydroxyl at carbon 14 where grayanotoxin I carries an acetate ester. The molecular formulas EFSA lists in its 2023 opinion make the relationship explicit: grayanotoxin I is C22H36O7 and grayanotoxin III is C20H34O6, a difference of C2H2O, which is exactly one acetyl group. Structurally, grayanotoxin III is the deacetylated analog of grayanotoxin I.

Grayanotoxin II differs more substantially. It carries an exocyclic methylene at the carbon 10 to carbon 20 position in place of the 10-hydroxyl that both grayanotoxin I and grayanotoxin III possess, and it carries no acetate. That is a change to the ring system rather than a substitution at a single position, and it is the structural reason grayanotoxin II is consistently treated as the less toxicologically significant of the three, a point returned to below.

One acetyl group is a small change by the standards of natural product chemistry, and it would be easy to conclude from that alone that the two compounds must behave identically. The structure-activity literature does not support that conclusion, nor does it support the opposite one. It does support the more limited statement made in the next two sections.

Both act at the same site on the voltage-gated sodium channel

Grayanotoxins are sodium channel toxins. This is worth stating plainly because mad honey is frequently described in popular coverage as a psychedelic or a hallucinogen, and that framing misdescribes the mechanism. Grayanotoxins bind the group II receptor site, commonly called site 2, on the voltage-gated sodium channel, at segment 6 of domains I and IV. Binding at that site interferes with the fast inactivation gate, so channels that would normally close after opening are held in a modified open state. The result is prolonged sodium influx and persistent depolarisation of excitable membranes, a pharmacological description of a cardiac and neuromuscular effect, not a perceptual one.

The literature describes grayanotoxin I and grayanotoxin III as the principal toxic isoforms acting at this site. No source reviewed here proposes a different binding site or mechanism for either. Evidence is thinner on the quantitative question: how tightly each binds, how binding differs between sodium channel subtypes, and whether the acetyl group at carbon 14 changes binding and unbinding kinetics. Reported affinity figures for grayanotoxin at this site sit in the region of ten micromolar, but they come from in vitro channel preparations rather than human tissue, and published comparisons between subtypes have largely used amphibian and rodent channels.

Why an in vitro affinity figure does not settle the question

A binding affinity measured on an isolated channel preparation tells you about the molecule and the channel. It does not tell you how much of the compound reaches that channel after a honey is eaten, how it distributes, or how quickly it is cleared, and EFSA states directly that it had no information on the biotransformation of grayananes in mammals. Two compounds with similar affinity at the same site can still produce different exposure profiles if one is absorbed or eliminated differently, and for grayanotoxin I and grayanotoxin III that comparison has not been made in humans. This is a genuine gap, not a technicality, and it is the reason the potency question below is answered with a risk-assessment convention rather than a measurement.

Relative potency: What the evidence actually supports

Three bodies of evidence bear on which of the two is more potent, and they do not all point the same way. Read together, they support the conclusion that the difference is small and not reliably resolved, which differs from the conclusion that the two are identical.

The animal lethality data

The older comparative toxicology places grayanotoxin I and grayanotoxin III together at the more toxic end of the grayanotoxin series, with intraperitoneal LD50 values in mice for the most toxic isoforms falling in a narrow band of roughly 0.87 to 1.3 milligrams per kilogram of body weight. Within that band, individual reported values place grayanotoxin III slightly below grayanotoxin I, making it marginally more toxic at that endpoint. Three limitations apply to that figure: it is a lethality endpoint in mice, it is an intraperitoneal injection rather than an oral exposure, and the separation between the two values is small relative to the variation normally seen between such studies. It does not establish a potency ranking that can be carried into human oral exposure.

The structure-activity data

The structure-activity work points, if anything, the other way. Hotta and colleagues, working on positive inotropic effect in guinea pig tissue in 1980, reported that the 3-beta-hydroxyl, 6-beta-hydroxyl, and 10-beta-methyl groups on the grayanane skeleton are essential for the effect, and that potency increased with a 10-alpha-hydroxyl group and with acylation of the 14-beta-hydroxyl group. Acylation at carbon 14 is precisely the feature that distinguishes grayanotoxin I from grayanotoxin III. At that endpoint, in that preparation, the acetyl group is potency-enhancing, which would place grayanotoxin I above grayanotoxin III.

These two findings are not a contradiction that needs resolving by picking a winner. They measure different things: lethality in one species via one route, and contractile effect in isolated tissue of another species. A compound can rank differently on those two endpoints without either result being wrong. Together, they show that ranking is endpoint-dependent and therefore not a molecular property that can be quoted as a single number.

How EFSA handled it

The EFSA CONTAM Panel resolved the question for regulatory purposes by not resolving it. In its 2023 opinion on grayanotoxins in certain honey, the Panel assigned a potency factor of 1 to both grayanotoxin I and grayanotoxin III because they have similar toxicity, and derived a single acute reference point applying to the sum of the two. That reference point is a BMDL10 of 15.3 micrograms per kilogram of body weight, based on the reduction of heart rate following intraperitoneal injection of grayanotoxin III in rats.

Two features of that derivation matter for anyone citing it. The reference point is anchored on animal data by an injected route, and the equal potency factor is a convention adopted under uncertainty, chosen because the available evidence did not justify weighting one isoform above the other. This is a defensible regulatory decision, but it is not the same as demonstrated equivalence. The Panel selected these two compounds partly because they are the grayananes most studied toxicologically and partly because commercial reference standards exist for them, which is a constraint of the analytical supply chain as much as a judgment about biological importance.

Why the assessment sums them rather than taking the higher of the two

Summing follows from the equal potency factor rather than a separate decision. When two compounds are judged to contribute comparably to the same effect through the same mechanism, the quantity that corresponds to the reference point is their combined total, and treating either one in isolation would understate the exposure. This same logic applies to other groups of structurally related contaminants, which is why the EFSA reference point is expressed against the sum rather than against whichever isoform is higher in a given sample.

It follows that two samples can reach the same summed total by very different routes, one dominated by grayanotoxin I and the other by grayanotoxin III, and that on the current assessment those two samples are not distinguished. Whether that convention is conservative or permissive for any individual sample depends on whether the true potencies are in fact equal, which is the uncertainty the potency factor absorbs rather than resolves. Anyone reporting a summed figure should state that it is a sum and name the compounds included, because a total presented without that detail cannot be checked against the reference point or another laboratory result.

Side by side: Grayanotoxin I and Grayanotoxin III

 

The final row summarises the comparison, which is why this page does not offer a ranking. The two isoforms sit close enough together that the answer changes with the endpoint measured.

Do they occur together in honey?

Usually, but not in a fixed ratio. The Ahn et al. 2022 analysis of 60 mad honey samples confiscated by Korean customs from travellers arriving from Nepal, which remains the largest quantitative dataset of its kind for Nepal-origin honey, detected grayanotoxin I and grayanotoxin III together in 33 of the 60 samples. The method measured only those two isoforms, and neither was detected in the other 27 samples.

The concentration ranges in that study are the more instructive part. Grayanotoxin I ranged across an 86-fold span and grayanotoxin III across a 255-fold span. Because the spans differ so much, the concentration of one isoform in a given sample poorly predicts the concentration of the other. A sample at the low end for grayanotoxin I is not reliably at the low end for grayanotoxin III, and a certificate of analysis that reports one figure does not allow the other to be inferred.

The dataset’s limits should be stated. It is a single study, on samples from one origin, collected in one regulatory context, and the authors did not establish why the variation is so wide. The analysis does not determine whether the range reflects different Rhododendron species, harvest timing, colony locations, post-harvest handling, or a combination. The co-occurrence pattern described above should therefore be read as what these 60 samples showed, not as an established property of mad honey in general.

Where Grayanotoxin II fits

Grayanotoxin II is the third isoform most often discussed, but it is rarely quantified: commercial reference standards exist only for grayanotoxin I and grayanotoxin III (EFSA 2023). It is structurally more distant from grayanotoxin I and grayanotoxin III, lacking the 10-hydroxyl that structure-activity work identifies as potency-enhancing. It is reported as less toxic than either. Ahn et al. (2022) did not measure it. Where it has been reported, levels were low: up to 1.1 µg/g in honey imported from Nepal (Lee et al. 2008) and a semi-quantitative 2.0 µg/g in one Turkish honey (These et al. 2015), both as reported by EFSA (2023).

EFSA puts the potency of grayanotoxin II an order of magnitude below that of grayanotoxin I and III, and built its reference point around the sum of grayanotoxin I and grayanotoxin III (EFSA 2023). This is not an argument that grayanotoxin II is absent or irrelevant; a laboratory that quantifies all three provides more information than one that quantifies two.

What this means when you read a lab report

The most common practical error this distinction creates is treating a single-isoform result as complete. A certificate of analysis that reports grayanotoxin III alone, or reports an undifferentiated grayanotoxin figure without saying which compounds it includes, has not measured what the EFSA reference point is expressed against. Because that reference point applies to the sum of grayanotoxin I and grayanotoxin III, a report covering only one cannot be compared against it.

A second limit applies even to a thorough report. EFSA notes that analytical standards are commercially available only for grayanotoxin I and grayanotoxin III, which means that a laboratory running a validated method for those two is silent on the other grayananes that may be present in the sample. This is a known and acknowledged constraint of the method rather than a failing of any particular laboratory, and it is a reason to read a clean result as evidence about two specific compounds rather than as evidence that the honey contains nothing.

Three related claims are worth recognising because they turn on the same confusion. A pollen analysis reporting a high percentage of Rhododendron pollen establishes floral origin, not nectar toxin content, and cannot substitute for quantification of either isoform. The phrase lab tested, used without naming the analyte, can describe a legitimate test for moisture or sugar profile in which grayanotoxin was never measured. And any suggestion that a person can establish what is in a jar by consuming some of it and observing the result describes a test that scales with the risk it claims to manage, which is why this page does not describe one.

 

Sources

• Ahn SY, Kim S, Cho H (2022), Determination of Grayanotoxin I and Grayanotoxin III in mad honey from Nepal using liquid chromatography-tandem mass spectrometry. Analytical Science and Technology 35(2):82-91.
• EFSA CONTAM Panel (2023), Risks for human health related to the presence of grayanotoxins in certain honey. EFSA Journal 21(3):7866.
• Lee SY, et al. (2008), Determination and monitoring of grayanotoxins in honey using LC-MS/MS. Korean Journal of Food Science and Technology 40:8-14.
• These A, et al. (2015), A case of human poisoning by grayanotoxins following honey ingestion: elucidation of the toxin profile by mass spectrometry. Food Additives & Contaminants: Part A 32(10):1674-1684.