Color development has played a significat role in the early studies on Cannabis (Cannabis sativa L.) and cannabinoids. Thus, the first phytocannabinoids were purified from Cannabis red oil, a deep-red high-vacuum distillation fraction of Cannabis extracts.^1,2 A red-purple color was also observed when fiber hemp or hashish was treated with methanolic KOH.³ Under these conditions, the development of a color is specific for Cannabis and Cannabis-derived products (marijuana, hashish),⁴ and the reaction has long been proved as an expeditious method for their identification in a forensic context (Beam test).⁴

The nature of the pigment from Cannabis red oil is still unclear, but color formation in the Beam test is the result of the aerobic oxidation of cannabidiol (CBD, 1) to the hydroxyquinone 2 (cannabidiolquinone, CBDQ, HU-331),⁵ a compound that has attracted considerable interest because of its selective anticancer activity^6,7 and catalytic inhibitory properties on topoisomerase IIα.⁸ While development of 2 as a drug was abandoned, possibly because of unfavorable stability properties (vide infra) and cellular toxicity,⁹ distinct lines of research rekindled interest in this compound. Thus, microsomial formation of 2 from CBD(1) has been associated with P450 covalent inhibition and perturbation of hepatic xenobiotics metabolism,¹⁰ and a similar process could also underlie the liver toxicity reported for high dosages of CBD.¹¹ Furthermore, 2 is formed during long-term storage of CBD under aerobic conditions,¹² and its availability is therefore important for quality control of this active pharmaceutical ingredient (API).

Despite the convergence of interest for CBDQ (2) from various areas of cannabinoid research, its only reported synthesis is the one inspired by the Beam test, that is, the aerobic oxidation of CBD in a cooled biphasic petroleum ether/5% ethanolic KOH system.^5,6 Under these conditions, yields are erratic, scale-dependent, and modest (ca. 20% at best),^5,6 while significant amounts of the dimeric quinone 3 are also formed by oxidative dimerization of CBDQ.⁵ Both reaction products, especially 3, are unstable and rapidly turn into a complex mixture of polar compounds.⁵ In our hands, the oxidation reaction was poorly reproducible and could not be scaled up over a few hundred milligrams of starting material, even when air or 80% oxygen was bubbled into the biphasic reaction system. A more reproducible behavior was observed with KH or LiH in tetrahydrofuran (THF) or toluene under heterogeneous conditions, but scale-up was still problematic. While Beam-type oxidation strategies were eventually abandoned, their mechanism is worth mentioning. Thus, the reaction is presumably triggered by formation of a phenolate anion, next oxidized to an electrophilic radical (4) that adds to dioxygen to form a hydroperoxy radical. The latter is reduced to the corresponding anion (5) by a second phenolate ion, and, after tautomerization to 6, the hydroperoxy anion is trapped by the para-carbonyl group. This generates the bridged keto-peroxyhemiacetal 7, whose α-deprotonation triggers cleavage of the peroxidic bond, eventually affording the hydroxylated quinone 2 via the hydrate 8 (Figure 1).

This process is reminiscent of the transformation of vitamin K hydroquinone into its epoxyquinone form,¹³ and the mechanism outlined in Figure 1 could explain the sensitivity of the reaction to radical traps like butylated hydroxytoluene (BHT) as well as the unreactivity of monoalkylated phytocannabinoids, like Δ⁹-tetrahydrocannabinol (Δ⁹-THC, 9) and cannabichromene (CBC, 10), where the prototropic equilibrium required for the formation of the peroxyhemiacetal is not possible (cf. the formation of 6 from 5 in Figure 1).

The reaction profile of the Beam test was basically replicated, without any substantial improvement of yield, by metal oxidants [FeCl₃, K₃[Fe(CN)₆], MnO₂, Cr⁶⁺-based reagents, CuCl, CuCl₂, Ag₂O, NH₄Ce(NO₃)₅] under both catalytic and stoichiometric conditions, as well as by peroxides (tert-butyl hydroperoxide (TBHP), basic H₂O₂), with significant amounts of the dimer 3 being always formed under basic conditions or during the long reaction times required to achieve a significant conversion. A surprising and notable exception was the behavior of the Takehira complex (CuCl₂-hydroxylamine),¹⁴ which afforded a mixture of the hydroxyiminodienone 11 and the chlororesorcinol 12. The regioselectivity of the formation of 11 was deduced from the diagnostic ³J heteronuclear multiple bond correlation (HMBC) cross-peaks of H-1″ with the hydroxyiminocarbonyl carbon.

The Takehira complex was originally developed for the oxidation of methylpolyphenols to their corresponding hydroxyquinones,¹⁴ a reaction of relevance for the industrial synthesis of vitamin E,¹⁴ and was later modified by replacement of hydroxylamine with other nitrogen bases.¹⁵ In control experiments, copper(II) chloride alone gave CBDQ (2) and the dimer 3 as the only reaction products, while the quinone 2 did not react with hydroxylamine, suggesting a role for hydroxylamine in the chemoselective halogenation reaction, possibly via the generation of an N-chlorinated species, and of copper(II) in the activation of the quinonecarbonyl carbon toward nucleophilic attack by hydroxylamine.

Hypervalent iodine derivatives have become increasingly popular for a wide range of oxidative reactions,¹⁶ and bis(trifluoroacetoxy)iodobenzene (BTIB) was reported to oxidize the mono-O-alkylated cannabinoid Δ⁹-THC (9), otherwise unreactive in Beam-type oxidations,⁶ to its corresponding hydroxyquinone.⁶ This λ³-iodane was also able to oxidize CBD to CBDQ, but λ⁵ iodanes like 2-iodoxybenzoic acid (1-hydroxy-1λ⁵,2-benziodoxole-1,3-dione, IBX, 13)¹⁷ and the Dess-Martin periodinane (DMP)¹⁸ gave much better and more reproducible yields, with a stabilized and not explosive version of IBX (SIBX)¹⁹ emerging as the reagent of choice. The superior behavior of SIBX compared to IBX might be related to the acidity of the stabilizing matrix (isophthalic and benzoic acids), which could help the hydrolytic cleavage of iodic esters formed in the reaction.¹⁹

The oxidation is presumably initiated by the sigmatropic rearrangement of the iodine–oxygen bond in the mixed λ⁵ iodane ester 14 formed by interaction of IBX and the C-1′ phenolic hydroxy group (Figure 2). The resulting C-2′ λ³-quinol 15, after oxidation to the corresponding λ⁵-iodane 16, is transformed by [3.3]-sigmatropic rearrangement of the carbon–oxygen bond into the C-4′ λ⁵-iodinane 17, with β-elimination eventually generating the hydroxyquinone 2 and a reduced λ-iodinane. Remarkably, dimerization was completely suppressed under iodinane oxidation, and yields in the range of 50–60% could be obtained at multigram reaction scale. CBDQ, an orange powder,²⁰ is unstable in solution, rapidly degrading in both protic (methanol) and aprotic (acetone, CHCl₃) solvents, with generation of the more polar dimer 3 next to a host of uncharacterized more polar products. On the other hand, it could be stored for at least 10 months as a powder at −18 °C in a sealed flask, or for additional time as a frozen benzene or dimethyl sulfoxide (DMSO) solution at 4 °C.²¹

The oxidation with SIBX is general for phytocannabinoids, and, apart from cannabigerol (18), it could also be applied to monoetherified compounds [cannabichromene (CBC, 10), cannabinol (19)] that are unreactive under Beam-test conditions, to afford their corresponding hydroxyquinones 20–22.

CBDQ (2) has been reported to be non-narcotic⁶ and lacks significant affinity for CB₁ and CB₂ receptors.⁹ Nevertheless, it showed powerful modulating activity on peroxisome proliferator-activated receptor gamma (PPAR-γ),^9,22 and various degrees of PPAR-γ activating activity were also shown by the other cannabinoquinoids (Table 1). However, dimerization was detrimental for activity, and dimeric quinones were devoid of significant activity in PPARγ-activity assays.²³ Dimeric quinones are axially chiral, and, since enantiomeric cannabinoids can show markedly different profiles of bioactivity,² the one from CBG (CBGQ (23)) was resolved by chromatography on a chiral-phase column packed with amylose-tris(5-chloro-2-methylphenylcarbamate). However, both the (aR) and the (aS) enantiomers turned out to be inactive.²³ Similarly, the hydroxyiminodienone 11 and the chlorinated resorcinol 12 were also devoid of activity.²³

In conclusion, we have developed a reproducible and scalable synthesis of cannabinoquinoids, including CBDQ (2), significantly enhancing access to this compound²⁴ of relevance not only for its bioactivity profile but also for the analytics of CBD, the study of its binding to P450 apoproteins, and its effects on liver function.