Forensic Toxicology

, Volume 36, Issue 2, pp 385–403 | Cite as

Pharmacological evaluation of new constituents of “Spice”: synthetic cannabinoids based on indole, indazole, benzimidazole and carbazole scaffolds

  • Clara T. Schoeder
  • Cornelius Hess
  • Burkhard Madea
  • Jens Meiler
  • Christa E. Müller
Open Access
Original Article



In the present study we characterized a series of synthetic cannabinoids containing various heterocyclic scaffolds that had been identified as constituents of “Spice”, a preparation sold on the illicit drug market. All compounds were further investigated as potential ligands of the orphan receptors GPR18 and GPR55 that interact with some cannabinoids.


The compounds were studied in radioligand binding assays to determine their affinity for human cannabinoid CB1 and CB2 receptors expressed in CHO cells, and in cAMP accumulation assays to study their functionality.


Structure-activity relationships were analyzed. The most potent CB1 receptor agonist of the present series MDMB-FUBINACA (12) (Ki = 98.5 pM) was docked into the human CB1 receptor structure, and a plausible binding mode was identified showing high similarity with that of the co-crystallized THC derivatives. MDMB-CHMCZCA (41) displayed a unique profile acting as a full agonist at the CB1 receptor subtype, but blocking the CB2 receptor completely. Only a few weakly potent antagonists of GPR18 and GPR55 were identified, and thus all compounds showed high CB receptor selectivity, mostly interacting with both subtypes, CB1 and CB2.


These results will be useful to assess the compounds’ toxicological risks and to guide legislation. Further studies on 41 are warranted.


Pharmacological evaluation of new synthetic cannabinoids Affinities for CB1 and CB2 receptors β-Arrestin assay at GPR18 and GPR55 cAMP accumulation assay Benzimidazole and carbazole Structure-activity relationships 


A challenging issue for forensic toxicologists and law makers is how to effectively respond to the constantly changing new psychoactive substances on the illicit drug market [1]. Among these, synthetic cannabinoids feature prominently [2, 3]. Between 2008, when so-called “Spice” products [4] containing synthetic cannabinoids began to appear on the drug market, and 2016, 169 new synthetic cannabinoids were confiscated and identified [2]. Most of them were discovered as powders, often in bulk amounts, while others were found in preparations of plant materials, e.g., minced herbs, onto which solutions of the cannabinoids had been sprayed [5]. These substances have been shown to bind to and in many cases activate cannabinoid (CB) receptors. CB receptors are divided into two subtypes, CB1 and CB2, which belong to the large family of rhodopsin-like class A G protein-coupled receptors (GPCRs) [6]. Both CB receptor subtypes are coupled to Gi proteins including a reduction in intracellular cAMP levels. The main psychoactive effects of cannabinoids are mediated by the CB1 receptor, which is predominantly expressed in the central nervous system [7], while CB2 receptor expression in the brain is restricted to microglial cells [8, 9]. CB2 receptors are highly expressed in the immune system, for example in tonsils and spleen [10, 11]. Activation of the CB2 receptor is considered as a new therapeutic option for the treatment of inflammatory diseases and pain [12, 13].

The plant-derived partial CB1 and CB2 receptor agonist Δ9-tetrahydrocannabinol (Δ9-THC, 1, Fig. 1) is used in therapy to target muscle spasms, nausea and cachexia [14]. The synthetic compound CP55,940 (2, Fig. 1) represents a potent full agonist at both receptor subtypes. A CB1 receptor antagonist, rimonabant, had been approved for the treatment of obesity but was later withdrawn from the market due to side effects resulting in depression and an increased suicide rate [15].
Fig. 1

Standard cannabinoid CB1/CB2 receptor agonists [12, 22]

The prevalence for the use of illegal psychoactive substances in Europe by 15–16 year-old teenagers was estimated in 2015 to be about 4% [5]. Synthetic CB1 receptor agonists are abused as an alternative to natural marijuana due to their psychoactive and analgesic effects. For synthetic cannabinoids more and more severe side effects and intoxications are reported; they are predominantly neurologic symptoms, but acute organ toxicity has also been observed [16]. In the USA, the principle of enumeration is used to restrict newly discovered synthetic cannabinoids, and every single synthetic cannabinoid has to be individually listed by name in the US List of Schedule I drugs [17]. In Germany new synthetic cannabinoids are legally controlled since November 2016 when the “Neue-Psychoaktive-Stoffe-Gesetz” (NpSG, New Psychoactive Substances Act) was adopted in [18]. Similar regulations exist in Austria and Switzerland [19, 20]. All corresponding compounds, the chemical structures of which are represented by a general formula in the statute with known structure-activity relationships (SARs), were restricted. Newly discovered SARs of synthetic cannabinoids will, therefore, provide a basis for future amendments. However, in many cases, only limited information is available regarding the activity of new substances. Both the affinity of a drug for its receptor and its ability to produce an agonistic response are important features, and these should be determined according to a compound’s chemical structure. For important classes of synthetic cannabinoids, at least four structural components, which have firstly been described by Huffman et al. and were later refined by the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA), are of importance (see Fig. 2 [3]): (1) a heterocyclic core consisting of indole or indazole with different substitutions; (2) a linker, e.g., an ester, amide or ketone; (3) a bulky lipophilic residue (R1), e.g., a heterocyclic or aryl substituent, but in newer synthetic cannabinoids a lipophilic amino acid can also be found; and (4) a residue (R2) which is a hydrophobic “side chain” attached to the nitrogen atom of the indole or the indazole ring system [21, 22]. The compound JWH-018 (3, see Fig. 1), a potent CB1 and CB2 receptor agonist, displays the basic features of this compound class and was one of the first synthetic cannabinoids identified in herbal blends for abuse [23, 24]. The common features of known synthetic cannabinoids are depicted in Fig. 2.
Fig. 2

Common structural features of synthetic cannabinoids. The figure was adopted from the European Monitoring Centre for Drugs and Drug Addiction (EMCDDA) [3] and modified

In a previous study [25], we had determined the pharmacological properties of 48 synthetic cannabinoids collected by the Institute of Forensic Medicine of the University of Bonn. In the present study, we investigated the affinities and functional properties of a new series of 42 synthetic cannabinoids, 16 of which have not been reported as cannabinoid receptor ligands before. The investigated set of compounds comprises four different core structures. The first three groups (A, B, C, see Table 1) represent differently substituted indoles and indazoles, which are structurally derived from the synthetic cannabinoids previously introduced by Huffman et al. and are widely distributed in illicitly sold "Spice" products. In the current study we investigated compounds with l-valinamide (AB)/l-tert-leucinamide (ADB or MAB), methyl-3,3-dimethylbutanoate (MDMB), methyl-3-methylbutanoate (MMB), and 2-methyl-2-phenylpropyl (cumyl) moieties as substituents in the R1 position. Further classes of compounds consist of carbazoles (E), substituted in position 3, and benzimidazole derivatives (F).
Table 1

Affinities of the investigated indoles, indazoles and carbazoles at the cannabinoid CB1 and CB2 receptors determined in radioligand binding assays

aFluorometric imaging plate reader membrane potential assay system from Molecular Devices (Sunnyvale, CA, USA)

bRadioligand binding study versus 0.5 nM [3H]CP55,940

cRadioligand binding study versus 0.62 nM [3H]CP55,940

dHomogenous time resolved fluorescence-based cAMP accumulation

Radioligand binding and cAMP functional studies on CB1 and CB2 receptors were complemented by CB1 receptor modeling and docking of the most potent CB1 receptor agonist of the present series to predict its interactions. We further tested all compounds for their ability to activate or block the two orphan GPCRs GPR18 and GPR55, both of which are known to interact with cannabinoids [26, 27, 28, 29]. We discuss SARs of the newly investigated compounds, integrating previously reported data, thereby providing a comprehensive analysis, which will help to predict properties of novel derivatives.



All compounds except for MDMB-CHMCZCA (41) were obtained from Cayman Chemical (Ann Arbor, MI, USA). According to the manufacturer, the purity of all compounds was declared to be > 95% as determined by liquid chromatography–tandem mass spectrometry (LC–MS/MS). MDMB-CHMCZCA (41) was purchased from We confirmed the purity of all compounds in our laboratories by liquid chromatography–ultraviolet-mass spectrometry (LC–UV-MS) measurements and found it to be generally ≥ 96%, except for MDMB-FUBINACA (12, 93%) and Cl-2201 (37, 86%).

Radioligand binding assays

Radioligand binding assays were performed as described previously [25]. Membrane preparations of Chinese hamster ovary (CHO) cells overexpressing the human CB receptor subtype CB1 or CB2 were incubated in the presence of the test compound and the radioligand [3H]CP55,940 (0.1 nM, see Fig. 1) (Perkin-Elmer Life Sciences, Rodgau-Jügesheim, Germany), for 2 h. Bound and unbound radioligand were separated by rapid filtration through glass fiber GF/C-filters (Perkin-Elmer, Boston, MA, USA), using a Brandel 96-well Harvester (Brandel, Gaithersburg, MD, USA). Radioactivity on the filters was determined by liquid scintillation counting. Three separate experiments were performed, each in duplicates.

cAMP accumulation assays

cAMP accumulation assays were performed also as previously described [25]. Briefly, CHO cells stably expressing the respective human CB receptor subtype CB1 or CB2 were seeded overnight. Then the phosphodiesterase inhibitor Ro-20-1724 [4-(3-butoxy-4-methoxyphenyl)methyl-2-imidazolidone; Sigma-Aldrich, St. Louis, MO, USA], and subsequently the test compound (10–1 µM) and forskolin (10 µM, Sigma-Aldrich) were added. After incubation for 15 min the buffer was removed, and the cells were lyzed. The amount of cAMP was determined in a radioligand binding assay by incubating 50 µL of the cell lysate with 3 nM [3H]cAMP in the presence of protein preparations from bovine adrenal glands (cAMP binding protein) [30]. Bound and unbound radioligand were separated by rapid filtration through GF/B filters, and radioactivity was determined by liquid scintillation counting. To test for antagonistic activity, test compounds were added to Hank’s buffered salt solution (HBSS) containing 10% dimethyl sulfoxide (DMSO), 10 min after the application of Ro-20-1724, and the mixture was incubated for 20 min at 37 °C. Then, the CB agonist CP55,940 was added at a concentration of 0.03 µM, and cAMP determination was carried out as described above [25].

β-Arrestin assays

β-Arrestin assays were performed in recombinant CHO cells expressing either the human GPR18 or the human GPR55 as described before using the β-galactosidase enzyme fragment complementation technology (β-arrestin PathHunter™ assay; DiscoverX, Fremont, CA, USA) [25].

Data analysis

Data were analyzed using GraphPad Prism Version 4.02-6.1, (GraphPad Software, San Diego, CA, USA).

Molecular docking

Molecular docking studies were carried out with the software package Rosetta ( using the 2017.08.59291 build [31, 32]. As templates the X-ray structures 5XRA and 5XR8 were employed [33]; fusion proteins and ligands were deleted and a conformer of MDMB-FUBINACA (12) was manually positioned in an initial model using the PyMOL Molecular Graphics System, Version (Schrödinger, Inc., New York, NY, USA). A conformer library of MDMB-FUBINACA (12) was calculated using the BCL Conformer:Generator [34]. Docking procedure and scripts for data processing are described in supplementary material. Docking scores were calculated using the Rosetta InterfaceAnalyzer. The best scoring models were clustered into a set of plausible binding poses. Results were compared to the pose of THC-like agonists in the template crystal structures 5XRA and 5XR8 and displayed using UCSF Chimera [35].

Results and discussion

Cannabinoid CB1 and CB2 receptor affinities

In the present study, CB1 and CB2 receptor affinities of a new series of synthetic cannabinoids were determined in radioligand binding studies, which provide an ideal basis for the analysis of SARs (Table 1). The investigated compounds comprise indole, indazole, benzimidazole and carbazole derivatives. For some of the compounds, EC50 values had previously been determined by functional assays; however, functional data are highly dependent on the expression level of the receptors or “receptor reserve”, while Ki values obtained in binding studies are largely independent of the employed cellular background [36].

The present set of compounds includes amino acid derivatives. These types of compounds were originally described in a patent and claimed as potential pain therapeutics [37]. In all cases, an alkyl or heteroaryl residue was introduced as R2, and the amino acid was coupled to an amino group in the R1 position (see Table 1; Fig. 2) [37]. The presented compounds feature a pentyl or 5-fluoropentyl side chain in position R2 (for Table 1; Fig. 2). MMB-018 (5), an indole derivative substituted with a valine methyl ester, showed affinity in the low nanomolar range with a Ki value of 15.1 nM at the CB1 receptor and an almost identical Ki value of 14.0 nM at the CB2 receptor. The corresponding indazole AMB (6) was more potent displaying subnanomolar affinity for both CB receptor subtypes (CB1 Ki = 0.866 nM; CB2 Ki = 0.973 nM), indicating the superiority of the indazole core. The 5-fluoropentyl derivatives MMB-2201 (7) and 5F-AMB (8) were similarly as potent as their pentyl analogues MMB-018 (5) and AMB (6), respectively, showing that the terminal fluorination of the pentyl side chain gives almost no effect. Compounds with a p-fluorobenzyl residue or a bioisosteric cyclohexylmethyl residue showed increased affinities in the subnanomolar range in the indazole series (FUB-AMB (9), CB1 Ki = 0.387 nM, and MA-CHMINACA (10), CB1 Ki = 0.339 nM) and were about equipotent at the CB2 receptor. Banister et al. [38] had already investigated these compounds and also 5F-AMB (8) in a fluorescence-based membrane potential assay and determined potencies in the nanomolar range (EC50 values ranging from 1.9 to 71 nM) in that assay, while our radioligand binding assay revealed higher affinities.

The valine methyl ester was replaced by a tert-leucine methyl ester in four of the investigated compounds: 5F-ADB (11), MDMB-FUBINACA (12), MDMB-CHMICA (13) and MDMB-CHMINACA (14), substituted with each 5-fluoropentyl (11), p-fluorobenzyl (12) and cyclohexylmethyl residue (13,14) for R2, respectively. MDMB-FUBINACA (12) was the most potent compound of the entire set of investigated compounds with a Ki value of 0.0985 nM at the CB1 receptor and 0.130 nM at the CB2 receptor. Banister et al. [38] had reported EC50 values of 3.9 nM at CB1 and of 55 nM at CB2 receptors determined in a fluorescence-based membrane potential assay for this compound [38]. MDMB-FUBINACA had caused the highest hypothermal response which the authors had ever observed in rats [38]. These results showed once more that functional assays often do not correctly predict compounds’ affinities. MDMB-CHMICA (13), which also showed subnanomolar affinities for CB1 and CB2 receptors, was previously found to be involved in fatal intoxications, and it was concluded that the compound could cause multiple organ failure with lethal outcome in combination with alcohol [39, 40]. The corresponding indazole MDMB-CHMINACA (14) again showed even slightly higher affinities for both receptors.

Next, compounds with a valinamide substitution (R1) were studied. These were somewhat less potent than the valine methyl esters [compare 5F-AB-PICA (15)/MMB-2201 (7); AB-CHMINACA (20)/MA-CHMINACA (10); and 5F-AB-PINACA (16)/5F-AMB (8)]. 5F-AB-PICA (15), a 5F-pentyl-indole derivative, displayed affinities of 35.0 nM and 89.0 nM for CB1 and CB2 receptors, respectively, while the corresponding indazole 5F-AB-PINACA (16) was more potent displaying affinities in the low nanomolar range. We further investigated the 5Cl-pentyl derivative 5Cl-AB-PINACA (17), which showed comparable Ki values to 5F-AB-PINACA (16) at 4.06 nM for CB1 and 12.0 nM for CB2. The m-fluorobenzyl and the o-fluorobenzyl derivatives (18 and 19) showed similar affinities at the CB1 receptor, as also previously reported by Buchler et al. [37], with Ki values in the nanomolar range, and somewhat lower affinity for the CB2 receptor. AB-CHMINACA (20) displayed low nanomolar CB1 and CB2 affinity in agreement with previous results by Wiley et al. [41].

5F-ADB-PINACA isomer 2 (26) contains a structural isomer of isoleucinamide with a different side chain. This modification resulted in a slight decrease in affinities to CB1 and CB2 as compared to 5F-ADB-PINACA (23), the corresponding tert-leucinamide. Furthermore, tert-leucinamides, have been investigated which contain a tert-butyl group. The 5-fluoropentyl-substituted indole derivative 5F-ADBICA (21) showed nanomolar affinities with a Ki of 2.72 nM at CB1 and 1.83 nM at CB2 receptors. This was in agreement with data published by Banister et al. [42], who had reported similar EC50 values. We found the corresponding indazole derivative 23 to be slightly more potent with Ki values at 1.43 nM for CB1 and 0.694 nM for CB2. Banister et al. had determined a higher potency at CB1 with an EC50 value of 0.24 nM in their membrane potential assay, but a slightly higher EC50 value at CB2 (2.1 nM). The p-fluorobenzyl-substituted indazole ADB-FUBINACA (24) showed even lower Ki values of 0.360 nM for CB1 and 0.339 nM for CB2. The indole ADB-CHMICA (22) was substituted in the R2 position with a cyclohexylmethyl residue and showed a Ki value of 1.24 nM for the CB1 and 0.628 nM for the CB2 receptor. The corresponding indazole MAB-CHMINACA (25), which had been introduced by Buchler et al. [37], was even more potent with a Ki value of 0.333 nM for CB1 and 0.331 nM for CB2, which fits well with data reported by Buchler et al. for CB1 (no data for CB2 had been published by them).

PX-1 (27) and PX-2 (28) are phenylalaninamide derivatives, PX-1 (27) with an indole core and PX-2 (28) with an indazole core structure. PX-2 (28) showed a Ki value for the CB1 receptor of 127 nM and was thus significantly less potent than the corresponding tert-leucinamide derivative 5F-ADB-PINACA (23). The Ki value at CB2 (17.4 nM) was also higher than the Ki value of 0.694 nM determined for 5F-ADB-PINACA (23). Indole derivative PX-1 (27) displayed a Ki value of 485 nM for CB1, corresponding to a fourfold decrease in affinity as compared to the indazole PX-2 (28). The Ki value at CB2 (164 nM) was about tenfold higher. This confirms that the indazole ring system generally leads to a higher affinity as compared to the indole core structure.

APP-FUBINACA (29) and APP-CHMINACA (30) had been introduced by Buchler et al. [37]. Both are indazoles varying in position R2. The p-fluorobenzyl derivative APP-FUBINACA (29) showed potencies for both CB receptor subtypes of around 50 nM, while the corresponding cyclohexylmethyl derivative APP-CHMINACA (30) was more potent displaying Ki values of 9.81 nM for CB1 and 4.39 nM for CB2.

Instead of an amino acid residue, the R1 position has also been substituted with a cumyl moiety. These types of compounds were first described by Bowden and Williamson [43] and it has recently been found in illicit drug material. For all three investigated cumyl derivatives (3133), we could demonstrate affinities in the low nanomolar range for the CB1 receptor. Bowden and Williamson had reported subnanomolar EC50 values in their functional assays using a homogeneous time-resolved fluorescence (HTRF)-based cAMP assay [43]. The indole derivatives Cumyl-PICA (31) and 5F-Cumyl-PICA (32) in our hands displayed potencies of around 25 nM for the CB2 receptor, while Cumyl-THPINACA (33) bearing a 4-tetrahydropyranylmethyl moiety (for R2) was more potent with a Ki value of 1.38 nM at the CB2 receptor, which was similar to its Ki value at the CB1 receptor.

The investigated series of compounds contained one member with a 3-oxycarbonyl linker: MO-CHMINACA (34), an indazole with a cyclohexylmethyl residue for R2 and a methoxycarbonyl-tert-leucine for R1. It displayed a Ki value of 10.4 nM at CB1 and 1.11 nM at CB2 receptors. The only other cyclohexylmethyl-substituted compound investigated by us was BB-22 (see our previous study [25]), which exhibited a Ki value of 0.217 nM for CB1 receptor; however it was substituted with a quinolone for R1 and contained an indole core.

Three 3-carbonylindoles (3537) were studied. FUB-JWH-018 (35), substituted with a naphthyl residue for R1 and possessing a p-fluorobenzyl residue for R2 displayed similarly high nanomolar affinities like the previously studied naphthoyl indazoles THJ018 and THJ2201 [25]. MAM-2201 and EAM-2201, which were substituted with methyl or ethyl in the 4-position of the naphthoyl residue, had shown subnanomolar affinities [25]. Here we report F-2201 (36) and Cl-2201 (37), the respective 4-fluoro- and 4-chloro derivatives. Both displayed high affinities at 1–2 nM for both CB1 and CB2 receptors. The previously described alkyl-substituted naphthoyl derivatives had shown similar potencies (compare MAM-2201 and EAM-2201) [25]. The substitutions can be ranked in the following order of potency at CB1: ethyl > fluoro > chloro > methyl, while for CB2 it was: ethyl > methyl > fluoro ≈ chloro.

The indole derivative mepirapim (38) belongs to the 3-amido-substituted derivatives, featuring a 4-methylpiperazinyl residue for R1. Mepirapim (38) was originally identified by Uchiyama et al. [44] and has been found in "Spice" preparations. We determined an affinity of 2650 nM for the CB1 receptor and 1850 nM for the CB2 receptor. Therefore, it can be regarded as a rather weak CB receptor ligand.

We further investigated three structurally dissimilar compounds, 3941, which contain a carbazole core substituted in position 3 with residues typically observed in position R1 of indazole- and indole-based compounds. EG-018 (39) and EG-2201 (40) feature a carbonyl linker connected to a naphthyl residue, whereas MDMB-CHMCZCA (41) is substituted with a methoxycarbonyl-tert-leucine residue through an amide linker. EG-018 (39) displayed low nanomolar affinities with Ki values of 7.17 nM for CB1 and of 2.27 nM for the CB2 receptor. EG-018 (39) can be compared to JWH-018 (3), which showed similar affinities. EG-2201 (40) was less potent at CB1 with a Ki value of 22.4 nM, but only slightly more potent at CB2 (Ki = 4.36 nM). MDMB-CHMCZCA (41) also displayed affinities in the low nanomolar range. The observed switch from indoles and indazoles to carbazoles can be interpreted as a reaction to the NpSG legislation and similar regulations in other countries that restricted the whole class of indoles and indazoles based on the known SARs. Recently, the synthetic cannabinoid Cumyl-PEGACLONE was identified as one of the first cannabimimetic compounds to circumvent these regulations; it consists of a γ-carboline, another new scaffold for cannabinoid receptor agonists [45]. Carbazoles (3941) represent a further new scaffold which circumvents restrictions applied by many, especially European, countries by simply exchanging the well-established bicyclic core structures of indole or indazole for a tricyclic carbazole ring system.

We further investigated the benzimidazole derivative FUBIMINA (42), which had previously been described by Wiley et al. [41], and determined a Ki value of 502 nM at the CB1 receptor, which is in the same range as the reported Ki value of 296 nM, and a Ki value of 99.0 nM for the CB2 receptor, which is slightly higher than the reported value of 23.5 nM [41].

The presently investigated set of compounds complements our previous efforts to study the SARs of synthetic cannabinoids [25]. Of special interest is the observed scaffold hopping. Carbazole derivatives with a high affinity for CB receptors circumvent restriction by current law and display a new lead structure for CB receptor ligands. Further insight into the SARs is required to describe the potency profile of this compound class in more detail.

cAMP accumulation assays

As a next step, we investigated the compounds in cAMP accumulation assays, to obtain information on their functionality (Fig. 3). CB receptors are Gi protein-coupled and thus reduce the levels of cAMP in the cells upon activation. We applied the compounds at