Volume 7, Issue 4 e202200291
Research Article
Open Access

Synthesis and Characterization of a Cannabinoid Type 2 Receptor Photoactivated Prodrug

Jiazhen Yin

Jiazhen Yin

Department of Chemistry, University of Otago, Dunedin, New Zealand

Search for more papers by this author
Raahul Sharma

Raahul Sharma

Department of Pharmacology and Clinical Pharmacology, School of Medical Sciences, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand

Search for more papers by this author
Assoc. Prof. Joel D. A. Tyndall

Assoc. Prof. Joel D. A. Tyndall

School of Pharmacy, University of Otago, Dunedin, New Zealand

Search for more papers by this author
Dr. Natasha L. Grimsey

Dr. Natasha L. Grimsey

Department of Pharmacology and Clinical Pharmacology, School of Medical Sciences, Faculty of Medical and Health Sciences, University of Auckland, Auckland, New Zealand

Search for more papers by this author
Dr. Andrea J. Vernall

Corresponding Author

Dr. Andrea J. Vernall

Department of Chemistry, University of Otago, Dunedin, New Zealand

Search for more papers by this author
First published: 22 December 2022
Citations: 1

Graphical Abstract

Breaking point: A cannabinoid type 2 receptor agonist prodrug has been developed, that can be rapidly cleaved upon exposure to λ=450–455 nm light. This prodrug has significantly reduced receptor binding compared to the released chromenopyrazole agonist. The coumarin products produced in the prodrug photolysis varied in different solvents and light conditions.


G protein-coupled receptor (GPCR) ligand prodrugs that can be released on demand with spatiotemporal control are powerful chemical tools that can assist in elucidating the consequences of GPCR activation or blockade at precise locations and times both in vitro and in vivo. Cannabinoid receptor prodrugs are of interest from both a drug delivery perspective and as tools to unravel the potential for differential signaling responses to be produced from cannabinoid receptor populations in distinct subcellular locations. Herein, the development and characterization of a cannabinoid type 2 receptor agonist prodrug is described, based on a 4-diethylamino-coumarylidenemalononitrilemethyl (DEACM-MN) photo caging moiety linked via a carbonate to a chromenopyrazole cannabinoid ligand. This prodrug showed rapid photolysis with 450–455 nm light and good stability in biologically relevant buffer. The formation of various coumarin products alongside drug release was studied in different conditions. Radioligand binding assays were conducted with the prodrug, which revealed a significantly decreased human cannabinoid type 2 receptor binding affinity than the active chromenopyrazole parent ligand.


The cannabinoid type 2 receptor (CB2R) is a G protein-coupled receptor (GPCR) that is highly expressed in microglia and in immune system cells,1 differing from the cannabinoid type 1 receptor (CB1R), which is mainly expressed in neurons in the central nervous system. CB2R has received great attention as a potential drug target but as yet, drugs selectively targeting CB2R are not in the clinic. For example, CB2R is linked to autoimmune and inflammatory diseases, neurodegeneration, osteoporosis and pain.2 Ligands and tool compounds can be utilized to better understand the role and spatiotemporal signaling profiles of CB2R, both in healthy and diseased tissues, and hold promise to unlock the clinical potential of drugs targeting CB2R. Prodrugs, defined as compounds with little or no biological activity that undergo a change to the ‘active’ ligands in vivo, are useful tool compounds that have been designed to address an array of factors. Cannabinoid receptor prodrugs have been developed for several reasons.3 Glycosylated cannabinoids such as prodrug ‘VBX-100’ are under development for the treatment of inflammatory gastrointestinal tract conditions, where the CB1R and CB2R agonist Δ-9-tetrahydrocannabinol (Δ-9-THC) is released in the large intestine with reduced systemic exposure.4, 5 Cannabinoid ligand Δ-9-THC has also been developed into a prodrug that is more hydrophilic than the poorly water soluble and lipophilic Δ-9-THC via conjugation with valine hemisuccinate, thereby allowing for ocular drug delivery lowering of intraocular pressure.6

Light-activated prodrugs are particularly useful as spatiotemporal release can be tightly controlled. A CB2R light-activated prodrug has been reported for anticancer applications, based on the CB2R selective antagonist-scaffold SR144528.7 This ‘dual-acting’ prodrug contained a phthalocyanine photosensitizer that was irradiated with 690 nm light to produce reactive oxygen species (ROS) and thereby also cleave the ROS-sensitive linker and release the active cannabinoid ligand, although functional cannabinoid ligand data is not reported. There have been many examples of photo-switchable GPCR ligands including for CB2R,8 that typically contain interchangeable active/inactive cis or trans isomers, as recently reviewed in Wijtmans et al.9 There are also examples of light-activated GPCR ligand prodrugs whereby a GPCR ligand is caged with a photocleavable moiety such as a nitrobenzyl, 7-nitroindoline or coumarin moiety, as reviewed in Ricart-Ortega et al.10 7-Diethylamino-4-hydroxymethylcoumarin (DEACM) is a commonly used light-activatable caging system for many applications, with examples including a light-activatable adenosine A2A receptor antagonist prodrug11 and a metabotropic glutamate receptor 5 negative allosteric modulator prodrug.12

We are motivated to develop a light-activated CB2R selective agonist prodrug because of our interest in studying the implications of spatiotemporal activation of CB2R. Development of a CB2R photocaged ligand has several advantages compared to a photo-switchable8 prodrug, including the ability to incorporate targeting or templating moieties that are separated from the active ligand upon uncaging and larger scope of light cleavage wavelength based on choice of photo-caging group. A photocaging group with maximum UV absorbance of 460–470 nm was selected to allow for future adaption into bioluminescence resonance energy transfer (BRET) applications with a luciferase/luciferin system.13 Instead of the commonly used DEACM photocaging group, we selected the comparatively red-shifted 4-diethylamino-coumarylidenemalononitrilemethyl (DEACM-MN) photocaging group to append to the CB2R selective agonist (prodrug 1, Figure 1).

Details are in the caption following the image

CB2R agonist-DEACM-MN prodrug 1 developed in this study, and previously reported CB2R agonist 2.14

In addition to the C−O bond cleavage that occurs to release the ‘drug’, a second branch point from the coumarin that is not cleaved by light can be used to introduce another functionality to the prodrug.15 In our work, we included a capped ‘linker’ in this position in order to probe if drug release with suitable speed still occurs and therefore if this position could be used, for example, to introduce a template for BRET.13b The CB2R agonist 2 (Figure 1), that was previously developed by our lab, has excellent CB2R affinity (hCB2R pKi=7.91±0.10)), good selectivity over CB1R (hCB2R pKi<5.30, over 407–fold), and behaves as a potent CB2R agonist (hCB2R pEC50=7.92±0.09, Emax=59.9±2.04 %).14 The phenol of 2 was selected as the attachment position for the DEACM-MN prodrug caging moiety since phenolic-substitution of this chromenopyrazole compound class results in very little/minimal CB2R activity.14, 16

Results and Discussion

The synthesis of CB2R selective agonist 2 (Figure 1) has been reported by our laboratory14 and can be made in 7 steps from readily available starting materials. However, there is only a 30 % yield14 in the last three steps in converting methyl ester 10 (10 shown in Scheme 1) to acetamide 2. Therefore, we first undertook light-activated studies in a ‘model’ system using the CB2R inactive compound 10. Previously reported coumarin 5 was prepared according to literature procedures,17 assumed to be the racemic mixture, and was acetylated using acetic anhydride to afford 6 (Scheme 1). In order to test the scope of having a longer linker with additional functionality in this position, 5 was coupled with 4 (which was prepared by alkylation of 3) to give amide 7. TBDMS-protected 6 and 7 were deprotected to afford DEACM-MN alcohols 8 and 9. The next coupling step was undertaken in a one-pot approach by first reacting N,N′-disuccinimidyl carbonate with 10 to form an N-hydroxysuccinimide-phenol intermediate, which then reacted with a coumarin alcohol (8 or 9) to form the carbonate-containing model prodrugs 11 and 12 in low yield. This low yield may be due to the excess of N,N′-disuccinimidyl carbonate used in the reaction, however a fold-excess of N,N′-disuccinimidyl carbonate was required to ensure the N-hydroxysuccinimide-phenol intermediate was formed quantitatively. Since a one-pot procedure, the excess unreacted N,N′-disuccinimidyl carbonate could have also reacted with the later added coumarin alcohol 8 or 9. The other species present immediately at the conclusion of the reaction time, aside from the desired product, were the phenol carbonate and coumarin alcohol carbonate species, and as expected in these mild reaction conditions malononitrile hydrolysis products were not observed. Activation of the phenol (10) first and then addition of the alcohol (8, 9) was preferred for steric hindrance reasons, and indeed when the alcohol (9) was treated with N,N′-disuccinimidyl carbonate followed by addition of phenol (10) minimal formation of 12 occurred (results not shown).

Details are in the caption following the image

Synthesis of model prodrugs 11 and 12. (i) 10-Bromodecanoic acid, NaI, NaH, DMF, 0–60 °C, 4 h; (ii) Ac2O, Et3N, CH2Cl2, rt, 12 h, 78 %; (iii) 4, HBTU, N,N-diisopropylethylamine, THF, rt, 12 h, 58 % (two steps from 3); (iv) Tetrabutylammonium fluoride, THF, 0 °C, 1 h, 98 % (8), 59 % (9); (v) N,N’-succinimidyl carbonate, N,N-dimethylpyridin-4-amine, CH2Cl2, rt, overnight, 13 % (11), 9 % (12).

Upon the successful synthesis and light-cleavage (described below) of model prodrugs 11 and 12, synthesis of the CB2R prodrug 1 was undertaken. The already synthesised 8 (Scheme 1) was coupled to CB2R agonist 2 (Figure 1) to form carbonate ester prodrug 1 (Scheme 2), again in moderate yield, using the conditions established for model prodrugs 11 and 12 (Scheme 1).

Details are in the caption following the image

Synthesis of prodrug 1. (i) 2, N-succinimidyl carbonate, N,N-dimethylpyridin-4-amine, CH2Cl2, rt, overnight, 14 %.

Model prodrugs 11, 12 and prodrug 1 were characterized by UV-VIS spectroscopy (Supplementary Figure S1) to determine if strong UV-VIS absorption did indeed occur in the intended 460–470 nm range. All three prodrugs (11, 12, 1) showed two strong absorption peaks at around 280 nm and 470–490 nm, attributed to the chromenopyrazole and the coumarin moiety respectively.

To measure if the coumarin-caged chromenopyrazoles could successfully be cleaved to release a chromenopyrazole scaffold upon irradiation with light, model prodrugs 11 and 12 were first studied. Model prodrugs 11 and 12 were dissolved in 9 : 1 acetonotrile:water and exposed to constant LED light of 450–455 nm. Reaction aliquots were removed at various time points and analyzed using peak integration from analytical reverse phase high-performance liquid chromatography (RP-HPLC) chromatograms. This data was used to plot the cumulative percentage cleavage of the prodrug over time (Figure 2A).

Details are in the caption following the image

Cumulative prodrug cleavage over time. A) Model prodrugs 11 and 12 (20 μM, 9 : 1 acetonitrile:water v:v) in the presence of LED light 450–455 nm. B) Prodrug 1 (20 μM, 9 : 1 acetonitrile:water v:v) in the presence of LED light 450–455 nm. C) Prodrug 1 (20 μM, 9 : 1 acetonitrile:water v:v) in the presence of LED light 525–530 nm. D) Prodrug 1 (20 μM, 9 : 1 acetonitrile:water v:v) in the presence of LED light 380 nm. E) Prodrug 1 (20 μM, 9 : 1 water:DMSO v:v) in the presence of LED light 450–455 nm. F) Stability of prodrug 1 (20 μM, 9 : 1 acetonitrile:water v:v) in ambient light (data shown as scatter as using the same exponential line fitting as for other data a line could not be generated) and in darkness. The cumulative cleavage of the prodrug (11, 12, 1) was calculated via RP-HPLC chromatogram peak integration of the prodrug peak. All data is the average (± error, note – errors bars are shown in A–F but in some cases are too small to see) of 3 independent experiments.

Both 11 and 12 were completely cleaved after 100 s of irradiation with 450–455 nm LED light (Figure 2A), and the equivalent quantitative appearance of chromenopyrazole 10 was observed by analytical RP-HPLC (Supplementary Figure S2). Model prodrug 12, which contains a longer chain extending from a third position of derivatization in the scaffold, showed comparable but slightly faster light cleavage than acetamide 11, demonstrating that further derivatization from this position is well tolerated and gives comparable prodrug cleavage and drug release.

Following the rapid cleavage of model prodrugs 11 and 12, the CB2R agonist prodrug 1 was synthesised (Scheme 2), evaluated for cleavage under various wavelengths of LED light in a solution of 9 : 1 acetonitrile:water v:v (Figure 2BD) and the uncaging quantum yield at the various wavelengths was determined (Table 1, refer to Supplementary Information for uncaging quantum yield calculations and details of photon flux measurement). Prodrug 1 was cleaved rapidly to completely release 2 within 60 s under 450–455 nm LED light (Figure 2B, Supplementary Figure S3 shows RP-HPLC chromatograms) with an uncaging quantum yield of 0.40. Under green LED light 525–530 nm (Figure 2C), cleavage of 1 was slightly slower (99 % cleavage within 2 min with 525–530 nm versus 100 % cleavage of 1 within 1 min under 450–455 nm LED light), with an uncaging quantum yield of 0.07. Under near-UV light (380 nm LED light), the cleavage of 1 was much slower (Figure 2D) and at this wavelength prodrug 1 also had a low uncaging quantum yield (0.04). These experiments confirmed that, as anticipated, prodrug 1 was indeed cleaved more quickly and had the best uncaging quantum yield with the 450–455 nm LED lamp.

Table 1. Photophysical and photochemical properties of prodrug 1.

LED light

Photon flux [Einstein s−1]/10−9

ϵabs [M−1 cm−1]/104


450–455 nm






525–530 nm




380 nm




  • [a] ϵabs, molar attenuation coefficient of 1 in 9 : 1 acetonitrile:water v:v at 453 nm. [b] ϵabs, molar attenuation coefficient of 1 in 9 : 1 water:DMSO v:v at 453 nm. [c] ϵabs , molar attenuation coefficient of 1 in 9 : 1 acetonitrile:water v:v at 528 nm. [d] ϵabs , molar attenuation coefficient of 1 in 9 : 1 acetonitrile:water v:v at 380 nm. [e] urn:x-wiley:23670932:media:cptc202200291:cptc202200291-math-0002 , uncaging quantum yield of 1 (20 μM) in 9 : 1 acetonitrile:water v:v. [f] urn:x-wiley:23670932:media:cptc202200291:cptc202200291-math-0003 , uncaging quantum yield of 1 (20 μM) in 9 : 1 water:DMSO v:v.

Prodrug cleavage experiments were then conducted in water from a stock of prodrug 1 in DMSO to aid solubility (9 : 1 water:DMSO), which were important for testing compatibility with future biological applications that would be carried out in aqueous environments. Upon exposure to LED light 450–455 nm, approximately 80 % of 1 was cleaved after 2 min, with over 99 % of 1 cleaved after 15 min (Figure 2E, Supplementary Figure S4A), with an uncaging quantum yield of 0.23. This was slightly slower than in an acetonitrile dominated solution (100 % cleaved within 1 min, Figure 2B) in otherwise equivalent conditions. This is in agreement with literature reports for the DEACM coumarin cage,18 where increasing water was reported to stabilize the excited state and therefore slow the excited state decay of the coumarin cage. The uncaging quantum yield of prodrug 1 is comparable to literature reports of similar photocages, for example Gandioso et al19 reported a DEACM-MN-derived caged aspartic acid with an uncaging quantum yield of 0.24 (505 nm, 1 : 1 acetonitrile:Tris buffer v:v). Similar uncaging quantum yields at lower wavelengths for DEACM-MN-based cages have also been reported, for example Fournier et al20 reported a “NdiEt-mcBA” model system with an uncaging quantum yield of 0.07 (365 nm, 1 : 1 acetonitrile:Tris buffer v:v).

The appearance of chromenopyrazole 2 as measured by HPLC peak integration was slightly different when prodrug 1 was cleaved in 9 : 1 water:DMSO (Supplementary Figure S4A). In acetonitrile-dominated conditions, the area of the peak corresponding to 2 on the analytical HPLC chromatogram increased as the peak corresponding to prodrug 1 decreased, but in 9 : 1 water:DMSO conditions, the peak area for 2 gradually increased up to around 120 s, then slightly decreased. Therefore, the stability of a synthetic sample of chromenopyrazole 2 in likewise conditions was carried out (Supplementary Figure S4B). This showed that over time (t=0 to 15 min), the area of the peak corresponding to 2 on the analytical HPLC chromatogram did gradually decrease in area (12 % reduction in peak area at 120 s) but peaks at other retention times were not present.

The stability of prodrug 1 (20 μM, 9 : 1 acetonitrile:water v:v) was assessed in ambient light (lab lights on, windows uncovered) and in darkness (Figure 2F, Supplementary Figure S5). Prodrug 1 was broken down slowly over time in ambient light, with approximately 50 % reduction of 1 (along with release of 2) after 8 hours. In the dark environment 1 was quite stable in the assay solution, with only approximately a 5 % reduction of 1 after 8 hours. Furthermore, the stability of prodrug 1 was evaluated in an environment to mimic that of an in vitro assay. In a 2-[4-(2-hydroxyethyl)piperazin-1-yl]ethane-1-sulfonic acid (HEPES) buffer in the dark at 30 °C, simulating the radioligand competition binding assay conditions (see below), prodrug 1 (10 μM) remained intact and release of chromenopyrazole 2 was not observed after 2 hours (Figure S6).

In all prodrug cleavage experiments (Figure 2), the requisite chromenopyrazole (2 or 10) was released, however, the coumarin product(s) produced alongside the chromenopyrazole varied under different conditions (solvents, light source) (Supplementary Figure S7, analytical HPLC chromatogram overlays). In water dominated solutions (water:DMSO, 9 : 1 v:v), experiments conducted with LED lamps produced coumarin alcohol 8 (Figure 3, pathway I) as the major coumarin product, confirmed by analytical HPLC chromatogram retention time, mass and robust nuclear magnetic resonance (NMR) characterization that matched that of the synthetic intermediate 8 (Scheme 1). According to literature reports of the coumarin scaffold photolysis mechanism,18a, 21 upon excitation, prodrug 1 undergoes heterolytic bond cleavage at the coumarin-methylene position to form carbocation 13 (Figure 3). Carbocation 13 then reacts primarily with water in water-dominated solutions to form coumarin alcohol 8 (Figure 3, pathway I).

Details are in the caption following the image

Proposed mechanism of prodrug 1 photolysis in the presence of light (Pathways I and II, ratio dependent on solvent system) and prodrug 1 hydrolysis in dark conditions.

When acetonitrile-dominated solutions (acetonitrile:water, 9 : 1 v:v) were exposed to light, 8 was formed as the minor coumarin product (Figure 3, Pathway I) along with a different major coumarin product. This predominating coumarin product, with the same molecular mass as 8, was isolated in a small quantity using semi-preparative RP-HPLC and is hypothesized from NMR spectra (Supplementary Figures S8 and S9) to be coumarin 17, a regioisomer of 8. The ratio typically formed across prodrug photolysis experiments was 6 : 1 17 : 8 in acetonitrile dominated solutions and 1 : 6 17 : 8 in water dominated solutions. Comparison of the 1H NMR, COSY and HSQC spectra of 8 to the proposed structure 17 revealed that the coumarin core was highly likely still intact in 17 as the 1H and directly attached 13C chemical shifts of Hc, Hd, He and Hc in 8 were very similar to Hc’, Hd’, He’ and Hc’ in 17, along with an almost identical NEt2 moiety. There were, however, several pronounced differences. The 1H chemical shift of the acetyl CH3 singlet was noticeably different (8 1.82 ppm; 17 2.18 ppm), which indicated 17 could have an O-acetyl functional group. The 1H NMR spectrum of 8 contained two heteroatom-H peaks (OH 5.99 ppm; NH 8.22 ppm), while 17 lacked these two signals but contained a broad singlet at 8.05 ppm that was attributed to an NH2 moiety. The 1H chemical shift of ‘Ha’ is quite different between the two coumarins (8 Ha 5.09 ppm; 17 Ha’ 6.23 ppm), again indicative that this proton is adjacent to an alcohol in 8 but an ester in 17 hence the downfield shift. The CH2 adjacent to Ha’ in 17 appeared to be (from HSQC, HBMS, COSY spectra data, Figure S9) underneath the water signal in the 1H NMR spectrum of 17, so further analysis of this peak was limited. An IR spectrum of 8 and 17 was run (Supplementary Figure S10), which revealed a distinctive strong absorption of an alcohol function group in 8 (3315.20 cm−1) and the lack of this distinctive absorption in 17.

Coumarin 17 could potentially form via Figure 3 Pathway II – when water is not the initial predominating nucleophile to react with carbocation 13. Instead, an intramolecular reaction could first occur to form a 4,5-dihydrooxazole five-membered intermediate 15, which could then be attacked by water. Coumarin 17 is essentially the N- to O-acyl transfer regioisomer of 8, however acyl migration is unlikely to have occurred after 8 had formed because typically O- to N- acyl migration (not the reverse) is more favorable and also because of the vastly different ratios of 8 versus 17 in different solvents. The cleavage of prodrug 12 with the long N-acyl chain with LED light (Supplementary Figure S2) did not produce coumarin alcohol 9 as a product but did produce a different coumarin of the same molecular mass as 9. This coumarin product was not further characterized by NMR spectroscopy but is likely to be the equivalent N- to O-acyl regioisomer of 9. While formation of intermediate 15 and product 17 may be unique to this prodrug connectivity, this does demonstrate the importance of considering the effect that carbocation stabilization has on product formation, in particular branching from the coumarin α-carbon. By example, coumarins containing an allylic system branched from the coumarin α-carbon have been reported to undergo double bond rearrangement to a more stable carbocation before ‘water trapping’ occurred.22 In the same report,22 the analogous phenomenon was also observed for allylic-BODIPY photocages.

Experiments with prodrug 1 in an acetonitrile dominated solution (acetonitrile:water, 9 : 1 v:v) in dark conditions did not produce detectable amounts of coumarin alcohol 8 or regioisomer 17 (via analytical HPLC chromatogram retention times). Instead, alongside production of chromenopyrazole 2 after 48 hours (Figure 2F), a small amount of a compound with a molecular mass ([M+H]+ 349.2) that was 18 less than coumarin alcohol 8 was detected that had a large (4 min) difference in retention time compared to 8 (Supplementary Figure S5B, retention time of 18 21.8 min), which could potentially be the dehydrated ene coumarin 18 (Figure 3), since equivalent vinyl coumarins have been reported to form in small amounts in previous reports.19, 23 In dark conditions, the carbonate of 1 is likely to undergo hydrolysis, but the mechanism whereby 18 potentially forms rather than alcohol 8 is not clear. It seems unlikely that 18 is an artifact of HPLC and/or mass spectrometry analysis since 8 was reliably detected in other experiments, even after extended experiment times, without ‘conversion’ to 18 as an artifact of measurement. The same coumarin product, possibly 18, was also detected in a small amount (approximately 5 % of all coumarin products) in prodrug stability in bench light experiments (Supplementary Figure S5A).

Radioligand competition binding assays were used to measure the CB2R affinity of known CB2R agonist 214 versus prodrug 1. The affinity of 2 (pKi 8.2±0.06) was consistent with previous reports.16 The affinity of prodrug 1, measured in a radioligand binding assay carried out under conditions to exclude light as much as practically possible, was approximately 1.6 log units lower (pKi 6.6±0.06; p<0.0001) than 2. Since RP-HPLC stability experiments (Figure S6) that mimicked the radioligand binding experiment did not show a decrease in 1 or an appearance of any chromenopyrazole 2 after 2 hours it is possible the detectable binding affinity measured for 1 is in fact due to engagement as an unmodified entity. To further probe this possibility, both 1 and 2 were docked into the cryo-EM structure of human CB2R (PDB ID: 6KPF)24 bound with an agonist (AM12033) (Figure 4). Chromenopyrazole 2 binds in a similar orientation to the experimental pose of the agonist AM12033, particularly the phenolic ring and alkyl tail (Figure 4B). The best three solutions for both 2 and AM12033 (docked as a positive control) were all within 1.5 Å heavy atom RMSD. Prodrug 1 does not dock into the binding site in the top seven poses (Figure 4C), but does in the lowest three ranked poses with negative (bad) docking scores (ChemPLP). The docking score of the highest ranked prodrug 1 pose (58.38) (not in the binding site) was approximately half of the values returned for AM12033 (101.99) and 2 (108.58). It therefore remains a possibility that 1 could engage with CB2R resulting in some measurable binding affinity, albeit it significantly worse than 2.

Details are in the caption following the image

A) The agonist (AM12033, magenta) was docked into the cryo-EM structure of the human CB2R (green ribbon) as a positive control, revealing a similar binding pose to the experimental AM12033 pose (orange). B) Chromenopyrazole 2 (yellow) binds in the same spatial region as AM12033 (orange). C) Prodrug 2 (purple) docked outside of the binding site, AM12033 (orange). The top ranked binding poses are shown. Tyr25 and Trp194 shown for perspective.

Although measurable CB2R affinity was detected for both 2 and 1, target engagement is expected to be improved considerably upon photo-cleavage of 1. This affinity differential is promising in terms of the potential for prodrug 1 to be utilized at a concentration that has little CB2R engagement in the prodrug form, but markedly enhanced engagement following photo-cleavage. This property may be exploited in vivo via tightly controlled spatiotemporal drug release to selectively increase CB2R activation in localized body regions, thereby reducing off-target effects. For example, photocleavable anticancer prodrugs, including those utilizing a coumarin caging moiety, have been developed to target cytotoxic effects to the tumor microenvironment, as reviewed in Dunkel et al.25 In addition to external light sources, prodrug 1 has the potential to be easily adapted into a prodrug that is triggered via light emitted from a luciferase. This in turn has scope to investigate unique CB2R signaling profiles, for example, with judiciously expressed luciferase, differential signaling between cell surface and intracellular CB2R2, 26 could be investigated. In order for bioluminescence to cleave prodrugs such as 1 a ‘template-approach,’ for example as reported by Lindberg13a and Chang13b et al, is likely required to bring the prodrug and source of light close enough to give light-mediated bond cleavage. As outlined in the introduction and exemplified by the results for model prodrug 12, derivatization from the second ‘branch point’ of our prodrug scaffold is well tolerated and can be used to introduce a template for BRET.13 For future applications, the stability of prodrug 1 should also be considered. Based on our prodrug 1 stability experiments (Figure 2F, Supplementary Figure S6) assays/experiments conducted with the exclusion of light for 48 hours or under ambient light for a much shorter time (1–2 hours) should result in little breakdown of 1 until purposefully triggered by approximately 450 nm light. Along with 2, photo-cleavage of 1 also produces (in aqueous conditions) coumarin 8, and the possibility that 8 itself possess biological activity must not be discounted in future uses of prodrugs such as 1. The biological activity of the coumarin ‘byproduct’ from prodrug photo-cleavage is often not widely or thoroughly investigated, however some studies do address this and have shown, for example, that DEACM has a good safety profile and is suitable for use as a photocage in vivo.27 4-Ethylacetamide coumarins have been reported as G protein-coupled melatonin receptors ligands,28 and there are reports of 3- but not 4-substituted coumarins as CB2R ligands.29


A photo-activated CB2R agonist prodrug 1 and model prodrugs 11 and 12 were designed, synthesised and evaluated. All three prodrugs showed promising stability and light-mediated cleavage profiles, in particular fast cleavage under 450–455 nm LED light. Along with chromenopyrazole release, different coumarins were detected as products in varying ratios depending on the conditions of prodrug cleavage. The affinity of prodrug 1 was determined to be significantly lower than that of the corresponding released chromenopyrazole 2. The photo-activated CB2R agonist prodrug 1 will likely prove a very useful prodrug to study CB2R in a triggered and controlled way using artificial light, and is of the appropriate photolysis wavelength to allow to future application to BRET systems with a luciferase.

Experimental Section

Chemical Methods

Chemicals were purchased from Sigma Aldrich, AK Scientific or Merck and were used without further purification. Thin layer chromatography (TLC) was carried out on silica gel plates 60 F254 and visualized under UV light at 254 nm and 356 nm, and with potassium permanganate dip. Flash column chromatography was performed using 40–63 μm silica. Proton (1H) and carbon (13C) nuclear magnetic resonance (NMR) spectroscopy was carried out using a 400 MHz Varian NMR spectrometer or a 600 MHz JEOL NMR spectrometer. Chemical shifts are listed on the δ scale in part(s) per million (ppm), referenced to CDCl3 (1H NMR: δ 7.26, 13C NMR: δ 77.16) or DMSO-d6 (1H NMR: δ 2.50, 13C NMR: δ 39.52) with residual solvent as the internal standard and coupling constants (J) recorded in hertz (Hz). Note – not all magnetically non-equivalent carbons were observed in 13C NMR spectrum for all compounds. Signal multiplicities are assigned as: s, singlet; d, doublet; t, triplet; q, quartet; dd, doublet of doublets; br, broad; or m, multiplet. High resolution electrospray ionisation mass spectra (HRMS-ESI) were obtained on a microTOFQmass spectrometer.

10-[(Furan-2-yl)methoxy]decanoic acid (4)

(Furan-2-yl)methanol (98 mg, 1 mmol), NaI (15 mg, 0.1 mmol) and 10-bromodecanoic acid (251 mg, 1 mmol) in DMF (6 mL) were cooled to 0 °C, and NaH (100 mg, 2.5 mmol) was slowly added. After 20 min the reaction mixture was heated to 60 °C and stirred for 4 h. The solvent was removed by evaporation under reduced pressure and the residue was partitioned between 1 M HCl (20 mL) and Et2O (20 mL). The aqueous phase was extracted with Et2O (2×20 mL), the combined organic layers dried over MgSO4 and evaporated to dryness to give a residue (239 mg) that used without further purification.

N-{2-[(tert-butyldimethylsilyl)oxy]-2-[2-(dicyanomethylidene)-7-(diethylamino)-2H-chromen-4-yl]ethyl}acetamide (6)

To a solution of 520 (racemic mixture) (105 mg, 0.24 mmol) in CH2Cl2 (4 mL) was added Ac2O (70 μL, 0.7 mmol) and Et3N (100 μL, 0.7 mmol) and the reaction mixture was stirred at rt for 16 h. The solvent was removed by evaporation under reduced pressure, the residue was partitioned between EtOAc (15 mL) and water (15 mL), the EtOAc layer was washed with brine (10 mL), dried over MgSO4 and evaporated to dryness The residue was purified by silica gel column chromatography eluting with hexane:EtOAc (1 : 1), affording 6 (88 mg, 0.18 mmol, 76 %) as an orange solid. 1H NMR (400 MHz, CDCl3) δ 0.01 (s, 3H), 0.07 (s, 3H), 0.96 (s, 9H), 1.23 (t, J=7.2 Hz, 6H), 2.05 (s, 3H), 2.86–2.93 (m, 1H), 3.44 (q, J=7.2 Hz, 4H), 3.81–3.87 (m, 1H), 5.21 (d, J=8.4 Hz, 1H), 6.04 (t, J=6.0 Hz, 1H), 6.61 (d, J=2.4 Hz, 1H), 6.77 (dd, Ja=8.8 Hz, Jb=2.4 Hz, 1H), 6.93 (s, 1H), 7.92 (d, J=8.8 Hz, 1H). 13C NMR (101 MHz, CDCl3) δ −5.1, −4.8, 12.4, 18.1, 23.1, 25.7, 44.9, 47.4, 54.5, 69.1, 97.2, 105.2, 107.0, 111.0, 114.1, 114.7, 125.8, 151.6, 153.2, 155.0, 170.9, 172.1. HRMS calculated for C26H36N4NaO3Si+ [M+Na]+, 503.2449; found, 503.2429.

N-{2-[(tert-butyldimethylsilyl)oxy]-2-[2-(dicyanomethylidene)-7-(diethylamino)-2H-chromen-4-yl]ethyl}-10-[(furan-2-yl)methoxy]decanamide (7)

To a solution of 4 (crude mass 239 mg) in THF (12 mL) was added N,N-diisopropylethylamine (0.27 mL, 1.47 mmol), HBTU (338 mg, 1.06 mmol) and 5 (350 mg, 0.8 mmol). The reaction mixture was stirred overnight at rt and then quenched with saturated aq. NaHCO3 (25 mL). The mixture was extracted with EtOAc (3×15 mL), the organic layers were combined, washed with water (20 mL) and brine (20 mL), dried over MgSO4 and evaporated under reduced pressure. The crude residue was purified by silica gel chromatography eluting with hexane:EtOAc (6 : 1), affording 7 (320 mg, 0.46 mmol, 58 % over two steps from 3) as a yellow solid. 1H NMR (400 MHz, CDCl3) δ −0.04 (s, 3H), 0.03 (s, 3H), 0.92 (s, 9H), 1.19 (t, J=7.2 Hz, 6H), 1.25 (m, 10H), 1.50–1.55 (m, 2H), 1.58–1.65 (m, 2H), 2.19 (t, J=8 Hz, 2H), 2.83–2.90 (m, 1H), 3.38–3.43 (m, 6H), 3.78–3.83 (m, 1H), 4.39 (s, 2H), 5.18 (d, J=8 Hz, 1H), 6.02 (t, J=6 Hz, 1H), 6.26 (d, J=3.2 Hz, 1H), 6.29–6.30 (m, 1H), 6.55 (d, J=2.4 Hz, 1H), 6.72 (dd, Ja=9.2 Hz, Jb=2.4 Hz,1H), 6.89 (s, 1H), 7.36 (s, 1H), 7.90 (d, J=9.2 Hz, 1H). 13C NMR (101 MHz, CDCl3) δ −5.1, −4.8, 12.4, 18.1, 25.3, 25.7, 26.0, 29.2, 29.3, 29.3, 29.3, 29.6, 36.4, 44.9, 47.3, 54.4, 64.7, 69.2, 70.4, 97.2, 105.2, 107.1, 108.9, 110.1, 111.0, 114.1, 114.7, 125.8, 142.6, 151.5, 152.0, 153.3, 155.0, 172.0, 173.9. HRMS calculated for C39H56N4NaO5Si+ [M+Na]+, 711.3912; found, 711.3902.

N-{2-[2-(dicyanomethylidene)-7-(diethylamino)-2H-chromen-4-yl]-2-hydroxyethyl}acetamide (8)

To a solution of 6 (230 mg, 0.48 mmol) at 0 °C in THF (15 ml) was added a solution of tetrabutylammonium fluoride (TBAF) in THF (0.72 mmol, 1 M, 0.72 mL THF). The reaction mixture was stirred at 0 °C for 10 min. The reaction was quenched with water (12 mL). The aqueous layer was extracted with EtOAc (3×15 mL). The organic layers were combined, washed with water (20 mL), brine (20 mL), dried over MgSO4 and evaporated under reduced pressure. The crude residue was purified by silica gel chromatography eluting with EtOAc:MeOH (99 : 1), affording 8 (171 mg, 0.47 mmol, 98 %) as a dark orange solid. 1H NMR (400 MHz, DMSO-d6) δ 1.15 (t, J=7.1 Hz, 6H, 2×CH2CH3), 1.82 (s, 3H, NHCOCH3), 2.97–3.03 (m, 1H, CHCH2NH, HA), 3.43–3.48 (m, 1H, CHCH2NH, HB), 3.49 (q, J=7.2 Hz, 4H, 2×CH2CH3), 5.07–5.11 (m, 1H, CHCH2NH, Ha), 5.98 (d, J=4.5 Hz, 1H, OH), 6.65 (d, J=2.6 Hz, 1H, coumarin H, He), 6.81 (s, 1H, coumarin H, Hb), 6.90 (dd, Ja=9.3 Hz, Jb=2.6 Hz, 1H, coumarin H, Hd), 7.92 (d, J=9.3 Hz, 1H, coumarin H, Hc), 8.22 (t, J=5.4 Hz, 1H, NH). 13C NMR (101 MHz, DMSO-d6) δ 12.3, 22.5, 44.2, 45.9, 51.3, 67.2, 96.3, 104.0, 107.3, 111.2, 114.3, 115.3, 126.3, 151.4, 154.7, 156.1, 170.1, 171.7. HRMS calculated for C20H22N4NaO3+ [M+Na]+, 389.1584; found, 389.1582.

N-{2-[2-(dicyanomethylidene)-7-(diethylamino)-2H-chromen-4-yl]-2-hydroxyethyl}-10-[(furan-2-yl)methoxy]decanamide (9)

According to the procedure described for 8, 7 (253 mg, 0.37 mmol) gave a crude residue that was purified by silica gel chromatography eluting with hexane:EtOAc (1 : 1), affording 9 (124 mg, 0.22 mmol, 59 %) as a dark orange solid. 1H NMR (400 MHz, CDCl3) δ 1.21 (t, J=7.2 Hz, 6H), 1.25 (m, 10H), 1.50–1.65 (m, 4H), 2.23 (t, J=7.6 Hz, 2H), 3.01–3.08 (m, 1H), 3.40–3.45 (m, 6H), 3.83–3.89 (m, 1H), 4.40 (s, 2H), 4.51 (br, 1H), 5.18 (d, J=7.2 Hz, 1H), 6.27–6.28 (m, 1H), 6.30–6.32 (m, 1H), 6.53 (m, 2H), 6.72 (dd, Ja=9.2 Hz, Jb=2.4 Hz,1H), 6.88 (s, 1H), 7.36 (d, J=1.2 Hz, 1H), 7.77 (d, J=9.2 Hz, 1H). 13C NMR (101 MHz, CDCl3) δ 12.4, 25.5, 25.9, 29.1, 29.1, 29.5, 36.3, 44.9, 46.5, 53.8, 64.6, 69.2, 70.3, 97.0, 104.8, 107.1, 109.0, 110.2, 111.0, 114.1, 115.1, 125.7, 142.6, 151.5, 151.9, 153.3, 154.9, 172.0, 175.2. HRMS calculated for C33H42N4NaO5+[M+Na]+, 597.3047; found, 597.3049.

Methyl 4-{9-[({1-[2-(dicyanomethylidene)-7-(diethylamino) -2H-chromen-4-yl]-2-acetamidoethoxy}carbonyl)oxy]-4,4-dimethyl-7-(2-methyloctan-2-yl)-1H,4H-chromeno[4,3-c]pyrazol-1-yl}benzoate (11)

To a solution 1016 (150 mg, 0.31 mmol) in CH2Cl2 (8 mL) was added N,N’-disuccinimidyl carbonate (DSC) (119 mg, 0.47 mmol), followed by the addition of N,N-dimethylpyridin-4-amine (DMAP) (57 mg, 0.47 mmol) and Et3N (95 μL, 0.67 mmol). The reaction was stirred at rt and monitored by TLC until all 10 was reacted. Then, 8 (114 mg, 0.31 mmol) in CH2Cl2 (3 mL) was added, followed by DMAP (57 mg, 0.47 mmol) and Et3N (95 μL, 0.67 mmol). The reaction was stirred overnight. The solvent was removed under reduced pressure and the crude residue was purified by silica gel chromatography eluting with hexane:EtOAc (2 : 1), affording 11 (35 mg, 0.04 mmol, 13 %) as an orange solid. 1H NMR (400 MHz, CDCl3) δ 0.83 (t, J=6.4 Hz, 3H), 1.05–1.10 (m, 2H), 1.19–1.27 (m, 18H), 1.50 (s, 3H), 1.54–1.58 (m, 2H), 1.71 (s, 3H), 2.04 (s, 3H), 3.02–3.08 (m, 1H), 3.43 (q, J=7.2 Hz, 4H), 3.71–3.77 (m, 1H), 3.85 (s, 3H), 5.80 (d, J=8.0 Hz, 1H), 6.07 (t, J=6.0 Hz, 1H), 6.35 (s, 1H), 6.56 (d, J=2.4 Hz, 1H), 6.71 (dd, Ja=8.8 Hz, Jb=2.4 Hz, 1H), 6.79 (s, 1H), 7.00 (s, 1H), 7.55–7.60 (m, 3H), 7.73 (d, J=8.8 Hz, 1H), 8.07 (d, J=8.4 Hz, 2H). 13C NMR (101 MHz, CDCl3) δ 12.4, 14.0, 22.6, 22.8, 24.5, 26.9, 27.1, 28.3, 28. 5, 29.8, 31. 7, 38.2, 44.2, 45.0, 52.4, 53.4, 55.3, 61.5, 74.0, 77.2, 97.3, 104.1, 106.6, 106.9, 111.1, 113.2, 113.6, 114.6, 123.1, 125. 7, 126.8, 128.7, 130.3, 135.5, 144.4, 145.4, 147.2, 151.4, 151.8, 153.6, 153.8, 155.1, 166.8, 171.1, 171.5. HRMS calculated for C50H56N6NaO8+[M+Na]+, 891.4052; found, 891.4018.

Methyl 4-{9-[({1-[2-(dicyanomethylidene)-7-(diethylamino) -2H-chromen-4-yl]-2-{10-[(furan-2-yl)methoxy]decanamido}ethoxy}carbonyl)oxy]-4,4-dimethyl-7-(2-methyloctan-2-yl)-1H,4H-chromeno[4,3-c]pyrazol-1-yl}benzoate (12)

According to the procedure described for 11, 10 (121 mg, 0.25 mmol), DSC (96 mg, 0.38 mmol), DMAP (2×46 mg, 0.76 mmol), Et3N (2×70 μL, 1 mmol) and 9 (124 mg, 0.22 mmol) gave a crude residue that was purified by silica gel chromatography with hexane:EtOAc:CH2Cl2 (1 : 1 : 1) to give 12 (21 mg, 0.02 mmol, 9 %) as an orange solid. 1H NMR (400 MHz, CDCl3) δ 0.83 (t, J=6.8 Hz, 3H), 1.06–1.09 (m, 2H), 1.20–1.31 (m, 28H), 1.53 (s, 3H), 1.56–1.61 (m, 6H), 1.71 (s, 3H), 2.19–2.23 (m, 2H), 3.04–3.12 (m, 1H), 3.40–3.47 (m, 6H), 3.72–3.76 (m, 1H), 3.86 (s, 3H), 4.43 (s, 2H), 5.79 (d, J=8.0 Hz, 1H), 5.94 (t, J=6.0 Hz, 1H), 6.29–6.33 (m, 2H), 6.36 (s, 1H), 6.57 (d, J=2.4 Hz, 1H), 6.72–6.75 (m, 2H), 6.79 (d, J=1.6 Hz, 1H), 6.70 (d, J=1.6 Hz, 1H), 7.39 (s, 1H), 7.55–7.60 (m, 3H), 7.74 (d, J=9.2 Hz, 1H), 8.07 (d, J=8.8 Hz, 2H). 13C NMR (101 MHz, CDCl3) δ 12.4, 14.0, 22.6, 24.5, 25.4, 25.5, 26.1, 27.0, 27.1, 28.4, 28.4, 29.3, 29.4, 29.6, 29.8, 30.9, 31. 7, 36.2, 38.2, 43.7, 44.2, 45.0, 52.4, 64.7, 70.4, 74.0, 77.2, 97.3, 104.1, 106.6, 108.9, 110.2, 111.2, 113.1, 114.5, 114.5, 123.1, 125.7, 126.8, 128.7, 130.4, 135.5, 142.6, 144.4, 145.4, 147.3, 151.4, 151.8, 152.1, 153.5, 153.7, 155.1, 166.7, 171.5, 174.2. HRMS calculated for C63H76N6NaO10+[M+Na]+, 1099.5515; found, 1099.5500.

N-{2-[2-(dicyanomethylidene)-7-(diethylamino)-2H-chromen-4-yl]-2-{[({1-[4-(acetamidomethyl)phenyl]-4,4-dimethyl-7-(2-methyloctan-2-yl)-1H,4H-chromeno[4,3-c]pyrazol-9-yl}oxy)carbonyl]oxy}ethyl}acetamide (1)

According to the procedure described for 11, 2 (60 mg, 0.12 mmol), DSC (48 mg, 0.18 mmol), DMAP (2×23 mg, 0.36 mmol), Et3N (2×50 μL, 0.72 mmol) and 8 (46 mg, 0.12 mmol) gave a crude residue that was purified by silica gel chromatography with MeOH:CH2Cl2 (1 : 20) to give 1 (15 mg, 0.017 mmol, 14 %) as an orange solid. 1H NMR (400 MHz, CDCl3) δ 0.82 (t, J=6.4 Hz, 3H), 1.03–1.10 (m, 2H), 1.14–1.30 (m, 18H), 1.52 (s, 3H), 1.54–1.58 (m, 2H), 1.71 (s, 3H), 1.95 (s, 3H), 2.13 (s, 3H), 2.96-3.03 (m, 1H), 3.42 (q, J=7.2 Hz, 4H), 3.74–3.79 (m, 1H), 4.40–4.57 (m, 2H), 5.92 (d, J=8 Hz, 1H), 6.44 (s, 1H), 6.54 (d, J=2.4 Hz, 1H), 6.70–6.73 (m, 2H), 6.82–6.90 (m, 2H), 6.99 (d, J=2 Hz, 1H), 7.33 (d, J=8.4 Hz, 2H), 7.41 (d, J=8.4 Hz, 2H), 7.58 (s, 1H), 7.92 (d, J=9.6 Hz, 1H). 13C NMR (150 MHz, CDCl3) δ 12.6, 14.2, 22.7, 22.9, 23.50, 24.7, 27.3, 28.7, 30.0, 31.9, 38.3, 43.2, 44.4, 44.7, 45.2, 54.9, 74.1, 77.6, 97.3, 103.7, 107.0, 107.7, 111.5, 113.8, 114.8, 115.2, 123.2, 126.3, 126.5, 127.6, 131.5, 135.1, 138.2, 140.5, 144.8, 148.8, 152.2, 153.8, 154.0, 155.3, 171.3, 171.8, 172.0. HRMS calculated for C51H59N7NaO7+ [M+Na]+, 904.4368; found, 904.4358.

2-Amino-1-[2-(dicyanomethylidene)-7-(diethylamino)-2H-chromen-4-yl]ethyl acetate (17)

A solution of 1 (3.3 mg, 0.004 mmol) in acetonitrile:water (20 mL, 9 : 1 v:v) was exposed to 450–455 nm LED light for 15 min as otherwise described in the prodrug light-mediated cleavage experimental procedure. The solvent was removed under reduced pressure and the residue was purified by semi-preparative RP-HPLC to afford 17 (0.8 mg, 0.002 mmol, 50 % from 1) as a red solid. 1H NMR (600 MHz, DMSO-d6) δ 1.15 (t, J=7.0 Hz, 6H, 2×CH2CH3), 2.18 (s, 3H, OCOCH3), 3.53 (q, J=7.4 Hz, 4H, 2×CH2CH3), 6.23–6.26 (m, 1H, CHCH2NH2, Ha’), 6.59 (s, 1H, coumarin H, Hb’), 6.71 (d, J=2.7 Hz, 1H, coumarin H, He’), 6.92 (dd, Ja=8.8 Hz, Jb=2.2 Hz, 1H, coumarin H, Hd’), 7.77 (d, J=8.8 Hz, 1H, coumarin H, Hc’), 8.04 (br, 2H, NH2).The CHCH2NH2 was underneath the large water peak, from the COSY and HSQC spectra of 17 these protons were both at 3.35 ppm. HRMS calculated for C20H22N4NaO3+[M+Na]+, 389.1584; found, 389.1588.

Prodrug light-mediated cleavage experiments

All experiments were conducted with a freshly prepared solution of prodrug in the requisite solvent mixture. A glass vial containing the prodrug solution (20 μM, typically 1 mL) was placed in an EvoluChemTM PhotoRedOx Box, placed over a magnetic stirrer and stirred at rt (approx. 20 °C). LED lamps used from EvoluChemTM were (380 nm: 8 mW/cm2, 450–455 nm: 34 mW/cm2, 525–530 nm: 10 mW/cm2) as reported by light measurements conducted by HepatoChem, Inc. An aliquot (50 μL) was taken from the vial for time=0 sec immediately prior to turning the LED lamp on and starting the experiment. Further aliquots (50 μL) were taken from the glass vial at different time points and analyzed by analytical RP-HPLC and LC–MS (either immediately or wrapped with aluminum foil and stored at −20 °C prior to analysis). Each experiment was conducted independently in triplicate. The percentage of cleaved prodrug was calculated by the following method: The peak corresponding to the prodrug was integrated at all of the assay timepoints, including at time zero. The peak integration at time zero was normalized to 100 % prodrug, and therefore at subsequent timepoints the percentage prodrug peak area decrease was calculated, with total disappearance of the peak corresponding to the prodrug being defined as 100 % prodrug cleavage. The error bars in Figure 2 represent the standard deviation of data from triplicate experiments and was calculated by the function in ORIGINPRO® and the default exponential line fitting was applied for all the data in Figure 2. Positive conformation of the chromenopyrazole release was confirmed by LC–MS and by retention time on analytical RP-HPLC using a standard sample of the released chromenopyrazole prepared synthetically.

Prodrug stability experiments without LED lamps

Ambient light – the prodrug (20 μM, typically 1 mL solution) was freshly prepared in the requisite solvent mixture and the vial was placed at rt in the fume hood with the fume hood light off without blocking out light from the laboratory ceiling lights (left on overnight) and ambient sunlight from uncovered laboratory windows. Dark stability conditions – the vial containing the prodrug solution (20 μM, typically 1 mL solution) was at rt and covered with foil during the whole experiment and aliquots were removed quickly with care to exclude light as much as possible. HEPES buffer conditions – prodrug 1 (10 μM, 300 μL) in 1 : 9 v:v DMSO:HEPES buffer (50 mM HEPES pH 7.4, 1 mM MgCl2, 1 mM CaCl2, 2°mg/mL fatty acid-free bovine serum albumin [BSA; ICPbio]) and placed in a multi-well plate and incubated inside a CLARIOstar® Plus plate reader at 30 °C in the dark for 2 h. All samples were analyzed as described for the prodrug light-mediated cleavage experiments.

Radioligand Binding Assay

Preparation of human CB2R (hCB2R)-expressing membranes from a HEK Flp-in (Invitrogen, R75007) HA-3TCS-hCB2 63R cell line and the radioligand binding assay were both adapted from a protocol described previously.30 Briefly, serial dilutions of displacers (2 and 1; final concentrations 10−10 M to 10−6 M), radioligand ([3H]-CP55,940 (5-(1,1-dimethylheptyl)-2-[5-hydroxy-2-(3-hydroxypropyl)cyclohexyl]phenol), final concentration 0.75 nM, PerkinElmer), and hCB2R-expressing membranes (final concentration 2.5 μg/point) were separately prepared in binding buffer (50 mM HEPES pH 7.4, 1 mM MgCl2, 1 mM CaCl2, 2°mg/mL fatty acid-free bovine serum albumin [BSA; ICPbio]). All manipulations of samples containing JZ-L-F were undertaken in as minimal ambient lighting as practically possible. JZ-L-F stocks and dilutions were prepared in opaque tubes, and assay plates were wrapped in foil between manipulations and for incubations. A CP55,940 condition was used to define maximum displacement (final concentration 10−6 M; Cayman). Vehicle (DMSO) was kept equivalent across all test conditions. All conditions were carried out in technical duplicate with three independent experiments for each condition.

All dilutions were dispensed into a 96-well polypropylene V-bottom plate (Gene Era Biotech). The dispensing order was radioligand (50 μL), displacer (50 μL), then hCB2R-expressing membranes (100 μL). The plate was then sealed, tap-mixed, and incubated for 1 hour at 30 °C. 50 μL of 0.1 % PEI (Sigma-Aldrich) was added to each well of a 96-well GF/C harvest plate (PerkinElmer) and incubated at room temperature for 1 hour. After the incubation, the harvest plate was washed with ice-cold wash buffer (50 mM HEPES pH 7.4, 500 mM NaCl, 1 mg/mL BSA). Washing was aided by a vacuum manifold set to ∼5 mmHg. Subsequently, the contents of the V-bottom plate were transferred to corresponding wells in the harvest plate. The V-bottom plate was then washed, and a second transfer was repeated. An additional three washes to the harvest plate were then conducted. The plate was then dried for ∼24 hours at 24 °C. The bottom of the harvest plate was then sealed, 50° μL Irgasafe Plus scintillation fluid (PerkinElmer) added per well, followed by sealing the top of the plate. Controls to verify that radioligand depletion was less than 10 % were also measured. Plates were read in a Wallac MicroBeta® TriLux Liquid Scintillation Counter (PerkinElmer) with a read time of 2 min/well.

GraphPad Prism (v8.0; GraphPad Software) was used to fit nonlinear regression curves (One site – Fit Ki) to the raw binding data and determine the pKi (negative log Ki) of each displacer. The Kd for the radioligand, [3H]-CP55,940, was set to 2.7 nM. The 10−6 M CP55,940 condition was used to define the bottom of the curve. Mean and SEM of pKi values from the three independent experiments were calculated. An unpaired two-tailed t-test was conducted in GraphPad Prism to determine whether a statistical difference between the affinities for 2 and 1 was present.

Molecular Modeling and ligand docking

Docking studies were carried out with the 2 and 1 using GOLD v2022.3.0 (CCDC). The cryo-EM structure of human CB2R (PDB ID: 6KPF, resolution 2.90 Å)24 bound with an agonist (AM12033) was used. The agonist AM12033 was also docked as a positive control and closely matched the AM12033 experimental pose. The ChemPLP scoring function was used with and a binding site radius of 15 Å centred around CZ of Phe183 in CB2R. Prodrug 2 and chromenopyrazole 1 were drawn in 3 dimensions and minimised (default forcefield) using Avogadro2[31] v1.97.0.


This work was supported by a University of Otago Research Grant. Jiazhen Yin was supported by a University of Otago Doctoral Scholarship. Raahul Sharma was supported by a University of Auckland Summer Research Scholarship. Natasha Grimsey was supported by a Health Research Council (NZ) Sir Charles Hercus Health Research Fellowship. Open Access publishing facilitated by University of Otago, as part of the Wiley - University of Otago agreement via the Council of Australian University Librarians.

    Conflict of interest

    The authors declare no conflict of interest.

    Data Availability Statement

    The data that support the findings of this study are available in the supplementary material of this article.